Reinhard Noll (Auth.) - Laser-Induced Breakdown Spectroscopy - Fundamentals and Applications-Springer-Verlag Berlin Heidelberg (2012)

Laser-Induced Breakdown Spectroscopy
•
Reinhard Noll
Laser-Induced Breakdown
Spectroscopy
Fundamentals and Applications
123
Dr. Reinhard Noll
Fraunhofer-Institut für Lasertechnik (ILT)
Steinbachstr. 15
52074 Aachen
Germany
reinhard.noll@ilt.fraunhofer.de
ISBN 978-3-642-20667-2 e-ISBN 978-3-642-20668-9

DOI 10.1007/978-3-642-20668-9
Springer Heidelberg Dordrecht London New York
Library of Congress Control Number: 2011940321
c Springer-Verlag Berlin Heidelberg 2012

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,
1965, in its current version, and permission for use must always be obtained from Springer. Violations
are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

In loving memory of my mother Maria,
who passed away November 12th, 2011
•
Preface
This book is meant as comprehensive source of the fundamentals and applications of

laser-induced breakdown spectroscopy (LIBS). It provides a systematic introduction
to the principles, the dominant process parameters, and the instrumental components
for LIBS. The effect of multiple pulses on material ablation, plasma dynamics, and
plasma emission is presented in detail. The double-pulse approach is described,
and the underlying physics by integral as well as temporally and spatially resolved
plasma diagnostics is clarified. A heuristic plasma modeling allows to simulate
complex experimental plasma spectra. These methods and findings form the basis
for a variety of applications to perform quantitative multielement analysis with
LIBS.
These application potentials of LIBS have really boosted in the last years ranging
from bulk analysis of metallic alloys and nonconducting materials, via spatially
resolved analysis and depth profiling covering measuring objects in all physical
states: gaseous, liquid, and solid. Dedicated chapters present LIBS investigations
for these tasks with special emphasis on the methodical and instrumental concepts,
as well as the optimization strategies for a quantitative analysis.
Requirements, concepts, design, and characteristic features of LIBS instruments
are described covering laboratory systems, inspection systems for inline process
control, mobile systems, and remote systems.
Based on this applied research and development, LIBS was pushed forward sig-
nificantly achieving limits of detection for quantitative trace element determination
and measuring frequencies not achieved so far. Industrial applications of LIBS sys-
tems are presented demonstrating the benefits of inline process control for improved
process guiding and quality assurance purposes. Among these the identification
testing of pipe fittings with LIBS impressively shows the first routine application
established where more than five million products were tested automatically in a
production line within a period of less than 2 years.
The author moved into this challenging subject in the early nineties and the
present book has grown out of work performed at the Fraunhofer-Institut für
Lasertechnik (ILT) in Aachen, Germany. Thanks are due to many colleagues
and students at ILT as well as former coworkers who continued their career in
vii
viii Preface
industry and universities for their contributions and discussions (in alphabetic
order): Ü. Aydin, H. Balzer, H. Bette, R. Bleich, A. Brysch, D. Eilers, R. Fleige,
C. Fricke-Begemann, C. Gehlen, M. Gröbner, C. Haas, J. Hertzberg, F. Hilbk-
Kortenbruck, M. Höhne, S. Hölters, P. Jander, C. Janzen, R. Klauke, O. Klein,
H. Krause, M. Kraushaar, T. Kuhlen, H. Kunze, A. Lamott, F. Legewie, A. Löbe,
J. Makowe, I. Mönch, L. Peter, S. Pflüger, R. Quay, P. Roth, R. Sattmann,
M. Sellhorst, M. Stepputat, N. Strauss, V. Sturm, J. Vrenegor, Q. Wang, P. Wer-
heit, R. Wester, E. Wiens, S. Winkelmann, and M. Wypadlo.
The author likes to express his special thanks to Mrs. Anna Mosna, who drew
many of the figures and illustrations shown in this book with great diligence.
Very special thanks go to the author’s wife for her help, patience, and encourage-
ment.
Aachen, Reinhard Noll

August 2011
Contents
1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 5
2 Laser-Induced Breakdown Spectroscopy . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 7
2.1 Principle and Measuring Parameters . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 7
2.2 Setup for Laser-Induced Breakdown Spectroscopy . . . . . . . . . . . . . . . . 12
2.3 Measuring Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 14
2.4 Applications and Methodical Extensions . . . . . . .. . . . . . . . . . . . . . . . . . . . 14
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 15
3 Process Parameters .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
3.1 Laser Pulse Energy and Repetition Rate . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
3.2 Temporal Characteristics of the Laser Pulse . . .. . . . . . . . . . . . . . . . . . . . 22
3.3 Laser Wavelength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 25
3.4 Beam Quality and Focusing . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 27
3.5 Ambient Gas Atmosphere . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 29
3.6 Spatial Resolution and Spatial Averaging . . . . . .. . . . . . . . . . . . . . . . . . . . 33
3.7 Direction of Incidence of the Laser Beam and Direction
of Observation of the Plasma Emission . . . . . . . .. . . . . . . . . . . . . . . . . . . . 38
3.8 Warming-up Pulses, Prepulses, and Measuring Pulses . . . . . . . . . . . . 39
3.9 State of the Measuring Object . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 43
4 Instrumental Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 47
4.1 Laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 47
4.2 Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 51
4.3 Beam Guiding Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 56
4.4 Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 61
4.5 Signal Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 65
4.6 Sample Stand and Measuring Chamber . . . . . . . .. . . . . . . . . . . . . . . . . . . . 66
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 70
ix
x Contents
5 Evaporation and Plasma Generation .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 75

5.1 Evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 75
5.2 Plasma Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 78
5.3 Absorption of Laser Radiation in the Plasma . .. . . . . . . . . . . . . . . . . . . . 79
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 81
6 Multiple Pulses for LIBS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
6.1 Double Pulses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 84
6.2 Triple and Multiple Pulses . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 92
6.3 Tailored Pulse Trains . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 94
7 Material Ablation .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
7.1 Crater Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 97
7.2 Ablated Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
7.3 Material Ablation in Different Ambient Gases
by Collinear Multiple Laser Pulses . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 102
7.4 Material Ablation by Collinear Double Laser Pulses
at Reduced Ambient Gas Pressures . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 110
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 117
8 Plasma Dynamics and Plasma Parameters . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 119
8.1 Expansion and Decay of the Plasma . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 119
8.2 Spatially Integrated Electron Density and Temperature . . . . . . . . . . . 132
8.3 Spatially Resolved Electron Density . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 152
8.4 Characteristic Time Scales and Local Temperature
Equilibrium Conditions of the Laser-Induced Plasma . . . . . . . . . . . . . 160
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 166
9 Plasma Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 167
9.1 Bremsstrahlung and Recombination Radiation . . . . . . . . . . . . . . . . . . . . 167
9.2 Line Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 169
9.3 Absorption and Optical Thickness . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 171
9.4 Line Broadening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 173
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 182
10 Modeling of Plasma Emission . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 185
10.1 Heuristic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 189
10.2 Calculation of Emission Spectra . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 193
10.3 Simulation of Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 198
10.3.1 Fe Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 198
10.3.2 Al Resonance Line with Self-reversal . . . . . . . . . . . . . . . . . . . . 201
10.4 Enthalpy of Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 203
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 204
11 Quantitative Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 207
11.1 Measuring Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 207
11.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 212
Contents xi
11.3 Measurement and Determination of Analytes . .. . . . . . . . . . . . . . . . . . . . 216

11.4 Recalibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 216
11.5 Interelement Correction . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 218
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 220
12 Combination of LIBS and LIF . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 228
13 Bulk Analysis of Metallic Alloys . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
13.1 Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
13.1.1 Solid Steel Samples . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
13.1.2 Solid Steel Samples with Scale Layer . . . . . . . . . . . . . . . . . . . . 235
13.1.3 Liquid Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 245
13.2 High-Alloy Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 252
13.3 Aluminum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 264
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 271
14 Bulk Analysis of Nonconducting Materials . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 275
14.1 Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 275
14.1.1 Identification of Plastic Materials . . . . .. . . . . . . . . . . . . . . . . . . . 275
14.1.2 Detection of Heavy Metals and Flame
Retardants in Technical Polymers . . . . .. . . . . . . . . . . . . . . . . . . . 285
14.2 Slag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 303
14.2.1 Converter Slags . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 304
14.2.2 Vacuum Slag . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 314
14.3 Soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 318
14.4 Cement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 326
14.5 Droplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 337
14.6 Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 349
14.7 Particulates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 355
14.8 Aerosoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 361
14.8.1 Aerosoles Sampled on Substrates . . . . .. . . . . . . . . . . . . . . . . . . . 361
14.8.2 Aerosoles in an Air Stream . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 372
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 380
15 Spatially Resolved Analysis.. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 387
15.1 High-Speed LIBS for Microanalysis . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 387
15.2 Element Mappings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 396
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 398
16 Depth Profiling .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 401
16.1 Measurement of the Thickness of Coatings . . . .. . . . . . . . . . . . . . . . . . . . 401
16.2 Characterization of Depth Profiles . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 416
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 426
17 LIBS Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 429
17.1 Laboratory Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 429
17.2 Inspection Systems for Inline Process Control .. . . . . . . . . . . . . . . . . . . . 437
xii Contents
17.3 Mobile Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 447

17.4 Remote Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 454
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 463
18 Industrial Applications.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 467
18.1 Identification Testing of High-Alloy Pipe Fittings . . . . . . . . . . . . . . . . . 467
18.2 Analysis of Slags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 469
18.3 Characterization of Inclusions and Segregations . . . . . . . . . . . . . . . . . . . 472
18.4 Recycling of Aluminum Scrap . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 484
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 488
A Annex . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 491
A.1 Displacement of Ambient Atmosphere . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 491
A.2 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 494
A.3 List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 498
A.4 Stark Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 511
A.5 Wavelengths and Elements . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 511
A.6 Spectral Line Selection for Boltzmann Plots . . .. . . . . . . . . . . . . . . . . . . . 511
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 533
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 535
Chapter 1
Introduction
Laser radiation is a high-quality form of electromagnetic energy enabling a mul-

titude of new methods and applications, such as material processing, biomedical
and communication technologies, and measuring methods. The advantages of lasers
in measuring technologies are the noncontacting measurement, high flexibility,
and high measuring speeds. Due to these features, laser measuring methods and
applications have encountered a dynamic development during the last years. Laser
measuring methods were introduced successfully in production technology, process
engineering, quality assurance, environmental technology, and life sciences. The
most important measuring quantities are:
• Geometrical quantities, such as distance, contour, shape, roughness, strain
• Dynamic quantities, such as velocities, vibration modes
• Thermodynamic and chemical quantities, such as temperature, density, concen-
tration, chemical bonds
The above-mentioned quantities are determined preferably with the use of spec-
troscopic methods. From the very first the invention of laser was tightly linked
with spectroscopic questions [1.1]. The special properties of laser radiation enable
a variety of new spectroscopic methods to analyze the chemical constituents of
a substance or to determine their physical state. Examples are laser absorption
spectroscopy, light detection and ranging (LIDAR), laser-induced fluorescence
(LIF), and coherent antistokes Raman spectroscopy (CARS) [1.2, 1.3]. On the one
hand, the tunability of the laser wavelength to atomic or molecular transitions allows
for a high selectivity, and on the other hand the high spectral brightness of laser
radiation enhances the detection sensitivity enabling the determination of traces.
Laser spectroscopic methods are a powerful tool for fundamental investigations,
such as high-resolution spectroscopy within the Doppler width to study the fine
structure of excited states [1.4]. Femtosecond laser pulses allow observing directly
the dynamics of chemical reactions. The Nobel prize for physics in the year 1999
was awarded to A. Zerwail for his pioneering work in this field [1.5]. In an
increasing degree laser spectroscopy enters into new application fields. Among
these are the remote investigation of harmful substances in the atmosphere, the
R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9 1, 1

© Springer-Verlag Berlin Heidelberg 2012
2 1 Introduction
monitoring of combustion processes or material-dependent production processes,

and the quality assurance of semi-finished products [1.6, 1.7, 1.8, 1.9].
For technical applications, laser spectroscopic methods are of special interest as
they are able to determine several species simultaneously with minimum equipment.
Laser-induced breakdown spectroscopy (LIBS) belongs to these methods [1.10].
Fundamentals and applications of LIBS are the subject of this book. LIBS is able
to analyze solid, liquid, and gaseous substances. In principle, laser absorption
spectroscopy and LIF are also able to determine several species, but in this
case different laser wavelengths are necessary, which cause a corresponding high
instrumental effort for technical applications.
Focusability and temporal modulation of the power of the laser beam are the
two main features of laser radiation which are exploited for LIBS. At irradiances
above 109 W=cm2 a solid substance to be studied evaporates locally within a short
time forming a transient plasma. In the plasma, the ablated atoms and ions are
excited and emit their characteristic line radiation [1.11]. Emission wavelengths
cover a broad spectral range from the deep ultraviolet to the infrared. A short time
after the invention of the laser the emission features of the laser-induced plasma
were investigated [1.12]. However, the used ruby lasers could only be operated at
low repetition rates. Additionally the emission of the plasma was observed time
integrated, so only a limited analytical quantification was achievable.
The dynamic technological developments in the field of solid state lasers, electro-
optic detectors, and signal processing during the last decade were successfully
utilized for LIBS. The analytical performance for a multielement analysis achieves
a level, which equals or is even better than that of classical methods [1.13].
The vital point is that due to optical excitation of the measuring radiation over
a distance, new fields of application were opened up, which are not accessible
for conventional physical methods, such as X-ray fluorescence analysis or spark
emission spectrometry. Such fields are, e.g., the inline analysis of liquid steel or the
material identification for sorting and quality assurance tasks [1.14, 1.15]. Quality
assurance close to the process on the basis of an automated inline measurement is
the first step to fast and efficient feedback actions which result in significant cost
savings and improved competitiveness.
Pulsed laser radiation can be used not only for analysis, but also for the
preparation of substances to be investigated. Surface layers, as, e.g., corrosion
layers, oxides, scale, contaminations, whose chemical composition differs from
one of the subjacent substances, can be ablated locally by the laser beam.
A mechanical preparation step, by milling, turning, or grinding, is not necessary in
many cases. This dual use of the laser beam as ablation and analyzing tool has a
high technological potential in terms of automation capabilities for inline inspection
systems.
With diode-pumped solid state lasers, compact radiation sources of high beam
quality are available offering electro-optical Q-switching with pulse repetition rates
of 1 kHz and beyond, average powers of several watts, and excellent beam qualities
[1.16]. The operating life time of these lasers amounts to more than 20,000 h. This
opens for LIBS the area of spatially resolved determinations of element distributions
1 Introduction 3
on the surface of macroscopic samples. An example of this approach is the analysis

of the cleanliness of steel samples, where microscopic inclusions in a steel matrix
are determined, consisting of hard materials as, e.g., carbides and nitrides. These
inclusions influence the processability and the material properties of steel grades.
While competing methods such as scanning electron microscopy (SEM) for a
spatially resolved energy-dispersive X-ray fluorescence analysis (SEM-EDX) or
the electron microprobe require high vacuum conditions and an extensive sample
preparation such as diamond polishing, the laser-based method measures under an
inert gas atmosphere and a simple preparation of the sample surface by milling or
grinding is sufficient.
The features of LIBS compared with those of other laser spectroscopic methods
are given in Table 1.1. Laser-induced breakdown spectroscopy is the only method,
which allows for a simultaneous multispecies analysis in all states of aggregation of
a substance.
Table 1.1 Comparative survey of laser spectroscopic methods with exemplary data of the used
laser sources and measuring quantities
Method LIBS LIF Raman CARS LAS
exc (nm) 266, 532, 200–800; 337 347, 694; 532, 607–609 760, 1,570,
1,064 488, 515, 1,800;
532; 266 3,160–
3,670;
10,600
Q .cm1 / 1 0.5; 1.2 <1 0.8; 180 <0:05
texc (ns) 5–100 10–cw 10–cw 10 6–cw
I.W=cm2 / 108 –1011 103 –106 103 –106 109 –1011
Measuring g, l, s, g, l, atoms, g, l, s, g, l, s, g, molecules
species atoms molecules molecules molecules
Measuring ci ci ci , T T; ci ci
quantity
Measuring >1 g=g >0:01 g=g >10 g=g, 300–3,500 K, >1 ng=g
range 300– 1 mbar–
3,500 K 100 bar
Measuring 10–1,500 10–1,000 10–1,000 50–1,000 500–4,000
distance
(mm)
Spatial >1 m >1 m >1 m >1 mm3 ./.
resolution
Minimum 10 s <10 s 10 ns 10 ns 100 ms
measuring
time
References [1.13], [1.19], [1.20], [1.22], [1.23], [1.25], [1.26] [1.7], [1.27],
[1.17], [1.21] [1.24] [1.28]
[1.18]
LIBS laser-induced breakdown spectroscopy, LIF laser-induced fluorescence, CARS coherent anti-
Stokes Raman spectroscopy, LAS laser absorption spectroscopy, exc wavelength of the laser used
for excitation, Q bandwidth of the laser, texc pulse duration of the laser beam, I irradiance at the
location of interaction, g gaseous, l liquid, s solid, ci concentration of species i , T temperature,
cw continuous wave
4 1 Introduction
500
ISI web of knowledge, Jan. 2011
400
number of publications
300
200
100
0
1980 1985 1990 1995 2000 2005 2010
year
Fig. 1.1 Number of publications in the field of laser-induced breakdown spectroscopy or LIBS in
the years 1980–2010 from a database search in ISI Web of Knowledge
The number of publications in the field of LIBS shows a strong growth in the last
years [1.29]. Figure 1.1 shows the result of a database search from January 2011
using the search term “laser-induced breakdown spectroscopy or LIBS” in the ISI
Web of Knowledge showing the nearly exponential growth of publications in the
last years reaching more than 450 publications dealing with LIBS in the year 2009.
The purpose of this book is to give a comprehensive introduction to the basic
methodological approaches, the relevant process parameters, and the instrumental
components being subject of Chaps. 2–4. The interaction of the laser beam with
the measuring object leading to evaporation and plasma formation is presented in
Chap. 5. The use of multiple pulses for LIBS has led to a significant progress in the
analytical performance, and Chap. 6 gives an overview of the various concepts and
achieved improvements. Their effect on the mass ablation and plasma dynamics
is discussed in Chaps. 7 and 8. Chapter 9 deals with the various emission pro-
cesses of the laser-induced plasma such as continuum emission and line emission.
A calculation of the plasma emission based on a heuristic model is presented in
Chap. 10. The procedures to be used for quantitative measurements with LIBS
are given in Chap. 11. The combination of the two laser spectroscopic methods
LIBS and LIF is the topic of Chap. 12. Chapters 13–16 present various analytical
investigations performed with LIBS ranging from the bulk analysis of metallic
matrices via nonconducting materials to spatially resolved analysis and depth
profiling. LIBS instruments are described in Chap. 17. Industrial applications of
LIBS are the subject of Chap. 18. Finally, the Annex comprises an estimate of the
necessary energy to displace the ambient atmosphere by the expanding plasma, lists
of abbreviations and symbols, a compilation of Stark data important for LIBS, and
finally a description of a spectral line selection method for Boltzmann plots.
References 5
References
1.1. Ch. Townes, The birth of the laser. Opto Laser Eur. 69, 20–30 (1999)
1.2. W. Demtröder, Laserspektroskopie (Springer-Verlag, Berlin, 1999)
1.3. C. Weitkamp, LIDAR – Range-Resolved Optical Remote Sensing of the Atmosphere
(Springer-Verlag, Berlin, 2005), ISBN 0-387-40075-3
1.4. C. Duke, H. Fischer, H.J. Kluge, H. Kremmling, Th. Kühl, E.W. Otten, Determination of
isotope shift of 190 Hg by on-line laser spectroscopy. Phys. Lett. 60A, 303–306 (1977)
1.5. G. Gerber, Die Momentaufnahme der Molekülspaltung. Phys. Blätter 55, 23–25 (1999)
1.6. M. Proffitt, A. Langford, Ground based differential LIDAR system for day or night
measurement of ozone throughout the free troposphere. Appl. Optics 36, 2568–2585 (1997)
1.7. R. Mihalcea, D. Baer, R. Hanson, A diode-laser absorption sensor system for combustion
emission measurements. Meas. Sci. Technol. 9, 327–338 (1998)
1.8. L. Paksy, B. Nemet, A. Lengyel, L. Kozma, J. Czekkel, Production control of metal alloys
by laser spectroscopy of molten metals. Part 1. Preliminary investigation. Spectrochim. Acta
B 51 279–290 (1996)
1.9. R. Noll, L. Peter, I. Mönch, V. Sturm, Automatic laser-based identification and marking
of high-grade steel qualities, in Progress in Analytical Chemistry in the Steel and Metals
Industries, ed. by R. Tomellni (European Communities, Luxembourg, 1999), pp. 345–351
1.10. D. Cremers, The analysis of metals at a distance using laser-induced breakdown spec-
troscopy. Appl. Spectrosc. 41, 572–579 (1987)
1.11. R. Adrain, J. Watson, Laser microspectral analysis: a review of principles and applications.
J. Phys. D Appl. Phys. 17, 1915–1940 (1984)
1.12. E. Runge, R. Minck, F. Bryan, Spectrochemical analysis using a pulsed laser source.
Spectrochim. Acta 20, 733–736 (1964)
1.13. V. Sturm, L. Peter, R. Noll, Steel analysis with laser-induced breakdown spectrometry in the
vacuum ultraviolet. Appl. Spectrosc. 54, 1275–1278 (2000)
1.14. C. Carlhoff, Laserinduzierte Emissionsspektroskopie für die Direktanalyse von flüssigem
Stahl im Konverter. Laser Optoelektr. 23, 50–52 (1991)
1.15. H. Sattler, Metallschrotte nach Analyse automatisch sortieren. Materialprüfung 35, 312–315
(1993)
1.16. P. Peuser, N. Schmitt, Dioden-gepumpte Festkörperlaser (Springer-Verlag, New York, 1995)
1.17. Final report of the Brite-Euram project: study of emission spectroscopy on laser produced
plasma for localised multielemental analysis in solids with surface imaging, Coordinator
Saclay, France, Contract MAT 1-CT-93–0029, 1996
1.18. L. Cabalin, D. Romero, J. Baena, J. Laserna, Effect of surface topography in the character-
ization of stainless steel using laser-induced breakdown spectrometry. Surf. Interface Anal.
27, 805–810 (1999)
1.19. W. Sdorra, A. Quentmeier, K. Niemax, Basic investigations for laser microanalysis. II.
Laser-induced fluorescence in laser-produced sample plumes. Mikrochim. Acta II 2, 201–
218 (1989)
1.20. R. Wamsley, T. O’Brian, K. Mitsuhashi, J. Lawler, Laser-induced fluorescence on HgC in
Hg-Ar discharges. Appl. Phys. Lett. 59, 2947–2949 (1991)
1.21. J. Bublitz, M. Dickenhausen, M. Grätz, S. Todt, W. Schade, Fiber-optic laser-induced
fluorescence probe for the detection of environmental pollutants. Appl. Optics 34, 3223–
3233 (1995)
1.22. P. Kasal, B. Mewes, D. Brüggemann, Lasergestützte Verbrennungsdiagnostik, VDI-
Broschüre zur Abschlußpräsentation von BMBF-Verbundprojekten, Düsseldorf, 1997, S.
1–13
1.23. B. Schrader (ed.), Infrared and Raman Spectroscopy (VCH Verlagsgesellschaft, Weinheim,
1995)
1.24. P. Jander, R. Noll, Automated detection of fingerprint traces of high explosives using
ultraviolet Raman spectroscopy. Appl. Spectrosc. 63, 559–563 (2009)
6 1 Introduction
1.25. A.C. Eckbreth, Laser Diagnostics for Combustion Temperature and Species (Abacus,
Tunbridge Wells, 1988)
1.26. R. Clark, R. Hester (eds.), Advances in Non-linear Spectroscopy (Wiley, Chichester, 1988)
1.27. M. Seiter, M. Sigrist, On-line multicomponent trace-gas analysis with a broadly tunable
pulsed difference-frequency laser spectrometer. Appl. Optics 38, 4691–4698 (1999)
1.28. T. Fink, S. Büscher, R. Gäbler, Q. Yu, A. Dax, W. Urban, Laser intracavity photoacoustic
spectrometer for trace gas analysis. Rev. Sci. Instrum. 67, 4000–4004 (1996)
1.29. R. Noll, Terms and notations for laser-induced breakdown spectroscopy. Anal. Bioanal.
Chem. 385, 214–218 (2006)
Chapter 2
Laser-Induced Breakdown Spectroscopy
In this chapter, the principle of laser-induced breakdown spectroscopy is presented.

The intention is to give an overview of the main parameters of the method,
which will be discussed in more detail in Chap 3. By means of a typical setup of
an analyzing device based on LIBS, the important instrumental components are
introduced. These components are the subject of Chap 4. In Sect. 2.3, the single
steps of an analysis are introduced.
2.1 Principle and Measuring Parameters
The principle of laser-induced breakdown spectroscopy is shown schematically in

Fig. 2.1.
A pulsed laser beam is focused onto the surface of a substance to be analyzed,
see (1) in Fig. 2.1. Radiation energy is locally coupled into the material (2) and
the material starts to evaporate (3). Within this material vapor and the surrounding
gas atmosphere a plasma is generated (4), leading to the excitation of the material
constituents and their spontaneous emission of radiation. The plasma decays and
emits element-specific radiation (5)–(7). This emission is resolved spectrally and
is detected by a spectrometer. For solid substances, a crater is formed finally (8).
The evaporated material is removed partially from the interaction zone driven by the
intrinsic dynamics of the plasma expansion and by an externally impressed gas flow.
The process denoted in phase (3) as “evaporation” is a simplified description and
refers to solid inorganic substances. In general, there is no pure sublimation from
the solid phase to the gaseous phase. Depending on the laser parameters – e.g., the
pulse duration – and the material properties a transient liquid phase may also exist.
Besides the evaporation and dissociation processes also particles are ablated, e.g.,
small particulates or droplets, which are ejected due to the impact of the pressure
exerted by the plasma and the accompanying shock waves on a melt layer [2.1]. For
organic substances, the initial material is disintegrated, fragmented, and dissociated.

8 2 Laser-Induced Breakdown Spectroscopy
Fig. 2.1 Principle of laser-induced breakdown spectroscopy shown in phases 1–8; LB D incoming
laser beam, S D sample, H D region of energy deposition, V D material vapor, P D plasma, E D
element-specific emission, CR D crater, PT D particles . The times given depict the temporal
evolution after start of irradiation of the laser pulse
tHE life time of the plasma depends on the laser beam parameters chosen, the
conditions of the surrounding gas atmosphere, and the substance to be analyzed.
The life time lies typically in the range of 0:5–10 s. The whole process depicted
in Fig. 2.1 can be repeated with frequencies of 10 Hz up to 1 kHz (see Chap 15,
Sect. 15.3).
Figure 2.2 shows schematically some of the measuring parameters and the
spectrum emitted from the plasma. The measuring parameters will be described
in more detail in Chap 3. The incident laser irradiance Ii L .t/ at the wavelength L
is a function of time (the spatial dependence is not discussed here). The temporal
structure of the laser pulse has a decisive influence on the generated plasma states
and hence the emitted spectrum, see also Sect. 3.2 and Chap. 6. The quantity s
describes the position of the beam waist of the laser beam in relation to the sample
surface for a solid or a liquid measuring object. A part of the incident irradiance of
the laser beam is reflected by the sample and the plasma in the backward direction,
and the other part is scattered in various directions. These intensities are denoted
with Ir L and IsL . Besides the irradiance Ii L .t/, the ambient gas pressure pg and
the type of the ambient gas have an influence on the plasma dynamics (see Sect. 3.5
2.1 Principle and Measuring Parameters 9
Fig. 2.2 Schematic illustration of some measuring parameters and the emitted spectrum; LB D
incident laser beam, S D sample, BW D beam waist, Ii L .t / D irradiance of the incident laser
beam, Ir L .t / D reflected laser irradiance, IsL .t / D scattered laser irradiance, pg D ambient gas
pressure, s D beam waist position, j D emission line of an analyte, r D emission line of a
reference line, tdelay D delay time between the laser pulse and the start of the integration window
to record the spectrum, tint D width of the integration window
Fig. 2.3 Schematic

illustration of the emission
spectra of the laser-induced
plasma for different time
delays with respect to the
irradiation of the laser pulse
and Chap 7). The emitted spectrum of the plasma I .t/ is also a function of time.
Figure 2.2, bottom right, shows schematically a spectrum detected at a time tdelay
with respect to the irradiation time of the laser pulse and integrated for a duration
of tint .
During the life time of the plasma the emission spectrum changes. Figure 2.3
illustrates schematically emission spectra of the laser-induced plasma for three
different time delays after the irradiation of the laser pulse. At the time t1 , the
plasma emits predominantly a continuous spectrum, caused by free–free transitions
of electrons. Only small peaks of the line intensities of atoms and ions are visible,
and the ratio of the peak intensity of an emission line to the neighboring intensity of
the spectral continuum emission is low. At the time t2 , the plasma has cooled down
and the intensity of the line emission, as well as the ratio peak intensity to continuous
background, increases significantly. At t3 , the plasma temperature decreases further

and the emission intensities decrease.
Figure 2.4a, b illustrates the incident laser pulse and the plasma emission at an
element-specific wavelength i as a function of time. Experimental investigations
have shown that the time delay between the rising of the laser pulse and the rising
of the plasma emission is small being in the order of <1ns. Hence, in the following
paragraphs we can assume that the signals shown in Fig. 2.4a, b start simultaneously,
see dashed vertical line. This start time is defined as the reference time for the
delay time tdelay . The variation in time of the plasma emission at an element-
specific wavelength shown in Fig. 2.4b reveals different signal components. The
full width at half maximum (FWHM) 1 of the first maximum lies typically in the
range of 40–500 ns. This peak is attributed to the strong continuum emission of the
plasma in the early phases (corresponds to tdelay D t1 in Fig. 2.3). After this first
generally distinct maximum, a further temporal variation is observed characterized
by a significantly greater time scale. In many cases, e.g., for atomic lines with upper
energy levels around 3 eV, a second maximum occurs, whose FWHM 2 is between
2 and 10 s. This signal component represents the element-specific line emission to
be evaluated for LIBS.
The following measurands are used for LIBS:
(a) The intensity of the plasma emission at discrete wavelengths as a function of
time is: I. D i ; t/, where i denotes a set of wavelengths of element-
specific lines (for simplicity of notation the index at the quantity I to denote
Fig. 2.4 Schematic

illustration of the temporal
variation of: (a) laser pulse,
(b) plasma emission at an
element-specific wavelength
i . I. D i ; t / D intensity
c
of the plasma emission at the
wavelength i ; 1 , 2 D
FWHM times of the two
signal components. (c) Gate
signal defining the position
and duration of the
integration window; tdelay ,
tint D cf. Fig. 2.2
2.1 Principle and Measuring Parameters 11
the spectral intensity is omitted). Usually this signal is integrated over a time
interval [tdelay , tdelay C tint ].
(b) A spectrum, which is integrated over a time interval [tdelay , tdelay C tint ]: S./ D
R tdelay Ctint
tdelay I.; t/dt with tdelay D start of integration interval with respect to the
start of the laser pulse and tint D temporal width of the integration window.
The measurands (a) and (b) are used for the quantitative determination of the
chemical composition of a substance. As a rule this composition is described
by the concentrations of the chemical elements expressed as the mass of the
analyte in relation to the total mass using the unit g/g or g=g for traces. For
higher concentrations, the unit m.-% is used in the following expressing the mass
percentage of an analyte. For quantitative measurements, the method has to be
calibrated using reference samples to determine the functional dependence between
the measurands and resultant quantities on the known concentrations of a set of
reference samples, cf. Sect. 11.2.
Figure 2.5 illustrates the procedure for a quantitative measurement. The left
diagram shows a spectrum S./ with a number of emission lines depending on
the composition of the sample to be analyzed. The spectral position of the lines
has to be allocated to the respective elements on the basis of literature data as, e.g.,
those given in [2.2, 2.3]. The height of a line is a measure of the concentration of
the respective element in the sample. However, this line intensity depends also on a
number of other factors, e.g., the laser pulse energy, plasma temperature, plasma
size, atomic parameters of the line transition, sample surface, detector response
function. Generally, the influence of these factors can be reduced by taking the ratio
of the intensity of an analyte line j to the line intensity of a dominant element r
of the sample, a so-called matrix element cf. (11.2), (11.3). This line acts as an
internal standard or reference. To gain quantitative results, the spectral signals
have to be calibrated using a set of samples with known chemical composition,
e.g., certified reference samples (CRM). Figure 2.5, right, shows a calibration curve
where the intensity ratio Ij =Ir is plotted as a function of the concentration cj of these
reference samples. In general, the calibration curve shows a nonlinear behavior. For
an unknown sample, the intensity ratio Ij =Ir is measured, yielding the concentration
of the analyte via the analysis function (also called working curve) which is the
Fig. 2.5 Left: emission spectrum of laser-induced plasma with analyte and reference line. Right:
calibration curve
inverse function of the calibration curve (for simplicity, this step is illustrated by
the arrows in Fig. 2.5, right). The procedure for quantitative measurements will be
described in more detail in Chap 11.
2.2 Setup for Laser-Induced Breakdown Spectroscopy
Figure 2.6 shows the principle setup for laser-induced breakdown spectroscopy.
A mirror guides the pulsed laser radiation to a focusing lens. The sample to be
analyzed is placed in a measuring chamber. As a rule the incident direction of
the laser radiation is oriented perpendicularly to the sample surface. The focused
radiation generates a plasma at the sample surface. The emission of this plasma is
observed in a direction, which includes an angle ˛O to the incident direction of the
laser radiation. In Fig. 2.6, the measuring radiation is transmitted via a fiber optics
to a spectrometer, where it is spectrally dispersed and converted to electrical signals.
The measuring chamber is gas tight. Laser radiation and measuring radiation
are transmitted via built-in windows. Via gas fittings the type of gas filling as well
as the gas pressure and gas exchange rate can be adjusted in a defined manner.
A translation stage moves the sample in relation to the incident laser beam to
measure at different locations on the sample surface.
The use of a measuring chamber is not a necessary precondition for laser-induced
breakdown spectroscopy. For inline analyzing tasks, the measuring chamber is
often set aside and the measurement is performed under atmospheric conditions.
In this case, the usable measuring radiation is limited to wavelength greater than
190 nm, since air absorbs shorter wavelength strongly (cf. Sect. 3.5). For quantitative
measurements with high requirements on measuring precision and uncertainty,
a measuring chamber is used to adjust the ambient gas conditions and the gas
exchange at the interaction region in a defined way.
Fig. 2.6 Setup for laser-induced breakdown spectroscopy; M D mirror, L D focusing lens, W1 D
window of the laser radiation, C D measuring chamber, P D laser-induced plasma, S D sample,
A D translation stage, G D gas fitting, ˛o D observation angle, W2 D window for the measuring
radiation, FO D fiber optics, D D detectors
2.2 Setup for Laser-Induced Breakdown Spectroscopy 13
A control unit triggers the laser and reads the signals of the detectors. To improve
the signal-to-noise ratio the plasma radiation is recorded only during the life time
of the plasma. For this purpose, the spectrally dispersed radiation is detected time
resolved and integrated over a time gate within the life time of the plasma. The
control unit adjusts the position and duration of that time gate (cf. tdelay and tint in
Fig. 2.4).
Figure 2.7 shows schematically the arrangement of components in an analyzing
system based on LIBS. The sample is positioned on a sample table. The laser beam
is focused onto the bottom side of the sample. The sample table can be set into
rotation, whereby the rotational axes and the optical axes of the incoming laser
beam are parallel and shifted laterally. Due to this eccentricity, a relative motion
between sample and laser beam occurs, where the locations of laser irradiation lie
on a circle. By this a spatial averaging is realized (cf. Sect. 3.6). Sample table and
measuring chamber are parts of the so-called sample stand (cf. Sect. 4.6).
Figure 2.8 shows a view of an analyzing system. The sample stand can be seen
on the left side at the top of the system.
Fig. 2.7 Arrangement of components in a LIBS-based analyzing system; T D sample table, S D

sample, ST D sample stand, FO D fiber optics, G D gas supply, D D detector, M D mirror
Fig. 2.8 Analyzing system

based on LIBS. Left: sample
stand
2.3 Measuring Procedure
A measuring procedure comprises the following steps:

1. Definition of a measuring method or retrieval of a measuring method which was
already defined. Putting the sample on the sample stand
2. Start of the measurement
3. Evaluation of the spectral signals
4. Display of the measuring results
A measuring method is defined in particular by the selection of measuring param-
eters and their temporal sequence, cf. Sect. 11.1. Measuring parameters are, e.g.,
the laser pulse energy, the number of prepulses and measuring pulses, and the gas
flow in the measuring chamber. These various parameters influencing the measuring
process will be discussed in more detail in Chap. 3.
A typical measuring sequence for a quantitative analysis of a sample including
data evaluation takes about 30 s–2 min. For an identification testing of work pieces,
inspection times of a few seconds or even fractions of a second are achievable, cf.
Chap. 18.
2.4 Applications and Methodical Extensions
Laser-induced breakdown spectroscopy is a versatile tool to analyze solid, liquid, or

gaseous substances. Figure 2.9 illustrates the different possibilities of applications
and methodical extensions of LIBS. The configuration described so far is shown in
item 1. The laser beam is focused onto a solid sample. By translation of the sample in
a direction perpendicular to the optical axis of the irradiated laser beam the sample
surface is scanned to obtain a spatially resolved information about the chemical
composition. By this approach maps of element distributions can be gained, cf. Sect.
18.3. The laser beam can be guided via fiber optics to a measuring head and then
be focused to analyze a liquid (item 2 in Fig. 2.9). Analysis of, e.g., liquid steel
by LIBS will be described in more detail in Sect. 13.1.3. The laser beam can also
be focused through a window into a closed tube to analyze aerosoles or gases, see
item 3 (cf. Sect. 14.6.). A methodical extension is the combination of LIBS with
separation techniques such as capillary electrophoresis (CE) or high-pressure liquid
chromatography (HPLC), see item 4. In this case, a molecule-specific separation is
followed by an element-specific analysis to improve speciation analytics of smallest
amounts of complex samples, cf. Sect. 14.5. LIBS can be combined with other laser
spectroscopic methods such as laser-induced fluorescence (LIF), see item 5. In a
first step, a pulsed laser having a fixed wavelength 1 ablates material and induces
a plasma, thus performing LIBS as described in the previous sections. After the
plasma plume has evolved, a second laser with a tunable wavelength 2 irradiates
the material cloud inducing element-specific transitions to generate fluorescence
References 15
1 2 3
measuring
head aerosoles,
gas flow
solid sample
scanning liquid
sample
4 5 6
CE, HPLC
λ1
droplets
λ2
ablation analysis
Fig. 2.9 Applications and methodical extensions of LIBS: 1 D element mapping of sample
surfaces by scanning, 2 D analysis of liquids, 3 D analysis of aerosoles and gases excited through
an optical window, 4 D combination of molecule-specific separation techniques with LIBS (CE D
capillary electrophoresis, HPLC D high-pressure liquid chromatography), 5 D LIBS combined
with laser-induced fluorescence (LIF), 6 D laser ablation of surface layers and subsequent analysis
of the bulk material
signals to detect selected species with high sensitivity, see Chap. 12. Finally, the
laser is a multipurpose tool allowing to locally ablate surface layers and to perform
chemical analysis of the underlying bulk material, see item 6 and Sect. 13.1.2.
References
2.1. A. Pakhomov, M. Thompson, D. Gregory, Laser-induced phase explosions in lead, tin and
other elements: microsecond regime and UV-emission. J. Phys. D Appl. Phys. 36, 2067–2075
(2003)
2.2. A. Zaidel, V. Prokofev, S. Raiskii, V. Slavnyi, E. Shreider, Tables of Spectral Lines
(IFI/Plenum, New York, 1970)
2.3. http://cfa-www.harvard.edu/amdata/ampdata/kurucz23/
Chapter 3
Process Parameters
This chapter describes the main process parameters and their influence on LIBS.
The effect on the plasma state and the emission spectrum will be described in more
detail in Chaps. 4, 6–8.
3.1 Laser Pulse Energy and Repetition Rate
The energy of the laser pulse constitutes the main part of the total energy avail-
able for evaporation, plasma formation, and generation of spontaneous emission.
Exothermic chemical reactions of the evaporated material with the ambient gas or
within the ambient gas may provide additional process energy.
Typical laser pulse energies lie in the range from 1 to 1,000 mJ depending on the
specific requirements. As a rule pulse energies below 50 mJ are used for small focal
length of the focusing optics (<100 mm), high spatial lateral resolution (<20 m),
high measuring frequencies (>100 Hz), or if the laser pulses are transmitted via fiber
optics. Pulse energies greater than 50 mJ are used for focal lengths ranging from
200 mm to several meters, low spatial lateral resolution (>200 m), and measuring
frequencies in the range of typically 10–100 Hz.
The repetition rate of the laser pulses defines the maximum measuring frequency
for LIBS. For flashlamp-pumped Nd:YAG lasers with electro-optical Q-switching,
this rate amounts to 10–100 Hz yielding pulses with a temporal width of typically
5–20 ns. Continuously pumped solid state lasers with acousto-optical Q-switching
achieve repetition rates beyond 10 kHz; however, the pulse energies are significantly
smaller and the pulse width is greater (>100 ns). Diode-pumped solid state lasers
with electro-optic Q-switching run with repetition rates of up to 2 kHz and beyond,
delivering pulses of about 1 mJ with pulse widths <10 ns. The average optical power
is typically in the range between 1 and 50 W.
Figure 3.1 shows the parameter range of repetition rates and laser pulse energies
used for LIBS to analyze metals, steel alloys, slags, and oxidic materials. The
laser parameters given in most of the scientific publications can be grouped around

18 3 Process Parameters
Fig. 3.1 Repetition rate versus laser pulse energy of Q-switched laser sources used for LIBS to
analyze metallic species, aluminum, steel, slag, and oxides (the numbers close to the data points in
the graph refer to the following references: 1 D [3.1], 2 D [3.2], 3 D [3.3], 4 D [3.4], 5 D [3.5],
6 D [3.6], 7 D [3.7], 8 D [3.8], 9 D [3.9], 10 D [3.10], 11 D [3.11], 12 D [3.12], 13 D [3.13],
14 D [3.14], 15 D see Chap. 15, 16 D see Sect. 18.4, 17 D see Sect. 17.4, 18 D see Sect. 16.1.
L1–L4 D limits, see text)
repetition rates of 20 Hz and pulse energies of 200 mJ. The dominant field of
application in this parameter range is bulk analysis. The diagonals plotted in Fig. 3.1
depict the average laser power. For inline applications of LIBS to directly analyze or
identify the chemical composition of a substance in a process line, typical average
laser powers are 5–10 W, see data points 1, 3, 12, and 13 in Fig. 3.1. This is a
consequence of the respective requirements: (a) fast measurement, (b) sufficiently
large working distance (distance between the last optical element of the measuring
system and the laser-induced plasma) in the range from 150 to 1,800 mm to assure
a flexible access to the measuring object.
High repetition rates (point 15 in Fig. 3.1) are advantageous for spatially resolved
analysis to scan macroscopic samples within a short time. In this case, the typical
requirements are: sample surface to be scanned 50 mm 100 mm, spatial lateral
resolution <20 m. The spatially resolved analysis will be discussed in detail in
Chap. 15.
Figure 3.1 shows as dashed lines various limits denoted with L1 to L4. The limit
L1 describes an upper value of the repetition rate of the laser given by restrictions
of the gas exchange in the interaction volume, see relation (3.1). L2 denotes a
limit for the repetition rate of the laser for single spectra evaluation, i.e., each
spectrum generated by each laser pulse is processed separately. This estimated
limit is given by the detector response and the processing speed of the subsequent
3.1 Laser Pulse Energy and Repetition Rate 19
12 300
10 250
pulse width [ns] – – – –

8 200
pulse energy [mJ]
6 150
4 100
2 50
0 0
0 2 4 6 8 10 12
Q-switch rate [kHz]
Fig. 3.2 Pulse energy and pulse width as a function of the Q-switch rate for a continuously
pumped Nd:YAG laser [3.15]
signal electronics. Maximum achieved repetition rates with single spectra evaluation
amount to 1 kHz, cf. Chap. 15. The limit L3 presents a lower limit for the laser pulse
energy of less than 100 J, where the intensity of the plasma emission becomes so
weak that noise levels of the detectors are approached. The limit L4 is given on the
one hand by pulse energies of about 1 J and on the other hand by an average laser
power of 100 W, both involving bulky Q-switched laser systems, which are at the
limit of technical feasibility and commercial availability.
Generally, a change in the repetition rate of a given laser source changes also
the beam parameters pulse energy, pulse width, and beam profile. As an example,
Fig. 3.2 shows energy and width of the pulses (FWHM) generated by a continuously
pumped Nd:YAG laser as a function of the repetition rate. With increasing repetition
rate, the level of inversion and hence the achievable amplification of the laser
medium between two consecutive pulses decreases with the consequence of a
reduced pulse energy and enlarged pulse width. Due to this behavior, the laser
power within a pulse is reduced from 66 kW for repetition rates <1 kHz to 14 kW
for repetition rates of 12 kHz. pulse energy and pulse width of the laser source
itself cannot be chosen independently, as they are linked with each other due to
the properties of the laser medium and the pumping source.
Nd:YAG lasers pumped by pulsed diode lasers (DPSSL D diode-pumped solid
state laser) can generate pulse energies of 2 mJ at 2 kHz repetition rate with electro-
optic Q-switching. The pulse width amounts to about 5 ns. Although the pulse
energy is only one-fifth of the one delivered by the continuously pumped Nd:YAG
6
RSD(EL) 180
5 RSD(τL)
EL 160
4 140
RSD [%]
EL [mJ]
3 120
2 100
80
1
single pulses, Uosc = 545V, Upc1 = 2500V
60
0
180 200 220 240 260 280
tfq [µs]
Fig. 3.3 Relative standard deviation RSD of the pulse energy EL and pulse width L as a function
of the delay time between flashlamp and Pockels cell triggering tfq . On the right scale, the
corresponding pulse energy is shown. Uosc D charging voltage of the capacitors driving the
flashlamp discharge, Upc1 D bias voltage at Pockels cell
laser from Fig. 3.2 at 2 kHz, the pulse power is greater by a factor of 6, due to the
reduced pulse width being shorter by a factor of 30.
The generation of plasma for LIBS requires a sufficient high irradiance
(unit W=cm2 / at the surface of the material to be analyzed. Typical values are
>107 W=cm2 . An estimation of the necessary irradiances will be given in Chap. 5.
The irradiance depends on the pulse energy, the pulse width, the beam diameter,
and the beam quality at the location of interaction with the sample, see Sect. 3.4.
Figure 3.3 illustrates the relative standard deviation (RSD) of the pulse energy
and pulse width of a flashlamp-pumped Nd:YAG laser as a function of the time
between the ignition of the flashlamp and the Q-switching tfq (cf. Sect. 4.1). There
is an optimum time tfq , where the highest pulse energies are achieved and at the
same time the RSD of the pulse energy describing the pulse-to-pulse fluctuations
of the pulse energy attains a minimum at about 2%. The RSD of the pulse width
amounts to about 3% at tfq D 220 s and increases strongly for greater tfq . These
features of a pulsed pumped laser demonstrate that a change in the laser operating
parameters – here shown for tfq – has an effect not only on the level of pulse energy
generated but also on the stability of the laser beam parameters.
Plasmas generated by two subsequent laser pulses at one sample location can
be considered as independent events, if the following conditions are fulfilled: (a)
the state of the ambient gas prior to the irradiance of the second pulse is identical
to the one persisting prior to the first pulse, (b) the surface of the sample to be
analyzed is not modified in a way, which influences the interaction of the laser
radiation with the material. The first condition is fulfilled, if the material ablated
3.1 Laser Pulse Energy and Repetition Rate 21
Fig. 3.4 The ablated material generated by a previous laser pulse has to be removed from the
interaction region prior to the irradiance of the next laser pulse, see text
by the laser pulse irradiation and/or recondensated material is transported off the
interaction volume by diffusion or gas flow. Furthermore, the ambient gas flows
back into this interaction volume and establishes a stationary and homogeneous gas
atmosphere prior to the arrival of the next pulse.
The time for a complete gas exchange in the interaction volume is denoted as
local gas exchange time ge . Figure 3.4 illustrates the situation. As an example we
examine the interaction of a collinear double pulse (cf. Chap. 6) with a steel sample.
For a burst energy of 2 140 mJ, the ablated mass of iron amounts to 300 ng, which
corresponds to 3:22 1015 iron atoms. After the plasma ignition, the material cloud
expands and attains a volume approximated here by a sphere having a diameter of
4 mm. The average iron species density is then 2 1017 cm3 . This number density
is by about two orders of magnitude smaller than the ambient gas density at 1 bar
of 2:5 1019 cm3 . Hence, there is a chance to push this material cloud away from
the interaction zone by a flow of the ambient gas. A practical limit will be the sound
velocity, which is for argon at room temperature 300 m/s. Thus, the condition (a)
described above can be expressed as follows:
2dmc
trep > ge D ; (3.1)
vsound
where trep D 1=rep is the time between two subsequent laser pulses, rep is the
laser pulse repetition rate, ge is the local gas exchange time, dmc is the diameter of
material cloud, and vsound is the velocity of sound of the ambient gas.
For (3.1), it is assumed that the material cloud should be shifted by twice its
diameter to assure a sufficient lateral translation. With dmc D 4 mm (cf. Fig. 3.4),
we obtain ge D 26 s and the limit for the repetition frequency is rep < 38 kHz.
The highest repetition frequency realized so far for LIBS amounts to 1 kHz, cf.
Chap. 15. This is still about a factor of 40 below the principal limit estimated using
relation (3.1).
To which extent the second condition (b) holds depends on the selected param-
eters: pulse energy, temporal structure of the laser pulse, focusing, ambient gas, as
well as on the properties of the sample to be analyzed. Qualitatively the situation
can be described as follows. The laser pulse causes a material ablation, i.e., the
local topology of the sample surface is modified. Depending on the selected
parameters and the studied sample material, these modifications can be described
phenomenologically as formation of a crater. Furthermore, the structural, physical,

and chemical state of the sample in this crater may have changed due to a series
of processes such as melting and resolidification of the material, ablation of oxidic
surface layers, chemical reactions of the sample species with constituents of the
ambient atmosphere and others. Hence, the subsequent laser pulse will encounter a
different local sample state, which in turn has an effect on the interaction process
and the state and dynamics of the generated plasma (cf. Chap. 8).
3.2 Temporal Characteristics of the Laser Pulse
A key parameter of the temporal characteristics of a laser pulse is the pulse width.
Figure 3.5 illustrates the main difference in the interaction of nanosecond and
femtosecond pulses with a sample for LIBS. For nanosecond pulses, the plasma
is generated while the laser irradiation still persists (cf. phase 2 in the first row of
Fig. 3.5). Hence, there is the possibility of an additional interaction between the
laser beam and the plasma, i.e., the plasma can be heated by the laser radiation. For
femtosecond pulses, the energy is deposited in the target within a very short time
before the material ablation and plasma formation starts. The laser beam plasma
interaction does not exist in this case (cf. phase 2 in the second row of Fig. 3.5).
A broad range of pulse widths was studied for laser ablation and LIBS varying
over nearly seven orders of magnitude, see Fig. 3.6. Looking at this diagram, it
is obvious that there is a cluster of data points at pulse widths in the nanosecond
range. In most cases, nanosecond pulses are generated by Q-switching of solid
state lasers. The pulse width depends on the Q-switching of the laser source and
the amplification of the laser medium. For solid state lasers with electro-optic
Q-switching, the typical pulse widths amount to 10–20 ns. Depending on the pulse
energy, this results in powers of the laser pulses ranging from 50 kW to 25 MW.
Fig. 3.5 Schematic illustration of the interaction of nanosecond (ns, top row) and femtosecond
laser pulses (fs, bottom row) with a sample for LIBS, for abbreviations see Fig. 2.1
3.2 Temporal Characteristics of the Laser Pulse 23
33
100n 266 nm SP 19 11
308 nm SP 32 12
9
532 nm SP 4 3
775 nm SP 16
10n 24
22 2
25
21
780 nm SP
800 nm SP 1
5 7 20
1054 nm SP 10 6
1n 1064 nm SP
pulse width [s]
23 8
800 nm DP 15
1064 nm DP 14
17, 27
100p
36
10p 29
1p 31
34
35 26
100f 30
18
28
10f
1μ 10μ 100μ 1m 10m 100m 1
laser burst energy [J]
Fig. 3.6 Pulse width versus laser burst energy studied for laser ablation and LIBS. SP D single
pulses, DP D double pulses (cf. Chap. 6). The numbers refer to the references as follows: 1 D
[3.1], 2 D [3.2], 3 D [3.3], 4–6 D [3.5–3.7], 7–9 D [3.8–3.10], 10–12 D [3.12–3.14], 13–36 D
[3.16–3.39]
The pulse duration influences the interaction of the laser beam with the measur-
ing object. For solid targets, the diameter of the crater is reduced for shorter pulse
durations, cf. Sect. 3.6, Fig. 3.18.
Figure 3.7 shows as an example the temporal shape of a laser pulse generated
by a Q-switched Nd:YAG laser [3.36]. The pulse width influences the relative
significance of the various interaction processes with the target material. For pulse
durations in the sub-nanosecond regime, i.e., picoseconds and femtoseconds, the
laser energy is dominantly coupled into the material causing a fast transition from
the solid state to the vapor phase prior to the formation of the plasma. By this
well-defined ablation geometries can be produced. However, the energy content in
the plasma is small which in turn results in a weaker and shorter line emission
[3.30]. For pulse durations beyond 50 ns and metallic targets, the melt phase
becomes increasingly important, i.e., the ratio between particles transported off the
interaction region in the melt phase in relation to those transported off as vapor
increases. For pulse durations between 1 and 50 ns, the laser energy partitioning for
material evaporation and energy coupling into the radiating plasma seem to be in
a favorable range for LIBS in terms of the total number of analytes excited into a
state, where a significant line emission persists for a few microseconds.
During a single flashlamp discharge cycle, which pumps the laser rod, multiple
laser pulses can be generated with a step-by-step control of the Pockels cell [3.36].
An example for a double pulse is shown in Fig. 3.8. A group of pulses generated
during a single flashlamp discharge is called a burst. The selected voltage levels for
Fig. 3.7 Temporal shape of a

laser pulse of a
flashlamp-pumped
Q-switched Nd:YAG laser
U [a. u.]
15 ns
t [ns]
250
double pulse
200 2 × 40 mJ, Δt = 6 μs
150
U [mV]
100
50 1st pulse 2nd pulse
0
-20 0 20 40 60 5980 6000 6020 6040 6060 6080
t [ns]
Fig. 3.8 Double pulses of a flashlamp-pumped Nd:YAG laser
the switching steps of the Pockels cell determine the energy of the respective pulses
in the burst. In this way different pulse energy ratios within a burst can be adjusted.
The temporal distance between two pulses within a burst is called interpulse
separation and denoted with t. The interpulse separations are set by the trigger
delays for the respective switching steps. The minimum interpulse separation t
amounts to 2 s with this approach. The maximum duration of a burst amounts to
100 s and is limited by the variation of the amplification as a function of time of
the flashlamp-pumped laser. The bursts are generated with the repetition frequency
of the flashlamp of, e.g., 10 Hz.
Figure 3.9 illustrates the different pulse modes: single pulses (SP), double pulses
(DP), and triple pulses (TP) and the describing parameters interpulse separation
t, cycle duration Trep , pulse energy Ej , pulse duration Lj [3.40]. Generating DP
3.3 Laser Wavelength 25
Fig. 3.9 Illustration of different pulse modes: single pulses (SP), double pulses (DP), and triple
pulses (TP) and the describing parameters
or TP within a single flashlamp discharge implies that the duration of the pulses
within a burst changes such that Lj .TP/>Lj .DP/>L .SP/. An alternative for the
generation of pulse bursts is the superposition of the laser beams of two laser
systems by, e.g., a polarizing beamPcombiner. The sum of pulse energies within
a burst is called burst energy Eb D Ej .
j
Typical t values are in the range of a few microseconds. A pulse structure with
time scales in the microsecond range has an effect on the transport processes in the
interaction region as well as on the laser-induced plasma, whose characteristic time
scale is of the same order of magnitude, cf. Sect. 8.5. Double or multiple pulses may
generate plasma states with more intense and longer lasting line emission which is
advantageous for LIBS. The effect of double and multiple pulses will be discussed
in more detail in Chaps. 6–8.
3.3 Laser Wavelength
The absorption properties of a substance depend on the wavelength of the laser

radiation. This natural absorption determines the portion of the laser pulse energy,
which is coupled into the material, especially in the phases (1) and (2) as shown in
Fig. 2.1 just before the initial state of aggregation of the substance to be analyzed
Fig. 3.10 Parameters of the interaction of laser radiation with a solid sample, LB D laser beam,
S D sample, ıopt D optical penetration depth (OPD), ıh D heat penetration depth (HPD) of laser
radiation in a solid sample (the drawing is not scaled), 2wb D laser beam diameter at the sample
surface
changes. The generation of material vapor and plasma modifies this absorption,
which is then dominantly influenced by the interaction of the laser radiation with
the material vapor, the plasma, and the particulates.
Furthermore, the surface properties of the sample, e.g., the roughness, influence
the absorption. For a specular reflecting surface, the absorptivity decreases from
0.64 at 200 nm to 0.36 at 1 m [3.41].
Figure 3.10 illustrates the optical penetration depth and the heat penetration
depth of laser radiation for a solid sample. The optical penetration depth ıopt is the
reciprocal of the absorption coefficient and describes that length along which the
incoming laser irradiance decreases to 37% of its initial value. The heat penetration
depth ıh is a measure of the propagation length of the thermal energy into the sample
after start of laser irradiation. The spatial and temporal run of the temperature inside
the sample depends on the temporal structure of the laser irradiance, as well as on the
laser beam profile and hence the geometry of the heat source. The heat penetration
depth as a function of time can be estimated as follows [3.42]:
p
ıh D 2 t; (3.2)
where is the thermal diffusivity, Œ D cm2 =s.

The heating of the sample can be described by the heat conduction equation,
which will be considered in more detail in Chap. 5.
Figure 3.11 shows ıopt and ıh as a function of the wavelength between 200 nm
and 1:2 m for aluminum and iron. The values for ıopt are calculated from tabulated
data of the complex index of refraction [3.43, 3.44]. For a first calculation of ıh ,
we have taken t D 10 ns, which is the typical order of magnitude of the full width
at half maximum of a Q-switched laser pulse. Thus, the values for ıh represent an
upper limit since only a part of the laser pulse energy is absorbed by the sample
via natural absorption. This is due to the induced plasma, whose emission can be
observed already after time delays of <1 ns after start of irradiation of the laser
pulse. On the other hand, the laser-induced plasma itself interacts with the sample
3.4 Beam Quality and Focusing 27
Fig. 3.11 Optical penetration

depth ıopt and heat
penetration depth ıh as a
function of the wavelength
for aluminum and iron
surface and acts as a local heat source transferring energy to the sample by radiation
and heat conduction. This secondary heat source is characterized by the size of
the plasma of typically a few millimeters, its life time being of the order of some
microseconds and energy density as a result of the temperature inside the plasma,
which are in the range of several thousand kelvin.
The absorption of polymer materials in the near infrared is determined by elec-
tronic as well as vibrational excitations. For Nd:YAG laser radiation at a wavelength
of 1:06 m, the optical penetration depth was determined experimentally [3.45].
For polyamide (PA 6), the optical penetration depths amounts to about 3.5 mm
in a temperature range of the polymer between 20 and 220ıC. For polycarbonate
(PC), ıopt is about 0.25 mm in this temperature range. These values are by orders of
magnitude greater than those for metals.
The absorption of laser radiation by the plasma as a function of the laser
wavelength will be discussed in more detail in Sect. 5.3.
3.4 Beam Quality and Focusing
The laser radiation is focused to increase the irradiance at the location of interaction
to values greater than 107 W=cm2 . Each substance exposed to these irradiances will
be transferred to the plasma state within a few nanoseconds.
The focusability of laser radiation depends on the beam quality and the wave-
length. In the ideal case of a so-called Gaussian beam, the laser beam can be focused
by an aberration free lens to the following beam waist radius:
2f
w0 D ; (3.3)
D
where w0 is the radius of beam waist, f is the focal length of the lens, is the
wavelength of the laser radiation, and D is the diameter of the illuminated aperture
of the lens.
At the beam waist radius w0 , the irradiance attains a level which is by a factor of
1=e 2 . 13:5%/ smaller than the one at the beam axis. Due to physical and technical
reasons, the diameter D of the illuminated aperture of a lens cannot be significantly
larger than the focal length f ; hence, the smallest possible beam waist diameter is
of the order of the used wavelength .
For a Gaussian beam the quantity:
w0 0 1
D ; (3.4)

where 0 is the beam divergence angle of a Gaussian beam, is a propagation invariant

insofar as the optics involved do not change the stigmatic character of the beam.
According to (3.4), the divergence angle increases for smaller beam waists.
The Gaussian beam is an idealization of a laser beam with the smallest possible
beam divergence. A real laser beam deviates from this ideal case, which is described
by the standardized beam propagation ratio M 2 [3.46]:
d‚
M2 D ; (3.5)
4
where M 2 is the beam propagation ratio, d is the diameter of laser beam, and ‚ is
the beam divergence angle (full angle) of the laser beam.
The smallest possible value of M 2 amounts to 1. For each real laser beam holds
2
M >1.
For a typical flashlamp-pumped Nd:YAG solid state laser follows with d D
10:8 mm and ‚ D 0:35 mrad: M 2 D 2:8. Figure 3.12 shows the experimentally
determined beam radius of a Nd:YAG laser beam as a function of the distance from
the focal plane, i.e., the location of the beam waist. The quantity zR describes the
Rayleigh length and p defines the distance from the focal plane at which the beam
radius increases to 2w0 .
For the average irradiance (average within a single laser pulse) at the location of
the beam waist holds:
EL EL D 2
If D 2
D ; (3.6)
L w0 4L f 2 2 M 4
where EL is the energy of the laser pulse, D is the diameter of the illuminated
aperture of the focusing lens, L is the temporal width (FWHM) of the laser pulse,
f is the focal length of the lens; is the wavelength of the laser radiation, and M 2
is the beam propagation ratio.
3.5 Ambient Gas Atmosphere 29
180
single pulses
w0 = (35 ± 4) μm
140
zR = (1.3 ± 0.2) mm
beam radius rb [μm]
100
60
20
-5 -3 -1 1 3 5
distance to focal plane Δs [mm]
Fig. 3.12 Beam radius of single pulses of a flashlamp-pumped Nd:YAG laser as a function of the
distance from the focal plane (pulse energy Eb D 80 mJ, focal length f D 200 mm)
For typical values (EL D 80 mJ, D D 10:8 mm, L D 20 ns, f D 200 mm,
D 1;064 nm, M 2 D 2:8), relation (3.6) yields an average irradiance of If D 1011
W=cm2 .
The diameter of the beam waist influences the achievable spatial resolution for
LIBS. In Fig. 3.12, this diameter amounts to 70 m. Due to the interaction of the
laser beam with a solid target, a crater is generated having a diameter greater than
the beam waist caused by heat conduction and material transport processes.
For a solid or liquid measuring object, the quantity s describes the distance
between the beam waist location and the surface of the target, cf. Fig. 2.2. A beam
waist lying at the surface corresponds to s D 0. For s>0, the beam waist is
within the sample and for s<0 the beam waist lies between the focusing optics
and the sample surface. In many cases, s>0 is chosen to avoid a gas breakdown in
front of the sample surface, which would lead to a plasma formation in the ambient
atmosphere and prevents an efficient ablation of the material to be investigated.
3.5 Ambient Gas Atmosphere
The ambient gas atmosphere influences the properties and the dynamics of the laser-
induced plasma, as well as the transmission of the generated line radiation along the
beam path to the spectrometer. Typical ambient gases applied for LIBS to analyze
solid or liquid substances are argon, nitrogen, and atmospheric air. Table 3.1 shows
physical data for these gases and helium.
Studies of the mass ablation efficiency show that the material ablated per laser
pulse (or laser burst) increases with rising ambient gas density at a pressure of
Table 3.1 Physical data of typical ambient gases used for LIBS
Gas Molar Gas Ionization Thermal Dynamic
mass density energy conductivity viscosity
(g) (kg=m3 ) (eV) (W/(mK)) (103 Ns=m2 )
Ar 39.95 1.784 15.76 0.0178 0.022
O2 32 1.429 12.06 0.0266 0.019
N2 28 1.25 15.58 0.026 0.018
He 4 0.179 24.59 0.15 0.022
The gas density refers to a pressure of 1 bar and a temperature of 293 K [3.41]
0.00
-0.02
UPMT [V]
-0.04 Fe 273.07 nm, argon
-0.06
0.0 0.2 0.4 0.6 0.8 1.0
t [μs]
0.00
-0.02
UPMT [V]
-0.04
Fe 273.07 nm, air
-0.06
0.0 0.2 0.4 0.6 0.8 1.0
t [μs]
Fig. 3.13 Photomultiplier (PMT) signals of the Fe II line emission at 273.07 nm as a function of
time, top: PMT signal of a laser-induced plasma generated in an argon atmosphere, bottom: for an
air atmosphere
1 bar (cf. Chap. 7) [3.47]. In the case of negligible plasma shielding, the plasma–
sample interaction is greater in an ambient gas of higher density because the plasma
expansion is restricted by the surrounding mass density which results in higher
density and temperature of the plasma in the vicinity of the sample surface leading
to enhanced material ablation.
Figure 3.13 shows photomultiplier (PMT) signals as a function of time detected
at a wavelength of 273.07 nm corresponding to a singly ionized iron line (Fe II)
generated by focusing a 1-mJ laser pulse onto an iron sample under two different
3.5 Ambient Gas Atmosphere 31
atmospheres: argon and air (the experimental setup is shown in Sect. 15.1). In air
the short living continuum peak (cf. Sect. 2.1) dominates the signal course and only
a small signal level of the element-specific emission is detected (for times >70 ns in
Fig. 3.13, bottom), whereas the situation in argon is completely different. A distinct
second maximum appears followed by a decay lasting several 100 ns. This second
maximum is attributed to the line emission, which is strongly enhanced for argon
in comparison with air. The line emission increases due to enhanced ablation and
higher plasma temperatures (see previous paragraph).
For a given ambient gas, e.g., air, the ablation rate increases for reduced ambient
pressures [3.48]. A reduction of the pressure from 1 bar to 0.1 mbar increases the
ablation rate by a factor of 5. This behavior is explained by the reduced plasma
shielding for lower ambient pressures and the reduced recondensation of the ablated
material in the interaction region due to an increased plasma volume whose center
is shifted to larger distances from the target surface. Further details are described in
Chap. 7.
Besides the mass effect of the ambient gas, the electronic configurations of
the gas species influence the absorption features for electromagnetic radiation.
Figure 3.14 shows the mass absorption coefficient of the gases oxygen, water vapor,
and ozone in the wavelength region from 100 nm to 30 m [3.49]. In the ultraviolet
region below 400 nm, the absorption of oxygen and water vapor increases strongly.
106
H2O
H2O
105
O2
104 O3
mass absorption coefficient [cm 2/g]
O3
103 O3
O2
O3
102 H2O
101 O3
100
10-1
O2 O2
10-2
10-3
10-4
102 103 104
wavelength [nm]
Fig. 3.14 Mass absorption coefficient of oxygen, ozone, and water vapor as a function of the
wavelength
30
vacuum
20
10
0 1.0
0.1 % O2
0.8
intensity [a.u]
0.6
0.4
0.2
1.0 0.0
C 193.10 nm
P 178.28 nm
S 180.73 nm
0.2 % O2
Fe 187.75 nm
0.8
0.6
0.4
0.2
0.0
178 180 182 184 186 188 190 192 194
wavelength [nm]
Fig. 3.15 Simulated emission spectra in the VUV range for a steel sample with emission lines
of carbon, phosphorus, and sulfur. Top: no absorption in the ambient atmosphere, middle/bottom:
0.1% and 0.2% oxygen content in an inert gas causing absorption of VUV emission lines
Using emission lines of the laser-induced plasma in this spectral range requires
vacuum conditions or an inert gas such as argon of high purity for a sufficient
high transmission between the source of line emission and the spectrometer. The
detection of line emissions in the vacuum ultraviolet (VUV) region is especially
of interest to analyze light elements in metallic matrices such as carbon, phos-
phorus, and sulfur [3.24]. This spectral range offers low background from the
Bremsstrahlung continuum and less interferences by other emission lines.
Figure 3.15 illustrates the strong effect of small oxygen content on the absorption
of VUV emission lines. The spectra shown are simulated plasma emission spectra
(cf. Sect. 10.3) for a steel sample with emission lines of carbon, phosphorus, and
sulfur in the spectral range from 178 to 194 nm. The spectrum shown at the top is
calculated without any absorption in the ambient gas (“vacuum”). For the other two
spectra, an inert gas atmosphere of 1 bar at 293 K is assumed with an oxygen fraction
of 0.1% and 0.2%, respectively, of the total particle density of the ambient gas. The
emitted radiation propagates along a distance of 180 mm through this ambient gas
before it enters the vacuum spectrometer. For the carbon line at 193.10 nm (see
dashed vertical line on the right side in Fig. 3.15) the peak intensity is reduced by
a factor of 5 from the top to the middle spectrum (note the different scales at the
y-axes of the three spectra), whereas this factor is more than 20 for the Fe 187.75-nm
line, 28 for S 180.73-nm line, and >40 for the P 178.28-nm line.
3.6 Spatial Resolution and Spatial Averaging 33
3.0×105
2.5×105
Argon
2.0×105
signal [cts]
1.5×105
1.0×105
air
5.0×104
0.0
165 170 175 180 185 190 195 200
wavelength [nm]
Fig. 3.16 Emission spectrum of a deuterium lamp for wavelengths below 200 nm transmitted
through a thin gas layer flushed with argon or air
Figure 3.16 illustrates the absorption properties of air looking at the emission
spectrum of a deuterium lamp in the near UV region, where the short distance
between the lamp and the spectrometer is flushed with argon or with air.
3.6 Spatial Resolution and Spatial Averaging
For LIBS, the laser beam is focused to induce a plasma. This intrinsically allows for
a high lateral spatial resolution. The point of analysis can be defined by the location
of the laser focus, i.e., the position of the waist of the converging laser beam. Two
kinds of spatial resolution have to be distinguished: the lateral resolution and the
longitudinal or depth resolution. Lateral and longitudinal resolutions are primarily
influenced by the laser beam parameters: focus diameter and Rayleigh length. Both
parameters cannot be selected independently but are linked to each other by the well-
known relation given for a Gaussian beam: zR D w20 =, where zR is the Rayleigh
length, w0 is the waist radius of the beam, and is the wavelength. In the following
paragraphs, we will first focus the discussion on the lateral spatial resolution.
Figure 3.17 illustrates the lateral spatial resolution of LIBS compared with
conventional spark optical emission spectrometry, where an electric spark is ignited
between an electrode and a metallic sample surface. While the spark jumps to any
position within a circle of about 5 mm diameter depending on the respective local
peaks of the electric field on the surface, the laser beam irradiates a well-defined
Fig. 3.17 Comparison of the spatial resolution of spark optical emission spectrometry (left) and
LIBS (right). E D electrode, SP D spark, S D sample, LB D laser beam, P D plasma
position. For a series of several hundreds of sparks – this is the case for the
photograph shown on the left side of Fig. 3.17 – the interrogated surface is a circle
with a diameter of 5 mm. Hence, the individual location of each spark is uncertain
within this 5 mm.
For the laser beam, the lateral extension of the interaction region is significantly
smaller. In the example shown in Fig. 3.17, the central part of the crater has a
diameter of less than 100 m. Of course this diameter depends among others on the
focusing conditions, cf. Sect. 3.4. With a tightly focused laser beam of high beam
quality, much smaller crater diameters can be achieved. High spatial resolution is
especially of interest for scanning microanalysis with LIBS, see Chap. 15.
The achievable lateral spatial resolution with LIBS is determined by the size of
the produced laser craters within the material. The crater diameter depends not only
on the laser focus diameter but also on the heat affected zone and therewith on the
thermal diffusivity and on the laser pulse duration, cf. Sect. 3.3. The dependence of
the crater radius produced by the laser pulse on the radius of the beam waist and the
heat penetration depth is described in a simplified approach for constant laser pulse
energy and beam profile:
rc / wb C ıh ; (3.7)
where rc is the radius of the produced laser crater, wb is the radius of the beam waist
on the surface of the specimen, and ıh is the heat penetration depth, see (3.2).
A shorter laser pulse duration will lead to a decreased heat penetration depth and
hence to a smaller crater radius.
Smaller laser craters can also be achieved by further reducing the applied laser
pulse energy while keeping the beam waist radius constant. However, this will lead
to smaller irradiances within the focus and therefore elements with high excitation
energies could not be excited sufficiently with the consequence of lower sensitivity
and higher detection limits.
Fig. 3.18 Microscopic photographs of craters produced in a steel sample by single laser pulses
generated with DPSSL systems. Left: nanosecond pulse with an irradiance of 1:2 TW=cm2 , right:
picosecond pulse, 25 TW=cm2
Figure 3.18 shows two photographs of laser craters produced by single laser
pulses of different pulse durations. The nanosecond pulse is produced by the DPSSL
described in Sect. 4.1 (see Table 4.3; Figs. 4.2 and 4.3). The picosecond pulse is
generated also by a DPSSL. The crater shown on the left side was generated
with a laser pulse of 5.5-ns duration, 2 mJ pulse energy, and a focus irradiance of
1:2 TW=cm2 , whereas the crater shown on the right side was generated with a laser
pulse of 40-ps duration, 0.4 mJ pulse energy, and a focus irradiance of 25 TW=cm2 .
The laser wavelength in both cases is 1,064 nm. The crater diameter is reduced from
16 m diameter for the nanosecond laser to 6 m for the picosecond laser, and the
diameter of the debris area is reduced from 60 to 30 m. Although the irradiance
within the laser focus is approximately 21 times higher for the picosecond laser
used in the right picture, the crater and the debris diameter are smaller than those
for the nanosecond laser in the left picture of Fig. 3.18. Hence, picosecond lasers
offer the potential to further improve the spatial resolution for scanning LIBS. An
even further increase in resolution would be gained with the use of femtosecond
lasers. However, an application of such lasers has to cope with the fact that these
lasers are at present more expensive by a factor of two than picosecond lasers.
While subnanosecond laser pulses allow to reduce the crater diameters, their
efficiency for plasma excitation for LIBS is expected to decrease. A compromise
has to be found to optimize the ablation geometry and the plasma excitation. There
is no reason to assume that the laser pulse durations applied so far for LIBS are
already at this optimum. The temporal pulse structure has a strong influence on
the ablated mass and the efficiency of plasma excitation (cf. Sect. 3.2 and Chap. 7).
Different approaches were studied based on combinations of nanosecond pulses,
which can easily be generated by conventional lasers. In the near future, further
progress in the development of DPSSL systems will open up new perspectives to
generate temporally modulated laser pulses within time scales from picosecond to
nanosecond. This will give the chance to find the optimum tailored pulse structure
for the various LIBS applications.
So far the lateral spatial resolution of LIBS considering the diameter of the laser-
induced crater in a solid target was discussed. The longitudinal spatial resolution or
depth resolution is determined by the depth of the crater.
Depth resolution is a key parameter for depth profiling of coated samples by
LIBS, which was studied by different research groups mostly with the aim to
achieve high depth resolution [3.47, 3.50–3.55]. The experiments were performed
on stationary samples by irradiating a single spot or a microline on the sample
with a series of laser pulses [3.53]. Each pulse ablates an amount of material, thus
penetrating step by step into the sample. From the spectra of the successive laser
pulses, information about the depth profile of the elements was obtained. To achieve
high depth resolution, the ablation depth per laser pulse was kept low (1–100 nm) by
using single pulses and large spot diameters or a line focus on the sample surface.
Figure 3.19 shows an overview of the achieved ablation rates with the corresponding
number of pulses per sample position. For depth profiling of moving coated samples,
a different approach is chosen, where collinear double pulses can penetrate with a
single burst coating thicknesses of up to 10 m and more (see data points with filled
triangles at n D 1 in Fig. 3.19, cf. Chap. 16 [3.56]).
For a representative analysis of an inhomogeneous sample, spatial averaging is
necessary. Spatial averaging with LIBS is possible via two principal approaches:
sequential and simultaneous. For sequential averaging, the laser focus is shifted
laterally relative to the measuring object, by moving the object or by moving of the
10000 9 double pulse
single pulse
ablation rate [nm / pulse]
1000 8
100 1 2
4
3 2
10 5
7
6
1
1 10 100 1000
number of pulses n
Fig. 3.19 Comparison of ablation rates as a function of the number of pulses per sample position
used in LIBS depth profiling experiments. Studies on static samples are plotted as filled squares
(the numbers close to the data points in the graph refer to the following references: 1–9 D [3.47,
3.50–3.55])
a b
Fig. 3.20 Principle of spatial averaging by scanning the laser beam across a sample surface by
rotation of the sample. S D sample, SR D sample rotation, AIL D area irradiated by the laser
beam, AIP D area where interaction of the laser-induced plasma with the sample surface takes
place; Ao , L D see text. (a) Overlapping areas after a few laser pulses and (b) laser track after
several hundred laser pulses
laser beam. In the first simple approach, the laser focus is scanned in a circle across
the surface of the measuring object, where the diameter of that circle is chosen to
be sufficiently large to cover the respective inhomogeneity scale of the object under
study. Figure 3.20 illustrates this approach, where the area irradiated by the laser
beam runs along a circle having a radius R on the surface of the sample. Depending
on the repetition rate of the laser, the radius of the beam focus and the angular
velocity of the sample rotation different degrees of area overlap can be adjusted
described by the following relation:
s
2
Ao 2 L L L
D D arccos 1 ; (3.8)
w2b 2wb wb 2wb
where is the degree of overlap, Ao is the overlapping area, see Fig. 3.20, wb is
the radius of laser beam on the sample surface, and L is the distance between the
centers of two adjacent laser beam positions.
The distance L is approximately given by the relation L !R=rep , where ! is
the angular frequency of the sample rotation, R is the radius shown in Fig. 3.20, and
rep the repetition frequency of the laser.
Figure 3.21 shows a photograph of such a scanned circle on a steel sample with a
series of LIBS craters along the circumference (for the setup see Chap. 2, Fig. 2.7).
Such an approach is especially of interest to perform a reliable bulk analysis of
inhomogeneous samples, e.g., slag samples, cf. Sect. 14.3 [3.57].
Besides moving the sample, the laser beam itself can be translated very fast
using a galvanometric scanner. By this different areas of a sample are irradiated
and moreover tailored crater geometries can be generated, cf. Sect. 13.1.2.
Fig. 3.21 Photograph of a

circular scan of the laser
focus across the surface of a
steel sample illustrating
spatial averaging, cf. Fig. 3.20
Fig. 3.22 Optical microlens

array to generate
simultaneously an array of
plasmas. LB D laser beam,
MA D microlens array, P D
plasma, and S D sample
With a microlens array a simultaneous generation of up to Np D 40 plasmas

is possible spread over an area of about 6 mm diameter, see Fig. 3.22 [3.58]. The
sampling area As is defined by the high-intensity spots of the laser beam at the
sample surface. With 40 spots and a focal length of the microlens array of 37 mm,
the single spot diameter for a laser beam with M 2 D 1 amounts to ds D 58 m.
The sampling area As is therefore: As D 0:25Np ds2 0:11 mm2 . For comparison,
the corresponding values for a single lens (f D 200 mm) with a beam diameter of
6 mm are: ds D 45 m and As D 0:25 ds2 0:0016 mm2 . Therefore, the sampling
area with the microlens array is approximately a factor of 66 higher than for the
single lens. The advantage of the microlens array is that it spatially averages over a
distance of approximately the diameter of the input laser beam.
3.7 Direction of Incidence of the Laser Beam and Direction

of Observation of the Plasma Emission
In many cases, the laser beam axis is oriented perpendicularly to the sample
surface to be analyzed. Figure 3.23 illustrates the more general case of a laser beam
irradiated under an angle of incidence of ˛L . For flat solid samples ˛L D 0ı is
3.8 Warming-up Pulses, Prepulses, and Measuring Pulses 39
Fig. 3.23 Incident laser beam and observation direction relative to the normal of the sample
surface. LB D laser beam, O D observation direction, N D normal to local sample surface,
S D sample, P D plasma, CR D crater, PCR D part of the plasma inside the crater
chosen offering the highest irradiances on the sample surface for a given laser pulse
energy, whereas ˛L ¤ 0ı has to be considered for nonflat measuring objects, e.g.,
scrap pieces in high-speed sorting applications, cf. Sect. 18.4. For a rotationally
symmetric beam, this leads to an elliptic cross section of the irradiance distribution
on the sample surface.
The observation angle ˛O determines the part of the plasma contributing to the
detected spectrum. For ˛O close to 90ı , the plasma is observed in side-on direction
and only those parts of the plasma contribute to the spectrum which are above the
sample surface and therefore not shielded by the generated crater (plasma part PCR
in Fig. 3.23 cannot be observed for ˛O D 90ı /. If only a few laser pulses are applied,
this effect of shielding plays only a minor role. However, in the case of a series of
laser pulses irradiated on the same sample location (in some applications several
hundreds up to thousands of pulses are applied), the shielding of the plasma by
the crater walls becomes more and more important. The realization of different
observation angles in sample stands is described in Sect. 4.6.
For many inline applications where only one direction of access is technically
feasible, ˛O is chosen close to ˛L , i.e., the plasma emission is observed antiparallel
to the incident laser beam, cf. Sects. 13.1.3, 14.1, and 14.6.
3.8 Warming-up Pulses, Prepulses, and Measuring Pulses
If a series of laser pulses are irradiated at one location on a sample to perform a LIBS
measurement, then we have to distinguish at least three groups of pulses. Before
irradiating the sample the laser has to be in a stabile operational mode. Besides a
sufficiently long switch-on time to assure a thermal stationary state of power supply
and laser head, the laser system has to generate a certain number of laser pulses to
achieve a stationary state of the laser medium, the cavity and the cooling system as
well. The number of pulses to achieve this state is called warming-up pulses Nwu
[3.40]. The laser pulses are blocked outside the laser by a shutter or beam dump
and do not irradiate the sample. The warming-up pulses assure a sufficient degree
14
12 ground side
scaled side
10
IC193 / IFe187
8
Npp Nmp
6
0
0 200 400 600 800 1000
pulse number
Fig. 3.24 Line ratio of the carbon line 193.1 nm to the matrix line of iron 187.7 nm as a function
of the number of laser pulses irradiated on a steel sample
of thermal stability of the laser operation with respect to beam profile, pulse energy,
and pointing stability.
The first series of laser pulses irradiating the sample under study is called
prepulses. The number of prepulses Npp causes a local ablation to remove surface
contaminations or surface layers which are not representative for the bulk com-
position of the sample. The spectral emission of the plasmas generated with the
prepulses is not used for the bulk analysis. After the prepulses, a series of measuring
pulses Nmp is irradiated to generate plasmas, whose spectral emission is evaluated
to determine the composition of the sample.
Figure 3.24 shows a series of spectral signals as a function of the pulse number
irradiated onto one location on a scaled steel sample (measuring parameters: double
pulses, Eb D 2 80 mJ, interpulse separation t12 D 30 s, beam waist position
s D 8 mm, tdel D 2 s, tint D 10 s, for further details see Sect. 13.1.2). The scale
layer has a composition which is different from the bulk composition. The spectral
signal shown is the ratio of the carbon line 193.1 nm to the matrix line of iron at
187.7 nm. For the scaled side, the spectral ratio shows a strong decrease for pulse
numbers 0–200, and beyond 500 pulses the ratio converges to a stationary value. If
the scale layer is removed by grinding, then the stationary value is achieved much
faster. Hence, in the case shown in Fig. 3.24, the number of prepulses is chosen
as, e.g., Npp D 500 to assure that the scale layer is penetrated completely and a
representative value for the bulk composition is gained.
3.9 State of the Measuring Object 41
Table 3.2 Examples for moving measuring objects in LIBS setups

Velocity of measuring Configuration Application, remark, references
object
0 mm/s Fixed sample on sample Bulk analysis of homogeneous
stand samples, gas exchange has to
be regarded; cf. Sects. 3.1 and
4.6 [3.24]
10–50 mm/s Sample on linear Scanning microanalysis, cf.
translational stages Chap. 15 [3.27]
12–75 mm/s Sample moved on sample Bulk analysis of
stand by excentric inhomogeneous samples, cf.
mechanism Sect.14.3 [3.57]
Up to 3 m/s Measuring object on Identification of scrap pieces or
moving belt conveyor raw materials, relative wind
forwards efficient gas
exchange [3.59]
3.9 State of the Measuring Object
Besides its elemental composition, the state of the measuring object can be
described by the following categories: (a) state of movement, (b) geometry and
surface properties (e.g., roughness), (c) state of aggregation, and (d) homogeneity.
Since a LIBS measurement can be performed within a few 10 s, there is an
intrinsic potential to measure objects which are moving. Table 3.2 gives an overview
of different states of movement of the measuring object studied with LIBS. In the
predominant number of LIBS studies published, the measuring object is fixed. For
scanning microanalysis, the sample is shifted relative to the laser beam to spatially
resolve element distributions. Circular sample movements are applied to perform a
spatial averaging for inhomogeneous samples, see Sect. 3.6. Highest velocities of
measuring objects were achieved for the identification of scrap pieces transported
on a belt conveyor with up to 3 m/s. For the stationary measuring objects or those
moving at slow velocities <100 mm=s the relative wind is negligible, whereas this is
not the case for measuring objects moving with speeds of several meters per second.
The effect of relative wind becomes obvious in the analysis functions shown
in Fig. 3.25 for static and moving aluminum reference samples, measured with the
setup described in Sect. 18.4 [3.59]. The coefficient of determination is better for
the moving samples than for the static samples, which is attributed to an improved
gas exchange in the interaction region by the relative wind (cf. Sect. 3.1, Fig. 3.4).
The geometry of the measuring object influences the measuring process. In
many cases, flat measuring objects are analyzed and the laser beam is irradiated
perpendicularly to the sample surface, i.e., parallel to the normal of the surface.
However, the general case, especially with respect to industrial applications of
LIBS, is measuring objects where the local surface element is inclined against the
propagation axis of the laser. In the following we consider the situation already
shown in Fig. 3.23, where the laser beam propagation axis encloses an angle ˛L
0.08
static samples r 2 = 0.983
0.06 moving samples

c(Cu) / c(Al) [1]
0.04
0.02
r 2 = 0.998
0.00
0.0 0.4 0.8 1.2 1.6 2.0 2.4
ICu 327.40 nm /IAl 305.71 nm [1]
Fig. 3.25 Analysis functions of static and moving aluminum samples at 3 m/s for copper
Fig. 3.26 Laser beam

irradiating a surface element
where the propagation axis of
the laser encloses an angle of
˛L with the normal of the
surface element. PI D point
of intersection of optical axis
and surface element, FL D
focusing lens, SE D surface
element, PB D projection of
beam profile onto the surface
element, OA D optical axis,
BW D beam waist, N D
normal of surface element,
d D distance between FL and
PI; s D beam waist
position, cf. Fig. 2.2
to the normal of the irradiated surface element. Figure 3.26 shows this situation in a
perspective view. If ˛L ¤ 0ı , the projection of the laser beam profile onto the surface
element has an elliptic shape and the irradiance is reduced in relation to the case
˛L D 0ı . Hence, the orientation of the surface element irradiated influences the
local irradiance and therefore the interaction process with the object. This effect has
to be taken into account especially for samples with nonflat surfaces or measuring
objects with a 3D geometry, e.g., scrap pieces, cf. Sect. 18.4.
LIBS is able to analyze matter in each state of aggregation: solid, liquid, and
gaseous. The relevance of various measuring parameters depends on the state of
aggregation of the measuring object. For example, the distance to the focal plane s
(cf. Figs. 2.2 and 3.11) determines the irradiance at the surface of a solid or liquid
References 43
metal sample causing the ablation and plasma ignition. For a gas, no surface can
be defined and the plasma ignition will start in those regions where the irradiance
exceeds the threshold for ionization and plasma formation.
After laser pulse irradiation at the surface of a free liquid surface or gaseous mea-
suring objects, the matter flows back into the interaction region and after a sufficient
time interval the same initial conditions of the measuring object are achieved. The
laser pulse liquid interaction at a free liquid surface can cause splashes and droplets
ejected from the liquid into the adjacent half space. These droplets may reach the
focusing objective and can cause damage. An approach to prevent such splashes
from reaching the optics is a gas flow to push these particles away [3.60–3.62].
Typical threshold irradiances in air leading to a breakdown initialized via free
electrons from multiphoton absorption and ionization and subsequent cascade
processes are in the order of 1010 –1013 W=cm2 depending on the wavelength of
the laser radiation and the experimental conditions [3.63].
For solid samples, the homogeneity of the sample is an important issue. The
extension of spatial variations of element concentrations has to be considered
carefully in order to decide which way of spatial averaging shall be applied to
perform a representative bulk analysis, cf. Sect. 3.6. In some cases, the sample has
to be prepared before analysis with LIBS. Slag samples are, e.g., crushed, milled,
and pressed into pellets to get a homogenization, cf. Sect. 14.2.
For the spatially resolved LIBS analysis of, e.g., metallic samples with inclu-
sions, the sample surface has to be prepared by grinding or milling, cf. Chap. 15.
However, the selection of the grinding paper has to be chosen carefully, since there
is a danger to contaminate the surface of the sample by species originating from the
grinding paper so that the measuring results are disturbed. For example, SiC paper
leads to a change in the histogram of carbon signals, whereas for corundum and
zirconium paper no such change is observed [3.64]. A similar effect occurs for Ca
histograms indicating that the resin used as binder in the SiC paper may lead to a
contamination of the sample surface. Milling of the sample surface to be scanned
by LIBS is an approach to avoid such effects.
References
3.1. E. Muller, A. Lauritzen, P. Eggimann, Laser spark optical emission spectrometry: A revolu-
tionary method for the iron and steel industry. Steel Times 6, 224–225 (1998)
3.2. GmbH & Co KG, Product Information of Spectro Analytical Instruments (GmbH & Co KG,
Kleve, 2000)
3.3. I. Whiteside, R. Jowitt, Laser liquid metal analysis, in Progress of Analytical Chemistry in
the Iron and Steel Industry (Commission of the European Communities, Brussels, 1992),
pp. 135–141
3.4. Final report of the Brite-Euram project: study of emission spectroscopy on laser produced
plasma for localised multielemental analysis in solids with surface imaging, Coordinator
Saclay, France, Contract MAT 1-CT-93–0029, 1996
3.5. K. Yamamoto, D. Cremers, M. Ferris, L. Foster, Detection of metals in the environment
using a portable laser-induced breakdown spectroscopy instrument. Appl. Spectrosc. 50,
222–233 (1996)
3.6. L. Paksy, B. Nemet, Production control of metal alloys by laser spectroscopy of the molten
metals. Part 1. Preliminary investigations. Spectrochim. Acta B 51, 279–290 (1996)
3.7. M. Sabsadi, P. Cielo, Quantitative analysis of aluminium alloys by laser-induced breakdown
spectroscopy and plasma characterization. Appl. Spectrosc. 49, 499–507 (1995)
27, 805–810 (1999)
3.9. C. Aragon, J. Aguilera, F. Penalba, Improvements in quantitative analysis of steel compo-
sition by laser-induced breakdown spectroscopy at atmospheric pressure using an infrared
Nd:YAG laser. Appl. Spectrosc. 53, 1259–1267 (1999)
3.10. M. Hatcher, Breakdown spectroscopy finds nuclear application. Opto Laser Eur. April,
20–22 (2000)
3.11. V. Sturm, L. Peter, R. Noll, J. Viirret, R. Hakala, L. Ernenputsch, K. Mavrommatis,
H. Gudenau, P. Koke, B. Overkamp, Elemental analysis of liquid steel by means of laser
technology. International Meeting on Chemical Engineering, Environmental Protection and
Biotechnology, ACHEMA 2000, Materials Technology and Testing, 9–11 (2000)
3.12. I. Mönch, R. Noll, R. Buchholz, J. Worringer, Laser identifies steel grades. Stainless Steel
World 12(4), 25–29 (2000)
3.13. H. Sattler, Metallschrotte nach Analyse automatisch sortieren. Materialprüfung 35, 312–315
(1993)
3.14. R. Noll, V. Sturm, L. Peter, I. Whiteside, Analysis using lasers. Proceedings of 49th
Chemist’s Conference, Scarborough, 22–27 (1997)
3.15. cw-Nd:YAG Laser BLS 600, Manufacturer Carl Baasel Lasertechnik GmbH, Postfach 1453,
82304 Starnberg
3.16. R. Sattmann, V. Sturm, R. Noll, Laser-induced breakdown spectroscopy of steel samples
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D Appl. Phys. 28, 2181 (1995)
3.17. L.M. Cabalin, J.J. Laserna, Atomic emission spectroscopy of laser-induced plasmas gener-
ated with an annular-shaped laser beam. J. Anal. At. Spectrom. 19, 445–450 (2004)
3.18. L. St-Onge, M. Sabsabi, P. Cielo, Quantitative analysis of additives in solid zinc alloys by
laser-induced plasma spectrometry. J. Anal. At. Spectrom. 12, 997–1004 (1997)
3.19. L. St-Onge, M. Sabsabi, P. Cielo, Analysis of solids using laser-induced plasma spec-
troscopy in double-pulse mode. Spectrochim. Acta B 53, 407–415 (1998)
3.20. B. Le Drogoff, J. Margot, S. Laville, M. Chaker, M. Sabsabi, T.W. Johnston, O. Barthelemy,
Influence of the laser pulse duration on laser-produced plasma properties. Plasma Sources
Sci. Technol. 13, 223–230 (2004)
3.21. A. Semerok, C. Dutouquet, Ultrashort double pulse laser ablation of metals. Thin Solid
Films 453–454, 501–505 (2004)
3.22. R. Noll, R. Sattmann, V. Sturm, Multi-Element-Analyse von Stahllegierungen mit Laser-
Emissionsspektroskopie, in Proceedings of 12 International Congress Laser 95 (Springer-
Verlag, Berlin, 1996), pp. 626–629
3.23. L. Peter, V. Sturm, R. Noll, R. Hakala, J. Viirret, B. Overkamp, P. Koke, Multi-elemental
analysis of low-alloyed steel by laser-induced breakdown spectrometry. SPIE 3823, 256–
265 (1999)
3.25. L. St-Onge, M. Sabsabi, P. Cielo, Towards quantitative depth-profile analysis using laser-
induced plasma spectroscopy: investigation of galvannealed coatings on steel. Spectrochim.
Acta B 55, 299–308 (2000)
3.26. L. St-Onge, R. Sing, S. Bechard, M. Sabsabi, Carbon emissions following 1:064 m laser
ablation of graphite and organic samples in ambient air. Appl. Phys. A 69, S913–S916 (1999)
3.27. H. Bette, R. Noll, High-speed laser-induced breakdown spectrometry for scanning micro-
analysis. J. Phys. D Appl. Phys. 37, 1281–1288 (2004)
3.28. M. Stepputat, R. Noll, On-line detection of heavy metals and brominated flame retardants
in technical polymers with laser-induced breakdown spectrometry. Appl. Optics 42, 6210–
6220 (2003)
References 45
3.29. B. Le Drogoff, J. Margot, M. Chaker, M. Sabsabi, O. Barthelemy, T.W. Johnston, S. Laville,

F. Vidal, Y. von Kaenel, Temporal characterization of femtosecond laser pulses induced
plasma for spectrochemical analysis of aluminium alloys. Spectrochim. Acta B 56, 987–
1002 (2001)
3.30. B. Le Drogoff, M. Chaker, J. Margot, M. Sabsabi, O. Barthelemy, T.W. Johnston, S. Laville,
F. Vidal, Influence of the laser pulse duration on spectrochemical analysis of solids by laser-
induced plasma spectroscopy. Appl. Spectrosc. 58, 122–128 (2004)
3.31. J. Kutzner, M. Silies, T. Witting, G. Tsilimis, H. Zacharias, Efficient high-repetition-rate
fs-laser based X-ray source. Appl. Phys. B 78, 949–955 (2004)
3.32. S. Klimentov, S. Garnov, T. Kononenko, V. Konov, P. Pivovarov, E. Dausinger, High rate
deep channel ablative formation by picosecond-nanosecond combined laser pulses. Appl.
Phys. A 69, S633–S636 (1999)
3.33. C. Momma, B.N. Chichkov, S. Nolte, F. von Alvensleben, A. Tünnermann, H. Welling,
B. Wellegehausen, Short-pulse laser ablation of solid targets. Opt. Commun. 129, 134–142
(1996)
3.34. P. Herman, A. Oettl, K. Chen, R. Marjoribanks, Laser micromachining of “transparent”
fused silica with 1-ps pulses and pulse trains. Proc. SPIE 3616, 148–154 (1999)
3.35. H.-U. Schmitz, R. Noll, M. Kraushaar, Laser-OES: A universal technique for slag analysis,
50th Chemist’s Conference, Ashorne Hill, 1999, 11–15
3.36. V. Sturm, R. Sattmann, R. Noll, Optical fiber transmission of multiple Q-Switch Nd:YAG
laser pulses with microsecond interpulse separations. Appl. Phys. B 63, 363–370 (1996)
3.37. V. Margetic, M. Bolshov, A. Stockhaus, K. Niemax, R. Hergenröder, Depth profiling of
multi-layer samples using femtosecond laser ablation. J. Anal. At. Spectrom. 16, 616–621
(2001)
3.38. S. Nolte, C. Momma, G. Kamlage, A. Ostendorf, C. Fallnich, F. von Alvensleben,
H. Welling, Polarization effects in ultrashort-pulse laser drilling. Appl. Phys. A 68, 563–
567 (1999)
3.39. B. Salle, O. Gobert, P. Meynadier, M. Perdrix, G. Petite, A. Semerok, Femtosecond and
picosecond laser microablation: ablation efficiency and laser microplasma expansion. Appl.
Phys. A 69, S381–S383 (1999)
3.40. R. Noll, Terms and notations for laser-induced breakdown spectroscopy. Anal. Bioanal.
Chem. 385, 214–218 (2006)
3.41. D.R. Lide, Handbook of Chemistry and Physics, 67th edn. (CRC Press, Boca Raton, 1986–
1987)
3.42. H. Carslaw, J. Jaeger, Conduction of Heat in Solids, 2nd edn. (Oxford University Press,
Oxford, 1959)
3.43. E. Shiles, T. Sasaki, M. Inokuti, D. Smith, Self-consistency and sum-rule tests in the
Kramers-Kronig analysis of optical data: applications to aluminum. Phys. Rev. B 22, 1612–
1628 (1989)
3.44. J. Weaver, E. Colavita, D. Lynch, R. Rosei, Low-energy interband absorption in bcc Fe and
hcp Co. Phys. Rev. B 19, 3850–3856 (1979)
3.45. R. Klein, Bearbeitung von Polymerwerkstoffen mit infraroter Laserstrahlung, D 82, Disser-
tation, Aachen, 1990
3.46. European Standard DIN EN ISO 11146–1, Lasers and laser-related equipment – test
methods for laser beam widths, divergence angles and beam propagation ratios – part 1:
stigmatic and simple astigmatic beams, 2005
3.47. J. Vadillo, C. Garcia, S. Palanco, J. Laserna, Nanometric range depth-resolved analysis of
coated-steels using laser-induced breakdown spectrometry with a 308 nm collimated beam.
J. Anal. At. Spectrom. 13, 793–797 (1998)
3.48. L. Peter, R. Noll, Material ablation and plasma state for single and collinear double pulses
interacting with iron samples at ambient gas pressures below 1 bar. Appl. Phys. B 86, 159–
167 (2007)
3.49. J. Pecker, Space Observatories (D. Reidel Publishing, Dordrecht, 1970)
3.50. D. Anderson, C. McLeod, T. English, A. Smith, Depth profile studies using laser-induced
plasma emission spectrometry. Appl. Spectrosc. 49, 691–701 (1995)
3.51. M. Mateo, J. Vadillo, J. Laserna, Irradiance-dependent depth profiling of layered materials
using laser-induced plasma spectrometry. J. Anal. At. Spectrom. 16, 1317–1321 (2001)
3.52. C. Garcia, M. Corral, J. Vadillo, J. Laserna, Angle-resolved laser-induced breakdown
spectrometry for depth profiling of coated materials. Appl. Spectrosc. 54, 1027–1031 (2000)
3.53. M. Mateo, L. Cabalı́n, J. Laserna, Line-focused laser ablation for depth-profiling analysis of
coated and layered materials. Appl. Optics 42, 6057–6062 (2003)
3.54. D. Papazoglou, V. Papadakis, D. Anglos, In situ depth and topography monitoring in LIBS
elemental profiling of multi-layer structures. J. Anal. At. Spectrom. 19, 483–488 (2004)
3.55. M. Mowery, R. Sing, J. Kirsch, A. Razaghi, S. Bechard, R. Reed, Rapid at-line analysis of
coating thickness and uniformity on tablets using laser induced breakdown spectroscopy.
J. Pharm. Biomed. Anal. 28, 935–943 (2002)
3.56. H. Balzer, M. Hoehne, V. Sturm, R. Noll, Online coating thickness measurement and depth
profiling of zinc coated sheet steel by laser-induced breakdown spectroscopy. Spectrochim.
Acta B 60, 1172–1178 (2005)
3.57. M. Kraushaar, R. Noll, H.-U. Schmitz, Slag analysis with laser-induced breakdown spec-
trometry. Appl. Spectrosc. 57, 1282–1287 (2003)
3.58. V. Sturm, Optical microlens array for plasma generation in spectrochemical analysis. J. Anal.
At. Spectrom. 22, 1495–1500 (2007)
3.59. Ü. Aydin, R. Noll, J. Makowe, Automatic sorting of aluminium alloys by fast LIBS
identification, in Seventh International Workshop Progress in Analytical Chemistry in the
Steel and Metal Industries, ed. by J. Angeli (Glückauf GmbH, Essen, 2006), pp. 309–314
3.60. L. Peter, V. Sturm, R. Noll, Liquid steel analysis with laser-induced breakdown spectrometry
in the vacuum ultraviolet. Appl. Optics 42, 6199–6204 (2003)
3.61. L. Barrette, S. Turmel, On-line iron-ore slurry monitoring for real-time process control of
pellet making processes using laser-induced breakdown spectroscopy: graphitic vs. total
carbon detection, Spectrochim. Acta B 56, 715–723 (2001)
3.62. S. Oh, T. Miller, F. Yueh, J.P. Singh, Comparative study of laser-induced breakdown
spectroscopy measurement using two slurry circulation systems. Appl. Optics 46, 4020–
4025 (2007)
3.63. A. Miziolek, V. Palleschi, I. Schechter (eds.), Laser-Induced Breakdown Spectroscopy
(Cambridge University Press, Cambridge, 2006), p. 212
3.64. Improved production control through rapid characterisation of non-metallic inclusions in
steel, Report, EUR 21627, 2005
3.65. A. Löbe, J. Vrenegor, R. Fleige, V. Sturm, R. Noll, Laser-induced ablation of a steel sample
in different ambient gases by use of collinear multiple laser pulses. Anal. Bioanal. Chem.
385, 326–332 (2006)
Chapter 4
Instrumental Components
4.1 Laser
Various laser types are applied for LIBS and laser ablation studies. The most
wide spread are flashlamp-pumped solid-state lasers with Nd:YAG as laser medium
operated in the Q-switch mode to generate high-energy laser pulses with pulse
durations in the nanosecond range. Table 4.1 gives an overview of the lasers and
laser parameters used for LIBS.
For further details concerning characteristics, design, construction, and perfor-
mance of solid-state lasers, see [4.29, 4.30]. In the following paragraphs, the main
features of solid-state lasers and here especially the Nd:YAG lasers are presented,
since these are of predominant importance for LIBS. The laser medium is a
doped insulator; dopants are rare earths or transition metals. The laser medium is
pumped optically with broadband noble gas or halogen lamps (emission range 200–
1,000 nm) or narrow band semiconductor lasers (806 nm).
Figure 4.1 shows the principal setup of a flashlamp-pumped solid-state laser.
The laser medium is usually a rod cooled with water. The flashlamps are arranged
parallel to this rod. The pump light is reflected and concentrated by the pump cavity.
The inner surface of the pump cavity has a highly reflecting coating or a scattering
surface to homogenize the illumination of the laser rod. For pulsed Nd:YAG lasers –
as typically applied for LIBS – Xe flashlamps are used with typical life times of
20–50 million flashes depending on the total energy dissipated during the flashlamp
discharge. The emission spectrum of the Xe flashlamp only partly matches the
excitation spectrum of Nd:YAG; hence, only a small fraction of the absorbed pump
power is transferred to the upper laser level. As a consequence less than 3% of the
electrical input power is converted into laser radiation. Diode-pumped systems offer
a better matching to the absorption band and reach 10–20% overall efficiency (laser
beam power in relation to electrical power input). Therefore, diode-pumped solid-
state lasers (DPSSL) systems save energy costs and secondary costs like cooling
engines. The reduced heat load of the laser crystal leads to smaller internal stresses

48 4 Instrumental Components
Table 4.1 Overview of laser types used for LIBS

Laser type (nm) Pulse energy, pulse Repetition References and remarks
duration rate (Hz)
Nd:YAG 1,064 1–1,000 mJ, 3–50 ns 1–100 Standard laser for LIBS,
laser medium is
pumped by flashlamps,
typical beam quality is
in the range of
M 2 D 3–6. Multiple
laser pulses can be
generated from a single
laser unit [4.1–4.10]
Nd:YAG, SHG 532 1–90 mJ, 5–14 ns 5–10 [4.11–4.16]
Nd:YAG, THG 355 10 mJ, 6 ns 5–30 [4.11, 4.17]
Nd:YAG, FHG 266 3–70 mJ, 4–8 ns 2–30 [4.18–4.21]
Nd:YAG 1,064 2.7 mJ, 25 ps [4.22]
Nd:YAG, SHG 532 1.5 mJ, 18 ps [4.22]
Nd:YAG, FHG 266 0.7 mJ, 13 ps [4.22]
Diode-pumped 1,064 2 mJ, 5 ns 1,000 Beam quality M 2 D 1:3
Nd:YAG [4.23]
Ti–sapphire, 400 150 fs [4.22]
SHG
Excimer XeCl 308 70–150 mJ, 5–50 [4.24, 4.25]
20–30 ns
Excimer KrF 248 40–70 mJ, 30 ns [4.26]
Excimer ArF 193 250 mJ, 20 ns 1–50 [4.27]
Dye 500 22 mJ, 30 ns 10 Pumped by a XeCl excimer
laser [4.28]
The range of parameters given in the columns – pulse energy, pulse duration, and repetition rate –
refer to data of different laser systems for the given laser type
SHG D second harmonic generation, THG D third harmonic generation, FHG D fourth harmonic
generation
Fig. 4.1 Principal setup of a flashlamp-pumped solid-state laser

4.1 Laser 49
Table 4.2 Typical Laser crystal Laser wavelength Pumping

diode-pumpable laser (nm) wavelength (nm)
crystals, their strongest laser
emission wavelengths and Nd:YAG 1,064, 1,319 808
suitable pumping Nd:YVO4 1,064, 1,340 809
wavelengths to be generated Nd:YLF 1,047, 1,053 792, 798
by laser diodes Yb:YAG 1,023–1,052 941
Cr:LiSAF 800–920 670, 760
and changes in the optical properties which improves the mode quality of the laser
beam (cf. Table 4.1, row 8).
The specific features of solid-state lasers in comparison with gas, liquid, or
semiconductor lasers are high power, short pulses, efficient frequency conversion,
and the capability to transmit the laser radiation via fiber optics (cf. Sect. 4.3).
The active ion in a Nd:YAG laser rod is Nd3C embedded in an yttrium aluminum
granate (YAG, Y3 Al5 O12 ) crystal. The Nd:YAG laser is a four-level system offering
a high amplification and suitable mechanical and thermal properties. The excitation
by optical pumping occurs into broad absorption bands and subsequent radiation-
less transitions to the upper laser level. The lifetime of this level (4 F3=2 ) is 230 s,
whereas the lower level has a lifetime of 30 ns only. The lower level lies 0.24 eV
above the ground level, and hence at room temperature the population of this level is
negligible. As a four-level system, the laser threshold of Nd:YAG is small. The laser
lines are broadened homogeneously by thermal lattice vibrations. The line width is
typically about 100 GHz at room temperature.
During the last years, the availability of high-power semiconductor laser diodes
has stimulated the development of all-solid-state laser systems [4.29, 4.31–4.33].
Instead of the spectrally broad emitting flashlamp, laser diodes pump the laser
crystal in a narrow spectral band selected in such a way to increase the efficiency of
the pumping by minimizing energy dissipation. These DPSSL in Q-switch operation
offer high pulse repetition rates of up to 1 kHz and more (cf. Chap. 15). Although
the pulse energy is still limited to a few millijoule, high intensities in the focal
spot can be achieved due to the excellent beam quality of DPSSL systems. The
potential benefits of DPSSL compared with flashlamp-pumped lasers for LIBS
are faster acquisition of spectral information, higher spatial resolution, improved
precision on the basis of significantly enlarged data ensembles (mean values, pulse
discrimination analysis), lower energy consumption, and increased life time. Laser
diodes have typical lifetimes of 10,000 h in continuous wave mode and 109 –1010
pulses in pulsed mode. In contrast to this typical flashlamp lifetimes amount to about
5 107 pulses. Typical diode-pumped laser materials and their important pumping
wavelengths are listed in Table 4.2 [4.29, 4.31].
The main advantage of diode-pumped solid-state laser systems is the high beam
quality in combination with the achievable pumping efficiency. This means that a
small laser focus diameter is easier to achieve with a DPSSL than with a flashlamp-
pumped system. To get a small focus with a flashlamp-pumped system requires the
use of apertures within the oscillator for mode selection, which in turn limits the
efficiency factor optical-to-optical down to 1–3%.
Fig. 4.2 Typical layout of a diode-pumped solid-state laser (DPSSL) in an end-pumped configu-
ration. DP D diode pump module, FO D fiber optic, L D lens, M D mirror, Q-Sw D Pockels cell
for Q-switch operation
Fig. 4.3 Photograph of the Nd:YAG DPSSL developed at ILT, data is given in Table 4.3
DPSSL systems can be pumped in a transversal configuration as it is the common

arrangement for flashlamp-pumped systems, as well as in a longitudinal (called
“end-pumped”) configuration, see Fig. 4.2. High beam quality and efficiency in an
end-pumped configuration are achieved by gain guiding. In this case, the pumped
volume and laser mode have a high degree of overlap. Figure 4.3 shows a photograph
of a DPSSL based on this principle developed at Fraunhofer ILT, Aachen. Table 4.3
lists the data of that laser.
Q-switched DPSSL have tunable repetition rates from below 1 Hz to several
kilohertz, favoring them for the generation of high repetitive laser pulses in the
4.2 Spectrometer 51
Table 4.3 Data of the Wavelength 1,064 nm

Nd:YAG diode-pumped laser
Repetition frequency 1 kHz
shown in Fig. 4.3
Pulse energy 2 mJ
Pulse duration 5.5 ns
Beam quality M 2 1:3
millijoule range. Preflashing of the laser to achieve thermal stability of the system
and by that to generate reproducible laser pulses is strongly reduced or even not
necessary, thus eliminating dead times.
4.2 Spectrometer
Spectrometers disperse the emitted radiation of the laser-induced plasma to get

a spectrum in terms of intensity as a function of the wavelength. The dominant
spectrometer types used for LIBS are: (a) Czerny–Turner, (b) echelle, and (c)
Paschen–Runge. For a detailed description of the various spectrometer types, the
reader is referred to the literature [4.34, 4.35].
Figure 4.4 shows the setup of a Czerny–Turner spectrometer. The incoming light
passing the entrance slit is collimated by a first concave mirror and directed onto a
plane grating. The dispersed radiation is imaged by a second concave mirror to a
detector plane.
The dispersion is described by the grating equation:
dg .sin ˛ C sin ˇ/ D n; (4.1)
where dg is the grating period, ˛ is the angle of incidence, the sign of ˛ is negative
if the incident beam is on the opposite side of the grating normal with respect to the
diffracted beam, ˇ is the angle of diffraction, n is the diffraction order, and is the
wavelength.
The linear dispersion in the detector plane at a central wavelength along an axis
x is given by:
d dg cos ˇ
D ; (4.2)
dx nf
where x is the coordinate in the detector plane, parallel to the dispersion direction,
and f is the focal length of the exit mirror.
For a spectrometer with a grating having 1,200 grooves/mm (d D 0:83 m/,
a deviation angle .ˇ ˛/ of 27:5ı wavelength, a central wavelength of 500 nm,
first order (n D 1), and a focal length of the second mirror of 640 mm, we obtain:
d=dx D 1:2 nm=mm.
The resolving power of a spectrometer is given by:

RD D nN; (4.3)

Fig. 4.4 Setup of a

Czerny–Turner spectrometer.
MCP D microchannel plate
Fig. 4.5 Schematic optical

setup of a Czerny–Turner
spectrometer. ES D entrance
slit, M1, M2 D mirrors, G D
grating, XS D exit slit
where is the spectral separation of two lines and N is the total number of
illuminated grooves of the grating.
The quantity describes the difference between two neighboring spectral lines
of equal intensity. These two lines can be resolved, if the Rayleigh criterion is
fulfilled, i.e., the maximum of one line falls on the first minimum of the other.
Taking as an example the grating from above with 1,200 grooves/mm, an illuminated
grating size of 100 mm, 500 nm, first order, we obtain from (4.3): R D 1:2 105
and D 4:2 pm.
Figure 4.5 illustrates schematically the beam propagation in a Czerny–Turner
spectrometer from the entrance slit to the exit slit. The optical setup is shown here in
an equivalent linear arrangement. The numerical aperture of the spectrometer seen
from the entrance slit is defined as follows:
NAES D nref sin ; (4.4)
where nref is the refractive index and is the angle shown in Fig. 4.5.
An analogous expression holds for the numerical aperture seen from the exit slit.
The f -number or f -value for the spectrometer entrance is given by:
s1
f -value D ; (4.5)
Dg
4.2 Spectrometer 53
where s1 is the distance between entrance slit and mirror M1, cf. Fig. 4.5, and Dg is
the effective diameter of the grating G, cf. Fig. 4.5.
For a circular grating, the effective diameter seen by the entrance slit is Dg D
Wg cos ˛, where Wg is the diameter of the grating. For a rectangular grating, an
area equivalent diameter is calculated. For f -values greater than 2, the following
approximation holds:
1
f -value : (4.6)
2NAES
As an example, the NA and f -value are for a Czerny–Turner system with 640 mm
focal length of the first concave mirror, a grating of 110 mm 110 mm; and an
angle of incidence of ˛ D 0:14 rad: NAES D 0:095, f -value D 5:3:
The Czerny–Turner configuration is the most widely used spectrometer type
for LIBS. It offers a high resolution at compact dimensions. In most cases, a
combination of a microchannel plate (MCP) and a photodiode array (PDA) is used to
detect the spectrum, cf. Sect. 4.4. However, the spectral range, which can be detected
simultaneously, is limited typically to 10–30 nm.
A much broader spectral range can be dispersed with an echelle spectrometer,
whose principle setup is shown in Fig. 4.6 [4.36]. Behind the entrance slit, the
incoming light is collimated by a first mirror and directed to a prism. The refracted
light irradiates the echelle grating under a flat angle. The echelle grating has a low
groove density of typically 80 grooves/mm. The diffracted light passes again the
prism and is imaged by a camera mirror to the detector plane (typical dimensions
25 mm 25 mm). The dispersion direction of the prism is perpendicular to the
dispersion direction of the echelle grating, and by this the high diffraction orders
are separated in the detector plane. A set of spectra are obtained for different
orders (see wavelength and order axis in Fig. 4.6). The orders are in the range
of n D 30–120. The spectral range detected simultaneously is 200–780 nm. The
resolving power including the MCP–PDA is about 20,000 [4.37].
Due to the dense packaging of lines and orders in the detector plane, there is a
danger of a cross talking between a strong spectral line and adjacent weak lines.
Since the sensitivity of the MCP–PDA is nearly equal for the whole detector plane,
there is no possibility for an individual adaption to these different line intensities.
Also the gating of the MCP with the parameters tdelay and tint (cf. Sect. 2.1) is the
same for the whole detector plane.
The Paschen–Runge spectrometer shown in Fig. 4.7 offers a wide spectral range,
a high spectral resolution, and the possibility for an individual adaption of the
detector sensitivity and gating parameters for each spectral line. Entrance slit,
grating, and a series of exit slits are mounted along the Rowland circle. The grating
has a spherical curvature with a radius equal to the diameter of the Rowland
circle. Behind the exit slits, photomultiplier tubes (PMT) are mounted detecting
the light passing through the respective exit slit. Typical diameters of the Rowland
circle are 500, 750, and 1,000 mm. The groove density of the applied gratings is,
e.g., 1,080, 1,200, 2,400, and 3,600 grooves/mm. Depending on the spectral lines,
Fig. 4.6 Setup of an echelle spectrometer
Fig. 4.7 Setup of a

Paschen–Runge spectrometer.
PMT D photomultiplier tube
different PMTs are selected with optimized cathode sensitivity especially in the deep
ultraviolet and near infrared region (cf. Sect. 4.4).
Table 4.4 gives an overview of typical data of the spectrometer types used for
LIBS.
Figure 4.8 illustrates the wavelength–time domains for different LIBS appli-
cations. The vertical axis shows the time delay with respect to the laser pulse
irradiation (at t D 0 s) and the duration of the integration of the spectral signals.
Czerny–Turner–MCP systems generally cover a rather narrow spectral range of
about 10–20 nm. The integration time is the same for the whole registered spectrum,
and hence these domains are drawn as closed boxes in Fig. 4.8, cf. numbers 1 and 2.
The echelle–MCP systems detect a broader wavelength range, where the integration
4.2 Spectrometer 55
Table 4.4 Typical data of the spectrometer types Czerny–Turner, echelle, and Paschen–Runge
Spectrometer type Czerny–Turner Echelle Paschen–Runge
Focal length (mm) 640 250 500
Grooves (mm1 ) 1,200 75 2,400
Diffraction order 1–2 30–120 1–3
f value 5 10 12
Width of spectral range 30 580a 460b
detected simultaneously
(nm)
Spectral resolution per <20 5–19 <20
pixel, exit slit width
(pm)
Detectors MCP–PDA MCP–CCD PMT
Size (mm mm mm) 740 350 350 320 140 160 940 790 345
MCP D microchannel plate, PDA D photodiode array, CCD D charge-coupled device, PMT D
photomultiplier tube
a
From 200 to 780 nm
b
From 130 to 590 nm
Fig. 4.8 Wavelength–time domains for different spectrometer types used for various LIBS
applications. The numbers refer to the following references: 1 D [4.38], 2 D [4.39], 3 D [4.40],
4 D [4.41], 5 D [4.42], 6 D [4.43], 7 D [4.23], 8 D [4.44]
time is also fixed for all wavelengths, see closed boxes 3 and 4 in Fig. 4.8. By
contrast Paschen–Runge systems comprise a broad spectral range and the delay
time and the integration time can be adjusted selectively for the different element
emission lines, see boxes with dashed horizontal lines for the numbers 5–8 in
Fig. 4.8. Furthermore, the sensitivity of each detector channel can be adapted
according to the intensity range of the respective emission line. In this way, the
dynamic range can be expanded by many orders of magnitude.
For some application, the observation of selected small spectral regions is
sufficient to identify a material [4.45]. In these cases, the plasma emission is filtered
spectrally by using band-pass filters which are centered at the interesting wavelength
and have a width of typically 10 nm. The radiation passing the filter is detected, e.g.,
with a photomultiplier.
4.3 Beam Guiding Optics
The beam guiding optics comprises all optical components used to guide the laser
radiation to the measuring object and to collect and guide the measuring radiation
to the spectrometer. Figure 4.9 gives an overview of the different variants of optical
setups applied for LIBS.
In configuration (1) of Fig. 4.9, the laser beam is focused by a lens onto the
surface of the measuring object and the light emitted by the plasma passes a window
and is collected by a fiber optics guiding the light to the spectrometer, see, e.g.,
[4.46]. The window protects the fiber end face from contaminations by deposition or
recondensation of particulates or vapor originating from the laser-induced material
ablation and plasma. Setup (1) is simple and relatively stable against variations of
the position of the plasma and the fiber end face due to the numerical aperture of the
fiber. For a typical fiber, the numerical aperture amounts to 0.22, which corresponds
Fig. 4.9 Variants of beam guiding optics applied for LIBS. LB D laser beam, FL D focusing lens,
P D plasma, S D sample, W D window, FO D fiber optics, L D lens, ES D entrance slit, DM D
dichroic mirror, M D mirror, CO D Cassegrain optics, GT D Galilean telescope, SM D spherical
mirror
4.3 Beam Guiding Optics 57
1.0
transmittance P (10 m)/P(0)
0.8
0.6
0.4
0.2
0.0
200 300 400 500 600 700 800 900 1000
wavelength [nm]
Fig. 4.10 Transmittance of an all-silica fiber optic (core and cladding are made of fused silica)
with a core diameter of 600 m and a length of 10 m as a function of the wavelength
to a full angle of the acceptance cone of 25ı . However, the solid angle collected from
the plasma light emission is very small, so the sensitivity achievable is reduced. In
setup (2), a lens is used to collect the light emitted by the plasma and to couple the
light into the fiber optics. In this way, the detected solid angle can be increased by
orders of magnitude. However, the lens has a chromatic aberration, i.e., the focusing
features are different for the various wavelengths emitted by the plasma. Hence, the
imaging conditions have to be selected carefully in order not to cut-off parts of the
emitting plasma volume. For this setup plasma position, lens, and fiber end face
have to be aligned precisely in order to achieve a high coupling efficiency.
Using fiber optics in the configurations (1) and (2) allows to position the
spectrometer up to several meters away from the measuring object. Especially
for LIBS applications under harsh environments such as the inline analysis of
measuring objects at high temperatures, this is a significant advantage. However, the
fiber optics limits the transmittable spectral range for wavelengths below 200 nm, cf.
Fig. 4.10. At configuration (3), the plasma emission is directly coupled into the
entrance slit of the spectrometer. Again the solid angle received is small and on
the other hand UV lines can be observed, if the surrounding atmosphere is an inert
gas having no absorption bands in the UV, e.g., argon. With this setup emission lines
down to 120 nm can be detected, which is of special interest for the analysis of light
elements in steel, see Sect. 13.1. Configuration (4) is similar to (2). If a magnesium
fluoride .MgF2 / lens is used, wavelengths in the deep UV can also be detected.
In setup (5), the measuring radiation propagating antiparallel to the incoming
laser radiation is collected by the focusing lens and reflected by a dichroic mirror
guiding the radiation to the spectrometer. For this configuration holds ˛L D ˛O ,
cf. Fig. 3.23. Observing the plasma emission along the same axis allows to look
into the crater formed by the laser interaction thus avoiding shielding of parts of
the plasma by the crater walls. This is especially of interest if series of laser pulses
are applied at the same location. Furthermore, setup (5) can be combined with a
scanner to deflect the laser beam into different directions, cf. Sect. 18.4. Due to the
chromatic aberration of the focusing lens, the coupling efficiency for the measuring
radiation depends on the wavelength. Hence, a compromise has to be found for the
spectral range of the plasma radiation of interest in a specific application.
Configuration (6) uses a spherical mirror to collect the plasma emission thus
avoiding chromatic aberrations. However, such a mirror is subject to spherical
aberrations. The spherical mirror allows for high collection efficiencies down to the
deep UV region which is of interest for microanalysis tasks where a tightly focused
laser beam at low pulse energy is used to detect inclusions in steel, cf. Sect. 18.3.
Setup (7) provides a chromatic- and spherical-aberration-free detection, which is
of importance for high-spatial-resolution measurements and for the simultaneous
monitoring of signals in different wavelength regions from a given spatial volume
[4.47]. In this case, we have ˛L D 0ı and ˛O D 90ı , which limits the applicability
to the analysis of gaseous media.
Finally, configuration (8) is used for remote LIBS analyzing systems, where the
laser beam is focused at distances of several meters by a Galilean telescope and
the plasma radiation is collected by a Newtonian telescope consisting of a spherical
primary mirror and a flat secondary mirror guiding the collected radiation to a fiber
optics [4.48].
A key feature of the refractive optical components used for the various con-
figurations shown in Fig. 4.9 to collect and guide the measuring radiation to the
spectrometer is the transmittance as a function of the wavelength. Figure 4.10 shows
the transmittance curve for a silica fiber optics with a length of 10 m. For wavelength
below 300 nm, the transmittance decreases significantly attaining values of about
0.2 at 200 nm. In the visible and near infrared regions, high transmittance values
greater than 0.8 exist. For wavelengths beyond 900 nm, absorption bands of OH are
effective.
For wavelength below 190 nm, fluorides are the only materials offering signifi-
cant transmittance. Figure 4.11 shows the transmittance of CaF2 and MgF2 windows
for wavelength down to 120 nm. At short wavelength, the cleanliness of the surfaces
plays an important role. A comparative study of the transmittance of CaF2 windows
with thicknesses of 2 and 20 mm shows that for wavelength greater than 140 nm the
volume absorption is negligible in relation to the surface absorption.
Fiber optics can also be used to transmit the laser radiation from the laser
source to the focusing lens. However, the transmittable laser pulse energy is limited
by laser-induced damage of the fiber core material and stimulated backscattering
processes. Table 4.5 summarizes threshold data for laser-induced damage of bulk
fused silica, which present an upper limit for the fiber damage thresholds. The
energy thresholds for laser-induced damage "d;th depend on the pulse width as
"d;th / tp0:4;:::;0:5 for bulk samples [4.53, 4.55]. The threshold values decrease
with increasing focal spot size which is explained by the increasing probability to
irradiate surface defects having lower damage thresholds.
4.3 Beam Guiding Optics 59
1.0
0.9
CaF2, 2 mm
transmittance
0.8
MgF2, 5 mm
0.7
0.6
0.5
100 150 200 250 300
wavelength [nm]
Fig. 4.11 Transmittance of CaF2 (thickness 2 mm) and MgF2 (thickness 5 mm) in the ultraviolet
spectral range [4.49–4.51]
Table 4.5 Damage threshold radiant exposure "d;th and corresponding irradiance Id;th for bulk
fused silica samples at 1:06 m wavelength
Tp (ns) 2r (mm) "d;th .J=cm2 / Id;th .GW=cm2 / References
18 0.0072 3,960 220 [4.52]
27 0.0090 3,240 120 [4.52]
15 1 60 4 [4.53]
30 1 2 210 7 [4.54]
80 1 130 1.5 [4.53]
tp D pulse width (FWHM), 2r D focal spot diameter
Table 4.6 shows the maximum fiber output energy per pulse for single, double,
and multiple pulses. In the rows 1–4, maximum transmitted energies are reported
for single Q-switch pulses when stimulated backscattering is not observed due to
the short fiber lengths, short pulse widths, and/or spectral width and spatial profile
of the laser. Measured values for single and double pulses at a fiber length of 5 m
are shown in rows 5 and 6. The damage threshold increases for a multimode spatial
profile of the laser and for a higher input beam divergence. The references [4.56,
4.57] only achieved the higher transmitted energy levels with a highly multimode
spatial profile of a Nd:glass laser and high input divergence 2in resulting in a high
output divergence of the transmitted beam. If a near-Gaussian spatial profile is used,
then much lower thresholds are achieved [4.57], see row 4. These radiant exposure
thresholds "max are about a factor of 1.7 higher than the values measured in [4.58]
for a Gaussian beam, see row 7, but the input divergence was five times smaller.
From this, no essential decrease of the fiber damage radiant exposure thresholds
60
Table 4.6 Maximum fiber output energy per pulse Emax , radiant exposure "max , and irradiance Imax for fused silica fibers of core diameter d
No. tp (ns) d (mm) L (m) N Emax (mJ) "max .J=cm2 / Imax .GW=cm2 / 2in .mrad/ Spatial profile References
1 16 0.4 1 1 100 80 5.0 50 MM [4.56]
2 25 0.4 1 1 97 77 3.1 64 MM [4.57]
3 25 1 1 1 >225a >29 >1:1 64 MM [4.57]
4 16 1 1 1 70 8.9 0.6 64 NG [4.57]
5 15 1 5 1 >38a >4:8 >0:32 <12 NG [4.58]
6 25 1 5 2 >60a >7:6 >0:31 <12 NG [4.58]
7 50–80 1 0.5 6 40 5.1 0.1 12 G [4.58]
8 50–80 1 0.5 6 >50 >6:4 >0:13 <12 NG [4.58]
tp D pulse width (FWHM), L D fiber length, N D number of pulses per burst, 2in D fiber input full-angle divergence
Spatial profile of laser output: MM D multimode, NG D near-Gaussian, G D Gaussian. Interpulse separations used for the measurements in row 6 are
t D 6 s and for rows 7, 8: 15 s
a
Limited by air breakdown ahead of fiber input face
4 Instrumental Components
4.4 Detectors 61
Table 4.7 SBS threshold energies Ep;th (in mJ) dependent on the spectral width Q of the laser
radiationa
Type Q < 0:002 cm1 Q D 0:1 cm1 Q D 0:65 cm1
S, WF 5.8 45 >90 .SP/
S, UV 3.4 26 >90 (SP), 300 (FPB)
G 0.75 8 >30 (SP)
S, WF D step-index fiber, water-free with enhanced near infrared transmission, UV D fiber
glass with enhanced short wavelength transmission, G D gradient-index fiber, SP D single pulse,
FPB D five-pulse burst with 10-s interpulse separations
a
For the broad spectral width, the fiber damage thresholds are given as a lower limit estimation of
the actually higher SBS thresholds. Fiber core diameter 0.6 mm, fiber length 5 m, input NA 0:07
for the individual pulses is obvious, due to the microsecond-spaced multiple pulse
irradiation, i.e., the transmissible energy of a N -pulse burst increases to about N
times the maximum energy of a single pulse.
For laser radiation with narrow spectral width stimulated Brillouin scattering
(SBS) limits the transmission of Q-switched laser pulses in large-core optical fibers
well below the threshold energy of fiber damage [4.59]. Table 4.7 summarizes
the SBS thresholds measured for different spectral widths of the laser radiation. The
threshold increases by a factor of 7:5 for the step-index fibers and 10 for the
gradient-index fiber, respectively, when the spectral width of the laser is changed
from <0:002 to 0:1 cm1 (i.e., by at least a factor of 50). For a laser with a spectral
width of 0:65 cm1 , the SBS thresholds exceed the damage thresholds of the fiber
input surface for all samples when single laser pulses are used.
Table 4.7 shows that energies of >90mJ for single pulses and 300 mJ for five-
pulse bursts are transmitted through 0.6-mm-diameter fibers of 5 m length, when a
broadband laser with a spectral width of 0:65 cm1 is used.
Fiber optic laser beam delivery was applied for remote material analysis by
LIBS using all-silica fibers with a core diameter of 550 m and a length of 75 m
[4.60]. Via this fiber Q-switched Nd:YAG laser pulses at 1,064 nm with 20-Hz pulse
repetition frequency and a pulse width of 6 ns were transmitted with pulse energies
of 5.5 mJ. At pulse energies greater than 9 mJ catastrophic damages occurred either
at the input face of the fiber, or more often, approximately 75 mm into the fiber.
4.4 Detectors
Various detector types are applied to register the spectrally dispersed radiation.
Table 4.8 lists the most import detectors and their typical data used in spectrometers
for LIBS measurements.
The PMT is a highly sensitive detector offering a large current amplification,
dynamic range, and measuring frequency. These detectors are used mainly in
combination with Paschen–Runge spectrometers, cf. Sect. 4.2, Fig. 4.7. Depending
on the respective LIBS application up to 45 and more PMTs are installed along the
Rowland circle of a Paschen–Runge spectrometer. PMTs allow for a time-resolved
Table 4.8 Detector types applied in spectrometers used for LIBS and their typical data and
features
Detector PMT CCD MCP–CCD
Number of 1 1,024 1;024 1;024.CCD/
sensitive
elements
Size of detector 4 mm 13 mm 14 m 14 m, ;25 mm.MCP/; 25:4 mm
element length of sensing 25:4 mm.CCD array/
area 14.4 mm
Spatial resolution 14 m 36 lp/mm
Window materials U; Q; MgF2 Glass, UV-enhanced Synthetic silica
fused silica
Quantum <65 <90 <17
efficiency (%)
Sensitivity 50–60 mA/W at 40 V=J=cm2 at 45 mA/W at 400 nm
peak 740 nm
wavelength
Gain Current Radiant emittance gain:
amplification: 9 103
3 105
Rise time 1.5 ns Integrating Integrating
Exposure control Possible >1 ms; 10 s 10 ns–10 s
Dynamic range >104 2,500:1 Limited by CCD
Measuring >1 kHz <500 Hz <20 Hz
frequency
Size ;13:5 mm40 mm 38:1 mm ;53 mm 19 mm.MCP/
10:03 mm
7:62 mm
Spectrometer Paschen–Runge Czerny–Turner, Czerny–Turner, echelle
Paschen–Runge
PMT D photomultiplier tube, CCD D charge-coupled device, MCP D microchannel plate,
U D UV glass, Q D synthetic silica
detection of the line emission; typical signals are shown in Fig. 3.13. The PMT
output signals are processed by a subsequent electronics where parts of the signal are
integrated as illustrated in Fig. 2.4. The high measuring frequency is deployed for
high-speed LIBS applications, see Chap. 15. The spectral PMT sensitivity depends
on the type of the cathode and window material. Figure 4.12 shows photocathode
sensitivity and quantum efficiencies as a function of the wavelength for typical
cathode and window materials. Spectral signals down to the deep vacuum ultraviolet
region with wavelength of, e.g., 130.22 nm for oxygen can be detected. For the major
part of industrial LIBS applications where the objects to be analyzed are metals,
PMTs are the detector type used preferentially.
CCD detectors consist of an array of single photosensitive elements thus they
are capable to detect a spectral range of, e.g., several tens of nanometers depending
on the dispersion of the spectrometer, see, e.g., Fig. 4.8. Their dynamic range is
much smaller than the one of PMTs. For best signal-to-noise ratios, CCD detectors
are cooled by, e.g., thermoelectric peltier elements. The CCD is an intrinsically
4.4 Detectors 63
Fig. 4.12 Photocathode sensitivity and quantum efficiency of various photocathode and window
materials. U D UV glass, Q D synthetic silica [4.61]
integrating detector type with integration times of typically 1 ms and more. Some
detector types allow for an exposure control with integration times down to
about 10 s.
Figure 4.13 shows the quantum efficiency as a function of the wavelength
depending on the light input direction. Front-illuminated CCDs (“front-sided” in
Fig. 4.13) have smaller quantum efficiencies than back-illuminated CCDs (“back-
thinned” in Fig. 4.13). Polysilicon and insulating layers usually deposited on the
photosensitive surface of a front-illuminated CCD limit the sensitivity for short
wavelengths because of the high absorption of the radiation, which leads to short
penetration depth (<2 nm), and absorption of the radiation in the gate material rather
than within the channel of the CCD [4.62, 4.63].
To detect UV light with front-illuminated CCDs, their photosensitive surface
is coated with a phosphor. The phosphor coating is prepared by sedimentation of
phosphor grains on the CCD with grain sizes in the range of 1–3 m. A typical
phosphor used consists of Gd2 O2 S:Tb having the maximum light emission at
545 nm with a decay time of 3.5 ms [4.64].
The MCP–CCD listed in the last column of Table 4.8 is a widely used detector
for LIBS studies. This combination of a microchannel plate and a CCD is also called
intensified CCD (D ICCD, MCP acts as an intensifier). The principle setup is shown
in Fig. 4.14. The irradiated photons generate free electrons at the photocathode,
which are accelerated toward the microchannel plate due to the applied electric field.
The MCP is a thin disc consisting of millions of small channels where each channel
has a diameter of 6 m. The channels serve as independent electron multipliers. The
Fig. 4.13 Quantum efficiency of CCD detectors without window. The spectral sensitivity of front-
illuminated CCDs can be extended to the UV region by coating of the photosensitive surface
(UV coat)
Fig. 4.14 Schematic setup of

a MCP-CCD. The gap
between the fiber optic plates
is shown here to illustrate the
different components. This
gap is zero in practice
input electrons impinging on the channel wall produce secondary electrons. This
process is repeated hundreds of times by the potential gradient across both ends of
the MCP and a large number of electrons are in this way released from the output
end. They are further accelerated onto the phosphor screen that reconverts electrons
into light. Finally, a fiber optic plate guides the light to the CCD.
Figure 4.15 shows an example of quantum efficiencies for image intensifiers
(MCP).
4.5 Signal Electronics 65
Fig. 4.15 Quantum efficiency of various image intensifiers [4.65]
4.5 Signal Electronics
For the detection and processing of the photomultiplier signals, a multichannel

integrator electronics MCI was developed [4.66]. Figure 4.16 shows the block
diagram, and the data is given in Table 4.9. The electronics is capable to process
up to 64 PMT signals in a pulse-by-pulse way. The anode currents of the PMT are
transmitted via coaxial cables to a gated multichannel integrator electronics. Each
channel is equipped with two different integration capacitors allowing to adapt the
sensitivity to different signal levels of the PMT output.
The dynamic range with respect to the ratio of the integration capacities (cf.
Table 4.9) and the digitalization resolution amounts to: .CB =CA / 215 D 1:5 106 .
The experimentally achieved dynamic range covers a range of integrated currents,
i.e., electric charges between 20 fC and 10 nC, corresponding to a ratio of 5 105 .
Taking into account the possibility to adapt the integration time tint and the voltage
supply of the PMT, dynamic ranges being greater than >107 are achievable.
The integrator electronics MCI is used to process PMT signals in a broad range
of industrial applications (cf. Chap. 18).
The development of the next generation of MCI, called MCI-2, was completed
in 2011 achieving maximum measuring frequencies of 3 kHz and beyond.
Fig. 4.16 Block diagram of the multichannel integrator electronics MCI. The electronics is able
to integrate the PMT output current in defined integration windows, digitizes the integrated electric
charges, and synchronizes the integration window with the laser source. PMT D photomultiplier,
PR D Paschen–Runge, A/D D analog-to-digital converter
Table 4.9 Data of the Number of channels up to 64

integrator electronics
Integration capacitors 22 pF.CA /; 1 nF.CB /
developed at Fraunhofer ILT
Delay time for the 240 ns–300 s
integration window,
tdelay
Width of the integration 50 ns–200 s
window, tint
Digitalization resolution 15 bit
Maximum measuring 1,000 Hz
frequency
Charge detection limit 20 1015 C
Sensitivity, delay time, and Programmable per
width of integration pair of channels
4.6 Sample Stand and Measuring Chamber
LIBS is a spectroscopic method which can be applied under ambient atmospheric

conditions, which is especially of interest for inline applications such as high-
speed identification of scrap pieces for material specific sorting, cf. Sect. 18.4. If the
analytical requirements with respect to sensitivity and precision are more stringent,
then well-defined ambient atmospheric conditions are necessary (cf. Sect. 3.5).
In those cases, a sample stand or a measuring chamber is used to establish a
controllable gas atmosphere in the interaction region. In the case of a measuring
chamber, the sample to be analyzed is brought into a gas tight vessel, see, e.g.,
Fig. 2.6. A sample stand is a measuring chamber with an opening in a sample table.
4.6 Sample Stand and Measuring Chamber 67
Fig. 4.17 Schematic drawing

of a sample stand and
measuring chamber with
respect to the parameters
defining the gas exchange.
C D measuring chamber,
S D sample, for the other
quantities see text
A solid sample having a plane surface is put on the sample table in such a way that
the opening is closed, see Fig. 2.7. The sample is then irradiated by the laser beam
within the area of the opening in the sample table.
A sample stand or a measuring chamber allows to control the following
parameters: gas type, gas flow rate, transport of the ablated material out of the
interaction region, transmission characteristics between the laser-induced plasma
and the spectrometer entrance window. Figure 4.17 illustrates schematically a
measuring chamber with respect to the parameters defining the gas exchange. The
time for a complete exchange of the gas volume inside the chamber – assuming that
the inlet gas flow rate is equal to the outlet gas flow rate – is simply given by:
VC
Tge D ; (4.7)
VPin
where Tge is the gas exchange time of the measuring chamber, VC is the volume of
measuring chamber, and VPin is the inlet gas flow rate.
For typical values (VC D 140 mm3 , VPin D 20 l= min), we obtain: Tge D 420 s.
This is the minimum necessary time required to completely exchange the gas
volume inside the measuring chamber. This time is greater than the local exchange
time estimated with relation (3.1) in Chap. 3. With an effective cross section of
the chamber of AC , the average gas flow velocity amounts to vf D VPin =AC . With
AC D 30 mm2 , this velocity is 11 m/s. A design goal of the sample stands described
in the following was to minimize VC and AC for short gas exchange times and high
gas flow velocities in the interaction region.
Figure 4.18 shows a sectional view of a sample stand integrated in a LIBS setup
for the analysis of light elements in steel samples requiring the observation of
emission lines in the vacuum ultraviolet spectral range (<200 nm). The plasma
is observed in side-on direction, and the detection angle ˛D amounts to 10ı . The
sample stand and the tube between the sample stand and the entrance window of the
spectrometer are purged with argon gas of high purity. Typical gas flow rates are in
the range of 10–20 l/min.
Due to the small detection angle radiation originating from the plasma region
being in contact with the surface of the sample is partly shielded by the crater formed
Fig. 4.18 Sectional view of a sample stand with side-on observation of the emitted radiation
integrated in a LIBS setup. VS D vacuum spectrometer, SE D signal electronics, L D laser,
LB D laser beam, PR D Glan-laser prism, BE D beam expansion, M D mirror, FL D focusing
lens, ST D argon-flushed sample stand, Ar D argon inlet, W1 D window of ST, P D laser-induced
plasma, Ar OL D argon outlet, ˛D D detection angle, OA D optical axis of VS, W2 D entrance
window of VS, E D entrance slit of VS
Fig. 4.19 Sample stand of Fig. 4.18 mounted at a Paschen–Runge spectrometer
thus reducing the portion of continuum radiation reaching the spectrometer. With
such a configuration, detection limits for carbon, phosphorus, and sulfur in low-
alloyed steel samples below 10 g=g were achieved, cf. Sect. 13.1 [4.67].
Figure 4.19 shows the sample stand of Fig. 4.18 mounted at a Paschen–Runge
spectrometer. A relative motion between the laser beam and the sample can be
achieved by a displacement of the sample table (upper part of the sample stand
in contact with the sample) in the plane perpendicular to the laser beam. By this a
4.6 Sample Stand and Measuring Chamber 69
Fig. 4.20 Sectional view of a sample stand with end-on observation of the emitted radiation
integrated in a LIBS setup. Other components as in Fig. 4.18
Fig. 4.21 Sample stand of Fig. 4.20 mounted at a Paschen–Runge spectrometer. For abbreviations
see Fig. 4.18
spatial averaging is realized, which is of interest for the analysis of inhomogeneous

samples, e.g., slag samples, see Sects. 3.6 and 14.2.
Figure 4.20 shows a sectional view of a sample stand with end-on observation of
the laser-induced plasma. Such a configuration is advantageous if samples have to
be analyzed having an oxide layer at the surface which is not representative for the
bulk composition, cf. Sect. 13.1.2. In this case, deeper craters have to be formed and
hence a greater part of the plasma lies inside the crater volume. In order to detect a
sufficient part of the line emission, an end-on observation is chosen. In this case, the
detection angle (cf. Fig. 4.18) amounts to ˛D D 78ı .
A 3D-CAD model of the sample stand is shown in Fig. 4.21 [4.68]. The gas
volume inside the sample stand is minimized to enhance the gas exchange rate, cf.
Sect. 3.1.
References
4.1. J. Millard, R. Dalling, L. Radziemski, Time-resolved laser-induced breakdown spectrometry

for the rapid determination of beryllium in beryllium-copper alloys. Appl. Spectrosc.
40, 491–494 (1986)
4.2. F. Leis, W. Sdorra, J. Ko, K. Niemax, Basic investigations for laser microanalysis: I. optical
emission spectrometry of laser-produced sample plumes. Mikrochim. Acta II 98, 185–199
(1989)
4.3. I. Whiteside, R. Jowitt, Laser liquid metal analysis, in Progress of Analytical Chemistry in
the Iron and Steel Industry, ed. by R. Nauche (Luxembourg Office for Official Publications
of the European Communities, Luxembourg, 1992), pp. 135–141, ISBN 92-826-3877-4,
EUR 14113
4.4. J. Aguilera, C. Aragon, J. Campos, Determination of carbon content in steel using laser-
induced breakdown spectrometry. Appl. Spectrosc. 46, 1382–1387 (1992)
4.5. M. Sabsadi, P. Cielo, Quantitative analysis of aluminum alloys by laser-induced breakdown
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D Appl. Phys. 28, 2181–2187 (1995)
4.7. G. Arca, A. Ciucci, V. Palleschi, S. Rastelli, E. Tognoni, Trace element analysis in water by
laser-induced breakdown spectroscopy techniques. Appl. Spectrosc. 51, 1102–1105 (1997)
4.8. A. Pichahchy, D. Cremers, M. Ferris, Elemental analysis of metals under water using laser-
induced breakdown spectroscopy. Spectrochim. Acta B 52, 25–39 (1997)
4.9. L. St-Onge, M. Sabsabi, Towards quantitative depth-profile analysis using laser-induced
plasma spectroscopy: investigation of galvanealed coatings on steel. Spectrochim. Acta B
55, 299–308 (2000)
4.10. A. Knight, N. Scherbarth, D. Cremers, M. Ferris, Characterization of laser-induced
breakdown spectroscopy (LIBS) for application to space exploration. Appl. Spectrosc. 54,
331–340 (2000)
4.11. C. Geertsen, A. Briand, F. Chartier, J. Lacour, P. Mauchien, S. Sjöström, Comparison
between infrared and ultraviolet laser ablation at atmospheric pressure – implications for
solid sampling inductively coupled plasma spectrometry. J. Anal. At. Spectrom. 9, 17–22
(1994)
4.12. W. Ho, C. Ng, N. Cheung, Spectrochemical analysis of liquids using laser-induced plasma
emissions: effects of laser wavelength. Appl. Spectrosc. 51, 87–91 (1997)
4.13. H. Zhang, F. Yueh, J. Singh, Laser-induced breakdown spectrometry as a multimetal
continuous-emission monitor. Appl. Optics 38, 1459–1466 (1999)
4.14. D. Romero, J. Romero, J. Laserna, Distribution of metal impurities in silicon wafers
using imaging-mode multi-elemental laser-induced breakdown spectrometry. J. Anal. At.
Spectrom. 14, 199–204 (1999)
4.15. R. Neuhauser, U. Panne, R. Niessner, Utilization of fiber optics for remote sensing by laser-
induced plasma spectroscopy (LIPS). Appl. Spectrosc. 54, 923–927 (2000)
4.16. P. Fichet, P. Mauchien, J. Wagner, C. Moulin, Quantitative elemental determination in water
and oil by laser-induced breakdown spectroscopy. Anal. Chim. Acta 429, 269–278 (2001)
4.17. L. Berman, P. Wolf, Laser-induced breakdown spectroscopy of liquids: aqueous solutions of
nickel and chlorinated hydrocarbons. Appl. Spectrosc. 52, 438–443 (1998)
References 71
4.18. J. Simeonsson, A. Miziolek, Spectroscopic studies of laser-produced plasmas formed in CO

and CO2 using 193, 266, 355, 532 and 1064 nm laser radiation. Appl. Phys. B 59, 1–9 (1994)
4.19. C. Geertsen, J. Lacour, P. Mauchien, L. Pierrard, Evaluation of laser ablation optical
emission spectrometry for microanalysis in aluminium samples. Spectrochim. Acta B 51,
1403–1416 (1996)
4.20. M. Guillong, I. Horn, D. Günther, Capabilities of a homogenized 266 nm Nd:YAG laser
ablation system for LA-ICP-MS. J. Anal. At. Spectrom. 17, 8–14 (2002)
4.21. L. St-Onge, V. Detalle, M. Sabsabi, Enhanced laser-induced breakdown spectroscopy using
the combination of fourth-harmonic and fundamental Nd:YAG laser pulses. Spectrochim.
Acta B 57, 121–135 (2002)
4.22. A. Semerok, C. Chaléard, V. Detalle, J. Lacour, P. Mauchien, P. Meynadier, C. Nouvellon,
B. Sallé, P. Palianov, M. Perdrix, G. Petite, Experimental investigations of laser ablation
efficiency of pure metals with femto, pico and nanosecond pulses. Appl. Surf. Sci. 138–139,
311–314 (1999)
4.24. K. Kagawa, K. Kawai, M. Tani, T. Kobayashi, XeCl excimer laser-induced shock wave
plasma and its application to emission spectrochemical analysis. Appl. Spectrosc. 48, 198–
205 (1994)
4.25. N. André, C. Geertsen, J. Lacour, P. Mauchien, S. Sjöström, UV laser ablation optical
emission spectrometry on aluminium alloys in air at atmospheric pressure. Spectrochim.
Acta 49B, 1363–1372 (1994)
4.26. N. Arnold, J. Gruber, J. Heitz, Spherical expansion of the vapor plume into ambient gas: an
analytical model. Appl. Phys. A 69, S87–S93 (1999)
4.27. Z. Hwang, Y. Teng, K. Li, J. Sneddon, Interaction of a laser beam with metals, Part I:
Quantitative studies of plasma emission. Appl. Spectrosc. 45, 435–441 (1991)
4.28. R. Knopp, F. Scherbaum, J. Kim, Laser-induced breakdown spectroscopy (LIBS) as an
analytical tool for the detection of metal ions in aqueous solutions. Fresenius J. Anal. Chem.
355, 16–20 (1996)
4.29. W. Koechner, Solid-State Laser Engineering, 6th ed., Springer Series in Optical Sciences,
vol. 1 (Springer, Berlin, 2006), XVI, p. 748
4.30. R. Diehl (ed.), High-Power Diode Lasers – Fundamentals, Technology, Applications
(Springer, Berlin, 2000), p. 416
4.31. P. Peuser, N. Schmitt, Diodengepumpte Festkörperlaser (Springer-Verlag, Berlin, 1995)
4.32. D. Hoffmann, G. Bonati, P. Kayser, P. Loosen, R. Poprawe, R. Dinger, C. Haas, H. Martinen,
Modular, fiber coupled, diode-pumped solid state laser with up to 5 kW average output
power, in Proceedings of Advanced Solid-State Lasers, OSA, Seattle, 2001, 33–35
4.33. K. Du, D. Li, H. Zhang, P. Shi, X. Wei, R. Diart, Electro-optically Q-switched Nd:YVO4
slab laser with high repetition rate and short pulse width. Opt. Lett. 28, 87–89 (2003)
4.34. A. Thorne, U. Litzen, S. Johansson, Spectrophysics (Springer-Verlag, Berlin, 1999)
4.35. J. Samson, Techniques of Vacuum Ultraviolet Spectroscopy (Pied, Lincoln, 1967)
4.36. H. Becker-Ross, S. Florek, Echelle spectrometers and charged-coupled devices. Spec-
trochim. Acta B 52, 1367–1375 (1997)
4.37. Echelle spectrometer 4000 EV/I from LLA Instruments GmbH, Berlin. http://www.lla.de
4.38. C. Aragon, J. Aguilera, F. Penalba, Improvements in quantitative analysis of steel compo-
sition by laser-induced breakdown spectroscopy at atmospheric pressure using an infrared
Nd:YAG laser. Appl. Spectrosc. 53, 1259–1267 (1999)
27, 805–810 (1999)
4.40. M. Sabsabi, V. Detalle, M. Harith, W. Tawfik, H. Imam, Comparative study of two new
commercial echelle spectrometers equipped with intensified CCD for analysis of laser-
induced breakdown spectroscopy. Appl. Optics 42, 6094–6098 (2003)
4.41. H. Balzer, S. Hölters, V. Sturm, R. Noll, Systematic line selection for online coating
thickness measurements of galvanised sheet steel using LIBS. Anal. Bioanal. Chem. 385,
234–239 (2006)
4.42. V. Sturm, A. Brysch, R. Noll, H. Brinkmann, R. Schwalbe, K. Mülheims, P. Luoto,
P. Mannila, K. Heinänen, D. Carrascal, L. Sancho, A. Opfermann, K. Mavrommatis, H.W.
Gudenau, A. Hatziapostolou, S. Couris, Online multi-element analysis of the top gas of a
blast furnace, in Seventh International Workshop Progress in Analytica Chemistry in the
Steel and Metal Industries, ed. by J. Angeli (Glückauf GmbH, Essen, 2006), pp. 183–188
4.44. R. Noll, I. Mönch, O. Klein, A. Lamott, Concept and performance of inspection machines
for industrial use based on LIBS. Spectrochim. Acta B 60, 1070–1075 (2005)
4.45. C. Bohling, K. Hohmann, D. Scheel, W. Schade, M. Reuter, G. Holl, Anti-personnel-mine
detection by laser-induced breakdown spectroscopy, in Conf. on Lasers and Electro-
Optics/Quantum Electronics and Laser Science and Photonic Applications, Systems and
Technologies, Technical Digest (CD), Optical Society of America, 2005, paper PWD3
4.46. K. Yamamoto, D. Cremers, M. Ferris, L. Foster, Detection of metals in the environ-
ment using a portable laser-induced breakdown spectroscopy instrument. Appl. Spectrosc.
50, 222–233 (1996)
4.47. F. Akamatsu, T. Wakabayashi, S. Tsushima, M. Katsuki, Y. Mizutani, Y. Ikeda, N. Kawahara,
T. Nakaiima, The development of a light-collection probe with high spatial resolution
applicable to randomly fluctuating combustion fields. Meas. Sci. Technol. 10, 1240–1246
(1999)
4.48. S. Palanco, J. Laserna, Remote sensing instrument for solid samples based on open-path
atomic emission spectrometry. Rev. Sci. Instrum. 75, 2068–2074 (2004)
4.49. P. Shaw, R. Gupta, T. Germer, U. Arp, T. Lucatorto, K. Lykke, Characterization of materials
using an ultraviolet radiometric beamline at SURF III. Metrologia 37, 551–554 (2000)
4.50. K. Knorr, Fluoridkristalle – alte Materialien mit glänzender Zukunft, Photonik 1/2002,
p. 62–64
4.51. Transmittance data for MgF2 from Korth Kristalle GmbH, Am Jägersberg 3, D-24161
Altenholz (Kiel), http://www.korth.de
4.52. L. Merkle, N. Koumvakalis, M. Bass, Laser-induced bulk damage in SiO2 at 1.064, 0.532,
and 0.355 m. J. Appl. Phys. 55, 772–775 (1984)
4.53. F. Rainer, R. Gonzales, A. Morgan, in Laser-Induced Damage in Optical Materials, ed. by
H. Bennett, L. Chase, A. Guenther, B. Newnam, M. Soileau, Proc. SPIE 1438, 58–73 (1989)
4.54. N. Boling, G. Dubé, Laser-induced inclusion damage at surfaces of transparent dielectrics.
Appl. Phys. Lett. 23, 658–660 (1973)
4.55. R. Taylor, K. Leopold, S. Mihailov, R. Brimacombe, Damage measurements of fused silica
fibres using long optical pulse XeCl lasers. Opt. Commun. 63, 26–31 (1987)
4.56. R Setchell, K. Meeks, W. Trott, P. Klingsporn, D. Berry, High-power transmission through
step-index multimode fibers, in Laser-Induced Damage in Optical Materials: 1990, ed. by
H. Bennett, L. Chase, A. Guenther, B. Newnam, M. Soileau, Proc. SPIE 1441 (1991) 61–70
4.57. W. Trott, K. Meeks, High-power Nd:glass laser transmission through optical fibers and its
use in acceleration of thin foil targets. J. Appl. Phys. 67 (1990) 3297–3301
4.58. V. Sturm, R. Sattmann, R. Noll, Optical fiber transmission of multiple Q-switch Nd:YAG
laser pulses with microsecond interpulse separations. Appl. Phys. B 63, 363–370 (1996)
4.59. V. Pashinin, V. Sturm, V. Tumorin, R. Noll, Stimulated Brillouin scattering of Q-switched
laser pulses in large-core optical fibers. Opt. Laser Technol. 33, 617–622 (2001)
4.60. A. Whitehouse, J. Young, I. Botheroyd, S. Lawson, C. Evans, J. Wright, Remote material
analysis of nuclear power station steam generator tubes by laser-induced breakdown
spectroscopy. Spectrochim. Acta B 56, 821–830 (2001)
4.61. Hamamatsu data sheets for the PMT types R 6350, R 6351, R 6354, R 7511, 1998
4.62. W. Franks, M. Kiik, A. Nathan, UV-responsive CCD image sensors with enhanced inorganic
phosphor coatings. IEEE Trans. Electron Dev. 20, 352–358 (2003)
References 73
4.63. L. Poletto, A. Boscolo, G. Tondello, Characterization of a charge-coupled-device detector

in the 1100 – 0.14 nm (1-eV to 9-keV) spectral region. Appl. Optics 38, 29–36 (1999)
4.64. T. Jüstel, C. Feldmann, C. Ronda, Leuchtstoffe für aktive Displays. Phys. Blätter 56, 55–58
(2000)
4.65. Princeton instruments data sheets for intensified spectroscopy cameras types HB GEN III,
UNIGEN and HQ GEN III, 2007
4.66. H. Bette, R. Noll, High-speed, high-resolution LIBS using diode-pumped solid-state lasers,
in Laser-Induced Breakdown Spectroscopy, ed. by A.W. Miziolek, V. Palleschi, I. Schechter
(Cambridge University Press, Cambridge, 2006), Chapter 14, pp. 490–515
4.68. R. Noll, V. Sturm, Ü. Aydin, D. Eilers, C. Gehlen, M. Höhne, A. Lamott, J. Makowe,
J. Vrenegor, LIBS – from research to industry, new frontiers for process control. Spec-
trochim. Acta B 63, 1159–1166 (2008)
Chapter 5
Evaporation and Plasma Generation
5.1 Evaporation
For LIBS of solid samples, the laser radiation evaporates the material and excites
spontaneous emission of the material species (cf. Chap. 2). The material is heated
locally to boiling or decomposition temperature to achieve an efficient evaporation.
The fraction of laser energy absorbed by the material contributes to this process.
The amount of temperature rise is influenced by the heat conduction of the material,
which transports the absorbed energy out of the region, where the laser energy is
coupled into the material.
The characteristics of the interaction between laser radiation and solid sample
are illustrated in Fig. 5.1. In the following paragraphs, two cases are considered.
(a) Surface absorption (ıopt ıh /
The optical penetration depth is much smaller than the heat penetration depth. For
metals and laser wavelengths in the visible and near infrared spectral region, ıopt
amounts to 106 –105 cm typically. The condition ıopt ıh is, e.g., fulfilled for
steel and times t > 109 s after start of the laser irradiation (the thermal diffusivity
of steel at room temperature is steel D 0:04 cm2 =s). In this case, the source term
in the heat equation can be described as a surface source. For a Gaussian intensity
profile of the laser beam, the source term is:
2r 2

2
Q.r; t/ D AI e wb ;
0 (5.1)
where A is the absorptance, I0 is the irradiance of the laser beam in the beam center
at r D 0, and wb is the beam radius, cf. Fig. 3.10.
In this case, a solution of the heat equation yields the temperature at the surface
and the beam center as a function of time [5.1]:
p
AI 0 wb 8t
T D p arctan ; (5.2)
2c wb

76 5 Evaporation and Plasma Generation
L L
MS PS
OPD OPD
HPD HPD
Fig. 5.1 Left: the optical penetration depth is smaller than the heat penetration depth; right: the
optical penetration depth is greater than the heat penetration depth. L D laser beam, MS D metal
sample, OPD D optical penetration depth, HPD D heat penetration depth, PS D polymer sample
where T is the temperature at the sample surface in the center of the laser beam,
is the thermal diffusivity, is the mass density, c is the specific heat capacity, and t
is the time.
For a given absorbed irradiance, (5.2) yields the time at which the temperature at
the sample surface attains, e.g., the boiling temperature and efficient evaporation
sets in. Figure 5.2 shows the absorbed irradiance as a function of time for steel
and aluminum according to (5.2). The heat conductivity c and the thermal
diffusivity are also functions of the temperature. To simplify the calculation of
these curves, average values were taken for the conductivity and diffusivity in
the temperature range from 300 K to the evaporation temperature Tv (the used
parameters are: (a) steel, average heat conductivity hci D 0:27 W=.cm K/,
average thermal diffusivity hi D 0:06 cm2 =s, T D Tv D 2;900 K; (b) aluminum,
hci D 1:79 W=cm K, hi D 0:64 cm2 =s, Tv D 2;700 K; the beam radius amounts
in both cases wb D 100 m). The absorbed irradiance necessary to reach evaporation
temperature decreases with longer irradiation times. Depending on the absorption
coefficient, the laser beam irradiance at the sample surface has to be chosen
sufficiently high to reach the evaporation threshold. For steel and aluminum, the
absorption coefficient has typically values between 0.03 and 0.1. For Q-switched
lasers with pulse durations of 10–100 ns, the necessary irradiances are greater
than 108 W=cm2 to evaporate steel or aluminum according to Fig. 5.2. The area of
irradiances accessible with commercially available Q-switched Nd:YAG lasers is
shown as hatched box in Fig. 5.2.
(b) Volume absorption (ıopt ıh /
The optical penetration depth is much greater than the heat penetration depth.
This is, e.g., the case for polymers. The optical penetration depth for polymers as
polyamide amounts to about 3.5 mm for the wavelength of the Nd:YAG laser [5.2].
With a thermal diffusivity of polyamide D 1:3 103 cm2 =s [5.3], (3.2) yields that
the optical penetration depth is greater than the heat penetration depth for times
t < 20 s.
5.1 Evaporation 77
1011 1011
1010 1010
absorbed irradiance AI0, (1-R)I0 [W/cm2]
Q- switched
Nd:YAG-laser
109 109
irradiance I0 [W/cm2]
polyamide
108 108
107 107
aluminum
steel
106 10
6
105 105
10-10 10-9 10-8 10-7 10-6 10-5 10-4
time [s]
Fig. 5.2 Absorbed irradiance as a function of time necessary to reach evaporation or decomposi-
tion temperature for various materials. The hatched box marks the area of irradiances accessible
with commercial Nd:YAG lasers. The parameters used for calculation of the curves are given in
the text
If the heat conduction is negligible, then the temperature rise can be calculated
directly for the locally absorbed laser irradiance. For the irradiance in the absorbing
medium holds: z

I D .1 R/I0 e ıopt ; (5.3)
where R is the reflectivity, I0 is the irradiance at the sample surface, and z is the
coordinate in the propagation direction of the laser beam with the sample surface at
z D 0.
The factor .1 R/ in (5.3) describes that part of the incoming irradiance,
which enters the sample. In contrast to the case (a) described above for the surface
absorption, the absorption A cannot be taken here, since in the considered case
of volume absorption the absorption is not limited to the surface alone. For the
absorbed energy per unit volume after a time t, we obtain from (5.3):
z
I0 ıopt
.dI =dz/t D .1 R/ e t: (5.4)
ıopt
This energy heats up the volume element leading to a temperature rise of T :
q D cT; (5.5)
where q is the heat energy per unit volume and T is the temperature rise.
With .dI =dz/t D q and the relations (5.4), (5.5) follows at the surface of the
workpiece .z D 0 mm/:
.1 R/I0 D cT ıopt =t: (5.6)
If we take for T the temperature difference between the decomposition or
evaporation temperature of the polymer and the initial temperature of the workpiece,
then (5.6) yields the necessary irradiance as a function of the irradiation time.
Figure 5.2 shows such a curve calculated for polyamide. The necessary irradiances
are significantly higher than for metals.
In the following paragraphs, we estimate the amount of material evaporated by
a laser pulse. A simplified energy balance is considered assuming that the absorbed
laser energy is contributing exclusively to the evaporation of the material. This
assumption neglects, e.g., the following terms: losses by heat conduction; specific
melting enthalpy; melt transport; recondensation of vapor; a part of the laser energy
contributes to the expansion of the material cloud, the acceleration of the ambient
gas and the formation of shock waves; a part is scattered and reflected by the vapor
and the plasma; a part contributes to the ionization and excitation of the species
in the plasma. Hence, the simplified energy balance yields an upper limit of the
evaporated mass:
AW Laser D V .Ev C cT /; (5.7)
where A is the absorptance, WLaser is the irradiated energy of the laser pulse, V is the
ablated volume, Ev is the specific evaporation enthalpy, and T is the temperature
difference between boiling temperature and room temperature.
For simplification, we furthermore neglect that the quantities A; ; c are gen-
erally not constant during the heating and evaporation phase. The values of the
initial state were taken instead. As numerical example (5.7) yields as upper limit
of the mass ablated by a laser pulse with AW Laser D 10 mJ: V D 1:6 106 g
(Ev D 6 103 J=g, c D 0:51 J=g K, T D 2;600 K). Experimentally determined
mass ablations lie in the range of micrograms per pulse depending on the pulse
energy, pulse width, burst mode, and target material, cf. Chap. 7.
5.2 Plasma Generation
The density of the evolving vapor can be estimated by use of the Clausius–
Clapeyron equation [5.4]. For orientation solely some typical numbers will be
indicated. Absorbed laser irradiances of 106 –107 W=cm2 result in vapor densities
in the range of about 1018 –1020 cm3 . Due to the high temperature of the vapor,
there always exist some free electrons, which can gain energy from the radiation
field and transfer this energy via collisions to the atoms of the vapor. Collisions
of electrons with atoms are a precondition for a net energy gain of the electrons
from the oscillating electric field of the laser beam. Collision-free electrons would
take up and loose energy periodically and hence on average would not gain any
kinetic energy. Within a short time, the electrons get high average energies, which
5.3 Absorption of Laser Radiation in the Plasma 79
are sufficient to partly ionize the atoms of the material vapor. In this way, the plasma
is induced, in which the atoms and ions are excited and emit their specific radiation.
The state of the plasma can be described with the use of balance equations for
particle densities, momenta, and energies. Model calculations show that within a
few nanoseconds after the start of the laser beam irradiation, the electron energies
reach values in the range of 0.25–1 eV [5.5]. Due to collisional ionization processes,
the electron density rises to a level of 1016 –1018 cm3 .
5.3 Absorption of Laser Radiation in the Plasma
The absorption of the laser radiation in the evolving material vapor and plasma
depends on the wavelength, which will be discussed in the following paragraphs in
more detail. The absorption coefficient can be calculated by means of the Maxwell
equations and the equation of motion for free nonrelativistic electrons [5.6]:
v 8
u s
u < ! 2 2 2 2
u1 p 1 !p 1
u2 1 C
u : ! 1 C .=!/2 ! ! 1 C .=!/2
˛ D 2k u (5.8)
u
t !p 2 1
1 ;
! 1 C .=!/2
where k D !=c is the wave vector in vacuum, ! is the angular frequency of the
laser radiation, c is the vacuum speed of light, !p is the plasma frequency, and is
the collision frequency of electrons with ions and atoms.
The plasma frequency is the characteristic eigenfrequency of the plasma describ-
ing the oscillation of the electrons with respect to the ions [5.7]:
s
ne e 2
!p D ; (5.9)
"0 me
where ne is the electron density, e is the elementary charge, "0 is the vacuum
permittivity, and me electron mass.
Taking the typical conditions of a laser-induced plasma for LIBS allows to
simplify the relation (5.8) under the following assumptions:
!; !p < !: (5.10)
For the absorption coefficient holds:
!p2
˛D p : (5.11)
c ! 2 1 .!p =!/2
The absorption coefficient is proportional to the collision frequency and approxi-

mately inversely proportional to the square of the frequency of the laser light. The
collision frequency comprises two terms:
D ei C en ; (5.12)
where ei is the collision frequency of electrons with ions, and en is the collision
frequency of electrons with atoms.
In the following paragraphs, the LIBS plasma of iron samples is taken as a
numerical example. The electron atom collision frequency for iron at a temperature
of 0.8 eV can be estimated to be ven D .5 1014 m3 =s/nFe I [5.8]. Typical values
of the iron atom density determined experimentally at a time of 1 s after the
irradiation of a double pulse are: nFe I 1016 cm3 [5.9]. Then we obtain for the
electron iron atom collision frequency: en 5 108 s1 . The electron ion collision
frequency is given by [5.10]:
p
1 4 2Z 2 e 4 ni ln ƒ
ei D p ; (5.13)
.4"0 /2 3 me .kB Te /3=2
where "0 ; e; me see (5.9), Z is the ion charge, ni is the density of ions, ln
is the Coulomb logarithm, kB is the Boltzmann constant, and Te is the electron
temperature.
For the Coulomb logarithm holds [5.10]:
3=2
3=2 .kB Te /
ln D ln 12"0 p : (5.14)
e 3 ne
With the experimental values (1 s after irradiation of a double pulse, cf. Chap. 8):
ne D ni D nFe II D 6 1016 cm3 , Te D 9;200 K follows: ln D 3:8 and ei D 9:4
1011 s1 . Hence, the collision frequency of the electrons with the ions is significantly
greater than the collision frequency between electrons and neutral particles. For the
plasma frequency follows in this case using relation (5.9): !p D 1:41013 rad=s. For
a Nd:YAG laser with a wavelength of 1:06 m, the angular frequency is ! D 1:78
1015 rad=s. Hence, the conditions (5.10) are fulfilled to apply the approximation
(5.11) to calculate the absorption coefficient (This estimation holds for tdelay D 1 s;
at earlier times being closer to the irradiation time of the laser pulse the situation is
completely different).
Figure 5.3 shows the absorption coefficient as a function of the electron density.
The temperatures given cover the range typically observed for LIBS plasmas where
the temperature is measured by use of Boltzmann plots (cf. Chap. 8). The range of
measured plasma parameters ne and Te is illustrated as hatched box (cf. Chap. 8
and, e.g., [5.11]). Assuming as an upper limit a plasma dimension of 5 mm, the
right axis shows the corresponding transmission of the laser radiation through the
plasma. Hence, for plasma conditions prevailing typically some 100 ns after the laser
References 81
103 0.007
102 0.607
transmission e-αdP for dp = 5 mm [1]

absorption coefficient α [m-1]
7000 K
101 8500 K 0.951
10000 K
20000 K
100 0.995
10-1 0.9995
10-2 0.99995
10-3 measured ne, Te at tdel= 1 μs 1
10-4
1015 1016 1017 1018 1019
electron density [cm-3]
Fig. 5.3 Absorption coefficient for laser radiation with a wavelength of 1:06 m in a plasma as
a function of the electron density. Parameter is the electron temperature of the plasma. The right
axis shows the transmission of the laser radiation through a plasma having a thickness of 5 mm.
The hatched box illustrates the typical range of electron densities and temperatures measured for
LIBS plasmas at a delay time of 1 s after irradiation of the laser pulse
pulse irradiation, the plasma is already highly transparent for the laser radiation of
a Nd:YAG laser. Therefore, a second collinear pulse of a double pulse irradiated
with an interpulse separation of t > 1 s reaches the surface of the measuring
object again with negligible absorption by the plasma generated by the first pulse
(cf. Chap. 8).
References
5.1. H. Carslaw, J. Jaeger, Conduction of Heat in Solids, 2nd edn. (Oxford University Press,
Oxford, 1959)
5.2. R. Klein, Bearbeitung von Polymerwerkstoffen mit infraroter Laserstrahlung, D82, Disser-
tation, Aachen, 1990
5.3. VDI-Wärmeatlas, 4. Auflage, Düsseldorf, 1984
5.4. R. Wester, Laser-induced evaporation as basic process of drilling, caving and cutting, Laser
und Optoelektronik 23, 60–63 (1991)
5.5. E. Beyer, Einfluß des laserinduzierten Plasmas beim Schweißen mit CO2 -Lasern,
Schweißtechnische Forschungsberichte, Band 2, Düsseldorf, Deutscher Verlag für
Schweißtechnik, 1985
5.6. P. Mulser, Hydrogen plasma production by giant pulse lasers. Z. Naturforsch. 25a, 282–295
(1970)
5.7. F. Chen, Introduction to Plasma Physics (Plenum, New York, 1974), ISBN 0–306–30755–3
5.8. M. Aden, Plasmadynamik beim laserinduzierten Verdampfungsprozeß einer ebenen Metall-
oberfläche, D82, Dissertation, Aachen, 1994
5.9. R. Sattmann, Spektrochemische Analyse von Stahl durch optische Emissionsspektrometrie
laserinduzierter Plasmen, D82, Dissertation, Aachen, 1997
5.10. A. Anders, A Formulary for Plasma Physics (Akademie-Verlag, Berlin, 1990), ISBN 3–05–
501263–1
5.11. L. Radziemski, T. Loree, D. Cremers, N. Hoffman, Time-resolved laser-induced breakdown
spectrometry of aerosols. Anal. Chem. 55, 1246–1252 (1983)
Chapter 6
Multiple Pulses for LIBS
Laser-induced breakdown spectroscopy was investigated already for a broad variety

of chemical analytical purposes in the 1970s to 1990s [6.1–6.6]. However, a
transfer to applications was often prevented because of insufficient analytical
performance compared with other atomic emission spectrometric methods such as
spark discharge or inductively coupled plasma (ICP) atomic emission spectroscopy.
Thus, at that time LIBS was used in a limited number of applications only, in
which the analytical requirements are low and the advantages of LIBS as a rapid,
noncontact, in situ method are essential [6.7].
To use LIBS in a broader field of applications, the analytical performance has to
be raised considerably. A crucial point is the laser parameters and the interaction of
the laser beam with the material to be analyzed. Most of the work reported is done
with pulsed pumped Q-switch Nd:YAG lasers, although other lasers have also been
used for LIBS (cf. Sect. 4.1). Single pulses, i.e., one laser pulse per pump pulse, are
the conventional approach to ablate and vaporize material and to induce the plasma.
Typical laser parameters are 10–100 Hz repetition rate, 8–50 ns pulse width, and
5–200 mJ pulse energy.
If single pulses are used for LIBS, then ablation and plasma excitation cannot
be optimized separately, in contrast to two-step techniques, such as laser ablation
with subsequent excitation by electrical sparks, microwaves, or inductively coupled
plasmas [6.8–6.10]. These two-step techniques require additional equipment close
to the interaction area. Spark excitation of the laser ablated material, for example,
requires electrodes located close to the samples. These reduce the flexibility and
may interfere with the analysis.
Uebbing et al. used a second Q-switch Nd:YAG laser to re-heat the vaporized
material [6.11]. The second laser was fired with controlled delay respective to the
first laser. The direction of the first laser beam was perpendicular to the sample
surface, whereas the second laser beam propagated parallel to the surface. Hence,
the second laser beam interacted only with the vapor above the sample. This
arrangement may be adequate for fundamental studies, but it is hardly suited for
a practical application of LIBS as an analytical method.

84 6 Multiple Pulses for LIBS
The use of controlled multiple laser pulses combines the advantages of a two-
step techniques without losing the flexibility of one-step LIBS. Multiple pulses
have been used in former approaches as well, but the separate pulses could not
be controlled actively [6.12, 6.13]. For that reason and due to worse pulse-to-pulse
stability compared with single pulses, these authors recommended single pulses for
LIBS. In the mid-nineties, a pioneering work demonstrated clearly the benefits of
actively controlled collinear multiple pulses for LIBS [6.14]. Since then double
pulses and multiple pulses were studied to improve the analytical performance of
LIBS – e.g., for the determination of traces in metallic matrices or to enable special
applications, e.g., the analysis of samples under water – or to increase the ablation
rate in laser material processing [6.15–6.19].
6.1 Double Pulses
The term “double pulse” is defined here as follows: two laser pulses having each a
width in the range of about 5–30 ns, which are separated in time by several 100 ns
up to several microseconds (cf. Sect. 3.2). This interpulse separation is denoted by
t (Fig. 3.9). Several methods are available to generate these double pulses: (a) a
single-pulsed laser with a modified electronic control of the Q-switch to generate
two pulses within a single flashlamp discharge, (b) superposition of the beams of
two pulsed lasers. For case (a), only one laser is necessary and the two pulses are
emitted collinearly. However, the interpulse separation is limited to about 0:5 s <
t < 180 s. Case (b) requires two lasers and a precise adjustment of the spatial
superposition of the two pulses. This approach offers a high flexibility in terms of
the wavelengths of the pulses, the pulse widths, and the sequence of pulses.
Figure 6.1 illustrates schematically the various configurations studied so far for
LIBS with double pulses. The arrows depict the direction of propagation of the laser
pulses, and the numbers show the temporal sequence. In the collinear case, both
pulses have the same axis of propagation and are directed orthogonally to the sample
1. collinear 2. orthogonal pre-ablation

2
2
1
1
Fig. 6.1 Double-pulse

configurations studied for 3. orthogonal re-heating 4. dual pulse crossed beam
LIBS. The arrows depict the 1 2 2
laser pulses and their
2
direction of propagation and 1 1
the numbers 1, 2 their
temporal sequence
6.1 Double Pulses 85
surface [6.14–6.16]. Various pulse combinations in terms of different wavelengths

and different pulse widths have been studied [6.18, 6.19]. In the case of orthogonal
preablation, a first laser pulse is irradiated parallel to the sample surface generating
a laser-induced plasma above the sample surface in the ambient atmosphere and the
second pulse is directed orthogonally to the sample surface [6.20]. For orthogonal
re-heating, the sequence of the pulses is exchanged, i.e., the laser pulses irradiates
the sample at first and then the re-heating pulse propagating parallel to the sample
surface is applied [6.21]. In the dual pulse crossed beam approach, the first laser
pulse is splitted into two beams irradiating the sample under an angle of incidence
of 45ı , and the second pulse follows in the same directions [6.22].
Investigations of collinear double pulses and triple pulses generated within a
single flashlamp discharge of a Nd:YAG laser for plasma excitation showed a
significant improvement in the detection limits [6.14,6.16,6.23]. With this approach,
detection limits for the elements C, P, S, Mn, Ni, Cr in low-alloyed solid steel
samples of less than 10 g=g were achieved for the first time using LIBS. Collinear
double and multiple pulses led to a significant increase in the ablated mass [6.14].
Comparing the ablated mass of a steel target using double pulses with 2 40 mJ and
t D 6 s with a single pulse of 1 80 mJ yields an increase by about a factor of
2.5. While for single pulses the ablated mass strongly saturates with increasing pulse
energy, no such saturation behavior is observed for double and multiple pulses up
to burst energies (denoting the sum of pulse energies within a double or multiple
pulse) of 320 mJ. Electron temperatures in the laser-induced plasma determined
spectroscopically are higher by 1,000–1,500 K using double pulses with respect
to those measured in the plasmas generated by single pulses of the same energy
(i.e., the energy of the single pulse is equal to the total energy of the double pulse).
Hence, the increased line emission observed using double-pulse excitation is linked
with the greater mass ablated and the higher temperatures achieved in the laser-
induced plasma [6.14].
For collinear double pulses studied for the analysis of solid metallic samples
under water, an increase in the ablated mass of aluminum samples by a factor of
3.8 was reported [6.15]. The observed increase in line emissions is attributed to
the formation of a bubble in the water layer above the sample surface generated
by the first pulse into which the emitting atoms can expand. Due to the absence of
atomic emissions from the plasma produced by the single pulse, it was not possible
to determine the temperature and electron density for underwater excitation. The
authors conclude that in this case the temperature is less than 3,000 K. For double-
pulse excitation, an electron temperature of 8,900 K was determined [6.15].
Studies of collinear double pulses interacting with aluminum samples in air
showed higher line intensities lasting longer than corresponding emissions gener-
ated by single laser pulses [6.17]. The time evolution of the line emission follows
an exponential decay, where the decay time constant for the Al II line 624.34 nm
and double-pulse excitation is more than twice that determined for single-pulse
excitation. Looking at the ionic nitrogen emission N II at 500.1 nm revealed greater
line intensities in the case of single-pulse excitation than for double pulses indicating
that the preplasma formation induced by the first pulse reduces the air present nearby
the surface and hence the concentration of nitrogen available for emission [6.17].
In the orthogonal preablation configuration, an enhancement of Fe lines of about
a factor of 30 was observed comparing preablation double pulses with single pulses
[6.20]. A temperature increase of up to 5,000 K was observed for preablation double
pulses. The signal enhancement is correlated with an increased crater volume, i.e.,
the ablated mass is greater for preablation double pulses than for single pulses
[6.20].
From a practical point of view, the collinear configuration is the most application
relevant approach for double pulses offering an easy access to the sample to be
analyzed. Hence, the investigations presented in the following are focused on the
collinear double pulse configuration.
For a qualitative assessment of the influence of multiple pulses on the spatial
structure of the laser-induced plasma, the visible plasma emission was observed by
time-integrated photographs. For the experiments, the multiple Q-switch pulses are
generated by a Nd:YAG laser with one oscillator (300 4 mm rod) and two amplifiers
(300 1=400 and 400 00 3=800 rods). The high voltage of the Pockels cell can be increased
in up to six steps with variable step height and time delay. Each voltage step results
in the emission of a Q-switch pulse. The time delay and the energy of the pulses
are adjustable. The minimum delay between consecutive pulses is about 2 s; the
maximum duration of a burst of pulses is given by the flashlamp pump pulse and
amounts to about 100 s. The laser emits the fundamental wavelength of 1,064 nm
with a burst repetition rate of 30 Hz. The pulse duration and maximum energy per
burst depend on the number of pulses per burst and the delays (which determine the
interpulse separations t, cf. Fig. 3.9). The following notation is used: E s ; Ebd , or
Ebm denotes the total energy of a single-, double- or multiple-pulse burst, and E1d
or E2d denotes the energy of the first or second pulse of a double pulse, and so on.
The maximum burst energy is measured with a volume-absorbing disc calorimeter
(Scientech model 38-0101). The energy distribution of the separate pulses of a burst
is determined by integration of a photodiode signal of the burst. Figure 3.8 shows
the photodiode signal of the laser output for a double pulse with E1d D E2d , Ebd D
80 mJ and t D 6 s interpulse separation. For 6 s interpulse separation, the pulse
duration (FWHM) of each pulse of a double burst amounts to about 25 ns. The pulse
duration for single pulses is about 15 ns.
The laser beam is focused on plane samples by a planoconvex lens with a focal
length of f D 200 mm. The diameter of the laser beam in the plane of the lens
amounts to about 12 mm. The sample surface is adjusted perpendicularly to the
laser beam. All experiments are performed under air and a pressure of 105 Pa. For
each measurement, the laser beam is focused on a new part of a ground surface of
the sample. A pure iron sample and samples with certified composition from the
Bureau of Analysed Samples, Ltd., UK (BAS), are used for the experiments.
For the photographs of the laser-induced plasma, the direction of observation is
oriented perpendicularly to the propagation direction of the laser beam, see Fig. 6.2.
A constant exposure time of 1 s is taken for all photographs, so the plasma radiation
induced by 30 laser bursts is accumulated. The photographs indicate the extent of

the light-emitting plasma.
Figure 6.2 shows photographs of the plasmas generated by single laser pulses
of E s D 80 mJ and double pulses of the same total energy Ebd D 80 mJ with an
interpulse separation of t D 6 s. The photographs in the left-hand column are
taken without sample. The middle of the air breakdown plasma is taken as the zero
point of s. Positive s corresponds to a smaller distance between focusing lens
and sample, that is, the focal region moves into the sample. In the experiments,
different s are realized by a shift of the focusing lens. The photographs taken for
s D 5 and ˙2 mm show radiating volumes, which are approximately a factor of
two greater for double pulses than for single pulses.
Another important fact is obvious from the photographs with s D 5 mm: for
the 80 mJ single pulse, the plasma induced in the surrounding atmosphere shields
off a major part of the laser energy. With double pulses, the plasma induced in the
atmosphere does not prevent a considerable part of the laser energy from reaching
the sample, and an intense plasma is also generated on the sample surface.
The plasma radiation is collected by an optical fiber bundle (4 mm diameter,
about 160 all-silica fibers, 200 m diameter per fiber, bundle length 4 m) without
additional optical elements. The distance of the entrance face of the fiber bundle
to the sample surface is about 67 mm, and the angle to the laser beam about 39ı .
The numerical aperture of the fiber amounts to 0.22, so the received radiation is
spatially integrated over the emitting plasma volume. The plasma light is guided to
a Paschen–Runge spectrometer with 750 mm focal length, 2,400 lines per millimeter
Fig. 6.2 Time-integrated photographs of the plasmas in the visible spectral range induced in the
atmosphere and on a steel surface by single and double laser pulses of 80 mJ burst energy. The
distance between the focal plane of the laser beam and the surface of the sample is denoted by s,
see text for details
grating, and photomultiplier tubes as detectors. The spectral resolution amounts to

0.03 nm in first order.
Figure 6.3 shows the output signal of a photomultiplier tube mounted in the
Paschen–Runge spectrometer at the Fe II 273.07 nm line for single and double
pulses of 80 mJ (interpulse separation t D 6 s, s D 5 mm/. Each curve is
averaged over 16 laser bursts on a pure iron sample after 84 prebursts. An analysis of
the signal induced by the single pulses (Fig. 6.3, top) shows that the decay following
the maximum can be described by an exponential function for times up to 0:3 s
after the maximum of the signal. The 1/e time constant amounts to about 90 ns. This
section of the signal is attributed to the continuum emission of the laser-induced
plasma (cf. Sect. 2.1). The subsequent signal behavior is characterized by decay
times greater than 800 ns. Line radiation is the dominant contribution to this part of
the signal.
The temporal behavior of the plasma emissionindexContinuum!emission of the
plasma induced by the first pulse of the double pulse is similar to that of the single
pulse, whereas the intensity is significantly smaller due to the reduced laser pulse
energy. In contrast to this, the plasma emission induced by the second pulse of the
0.0
-0.2
U [V]
-0.4
tint = 2 μs
Fe 273.07 nm
single pulse
-0.6
1 × 80 m J tdelay = 0.8 μs
-0.8
0.0
-0.2
U [V]
-0.4 tint = 2 μs
Fe 273.07 nm
double pulse tdelay = 0.8 μs
-0.6 2 × 40 m J, Δt = 6 μs
-0.8
0 2 4 6 8 10
t [µs]
Fig. 6.3 The time profile of the plasma emission at the spectral position of Fe II 273.07 nm for a
single and a double pulse of 80 mJ burst energy. The photomultiplier signals have negative polarity
double pulse reveals a completely different behavior. About 0:4 s after the first
maximum, the time profile of the photomultiplier signal attains a second maximum.
At this time, the number of singly ionized species and their excitation achieves a
maximum. For times greater than 0:4 s (relative to the start of the emission induced
by the single pulse or to the start of the emission induced by the second pulse of the
double pulses), the line intensity is approximately a factor of two greater for the
double pulse than for the single pulse, although the laser pulse energy of the second
pulse is only half of the single pulse energy.
For analytical measurements, the ratio of the analyte signal, i.e., the element
specific emission, to the background caused by the continuum emission is a key
issue. Figure 6.4 shows the photomultiplier signals of the Mn II line at 293.31 nm
for a pure iron sample (Hu 85) and a low-alloyed steel sample with a Mn content of
0.62% (BAS 451). All experimental parameters are the same as for Fig. 6.3.
The emission maxima which coincide with the laser pulses are the same for
both samples. Obviously the origin of this emission is the element-independent
50
0.00
1 × 80 m J 40
BAS 451
-0.05
I BAS 451 / I Hu 85
I Mn 293.31 [a.u.]
Hu 85
30
-0.10
20
I BAS 451 / I Hu 85
-0.15
10
-0.20 0
50
0.00
2 × 40 m J 40
BAS 451
-0.05
I BAS 451 / I Hu 85
I Mn 293.31 [a.u.]
Hu 85
30
-0.10
20
-0.15
10
I BAS 451 / I Hu 85
-0.20 0
0 5 10 15
t [μs]
Fig. 6.4 Emission of the Mn II line at 293.31 nm for single (top) and double pulses (bottom) as a
function of time and two samples, where BAS 451 (solid line) has a Mn concentration of 0.62%
and Hu 85 (dashed line) is a pure iron sample. The right axis shows the ratio of the PMT signals
of the two samples
continuum radiation. After about 100 ns, differences in the PMT curves occur. This
difference is due to the manganese line radiation. Again significantly higher Mn line
intensities show up after the second pulse of the double pulse.
The ratio of line to background emission is also shown in Fig. 6.4. For a single
pulse after 1 s, this ratio amounts to 7, and 1 s after the second pulse of the
double pulse it achieves a value of 20. Besides an increase in the line intensity
with double pulses, the ratio of line to background radiation improves significantly.
Table 6.1 shows the time constants of the PMT curves shown in Fig. 6.4. All
curves can be approximated by an exponential decay with two time constants.
For the fitting of the measuring values obtained for the BAS 451 samples after
the second pulse of the double pulse, the time regime after the second emission
maximum was evaluated.
In the time interval up to 200–500 ns after the laser pulse, all four curves (solid
and dashed lines of the Mn intensity shown in Fig. 6.4) show an exponential decay
with time constants less than 100 ns. After the 40 mJ pulses, the time constant is in
the range of 30–40 ns, and for the 80 mJ pulse 60 ns. The higher value is attributed to
an enlarged hot core of the plasma and hence smaller energy losses. At later times,
the background radiation (Hu 85) decays with time constants of about 400–850 ns.
The line radiation (BAS 451) decays after the 80 mJ single pulses and the second
pulse of the double pulses with a time constant of about 1:5 s. The smaller time
constant after the first pulse of the double pulse amounts to about 900 ns which is
attributed to the higher energy losses of the plasma having a smaller size.
The effect of double pulses for the quantitative microchemical analysis of
steel with LIBS was investigated [6.14]. With the Paschen–Runge spectrometer,
multielement analysis of low-alloyed steel samples (BAS 451–460, 097) were
performed. Up to 14 elements were analyzed simultaneously. The element signals
were integrated with tdelay D 0:8 s and tint D 2 s. The number of prepulses was
Npp D 20, and the number of measuring pulses Nmp D 128. For example, Fig. 6.5
shows the calibration curve for silicon (ratio of line intensities of Si 212.41 nm and
Fe 273.07 nm, this line ratio is used as internal standard, cf. Chap. 11) for single and
double pulses of 80 mJ burst energy. Each sample is measured three to six times.
On average, the relative standard deviation of the measurements is smaller for the
double pulses (for Si, 1.8% for single and 1.5% for double pulses), see Fig. 6.5,
middle. The calibration curves are given by quadratic fits to the measured values,
with the correlation coefficient r as a measure of the quality of the approximation
of the calibration curve to the measured values. For double pulses, r increases
Table 6.1 Time constants of the exponential decay of the Mn signals of Fig. 6.4 for the first 500 ns
after the laser pulses and for later times
Pulse mode (mJ) Sample 1. Pulse, time constant (ns) 2. Pulse, time constant (ns)
1 80 Hu 85 60 400 – –
BAS 451 60 1,500 – –
2 40 Hu 85 33 360 40 850
BAS 451 32 880 30 1,450
0.12
I Si 212.41 / I Fe 273.07
0.08
1 × 80 mJ, r = 0.984
0.04 2 × 40 mJ, Δt = 6 μs,
r = 0.999
0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6
250
200
s[Si] [μg / g]
150
100
50
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
1000
750
500
Δ[Si] [μg / g]
250
-250
-500
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Si concentration [m.–%]
Fig. 6.5 Top: Silicon calibration curve for single and double pulses; middle: standard deviation
of measured concentrations; bottom: difference between measured concentration and certified
concentration of the reference samples
to 0.999 in comparison with 0.984 for single pulses. The measuring values lie
closer to the calibration curve, and the correctness of the analysis is improved, see
Fig. 6.5, bottom. Finally, for double pulses, the calibration curve has a factor of two
lower background equivalent concentration (BEC, cf. Sect. 11.2, (11.19), Fig. 11.3,
reduction from 2,100 to 790 g=g) and a higher gain, which further improve the
analytical performance. The limit of detection using the 3s-criterion is improved
from 174 to 19 g=g [see (11.16)].
As shown in Fig. 6.2, the plasma induced by double pulses has a greater extension
than that generated by single pulses using the same total laser energy. Hence,
0.8
0.7 1 × 80 mJ, r = 0.998

2 × 40 mJ, Δt = 6 μs,
0.6
r = 0.999
I Mn 293.31 / I Fe 273.07
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Mn concentration [m.-%]
Fig. 6.6 Manganese calibration curves for single and double pulses
self-absorption of spectral lines may occur which would result in a flattening of the
calibration curves. This effect is more pronounced for double pulses as shown exem-
plary for a manganese calibration curve in Fig. 6.6. For single pulses, the calibration
curve is nearly linear up to manganese concentrations of 1.4 m.-%, whereas the
curve for double pulses flattens significantly in the studied concentration range. Self-
absorption at higher concentration levels of analytes can be avoided by selection of
weaker spectral lines or lines having higher energy levels of the final state. However,
using intense spectral lines with double-pulse excitation is especially of interest for
the analysis of trace elements or low analyte concentrations, cf. Chap. 13.
The plasma dynamics and parameters of plasmas generated by single and double
pulses will be presented in Chap. 8.
6.2 Triple and Multiple Pulses
Collinear triple pulses can be generated within a single flashlamp discharge of a

Nd:YAG laser [6.16]. The laser emits three equal-energy pulses of 16 ns duration
(FWHM) with an interpulse separation of 25 s between the first and second pulse,
and 40 s between the second and the third pulse, respectively. The burst energy
amounts to 300 mJ. The analytical performances achieved with this approach are
described in detail in Sect. 13.1.1.
The effect of multiple pulses on the line emission of Fe I 532.80 nm is shown
in Fig. 6.7. The multiple pulse comprises four pulses with a total burst energy of
160 mJ. The intensity levels of the element-specific signal part after the third and
fourth pulse are about the same and they are greater than after the second pulse.
Hence, for an n-fold multiple pulse (n > 2) and use of several time integration
6.3 Tailored Pulse Trains 93
0.8
4 × 40 mJ, Δt = 6 μs
I Fe 532.80nm [a.u.] 0.6
0.4
0.2
0.0
0 5 10 15 20 25 30 35
t [μs]
Fig. 6.7 Intensity of the line Fe I 532.80 nm as a function of time for a fourfold multiple pulse
with 440 mJ burst energy and interpulse separations of 6 s (the PMT output is shown as positive
signal in this diagram)
Fig. 6.8 Temporal structure of a tailored pulse train as a general approach for bursts to optimize
element-specific emission
gates during such a burst, the detected element-specific intensity can be increased
approximately by a factor of (n 1) with respect to the analyte intensity of the
second pulse of a double pulse.
6.3 Tailored Pulse Trains
So far we have considered bursts comprising a series of two to four Q-switch

pulses with typical pulse widths of a few nanoseconds and interpulse separations
in the microsecond regime. In a more general approach, a burst can consist of
a series of pulses having different pulse widths and interpulse separations as
illustrated exemplarily in Fig. 6.8. By such an approach, a further optimization of
400
aluminum
350 C75 (1.1750)
steel
300
250
crater depth [μm]
200
150
100
50
1 × single pulse 1 × burst

burst mode
Fig. 6.9 Crater depths achieved with a single pulse and a single optimized burst for different
materials
the interaction of bursts with matter can be achieved to maximize the ablation depth
and/or the element-specific emission of interest.
Of course, such an approach requires a more sophisticated laser beam source.
Especially, for the bulk analysis of objects having surface layers of a different
composition – e.g., lacquers, oxides, anodized layers – burst structures are required
being capable to penetrate this surface layers effectively and to ablate and excite the
material underneath (cf. Sect. 18.4).
Figure 6.9 shows the results of an optimized burst structure in terms of the crater
depth achieved with a single Q-switch pulse compared with an optimized burst with
total laser burst energy in the 100 mJ range. By temporal pulse tailoring the crater
depth can be increased with a single burst by a factor of 3 for steel to 7 for aluminum
[6.24]. Such potentials of tailored pulse trains are, e.g., of interest for high-speed
applications of LIBS for material recycling tasks, cf. Sect. 18.4.
References
6.1. C. Aragon, J. Aguilera, J. Campos, Determination of carbon content in molten steel using
laser-induced breakdown spectroscopy. Appl. Spectrosc. 47, 606–608 (1993)
6.2. J. Belliveau, L. Cadwell, K. Coleman, L. Hüwel, H. Griffin, Laser-induced breakdown
spectroscopy of steels at atmospheric pressure and in air. Appl. Spectrosc. 39, 727–729
(1985)
References 95
6.3. K. Grant, G. Paul, J. O’Neill, Quantitative elemental analysis of iron ore by laser-induced
breakdown spectroscopy. Appl. Spectrosc. 45, 701–705 (1991)
6.4. R. Kirchheim, U. Nagorny, K. Maier, G. Tölg, Laser microprobe spectrometry of single-
crystal metals and alloys. Anal. Chem. 48, 1505–1508 (1976)
6.5. J. Millard, R. Dalling, L. Radziemski, Time-resolved laser-induced breakdown spectrometry
for the rapid determination of beryllium in beryllium-copper alloys. Appl. Spectrosc. 40,
491–494 (1986)
6.6. R. Wisbrun, R. Niessner, H. Schröder, Laser-induced breakdown spectrometry as a fast
screening for environmental analysis of trace amounts of heavy metals in soils. Anal. Meth.
Instrum. 1, 17–22 (1993)
6.7. D. Cremers, D. Romero, An evaluation of the factors affecting the analysis of metals using
laser-induced breakdown spectroscopy (LIBS). Proc. SPIE 644, 7–12 (1986)
6.8. W. van Deijck, J. Balke, F. Maessen, An assessment of the laser microprobe analyzer as
a tool for quantitative analysis in atomic emission spectrometry. Spectrochim. Acta 34B,
359–369 (1979)
6.9. T. Ishizuka, Y. Uwamino, Atomic emission spectrometry of solid samples with laser
vaporization-microwave induced plasma system. Anal. Chem. 52, 125–129 (1980)
6.10. M. Thompson, J. Goulter, F. Sieper, Laser ablation for the introduction of solid samples into
an inductively coupled plasma for atomic emission spectrometry. Analyst 106, 32–39 (1981)
6.11. J. Uebbing, J. Brust, W. Sdorra, F. Leis, K. Niemax, Reheating of a laser-produced plasma
by a second pulse laser. Appl. Spectrosc. 45, 1419–1423 (1991)
6.12. E. Piepmeier, H. Malmstadt, Q-switched laser energy absorption in the plume of an
aluminium alloy. Anal. Chem. 41, 700–707 (1969)
6.13. R. Scott, A. Strasheim, Laser induced plasmas for analytical spectroscopy. Spectrochim.
Acta 25B, 311–332 (1970)
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D. Appl. Phys. 28, 2181–2187 (1995)
6.15. A. Pichahchy, D. Cremers, M. Ferris, Elemental analysis of metals under water using laser-
induced breakdown spectroscopy. Spectrochim. Acta B 52, 25–39 (1997)
6.17. F. Colao, V. Lazic, R. Fantoni, S. Pershin, A comparison of single and double pulse laser-
induced breakdown spectroscopy of aluminum samples. Spectrochim. Acta B 57, 1167–
1179 (2002)
Acta B 57, 121–135 (2002)
6.19. S. Klimentov, S. Garnov, T. Kononenko, V. Konov, P. Pivovarov, F. Dausinger, High rate
deep channel ablative formation by picosecond-nanosecond combined laser pulses. Appl.
Phys. A 69, S633–S636 (1999)
6.20. D. Stratis, K. Eland, S. Angel, Effect of pulse delay time on a pre-ablation dual-pulse LIBS
plasma. Appl. Spectrosc. 55, 1297–1303 (2001)
6.21. D. Stratis, K. Eland, S. Angel, Dual-pulse LIBS using a pre-ablation spark for enhanced
ablation and emission. Appl. Spectrosc. 54, 1270–1274 (2000)
6.22. A. Kuwako, Y. Uchida, K. Maeda, Supersensitive detection of sodium in water with use of
dual-pulse laser-induced breakdown spectroscopy. Appl. Optics 42, 6052–6056 (2003)
6.23. R. Noll, R. Sattmann, V. Sturm, S. Lüngen, H.-J. von Wachtendonk, Schnelle Multielement-
analyse in der Stahlschmelze mit laserinduzierter Emissionsspektrometrie. Stahl u. Eisen
117, 57–62 (1997)
6.24. DE 10 2008 032 532, Verfahren und Vorrichtung zum präparierenden Lasermaterialabtrag,
Patent application, Publication date 11.2.2010, 13 p.
Chapter 7
Material Ablation
The interaction of the pulsed laser beam with the material causes an ablation of
mass from a liquid or solid measuring object. This mass is partially evaporated and
transferred into the plasma state, where the species are excited to emit element-
specific radiation used for LIBS. However, a part of the ablated mass leaves the
measuring object as material vapor, particle, or liquid. At a later stage, this vapor
condenses, forms particulates, or precipitates in the neighborhood of the interaction
region. The particles ejected may be carried away by a gas flow or they propagate
back to the surface of the measuring object forming a deposition. If a melt phase
occurs – e.g., in case of the interaction of pulsed laser radiation with metals – a part
of the mass is ejected as a liquid forming splashes or droplets which resolidify in
the surrounding atmosphere or on the surface of the measuring object. Due to the
complexity of processes, it is generally a difficult task to determine quantitatively
the amount of ablated mass for single laser pulses or laser bursts.
In Sect. 7.1, the formation of a crater in a solid object will be described.
Section 7.2 presents measurements of the ablated mass for single and multiple pulses
at different ambient pressures and gas types.
7.1 Crater Formation
Figure 7.1 illustrates schematically a cross section of a crater formed in a metallic

sample after the interaction with a series of laser pulses. The position of the
original surface RP can be used to define the upper border of the actual crater
volume. The material above RP, e.g., the collar formed around the crater rim (CL in
Fig. 7.1) and the splashes and droplets (SP in Fig. 7.1), is due to resolidifcation and
recondensation processes occurring during and after the laser–matter interaction. In
case of a rotationally symmetrical geometry, the diameter d1 describes the size of
the crater at the plane RP corresponding to the area of high irradiance of the laser
beam and d2 is the diameter of the area affected by splashes, recondensed particles,

98 7 Material Ablation
Fig. 7.1 Schematic illustration of the cross section of a crater formed in a metallic sample after
interaction with a series of laser pulses. S D sample, RP D original surface of the sample serving
as reference plane, CR D crater underneath RP, CL D collar formed by solidified melt ejected
along the crater walls during the interaction with the laser beam, SP D splashes and particles
deposited on the surface, d1 , d2 D diameters, see text; h D depth of the crater
Fig. 7.2 SEM image of a

crater formed in a steel
sample after interaction with
ten bursts, 2 40 mJ,
s D 0 mm
depositions, and secondary interactions between the formed plasma and the sample
surface.
Figure 7.2 shows an image of a crater formed by a series of ten double pulses in
a steel sample taken with a scanning electron microscope (SEM). Evaluating this
image yields for the diameters described above: d1 D 73 m and d2 D 390 m.
The diameter d1 is approximately equal to the determined diameter of the laser
beam at the plane RP of 70 m (cf. Fig. 3.12). While SEM images offer a high
lateral resolution, it is not possible to determine the depth of deep craters (the depth
of shallow craters can be determined with stereo SEM).
Figure 7.3 shows a metallographic cross section of a crater taken with an optical
microscope. The collar of the crater is clearly visible. Such cross sections enable the
determination of the crater dimensions diameter and depth. However the necessary
grinding process is time consuming and the achievable accuracy depends on the
accuracy to which extent the middle of the crater is met. The results of the
determination of the crater volume by light microscopy and stereo SEM are shown
exemplarily in Table 7.1 for single and double pulses irradiated onto a steel sample.
7.2 Ablated Mass 99
Fig. 7.3 Optical microscope

image of a cross section of a
crater formed after 200 bursts,
1 80 mJ, s D 0 mm
Table 7.1 Determination of Nb Eb .mJ / V=Nb (m3 ) light V=Nb (m3 ) stereo
the crater volume per burst by microscopy SEM
light microscopy and stereo
SEM for single and double 50 1 80 3,400 5,300
pulses 1 2 40 53,000 48,000
10 2 40 34,600 77,900
Nb D number of laser bursts, Eb D burst energy, V =Nb D
crater volume per burst, SEM D scanning electron
microscope, s D 0 mm
Differences in the determined crater volume per burst of up to a factor of two occur
comparing the two methods. Yet the data show clearly the increase in the crater
volume for double pulses by a factor of 9 to 16.
The shape of the crater depends among others on the laser beam profile. A spatial
beam profiling of the laser allows to modify the crater shape. A Gaussian beam
profile of a laser source can be formed to a top hat beam profile by, e.g., slightly
focusing the beam on a 600-m pinhole, which cuts off the outer part of the
Gaussian beam, and leads to a near flat top beam. A disadvantage of this method
is the loss of 95% of the burst energy. The aperture is then imaged onto the surface
of the specimen with a demagnification factor of 10. Figure 7.4 shows an image of
a crater generated with such a top hat laser beam profile on a galvanized sheet steel.
A crater with a flat crater bottom is formed.
7.2 Ablated Mass
Various methods have been applied to determine the ablated mass. Table 7.2 gives
an overview of results for the ablated masses for various materials, laser pulse
parameters, and ambient conditions. The masses ablated per pulse range from a
few 100 pg to about 10 g depending on the experimental conditions.
Fig. 7.4 LIBS crater

generated with a top-hat
beam profile using two
double pulses, Eb D 8:5 mJ,
crater diameter d1 155 m
Piepmeier et al. showed that the amount of ablated material at atmospheric

pressure was relatively constant over the range of energies used [7.3]. The stated
uncertainty of the mass determination is about 30%. At 1.3 mbar, the amount of
copper vaporized increased significantly with laser energy. With a 100-mJ laser
pulse, the amount of ablated copper is about an order of magnitude greater at
1.3 mbar, 300 ng, than at 1,000 mbar. Sattmann et al. studied the influence of single
and double pulses on the ablated mass [7.7]. The number of bursts which are
necessary to drill through a foil of stainless steel of 100-m thickness is determined
experimentally. This number is approximately inversely proportional to the material
removal per burst. The mean drilling depth per burst is the foil thickness divided by
the mean value of the number of laser bursts until penetration. It should be noted that
the material removal for a foil is different from that of a semi-infinite sample because
of restricted heat transfer conditions, but is expected to behave similarly. Semerok
et al. used polished samples and measured the craters formed by means of a phase-
stepping profilometer with lateral resolution of 0:5 m and depth resolution of
0:01 m [7.8]. Crater volumes were measured with reference to the initial surface
of the sample (without “corona” around the crater). The crater depth as a function
of the fluence and the crater volume as a function of the pulse energy were studied.
Ablation efficiency dependence on pulse duration and wavelength was discussed.
In the following paragraphs, the material ablation studied for single and multiple
pulses will be presented in more detail. Figure 7.5 shows the drilling depth d per
burst as a function of the laser burst energy at an ambient air pressure of 60
and 1,013 mbar for single-, double-, and four-pulse bursts (s D 0 mm for these
measurements). The interpulse separation for double- and four-pulse bursts is 6 s.
The mass ablated per burst (right-hand ordinate) is determined under the assumption
of a bore hole diameter of 200 m, which is the mean diameter of the holes,
measured with a microscope. For an ambient pressure of 1,013 mbar, the material
ablation of single pulses cannot be raised by increasing the laser burst energy as
7.2 Ablated Mass 101
Table 7.2 Overview of determinations of ablated mass per laser pulse (or per laser burst) for solid
samples
Reference Material EL (mJ) L .ns/ m (g) Remarks
Lorenzen [7.1] Steel 25–300 6–9 0.1–7 1,064 nm, ablated area
0:1–6 mm2 , n.s.
De Young [7.2] Al, Cu 400 8 0.4–1.5 High vacuum,
109 W=cm2 ,weighing
Piepmeier [7.3] Cu 3–120 60 0.04 Pressure dependence,
p D 1;000 mbar
Cu 3–120 60 0.04–0.31 p D 1:3 mbar, dissolution
of condensed matter and
flame atomic absorption
spectroscopy
Felske [7.4] Al 1.9 Mechanical Q-switching,
Cu 2.7 1:8 1010 W=cm2 ,
Steel 01.4 mass calculated from
crater volume
Leis [7.5] Steel 5 8 0.03 1,064 nm, crater diameter
100 m, p.Ar/ D
140 mbar, n.s.
Jowitt [7.6] Steel 750 20 0.03 1,064 nm, weighing,
weight of ablated mass
is approximately
proportional to the laser
pulse energy in the
range of 500–800 mJ
Sattmann [7.7] Steel 40–320 25 SP: 0.1–0.14 1,064 nm, drilling of steel
DP: 0.3–0.7 foil, first demonstration
of enhanced mass
ablation with DP,
t D 6 s
Semerok [7.8] Fe 0.06 4 532 nm, 1,064, 532, 266 nm, laser
d D 8 m: spot diameter d
710 pg
25 ps 532 nm, Phase-stepping
d D 9:5 m: profilometer to
188 pg determine crater volume
EL D laser pulse or laser burst energy, L D pulse width of the laser pulses; in case of double
pulses, the interpulse separation t is stated as well. If not stated otherwise the experiments
were performed at normal atmospheric conditions and single pulses were applied. SP, DP D
single/double pulses, n:s: D method of determination of the ablated mass is not specified
observed by Piepmeier et al., see [7.3]. This behavior is attributed to a shielding of

the target by a plasma formed in the atmosphere in front of the sample. The material
removal of double pulses is higher than that for single pulses with the same energy,
and it can be further raised by using four-pulse bursts. No saturation of the removal
is observed for increasing burst energy of double pulses up to Ebd D 320 mJ. Due
to the lower peak intensity of the double pulse, shielding effects are reduced which
favors – among other effects to be discussed later – an increase of the mass ablation.
25
60 1013 mbar 1.4

20 single pulse
1.2
double pulse
four-pulse burst
1.0
d / burst [μm]
m / burst [μg]
15
0.8
10 0.6
0.4
5
0.2
0 0.0
0 100 200 300 400
Eb [mJ]
Fig. 7.5 Drilling depth d per burst in a steel foil as a function of the burst energy for single-,
double-, and four-pulse bursts at 60 and 1,013 mbar ambient atmospheric pressure. The interpulse
separation of double- and four-pulse bursts is 6 s. The right axis shows the corresponding ablated
mass per burst. s D 0 mm
In the following paragraphs, we compare the ablated mass of a double pulse with
energy Ebd with the ablated material for two single pulses with half of the energy of
the double pulse, E s D Ebd =2. Figure 7.5 shows that, e.g., a 240mJ double pulse
removes about 0:32 g (at 1,013 mbar ambient pressure), whereas two 40-mJ single
pulses ablate only about 2 0:1 g. The significant difference between the ablation
of a double pulse with energy Ebd and the ablation of two single pulses with energy
E s D Ebd =2 shows that the material ablation is strongly affected by the combined
effect of the subsequent pulses of a double pulse at interpulse separations of 6 s.
Figure 7.5 also shows results for a pressure of 60 mbar. Due to the reduced ambi-
ent pressure, the material vapor can escape faster from the interaction region, which
results in higher specific drill depths for all studied laser parameters. Furthermore,
shielding effects by a plasma in the ambient atmosphere are diminished.
7.3 Material Ablation in Different Ambient Gases

by Collinear Multiple Laser Pulses
Typically the interaction region is flushed with an ambient gas to avoid a breakdown
at residual particulates above the sample surface which are generated by preceding
laser pulses [7.9,7.10,7.11,7.12,7.13,7.14]. Furthermore the plasma conditions like
electron density and electron temperature are strongly dependent on the ambient
gas [7.13,7.15]. Often inert gases such as argon are used because they offer a higher
transmittance for wavelengths below 200 nm [7.9, 7.10, 7.11] compared with air.
7.3 Material Ablation in Different Ambient Gases by Collinear Multiple Laser Pulses 103
The oxygen content of 20% absorbs significantly wavelengths below 200 nm (cf.
Fig. 3.16). This wavelength region is interesting especially for the elements C, P, S,
N, and O, which are important, for example, for the analysis of steel samples which
are used for process monitoring during steelmaking.
The repetition rates of Q-switched solid-state lasers with pulse energies between
100 and 500 mJ and a pulse length of 10 ns are typically in the range of 10–50 Hz.
Many studies about laser ablation were performed using these parameters [7.7,7.12,
7.13,7.14,7.16,7.17]. Sattmann et al. describe the influence of interpulse separation
and burst energy of collinear double pulses on laser ablation [7.7]. Russo studied
the influence of the pressure of helium and argon atmospheres [7.16]. The lower
ionization energy of argon leads to a more severe plasma shielding and thus smaller
crater depths compared with those obtained in a helium atmosphere. Mass ablation
rate depends on the laser pulse duration and the power density on the sample surface
[7.17]. The type of ambient gases like helium, argon, and neon have an effect on
the particle size distribution showing that in helium significantly smaller particles
occur compared with Ar [7.14]. Aguilera et al. looked for the effect of ambient
gases on the temporal evolution of electron temperatures and electron densities of
the laser-induced plasmas [7.13]. The decay times are shorter in helium than in air
and argon.
In the following paragraphs, the influence of single, collinear double, and triple
laser pulse bursts in different ambient gases consisting of mixtures of argon, nitro-
gen, and oxygen on the ablation of a steel sheet is described in more detail [7.18].
The experimental setup is shown in Fig. 4.18, Chap. 4. The laser beam generated
by a Q-switched Nd:YAG laser working at the fundamental wavelength 1,064 nm
is guided via a Glan-laser prism, a twofold beam expansion, and a mirror to a
lens focusing the laser beam into a gas-flushed sample stand onto the sample.
The Glan-laser prism is used to attenuate the laser pulse energy while keeping
the operational conditions of the laser source constant. Two lasers are used for the
experiments. Laser 1: Surelite SLI-10 (Continuum, Santa Clara, CA), repetition rate
of 10 Hz, with double pulse option, interpulse separation 30 s. Laser 2: YM/R-
201 (Innolas, Krailling, Germany), repetition rate of 50 Hz, with double and triple
pulse option, interpulse separations 5–50 s.
The plasma light is detected via a direct light channel with a vacuum Paschen–
Runge spectrometer without further optical imaging. In the Paschen–Runge spec-
trometer, the plasma light is dispersed spectrally covering a spectral range from 178
to 406 nm (grating, 2,700 l/mm) with a spectral resolution of 20 pm. The distance
of the sample surface to the entrance slit of the spectrometer is 300 mm. The
plasma signals are detected at a detection angle of ˛D D 10ı between the optical
axis of the spectrometer and the sample surface. Twenty photomultipliers are located
on the Rowland circle (diameter 0.5 m) to detect 19 element lines and the zeroth
order signal. A multichannel-integrator electronics (see Sect. 4.5) integrates the
signals in programmable time windows for each generated plasma (SE in Fig. 4.18).
The flow rate of the ambient gases through the sample stand amounts to about
17 l/min. The gas mixtures studied were provided in gas bottles from Linde AG with
specified composition. A series of laser pulses are irradiated to one sample position.
The focal length of the plano-convex lens is f D 150 mm and the focal position
is s D 8 mm within the sample to avoid a gas breakdown in front of the sample
surface. The experiments are carried out with the above-mentioned two laser sources
in different pulse modes as illustrated in Fig. 3.9, Chap. 3.
The pulse ratio and the interpulse separation within a burst can be varied. Double-
and triple-pulse bursts (DP/TP) with equidistant interpulse separations, uniform
energy distribution (i.e., pulse ratios D 1), and equal total energy (i.e., burst energy)
are used. The pulse width for laser 1 in single-pulse mode (SP) is SP D 6 ns and in
DP mode 10 ns for both pulses. The interpulse separation was set to t1 D 50 s,
where laser 1 is able to generate stable DP with uniform energy distribution (burst
energy 160 mJ D 80 C 80 mJ). For laser 2, the pulse widths for SP is SP D 7 ns
and for DP, TP 14–40 ns for each pulse. Comparing the different burst modes,
it is not possible to keep the pulse widths constant because the double- and triple-
pulse bursts are generated with one laser system within a single flashlamp discharge.
The pulse widths depend on the flashlamp voltage, the delay between the start of the
flashlamp discharge and the Q-switch trigger as well as on the interpulse separations.
Generally the material ablation will be influenced by both the pulse structure (SP,
DP, TP) as well as by the individual pulse widths. However, with the setup used here,
there is no possibility to vary the pulse widths independent from the pulse structure.
Nevertheless, the use of a single laser system to generate collinear multiple pulses
is of great practical interest since it simplifies the setup of industrial LIBS systems.
The line emissions are integrated with a time delay of tdelay D 4 s and an
integration width of tint D 10 s after the last pulse of a burst. Five measurements
at different sample locations are carried out for each parameter set.
The efficiency of the laser ablation process is described by the ablation burst
number Nabl; 3s , which is defined as the number of laser bursts needed to penetrate
a steel sheet of a given thickness. In order to quantify Nabl; 3s experimentally in a
precise and reproducible way, a block of copper is put onto the rear side of the steel
sheet, see Fig. 7.6, and during the ablation sequence LIBS signals of the copper line
324.8 nm are detected.
The steel sheet has a low copper content and therefore only a background Cu
signal is detected during the ablation of the steel sheet. Once the steel is penetrated,
the Cu signal rises strongly, see Fig. 7.7.
Fig. 7.6 Schematic drawing

of steel sheet and copper
block with laser-generated
crater. If the depth of the
crater is greater than 100 m
a copper signal is detected
indicating the penetration of
the steel sheet
8000
7000
6000
ICu 324.75 [counts]
5000
4000 <ICu>
<ICu> + 3s
3000
2000 Nabl, 3s
1000
0 200 400 600 800 1000 1200 1400
laser burst number
Fig. 7.7 Intensity of the copper line 324.75 nm as a function of the laser burst number. At a burst
number of Nabl; 3s , the intensity increases to hICu iC 3s, where hICu i is the average copper intensity
prior to penetration and s the corresponding standard deviation
First, we determine the average background Cu signal hICu i prior to penetration

of the steel sheet and the corresponding standard deviation s (in the example shown
in Fig. 7.7 from pulse 101–700). We define Nabl; 3s as the lowest ablation burst
number where the Cu signal exceeds the sum (hICu i C 3s). Hence, the probability
of a wrong result due to statistical fluctuations in the background Cu signal is less
than 1%.
Steel sheets with known thickness are used as samples for the experiments to
determine the ablation burst number Nabl; 3s . The steel sheets are made of cold-
rolled, spring hard metal sheet X5CrNi18 (1.4301) with a well-defined thickness.
The sheets are 13 mm wide with a thickness of 100 ˙1:2 m. Their density amounts
to 7:9 g=cm3 . The concentration ranges of the elements are listed in Table 7.3 [7.19].
From the composition given in Table 7.3, the minimum iron content is greater than
cFe D 66:8 m.-%.
Gas mixtures with varying proportions of argon are chosen to examine a range of
chemical reactivity and gas densities. An idea to enhance the ablation process is to
use a mixture of argon and oxygen as ambient gas. An exothermic reaction between
ablated material in the outer plasma region may contribute an additional energy
supply for the plasma and thus enhance the plasma–matter interaction. Therefore,
five ambient gases are studied, see Table 7.4, to determine the influence of gas
composition and density on the ablation efficiency.
Figure 7.8 shows the ablation burst number Nabl; 3s as a function of the gas density
for steel sheets with a thickness of 100 m, single- and double-pulse bursts of
Table 7.3 Concentration Element Concentration range (m.-%)

ranges of alloy elements in
the steel sheet of grade C 0:07
X5CrNi18 (1.4301) [7.19] Si 1:00
Mn 2:00
P 0:045
S 0:015
Cr 17.0–19.5
Ni 8–10.5
N 0:11
Table 7.4 Composition of No. Ar O2 N2 Gas density

used mixtures of gases (vol.-%) (vol.-%) (vol.-%) (kg/m3 )
1 0 20 80 1:29
2 50 50 0 1:61
3 92 8 0 1:76
4 99 1 0 1:78
5 99.999 0 0 1:784
5000 1 SP
DP
4000
Nabl, 3s
3000
2
2000 3
1 4
5
2
1000 3 4
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

gas density [kg / m3]
Fig. 7.8 Ablation burst number Nabl; 3s as a function of the gas density. Laser 1, Eb D 160 mJ,
t1 D 50 s, s D 8 mm. Errors bars mark 3s standard deviation. The numbers 1–5 at the data
points refer to the gas mixtures listed in Table 7.4
Eb D 160 mJ generated with laser 1. The interpulse separation of the DP amounts

to t1 D 50 s.
A nearly linear relationship between the ablation burst number Nabl; 3s and the
gas density is observed for both SP and DP. The ablation burst number for DP is
always smaller than for SP and can be reduced by a factor of 2.8 in ambient gas
no. 1 (cf. Table 7.4) using DP instead of SP. For gas no. 5, the factor of improvement
is 1.6. The influence of the burst mode on the ablation process decreases with rising
gas density. The ablation burst number plotted as a function of the oxygen content
shows no clear correlation. There is certainly a reaction of the oxygen with the
plasma constituents but the effect might only take place in the outer region of the
plasma and has no significant influence on the steel ablation.
The crater diameters are in single-pulse mode dcrater;SP 1;300 m and in
double-pulse mode dcrater;DP 750 m for all mixtures of gases. In the studied
parameter regime, the gas density of the ambient gases has no effect on the crater
diameters. Therefore, the ablation burst number is a measure for the steel ablation
and approximately proportional to the ablated depth for both SP and DP.
The smaller crater diameters for double-pulse bursts are an effect of the reduced
plasma–sample interaction. Also St-Onge et al. observed smaller crater diameters
using DP instead of SP [7.20]. An explanation of the reduced plasma–sample
interaction for DP is given in [7.21]. For double pulses, the first pulse leads to a
local dilution of the ambient gas in the interaction region between laser radiation
and sample. The second pulse of the double-pulse burst interacts again with the
sample surface and generates a plasma which detaches from the sample surface
providing an efficient way to transport material away from the interaction region.
For this the thermal influence of the plasma on the sample decreases compared with
single pulses. The plasma generated by single pulses still stays in contact with the
sample surface causing an enhanced thermal interaction with the sample.
The material ablated per burst increases with rising gas density (a reduction of the
oxygen and an increase of the argon content enlarge the gas density of the ambient
gas). In the case of negligible plasma shielding, the plasma–sample interaction
is greater in an ambient gas of higher density because the plasma expansion is
restricted by the surrounding mass density which results in higher density and
temperature of the plasma in the vicinity of the sample surface leading to enhanced
steel ablation.
On the contrary, investigations on copper samples showed that the crater depth
and the ablated mass are larger in air than in argon atmosphere [7.15]. This behavior
was attributed to the high electron density in case of an argon atmosphere which
results in an optical shielding of the sample surface. In [7.15] the laser beam is
focused on the sample surface (i.e., s D 0 mm, instead of 8 mm as applied here),
resulting in laser radiant exposures of about 150 J=cm2 , which are more than a factor
of three higher than those used in this study. The shielding effect may be effective in
[7.15], whereas it plays a minor role here. Furthermore, the gas flow rate applied in
[7.15] amounts to 100 ml/min, which is more than two orders of magnitude less
the one we used (17 l/min) while the laser repetition rate is comparable (20 Hz
in [7.15] and 10 Hz here). The high flow rate allows for an effective transport of
the ablated or condensated particulates out of the interaction region, thus avoiding
any accumulation or redeposition effects. This is a precondition for reproducible
measurements of the ablation as manifested in the small error bars in Fig. 7.8.
Vadillo et al. studied also the ablation behavior in different ambient gases [7.22].
Focusing a laser of 581 nm on a Fe foil (i.e., s D 0 mm), higher ablation rates
were found in an air atmosphere than in an argon atmosphere as well as higher

ablation rates for low pressure than at atmospheric pressure. The results obtained
were explained as an effect of the redeposition of rejected material at atmospheric
pressure. However, no flow rate of the gas inside the measuring chamber is stated in
[7.22]. If the flow rate was zero, then an accumulation of particulates in the vicinity
of the interaction region originating from previous laser pulses will occur, causing a
scattering and absorption of the laser radiation of subsequent pulses thus leading to a
reduced interaction of the laser radiation with the sample surface. Due to the higher
gas density, this effect is expected to be stronger for argon than for air (additionally
in air a chemical reaction between the fine iron particulates and the oxygen will take
place). This may explain the difference in the observed ablation behavior in air and
argon to the one found here, where a high flow rate was used. Also the experiments
described in [7.23] were performed with a very low flow rate of 50 ml/min and
fluences being a factor of two greater than those used here. Mass ablation was
quantified by an electronic microbalance, which implies that the result is to be
interpreted as a net mass loss, i.e., the mass vaporized minus the mass redeposited.
A further difference of the setup used here (cf. Fig. 4.18) in contrast to those
described in [7.15, 7.22, 7.23] is the orientation of the normal of the sample surface
in relation to the axis of gravity. In [7.15,7.22] the laser is irradiated onto the sample
from above (i.e., surface normal and gravity vector are antiparallel), whereas the
laser propagates from the bottom to the sample in the setup shown in Fig. 4.18
(i.e., the surface normal and the gravity vector are parallel). The latter configuration
favors transport of the ablated material or condensated particulates away from the
sample surface, thus reducing a potential redeposition.
Besides the gas density also other physical properties of the gas mixtures used
were considered, which may show a correlation with the ablation burst number, such
as ionization potential, chemical reactivity (via the oxygen content), and thermal
conductivity. Only for the latter quantity, a clear correlation was found in terms of
a decrease of the ablation burst number with decreasing thermal conductivity (gas
mixture no. 1 has the highest thermal conductivity of about 0.026 W/m K, and gas
mixture 5 has 0.018 W/m K). However, the interpretation given above discussing
the effect of the higher gas density is congruent to the effect of a lower thermal
conductivity which reduces the heat flow from the plasma to the ambient gas thus
favoring higher temperatures in the plasma.
Furthermore, the influence of the burst mode and interpulse separation on laser
ablation was studied. In these measurements, SP, DP, and TP are applied with
symmetric burst energy partition and equal interpulse separation for triple-pulse
bursts generated with laser 2. The ablation burst numbers Nabl; 3s are measured
on steel sheets with a thickness of ds D 100 m and a total burst energy of
Eb D 105 mJ. (This burst energy is the maximum possible burst energy for TP
and interpulse separations of t1 D t2 D 5 s. For all other SP, DP, and TP
configurations, laser 2 is able to generate higher burst energies, which are then
attenuated externally down to 105 mJ to keep the burst energy for the different pulse
structures studied constant.) The ambient gases no. 4 and 5, see Table 7.4, are used
to examine the influence of an oxygen content on the ablation process. Figure 7.9
2000
gas no. 4
1800
gas no. 5
SP
1600
1400
1200
Nabl, 3s
1000
DP
800
600
400
TP
200
0
0 10 20 30 40 50
interpulse separation Δt i [μs]
Fig. 7.9 Ablation burst number Nabl; 3s as a function of the interpulse separation and burst mode
for ambient gases no. 4 (99 vol.-% argon, 1 vol.-% O2 / and 5 (99.999 vol.-% argon) of Table 7.4.
Laser 2, Eb D 105 mJ, s D 8 mm
shows the ablation burst numbers as a function of the interpulse separation ti for
SP, DP, and TP. The ablation burst number for SP is plotted at t D 0 and marked
with a horizontal line.
No significant difference for gas no. 4 and 5 is observed. Hence, the gas density
seems to have the dominant influence on depth penetration, whereas the oxygen
content plays only a minor role for the studied gas mixtures.
For all interpulse separations ti , the ablation burst numbers are significantly
smaller for DP and TP than for SP. Using DP with an interpulse separation of t1 D
50 s – like in the measurement with laser 1 – the ablation burst number Nabl; 3s for
DP is reduced by a factor of 1.4. With DP and an interpulse separation of t1 D
15 s, the depth penetration increases by a factor of 2.4 in comparison with SP.
Using TP with interpulse separations of t1 D t2 D 50 s, the ablation burst
number decreases by a factor of 3.1. With TP and interpulse separations of t1 D
t2 D 15 s, the ablation burst number decreases in ambient gases no. 4 and 5 by
a factor of 6.3 in comparison with single pulses. Surprising are the measurements
with interpulse separations of t1 D t2 D 5 s. A further lowering of the ablation
burst number was expected. However, the ablation burst number Nabl; 3s increases for
DP and stays about constant for TP.
In Fig. 7.10, ablation burst numbers Nabl; 3s of measurements in ambient gas
no. 5 performed on different days are illustrated to show the reproducibility of the
results. The days are labeled with the letters A, B, and C. “A” marks the first day
2000
1800 Day A
Day B
1600
Day C
1400
1200
Nabl,3s
1000
800
600
400
200
0
SP DP 30 μs DP 5 μs TP 30 μs TP 5 μs
Fig. 7.10 Ablation burst numbers Nabl; 3s for SP, DP, and TP measured at different days. Laser 2,
Eb D 105 mJ, SP/DP/TP D single/double/triple pulse
of measurements, “B” is 8 days after “A,” and “C” is 11 days after day “A.” The
error bars represent the standard deviation of five measurements on the samples
(the same sample type is used as before). The labeling “DP 30 s” means that the
ablation burst numbers Nabl; 3s are measured in double-pulse mode with an interpulse
separation of t1 D 30 s. The other labels are given in an analogous way. In the
first part of the figure, the SP measurements are presented showing a difference
in the ablation burst numbers for SP on day A and B which is about a factor 2
greater than the standard deviation. In the second part of Fig. 7.10, the ablation burst
numbers using DP are shown. There is no significant difference between the results
using interpulse separations for t1 D 30 s or t1 D 5 s. For TP again, no
significant difference for bursts with interpulse separations of t1 D t2 D 30 s
and t1 D t2 D 5 s was found. The error bars for t1 D t2 D 5 s are
significantly greater. The reason is that for TP, the LIBS signals used to determine
the ablation burst number Nabl; 3s show a higher scattering attributed to a reduced
stability of laser 2 in the TP mode.
7.4 Material Ablation by Collinear Double Laser Pulses

at Reduced Ambient Gas Pressures
Several authors have studied the influence of ambient gases and different gas
pressures on laser ablation with single pulses [7.5, 7.24]. Argon as ambient gas in
the pressure range from 0.1 to 1,000 hPa was studied for LIBS using single Nd:YAG
7.4 Material Ablation by Collinear Double Laser Pulses at Reduced Ambient Gas Pressures 111
laser pulses having an energy of 5 mJ [7.5]. The target materials were low-alloyed
steel samples. For the silicon line Si I 288.2 nm, the authors found a pronounced
maximum at an ambient pressure of 140 hPa attributed to an optimum atomization
process of the ablated and expanding sample material interacting with the argon
buffer gas. The expansion dynamics of the plasma generated by laser ablation of
an aluminum target was investigated using frequency-doubled radiation from a
Q-switched Nd:YAG laser [7.24]. The ambient gas studied was air at pressures
ranging from 106 to 100 torr. In the pressure region between 106 and 2 102 torr
the plasma expands freely. At pressures of 101 torr, the plasma breaks away from
the target surface and a contact boundary is formed between plasma and a shock
wave in the gas. At 1.3 torr, the onset of an instability is observed perturbing the
interface between plasma and ambient gas. At pressures between 10 and 100 torr
a plume confinement is observed. The dynamics of plume boundaries generated
by irradiation of steel targets using a KrF laser at 248 nm were studied at ambient
argon pressures between 3 105 and 20 mbar [7.25]. The authors showed that
over a wide range of parameters the dynamics can be described by a generalized
shock wave expansion model. Ambient gas effects were also studied for resonance-
enhanced laser-induced plasma spectroscopy irradiating potassium iodate pellets
with frequency-doubled laser pulses from a Nd:YAG laser followed after 27 ns by
a dye laser pulse at 404.4 nm to resonantly excite the K atoms [7.26]. By using a
rarefied atmosphere of heavy inert gas, the hot plume could be thermally insulated
and confined.
Most studies dealing with the effect of ambient gas pressure were restricted to
the interaction of single laser pulses. However, Sattmann et al. studied the material
ablation for single and collinear double pulses as a function of the burst energy for
air pressures of 60 and 1,013 mbar giving a first indication that the saturation effect
observed for the material ablation as a function of the irradiated laser pulse energy
using single pulses at 1,013 mbar does not occur at pressures of 60 mbar and can be
overcome by double pulses [7.7].
In the following studies are presented to clarify the influence of the ambient
gas pressure for collinear double pulses interacting with a solid iron target [7.27].
Material ablation is determined as a function of the ambient gas pressure being
varied between 0.1 and 1,000 mbar.
The experimental setup is shown in Fig. 7.11. A flashlamp-pumped Q-switched
Nd:YAG laser emits at 1,064 nm and has a repetition rate of 10 Hz. The Pockels cell
voltage allows for a stepwise increase of the voltage up to the /4-voltage to generate
a burst of laser pulses with adjustable interpulse separations and pulse heights during
a single flashlamp discharge. The laser beam is guided via two mirrors and two
beam splitters to attenuate the burst energy. The resulting burst energy is measured
with a volume absorbing disk calorimeter (Scientech 38–0101). A lens of 200-mm
focal length focuses the laser beam through a window onto a sample positioned
inside a measuring chamber. This chamber can be evacuated by a vacuum pump.
Via a gas inlet the chamber is filled with pressures ranging from 0.1 mbar to 1 bar.
The pressure is measured with two pressure gauges with measuring ranges of 0.1–
27 mbar, 1–1,060 mbar, respectively (Wallace & Tiernan). The light emitted by the
Fig. 7.11 Experimental M1 M2

setup. LS D laser system,
M1, M2 D mirrors, BS1,
BS2 D beam splitters, BS2
CM D calorimeter, L D lens, BS1
W1, W2 D windows, LS
S D sample, P D plasma, CM
MC D measuring chamber, L
F D filter, FO D fiber optics, FO
W1
G D gas inlet, VP D vacuum
pump, PG D pressure gauge P F
W2
G
S
MC PG
VP
plasma is guided to the spectrometer with a fiber optics positioned at a distance of

75 mm from the plasma center and oriented at an angle of 45ı to the direction of
the incoming laser beam. The fiber optics consists of a bundle of quartz fibers each
with a core diameter of 200 m. The diameter of the bundle amounts to 3 mm. In
front of the fiber, an edge filter is inserted to assure that with the spectrometer only
emission lines in first diffraction order are observed (Schott type GG 395). With
the numerical aperture of the fibers and the distance of 75 mm, the diameter of the
receiving cone at the plasma location amounts to about 30 mm; hence, it is expected
that the emission received is spatially integrated over the whole plasma volume.
The laser beam profile was measured with a CCD camera being shifted axially
along the beam propagation axis to determine the Rayleigh length and the beam
radius. The converging beam behind the focusing lens is further attenuated by
reflection at a wedge prism and a stack of gray filters in front of the CCD camera.
Table 7.5 shows the results for the beam waist radius and the Rayleigh length
for single and double pulses of the same total energy of 80 mJ and an interpulse
separation of t D 6 s. No significant differences were found.
For the measurements, the following parameters were used: single pulses 80,
160 mJ, pulse width 20 ns; double pulses 2 40; 2 80 mJ, width 40 ns; interpulse
separation t D 6 s, energy ratio 1:1; position of the beam waist relative to the
sample surface s D 10 mm (i.e., the focus lies underneath the sample surface),
detection after the single pulse or the second pulse of a double pulse with tdelay D
1 s, tint D 1 s. Time-resolved studies are out of the scope of this study.
The following samples were used: (a) high-alloyed steel foils of the grade 1.4301
(18% Cr, 9% Ni) with a thickness ranging from 25 to 300 m and (b) pure iron of
Thyssen Krupp Stahl AG, TRP 1043.
The material ablation was determined by counting the number of bursts necessary
to drill through high-alloyed steel foils. Figure 7.12a, b shows the number of bursts
as a function of the foil thickness for air pressures of 1 bar and 0.1 mbar. Each data
Table 7.5 Beam waist radius w0 and Rayleigh length zR determined for single and double pulses
focused by a 200-mm lens
w0 (m) zR (mm)
Double pulse 2 40 mJ 32 ˙ 4 1:3 ˙ 0:2
Single pulse 1 80 mJ 35 ˙ 4 1:3 ˙ 0:2
a
6000
p = 1 bar, air
m1 = (46 ± 2) μm–1
5000
number of bursts b [1]
2 x 80 mJ
1 x 160 mJ
4000
3000 3x
3x
2000
m2 = (12.9 ± 0.6) μm–1
1000
b 1500
p = 0.1 mbar, air
m1 = (8.23 ± 0.6) μm–1

1200
900
3x
600 m2 = (4.0 ± 0.12) μm–1
300
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30
foil thickness dF [mm]
Fig. 7.12 Number of bursts needed to drill through high-alloyed steel foils as a function of the foil
thickness for air pressures of (a) 1 bar and (b) 0.1 mbar for single and double pulses. Burst energy
Eb D 160 mJ
point is the average of seven measurements. The error bars correspond to 99%
confidence intervals and are plotted enlarged by a factor of 3 for better visibility.
Linear relationships are found revealing significant differences in the slopes between
single and double pulses and for the different pressures. For double pulses, the
number of bursts necessary to drill through the steel foils is in general significantly
smaller compared with single pulses. These differences cannot be explained by
different beam profiles, since these were nearly identical for single and double
pulses (see above). Furthermore, during drilling the effective laser irradiance is
not changing significantly with increasing drilling depth. Taking the determined
Rayleigh length zR D 1:3 mm and s D 10 mm the reduction of the beam diameter
at the maximum interaction depth of 300 m is 3% with respect to the beam
diameter at the surface of the foil. This corresponds to an increase of the irradiance
by 6% while penetrating through the foil from the front to the backside. This
increase has obviously no influence on the observed linear relationship between
the number of bursts and the foil thickness shown in Fig. 7.12.
The potentially restricted heat conduction in the foil is estimated as follows.
The foil can be considered to be a semi-infinite material as long as the following
condition is fulfilled: p
df 4 L ; (7.1)
where df is the foil thickness, is the thermal diffusivity of the sample material, and
L is the duration of the laser pulse.
The right-hand side of relation (7.1) describes the depth in a semi-infinite sample
where the temperature rise is 3 103 of the value achieved at the surface. With D
0:04 cm2 =s and L D 40 ns, relation (7.1) yields: df 1:6 m. At the beginning of
the drilling this is fulfilled for all foil thicknesses used down to the thinnest foil
having a thickness of 25 m.
During drilling, the successive ablation leads to a local stepwise decrease of the
residual foil thickness and hence after a certain number of laser pulses irradiated at
one location the residual thickness may approach the estimated limit of 1:6 m and
condition (7.1) no longer holds. Figure 7.12b shows that for double pulses and the
thinnest foil of 25 m, the number of bursts determined is about 50. For a rough
estimate, we calculate the ablated depths per burst by 25 m=50 D 0:5 m/burst.
From this we expect that for the last three bursts, condition (7.1) is not fulfilled.
With respect to the total number of bursts applied, this is obviously only a minor
effect. Hence, no significant influence of a restricted heat conduction is expected
which is confirmed by the observed linear behavior of the numbers of bursts as a
function of the foil thickness as shown in Fig. 7.12.
A reduction of the pressure from 1 bar to 0.1 mbar reduces the number of bursts
per ablated micrometer (i.e., the slope of the regression lines plotted in Fig. 7.12) by
a factor of 5 for single pulses and a factor of 3 for double pulses. The best fit straight
lines do not go through the origin but intersect the abscissa at a final value of the foil
thickness df 0 . For a given pressure the intersection point has for single and double
pulses about the same value. From the curves of Fig. 7.12, we determine for 1 bar
df 0 D 27 ˙ 4 m and 16 ˙ 5 m for 0.1 mbar. These values are much greater than
the material thickness ablated per burst (reciprocal of the slopes in Fig. 7.12) which
lies in the range of 20–250 nm. The quantity df 0 is interpreted as the extrapolated
foil thickness where the induced plasma pressure of the first laser burst is sufficiently
high to break the foil. The thickness df 0 is greater for 1 bar than for 0.1 mbar, since
the ambient gas acts as a mass load for the expanding plasma leading to a local
confinement and thus increasing the pressure exerted onto the foil.
Figure 7.13 shows the number of bursts needed to drill through a foil of 50 m as
a function of the pressure for two different burst energies. For double pulses, only a
weak dependence is observed becoming manifest in an increase of the burst number
by about a factor of two while approaching 1,013 mbar. Considering the error bars,
the number of burstsindexBurst!number of bursts does not change significantly
while increasing the burst energy from 80 to 160 mJ in the pressure range <50 mbar
and at 1 bar for single pulses and double pulses. For a burst energy of 80 mJ, the
number of bursts for double pulses is always less than the number of bursts for single
pulses, see Fig. 7.13b. Figure 7.13a (Eb D 160 mJ) shows a different behavior. At
a pressure of 500 mbar, the number of bursts is greater for double pulses than for
single pulses. For pressures <100 mbar, the number of bursts is independent of the
pressure. For Eb D 80 mJ, the number of bursts does not change for pressures in the
range <50 mbar.
a 1500
2 × 80 mJ 2x
1200 1 × 160 mJ
900
600
10 x
300
b 1500
2 × 40 mJ
1200 1 × 80 mJ
900
600
10 x
300
0
0.1 1 10 100 1000
air pressure p [mbar]
Fig. 7.13 Number of bursts needed to drill through a high-alloyed steel foil with a thickness of
50 m as a function of the ambient air pressure for single and double pulses of burst energies
(a) 160 mJ and (b) 80 mJ
Figure 7.13a shows that the number of double pulses to drill through the 50-
m foil at 1,013 mbar is about 300. The same number is obtained for single pulses
and pressures 100 mbar. This means at 1 bar, the material ablation of double
pulses is similar to the one observed in the case of single pulses (of the same total
energy) irradiated at a reduced ambient pressure of 100 mbar. This suggests the
interpretation that double pulses act as if the ambient pressure is reduced. This
is consistent with the findings of a paper studying the space- and time-resolved
dynamics of plasmas generated by collinear double pulses [7.21].
The regression lines of Fig. 7.12 can be described by the following relation:
bi .p/ D mi .p/Œdf df 0 .p/; (7.2)
where bi .p/ is the number of bursts; i D 1 for single pulses, i D 2 for double
pulses; mi .p/ is the slope of regression line; df is the foil thickness; and df 0 .p/ is
the extrapolated foil thickness for bi .p/ D 0.
While the slope depends on the burst mode i , the quantity df 0 .p/ is independent
of i , cf. Fig. 7.12 for Eb = 160 mJ. Assuming that the latter holds also for a burst
energy of 80 mJ, we see from (7.2) that the ratio of the ablation rates defined by
a2 =a1 D m1 =m2 equals the ratio b1 =b2 for df D const. Figure 7.14 shows the ratio
of the ablation rates a2 =a1 as a function of the pressure determined for a steel foil
with a thickness of df D 50 m.
At 1,013 mbar, the ablation rate for double pulses is by a factor of 3.6 (burst
energy of 160 mJ) to 4.7 (80 mJ) higher than the ablation rate for single pulses. For
pressures <100 mbar, the ratio of ablation rates approaches a value of 2 which is
equal to the ratio of pulse numbers. In this pressure regime, the effect of double
pulses is merely an additive one, i.e., the conditions generated by the first pulse
of the double pulse have no significant influence on the interaction of the second
7
burst energy
6
ratio of ablation rates a2 /a1
160 mJ
5 80 mJ
0
0.1 1 10 100 1000
pressure p [mbar]
Fig. 7.14 Ratio of ablation rates of double pulses to single pulses as a function of ambient air
pressure for a foil thickness of 50 m and two burst energies
References 117
pulse irradiated 6 s later. Hence, the two pulses of a double pulse burst can
be considered to cause independent interaction processes. However, in the range
100 mbar < p < 1; 013 mbar, the situation is completely different; here the ratio
of ablation rates shows a strong variation with changing ambient pressure. For the
majority of data points in Fig. 7.14, we find a2 =a1 > 1. The only exception occurs
for a burst energy of 160 mJ and pressures of 250 and 500 mbar. In this case, the
ratio a2 =a1 is less than 1, i.e., the ablation rate for single pulses is higher than the
one for double pulses. At a pressure of 700 mbar and a burst energy of 80 mJ, the
ratio of ablation rates achieves a maximum value of 6, i.e., the efficiency of ablation
is increased by a factor of 6 using double pulses instead of single pulses of the
same energy. The position of the maximum is assumed to depend on the interpulse
separation, the burst energy, and the pulse height ratio within a double-pulse burst.
References
7.1. C. Lorenzen, C. Carlhoff, U. Hahn, M. Jogwich, Applications of laser-induced emission

spectral analysis for industrial process and quality control. J. Anal. At. Spectrom. 7, 1029–
1035 (1992)
7.2. R. De Young, W. Situ, Elemental mass spectroscopy of remote surfaces from laser-induced
plasmas. Appl. Spectrosc. 48, 1297–1306 (1994)
7.3. E. Piepmeier, D. Osten, Atmospheric influences on Q-switched laser sampling and resulting
plumes. Appl. Spectrosc. 25, 642–652 (1971)
7.4. A. Felske, Über einige Erfahrungen bei der Makrospektralanalyse mit Laserlichtquellen.
Spectrochim. Acta 27B, 1–21 (1972)
7.5. F. Leis, W. Sdorra, J. Ko, K. Niemax, Basic investigations for laser microanalysis: I. Optical
emission spectrometry of laser-produced sample plumes. Mikrochim. Acta II 98, 185–199
(1989)
7.6. R. Jowitt, Direct analysis of liquid steel by laser, Proceedings of 38. Chemistry Conf., BSC
Teesside Laboratories, 1985, 19–29
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D Appl. Phys. 28, 2181–2187 (1995)
7.8. A. Semerok, C. Chaléard, V. Detalle, J. Lacour, P. Mauchien, P. Meynadier, C. Nouvellon,
B. Sallé, P. Palianov, M. Perdrix, G. Petite, Experimental investigations of laser ablation
efficiency of pure metals with femto, pico and nanosecond pulses. Appl. Surf. Sci. 138–139,
311–314 (1999)
7.11. J. Vrenegor, R. Noll, V. Sturm, Investigation of matrix effects in LIBS plasmas of high-alloy
steel for matrix and minor elements. Spectrochim. Acta B 60, 1083–1091 (2005)
7.12. J. Aguilera, C. Aragón, F. Penalba, Plasma shielding effect in laser ablation of metallic
samples and its influence on LIBS analysis. Appl. Surf. Sci. 127–129, 309–314 (1998)
7.13. J. Aguilera, C. Aragón, A comparison of the temperatures and electron densities of laser-
produced plasmas obtained in air, argon, and helium at atmospheric pressure. Appl. Phys.
A 69, S475–S478 (1999)
7.14. I. Horn, D. Günther, The influence of ablation carrier gasses Ar, He and Ne on the particle
size distribution and transport efficiencies of laser ablation-induced aerosols: implications
for LA–ICP–MS. Appl. Surf. Sci. 207, 144–157 (2003)
7.15. W. Sdorra, K. Niemax, Basic investigations for laser microanalysis: III. Application of
different buffer gases for laser-produced sample plumes. Mikrochim. Acta 107, 319–327
(1992)
7.16. R. Russo, Laser ablation. Appl. Spectrosc. 49, 14A–28A (1995)
7.17. M. Guillong, I. Horn, D. Günther, Description and characterization of a homogenized high
power 266 nm Nd:YAG laser ablation systems for LA-ICP-MS. J. Anal. At. Spectrom. 17, 8–
14 (2002)
7.18. A. Löbe, J. Vrenegor, R. Fleige, V. Sturm, R. Noll, Laser-induced ablation of a steel sample
in different ambient gases by use of collinear multiple laser pulses. Anal. Bioanal. Chem.
385, 326–332 (2006)
7.19. C. Wegst, Stahlschlüssel 2001, Marbach, Stahlschlüssel Wegst GmbH, 19. Auflage
Acta B 57, 121–135 (2002)
7.21. R. Noll, R. Sattmann, V. Sturm, S. Winkelmann, Space- and time-resolved dynamics of
plasmas generated by laser double pulses interacting with metallic samples. J. Anal. At.
Spectrom. 19, 419–428 (2004)
7.22. J. Vadillo, J. Romero, C. Rodriguez, J. Laserna, Depth-resolved analysis by laser-induced
breakdown spectrometry at reduced pressure. Surf. Interface Anal. 26, 995–1000 (1998)
7.23. Y. Iida, Effect of atmosphere on laser vaporization and excitation processes of solid samples.
Spectrochim. Acta B 45, 1353–1367 (1990)
7.24. S. Harilal, C. Bindhu, M. Tillack, F. Najmabadi, A. Gaeris, Internal structure and expansion
dynamics of laser ablation plumes into ambient gases. J. Appl. Phys. 93, 2380–2388 (2003)
7.25. J. Heitz, J. Gruber, N. Arnold, D. Bäuerle, N. Ramaseder, W. Meyer, J. Hochörtler, In-situ
analysis of steel under reduced ambient pressure by laser-induced break-down spectroscopy.
Proc. SPIE 5120, 588–595 (2003)
7.26. S. Lui, N. Cheung, Resonance-enhanced laser-induced plasma spectroscopy: ambient gas
effects. Spectrochim. Acta B 58, 1613–1623 (2003)
167 (2007)
Chapter 8
Plasma Dynamics and Plasma Parameters
In this chapter, investigations are presented to clarify the underlying physical

processes of the observed improvements achieved with collinear double pulses
studying the space- and time-resolved dynamics of the plasmas generated by laser
double pulses interacting with metallic samples, as well as the plasma state in terms
of electron density and temperature.
8.1 Expansion and Decay of the Plasma
The experimental methodology is based on high-speed framing and streak images to

observe the spatial and temporal development of the plasma using the optical emis-
sion of the plasma [8.1]. Furthermore, the spatiotemporal changes of the refractive
index of the plasma are investigated using a Mach–Zehnder interferometer, where
the shifts of the interference fringes are detected with a high-speed electro-optic
camera.
The experimental setup for the time-resolved photography of the luminous laser-
induced plasma is shown in Fig. 8.1. A Nd:YAG laser (Lumonics, model HY
1200, modified) with multipulse option, operated at 10 Hz and 1,064 nm emission
wavelength is used to generate the plasmas on the surface of a plane metallic sample
in ambient air. Details of the laser are described in [13.4]. For the experiments
presented in this chapter, the plasma dynamics resulting from the laser–sample
interaction with single and double Q-switched Nd:YAG laser pulses is investigated.
The pulse width of a single pulse amounts to 15 ns, whereas each pulse of a
double pulse has a width of 25 ns. In the case of double pulses, the energy
ratio adjusted is 1:1. The interpulse separation of the double pulses amounts to
t D 6 s. The pulses are focused by a planoconvex lens of 200 mm focal
length on pure iron targets (Fe concentration > 99:99%). The beam waist of the
laser focus lies inside the sample, 5 mm beneath its surface (s D 5 mm). The
diameter of the laser beam on the surface of the sample is 280 m. The pulse
energies are adjusted by use of a Glan–Thompson polarizer and measured with

120 8 Plasma Dynamics and Plasma Parameters
GT
M M
Nd:YAG Laser
M CL
LM SC
P
S OA
Fig. 8.1 Experimental setup for high-speed photography of the luminous laser-induced plasma.
M D mirror, GT D Glan–Thompson polarizer, CL D volume absorbing disc calorimeter, L D
lens, P D laser-induced plasma, S D sample, OA D optical axis of streak camera, LM D long
distance microscope, SC D streak camera
a volume absorbing disc calorimeter (Scientech 38–0101). The optical axis of

the streak camera is oriented perpendicularly to the propagation direction of the
laser beam irradiating the sample. The laser-induced plasma is imaged by a long-
distance microscope on the streak camera (Hadland, model Imacon 790) with a
S 20 photocathode having the dimensions 9 8 mm2 . The distance between the
laser-induced plasma and the front end of the microscope is about 160 mm. The
streak camera can be operated in framing mode and in streak mode. In streak mode,
the slit of the streak camera is oriented collinearly to the propagation axis of the
irradiated laser beam. The magnification object-to-image was determined using a
line graticule placed at the position of the laser propagation axis just above the
sample and observed with the streak camera in the focus mode. The experiments
are carried out under normal atmospheric conditions.
The experimental setup used for the interferometric investigations is shown
schematically in Fig. 8.2. The components to generate the laser-induced plasma
are identical to Fig. 8.1, but the sample is now located in a gas-tight measuring
chamber allowing to perform experiments with air pressures of p0 D 100 and
1,013 mbar. The Mach–Zehnder interferometer is oriented in such a way that one
arm passes through the interaction volume above the sample surface perpendicular
to the irradiated Nd:YAG laser beam. The radiation source for the interferometry is
a cw argon ion laser (Coherent, model Innova 100–10) operated at an emission
wavelength of 514.5 nm. The interference fringes are observed with the streak
camera. To avoid damage of the streak camera due to continuous illumination by
the argon laser beam, a mechanical shutter and an acousto-optic modulator (AOM)
are used to extract pulses of 10 s duration of the argon laser beam synchronized
to the triggering of the Nd:YAG laser. The first-order diffracted beam of the AOM
8.1 Expansion and Decay of the Plasma 121
M1 M2
GT
Nd:YAG
Laser BS2
BS1
Argon Ion CL
Laser M5 M6
L1
z
P x
y
BS3
M9
SH L3 M7 SC
S
L2 MC IF
EF
~67° L4 L5
AOM
M8 M10
BS4
M3 M4
A
Fig. 8.2 Experimental setup for the interferometric measurement of the laser-induced plasma.
SH D shutter, AOM D acousto-optic modulator, A D aperture, BS1 to BS4 D beam splitters,
MC D measuring chamber, EF D edge filter, IF D interference filter. For further abbreviations see
Fig. 8.1
is used for the interferometer, whereas the zeroth and higher orders are blocked
by an aperture (see A in Fig. 8.2). The telescope (L2 and L3 in Fig. 8.2) serves to
adjust the diameter of the illuminated area at the location of the plasma. To cover
the total plasma geometry during the expansion phase, an illumination diameter of
20 mm was chosen. The second telescope (L4 and L5) and the internal optic of
the streak camera form an image of the plasma on the photocathode. The beam
deviation of the argon laser beam due to refractive index gradients in the interaction
region of the Nd:YAG laser beam may cause uncertainties in the Abel inversion of
the interferograms; however, accurate results are achieved by proper imaging of the
plasma plane on the photocathode of the streak camera. The interference filter has a
maximum transmission at the wavelength of the argon ion laser and a spectral width
of 10 nm (FWHM). The edge filter (Schott, type OG 515) attenuates additionally
the intense emission of the laser-induced plasma.
Nd:YAG laser, shutter, AOM and streak camera are synchronized using a
personal computer as master trigger and three digital delay generators. Before the
start of each measurement, a minimum of 50 flashlamp discharges of the laser is
activated to stabilize the pulse energy to a standard deviation of 2%.
The framing and streak images shown in the following are gained with the setup
of Fig. 8.1. Figures 8.3 and 8.4 show framing photographs of the evolving plasma
induced by a 2 40 mJ laser double pulse in air at 1 bar. The streak camera is
operated with a framing rate of 2 107 frames/s, i.e., the temporal separation
between subsequent frames amounts to 50 ns. The exposure time per frame is 10 ns.
The magnification and the setting of the aperture of the microscope are kept constant
to assure comparability of the recordings. Figure 8.3 shows the frames after the first
pulse of the double pulse. The first pulse reaches the sample between frame 1 and 2
(Numbers refer to the sequence of frames as indicated on the top left of Fig. 8.3.
Fig. 8.3 Framing images of the laser-induced plasma after the first pulse of a double pulse. Expo-
sure time 10 ns per frame, time between subsequent frames 50 ns. The time sequence of the frames
is shown on the top left with the numbers 1–10. The view detected by the frames is illustrated
schematically on the top right. The white horizontal line in each frame indicates approximately the
position of the sample surface. Laser parameters: 2 40 mJ, interpulse separation 6 s
Fig. 8.4 Framing images of the laser-induced plasma after the second pulse of a double pulse.
Experimental settings and arrangement of frames are the same as for Fig. 8.3
The arrangement of frames shown in Fig. 8.3 was chosen according to the usual
orientation of the time axis from left to right. The original arrangement of the
frames yielded by the streak camera is different.) Frames 2–4 are overexposed due
to the high luminosity of the plasma at this phase. Figure 8.4 shows the frames after
the second pulse of the double pulse. The second pulse of the double pulse reaches
the sample at the exposure of frame 1 (frame numeration analog to Fig. 8.3).
The plasma geometry and dynamics observed in Figs. 8.3 and 8.4 differ signifi-
cantly from each other. After the first pulse, a nearly hemispherical plasma geometry
appears staying in contact with the sample surface for times up to 400 ns after arrival
of the first laser pulse. The extension of the plasma in the direction of the laser
beam axis amounts to 1 mm (frame 5 corresponding to an elapsed time of 150
ns after the first visible plasma emission in frame 2). After the second pulse, the
plasma emission starts at the sample surface (see frame 1 in Fig. 8.4). This is a clear
indication that the second laser pulse primarily interacts with the sample surface
and not with a residual plasma originating from the first pulse. The plasma then
splits into two regions, where one region lasts at the sample surface and disappears
after about 100 ns. The other region attains a layer-like shape, propagating toward
the incoming laser beam. At 150 ns after the irradiation of the second pulse of
the double pulse, an inhomogeneous distribution of the luminosity in this layer is
observed looking like a turbulent structure which tends to decrease at later times.
After 200 ns, the expansion toward the incoming laser beam stops. After 350 ns,
the luminous plasma starts to expand in the backward direction toward the sample
surface. At 450 ns, the plasma shows a more homogeneous luminosity distribution
and its shape differs significantly from the one observed in the time interval between
150 and 250 ns. The plasma has nearly reached the sample surface again.
The further temporal development can be observed by streak photographs as
shown in Fig. 8.5. The observation geometry is the same as for the Figs. 8.3 and
8.4. The streak velocity amounts to 50 ns/mm on the screen of the streak camera.
Fig. 8.5 Streak photographs

after first and second laser
pulse of a double pulse.
Experimental parameters are
the same as for Fig. 8.3
The time scale runs from left to right (In the original streak images delivered from
the streak camera, the time scale runs from right to left. These images are presented
here side-inverted according to the usual direction of the time axes). The white line
illustrates the position of the surface of the sample. The luminous regions beneath
the white line are due to an overexposure of the streak images for times close to
the arrival of the laser pulse. The upper image shows the laser-induced plasma
after the first pulse, and the lower image shows the plasma after the second pulse.
After the first pulse, the plasma stays in contact with the sample surface, while
the boundary of the luminous plasma expands to about 1 mm above the sample, as
observed in the framing images of Fig. 8.3. The luminous layer evolving after the
second pulse starts at the sample surface and expands up to a distance of about 2 mm
from the sample surface. After 300 ns, the front of the luminous layer changes the
direction and propagates back toward the sample. Approximately 600 ns after the
irradiation of the second laser pulse, the luminous front reaches again the sample
surface (marked by a white triangle in the lower image of Fig. 8.5). At the surface, a
partial reflection of the plasma front is visible. The observed dynamics of the plasma
expansion after the second pulse is a clear indication that the first laser pulse has led
to a transient modification of the ambient particle density above the sample thus
enabling a greater expansion of the sample material ablated by the second pulse.
The observed reflection-like behavior of the luminous front at about 300 ns after the
second laser pulse may be caused by a radially increasing particle density in the
surrounding atmosphere as a result of the interaction of the first laser pulse. Streak
photographs taken with a greater aperture of the microscope show that for times
beyond 1 s after the second laser pulse the luminosity of the plasma becomes more
and more homogeneous and the structure and geometry of the plasma stays nearly
constant for more than 5 s.
In the time range shown in Figs. 8.3–8.5, the emission spectrum of the plasma
changes from a continuum emission at early times to pronounced line emissions
after about 800 ns (see Chap. 6, Fig. 6.3). The exponential decay of the continuum
emission can be described by a decay time constant of about 90 ns corresponding to
the time range covered by the frames 2 and 3 of Fig. 8.3 and frames 1, 2, and 3 of
Fig. 8.4.
Framing and streak images of Figs. 8.4 and 8.5 show clearly that the dominant
effect of the second pulse is again an interaction with the sample surface and not a
re-heating of a residual plasma above the sample originating from the first pulse. In
a simplified approach, the nonlinear character of the interaction of the laser beam
with the residual plasma, which may lead to heating and ionization processes is not
considered. The absorption of laser radiation in a plasma can then be described by
the absorption coefficient using the formula (5.11) assuming ! and !p < !.
For this estimate, the absorption by microscopic particulates originating from
condensed iron vapor or solidified melt droplets thrown out of the interaction region
by the first pulse is neglected. The collision frequency is given by relation (5.12).
The electron–ion collision frequency ei can be estimated using the formula (5.13)
for the electron–ion momentum transfer in a plasma using the screened Rutherford
cross section and assuming a Maxwellian electron distribution. The electron-neutral
collision frequency is estimated taking the geometric cross section of neutral iron
atoms. With the following plasma parameters determined spectroscopically at a time
delay of 1 s after a single pulse of 80 mJ: electron density ne D 7 1016 cm3 ,
neutral iron density nFeI D 6 1016 cm3 , electron temperature Te D 8;300 K, we
obtain: ei D 1:3 1012 s1 , and en D 1:4 109 s1 . Using relation (5.11), the esti-
mated absorption coefficient due to ablated material originating from the first pulse
amounts to ˛ D 0:3 m1 . Taking as an upper limit a plasma dimension of 5 mm,
this corresponds to a transmission of more than 99% of the irradiated laser intensity
of the second pulse. In this case, a dominant part of the laser energy may propagate
to the sample surface again where it interacts with a preablated sample surface.
Hence, the plasma emission induced by the second pulse should start at the sample
surface, which is consistent with the experimental observations of Figs. 8.4 and 8.5.
The expansion of the laser-induced plasma can be described by the Sedov-
modelindexSEDOV-model for shock waves after a strong explosion in a homo-
geneous atmosphere [8.2, 8.3]. A dimensional analysis leads to the following
relationship for the expansion z of the shock wave as a function of time t:
1 2
z D 0 .E0 =0 / 2C t 2C ; (8.1)
where E0 is the energy deposited in a small volume, 0 is the density of the

homogeneous undisturbed ambient gas, and 0 is a function of the specific heat
ratio of the surrounding gas.
The parameter describes the geometry of the energy release and the shock wave
expansion with D 3 for a point-like energy release and a spherical shock wave
expansion and D 1 for a planar shock wave expansion. Hence, for a spherical
shock wave, the expansion should follow the relation z / t 0:4 and for a planar
expansion z / t 0:66 . The exponent of t is always expected to be twice that of E0 =0 .
To determine the expansion velocity of the luminous plasma with a higher
accuracy, streak images of the expanding plasma are taken with streak velocities of 1
and 5 ns/mm, while keeping the other experimental parameters constant. Figure 8.6
shows a double-logarithmic plot of the determined expansion after the first and the
second laser pulse of a double pulse. The time zero was chosen at the start of the
plasma emission detected in the streak images. The error bars in the z-direction
indicate the reading error in determining the front position on the enlarged streak
images. Uncertainties on the time axis are caused by deviations from the nominal
streak velocity which are specified to be less than 5% and by reading errors. The
corresponding error bars are not shown in Fig. 8.6, since they are significantly
smaller than those in the z-direction. After the first pulse, two regions of different
dynamics can be distinguished. For times up to 25 ns after the start of the interaction
of the first laser pulse, the data points can be fitted by a linear regression described
by the function z / t .0:76 ˙ 0:02/ . A linear fit to the data points after 25 ns yields
z / t .0:44 ˙ 0:04/ . Under the assumption that the luminous plasma front coincides
with the position of the shock front at the early expansion phase, we can compare
the determined dynamics with the prediction of Sedov’s model. For times 25 ns
Sedov’s model cannot be applied, since the laser pulse still persists and couples
1. laser pulse
end of 1. laser pulse
1
z [mm]
z ~t (0.44 ± 0.04)
z ~t (0.76 ± 0.02)
0.1
1 10 100
2. laser pulse
1 z ~t (0.67 ± 0.02)
z [mm]
0.1
1 10 100
t [ns]
Fig. 8.6 Height z of the luminous plasma front above the sample surface as a function of time
determined from side-on streak images. Top: after the first laser pulse, bottom: after the second
laser pulse. Experimental parameters are the same as for Fig. 8.3
energy into the interaction region. Beyond 25 ns, the determined exponent is close
to the theoretical value for a spherical expansion of a shock front.
For the second laser pulse, the determined data points are shown in the lower
diagram of Fig. 8.6. Applying a single fit using the function z / t ˇ yields an
exponent of ˇ D 0:67 ˙ 0:02 which is consistent with a planar shock front
expansion. (A close look to the data points in the lower diagram of Fig. 8.6 shows
that there might be a small change in the dynamics after a time of about 60 ns.
Applying the fit only to the data points with t > 60 ns yields ˇ D 0:53 ˙ 0:02.
However, the change in the dynamics after the second pulse at a time of about
60 ns is significantly smaller than the one observed at about 25 ns after the first
laser pulse.) These values indicate different expansion geometries for the plasmas
induced by the first and the second laser pulse and are in qualitative agreement with
the results of the framing images shown in Figs. 8.3 and 8.4.
The derivatives of the regression curves of Fig. 8.6 yield the expansion velocities
of the luminous plasma front in the z-direction. In the time interval from t D 30 ns
to t D 60 ns after the first laser pulse, this velocity decreases from 0:8 106 to
0:6 106 cm=s (corresponding to Mach numbers of 23 and 17), whereas in the same
time interval after the second laser pulse the expansion velocities are 1:5 106 and
1:2106 cm=s, respectively. Hence, after the second pulse the expansion is by about
a factor of 2 faster than after the first pulse.
The experimental setup shown in Fig. 8.2 enables the observation of changes
of the refractive index n caused by the expanding plasma in the measuring arm of the
Mach–Zehnder interferometer. In a first step, without the laser-induced plasma,
the mirrors of the interferometer are tilted to adjust the number and orientation of
the interference fringes. A fringe orientation parallel to the sample surface is chosen.
The slit of the streak camera crosses these fringes perpendicularly. Figure 8.7 shows
streak images of the interferograms for a laser double pulse with 2 80 mJ, t D
6 s and s D 5 mm. The sample surface is located at z D 0 mm. The expanding
plasma and shock front leads to a displacement of the interference fringes. After
the first laser pulse, a disturbance of the fringe pattern propagates in positive
z-direction. At the leading edge of this disturbance, a strong displacement of the
interference fringes toward the sample surface is observed. Reference measurements
with glass plates in the measuring arm of the interferometer show that this direction
of fringe displacement corresponds to an increased index of refraction. Hence, the
displacement of the fringes observed at the leading edge of the expanding fringe
disturbance is attributed to an increased particle density caused by the shock wave.
After the second laser pulse, an analog displacement of the interference fringes
is observed at the leading edge of the evolving disturbance in the fringe pattern
indicating the formation of a second shock front propagating behind the shock front
induced by the first laser pulse. The reflection-like behavior of the luminous plasma
observed in Fig. 8.5 at about 300 ns after the impact of the second pulse does not
show up in a corresponding structure of the fringe patterns observed in Fig. 8.7.
Fig. 8.7 Interferograms registered with the streak camera after the first and second laser pulse of
a double pulse. Laser parameters: 2 80 mJ, interpulse separation 6 s, p0 D 1; 013 mbar
A potential reason might be a too small change in the refractive index caused by the
back-traveling plasma, which cannot be resolved by the setup used.
For a comparison of the experimental data with the Sedov model, the expansion
z of the shock front as a function of time t is shown in a double-logarithmic plot in
Fig. 8.8. The distortion of the interference fringes can be detected clearly for more
than 12 s after the laser pulse interaction (see Fig. 8.7 bottom, shock front induced
by the first laser pulse at about z D 7 mm) in contrast to the luminosity of the plasma,
which can be observed for a few microseconds only. The exponent describing the
expansion of the shock front in the time range from 200 ns to 12 s after the first
pulse amounts to 0:45 ˙ 0:01. This result agrees with the exponent determined from
the streak images of the luminous plasma at the early times between 30 and 65 ns
after the first laser pulse, cf. Fig. 8.6. The time exponent of the shock front expansion
after the second laser pulse is 0:35 ˙ 0:04 and hence is significantly smaller than the
one determined from the luminous front at early times, cf. Fig. 8.6.
For the determination of the time exponent for the shock front propagation after
the first laser pulse, functions of the following form are fitted to the measuring data:
z D Ata : (8.2)
10
1. laser pulse
z [mm]
z~t (0.45 ± 0.01)
1
10
2. laser pulse
z [mm]
z ~t (0.35 ± 0.04)
1
0.1 1 10
t [μs]
Fig. 8.8 Height z of the shock front above the sample surface as a function of time determined
from interferograms observed in side-on direction with the streak camera, cf. Fig. 8.7. Top: after
the first laser pulse, bottom: after the second laser pulse
Since the shock front propagates into an undisturbed ambient atmosphere, the
preconditions of Sedov’s model are fulfilled and a comparison of (8.2) with (8.1)
yields the following relations: a D 2=.2 C / and A D 0 .E0 =0 /a=2 . For the
latter parameter holds also the relation: A / .E0 =p0 /a=2 , where p0 is the ambient
pressure, which can be changed in the experiment, cf. Fig. 8.2. In the following
paragraphs, it is assumed that the energy E0 deposited locally is proportional to
the irradiated laser pulse energy E1 (E1 denotes the pulse energy of a single pulse
or the pulse energy of the first pulse of a double pulse). The fit parameters A
and a are determined for measurements with the following sets of parameters: (a)
single laser pulses with E1 D 20, 40, 80, 120, and 160 mJ at p0 D 1 bar ambient
pressure, (b) single laser pulses with E1 D 40 mJ at ambient pressures of p0 D
100, 500, and 1,013 mbar. For these parameters, the determined exponent a varies
between 0.36 and 0.50 with an average value of aN D 0:43 ˙ 0:05. This value
agrees with the expected exponent of 0.44 for a spherical shock front propagation
according to Sedov’s model. Relation (8.2) describes well the dependence of the
shock wave position as a function of time for distances of up to 7 mm from the
sample surface, which is still below the critical radius rc D .E=p0 /1=3 , which
amounts to rc D 9:3 mm for E D 80 mJ and p0 D 1;013 mbar [8.3]. At rc , the
counterpressure of the ambient gas can no longer be neglected and the self-similar
solution of the problem of a strong explosion no longer holds.
Figure 8.9 shows the values of the fit parameter A determined for the parameter
sets described above in a double-logarithmic plot. The linear regression shown
in Fig. 8.9 yields: A / .E1 =p0 /0:23 ˙ 0:02 . As expected by the Sedov model, the
exponent is about half of a. N Hence, the dynamics of the shock front propagation
after the first laser pulse can be described in a wide range of laser pulse energies and
ambient pressures by Sedov’s model for a spherical expansion.
4
fit parameter A [a.u.]
A ~ (E1 / p0) (0.23 ± 0.02)
1
10 100
E1 / p0 [mJ / mbar]
Fig. 8.9 Fit parameter A as a function of the ratio of the laser pulse energy E1 and the ambient
pressure p0 for the first laser pulse
For the description of the expansion of the shock wave after the second laser
pulse, in analogy to relation (8.2), the following function is used: z D Btb . Since the
ambient atmosphere is not homogeneous for the second shock wave, the parameters
B and b cannot be interpreted with respect to (8.1). The following measuring
parameters were used to determine B and b: (a) double pulses with E1 D E2 (E2
denotes the pulse energy of the second pulse of a double pulse) and E2 D 20, 40,
60, 80, and 100 mJ at p0 D 1;013 mbar ambient pressure, (b) double pulses with
E1 D E2 D 40 mJ at ambient pressures of p0 D 100, 500, and 1,013 mbar. The
determined fit parameter b varies between 0.27 and 0.41, with an average value of
bN D 0:32 ˙ 0:05.
Figure 8.10 shows the determined parameter B as a function of the laser pulse
energy of the second pulse E2 at a pressure of p0 D 1;013 mbar (pressure in the
ambient atmosphere before the impact of the first pulse). A regression curve yields:
B / .E2 /0:30 ˙ 0:02 . The parameter B is not plotted as a function of .E2 =p0 / since
there is no homogeneous pressure distribution in the interaction region at the time
the second laser pulse is irradiated onto the sample. The determined exponent is
approximately equal to b. N This behavior differs significantly from the shock wave
propagation after the first laser pulse.
The propagation of shock waves in an inhomogeneous atmosphere was described
by Sedov assuming the following relation for the density variation of the ambient
atmosphere by a power law [8.2]:
D .r/ D Dr' ; (8.3)
where D is a positive constant, r is the radius, and ' is a dimensionless constant.

The propagation of a spherical shock wave as a function of time is then given by the
following power law:
r / t 2=.5'/ : (8.4)
Taking the experimentally determined exponent bN for the shock wave after the
second pulse and relation (8.4), we obtain ' D 1:25. Equation (8.3) then yields
.r/ / r 1:25 , which describes a density increasing radially from the center. While
5
fit parameter B [a.u.]
B ~ E2(0.30 ± 0.02)
Fig. 8.10 Fit parameter B as

2
a function of the energy E2 of 10 100
the second laser pulse at a
pressure of p0 D 1;013 mbar E2 [mJ]
the first shock wave propagates into a spatially homogeneous ambient atmosphere,
the second shock wave encounters – as a result of the interaction of the first laser
pulse – a locally rarefied ambient atmosphere, allowing for higher propagation
velocities of the second shock wave.
So far the expansion of the plasma from a flat surface of the sample was studied.
If a series of laser pulses is irradiated on the same spot on the surface of a sample,
then a crater is formed, whose geometry influences the plasma expansion and
dynamics. Instead of a free expansion into a half space the plasma is partially
confined by the crater walls formed by the preceding laser pulses. Figure 8.11
illustrates this change in the plasma dynamics with side-on images of the luminous
plasma. The exposure time of the images taken with an intensified high-speed CCD-
camera was set to 50 ns. The position of the sample surface is marked by the
white vertical line on the left of the images, and the laser beam is irradiated from
the right. The experimental parameters were: single pulses, 160 mJ, s D 5 mm,
tdelay D 5 s, ambient atmosphere: air. Figure 8.11a is taken for a flat sample surface
and Fig. 8.11b in case of a crater formed by preceding laser pulses having a diameter
of about 1 mm and a depth of 0.9 mm. The change in the plasma geometry is clearly
visible. Obviously the crater effects a confinement of the plasma leading to faster
expansion toward the incoming laser beam. The images were taken at the same
delay time tdelay showing that the plasma volume, the center of the plasma, and the
expansion velocity are changed depending on the two considered cases of a plasma
induced on a flat surface or inside a crater.
The partition of the laser burst energy in a collinear double pulse studied so far
was 1:1, i.e., each single pulse of the double pulses has the same energy. A change
in this partition has an influence on the plasma dynamics as well. Figure 8.12 shows
side-on images of the luminous plasma at different delay times, where the three
images of column (a) are taken for a single pulse with 80 mJ, column (b) for a
symmetric double pulse with 2 40 mJ and column (c) for an asymmetric double
pulse with E1 D 10 mJ and E2 D 70 mJ. While the single pulse causes the largest
plasma in the beginning (for tdelay D 0 s), it decays faster leaving a small emitting
volume after 0:5 s which is still in contact with the sample surface. In case of a
Fig. 8.11 Side-on images of

the luminous plasma evolving
from the sample surface
(white vertical line on the
left) for different local sample
surfaces. The laser beam is
irradiated from the right.
(a) Flat sample surface,
(b) surface with a crater
formed by preceding laser
pulses. For further details see
text
Fig. 8.12 Comparison of the

plasma emission for (a) a
single pulse of 80 mJ, (b) a
symmetric double pulse with
2 40 mJ, and (c) an
asymmetric double pulse with
E1 D 10 mJ and
E2 D 70 mJ. The numbers in
the upper right corner of each
image show the delay time
tdelay in microseconds of the
start of the exposure with
respect to the last pulse
irradiated
symmetric double pulse shown in column (b) of Fig. 8.12, the plasma dimensions
are smaller in the beginning and after 0:5 s a spherical plasma is visible which
is clearly detached from the sample surfaces and shows a volume being greater
than the one in case (a) at the same delay time. For the asymmetric double pulse,
the energy of the second pulse is significantly greater than in case (b); hence, the
plasma is larger and emits stronger in the beginning. At later times, a spherical
plasma is formed covering an even larger volume than in case (b) and still staying in
contact with the sample surface at 0:5 s. Furthermore, the plasma generated with
asymmetric double pulses shows the strongest emission after 0:5 s compared with
the other cases.
The exposure time itself of each frame was kept constant and amounts to 50 ns.
The further measuring parameters are: s D 5 mm, t D 6 s. The laser is
irradiated from the top, and the black horizontal line marks the position of the
sample surface.
This behavior shows that a changed partition of the burst energy among the two
pulses of a collinear double pulse allows to optimize the plasma dynamics with
respect to reduced decay times, enlarged plasma volumes, and increased emission
features.
8.2 Spatially Integrated Electron Density and Temperature
In order to investigate the effects of single- and double-pulse bursts on the laser-
induced plasma, spatially and temporally integrated emission intensities, electron
temperatures, and electron densities are measured. The time interval for integra-
tion, shown in Fig. 6.3, is chosen in accordance with the needs of quantitative
multielement analysis with LIBS, namely high line-to-background ratio of the
element-specific radiation. The delay of 0:8 s with respect to the peak emission
8.2 Spatially Integrated Electron Density and Temperature 133
excludes the contribution of the continuum radiation. The width of the integration
interval of 2 s is in the range of the decay times of the line radiation, see Fig. 6.3.
Hence, the temporal integration covers the major part of the line emission.
To determine the electron temperature and density, iron emission lines in the
spectral region around 536 nm were selected (the other experimental parameters are
described in Sect. 6.1). The plasma radiation is collected by an optical fiber bundle
without additional optical elements. Due to the numerical aperture of the fiber
bundle, the received radiation is spatially integrated over the whole emitting plasma
volume. The fiber bundle is divided into four arms with about 40 fibers each. A fiber
bundle output having a rectangular cross section is mounted at the entrance slit of a
Czerny–Turner spectrometer (Jobin Yvon HR 640) with 640 mm focal length, 2,400
lines per millimeter grating and an intensified photodiode array detector (Princeton
Instruments). The spectral resolution is 0.045 nm at 540 nm. The spectra of the 81st
to the 100th pulse on a ground surface are accumulated to improve the signal-to-
noise ratio. The electron temperature is deduced from Boltzmann plots, assuming
that the plasma is in local thermal equilibrium as stated by Adrain et al. [8.4], and
the electron density by measuring the line width of a Stark-broadened emission
line [8.5]. Spectral line selection methods for Boltzmann plots are described in
Sect. 19.6. The Fe lines used for the evaluations are listed in Table 8.1.
Figure 8.13 shows the whole spectrum detected simultaneously covering the
spectral range of the lines listed in Table 8.1. Figure 8.14 shows a part of this
spectrum around the iron line Fe I 538.34 nm with the data points and fit curves.
The intensities of the lines are determined by a fit of Lorentzian curves to the
measured spectrum. The uncertainty in the electron temperature measurement is
estimated by the evaluation of the temperature from the maximum and minimum
slopes of straight lines through the error-affected measuring points in the Boltzmann
plot. Figure 8.15 shows the Boltzmann plot for the spectra of Fig. 8.13. The quantity
shown on the ordinate is:
I
y D ln cB (8.5)
Anm gn
with I measured intensity of a spectral line assumed to be proportional to the
emission coefficient, cf. (9.3) in Chap. 9, unit: 1; wavelength, Anm Einstein
coefficient of the transition n ! m, gn statistical weight of the upper level, cB
Table 8.1 Wavelength , (nm) En (eV) Anm (108 s1 ) gn

excitation energy En , Einstein
coefficient Anm , and statistical 532.418 5.54 0.15 9
weight gn of the upper level n 532.804 3.24 0.0115 7
of the selected iron lines for 537.150 3.27 0.0105 5
the determination of the 538.337 6.62 0.59 13
electron temperatures from 539.713 3.21 0.00259 9
Boltzmann plots 540.578 3.28 0.0109 9
The width of the Fe line at 538.34 nm caused by Stark
broadening is used to evaluate the electron density
[8.6, 8.7]
Fe 532.80
Fe 532.42
3x104 1 x 80 mJ
Fe 537.15
2 x 40 mJ
Fe 540.58
Fe 538.34
I [counts]
2x104
Fe 539.71
1x104
0
532 534 536 538 540
λ [nm]
Fig. 8.13 Simultaneously detected spectral range, cf. Table 8.1. Measuring parameters: SP with
1 80 mJ and DP with 2 40 mJ, t D 6 s, tdelay D 0:8 s, tint D 2 s, air atmosphere at
1,013 mbar, s D 5 mm
20000
1 x 80 mJ
2 x 40 mJ
15000
Fe 538.34
I [counts]
10000
5000
0
536 537 538 539
λ [nm]
Fig. 8.14 Measured spectra (data points) and Lorentz fit curves (lines) in the spectral range of the
line Fe I 538.34 nm, measuring parameters are same as for Fig. 8.13
constant factor for a given species in a given ionization state and fixed process and
instrumental parameters, unit: m1 s1 .
The errors in each measuring point arise from the standard deviation of the line
intensities of 5–10% and the stated errors of the Einstein coefficients in the range of
30%. The resultant measurement uncertainty in the temperature is about 10%. The
repeatabiltiy in terms of standard deviation of the temperature values of subsequent
measurements determined from least square fits of a straight line to the measured
points in the Boltzmann plot amounts to less than 5%.
17
m = -1.215 ± 0.036
16 Te = (9540 ± 280) K
l n ( I λc B / A nm g n )
15
m = -1.395 ± 0.042
14 Te = (8310 ± 250) K
13
1 x 80 mJ
12 2 x 40 mJ
3 4 5 6 7
En [eV]
Fig. 8.15 Boltzmann plots of the spectra shown in Fig. 8.13 to determine the electron temperature
For the Boltzmann plots shown in Fig. 8.15, the determined slopes of the fit
curves yield the temperatures of the plasmas as follows: Te D 9;540 ˙ 280 K for
double pulses and 8;310 ˙ 250 K for single pulses (the errors given here refer to the
uncertainty in fitting the data points to straight lines).
The error in the electron density measurement is influenced by the uncertainty in
the spectral line fit and the errors in the Stark coefficients taken from Freudenstein
and Cooper [8.7]. Additionally, it has to be considered that the Stark broadening
constant stated in [8.7] depends on the temperature, cf. Sect. 10.3. The estimated
total error amounts to about 20%. The repeatability in terms of standard deviation
is 5% and less. The error bars in the following diagrams refer to the repeatability of
the electron temperature and density measurements.
Table 8.2 summarizes the measuring results for single and double pulses with a
burst energy of 80 mJ at a delay time tdelay D 0:8 s. The quantities listed in the
first column are: the intensity of the iron emission at 532.42 nm determined from a
Lorentz fit in arbitrary units, the electron temperature, the electron density, the total
number of iron particles (atoms and ions), the degree of ionization (r D ne =nFe /, the
total number of iron atoms in the neutral state (i.e., the number of Fe I atoms), the
Boltzmann factor and the total number of iron atoms in the neutral state populating
the excited state n.
The ratio of the measured intensities of Fe 532.42 nm for double pulses and single
pulses amounts to 3;120=1;410 2:2 ˙ 0:4. In the following, this value will be
compared with the expected ratio based on the values for Te and ne determined
spectroscopically and NFe . The simplifying assumptions for this calculation are:
complete atomization of the ablated material in the plasma, population of the excited
levels and the ionization can be described by the Boltzmann distribution, Saha
equation, respectively; the plasma is optically thin for the considered emission line.
In a first step, the degree of ionization r D ne =nFe is calculated using the Saha
Table 8.2 Spectroscopically determined values for Te , ne at a time of tdelay D 0:8 s and
calculated quantities using the experimentally determined mass ablation per burst
Burst mode, SP, 1 80 mJ DP, 2 40 mJ Remarks
burst energy
IFe
S
532:42 ; IFe 532:42
D
1; 410 ˙ 150 3;120 ˙ 300 Determined
(a.u.) spectroscopically
Te (K) 8; 310 ˙ 100 9;540 ˙ 100
ne (cm3 ) (6:8 ˙ 0:2) 1016 (6:3 ˙ 0:2) 1016
NFe [1] (1:4 ˙ 0:2) 1015 (3:5 ˙ 0:4) 1015 Deduced from ablated
mass per burst (cf.
Fig. 7.5, 130 ng for SP
and 330 ng for DP and
the molar weight of
iron 56 g/mol)
r [1] 0:56 ˙ 0:05 0:857 ˙ 0:007 Calculated quantities
NFe I [1] (6:2 ˙ 1:2) 1014 (5:0 ˙ 0:6) 1014
4
fB [1] (4:4 ˙ 0:4) 10 (1:19 ˙ 0:09) 103
11
NFe In [1] (2:7 ˙ 0:8) 10 (6:0 ˙ 1:2) 1011
nFe (cm3 ) 1:2 1017 7:4 1016
3 16
nFe I (cm ) 5:3 10 1:1 1016
3 17
nN (cm ) 6:8 10 6:2 1017
nN =nFe 5.6 8.4
Experimental parameters: see Fig. 8.13. For the error of the temperature the repeatability in terms
of the standard deviation of the temperature values determined under repeatability conditions is
taken
fB D exp.En =kT e / Boltzmann factor, NFe I n number of atoms in the upper state n; nFe , nFe I iron
and iron atom density; nN nitrogen density
equation. With the determined parameters Te and ne , the partition functions and
the lowering of the ionization energy follow r D 0:56 ˙ 0:05 for single pulses
and r D 0:857 ˙ 0:007 for double pulses. Furthermore, the Saha equations shows
that the second ionization stage is negligible, i.e., nFe III =nFe II < 103 ; hence, the
relation nFe II ne holds. With nFe I D nFe nFe II follows:
nFe I D nFe .1 r/: (8.6)
Assuming that this local relation holds for the whole plasma volume (which is
a strong simplification, since the plasma is not homogeneous in densities and
temperatures) allows to rewrite (8.6): NFe I D NFe .1 r/. With r and NFe , we
obtain NFeI D .6:2 ˙ 1:2/ 1014 for single pulses and NFeI D .5:0 ˙ 0:6/ 1014
for double pulses (see row no. 6 in Table 8.2).
The population of the upper level is given by the Boltzmann distribution for the
measured values of Te and the energy of the upper level En of the transition, i.e.,
5.54 eV for the Fe line 532.42 nm (cf. Table 8.1). The expected intensity ratio is:
En

IFe
D
532:42 NFe
D
Ie
kT D
e uS
D En
D 1:8 ˙ 1:0; (8.7)
IFe 532:42
S S
NFe I e e u
S kT D
where uS , uD is the temperature dependent partition function for single and double
pulses excitation, cf. Sect. 9.2, (9.3), (9.6). Within the error limits, this value
coincides with the experimentally determined intensity ratio of 2:2 ˙ 0:4.
The total particle density can be calculated with the values for ne and r: nt D
ne C nFe I C nFe II D .1 C 1=r/ne yielding 1:9 1017 cm3 for single pulses and
1:4 1017 cm3 for double pulses. With the equation of state for ideal gases, a
pressure inside the plasma of 220 hPa for single pulses and 185 hPa for double
pulses can be estimated. These values are about a factor of 5 below the ambient
pressure. If a pressure equilibrium with the ambient gas is assumed after a delay
time of 0:8 s, then there must be additional particles from the ambient atmosphere
inside the plasma with a density of 6:8 1017 cm3 for single pulses and 6:2
1017 cm3 for double pulses. These values are by a factor of 5.6 and 8.4 higher than
the respective iron particle density for single and double pulses. For the measured
electron temperature Te and electron density ne ; the degree of ionization of nitrogen
is below 1%; hence, the electron density is mainly a consequence of the ionization
of the iron particles. The approximation nFe II ne given above is still valid. The
density of particles in the ambient atmosphere amounts to 2:5 1019 cm3 . Hence,
more than 97% of the ambient atmosphere particles are displaced by the plasma
(cf. Annex, Sect. A.1).
The size of the plasma can be estimated on the basis of the spectroscopically
determined iron particle density nFe and the ablated mass assuming that the plasma
parameters and the iron particle density are homogeneous inside the plasma and
the ablated material is completely atomized in the plasma. For the iron particle
density follows nFe D 1:2 1017 cm3 for single pulses and nFe D 7:4 1016 cm3
for double pulses. Assuming a hemispherical plasma yields a radius of 1.8 mm
and 2.9 mm for single pulses and double pulses, respectively. Qualitatively this is
consistent with the observations shown in Fig. 8.5, where a greater extension of
the plasma after the interaction with a double pulse is detected compared with the
plasma extension for a single pulse. The plasma radii estimated here are greater than
the observed extension of the luminous plasma in the framing and streak images
of Figs. 8.3–8.5. However, we have to consider that the ablated mass used for this
estimation (cf. Fig.7.5) is an upper limit for the mass ablated for a semi-infinite
body (instead of a thin foil) and furthermore that not all particles ablated are excited
inside the plasma volume.
The temporal variation of the line intensities of Fe I 532.42 nm and Fe I
532.80 nm, the electron temperature, and electron density after the second pulse
of a double pulse is shown in Fig. 8.16. Since the integration time of 2 s is of the
order of the times estimated in the following, the latter are upper limits for the real
decay times, which have to be determined by a deconvolution procedure.
The line intensity of the Fe I 532.42 nm having an upper energy level of 5.54 eV
decays monotonously, whereas the line Fe I 532.80 nm with a smaller upper energy
level of 3.24 eV attains a maximum after 3 s. The measuring points of the
electron density can be fitted by an exponential curve with a time constant of 5:7 s.
The semi-logarithmic plot of the electron density shows that the temporal variation
1500
1250 Fe 532.42 nm
Fe 532.80 nm
I [counts] 1000
750
500
250
0
10000
9000
Te [K]
8000
7000
6000
10.0
n e [1016 cm- 3 ]
1.0
0.1
0 5 10 15 20
t delay [μs]
Fig. 8.16 Line intensity I of Fe 532.42 nm, Fe 532.80 nm, electron temperature Te and density
ne as a function of the time delay tdelay after the second pulse of a double pulses with 2 40 mJ
(tint D 2 s, t D 6 s, s D 0 mm; ambient atmosphere: air at normal pressure)
can be described by two time constants. Up to about 6 s, the time constant is
2:5 s, whereas for later times the time constant is about 30 s.
In the following, we compare the measured intensity of the Fe I 532.42 nm line
with the intensity to be expected from the measured plasma parameters. Table 8.3
shows the measured values for the delay times 1 and 20 s after the second pulse of
a double pulse.
The experimentally determined intensity IFe D
532:42 decays after 20 s to a value,
which is (18 ˙ 5)% of the initial value at 1 s. Assuming an optically thin plasma,
the line intensity is proportional to the total number of emitting particles. For the
estimation, we consider the following two simplifying cases of a homogeneous
plasma: (a) the total number of iron particles NFe is constant in the time period from
1 to 20 s and (b) the plasma volume contributing to the line intensity is constant.
Table 8.3 Spectroscopically determined values of IFe

D
532:42 , Te , ne at the delay times of 1 and
20 s and calculated quantities
tdelay .s/ 1 20 Remarks
IFe
D
532:42 (a.u.) 1;240 ˙ 150 220 ˙ 30 Determined spectroscopically
Te (K) 9;200 ˙ 100 6;600 ˙ 100
ne .cm3 / .5:5 ˙ 0:2/ 1016 .0:5 ˙ 0:02/ 1016
r [1] 0:86 ˙ 0:02 0:47 ˙ 0:05 Calculated quantities
fB [1] .9:1 ˙ 0:7/ 104 .5:8 ˙ 1:6/ 105
nFe I .cm3 / .9:0 ˙ 1:5/ 1015 .5:5 ˙ 1:7/ 1015
nFe .cm3 / .6:4 ˙ 0:4/ 1016 .1:0 ˙ 0:2/ 1016
nFe I D density of iron atoms, nFe D density of iron particles (atoms and ions), further abbre-
viations see Table 8.2
For case (a), we use NFe I D NFe .1 r/, where r is calculated with the Saha
equation, then the following relation holds:
Em

20 s 20 s
IFe 532:42 .1 r 20 s /e kT e u1s
1 s
D E
D 0:36 ˙ 0:08: (8.8)
IFe 532:42 1ms
.1 r 1 s / e kT e u20 s
The intensity of the line Fe 532.42 nm at 20 s is then (36 ˙ 8)% of the value at
1 s. Since the experimentally measured intensity decreases to 18%, the total
number of iron particles must have been reduced by condensation, diffusion out
of the radiating volume, or cluster formation to a value of about 50% of the initial
total number of iron particles at 1 s.
For case (b), a constant plasma volume, we take nFe I instead of NFe I :
Em
20 s 20 s kT 20 s
IFe 532:42 nFe I e e u1s
1 s
D D 0:06 ˙ 0:03 (8.9)
IFe 532:42 1 s E1ms
nFe I e kT e u20 s
showing a decrease to (6 ˙ 3)% of the initial value. This value is smaller than
the measured decrease to (18 ˙ 5)% which is an indication that the iron particles
are dispersed over a greater volume in the time interval from 1 to 20 s. Hence,
from (8.9) follows a volume increase by a factor of 3 to get an intensity decrease
consistent with the experimental findings.
Table 8.3 shows that the density of iron atoms decreases in the considered time
interval; however, this decrease is much smaller than the decrease of the total iron
particle density. This effect is attributed to the recombination of Fe ions.
The ratio of the line intensities IFe 532:80 .t/=IFe 532:42 .t/ (see Fig. 8.16 top) –
assuming optically thin emission – allows also to determine the temperature T .t/.
If we further assume that the plasma is homogeneous, then the intensity of an
iron atom line is proportional
to the totalnumber of emitting iron atoms NFe I [cf.
(9.3)]: IFe 532:42 / NFe I eEj =kT =u0Fe .T / . Knowing T .t/; IFe 532:42 .t/ allows then
Fig. 8.17 Electron density

and number of iron atoms as a
function of time. The electron
density data is the same as in
Fig. 8.16, bottom. The
number of iron atoms is
deduced from the intensities
of two iron atom lines (see
Fig. 8.16, top)
to calculate NFe I .t/. Figure 8.17 shows the electron density and the total number
of iron atoms as a function of time. The strong monotonous decay of the electron
density is accompanied by an increase in the number of iron atoms for delay times
up to 5 s. This increase is mainly fed by recombining iron ions.
The electron temperature decrease as a function of time shown in Fig. 8.16
is plotted in Fig. 8.18 again in a double logarithmic diagram showing that this
temperature decrease can be described by a power law as Te / t 0:11 . This decrease
is much smaller than a cooling expected for a pure adiabatic expansion of the plasma
volume, which is described by:
1
T1 V2
D ; (8.10)
T2 V1
where T1 ; V1 and T2 ; V2 are the temperature and volume of the plasma at time
t1 ; t2 and is the adiabatic coefficient. Taking the volume increase of a factor of
3 estimated above and the adiabatic coefficient of a monoatomic gas of 1.67 relation
(8.10) yields a temperature drop by a factor of 2.1 (see open square in Fig. 8.18),
which is much stronger than the observed decrease by a factor of 1.4. Obviously
the cooling of the plasma is reduced which is interpreted as a consequence of the
decreasing degree of ionization releasing recombination energy contributing to a
post-heating of the plasma.
Figure 8.19 shows the spatially integrated intensity of the Fe 532.42 nm line, the
electron temperature, and density for single and double pulses (interpulse separation
of double pulses t D 6 s, s D 5 mm) as a function of burst energy under
1,013 mbar atmospheric pressure [6.14]. The intensity of the iron line Fe 532.42 nm
is a factor of about two higher if a double pulse of 2 40 mJ is used instead of a
single 80 mJ pulse (cf. also Fig. 6.3). For single pulses, the intensity increases up to
pulse energies of about 70 mJ and remains nearly constant for higher energies. With
double pulses, the intensity grows monotonically over the investigated energy range
50–350 mJ.
Fig. 8.18 Temperature of the

laser-induced plasma after the
second pulse of a double
pulse as a function of time in
a double logarithmic diagram
(same data as Fig. 8.16,
middle) compared with the
temperature decrease
estimated for an adiabatic
plasma expansion with a
volume increase by a
factor of 3
The electron temperature determined for single pulses shows a saturation

behavior for energies greater than 40 mJ at 8,300 K; for double pulses at 160 mJ
at 10,200 K. The electron density is approximately equal for single and double
pulses with energies up to 150 mJ. For greater burst energies, it amounts to about
9 1016 cm3 for double pulses and (6–7) 1016 cm3 for single pulses.
The plasma parameters for double pulses of 2 40 mJ with variable interpulse
separations are shown in Fig. 8.20 [6.14]. The experimental parameters are the same
as for Fig. 8.19. Two limits were taken into account: (a) an interpulse separation
of zero corresponds to a single pulse of 80 mJ and (b) an infinite interpulse
separation corresponds to quasi-single pulses of 40 mJ (interpulse separation 33 ms,
corresponding to the burst repetition rate of 30 Hz).
Again, the Fe 532.42 nm intensity and the electron temperature show signif-
icantly higher values for double pulses than for single pulses. For the double
pulses, the electron temperature is approximately constant at about 9,300 K in the
investigated range of interpulse separations. The line intensity attains a minimum
for double pulses with interpulse separations of about 5–10 s. The behavior of the
electron density is similar to that of the line intensity, with the exception that the
electron density generated by a single 80 mJ pulse is higher than those for double
pulses with interpulse separations below about 15 s.
The influence of the focal position relative to the sample surface is shown in
Fig. 8.21 for single and double pulses of 80 mJ (interpulse separation of double
pulses 6 s, 1,013 mbar atmospheric pressure). The focusing lens is shifted for these
measurements, so the distance between sample surface and fiber optics remains
constant (see Sect. 6.1). For single pulses, the maximum of the Fe 532.42 nm
intensity is observed close to s 9 mm. For smaller or even negative s,
the line intensity drops to a very small value. This corresponds to Fig. 6.2, which
shows a plasma separated from the sample for negative s. In contrast to this,
6000 single pulses

double pulses, Δt = 6 μs
IFe 532.42 [counts]
4000
2000
11000
10500
10000
Te [K]
9500
9000
8500
8000
10
9
ne [1016 cm- 3]
5
0 100 200 300 400
Eb [mJ]
Fig. 8.19 The spatially integrated intensity of the Fe 532.42 nm line, electron temperature, and
density for single and double pulses as a function of the burst energy in the time period shown in
Fig. 6.3
the behavior for double pulses is nearly symmetric to s 1 mm, revealing

the symmetry of the intensity distribution of the laser beam with respect to the
focal plane. Three local maxima are observed: at s 1, 12, and 11 mm.
The maxima at s D 12 and 11 mm have nearly the same width. Obviously
at s D 12 and 11 mm, the intensity level at the target surface and the size of
the irradiated area induce a plasma dynamic which offers optimum conditions for
the emission of the iron line at 532.42 nm. The maximum at 12 mm is slightly
smaller than the one at 11 mm, which may be the consequence of a small shielding
effect of an atmospheric plasma in the case of s < 0 mm. The reason for the
central peak at s 1 mm is not yet clear. Its half width is about 2 mm, which
6000
IFe 532.42 [counts] 4000
2000 2 x 40 mJ
1 x 80 mJ
1 x 40 mJ
0
10000
9500
Te [K]
9000
8500
8000
9
ne [1016 cm-3]
6
0 20 40 60 80 inf.
Δt [µs]
density for double pulses of 80 mJ burst energy as a function of the interpulse separation
is comparable with the Rayleigh length of the laser beam of about 1 mm, evaluated
under the assumption of a Gaussian beam profile.
The electron temperature and density with single pulses attain absolute maxima
for s 3 and 7 mm, respectively, and local maxima at s 21 mm. For double
pulses, there are two maxima in Te , of which the maximum at s D 2–3 mm is
significantly greater than the one at s D 6 mm. Again this is an indication
that, for negative s, an atmospheric plasma evolves, which attenuates the laser
intensity at the sample surface and leads to reduced electron temperatures. Unlike
the Fe intensity, the temperature for double pulses is not symmetric with respect to
s –1 mm, but at s –12 and 11 mm, corresponding to the maxima of the Fe
line intensity, its value is equal to about 8,800 K. The electron density also seems to
attain maxima at s –12 and 11 mm.
6000 1 x 80 mJ
2 x 40 mJ, Δt = 6 μs
IFe 532.42 [counts]
4000
2000
10000
9000
Te [K]
8000
7000
10
8
ne [1016 cm-3]
0
-20 -10 0 10 20 30
Δs [mm]
density for single and double pulses of 80 mJ burst energy as a function of the focus position s
With the setup shown in Fig. 7.11, the dependence of the plasma parameters
on the ambient pressure was studied. The sample used is a pure iron sample. The
following parameters were used: single pulses 80, 160 mJ, pulse width 20 ns; double
pulses 2 40, 2 80 mJ, width 40 ns; interpulse separation t D 6 s, energy
ratio 1:1; position of the beam waist relative to the sample surface s D 10 mm,
detection after the single pulse or the second pulse of a double pulse with tdelay D
1 s, tint D 1 s. Excitation temperatures are determined by Boltzmann plots using
Fe I lines in the range between 427 and 542 nm with upper excitation energies
between 3.21 and 6.76 eV. In the case of local temperature equilibrium (LTE, see
Sect. 8.5), the excitation temperature is equal to the electron temperature. Evaluation
of the line width of the Stark broadened atomic iron line at 538.337 nm yields
electron densities. To assess the degree of absorption of this line in the plasma,
we estimated the effective absorption coefficient, which is proportional to the value
kt defined in [8.8], see also (9.31). For temperatures in the range of 7,500–9,700 K,
we obtain kt < 0:8 1030 m3 and hence absorption plays only a minor role and
no significant deviations from the Boltzmann plot are to be expected. For each burst
energy and pressure value measurements at five different sample locations were
conducted. At each sample location, five measurements are performed, where each
measurement comprises a series of ten bursts. Hence, for each data point shown
in the following diagrams, the spectral signals generated by 5 5 10 D 250
bursts are evaluated. Figure 8.22 shows the intensity of the Fe I line at 538.34 nm
and the continuum intensity measured at 534.6 nm – which lies within a spectral
region where no line emission was observed – as a function of the pressure for a
burst energy of 80 mJ.
The line intensity at 1 bar for double pulses is by a factor of 4.6 greater than
the one for single pulses and decreases monotonously for reduced pressures. Single
pulses show a different behavior. The line intensity increases for reduced pressures,
attains a maximum at about 100 mbar, and then decreases for further pressure
reductions. Error bars are shown enlarged for better visibility. For double pulses,
35
a
line intensity I Fe538.34 [counts*nm]
30 2 x 40 mJ
1 x 80 mJ
25 2x
20
15
10
4x
5
0
30
continuum intensity I cont534.6 [counts]
b
25
2x
20
15
10 4x
0
0.1 1 10 100 1000
pressure p [mbar]
Fig. 8.22 Line intensity of the iron line Fe I 538.34 nm (a) and continuum intensity at 534.6 nm
(b) as a function of pressure for a burst energy of 80 mJ
the error bars are significantly smaller than for single pulses. The physical reason of
these reduced error bars is not clear.
The maximum in the line intensity of atomic iron at a reduced air pressure was
also observed by Y. Iida, who studied the interaction of single pulses with a standard
Al alloy containing 0.97% iron in the pressure range from 5 to 760 torr [8.9]. At an
air pressure of 200 mbar, a maximum of the line emission Fe I 374.949 nm was
found, which is by about a factor of 10 higher than the value measured at 1 bar,
whereas here an enhancement by a factor of 3.2 was measured (cf. Fig. 8.22).
The continuum intensity measured in the line-free region at 534.6 nm decreases
for reduced pressures with the only exception for single pulses while going from
1,000 to 500 mbar. For all data points taken between 0.1 mbar and 1 bar, the
continuum intensity is higher for single pulses than for double pulses. At 0.1 mbar
and at 1 bar, the continuum intensity does not differ significantly between single
and double pulses. The ratio of line-to-continuum intensity is for all pressures
greater than 0.1 mbar higher for double pulses than for single pulses. The greatest
improvement of the line-to-continuum ratio amounts to a factor of 4.2 and is
achieved for double pulses at 1 bar. An increase of the burst energy from 80 to
160 mJ does not significantly change the line intensities measured at 1 bar for single
and double pulses (not shown here). However, the continuum intensity increases by
a factor of 3 for single and double pulses.
The line intensity as a function of the pressure shown in Fig. 8.22 is not directly
determined by the material ablation shown in Fig. 7.13. The line intensity is also
influenced by the plasma state in terms of temperature, density, and size. Electron
temperatures, electron densities, and ionization degrees are determined for all 25
measurements performed for each pressure value and burst energy. First, it was
checked to which extent the measuring values change for the five consecutive
measurements at one sample location. A possible reason for such a change could be
the formation of the crater, which may influence the conditions of the plasma gener-
ation for subsequent measurements and the interaction between plasma and sample
surface. No significant change was observed within the measuring uncertainty.
Figure 8.23a–c shows the determined electron temperature, electron density, and
ionization degree as a function of the pressure for a burst energy of 80 mJ. The error
bars correspond to the standard deviation of the average of the 25 measurements.
The ionization degree was calculated using the determined electron temperature,
electron density, and the Saha equation assuming that only the first ionization step
is significant, i.e., the density of singly ionized iron is equal to the electron density:
nFe II D ne . The ionization degree is then defined as: ı D ne =.nFe I C ne /, where
nFe I is determined from the Saha equation using tabulated partition function and
Unsöld’s formula to calculate the lowering of the ionization energy [8.10].
The highest temperatures of nearly 10,000 K were achieved for double pulses
and a pressure of 1 bar. For single pulses, the temperature at 1 bar is about 15%
less. The reduced temperature for single pulses is attributed to a shielding effect of
a plasma ignited in the ambient gas which absorbs a part of the laser energy thus
reducing the available energy for the ablation of the sample material and formation
of the iron plasma. For double pulses, a reduced shielding occurs since the material
11000
electron temperature Te [K]

a
10000
2 x 40 mJ
1 x 80 mJ
9000
8000
3x
7000
electron density ne [1016cm-3]
10 b
3x
1
0.1
1.2
c
degree of ionization d
1.0
0.8
0.6
0.4
0.2
0.1 1 10 100 1000
pressure p [mbar]
Fig. 8.23 (a) Electron temperature, (b) electron density, and (c) degree of ionization as a function
of pressure for a burst energy of 80 mJ
ablated by the first pulse partly pushes the ambient gas out of the interaction region
[8.1]. Hence, the energy of the second pulse can be transferred more effectively to
the ablated iron vapor leading to higher temperatures of the iron plasma (although
the energy of the second pulse is just half of that of the single pulses).
For double pulses, the electron temperature decreases monotonously while
lowering the pressure and stabilizes at about 7,000 K for pressures less than 10 mbar.
For single pulses, the temperature rises to 9,700 K being nearly constant for
pressures between 500 and 10 mbar and then drops again for pressures <10 mbar.
The rise of the electron temperatures observed for single pulses and pressures
below 1 bar is attributed to a reduced shielding effect of the ambient gas plasma
due to the lower initial gas particle density. The ambient gas plasma acts also as a
energy buffer transferring a fraction of its energy content to the adjacent material
plasma. The observed temperature decrease for single pulses and pressures below
10 mbar may be an indication that this effect becomes negligible for such low
pressures. For double pulses, the buffer effect is expected to be significantly smaller
since the first pulse leads to a localized reduction of the ambient gas density. The
monotonous decrease of the electron temperature with reduced ambient pressures
may be interpreted as a geometric effect. The first pulse generates a depleted gas
volume above the sample surface, which is greater for lower initial pressures.
The plasma generated by the second pulse expands into this preformed enlarged
volume leading to reduced temperatures. The relative change of the volume of
emitting Fe I atoms can be estimated by the ratio of the number of emitting Fe I
atoms to the density of Fe I atoms calculated from the Saha equation. The number
of emitting Fe I atoms is proportional to the measured line intensity of the Fe I
538.34 nm line (see Fig. 8.22) divided by eEj =kT =uo .T /, where Ej is the upper
excitation energy, k the Boltzmann constant, T the temperature, and uo .T / the
partition function of Fe I and assuming that the plasma can be considered as
optically thin for this line. This estimation yields a monotonous increase of the
volume of emitting iron atoms for double pulses with decreasing ambient pressure.
At a pressure of 10 mbar, the estimated emission volume is by a factor of 8 greater
than at 1 bar. Assuming an adiabatic cooling taking this volume ratio and three
degrees of freedom would cause a temperature reduction by a factor of 4. The
observed temperature reduction is significantly smaller probably due to additional
energy released by recombination processes. Another possible explanation is the
lack of shock heating as the ambient gas pressure is reduced.
The electron temperature curves show that double pulses at 1 bar lead to electron
temperatures which are approximately the same as those achieved for single pulses
at reduced pressures of about 100 mbar. The obvious interpretation is that the first
pulse in a double pulse causes a transient localized reduction of the ambient particle
density by about a factor of 10, creating conditions – in terms of the electron
temperature – similar to those observed for single pulses and pressures of about
100 mbar.
The electron temperatures and electron densities reported here for single pulse
excited plasmas at 1 bar are similar to those found in an experiment, where single
pulses of 100 mJ were focused on a low-alloy steel sample in an air atmosphere at
1 bar yielding temperatures in the range between 8,000 and 10,000 K and electron
densities of 4–11 1016 cm3 at a delay time of 3 s [8.11].
The electron density at 1 bar is about the same for double and single pulses, see
Fig. 8.23b. For double pulses, the electron density decreases for lower pressures. For
pressures <10 mbar, the resolution of the spectrometer is not sufficient to determine
the line width. For single pulses, the electron density attains a maximum at 500 mbar
and then decreases monotonously. The electron density for double pulses at 1 bar is
about the same as the one determined for single pulses at pressures around 100 mbar.
The material ablation increases for reduced pressures (cf. Fig. 7.13), the degree
of ionization is nearly constant or increases slightly (see Fig. 8.22c); hence, the
reduction of electron densities is linked with an increase of the plasma volume. The
decreasing electron density with reduced ambient pressure corresponds to the course
of the intensity of the continuum emission shown in Fig. 8.22, bottom. The electron
density decays by more than an order of magnitude while reducing the pressure
from 1,000 to 1 mbar; hence, an even stronger decrease of the continuum radiation
is expected, since the emission coefficient for free–free Bremsstrahlung scales as
"cont n2e [cf. (9.1)]. However, at the same time, the emitting volume increases,
leading to the diminished reduction of the continuum emission with lower pressures
(cf. Fig. 8.22b).
Figure 8.23c shows the degree of ionization as a function of the pressure. The
degree of ionization is higher at 1 bar for double pulses than for single pulses.
The curves cross at 100 mbar, indicating again that double pulses cause a plasma
state comparable with the one generated with single pulses at a reduced pressure of
100 mbar.
Argon was used as ambient gas to study the influence of collinear double pulses
on the excitation of the surrounding atmosphere as a function of the pressure.
Figure 8.24a–c shows the investigated spectral range centered around 430 nm. The
spectra shown in Fig. 8.23a, b are LIBS spectra gained with single pulses of 160 mJ
and the pure iron sample in air and argon atmosphere, respectively. The spectrum
shown at the bottom was detected with the same spectrometer setup observing
the emission of a low-pressure argon lamp. The emission spectrum of the latter
comprises only emission lines of atomic argon. The line detected at 434.81 nm in the
LIBS spectrum shown in the middle is attributed to an Ar II emission substantiated
as follows. The line at 434.81 nm is not observed with air as ambient atmosphere
(cf. spectrum at the top in Fig. 8.24). Hydrogen is a potential contaminant for argon.
The only line of hydrogen in this spectral range is the Balmer H line at 434.05 nm.
However, the deviation of the spectral position of 0.8 nm from the detected emission
is too big to be attributed to a Stark shift. Furthermore, hydrogen lines show a
stronger broadening at these plasma conditions than argon or iron lines. Estimating
the Stark width (FWHM) of the H line for a temperature of 10,000 K and an
electron density of 1016 cm3 yields 1.06 nm, which is significantly greater than
the width of the line observed at 434.81 nm in the spectrum shown in the middle
of Fig. 8.24 [8.5]. The ratio of the intensities of the two Ar I lines (spectrum in
Fig. 8.24c) at 434.52 and 430.01 nm amounts to 0.3. Such a relation would be
expected for an excitation temperature of about 4,600 K in the argon discharge
lamp (using line data from [8.12]). For higher temperatures to be expected in the
LIBS plasma this ratio increases. However, in the LIBS spectrum taken in argon
atmosphere a line pair with this intensity ratio at these spectral positions is not
observed. The line observed in the LIBS spectrum at an argon pressure of 750 mbar
at 434.81 nm lies – within the uncertainty of the wavelength calibration of the
spectrometer – at the position of an Ar II line. This identification is further supported
by the small line appearing on the right side of the Fe I 432.58 nm line. This line lies
at 433.12 nm, which is also an Ar II line. Taking the parameters of these two Ar II
lines, the expected intensity ratio I433:12 =I434:81 is for temperatures between 10,000
10000
a pair = 750 mbar
intensity ID [counts]
8000
6000
4000
2000
0
10000
b
Fe I 427.18 pargon = 750 mbar
8000
Fe I 430.79
Fe I 432.58
6000
Ar II 433.12
4000 Ar II 434.81
2000
0
2500
c argon lamp
2000
Ar I 430.01
1500
1000
Ar I 434.52
500
0
426 428 430 432 434 436
wavelength l [nm]
Fig. 8.24 Spectra of the pure iron sample gained with single pulses of 160 mJ in air atmosphere
(top) and argon atmosphere (middle). Bottom: spectrum of a low-pressure argon lamp in the same
spectral region
and 20,000 K about 0.15–0.16, which is in agreement with the observed ratio. For
temperatures in the range of 7,000–20,000 K, the Ar II 434.81 line has a kt value of
<5 1032 m3 , and hence this line is also not affected by self absorption (see above
discussion for the Fe I 538.34 line).
Figure 8.25 shows the intensity of the Ar II 434.81 nm line as a function of the
argon pressure for single and double pulses with a burst energy of 160 mJ. The
line intensity attains a minimum for single pulses at 10 mbar. In the pressure range
from 10 to 1,000 mbar, the intensity increases linearly with the pressure. For double
pulses, the line was not observable at a pressure of 50 mbar. In the range from 100 to
1000
2 x 80 mJ
intensity I Ar434.81 [counts*nm] 1 x 160 mJ
100
~p
~p 2
10
1
0.1 1 10 100 1000
pressure p [mbar]
Fig. 8.25 Intensity of the Ar II 434.81 nm line as a function of the argon pressure for single and
double pulses with a burst energy of 160 mJ
1,000 mbar, the line intensity increases with pressure following a power law p 2 .
At 1 bar, the line intensity of the Ar II line is for double pulses about a factor of
6 smaller than for single pulses. The Ar II line intensity induced by single pulses
reaches this level at a pressure between 100 and 200 mbar. Obviously double pulses
lead to plasma conditions, which are similar to those generated with single pulses at
reduced ambient particle density.
For pressures below 10 mbar, the intensity rises again for single and double
pulses. The cause of this behavior is not known.
Figure 8.26a, b shows the line widths (HWHM) of the lines Ar II 434.81 and Fe I
432.58 nm as a function of the argon pressure. For pressures below 1 mbar, the line
width converges to the instrumental width of 0.027 nm for that spectral range. The
Ar II 434.81 nm line shows Stark broadening. Taking the broadening parameters
given in the literature for this line, the electron density can be determined, see
vertical scale on the right side of Fig. 8.26 [8.13]. Stark parameters for the Fe I
432.58 line are not available [This line was studied to compare its behavior with
the Ar 434.81 line. The Fe 432.58 line lies in the same spectral window detected
(cf. Fig. 8.24b) and shows no strong overlap with any other lines]. The electron
density decreases at lower pressures. For double pulses, the electron density is
always smaller than for single pulses. Comparing the values shown in Fig. 8.26
with those of Fig. 8.23 shows that the electron density determined from the Ar
II line is significantly greater than the one determined with the Fe I 538.34 nm
line. Determining electron densities with Ar II line implies that this electron
density persists in regions, where Ar II dominates the plasma and these regions
are expected to be spatially separated from the iron containing core of the plasma.
The expanding plasma consisting of ablated iron species pushes the surrounding
ambient gas particles outward, inducing a shock wave and collecting the ambient
gas in a shell surrounding the core plasma. If for this argon shell, a minimum argon
particle density corresponding to the initial gas pressure of 1 bar at room temperature
0.13 10
a
line width HWHM Δλ [nm]
0.11 8
2 x 80 mJ
ne [1017 cm-3] Argon

Ar II 434.81, ne
0.09 6
Fe I 432.58
0.07 4
0.05 instrumental width 2

Δλinstr = 0.027 nm
0.03 0
0.01
0.13 10
b
line width HWHM Δλ [nm]
0.11 8
1 x 160 mJ
ne [1017 cm-3] Argon

Ar II 434.81, ne
0.09 6
Fe I 432.58
0.07 4
0.05 2
0.03 0
0.01
0.1 1 10 100 1000
pressure p [mbar]
Fig. 8.26 Line widths of the Ar II line 434.81 nm and Fe I line 432.58 nm (see Fig. 8.24b) as a
function of the argon pressure. Top: double pulses 2 80 mJ, bottom: single pulses 1 160 mJ.
The right scale refers to the electron density determined from the Stark width of the Ar II line
(2:5 1019 cm3 ) is assumed, then the electron density for a given temperature
using the Saha equation can be estimated. For 13,000 K (which is in the range to be
expected from the spectrum in Fig. 8.24b, looking at the ratio of the two Ar II lines),
an electron density of 8:7 1017 cm3 is obtained, which is approximately equal
to the value determined for single pulses at 1 bar (see Fig. 8.26b). Since the second
plasma induced by a double pulse expands into a locally rarefied argon particle
density as an effect of the preceding laser pulse, the electron density level achieved
in the argon shell is expected to be smaller than for single pulses in agreement with
the measurements (cf. Fig. 8.26a).
8.3 Spatially Resolved Electron Density
The index of refraction caused by the laser-induced plasma is determined from the
interference patterns detected in the framing mode of the streak camera. Figures 8.27
and 8.28 show framing images of the interference patterns after the first and the
8.3 Spatially Resolved Electron Density 153
Fig. 8.27 Framing images of interferograms detected in a time interval comprising the first and
the second laser pulse of a double pulse. In the upper right, a schematic view of the orientation
of the fringes relative to the plasma and the sample is shown. The diagram in the lower right shows
the temporal sequence of the frames with the delay times tdelay starting at 0 s for the first frame.
Experimental parameters are the same as for Fig. 8.7
Fig. 8.28 Framing images of interferograms detected after the second laser pulse of a double
pulse. All other experimental parameters are the same as for Fig. 8.27
second laser pulse of a 2 80 mJ double pulse and an ambient pressure of p0 D

1;013 mbar [8.1]. The framing rate is 106 s1 , corresponding to a time separation
between subsequent frames of 1 s. The exposure time amounts to 200 ns/frame.
For a better use of the image format of the streak camera, the camera is rotated by
90ı around the axis of observation. The mutual distance between the interference
fringes is increased in relation to the one chosen for the streak images in the previous
section to simplify the evaluation of the interferograms. The geometry and sequence
of the frames is shown on the right side of Fig. 8.27. The time of irradiation of the
laser pulses cannot be determined exactly from the framing images. In Fig. 8.27, the
first pulse arrives at the sample between the frames no. 1 and 2, and the second pulse
between frames no. 7 and 8. In Fig. 8.28, the second pulse hits the sample between
frame no. 1 and 2.
The boundaries of the displaced interference fringes show a well-defined hemi-
spherical shape. At the front of the shock wave, the fringes are displaced toward
the sample surface, which is due to the increase in the refractive index caused by the
compressed air and sample particles. Inside the sphere – defined by the leading edge
of the shock wave – the fringes are displaced in the opposite direction, i.e., toward
the positive z-direction, compared with their original position. This displacement
corresponds to a refractive index <1 indicating the presence of free electrons and/or
a depletion of the ambient atmosphere. The fringe displacements are symmetric
with respect to the z-axis; hence, the method of Abel inversion can be applied to
determine the refractive index distribution [8.14].
The fringe displacement is given by:
Z
1
ı.y; z; t/ D .np na /dx; (8.11)
Vp
where x; y; z are the coordinates shown in Fig. 8.2, is the wavelength of the argon
ion laser, np is the refractive index of the plasma, and na is the refractive index of
the ambient atmosphere in the reference arm of the interferometer. The integration
runs along the line of sight across the plasma volume Vp . The refractive index of
the plasma can be described by the following relation [8.14] (in this chapter, capital
letters are used for the electron density and particle densities in order to avoid a
mix-up with refractive index quantities):
e 2 2 X
np 1 D 2 2
Ne C 2 Nj ˛j ; (8.12)
8 me c "0 j
where Ne is the electron density, Nj are the densities of neutral atoms and ionic
species in the plasma, ˛j the polarizabilities.
The polarizability of atomic iron is ˛Fe I D 13 1030 m3 [8.15]. The contri-
bution of the iron atoms to the refractive index can then be estimated with (8.12):
.np 1/Fe I D C8:2 1023 NFeI =cm3 , where NFeI is the particle density of
atomic iron. The contribution of the free electrons amounts to .np 1/Ne D 12
1023 Ne =cm3 .
With an average degree of ionization in the plasma determined spectroscopically

of 90% the ratio j.np 1/NFe I j=j.np 1/Nej amounts to 0.08; hence, the contribution
of the atomic iron particles can be neglected. No data for the polarizability of
singly ionized iron atoms have been found in the literature. However, calculations
of polarizabilities for the transition element copper showed that the polarizability of
a copper ion is less than one-tenths of the value calculated for copper atoms [8.16].
Assuming that similar relations hold for iron only a small contribution of the iron
ions to the refractive index is expected, which is in the order of the one estimated
for the iron atoms.
A reduction of the ambient gas particle density behind the shock wave will lead
to a reduced refractive index. The order of magnitude of this effect is estimated
assuming that the shock wave reduces the ambient gas particle density completely,
i.e., the shock wave leaves a vacuum behind. Under this simplified assumption, the
ratio of the fringe displacements due to vacuum ıv and free electrons ıe amounts to:
ıv =ıe D j1 na j=j.np /Ne na j D 0:7, for na D 1:0002779, D 514:5 nm, and
Ne D 1018 cm3 . If a complete displacement of the ambient atmospheric particle
density occurs, then the expected fringe shift is of the same order of magnitude as
the one expected by the free electrons of the plasma. Since both effects go into the
same direction, the experimentally determined fringe shifts can only be attributed
to the free electrons of the plasma by the ratio ıe =.ıe C ıv / D 0:6. Figure 8.29
shows the determined distribution of the refractive index 4 s (left) and 5 s (right)
after the irradiation of one 80 mJ laser pulse. The formation of the shock front in a
spherical geometry can be clearly seen by the values of the refractive index n > 1.
Behind the shock front, a volume of reduced refractive index becomes visible.
Fig. 8.29 Refractive index distribution determined from the framing interferograms. Left: 4 s
after a single 80 mJ laser pulse. Right: 5 s after a single 80 mJ laser pulse
Fig. 8.30 Refractive index distribution determined from the framing interferograms. Left: 2 s
after a single 80 mJ laser pulse. Right: 2 s after the second pulse of a 2 40 mJ double pulse
Figure 8.30 shows refractive index distributions 2 s after a single 80 mJ laser

pulse (left) and after the second pulse of a 2 40 mJ double pulse (right). The
formation of two concentric shock waves can be seen in the right diagram, where
the second shock wave is weaker than the first. Interpreting the central region
with n < 1 as the plasma volume, the volume ratio amounts to 3.3, i.e., the
plasma volume after the double pulse is by a factor of 3.3 greater than in the
case of the single pulse (for the same total laser pulse energy irradiated). Framing
interferograms taken at a reduced ambient pressure of 500 mbar show a reduced
depth of the refractive index trough.
A reduction of the ambient pressure has a significant influence on the depth of
the refractive index distribution. Figure 8.31 shows the refractive index distribution
determined from framing interferograms gained at ambient pressures of 1,000 mbar
(left) and 500 mbar (right), measuring parameters: 2 40 mJ, t D 6 s, s D
5 mm, 2 s after the second pulse. A reduction of the ambient pressure manifests in
a reduced shift of the interference fringes in the plasma and therefore to a smaller
deviation of the refractive index from the one for air in the reference arm of the
interferometer.
In the following, we assume that only the free electrons lead to the reduced
refractive index region with n < 1, then the relation (8.12) can be used to calculate
the electron density distribution. Figure 8.32 shows the result using the data of
Fig. 8.31 for the ambient pressures of 1,000 and 500 mbar. For an ambient pressure
of 1,000 mbar, the electron density attains a value of 3 1018 cm3 at the center
of the plasma sphere, for 500 mbar only about half of this value is reached. The
estimated relative error is 20%.
Fig. 8.31 Refractive index distribution determined from framing interferograms gained at an
ambient pressure of 1,000 mbar (left) and 500 mbar (right), 2 40 mJ, 2 s after the second pulse
of a double pulse (the left diagram is identical to the right diagram of Fig. 8.30)
Fig. 8.32 Left: Electron density distribution calculated from Fig. 8.31, left for 1,000 mbar ambient
pressure; right: from Fig. 8.31, right for 500 mbar, assuming that the refractive index inside the
second shock wave n < 1 is caused by free electrons only. Measuring parameters: 2 40 mJ, 2 s
after second pulse of double pulse
Taking into account the above-discussed reduction of the ambient gas particle
density behind the shock wave, the electron density is only 60 % of 3
1018 cm3 , i.e., 2 1018 cm3 . The electron density determined interferometrically
after a single pulse of 80 mJ and 2 s is about the same. For the comparison of the
electron densities measured interferometrically with those determined by spatially
integrated emission spectroscopy of a Stark broadened iron line, the electron density
distribution – as shown in Fig. 8.32 – is approximated by parabolic profiles along the
z-direction at y D 0 and along the y-direction at z D – where is the minimum
distance between the boundary of the evaluable interferogram and the surface of the
sample being about 0.5 mm – as follows:

Ne D Ne;max 1 .=re /2 ; (8.13)
where is the spatial coordinate z or y (see Fig. 8.2), re is the radius of the parabolic
distribution, and Ne;max is the maximum electron density. The parameters Ne;max and
re of both parabolas for y D 0 mm and z D are averaged. For an ambient pressure
of 100 mbar, only one parabola can be evaluated.
An integration over the volume of the electron density distribution approximated
in this way yields the total number of electrons within the plasma half-sphere having
a radius of re . By division through the plasma volume an average electron density
can be estimated. For the integrated average electron densities, an uncertainty of
˙20 % is assumed.
The effect of neutral iron atoms on the refractive index requires a correction of the
determined electron densities to higher values. This correction amounts to about 7%
and is therefore significantly smaller than the error of the single measuring values
of the refractive index of 20 %. Hence, this correction is neglected.
Figure 8.33 shows the integrated average electron densities 2 s after the second
laser pulse (p0 D 1; 000 mbar and p0 D 500 mbar) and 2 s after a single pulse
(p0 D 100 mbar) as a function of the ambient pressure. The values deduced from
the parabolic approximation and volume integration, the lower limits of the electron
densities taking into account the correction due to the displaced air inside the plasma
volume, and the electron density values determined by spectroscopy are shown.
100 1000
averaged electron density [1017cm–3]
interferometry spectroscopy
electrons only Fe 538.34 nm
10 10
air displacement
EL = 1 x 40 mJ
1 1
tdelay = 1. pulse + 2 μs
0.1 0.1
100 1000
p0 [mbar]
Fig. 8.33 Average electron density as a function of the ambient pressure determined by interfer-
ometry and spectroscopy. Measuring parameters for the interferometric data, 500 and 1,000 mbar:
240 mJ, t D 6 s, tdelay D 2 s, s D 5 mm; 100 mbar: 140 mJ, tdelay D 2 s, s D 5 mm;
for spectroscopy: Stark broadening of the iron line Fe I 538.34 nm, 2 40 mJ, t D 6 s,
tdelay D 2 s, s D 10 mm (see also Fig. 8.23)
Interferometric measurements performed at s D 10 mm showed that within

the measuring uncertainty there is no difference in the electron density compared
with the values determined for s D 5 mm. The electron densities gained from
interferometry are an order magnitude greater than those deduced spectroscopically.
This discrepancy cannot be attributed to the different s values. On the other hand,
the pressure dependence shows qualitatively the same behavior.
Figure 8.34 illustrates a model to describe qualitatively the observed difference in
electron density values found. The curves show the density distribution of electrons,
neutral iron atoms (Fe I), and singly ionized iron ions (Fe II) as a function of
the plasma radius for single (left) and double pulses (right). The plotted numbers
are values determined by the interferometric measurements. The electron density
distribution can be approximated by a parabolic distribution. The maximum electron
density at the plasma center is the same for single and double pulses, but the plasma
is greater in case of the double-pulse excitation. If there is no difference in the
electron density at the center of the plasma, then the density of singly ionized
iron (Fe II) at the center is also the same due to the charge neutrality condition.
Spectroscopic studies showed that the average degree of ionization is lower for
single pulses than for double pulses. Hence, the density of iron atoms shown for
double pulses is smaller than for single pulses.
The spectroscopically determined Stark broadening of the neutral iron atom line
is the result of a superposition of all emission regions inside the plasma. However,
the outer regions will contribute dominantly to this broadened line since there
greater densities of Fe I prevail and the absolute number of Fe I atoms is also
Fig. 8.34 Qualitative presentation of the density distribution of electrons, neutral iron atoms (Fe I),
and singly ionized iron ions (Fe II) in laser-induced plasmas for single and double pulses of the
same total pulse energy. The density distributions are shown as a function of the plasma radius.
SP D single pulse, DP D double pulse. Parameters: SP 1 80 mJ, tdelay D 2 s, DP 2 40 mJ,
t D 6 s, tdelay D 2 s; for both: p0 D 1 bar, s D 5 mm
greater, as the volume of a spherical shell scales as dV D 4 r 2 dr (see hatched

region in Fig. 8.34 illustrating the interval where the neutral iron particle density
attains a maximum). As a result, the electron density determined spectroscopically
by a neutral iron line cannot be considered as a representative average electron
density of the plasma, but as an electron density to be allocated to the outer plasma
regions. Hence, the values determined spectroscopically for the electron density are
underestimating the real average electron density. In this way, the spatial separation
of the density distributions of Fe I and Fe II may explain qualitatively the found
difference. The interferometric measurement is – taking into account the simplifying
assumptions described above – independent of the distribution of neutral particles
and ions inside the plasma.
The investigations presented in this chapter have shown that the effect of
collinear double pulses is a local transient reduction of the particle density in the
vicinity of the sample surface caused by the first pulse. The second pulse interacts
predominantly with the sample surface. No re-heating of a residual plasma from the
first pulse was observed. The material ablated by the second pulse expands faster
into the cavity of reduced particle density originating from the first pulse attaining
a significantly enlarged plasma volume. The electron densities are approximately
the same for single and double pulses of the same total energy. Considering the
larger plasma volume and the fact that the mass ablated using double pulses is
also increased significantly (cf. Sect. 7.2) favors a more intense and longer lasting
emission of element specific radiation to be used for LIBS analysis.
8.4 Characteristic Time Scales and Local Temperature

Equilibrium Conditions of the Laser-Induced Plasma
The plasma induced by the laser pulse for LIBS is a nonstationary plasma governed
by different characteristic time scales as illustrated in Fig. 8.35 ranging from
picoseconds to several tenths of a second, thus spanning 11 orders of magnitude.
The temporal width of the initiating laser pulse L lies typically in the range of a few
nanoseconds to several 10 ns. After ablation and ionization of the irradiated material,
the electrons collide with each other described by the relaxation time ee , which is of
the order of picoseconds. Hence, the electrons attain very fast a Maxwellian velocity
distribution. The energy exchange via collisions with the neutral particles or ions
characterized by k0 and k1 takes much longer. After the time e;0 , the population
density of the discrete energy levels of heavy neutral particles can be described
by the Boltzmann statistics with a temperature Te . The typical plasma life time
Plasma amounts up to several 10 s. At early times the spectral plasma emission is
dominated by free–free (ff) and free–bound (fb) transitions of electrons being taken
over by line emission at later times. The bar marked with mod describes the time
scale of double and multiple pulses (in the case of separated single pulses within
a double-pulse burst this time corresponds to the interpulse separation t or for
8.4 Characteristic Time Scales and Local Temperature Equilibrium Conditions 161
Fig. 8.35 Time scales of the laser-induced plasma for LIBS. L D temporal width of laser pulse,
0 1
ee D electron–electron relaxation time, k D electron–neutral relaxation time, k D electron–ion
relaxation time, e;0 D time for population of discrete levels according to Boltzmann statistics,
Plasma D life time of plasma, ff D free–free emission, fb D free–bound emission, mod D time
scale of multiple pulses, ge D gas exchange time, trep D repetition time of laser pulses
multiple pulses to the sum of interpulse separations, cf. Fig. 3.9). This time domain
overlaps with those of the plasma life time and the line emission. For repetitive
plasma ignition, the gas exchange time ge [cf. Chap. 3, relation (3.1)] has to be
considered to assure an effective removal of the material ablated by the previous
laser pulse. The bar trep describes the time between consecutive laser pulses or laser
bursts corresponding to laser pulse repetition rates between a few hertz up to 1 kHz
applied so far for LIBS.
In the following, characteristic spatial and temporal quantities describing the
state of the laser-induced plasma will be discussed in more detail. A plasma can
be described as classical if the thermal De Broglie wavelength of the electrons,
h
B D p (8.14)
2 me kB Te
with h Planck constant, me electron mass, kB Boltzmann constant, Te electron

temperature, is smaller than the mean distance n between the free electrons
1=3
n ne . In this case, quantum effects, e.g., degeneration of the electron gas,
can be neglected. Furthermore, the average distance between particles is compared
with the average collision length given by the Landau length
e2
L D (8.15)
4 "0 kB Te
with e elementary charge, "0 dielectric constant. Figure 8.36 shows these char-
acteristic length scales in a ne Te diagram together with the plasma parame-
ters determined experimentally (cf. Fig. 8.23a, b for double and single pulses).
1018
ln = 60 lB
ln = 11 lL
electron density ne [cm-3]
1017
1016
1015
ln = 250 lB
ln = 40 lL
1014
6000 7000 8000 9000 10000 11000
Fig. 8.36 Electron density and electron temperature measured for double and single pulses. The
curves illustrate the characteristic plasma length scales De Broglie wavelength B and Landau
length L in relation to the mean particle distance n . Filled triangles D double pulses, open
triangles D single pulses. Triangles pointing upward: burst energy 160 mJ, triangles pointing
downwards: burst energy 80 mJ
The measured parameters lie in a parameter field where the De Broglie wavelength
is by a factor of 60–250 and the Landau length by a factor of 11–40 smaller than
the average particle distance. Hence, the laser-induced plasma can be treated as a
classical and ideal plasma, where the interaction energy between charged particles
is small compared with the kinetic energy.
Another description of ideality can be defined by the condition that the number
of electrons in the Debye sphere is significantly greater than 1. The radius of the
Debye sphere is given by the length along which the potential of an ion inside the
plasma decays by a factor of 1=e. For the number of electrons in the Debye sphere
s
holds

4 "0 kB 3=2 Te3
ND D : (8.16)
3 2e 2 ne
Figure 8.37 shows the ne Te diagram with curves for different ND . For the laser-
induced plasmas studied ND varies between 1 and 8; hence, the condition ND 1
is not fulfilled for all plasmas. Hence, the assumption of a continuous charge density
distribution inside the Debye sphere is not completely fulfilled. Such a continuous
charge distribution is, e.g., assumed to calculate the reduction of the ionization
energy or to calculate cross sections. For simplification, the fulfillment of the Debye
approximation will be assumed in the following paragraphs.
For a local temperature equilibrium (LTE) in an optically thin plasma, the
probability of a collision-induced transition is much greater than the probability
of a radiating transition. This condition is fulfilled, if the following relation holds
1018
ND = 1
electron density ne[cm-3]
1017
1016
ND = 2
1015
ND = 8
1014
6000 7000 8000 9000 10000 11000
Fig. 8.37 Electron density and electron temperature measured for double and single pulses (same
data points as in Fig. 8.36). The curves illustrate the Debye number ND
for all energy levels of an atom with a principal quantum number greater or equal
n [8.5]:
7
18 .z C 1/ kB Te 1=2
ne 7 10 cm3 ; (8.17)
n17=2 EH
where z is the ionization stage, z D 0 for neutral particles, z D 1 for singly ionized
particles, etc., n is the principal quantum number, EH is the ionization energy of
hydrogen.
Relation (8.17) is valid exactly only for the hydrogen atom. For transitions of iron
atoms studied, e.g., in Sect. 8.2, we have n 3 and z D 0. All experimentally found
electron temperatures of the material plasma were less than Te D 11; 000 K (cf.
Figs. 8.15, 8.16, 8.19–8.21, 8.23). Relation (8.17) yields then: ne 2 1014 cm3 .
This value is less than the determination limit of the electron density using the Stark
broadened iron lines, which amounts to about 2 1015 cm3 . Hence, in these cases,
where the electron density can be determined using Stark broadened iron lines, the
LTE condition is fulfilled.
Figure 8.38 shows the relation (8.17) for the considered iron atoms in a ne Te
diagram together with the plasma trajectory taking the data of Fig. 8.16. Parameter
of the data points is the time delay tdelay after the second pulse of the double
pulse. The bar alongside the plasma trajectory indicates the typical integration time
window used for quantitative analysis. Figure 8.38 illustrates clearly that the LTE
condition is fulfilled for all relevant time delays.
For the Ar II line at 434.81 nm (cf. Fig. 8.26), we have n D 4 and z D 1, so
relation (8.17) results in: ne 5 1015 cm3 . Again this condition is fulfilled for
electron densities determined with the help of the Stark broadened Ar II line (cf.
Fig. 8.26).
Fig. 8.38 Local temperature

equilibrium condition (LTE)
according to relation (8.17)
and plasma trajectory in a
ne Te diagram (data points
are taken from Fig. 8.16). The
bar alongside the plasma
trajectory indicates
exemplarily the integration
time window for quantitative
LIBS analysis
For nonstationary plasmas, the relaxation times of the various plasma species are
considered to estimate whether a LTE persists. For electrons, the relaxation time to
establish a Maxwell velocity distribution is given by [8.17]:
ee 3:16 104 We3=2 n1

e s; (8.18)
where We is the electron energy in eV and ne is the electron density in cm3 .

For We D 1 eV, ne D 1016 cm3 , relation (8.18) yields ee 3 1012 s. Hence,
for the electrons, an electron temperature is established very fast compared with
the typical life time of a laser-induced LIBS plasma of several microseconds. The
equilibration time to obtain LTE between electrons and atoms is given by [8.18]:
" 3=2 #
0 7 EH nt M
7:5 10 ne s; (8.19)
k
kB T n1 me
where nt is the total density of heavy particles, n1 is the density of ions, ne is the
electron density in cm3 , and M is the mass of atoms or ions.
The corresponding time for singly ionized particles is:
" 3=2 #
1 EH M
7:5 107 ne s (8.20)
k
kB T me
Equations (8.19) and (8.20) are approximations, and they are exactly valid only in
the case of hydrogen atoms.
So far we have discussed the time necessary to obtain a kinetic temperature of
the particles. The equilibration time for a distribution of energy over the discrete
terms of heavy particles according to Boltzmann’s law is given by:

1=2
z3 nz 7 Ez1;2 kB Te Ez1;2
e;z1 1:1 10 exp s; (8.21)
f12 ne nz C nz1 z2 EH z2 EH kB Te
where z is the stage of ionization, z D 1 for neutrals, z D 2 for singly ionized atoms,
etc., f12 is the oscillator strength of the resonance line, and Ez1;2 is the excitation
energy of the resonance line.
Figure 8.39 shows again the ne Te diagram with curves for k0 , k1 , and e;0 . For
the calculation of k0 , an ionization degree of 40 % is assumed. For the mass M in the
relations (8.19) and (8.20), the value for iron is taken M D 56 mp, where mp is the
proton mass. The time e;0 is calculated for the Fe I line at 549.713 nm. Assuming
n1 n0 for the plasma, then n1 =.n1 Cn0 / is the degree of ionization. As upper limit
for the degree of ionization 1 is taken. Figure 8.39 shows that the relaxation times
for the kinetic energy between electrons and neutral particles k0 are in the range
between 30 ns and 1s for an ionization degree of 40%. The corresponding times
for ions k1 amount 12–400 ns. However, for those plasmas (data points in Fig. 8.39)
with relaxation times k0 close to 1s the degree of ionization is between 80% and
100% and not 40% as assumed for the calculation (cf. Fig. 8.23c). According to
relation (8.19), the relaxation time of the kinetic energy is by a factor of 2–2.5
shorter and amounts 0:4–0:5 s. Hence, for the investigated plasmas, a kinetic
equilibrium between electron and heavy particles is achieved 500 ns after the laser
pulse irradiation.
The relaxation times for the excited states e;0 of the neutral particles lie in the
range between 2 and 300 ns. For the spectroscopic results shown, e.g., in Figs. 8.22
and 8.23, the emission spectrum was detected with tdelay D 1 s and tint D 1 s,
1018
tk0 = 30 ns, te,0 = 2 ns
tk1 = 12 ns
electron density ne [cm–3]
1017
1016
te,0 = 300 ns
1015
tk0 = 1 μs,
tk1 = 0.4 μs
14
10
6000 7000 8000 9000 10000 11000
Fig. 8.39 Relaxation times in the ne –Te diagram, see text

so a LTE can be assumed. For those plasma states where the relaxation time is not
significantly smaller than 1 s, deviations of the LTE are to be expected.
References

Spectrom. 19, 419–428 (2004)
8.2. L. Sedov, Similarity and Dimensional Methods in Mechanics, 4th edn. (Gostekhizdat,
Moscow, 1957), English transl.: ed. by M. Holdt (Academic, New York, 1959)
8.3. Y. Zel’dovich, Y. Raizer, in Physics of Shock Waves and High-Temperature Hydrodynamic
Phenomena, vol. I, ed. by W. Hayes, R. Probstein (Academic, New York, 1966)
8.4. R. Adrain, J. Watson, P. Richards, A. Maitland, Laser microspectral analysis of steels. Opt.
Laser Technol. 12, 137–143 (1980)
8.5. R. Griem, Plasma Spectroscopy (McGraw-Hill, New York, 1964)
8.6. J. Fuhr, G. Martin, W. Wiese, S. Younger, Atomic transition probabilities for iron, cobalt, and
nickel (a critical data compilation of allowed lines). J. Phys. Chem. Ref. Data 10, 305–565
(1981)
8.7. S. Freudenstein, J. Cooper, Stark broadening of Fe I 5383 Å. Astron. Astrophys. 71, 283–288
(1979)
8.8. J. Aguilera, J. Bengochea, C. Aragón, Curves of growth of spectral lines emitted by a
laser-induced plasma: influence of the temporal evolution and spatial inhomogeneity of the
plasma. Spectrochim. Acta B 58, 221–237 (2003)
8.9. Y. Iida, Effects of atmosphere on laser vaporization and excitation processes of solid
samples. Spectrochim. Acta 45B, 1353–1367 (1990.)
8.10. H.-W. Drawin, P. Felenbok, Data for Plasmas in Local Thermodynamic Equilibrium
(Gauthier-Villars, Paris, 1965)
8.11. K. Monge, C. Aragón, J. Aguilera, Space- and time-resolved measurements of temperatures
and electron densities of plasmas formed during laser ablation of metallic samples. Appl.
Phys. A69, S691–S694 (1999)
8.12. http://physics.nist.gov/PhysRefData
8.13. N. Konjevic, W. Wiese, Experimental Stark widths and shifts for spectral lines of neutral
and ionized atoms. J. Phys. Chem. Ref. Data 19, 1307–1385 (1990)
8.14. R. Boxmann, Interferometric measurement of electron and vapor densities in a high-current
vacuum arc. J. Appl. Phys. 45, 4835–4846 (1974)
8.15. S. Fraga, J. Karwowski, K. Saxena, Hartree-Fock values of coupling constants, polarizabil-
ities, susceptibilities and radii for the neutral atoms, helium to nobelium. Atom. Data Nucl.
Data Tables 12, 467–477 (1973)
8.16. S. Fraga, K. Saxena, B. Lo, Hartree-Fock values of energies, interaction constants, and
atomic properties for the ground states of the negative ions, neutral atoms, and first four
positive ions from helium to krypton. Atom. Data 3, 323–361 (1971)
8.17. A. Eberhagen, Elektromagnetische Strahlung aus Plasmen hoher Temperatur und Dichte.
Z. Ang. Physik XX, 244–257 (1966)
8.18. W. Lochte-Holtgreven, Plasma Diagnostics (North-Holland Publishing, Amsterdam, 1968)
Chapter 9
Plasma Emission
This chapter describes the dominant radiation processes of the laser-generated

plasmas such as Bremsstrahlung, recombination radiation, and line radiation. For
the line emission being the relevant emission signature used for LIBS, the various
line broadening mechanisms are presented and absorption processes are discussed.
9.1 Bremsstrahlung and Recombination Radiation
Transitions between free states of electrons moving in the Coulomb fields of the
ions generate radiation. The emission coefficient of free–free transitions is given
by [9.1]: r
!;d .˛a0 /3 EH EH „! X 2 z
"ff D 16 e kT Ne z Na ; (9.1)
3.3/1=2 kT z;a
where ˛ is the fine structure constant, a0 is the Bohr radius, EH is the ionization
energy of hydrogen, T is the plasma temperature, „! is the photon energy, Ne is
the electron density, z is the charge state of the ions, z D 0 corresponds to
neutral atoms, z D 1 to singly ionized species, etc., and Naz is the density of ion
species a with charge z, if Gaunt corrections, reduction of the ionization energy,
and advance of the series limit are neglected. The unit of the emission coefficient
(9.1) is ["ff!;d D W=.m3 sr rad s1 /, i.e., power per unit volume per steradian per
angular frequency interval. Relation (9.1) can be transformed to an emitted power
per unit volume per wavelength interval "ff ; Œ"ff D W=.m3 m/ by multiplying the
right side with .4/2c=2 , where c is the velocity of light and the wavelength of
the emitted radiation. Figure 9.1 illustrates the Bremsstrahlung emission described
by the calculated emission coefficient "ff using experimentally determined plasma
parameters. The Bremsstrahlung emission shows a strong decay toward wavelengths
below 400 nm, and it decreases by more than an order of magnitude within the
first 4 s. This type of emission generates a continuous background in the LIBS
spectra. To achieve high ratios between line emission coefficients and continuous

168 9 Plasma Emission
104
1
103
8
0
1
102 2
e ff , e ji / ΔλFWHM, e bb [Wcm-3 nm-1]
4
Planck emission 8
101
100 tdelay [μs]

Fe I line 538.34 nm
λ
10-1
0
10-2 1
2
λ
10-3 4
8
10-4
Bremsstrahlung
-5
10
100 200 300 400 500 600 700 800 900
wavelength [nm]
Fig. 9.1 Calculated Bremsstrahlung emission described by the emission coefficient for free–free
transitions "ff , line emission coefficient "ji , and equivalent Planck emission coefficient "bb as a
function of the wavelength using the plasma parameters from Fig. 8.16 for delay times tdelay from
0 to 8 s after the second pulse of a double pulse. The calculated peak height of the emission line
is plotted for tdelay D 8 s. The peak heights of this line for the different delay times are indicated
by horizontal dashes
Bremsstrahlung emission, it is favorable to use short line emission wavelengths –

as far as is by other restrictions possible – and a sufficient delay time tdelay after the
plasma ignition.
Transitions between free states of electrons and bound states generate the
recombination radiation described by the following emission coefficient [9.1]:
3=2 X z4 2
!;d 32 .˛a0 /3 EH EH „! 2
"fb D e kT Ne ez EH =n kT Nza ; (9.2)
3 3.3/1=2 kT z;a;n
n 3
where n is the main quantum number.

The overall contribution of the recombination radiation can be obtained by
integrating relation (9.2) over all emitted frequencies and comparing this result
with the overall contribution of the Bremsstrahlung radiation. It turns out that the
recombination radiation is larger by a factor of E1 z1;a
=kT, where E1z1;a
is the
ionization energy of the species a for the ionization stage z 1. Considering an iron
9.2 Line Emission 169
0;Fe
plasma where only the first ionization stage is of relevance, we have E1 D 7:9 eV.
Comparing this with a typical plasma temperature of 0.7 eV (8,100 K, cf. Fig. 8.16)
yields a factor of about 10. Hence, the total power radiated over all frequencies
is by about an order of magnitude higher for recombination radiation than for the
Bremsstrahlung emission.
9.2 Line Emission
In an optically thin plasma, the atoms and ions emit line radiation described by the
line emission coefficient [9.2]:
1 gj Ej
";d D ./ Aji hji Naz z e kT ; (9.3)
ji
4 ua .T /
where ./ is the line profile as a function of the frequency, Aji is the Einstein
coefficient for the transition from upper level j to lower level i , h is the Planck
constant, hji is the photon energy of the transition from j to i , Naz is the density
of species a with charge z, gj is the statistical weight of upper level j , uza .T / is
the partition function of species a with charge z, and Ej is the energy of the upper
level j .
The unit of ";d
ji is W=.m3 sr s1 /. The line shape function ./ depends on
the type of broadening being effective for the respective line. The function ./ is
normalized as follows: Z C1
./d D 1: (9.4)
1
In case of natural line broadening or a pressure broadened line in the impact
approximation, the function is given by the Lorentz expression [9.3]:
1 H
L ./ D ; (9.5)
2 . 0 V /2 C . H =2/
where H is the full frequency width at half maximum, V is the frequency shift
of the line maximum in relation to an unperturbed line, and 0 is the frequency of
the line maximum for the unperturbed line.
The partition function uza .T / of species a in charge state z is given by:
Dn
iX z
Ea;i
uza .T / D ga;i
z
e kT ; (9.6)
i D1
where i D 1 denotes the ground state and i D n the highest level which is still
bound. To estimate n , the lowering of the ionization energy is calculated using
Unsöld’s formula:
1=3
4
Ezion D 3e 2 .z C 1/2=3 ne ; (9.7)
3
where ne is the electron density. The values uza .T; Ez / are tabulated [9.4]. Further
data for iron and several ionization states of iron are given in [9.5]. Figure 9.2
shows the partition function of neutral (solid line) and singly ionized iron (dashed
line); parameter is the lowering of the ionization energy. For electron densities
between 0.5 and 7 1016 cm3 , EFe0 varies between 0.12 and 0.28 eV. For plasma
temperatures below 10,000 K, this has only a minor influence on the partition
function of neutral iron atoms. For singly ionized iron, the influence of EFe1 is
negligible in the relevant temperature range.
Between the Einstein coefficient Aji and the oscillator strength fij , the following
relation holds [9.6]:
2e 2 gi fij
Aji D ; (9.8)
me c"0 gj 2ji
where e is the elementary charge, me is the electron mass, c is the velocity of light,
"0 is the dielectric constant, gi ; gj is the statistical weight of lower and upper level,
fij is the oscillator strength, and ji is the wavelength of the transition from j to i .
As an example the line emission coefficient for the Fe I line at 538.34 nm (cf.
Fig. 8.14) is estimated with (9.3) by:
RR g Ej
"ji ";d d d Aji hcji Naz uza .Tj / e kT

"ji D
ji
D D ;
FWHM FWHM FWHM
180 12
160 DEFe [eV] 0.1

10
140
partition function Fe [1]
N1
partition function N [1]
0.25
120 0.5
8
100
80 6
Fe 1
60 N0
40 4
Fe 0
20
2
6000 8000 10000 12000 14000 16000
temperature [K]
Fig. 9.2 Partition function for neutral (Fe 0) and singly ionized iron (Fe 1) as a function of the
temperature, parameter: lowering of the ionization energy according to (9.7). Partition function of
neutral (N 0) and singly ionized nitrogen (N I) for EN D 0:1 eV
9.3 Absorption and Optical Thickness 171
where the integration runs over the line and the full solid angle. The line width
(FWHM) is estimated by the Stark width (cf. Sect. 9.4) of that line. The line data
is: Aji D 5:9 107 s1 , gj D 13, and Ej D 6:63 eV [9.3]. The neutral iron
density Naz D n0Fe is determined from ne and the Saha equation. For the temperature
and the electron density as a function of time, we take the data of Fig. 8.16. The
partition function uza .T / D u0Fe .T / is taken from [9.4]. The determined line emission
coefficient for the 538.34 nm line is plotted in Fig. 9.1 for delay times from 0 to 8 s.
While the Bremsstrahlung decays over the first 8 s by many orders of magnitude,
the line emission shows only a slight decrease. The neutral iron line decay is less
pronounced since the neutral iron number density is additionally fed by recombining
iron ions.
The Bremsstrahlung and line emission is compared with the black body emission
in case of an optically thick plasma. The spectral radiance of a black body is given
by Planck’s law [9.7]:
2hc2 1
L;d D 5 h=kT : (9.9)
e 1
The unit of L;d is W=.m2 sr m/. The spectral radiance of a black body is an
emitted power per unit surface area per steradian per wavelength interval. For the
comparison with the Bremsstrahlung and line emission coefficient, an equivalent
emission coefficient is defined assuming that the plasma is described by a sphere
with a radius r:
Ap 3
"bb D 4 L;d D 4 L;d ; (9.10)
Vp r
where Ap D 4 r 2 is the surface area of the plasma sphere with a radius r and
Vp D . 4 3
3 /r is the volume of the plasma sphere.
The unit of the quantity "bb is W=.m3 m/. Assuming r D 2 mm, "bb is plotted in
Fig. 9.1 taking the temperature for the delay times of 1 and 8 s. The Planck curve
describes the upper limit of the plasma emission. However, in this case, there is no
line information available and the emission depends only on the temperature. If the
real laser-induced plasma is not optically thick, its emission will always be below
the Planck curve. Hence, for LIBS, the plasma state has to be optimized in such
a way that high line intensities are achieved which are well above the continuum
emission and still far enough away from the Planck limit, where self-absorption
dominates and the spectral information is lost.
9.3 Absorption and Optical Thickness
For the absorption coefficient including the contribution of the induced emission
holds the following relation [9.7]:
h z h

˛ ./ D Na;i ./Bij 1 e kT ; (9.11)
c
where Na;iz
is the density of species a with charge z in the lower state i and Bij is the
Einstein’s transition probability for absorption.
The unit of ˛ ./ is m1 . Inserting Bij D Aji .gj =gi /.c 3 =8h 3 / and relation
(9.8) in (9.11) yields:
e2 h

˛ ./ D fij Na;i
z
./ 1 e kT : (9.12)
4me c"0
The population density of the lower state Na;i

z
can be expressed by the Boltzmann
formula: gi Ei
Na;i
z
D Naz z e kT : (9.13)
ua .T /
Furthermore, we convert (9.12) to ˛ ./, i.e., the absorption coefficient in the
wavelength interval d by use of the relation ./ D ./.c=2 /. Then, we get
with (9.13):
e2
2 z gi
E
kTi
˛ ./ D kT
hc
fij ./ N a e 1 e : (9.14)
4me c 2 "0 ua .T /
z
As an example this expression is evaluated for the Fe I 538.34 nm line by

approximating the line profile function as follows: ./ D 1= Stark FWHM , where
Stark
FWHM is the Stark width (cf. Sect. 9.4). Taking the data of Fig. 8.16, we obtain
for ˛ ./ in the first 8 s values ranging from 51 to 59 m1 .
For local temperature equilibrium (cf. Sect. 8.4), the ratio of the line emission
coefficient (9.3) to the absorption coefficient (9.12) is given by Kirchhoff’s law:
";d
D L;d ;
ji
(9.15)
˛
where L;d is the spectral radiance according to Planck’s law [cf. eq. (9.9)], with
respect to the frequency interval: L;d D L;d .2 =c/, which can be verified easily
by use of the relations (9.3), (9.12), and (9.13).
The optical thickness for a nonhomogeneous plasma along a line of sight in the
x-direction is defined as: Z s

s D ˛ .; x/dx; (9.16)
0
where s is the geometrical thickness of a plasma layer and ˛ .; x/ is the absorption
coefficient as a function of x.
For a geometric plasma extension of s D 2 mm, a homogeneous plasma, and the
calculated absorption coefficients, an optical thickness of s 0:11 results for the
chosen example of the Fe I 538.34 nm line. For a plasma layer of thickness s and
a constant plasma temperature, the following spectral radiance is emitted along the
line of sight running parallel to the plasma extension s:
Z

I .s/ D L;d et dt D L;d 1 e ; (9.17)
0
9.4 Line Broadening 173
I .s/
L;d D ˛ sL;d D ";d
ji s for
1;
I .s/ L;d for

1;
where dt D ˛ dx and L;d is Planck’s law, see (9.9), here related to the frequency
interval.
The first approximation holds for an optically thin plasma, and the latter for an
optically thick plasma. Hence, the plasma can be considered as optically thin for the
case of the Fe I 538.34 nm line.
9.4 Line Broadening
The relevant broadening mechanisms for emission lines of LIBS plasmas are the
Doppler broadening and the Stark broadening. The Doppler broadening is an
inhomogeneous broadening and the line profile can be described by a Gaussian
function. The line width (FWHM) is given by [9.7]:
1=2
8kT ln 2
D D ji 0 ; (9.18)
Ma c 2
where Ma is the mass of the emitting atom species a and ji 0 the central wavelength
of the transition from j to i . For an iron atom, a temperature of 8,500 K, and D
400 nm, we have D D 3:5 pm.
Collisional broadening in a plasma by collisions with charged particles causes the
so-called Stark broadening. The line profile can be described by a Lorentz function
[cf. eq. (9.5)]. Multielectron species do not have a permanent dipole moment. The
external electric field induces a dipole moment. The change of the emission fre-
quency is in this case proportional to the square of the electrical field strength. This
behavior is called quadratic Stark effect. The broadening of isolated transitions of
neutral atoms and singly ionized ions is mainly caused by collisions with electrons.
Hence, half-widths can be calculated by the electron collision approximation. The
quasi-static broadening by ions is taken into account as a small correction. The line
width (FWHM) for neutral atoms is approximately given by [9.8]:

3 1=3 ne
Stark D 1 C 1:75A 1 ND WFWHM ; (9.19)
4 1016 cm3
where Stark is the full width at half maximum (FWHM) of the line, A is the
dimensionless coefficient, and ND is the number of particles in the Debye sphere,
cf. (8.16), and WFWHM is the Stark broadening parameter for the full width (FWHM).
In relation (9.19), we used the parameter WFWHM instead of W D WFWHM =2,
which is usually tabulated in theoretical calculations [9.9], since often experimen-
tally determined Stark broadening parameters are given as the full width at half
maximum, see, e.g., [9.10, 9.11]. The line shift is given by [9.7]:
ne ne 1=4
shift D D ˙ A
1016 cm3 1016 cm3

3 1=3 ne
1 ND WFWHM ; (9.20)
4 1016 cm3
where D is the Stark shift parameter.
The coefficients WFWHM , A, and D are independent of the electron density ne ,
and are slowly varying functions of the electron temperature. A wavelength shift is
defined to be positive, if the wavelength shifts to longer wavelength (red shift). The
minus sign in the shift formula (9.20) applies to the high temperature range of those
few lines that have a negative value of D=WFWHM at low temperatures. For singly
ionized ions, the factor (3/4) is to be substituted by 1.2.
The formulas (9.19) and (9.20) are good approximations (20–30%) in the
following parameter range [9.8]:
ne 1=4
ND 2 and 0:05 < A < 0:5: (9.21)
1016 cm3
The first condition is estimated for a temperature of 7,000 K and an electron
density of 1017 cm3 (the temperature is at the lower limit of the observed plasma
temperatures – see Figs. 8.16 and 8.18 – and the electron density is at an upper
limit) yielding ND D 2:5. For the second condition, the range of observed electron
densities between 3 1015 cm3 and 1017 cm3 – see, e.g., Fig. 8.23b – is taken into
account, which results in 0:05 < A .0:74; : : : ; 1:8/ < 0:5. From this, we have
0:07 < A < 0:3. The dimensionless parameter A is tabulated for isolated neutral
atom lines in Appendix IV, pp. 320–362 of [9.9].
Neglecting the ionic part of the broadening relation, (9.19) reduces to:
ne
Stark D WFWHM : (9.22)
1016 cm3
This relation can be used to determine the electron density taking a measured line
width – (FWHM) – after correction due to apparative and Doppler broadening – and
a known Stark broadening coefficient WFWHM [9.10–9.12].
The different normalized line profiles [cf. (9.4)] are plotted for Doppler broaden-
ing D ./ and Lorentz broadening L ./ in Fig. 9.3 taking as an example the Fe I,
538.34 nm line with the same widths (FWHM) for both profiles of 0.05 nm (this
value is chosen for illustration only). The additionally shown Voigt profile V ./
has the same width (for the calculation of V ./ a Doppler width of 0.03 nm and a
Lorentz width of 0.031 nm is taken), this function will be described in the following.
In general, the Doppler broadening and Stark broadening are superimposed. The
resulting line shape is described by the normalized Voigt profile as follows:
p
2 ln 2=
V ./ D K.u; a/ (9.23)
D
2 × 1010
Doppler
Voigt
Γi(λ) [m-1]
1 × 1010 Lorentz
0
538.25 538.30 538.35 538.40 538.45
wavelength [nm]
Fig. 9.3 Calculated normalized line profile functions of the same width (FWHM) for Doppler,
Lorentz, and Voigt
with the Voigt profile function:

Z C1 2
a et
K.u; a/ D dt (9.24)
1 .u t/2 C a2
and the variables:
p
2 ln 2
uD ji 0 ; (9.25)
D
p N C L p L
a D ln 2 ln 2 ; (9.26)
D D
where N is the natural line width, L is the Lorentz line width, and D is the
Doppler line width.
Figure 9.4 shows the Voigt profile function calculated for the experimentally
determined temperature and electron density of Fig. 8.16 at the delay times 1, 4,
and 8 s. The line profile is narrowing for later delay times due to the strongly
decreasing electron density which results in a reduced Stark broadening. On the
other hand, the Doppler widths are nearly constant in the time interval from 1 to
8 s; hence, the parameter a [cf. (9.26)] drops significantly.
Experimentally measured line shapes can be fitted by the Voigt function, see
Fig. 9.5. The coefficient of determination in the shown example of the Fe 438.35 nm
line is better than 0.999. The fit yields the widths of the Gaussian and the Lorentzian
part of the Voigt profile [cf. (9.26)]: m G D 0:021 nm, L D 0:018 nm. The index
m
“m” indicates that these values are gained from the measurement.
The measured widths have to be corrected by the apparative broadening to obtain
the correct line width of the Doppler and Lorentz part using the following relations:
2 2
m
G D 2D C aG ; (9.27)
3.68, 8
Voigt profile function K(u, a) [1] 30
Fe I, 538.34 nm
parameter a, t delay [μs]
20
6.97, 4
10
19.6, 1
0
538.0 538.2 538.4 538.6
wavelength [nm]
Fig. 9.4 Calculated Voigt profile function for the Fe I, 538.34 nm line and the parameters of
Fig. 8.16
10000
measured spectrum
Voigt function
8000
intensity I Fe 438 [cts]
6000
4000
2000
438.2 438.3 438.4 438.5

100
residues
-100
438.2 438.3 438.4 438.5
wavelength l [nm]
Fig. 9.5 Experimentally measured line shape of the Fe I line 438.34 nm and fit of the data by a
Voigt function
m
L D L C aL ; (9.28)
where m
G, m
are the measured Gaussian, Lorentzian line widths (FWHM);
L
D , L are the correct Doppler, Lorentz line widths (FWHM), and aG , aL
are the apparative broadening described by a Gauss, Lorentz function, respectively.
For the determination of the apparative broadening, the spectrometer-detector
system should be operated in a configuration that is also used for the LIBS
measurements. This refers, e.g., to the width of the entrance slit, the spectral range
studied, and the settings of the detector (amplification, gating). In the interesting
spectral range, the Lorentz and Gauss widths of a set of lines of the irradiated sample
are determined as a function of the burst energy. Figures 9.6 and 9.7 show the results
for an echelle MCP–CCD system (cf. Sect. 4.2). In Fig. 9.6, the measured Lorentz
line width is plotted – determined after fitting the emission line by a Voigt profile –
as a function of the burst energy for four iron lines from which the 381.30 nm
line shows the smallest broadening. The Lorentz line widths decrease for smaller
burst energies due to a decreasing electron density. The data of the narrowest line
381.30 nm is fitted by a quadratic polynomial yielding the Lorentz width for a burst
energy of 0.5 mJ: m L D .2:7 ˙1:5/ pm. This value is an upper limit of the Lorentz
part of the apparative broadening, i.e., aL .2:7 ˙ 1:5/ pm.
Figure 9.7 shows the Gauss widths of the same lines. These widths do not depend
on the burst energy but are approximately constant. The Doppler broadening being
of the order of 3.5 pm and less does not contribute significantly to these widths
[cf. (9.18) and (9.27)]. Hence, the average Gauss width is: m G aG D
.20:1 ˙ 0:8/ pm. The total apparative broadening V can then be estimated using
a
16
14
Lorentz line width ΔλL [pm]
12
m
10
4
Fe 381.30 Fe 384.99
2
Fe 382.78 Fe 381.58
0
0.5 1.0 1.5 2.0 2.5 3.0
burst energy E b [mJ]
Fig. 9.6 Lorentz line widths m

L of a set of iron lines as a function of the burst energy for an
echelle MCP–CCD system to determine the Lorentz part of the apparative broadening
24
22
Gauss width ΔλmG [pm]
20
18
16
Fe 381.30 Fe 384.99
Fe 382.78 Fe 381.58
14
0.5 1.0 1.5 2.0 2.5 3.0
burst energy Eb [mJ]
Fig. 9.7 Gauss line widths m

G of a set of iron lines as a function of the burst energy for an
echelle MCP–CCD system to determine the Gaussian part of the apparative broadening
the following approximation of the width (FWHM) of a Voigt profile for given
Gauss and Lorentz parts [9.13]:
q
2 2
aV D 0:5346 aG C 0:2166 aG C aL : (9.29)
Hence, for the total apparative broadening, we obtain for the example discussed:
aV 20:5 pm.
In case of a nonnegligible optical thickness and a homogeneous plasma, the
shape of the emitted line in the vicinity of the line maximum is proportional to
the following expression [cf. (9.16) and (9.17)]:
./s
./ D 1 e˛ ; (9.30)
where s is the extension of the plasma and ./ is a nonnormalized shape function.
This function is shown in Fig. 9.8 for the Fe I 538.34 nm line and different particle
densities Naz D nFeI ranging from 1017 to 2 1019 cm3 (electron temperature and
plasma size are kept constant using the values: T D 7;300 K, s D 2 mm. Further
assumptions for the calculations are described in the following as case (b).)
With increasing particle density, the absorption becomes more and more impor-
tant. The line is broadening and the peak flattens. The peak of the line approaches
a value of 1 corresponding to the spectral radiance of the Planck function [cf.
(9.17)]. The nFeI values taken for Fig. 9.8 are much higher than the experimentally
determined values, which are typically in the range of 5 1015 to 5 1016 cm3 (cf.
Tables 8.2 and 8.3).
parameter nFe I [1017cm– 3]

1.0
200
0.8
(1–e–α(λ)s) [1]
0.6 80
0.4
0.2 20
1 5
0.0
537.5 538.0 538.5 539.0
wavelength [nm]
Fig. 9.8 Calculated line shape of the Fe I, 538.34 nm line according to (9.30), parameter: iron
particle density
The radiance of the emission line can be calculated by integrating relation (9.17)
over the line assuming that the variation of L;d is small in the narrow wavelength
range of the spectral line and that the plasma is homogeneous:
Z 1
ji ;d ./s z
I.s/ D L 1 e˛ d D Lji ;d Na ; T; s : (9.31)
0
The integral .Naz ; T; s/ is a function of the particle density Naz of the emitting
species, the temperature, and the plasma extension. This integral is calculated as a
function of the iron atom density Naz D nFeI for the Fe I, 538.34 nm line assuming
a constant temperature of 7,300 K and a constant plasma size of 2 mm. Two cases
were considered: (a) the parameter a [see (9.26)] is kept constant at the value 3.68
corresponding to the experimentally observed Stark widths after a delay time of 8 s
(cf. Figs. 8.16 and 9.4); (b) while varying nFeI in a constant plasma volume at a fixed
temperature T , the electron density cannot be constant but varies according to the
Saha equation. Hence, the Stark width changes as well and therefore the parameter
a too. Figure 9.9 shows the results. When the particle density nFeI is increased, the
integral and consequently the line radiance increases as well. In case (a), the slope
changes for high particle densities, which is a consequence of self-absorption. How-
ever, in case (b) where the line broadening induced by the higher electron density is
considered, this change is much less pronounced. In the range of particle densities
nFeI shown in Fig. 9.9, the parameter a varies in this case between 2.5 and 37.
For quantitative analysis with LIBS, the analyte emission lines have to be
selected carefully to achieve a good response of the line intensity versus the analyte
10-10 a = 2.5 - 37
Fe I, 538.34 nm
T = 7300 K, s = 2 mm
integral L [m]
a = 3.68
10-11
10-12
1016 1017
nFe I [cm-3]
Fig. 9.9 Integral as a function of the particle density nFeI for the Fe I 538.34 nm line assuming a
homogeneous plasma with a size of 2 mm and a constant temperature of 7,300 K. The parameter a
is kept constant or varies, see text. The horizontal bar indicates the range of nFeI values determined
spectroscopically (cf. Tables 8.2 and 8.3)
concentration and to avoid self-absorption. Figure 9.10 shows a section of LIBS

emission spectra of a set of aluminum samples with a Mg content varying over
more than four orders of magnitude. Two Mg lines are visible where the line at
279.078 nm increases monotonously with increasing Mg concentration, whereas the
line at 279.553 nm changes its shape showing a dip at high Mg contents. The inset
diagram shows the response as a function of the Mg concentration. The 279.553 nm
line shows a strong saturation behavior.
The dip is an indication that the LIBS plasma is not homogeneous but consists
of regions having different particle densities and temperatures. Outer colder plasma
regions absorb radiation from inner parts. The spectral width of this absorption is
smaller than the emission from the core due to reduced line broadening at the outer
plasma regions. At the center of the line, the optical thickness is greater; hence, we
look only at the outer regions of the plasma, where the temperatures are lower and
therefore the emission is weaker.
In order to assess the sensitiveness of a line for self-absorption, the absorption
coefficient (9.14) is factorized as follows:
˛ ./ D ./Naz kt ; (9.32)
where the quantity kt depends on the atomic data of the line and the temperature
only:
e2 gi Ei hc

kt D 2
fij 2 z e kT 1 e kT : (9.33)
4me c "0 ua .T /
Table 9.1 shows the atomic data of the two Mg lines of Fig. 9.10 and the calculated
kt values for two temperatures [9.14, 9.15].
5 × 104 10000
intensity [arb. u.]

1000
4 × 104
100 279.078 nm
intensity [a.u.]
279.553 nm
3 × 104 1 10 100 1000 10000

Mg content [μg / g]
increasing
Mg content
2 × 10 4
1 × 104
0
278.6 278.8 279.0 279.2 279.4 279.6
wavelength [nm]
Fig. 9.10 LIBS spectra of a set of aluminum samples with varying Mg concentration
Table 9.1 Atomic data of the Mg II lines of Fig. 9.10 and kt values for two temperatures T1 D
9;000 K and T2 D 6;000 K
Mg line (nm) Ionization stage gi Ei .eV/ fij kt .T1 /.m3 / kt .T2 /.m3 /
279.078 II 2 4.423 0.953 2:16 1030 1:26 1031
28
279.553 II 2 0 0.629 4:30 10 4:35 1028
gi D statistical weight of the lower level, Ei D energy of the lower level, fij D oscillator strength
The Mg II 279.553 nm line is a resonance line and the kt values are by orders of
magnitude greater than for the line 279.078 nm indicating their strong sensitiveness
to self-absorption. As a first estimate whether a line is affected by self-absorption in
a LIBS plasma, its kt value calculated for typical temperatures should be less than
about 5 1030 m3 .
The Mg II line at 279.078 nm is Stark broadened and theoretical as well as
experimental Stark width data is available [9.8, 9.14]. Considering the apparative
broadening, the determined electron density is in the range of 1:2–1:5 1017 cm3 .
Using relation (9.3), we can estimate the temperature from the line ratio of the
279.078 nm line to the 279.553 nm line for the sample having a Mg content of
100 g=g. Due to the fact that the latter line is affected by absorption, this estimate
yields an upper limit of the temperature. Taking into account the different Stark
widths of these lines, we obtain T < 13;000 K.
Figure 9.11 shows another part of the LIBS spectrum with an Al I resonance line
with line reversal demonstrating the shift between the absorption and emission max-
imum of 26 pm. The emission of the plasma core is shifted to longer wavelength.
By use of relation (9.20), neglecting the ion correction term, we can estimate the
4 × 104 absorption max. 394.401 nm

emission max. 394.427 nm
3 × 104
intensity [a.u.]
2 × 104
1 × 104 spectral shift
0
393.5 394.0 394.5 395.0 395.5
wavelength [nm]
Fig. 9.11 Al I line at 394.401 nm with line reversal
electron density difference between the plasma core and the plasma shell. With the
Stark shift parameter for the Al I 394.401 nm line taken from reference [9.9], we
obtain for the electron density difference ne D 1:07 1017 cm3 , a value which
is in the same order as the electron density determined from the Stark width of the
Mg II line 279.078 nm (cf. Fig. 9.10).
References
9.1. H. Griem, Plasma Spectroscopy (McGraw-Hill, New York, 1964)

9.2. C. Breton, J. Schwob, Ch. 8, in Vacuum Ultraviolet Radiation Physics, ed. by N. Damany,
J. Romand, B. Vodar (Pergamon Press, Oxford, 1974)
9.3. J. Fuhr, G. Martin, W. Wiese, S. Younger, Atomic transition probabilities for iron, cobalt,
and nickel. J. Phys. Chem. Ref. Data 10, 305–524 (1981)
9.5. J. Halenka, J. Madjev, Tables of partition functions of iron, Fe I–Fe X. Acta Astronom. 52,
199–206 (2002)
9.6. Bergmann-Schäfer, Lehrbuch der Experimentalphysik, Band IV, Teil 2, Aufbau der Materie,
Hrsg. H. Gobrecht, Walter de Gruyter, Berlin, 1981
9.7. A. Anders, A Formulary for Plasma Physics (Akademie-Verlag, Berlin, 1990)
9.8. G. Bekefi, Principles of Laser Plasmas (Wiley, New York, 1976)
9.9. H. Griem, Spectral Line Broadening by Plasmas (Academic, New York, 1974)
9.10. N. Konjevic, M. Dimitrijevic, W. Wiese, Experimental Stark widths and shifts for spectral
lines of neutral atoms (a critical review of selected data for the period 1976 to 1982). J. Phys.
Chem. Ref. Data 13, 619–647 (1984)
atoms (a critical review of selected data for the period 1983 to 1988). J. Phys. Chem. Ref.
Data 19, 1307–1385 (1990)
9.12. S. Freudenstein, J. Cooper, Stark broadening of Fe I 5383 Å. Astron. Astrophys. 71, 283–288
(1979)
References 183
9.13. J. Olivero, R. Longbothum, Empirical fits to the Voigt line width: a brief review. J. Quant.
Spectrosc. Radiat. Transf. 17, 233–236 (1977)
9.14. R. Kurucz, B. Bell, Atomic Line Data (Smithsonian Astrophysical Observatory, Cambridge,
1995), http://cfa-www.harvard.edu/amp/tools.html
9.15. S. Bukvic, A. Sreckovic, S. Djenize, Mg II h and k lines Stark parameters. New Astron. 9,
629–633 (2004)
Chapter 10
Modeling of Plasma Emission
A modeling of the emission spectra of laser-induced plasmas for LIBS can

contribute to a better understanding of the relevant parameters influencing the
observed measuring radiation and potentially be an instrument for a quantitative
analysis as well. In a rigorous treatment, such a modeling of the plasma emission
should be part of a physical model of the LIBS method including plasma generation
and plasma emission. Such a description is beyond the scope of this chapter.
Figure 10.1 illustrates schematically a physical model with the input quantities
given by the experimental conditions and the desired output quantities.
The laser beam parameter IL .R; t/ is the irradiance of the incident laser beam as
a function of coordinate R and time t. For simplicity, a rotationally symmetric laser
beam profile is assumed and R is a radial coordinate giving the distance from the
beam propagation axis. In practice, the quantity IL .R; t/ is often not known or only
known approximately. The laser pulse or laser burst energy EL can be measured
quite easily, so at least the spatially and temporally integrated value of IL .R; t/ is
known.
The wavelength of the laser radiation has an effect on the beam focusing and
the absorption of the radiation by the specimen, the material vapor, and the plasma.
Further laser parameters such as the polarization state and beam pointing stability
may also have an effect on the interaction process with the specimen.
The ambient atmosphere can be characterized by the pressure pg .r; t/. Prior to
the irradiation of the first laser pulse, this pressure can be assumed to be constant.
However, in case of pulse bursts consisting of two or more pulses irradiated within
a time interval of a few microseconds, the preceding pulses induce a local variation
of the ambient particle density and pressure which influences the dynamics of the
plasma expansion for the subsequent pulses (cf. Chap. 8) [10.1]. In this case, the
pressure depends on time t. The ambient atmosphere consists of different species
a, whose densities are described by Na;g .r; t/. Besides the dependence on t, these
densities are a function of the coordinate r. In an even more rigorous treatment,
the mixing of particles ablated by the laser and particles from the initial ambient
atmosphere has to be considered. Hence, in a pulse burst, the subsequent pulses

186 10 Modeling of Plasma Emission
Fig. 10.1 A physical model for the LIBS method with the input and output quantities
encounter a pressure pg .r; t/ to which not only the species of the ambient gas
but also the ablated and vaporized particles will contribute. Further effects – not
shown in Fig. 10.1 – arise from residual particulates originating from ablations by
previous laser pulses which are still in the interaction volume and act as scattering
and absorption centers for subsequent laser pulses. These scattering and absorption
processes again depend on the wavelength of the irradiated laser beam. A gas flow
in the interaction region will also influence the extent to which such particulates are
moved out of the propagation path of the laser beam (cf. Sect. 3.1, Fig. 3.4).
Last but not least the material parameters are essential input quantities too. The
concentrations of the species a in the specimen are described by the quantities
ca . The initial absorption of the laser beam depends on the natural absorption
coefficient ˛./. The specific heat capacity cp influences the temperature rise at
the surface of the specimen. The specific melt and evaporation enthalpies Em ; Ev
as well as the thermal diffusivity are necessary input quantities to describe the
melting and evaporation of the specimen leading to the material ablation and crater
formation (in case of, e.g., a metallic solid specimen). Not shown in Fig. 10.1 are
further material properties such as the surface roughness, crater formation, or the
inclination of the surface of the specimen versus the incident laser beam axis.
Additionally most of these material parameters are a function of the temperature
T and these dependencies are not known for many materials.
The desired output parameters are the electron density Ne .r; t/, the particle
densities of species a with charge zW Naz .r; t/, the electron temperature Te .r; t/,
the particle temperatures Taz .r; t/, and last but not least the spectrum emitted by
the plasma I .t/ as a function of time and – for comparison with experimental
data – the quantity deduced thereof St ./ describing the spectrum integrated over a
time interval tint starting after a delay time tdelay after the last irradiated laser pulse.
The quantities Te .r; t/ and Taz .r; t/ require the establishment of local equilibrium
conditions among the ensemble of particles concerned.
10 Modeling of Plasma Emission 187
The main interactions and processes to be considered in the physical model are
listed on the left side of Fig. 10.1. Some aspects of these processes have already been
studied, e.g., evaporation [10.2], plasma generation [10.3], ambient gas pressure
[10.4, 10.5], plasma dynamics [10.1], and interaction of laser radiation and plasma
[10.6]. Several papers were published treating approaches for a physical model of
the LIBS plasma, e.g., [10.7, 10.8]. Considering the fact that many of the necessary
material parameters for a physical model of the LIBS plasma formation are not
known, the focus is put here on a heuristic modeling of the spectral plasma emission
after the interaction of the laser pulse with the specimen.
Several papers were published in the last years presenting models to describe
the postbreakdown phase of a LIBS plasma [10.9–10.12]. A spherical plasma is
considered, which is assumed to be in local thermodynamic equilibrium (LTE). The
idea is to describe the dynamic plasma evolution after the interaction of the laser
pulse with the target.
In a simplified theoretical approach to model an inhomogeneous optically thick
laser-induced plasma, it is assumed that the initial temperature distribution can be
described by a parabolic function with the temperature maximum at the center of
the plasma [10.9]. Furthermore, an expansion law for the plasma radius into an
ambient vacuum is defined. The number of fitting parameters for the temperature
profile and the plasma size amounts to 5. The plasma composition is assumed to be
constant. An approximate solution of the radiation transport equation was used and
spectra were calculated for a binary system of Si and C in the wavelength interval
from 287 to 289 nm comprising a single strong Si line at 288.16 nm. Further model
outputs are spatial and temporal distributions of atom, ion, and electron number
densities. No direct experimental verifications were shown. The authors stated that
the model inputs could easily be measured in an experiment. This may be possible
for a fundamental study, for a practical application of the model the additional effort
to determine experimentally the input parameters – such as total number of atoms
in the plasma volume, the temperature function, the plasma expansion function, and
the atomic stoichiometry of plasma species – is assessed to be prohibitive, since this
would require an extensive additional apparative equipment.
In a further model expansion, the gas dynamic equations were coupled with the
equation of radiative transfer for a spherical plasma evolving into an ambient vac-
uum [10.10]. Partition functions from the tables of Drawin and Felenbok were used
[10.13]. The authors stated uncertainties of ratios of partition functions being less
than 50%. This means that for a comparison of calculated line ratios to measured
line ratios, deviations may occur of up to 50% even in the case of availability of a
perfect postbreakdown plasma model. In this regard, expectations of the practical
usefulness of a model have to be oriented to a more qualitative description as long
as significantly more accurate partition function data are not available. The model
output is a series of synthetic spectra in the spectral range of 280–290 nm which
includes three spectral lines from the binary Si–C system taken. The intention of
the authors is to vary the initial conditions until a close fit between the synthetic
and the experimental spectra is obtained; however, no experimental verification was
presented.
An experimental verification of a radiative model of laser-induced plasma

expanding into vacuum was reported for a pure iron foil and a set of eight Al
samples [10.11]. The goal was to predict the composition of the plasma from fits of
synthetic spectra to measured spectra. Instrumental parameters such as slit function,
diffraction grating efficiency, and detector response function were considered. The
experiments were carried out at an ambient pressure of 0.5 torr, which results in
rather large observed plasma diameters of about 3 mm achieved after 200 ns. At
a delay time of 40 ns and an integration window of 55 ns, a spatially integrated
temperature was determined using Fe II lines, yielding 16,500 K with an error of
13%. A qualitative comparison of a synthetic iron spectrum in the range from 220
to 300 nm calculated for an average temperature of 16,500 K with the experimental
spectrum showed similarities; however, a detailed look reveals strong deviations in
the range of 50% to several 100%. Using Fe II lines for the experimental temperature
determination certainly has led to an overestimation of the temperature.
For the aluminum alloys, the authors could demonstrate fits with correlation
coefficients greater than 95% for a spectral region between 278 and 288 nm
comprising up to six lines [10.11]. They found a systematically higher intensity
for the calculated Mg II 280.27 nm line than in the experiment and attributed this
to an imprecise value of the line transition probability. However, taking the data
from the Kurucz-files for the two Mg II lines 279.55 and 280.27 nm, assuming the
same line profile, negligible self-absorption and calculating the Boltzmann factors
a ratio of IMg II279:55 =IMg II280:27 D 2:05 is obtained for a temperature of 22,000 K
(the influence of the temperature for this estimation is very low, since the two upper
energy levels of the two Mg II lines differ only by about 0.01 eV) [10.14]. From the
experimental spectra shown in Fig. 7 of [10.11], a line ratio of about 1.9 is deduced,
which is quite close to the expected value.
The two Mg II lines considered are resonance lines, hence they may be affected
by self-absorption. The ratio of the two lines in the synthetic spectrum of sample
AA3 shown in Fig. 7 [10.11] is significantly less than 2 indicating the effect of
self-absorption implemented in the applied model. However, the measured ratio
for sample AA3 is consistent with the assumption of negligible self-absorption of
these two Mg II lines at this level of Mg concentration (for sample, SM 10 having a
fivefold higher Mg concentration the experimental line ratio shown in Fig. 7 [10.11]
indeed decreases to about 1.75 showing that self-absorption is becoming important
at these Mg concentration levels). Hence, the discrepancy found by the authors
cannot be attributed to imprecision of the available line transition probabilities.
In a further paper of the same group, the radiative model of the postbreakdown
laser-induced plasma expanding into ambient gas was studied [10.12]. A qualitative
comparison between synthetic spectra and measured spectra at different delay times
.0:1–4 s/ was shown for a SiC plasma in an argon atmosphere at atmospheric
pressure. The spectral range covered two resolved emission lines of silicon and
carbon (Si I 288.16 nm and C II 283 67 nm/283.76 nm). The authors concluded
that the determination of intrinsic plasma characteristics from the experimentally
determined parameters can in principle be solved. As a limitation they proposed to
introduce at least two temperatures for electrons and heavy particles.
10.1 Heuristic Model 189
The profiles of single Ti ion lines generated by irradiating UV laser pulses at

308 nm on titanium targets in low-pressure nitrogen atmospheres (1 mbar) were
studied at the initial high-density plasma evolution (200 ns) [10.15]. At this phase,
the plasma expands fast into the low-density ambient atmosphere. Single observed
asymmetric and self-reversed line profiles of Ti II lines could be described by a two
zone one-dimensional plasma approach. (The oscillator strength of 0.0165 given in
[10.15] for the Ti II 346.150 nm line is not correct, and the correct value is 0.0141,
cf. e.g., [10.14].)
A one-dimensional two region plasma description was studied to fit a set of
curves of growths (COG) of five iron atomic lines in the spectral region of 373–
382 nm [10.16]. The number of parameters used to fit the observed COG is 7 for
the two region plasma model. The authors could show that for intense lines, at
high concentrations of the element in the sample and at initial times of the plasma
evolution it is necessary to use a model of the laser-induced plasma that includes
regions with different values of temperatures, electron densities, and atom densities.
From the fitting of the set of COGs, a ratio of number densities of emitting neutral
atoms of the inner to the outer region of 200 were determined, while the length of
the central plasma region is stated as 1 mm and the length of the outer regions as
4 mm. Ambient gas effects were not studied.
Instead of modeling the postbreakdown dynamics of the LIBS plasma and calcu-
lating emission spectra linked to this dynamics in the following a simplified heuristic
model is presented where no plasma expansion is considered and the inhomogeneity
of the plasma is described by a shell model of the plasma surrounded by an ambient
gas. This approach is led by experimental observations. For quantitative LIBS
analysis, the spectrum emitted by the plasma expanding into the ambient atmosphere
is typically detected at a delay time being in the range of 1 s after the interaction
of the laser pulses with the specimen. At this time region, the plasma dynamics has
already slowed down significantly.
10.1 Heuristic Model

In the heuristic model, it is assumed that the composition of the specimen, the
plasma, and the plasma state is known. The goal is to calculate the spectral plasma
emission for an inhomogeneous plasma surrounded by an ambient atmosphere.
The plasma can be described by a simplified spherical shell model as illustrated
in Fig. 10.2 [10.17]. A spherical description is assumed, since the experimentally
observed plasma geometry 500 ns after the irradiation of the laser pulse (or the last
pulse of a pulse burst) has nearly a spherical shape. Deviations from a spherical
geometry certainly exist in the time interval up to 500 ns, see, e.g., Figs. 4 and
5 in [10.1]. However this time domain is usually not used for quantitative LIBS
measurements. While for single pulses the plasma stays in contact with the sample
surface, the plasma is detaching from the sample surface for collinear double pulses
as shown in Fig. 8.12 by high-speed images of the laser-induced plasma observed in
side-on direction. Especially for double pulses, the assumption of a spherical plasma
Fig. 10.2 Simplified plasma

model for the heuristic
approach. (a) The plasma is
described by two shells: an
inner sphere or plasma core
S1 and an outer shell S2. The
ambient atmosphere is
described by shell S3; (b)
densities, and (c) temperature
distribution in the regions
S1–S3. “Copyright 2009,
American Institute of Physics
[10.17]. Reprinted with
permission”
geometry for delay times greater than 0:5 s is well justified by these experimental
observations.
The laser-induced plasma is described by two shells: the inner sphere or plasma
core S1 and an outer shell S2. By this assumption the experimental fact of a
nonhomogeneous plasma is described in the most simple approach analogous to
[10.15]. Within these shells, spatially homogeneous values of the physical quantities
are assumed. For example, sphere S1 contains mainly the ablated material, whereas
in shell S2 particles of the ablated material and the ambient atmosphere can be taken
into account. Furthermore, it is assumed that these shells contain only atoms, ions,
and electrons. Molecules and particulates are not taken into account. Shell S3 is used
to describe the stationary ambient atmosphere in the region between the plasma and
the detector plane of a spectrometer operated under the same atmospheric conditions
or the entrance slit of a vacuum spectrometer. This atmosphere absorbs a part of the
emitted radiation of the shells S1 and S2 depending on the wavelength. The ambient
gas may consist of mixtures of atoms and molecules.
Figure 10.3 illustrates schematically the heuristic model with the input quantities
given by the experimental conditions and the desired output quantities. The input
quantities and assumptions for the heuristic model are listed in detail in Table 10.1.
The assumption of a constant plasma geometry is a strong simplification. In
a self-consistent physical model, the plasma dynamics in terms of the plasma
expansion and the temporal evolution of the particle densities and temperatures are
interrelated. In order to reduce the number of input parameters for the heuristic
model, the plasma geometry is kept constant. From the experimental point of view,
the spectral emission used for LIBS is observed after a delay time of several
100 ns after the laser pulse irradiation. At this time regime, the plasma has already
attained its maximum extension and starts to decay with a moderate velocity. The
characteristic time of this decay can be estimated as follows:
10.1 Heuristic Model 191
Fig. 10.3 The heuristic model to calculate the spectral plasma emission
D dPlasma =vth ; (10.1)
where dPlasma is the diameter of the spherical plasma and vth is the thermal velocity of
the atoms and ions in the plasma. Relation (10.1) yields the order of magnitude of
the time after which the plasma disintegrates due to the nondirectional movement
of the heavy particles. For a plasma of iron atoms and ions at a temperature of
9,000 K and a diameter of dPlasma D 2 mm, the lifetime estimated by (10.1) is D
1:1 s. For integration times of the plasma emission being smaller than this value,
the change of the plasma geometry will not play an important role.
The particle densities are described as follows:
2 3
ca "a 1; 2
6 ma 7
Na 1; 2 D 6 7
4 P cb "b 1; 2 5 Ntot1; 2 ; (10.2)
b mb
ma;b , mass of an analyte atom; and Ntot1;2 , total particle density (all species)
originating from the specimen in sphere 1, shell 2.
For the ambient gas density in sphere 1 and shell 2 holds:
g g g
Na 1; 2 D ka 1; 2 Ntot 1; 2 ; (10.3)
g
ka 1; 2 , fraction of the density of species a of the ambient gas in sphere 1, shell 2; and
g
Ntot1;2 , total particle density of the ambient gas in sphere 1, shell 2.
For the density of species a of the ambient gas in shell 3, we set:
g g g
Na 3 D ka 3 Ntot3 ; (10.4)
Table 10.1 List of input quantities and their description for the heuristic model
Quantity Description Units
ca Concentration of the element a in the specimen kg/kg
r1 Radius of plasma core 1. The sphere 1 extends from r D 0 to r1 . For m
simplicity the radius r1 is kept constant, i.e., no dynamics of the
plasma geometry is considered
r2 Outer radius of shell 2. The thickness of shell 2 amounts to: r2 –r1 . The m
radius r2 is kept constant
r3 Outer radius of shell 3. The thickness of shell 3 amounts to: r3 –r2 . m
This distance describes the propagation path of the emitted light
through the ambient gas between the plasma and the detector of
the spectrometer or the spectrometer entrance slit in case of a
vacuum spectrometer
"a 1; 2 Weighting factor describing the proportion of an element a in the 1
plasma in relation to its proportion in the specimen. If the
laser-induced evaporation process does not depend on the
constituents, structure or chemical bonds, then the concentration
relations inside the plasma are identical to those of the original
specimen. In this case holds: "a;1;2 D 1. The relation "a;1;2 < 1
corresponds to a depletion of the respective element in the plasma
and "a;1;2 > 1 to an enrichment
Na 1; 2 Particle density of species a originating from the specimen in sphere 1 1=m3
or shell 2. Total density of particles
P originating from the specimen
in sphere 1 or 2: Ntot 1;2 D Na 1;2
a
g
Na 1;2 Particle density of species a originating from the ambient gas in 1=m3
sphere 1 or shell 2
g
Na 3 Particle density of species a (atoms or molecules) of the ambient gas 1=m3
in shell 3. For simplification, it is assumed that this particle density
is stationary
T1; 2 Temperature in sphere 1 and shell 2. A local temperature equilibrium K
is assumed
Tg Temperature of the ambient gas assumed to be stationary K
a Apparative broadening given as full width at half maximum (FWHM). pm
Typical values are in the range of 10–70 pm
./ Spectral characteristics of the observation channel given, e.g., by the 1
features of a light receiving and transmitting optics such as lenses,
mirrors or fiber optics, reflectivity of the grating, and spectral
sensitivity of the detector. The function ./ is normalized:
0 < ./ 1. ./ can be defined in an interval .min , max / by a
series of data points .i /
(order) Weighting factor between 0 and 1 describing the fraction of diffracted 1
power for a given order of the grating spectrometer. If .1:/ D 1,
then only the first diffraction order is considered; if .1:/ D 0:8
and .2:/ D 0:2 then the first order contributes 80% and the
second 20% to the observed total spectrum
g g
ka 3 , fraction of the density of species a of the ambient gas in shell 3; and Ntot3 , total
density of the ambient gas in shell 3.
10.2 Calculation of Emission Spectra 193
10.2 Calculation of Emission Spectra
The key assumption in what follows is that the plasma is in LTE. The electron and
ion densities can thus be determined using a system of Saha equations:
!
Ne NazC1 2 uzC1
a T; Ea;
ion
zC1 Ea;
ion
z Ea; z
ion
D 3 exp ; (10.5)
Naz e uza .T; Ea;
ion /
z kB T
subject to the constraint of constant total number of heavy particles:

nza 1
X
Na D Naz ; z D 0; : : : ; nza 1 (10.6)
zD0
and of charge neutrality:

a 1 nX
nX za 1
Ne D zN za ; (10.7)
aD0 zD0
where nza is the number of considered ionization states of species a.
The uza .T; Ea; z / are the partition functions of the corresponding ions [10.13].
ion
e is the thermal De Brogli wavelength of the electrons and Ea; z is the reduction
ion
of the ionization energy, depending on the plasma density. The population density
of ionization stage z in the excited state n is given by:

ga; zIn Ez
Na;z n D Naz z exp a; n ; (10.8)
ua T; Ea; ion
z kB T
ga; zI n is the degeneracy and Ea;n

z
, the energy of the excited state n. For a given
particle density and temperature, the unknowns are the ion densities. The system of
equations (10.5) is solved by the Newton method. For molecules, e.g., nitrogen N2 ,
the dissociation reactions and molecular ionization processes are not considered. In
the case of nitrogen, only atomic nitrogen and its ionization stages are taken into
account. Detailed calculations of atmospheric plasmas show that for temperatures
beyond 8,000 K and pressures of 1 bar more than 90% of the nitrogen molecules are
dissociated [10.30].
The variation of the spectral radiance in the plasma is described by the radiation
transport equation:
dI./
D "./ ˛./I./; (10.9)
ds
where I./ is the spectral radiance with the unit W=.m2 sr m/, ds the incremental
propagation path, ˛./ the absorption coefficient with the unit 1/m and "./ the
emission coefficient. Relation (10.9) is a simplification of the general case of a radi-
ation bundle propagating inside a plasma, which is attenuated not only by absorption
but also by scattering of the radiation out of the direction of propagation. The total
attenuation is then governed by the extinction coefficient given by the sum of the
absorption coefficient and the scattering coefficient: k./ D ˛./ C ./. In the
following, the contribution of the scattering coefficient is neglected and the relation
(10.9) is used. The emission coefficient and the absorption coefficient comprise con-
tributions of line transitions (bound–bound), free–free transitions (Bremsstrahlung),
as well as free–bound transitions. The emission coefficient is thus given by:
"./ D "l ./ C "ff ./ C "fb ./; (10.10)
with the line emission coefficient "l ./, the emission coefficient of free–free
transitions "ff ./ and the emission coefficient of free–bound transitions "fb ./.
The index d is not used here for simplicity of notation, cf. Sect. 9.1, (9.1), (9.2)
and Sect. 9.2, (9.3). The line emission coefficient itself is a sum of line emission
coefficients of all species, ionization states, and all transitions inside the plasma:
X
"l ./ D "a; zInm ./; (10.11)
a; zI nm
where "a; zI nm is the line emission coefficient of the species a in the charge state z
for the transition n ! m. The index explicitly shown in (10.11) indicates that
this emission coefficient has the unit emitted power per unit volume per wavelength
interval.
The line emission is broadened by Doppler and Stark effect. The superposition
of these broadening contributions is described by Voigt profiles [cf. Sect. 9.4,
(9.23)]. For the quadratic Stark broadening of multielectron systems, the change
in the emission frequency is proportional to the square of the local electric field.
Stark broadening is taken into account for those lines for which experimental or
theoretical data is available [10.16, 10.19–10.26]. Stark broadening of hydrogen has
to be considered separately. Due to the degeneration of the hydrogen levels, the
existing dipole moment of the hydrogen atoms aligns in the external electric field
and the shift of emission frequencies is proportional to the local electric field. The
coefficients describing the relation between electron density and Stark width are
tabulated in [10.27].
In case of a bound–bound transition for a single line of species a in charge state
z (the indices a, z are omitted for simplicity on the right sides of the following
equations) holds:
1 c
"a; zI nm ./ D Anm Nn h nm ./; (10.12)
4
h
˛a; zI nm ./ D .Bmn Nm Bnm Nn / nm ./;
c
or with:
3
Bnm D Anm ;
8 h
gn Bnm D gm Bmn

Nm Nn 4
˛a;zInm ./ D gn Anm nm ./; (10.13)
gm gn 8 c
with h, the Planck’s constant; c, the vacuum speed of light; nm ./, the line
profile; Anm , the Einstein’s coefficient of spontaneous emission; Bnm , the Einstein’s
coefficient of stimulated emission; and Bmn , the Einstein’s coefficient of absorption.
Nn is the density of the upper level of the line and Nm , the population density of the
lower level and gn and gm , the degeneracies.
Besides line radiation, the contributions of continuous free–free and free–bound
radiation are included. The contribution of free–free transitions (Bremsstrahlung) is
in a hydrogen approximation given by [10.27]:
XX
"ff ./ D Cff z2 Na;z ; (10.14)
a z
s
3
128 2 .˛a0 / EH EH c hc
Cff D p exp Ne ; (10.15)
3 3 kB T 2 kB T
˛ is the fine structure constant; a0 , the Bohr radius; and EH is the Rydberg constant.
The bound–free (recombination) contribution to the total radiation is given by:
a 1 nX
nX za 1 nCn
X max 2
z4 z EH
"fb ./ D Cfb exp Naz ; (10.16)
aD0 zD1 nDn
n3 n2 kB T
2 EH
Cfb D Cff : (10.17)
3 kB T
The lower summation limit in relation (10.16) for the principal quantum number n
is given by: s
z2 EH
n D : (10.18)
hc
Instead of summing up over all quantum numbers n n , the summation is trun-
cated at nmax D 30.
In case of LTE, the absorption coefficients ˛ can be deduced from Kirchhoff’s
law:
"./
L;d D ; (10.19)
˛./
with the Planck function (cf. Sect. 9.2):
2hc2 1
L;d D : (10.20)
5 hc
exp 1
kB T
Assuming that the emission coefficient and the absorption coefficient are constant
within core 1 or shell 2, the solution of the radiation transport equation (10.9) is:
"./
I.; s/ D I.; s D 0/ exp.˛s/ C Œ1 exp.˛ s/ : (10.21)
˛./
In the limiting case ˛s 1, the plasma is optically thin and the radiation increases
linearly with the path length s (initial value I.; s D 0/ not considered in the
following):
I./ D "./s; (10.22)
whereas in case that ˛ s 1 the plasma is optically thick and the spectral radiance
becomes independent of the path length:
"./
I./ D ; (10.23)
˛./
which in case of thermodynamic equilibrium results in Kirchhoff’s law, cf. (10.19).
Equation (10.21) is evaluated for every volume element of the plasma including
reabsorption along the path through the plasma and the contributions of all
volume elements are added to get the total irradiance. Besides line radiation, the
contributions of continuous free–free and free–bound radiation are included. The
necessary atomic line data such as Einstein coefficients, lower and upper energy of
the transition levels, and lower and upper statistical weights for that transition are
taken from the Kurucz data base [10.14].
Figure 10.4 shows the shell model with the integration paths applied. The
radiation transport equation (10.9) is integrated from points 1 to 2 using the
emission coefficient and the absorption coefficient of shell 2. The spectral
radiance I.; s D 0/ is set to zero at point 1. From points 2 to 3, the radiation
transport equation is integrated using the emission coefficient and absorption
coefficient of the plasma core 1. The initial radiance is set to the result of the
first integration path. Finally, the integration runs from points 3 to 4 using the
values of shell 2 again and taking the result of the irradiance from the previous
integration path. If the propagation path does not intersect the plasma core, then
the integration runs from points 1 to 4. The obtained radiance is multiplied by the
cross section A D .ro2 ri2 /, see Fig. 10.4, yielding a spectral radiant intensity
[unit: W/(sr m)] considering r3 >> r2 ; r1 . Between r D 0 and r D r2 ten bundles
of equal width r D ro ri D r2 =10 are taken for the calculation.
Between point P4 and the entrance of the spectrometer P5 at r3 , the absorption
of the calculated emission spectrum of the plasma in the ambient atmosphere is
calculated. In case of argon as ambient gas, the absorption is calculated from
the available atomic line data: Einstein coefficients, wavelengths, and statistical
weights [10.14]. For the gases NO, N2 , O2 , O3 , and H2 O, the wavelength-dependent
absorption coefficient is taken from experimental data [10.29].
The calculated spectral radiant intensity at the distance r3 cannot directly be used
to simulate a measured spectrum. The receiving solid angle (e.g., the slit aperture of
Fig. 10.4 Shell model of the plasma and integration paths of the radiation transport equation, see
text. “Copyright 2009, American Institute of Physics [10.17]. Reprinted with permission”
the spectrometer at the distance r3 ), the apparative broadening of the spectrometer,

and the spectral characteristics have to be taken into account to calculate the
simulated spectral radiant flux to be compared with a measured spectrum (assuming
r3 r2 ; r1 ): Z 1
dAs
Im ./ D 2 ./ f .; 0 ; a /I3 .0 /d0 ; (10.24)
r3 0
where Im ./ is the calculated spectral radiant flux, unit: W/m; I3 ./, the calculated
spectral radiant intensity at radius r3 , unit: W/(m sr); dAs , the receiving entrance
aperture of the spectrometer, unit: m2 ; ./, the spectral characteristics of the
observation channel, unit: 1; f .; 0 ; a /, the function describing the apparative
broadening, unit: m1 ; and a is the full width at half maximum (FWHM) of the
apparative broadening.
The number of free parameters of the presented heuristic model is eight
(assuming only a single species of the ambient atmosphere and "a1;2 D 1 correspond-
ing to a stoichiometric laser ablation): Ntot1;2 , Na 1; 2g , T1; 2 , r1 , and r2 . From the
experimental point of view, there are further restrictions assuming that the following
integral parameters can be determined experimentally with a moderate effort: the
total mass of ablated particles (links r1 , r2 with Ntot1;2 ), the size of the plasma
(corresponds to r2 ), the average plasma temperature gained with a Boltzmann plot
using optically thin lines (links Ntot1;2 , Na 1; 2 g , r1 , r2 , and T1; 2 ). If furthermore a
pressure equilibrium with the ambient gas at the detection time of the spectrum
(links Ntot1;2 , Na 1; 2 g , Ne1;2 , and T1; 2 ) is assumed in total five additional equations
are available; hence, the number of free adjustable parameters can in principle be
reduced to three.
The developed software for the heuristic model is called SPES which stands for
spectra evaluation and simulation.
10.3 Simulation of Spectra
10.3.1 Fe Spectrum
Figure 10.5 shows the experimental double pulse LIBS spectrum (solid line) of
an iron sample in the wavelength range from 531 to 541 nm (cf. Fig. 8.13). This
spectrum comprises various iron lines with a broad range of excitation energies from
3.2 to 6.6 eV and the Fe line 538.34 nm which is nearly isolated from neighboring
lines and shows a strong Stark broadening with known parameters [10.26, 10.28].
For the calculation, the parameters given in the figure caption of Fig. 10.5 are
used. In total 92 Fe I and II lines were taken for the calculation based on the data of
[10.14]. The iron particle densities and the shell geometry result in a total Fe mass
of 64 ng which is close to the measured ablated mass per burst of about 50–57 ng
which can be estimated from published data [10.5]. The particle densities inside
the plasma core and the plasma shell result in a total plasma pressure of 1 bar
which is to be expected for the time domain tdelay > 500 ns after irradiation of the
laser burst, where the plasma expansion has slowed down significantly. Nitrogen
contributes significantly to this pressure balance. Furthermore, the portion of the
electron density originating from nitrogen N II in relation to the portion of Fe II
amounts to about 20% in the plasma core.
Fig. 10.5 Measured iron spectrum (solid line, measurement parameters: see text) and calculated
spectrum (dotted line) using the following values: T1 D 12;000 K, NFe 1 D 1:2 1017 cm3 ,
NN 1 g D 3:5 1017 cm3 , Ne 1 D 1:4 1017 cm3 , T2 D 9;500 K, NFe 2 D 6 1016 cm3 ,
NN 2 g D 6:5 1017 cm3 , Ne 2 D 5:5 1016 cm3 , r1 D 1:0 mm, r2 D 1:2 mm, a D 72 pm.
For the lines marked with an “S” Stark width broadening parameters are published (cf. Table 10.2,
column 3). “Copyright 2009, American Institute of Physics [10.17]. Reprinted with permission”
10.3 Simulation of Spectra 199
For LTE the condition (8.17) should be fulfilled for all energy levels of an atom
with a principal quantum number greater or equal n. For the electron densities and
temperatures resulting from the simulation of Fig. 10.5 (see figure caption) condition
(8.17) is fulfilled for iron atoms and singly ionized iron ions assuming as lowest
principal quantum number 3.
The determined spatially integrated electron densities and temperatures gained in
experiments, see Sect. 8.2, Table 8.2, are consistent with those of the simulation (see
figure caption of Fig. 10.5). The experimental values lie between the core and shell
values yielded by the simulation. They are closer to the shell values (temperature,
total iron density, iron atom density, nitrogen atom density, and electron density),
which is attributed to the fact that the experimentally determined temperature was
based on atomic iron lines only. Furthermore, the total number of iron particles of
core and shell from the simulation is 6:9 1014 which deviates less than 30% from
the value 5 1014 stated in Table 8.2. For the ratio of the total nitrogen to total iron
particle density in the shell, the simulation yields 10.8 to be compared with 8.4 in
Table 8.2.
For the line 538.34 nm, the average Stark broadening constant (referring here to
the FWHM) Cs D 0:212 1017 nm cm3 published by Freudenstein et al. was used,
which was determined for an average temperature of 9,500 K [10.28]. Of course this
is again a simplification since the quantity Cs itself depends on the temperature, see,
e.g., Fig. 5 in [10.26], where measured line widths of the Fe I 538.34 nm line are
shown for temperatures between 6,000 and 14,000 K with values of Cs spanning the
range 0:06 1017 to 0:28 1017 nm cm3 . Table 10.2 shows an overview of Fe I
multiplets and corresponding Stark broadening data – as far as available – in the
spectral range considered for Fig. 10.5. For the multiplet z5 Go –e5 H, it is obvious
that the Stark widths differ significantly within the multiplet which is in contrast to
statements saying that Stark widths should normally be nearly the same for all lines
within multiplets (see, e.g., [10.25], p. 823). For the calculation, the Stark width
broadening constants shown in column 4 of Table 10.2 are used.
For the first line of the multiplet z5 Go –e5 H, the average of the values available for
the lines 536.747 and 536.996 nm are taken for the calculation. For line 538.337 nm,
the value of [10.28] is used, since this data is mostly deployed in LIBS studies and
differs only by 5% from the average value calculated on the basis of the data given in
[10.26]. For multiplet z5 Do –e5 D where a Stark shift value is published for the third
line, this value is used for the first two lines as well. For the line 540.415 nm, the
average Stark width broadening constant is taken in column 3 of Table 10.2 using
the values published in [10.24] for the multiplet 540.4 nm. Since there is only a
negligible difference to the value for the 540.411 nm line, the same Stark broadening
constant of 0:126 1017 nm cm3 is used for the calculation.
From this discussion, we have to notice that the available Stark data are
fragmentary and have significant uncertainties. Hence, any comparison between
calculated and experimentally observed spectra has to consider this starting position.
Also a single shell modeling was tested, but in this case it was not possible to
achieve relative line intensities between the iron ion lines and atomic lines that are
similar to the observed ones.
Table 10.2 List of Fe I lines in the spectral range shown in Fig. 10.5 belonging to multiplets where
Stark broadening data is partially available
Multiplet Fe I Stark width Stark width STARK Stark Reference;
line broadening broadening shift shift remark
(nm) constant constant (nm) used
(1017 used for for
nm cm3 ) calculation calcu-
(1017 lation
nm cm3 ) (nm)
z5 Go –e5 H 536.486 – 0.132 – – [10.14], n.a.
536.747 0.147 0.147 – – [10.26]
536.996 0.116 0.116 – – [10.26]
538.337 0.203 0.212 – – [10.26], ref.
[10.28] states
0.212
1017 nm cm3
z5 Do –e5 D 532.418 – 0.090 – 0.050 [10.14], n.a.
533.993 – 0.090 – 0.050 [10.14], n.a.
539.317 0.090 0.090 0.050 0.050 [10.26]
z5 Go –f5 G 540.050 – 0.126 – – [10.26]
540.411 0.126 0.126 – – [10.26], this line
overlaps with
540.415 nm
belonging to
another
multiplet and
having nearly
the same upper
energy level
and oscillator
strength
z3 Go –e3 H 540.415 0.127 0.126 – – [10.24]
The Stark width broadening constant shown in the third column is the average of the ratios of the
Stark width and the electron density measured for different temperatures for a given line taken
from reference [10.26]. The Stark shift refers to an electron density of 1016 cm3 . Only those
data of [10.26] were taken, where an uncertainty classification of grades A, BC or B is stated
corresponding to uncertainties being within 15%, 23% and 30%. n:a:, no Stark data available
Figure 10.5 shows that all experimentally observed lines can be reproduced in
the calculation. Comparing the measured and calculated line shape of the Fe I
538.34 nm line shows a high degree of overlap. Also a small satellite on the left
wing of this line shows up in the calculation being approximately in the right relation
to the Fe I 538.34 nm line. The continuum base level of the calculated spectrum is
slightly lower than the measured one. The origin of this discrepancy is not yet clear.
A potential reason could be light scattering in the spectrometer which could raise
the continuous background and is not considered in the calculation so far.
The peak positions of the calculated lines of the multiplet z5 Do –e5 D coincide
approximately with those of the experimental spectrum showing that the assumed
10.3 Simulation of Spectra 201
Stark shift is appropriate. If the Stark shift is not taken into account offsets between
the calculated and measured peak positions were clearly visible.
The small double peak close to 533.99 nm is reproduced also in the calculation.
The relative height of the peaks depends on the chosen radius r2 , a slight change
of this parameter by 10% leads to a visible change in the relation of the peak
heights. Discrepancies in the peak height and width of the line Fe I 532.42 nm of
the multiplet z5 Do –e5 D occur.
While for multiplet z5 Go –e5 H the measured and calculated line shapes of the
538.34 nm line coincide well, discrepancies occur for the other three lines of this
multiplet. Again this may partly be attributed to different uncertainties given for the
respective Stark data, where for the 538.34 nm line an uncertainty of 15% is stated,
compared with an uncertainty of 30% for the other three lines.
A significant difference between calculated and measured spectrum is observed
for the line 540.411 nm of the z5 Go –f 5 G multiplet. This line overlaps with
540.415 nm which is a Fe I line from another multiplet, for which however no
Stark data is available.
Simulations of the shown Fe spectrum were undertaken to reduce these dis-
crepancies by variation of the Stark parameters and Einstein coefficients for the
lines 532.4178, 532.8038, 540.4112, 540.4149, and 540.5774 nm. The discrepancies
could be reduced significantly; however, this requires variations of Stark parameters
and Einstein coefficients being partly greater than 50%. Hence, it is concluded that
the found discrepancies may be attributed partly to the uncertainty of the available
atomic parameters and partly to the strong simplification of the spherical stationary
plasma model chosen.
While the model is able to simulate even detailed structures of the studied Fe
spectrum, discrepancies between simulation and experiment remain. Comprehen-
sive and accurate Stark data are a necessary precondition to further improve the
simulation capability for such complex multiline spectra.
10.3.2 Al Resonance Line with Self-reversal
Figure 10.6, top, shows a LIBS spectrum of an Al I resonance line with line reversal
taken with a laser burst irradiating an aluminum sample and detected with a
high-resolution echelle spectrometer with the following experimental parameters:
collinear double pulses 2 100 mJ, t D 31 s, tdelay D 2 s, tint D 5 s,
air atmosphere at 1,013 mbar, and s D 25 mm. (This spectrum is identical to
the one shown in Fig. 9.11.). Fitting the wings of this line with a Lorentz function
allows to estimate the shift between the absorption and emission maximum being
of the order of 26 pm. This Al line has a Stark red shift. By use of the relation
describing the Stark shift as a function of the electron density given in [10.18] [cf.
(9.20)], neglecting the ion correction term, the electron density difference between
the plasma core and the plasma shell can be estimated.
Table 10.3 shows the available Stark data for this line. With the Stark shift
parameter for the Al I 394.401 nm line 0:0242 1017 nm cm3 , we obtain for the
Fig. 10.6 Al I line at 394.4 nm with line reversal. Top: Experimental spectrum (solid line) and
Lorentz fit curve to the wings (dash-dot line). Bottom: Experimental spectrum and calculated line
profile (dotted line) using the following values: T1 D 10;500 K, NAl 1 D 1 1017 cm3 , NN 1 g D
5:90 1017 cm3 , Ne 1 D 1:4 1017 cm3 , T2 D 6; 970 K, NAl 2 D 4:2 1016 cm3 , NN 2 g D
9:981017 cm3 , Ne 2 D 2:21016 cm3 , r1 D 1:0 mm, r2 D 1:2 mm, a D 10 pm. “Copyright
2009, American Institute of Physics [10.17]. Reprinted with permission”
electron density difference between core and shell corresponding to a shift of 26 pm:
Ne D 1:07 1017 cm3 .
Figure 10.6, bottom, shows a comparison between experimental and calculated
line profile showing a good agreement. The simulation yields the electron densities
in the core and shell: Ne1 D 1:38 1017 cm3 and Ne2 D 2:22 1016 cm3 . This
corresponds to a difference of Ne1 –Ne2 D 1:16 1017 cm3 , which is within 10%
equal to the result of the estimation given above.
In Sect. 3.5, Fig. 3.15 shows simulated emission spectra in the vacuum ultraviolet
spectral range (VUV) calculated with the presented heuristic model demonstrating
the strong effect of a small oxygen content on the absorption of emission lines in
the VUV. For further details, see Sect. 3.5.
10.4 Enthalpy of Plasma 203
Table 10.3 Stark data of the Al I line 394.401 nm published for different electron densities and
temperatures
Multiplet, T(K) ne (1017 cm3 ) Stark width Stark shift Reference;
Al I line broadening constant accuracy
394.401 nm constant (1017 nm cm3 )
(1017 nm cm3 )
2 o 2
P – S 13,200 1.28 0:0422a 0.0242 [10.19]; 30–50%
9,670 1.42 0.0327 0.0099 [10.19]; 50%
11,000 0.1 0:0370a [10.19]; 30%
13,600 4.5 0.0311 0.0207 [10.19]; >50%
11,700 2.5 0.0336 0.0168 [10.20]; 50%
Only those values marked with “a ” having accuracies between 30% and 50% are taken to calculate
an average Stark width broadening constant to be used for the simulation: 0:0396 1017 nm cm3 .
For the shift the value 0:0242 1017 nm cm3 is taken for the simulation
For significant improvements of the model with the aim to simulate precisely
experimental LIBS spectra much more accurate and precise Stark data are needed.
10.4 Enthalpy of Plasma
The enthalpy of the plasma is calculated using the data gained from the simulation
of the iron spectrum. The total enthalpy is the sum of the enthalpies of the plasma
core (index 1) and the plasma shell (index 2):
H D H1 C H2 :
Considering only the first ionization stage of the particles, the enthalpy of the plasma
core is (and in an analogous way for the shell) given by:

5 2 @
H1 D NFe0 V1 kT1 C kT1 ln uFe0
2 @T1

5 @
C NFe1 V1 kT1 C EFe0
ion
C kT12 ln uFe1 C
2 @T1
!
5 END2 2 @
C NN0 V1 kT1 C C kT1 ln uN0
2 2 @T1
!
5 END2 2 @ 5
C NN1 V1 kT1 C C EN0 C kT1
ion
ln uN1 C Ne1 V1 kT1 ;
2 2 @T1 2
(10.25)
where NFe0 ; NFe1 ; NN0 ; NN1 are the densities for iron atoms, iron ions with z D 1,
nitrogen atoms and nitrogen ions with z D 1, V1 is the volume of the plasma core,
EFe0
ion
; EN0
ion
the ionization energies for iron and nitrogen atoms,END2 the dissociation
energy of nitrogen molecules and Ne1 the electron density in the core. In relation
(10.25), the enthalpy of molecular nitrogen is neglected. For a pressure of 1 bar and a
temperature of 9,500 K as yielded in the simulation of the iron spectrum for the shell
(cf. Fig. 10.5), calculations show that more than 99% of all nitrogen molecules are
dissociated [10.30]; hence, the energy content of molecular nitrogen in the plasma
can be neglected. The enthalpy of the plasma shell is calculated in an analogous way.
Taking the values of the simulation of the iron spectrum from Fig. 10.5 and
evaluating the relation (10.25) by use of parabolic fit functions for the partition
functions yields: H1 D 3:3 mJ and H2 D 2:7 mJ. Referring the total enthalpy to
the process energy irradiated initially, i.e., the burst energy of 2 40 mJ used in
the experiment to generate the iron spectrum, yields a percentage of 7.5%. Hence,
0:8 s after the last laser pulse, the plasma enthalpy still comprises 7.5% of the
total energy irradiated. In Sect. 19.1, the necessary energy to displace the ambient
atmosphere is estimated.
References

Spectrom. 19, 419–428 (2004)
10.2. Z. Hwang, Y. Teng, K. Li, J. Sneddon, Interaction of a laser beam with metals Part I:
quantitative studies of plasma emission. Appl. Spectrosc. 45, 435–441 (1991)
10.3. S. Harilal, C. Bindhu, C. Riju, V. Nampoor, C. Vallabhan, Electron density and temperature
measurements in a laser produced carbon plasma. J. Appl. Phys. 82, 2140–2146 (1997)
10.4. Y. Iida, Effects of atmosphere on laser vaporization and excitation processes of solid
samples. Spectrochim. Acta 45B, 1353–1367 (1990)
167 (2007)
10.6. J. Aguilera, C. Aragon, F. Penalba, Plasma shielding effect in laser ablation of metallic
samples and its influence on LIBS analysis. Appl. Surface Sci. 127–129, 309–314 (1998)
10.7. M. Aden, E. Beyer, G. Herziger, H. Kunze, Laser-induced vaporization of a metal surface.
J. Phys. D: Appl. Phys. 25, 57–65 (1992)
10.8. A. Bogaerts, Z. Chen, D. Autrique, Double pulse laser ablation and laser-induced break-
down spectroscopy: a modeling investigation. Spectrochim. Acta B 63, 746–754 (2008)
10.9. I. Gornushkin, C. Stevenson, B. Smith, N. Omenetto, J. Winefordner, Modeling an
inhomogeneous optically thick laser induced plasma: a simplified theoretical approach.
10.10. I. Gornushkin, A. Kazakov, N. Omenetto, B. Smith, J. Winefordner, Radiation dynamics
of post-breakdown laser induced plasma. Spectrochim. Acta B 59, 401–418 (2004)
10.11. I. Gornushkin, A. Kazakov, N. Omenetto, B. Smith, J. Winefordner, Experimental verifi-
cation of a radiative model of laser-induced plasma expanding into vacuum. Spectrochim.
Acta B 60, 215–230 (2005)
10.12. A. Kazakov, I. Gornushkin, N. Omenetto, B. Smith, J. Winefordner, Radiative model of
post-breakdown laser-induced plasma expanding into ambient gas. Appl. Optics 45, 2810–
2820 (2006)
References 205
10.14. R. Kurucz, B. Bell, Atomic Line Data, Kurucz CD-ROM No. 23 (Smithsonian Astrophysical
Observatory, Cambridge, 1995). www.cfa.harvard.edu
10.15. J. Hermann, C. Boulmer-Leborgne, D. Hong, Diagnostics of the early phase of an
ultraviolet laser induced plasma by spectral line analysis considering self-absorption.
J. Appl. Phys. 83, 691–696 (1998)
10.16. C. Colón, G. Hatem, E. Verdugo, P. Ruiz, J. Campos, Measurement of the Stark broadening
and shift parameters for several ultraviolet lines of singly ionized aluminum. J. Appl. Phys.
73, 4752–4758 (1993)
10.17. R. Wester, R. Noll, Heuristic modeling of spectral plasma emission for laser-induced
breakdown spectroscopy. J. Appl. Phys. 106, 123302-1–123302-10(2009)
10.18. G. Bekefi, Principles of Laser Plasmas (Wiley, New York, 1976)
10.19. N. Konjevic, M. Dimitrijevic, W. Wiese, Experimental Stark widths and shifts for spectral
lines of neutral atoms. J. Phys. Chem. Ref. Data 13, 619–647 (1984)
and ionized atoms. J. Phys. Chem. Ref. Data 19, 1307–1385 (1990)
10.21. H. Griem, Spectral Line Broadening by Plasmas (Academic, New York, 1974), Appendix
IV, p. 320
10.22. J. Aguilera, J. Bengoechea, C. Aragon, Curves of growth of spectral lines emitted by a
laser-induced plasma: influence of the temporal evolution and spatial inhomogeneity of the
plasma. Spectrochim. Acta B 58, 221–237 (2003)
10.23. A. Bartecka, T. Wujec, J. Halenka, J. Musielok, Experimental Stark-broadening studies of
the N I multiplet (1D) 3s2D – (1D) 3p2P0 at 7905.5 Å. Eur. Phys. J. D 29, 265–271 (2004)
10.24. A. Lesage, J. Lebrun, J. Richou, Temperature dependence of Stark parameters for Fe I
lines. Astrophys. J. 360, 737–740 (1990)
10.25. V. Bakshi, R. Kearney, Measurements of Stark widths of some Ar I transitions in a
Dc Argon plasma jet at atmospheric pressure. J. Quant. Spectrosc. Radiat. Transfer 42,
405–413 (1989)
10.26. N. Konjevic, A. Lesage, J. Fuhr, W. Wiese, Experimental Stark widths and shifts for
spectral lines of neutral and ionized atoms. J. Phys. Chem. Ref. Data 31, 819–927 (2002)
10.27. H. Griem, Plasma Spectroscopy (McGraw-Hill, New York, 1964)
10.28. S. Freudenstein, J. Cooper, Stark broadening of Fe I 5383 Å. Astron. Astrophys. 71,
283–288 (1979)
10.29. J. Pecker, Space Observatories (D. Reidel Publishing, Dordrecht, 1970)
10.30. J. Bacri, S. Raffanel, Calculation of some thermodynamic properties of air plasmas: internal
partition functions, plasma composition, and thermodynamic functions. Plasma Chem.
Plasma Process. 7, 53–87 (1987)
Chapter 11
Quantitative Analysis
Quantitative analysis with LIBS is defined here as the determination of concentra-

tions of chemical elements in a specimen. A more generic definition includes as well
the application case of material identification where a decision is taken in terms
of an allocation of a specimen to a material class by evaluating whether a set of
determined concentrations falls into or is close to a set of predefined concentration
intervals or is smaller or greater than predefined concentration thresholds or not.
This chapter is focused on quantitative analysis based on calibration. Calibration-
free LIBS approaches were studied, but the achieved analytical performance
in terms of accuracy levels and correctness of the determined concentrations -
especially in the case of trace detection - are still significantly worse than those
obtained with a quantitative LIBS analysis based on calibration [11.1].
The principle steps for a quantitative analysis with LIBS are shown in Fig. 11.1,
which are described in more detail in the following sections.
11.1 Measuring Method
First of all, the measuring method has to be defined comprising the setup, process
parameters, and evaluation procedure. The setup in turn is described by the
instrumental components applied, cf. Chap. 4. Their respective settings belong to
the process parameters of the measuring method, cf. Chap. 3. Table 11.1 shows a
selection of the most important process parameters to be defined and documented.
Furthermore, the sample taking, sample preparation, and sample surface condi-
tion have to be defined.
The measured quantity is a line intensity of an analyte line i described by:
ZZ
Ii D I .; t/ddt; (11.1)
tdelay ; tint ; i
where I .; t/ is the spectral radiant flux detected by the spectrometer.

208 11 Quantitative Analysis
Fig. 11.1 Block diagram of the various steps for a quantitative LIBS analysis. The blocks marked
with the letter “C” refer to the calibration procedure and those with “QA” to the quantitative
analysis
Generally, the quantity I .; t/ has to be corrected first to take into account,
e.g., the spectral characteristics of the detector, continuum background radiation,
and line interferences. The integral (11.1) runs over the analyte line i at the
wavelength i and the time interval [tdelay , tdelay C tint ]. The line emission depends
on the line emission coefficient [cf. Sect. 9.2, relation (9.3)], which is in turn a
function of the density of species and the temperature. The line emission coefficient
is a local quantity and the radiation transport equation has to be solved [cf.
Sect. 10.2, (10.9)] to determine the spectral radiant flux. The density of species
in the laser-induced plasma may vary from laser pulse to laser pulse due to
changing measuring conditions, e.g., the ambient atmosphere, the microstructure
of the sample surface (roughness, crater formation), varying laser pulse energies,
etc. Hence, for quantitative LIBS measurements, in general, line ratios are used for
calibration, e.g.:
Iia
Qi;r D ; (11.2)
Irb
11.1 Measuring Method 209
Table 11.1 Selection of process parameters to be defined for quantitative LIBS analysis
Process Symbol Definition Remarks
parameter
Laser Unit: m
wavelength
Repetition rep The number of laser pulses or cf. Fig. 3.1 and Table 4.1. The
frequency bursts per time interval gas exchange time has to be
chosen according to relation
(3.1). Unit: Hz
Beam M2 cf. Sect. 3.4 The beam propagation ratio is a
propagation measure of the beam quality.
ratio The minimum value is 1.
ISO 111146–1 describes
measurement arrangement
and test equipment to
determine M 2
Focusing optics f Focal length
Aperture of D Diameter of the illuminated The quantities f , D, M 2 , and
focusing aperture of the focusing lens L determine the irradiance
optics at the beam waist, see
Sect. 3.4
Burst mode SP, DP SP D single pulse, DP D For double pulses, the
double pulse, cf. Chap. 6 configuration has to be
given: collinear, orthogonal
preablation, orthogonal
re-heating, dual pulse
crossed beam, cf. Sect. 6.1
Burst energy Eb Total energy of a burst, i.e., the Average and standard deviation
sum of the pulse energies of the burst energy should be
within P
the burst: measured. Unit: J
Eb D Ej , see Fig. 3.9
j
Pulse energy Ej Energy of the j -th pulse within Unit: J
the burst, see Fig. 3.9
Pulse duration L ; Lj Pulse duration of a laser pulse Unit: s
given as FWHM. If bursts
are applied Lj denotes the
pulse duration of the j -th
pulse within the burst, see
Fig. 3.9
Interpulse t Time between the two pulses of Unit: s
separation a double pulse (from
maximum to maximum). If
bursts with more than two
pulses are used, an index is
introduced so that tj
denotes the interpulse
separation between the j
and j C 1 pulse, see Fig. 3.9
(continued)
Table 11.1 (continued)

parameter
Warming-up Nwu Number of laser pulses The warming-up pulses assure a
pulses generated to stabilize the sufficient degree of thermal
thermal state of the laser. stability of the laser
The laser pulses are blocked operation with respect to
outside the laser by a shutter beam profile, pulse energy,
or beam dump and do not pointing stability
irradiate the sample
Prepulses Npp Number of laser pulses The spectral emission of the
irradiating the sample plasmas generated with the
leading to a local ablation to prepulses can be utilized to
remove surface check whether a
contaminations or surface nonrepresentative surface
layers which are not layer is penetrated, cf.
representative for the bulk Sect. 13.1.2. If multiple
composition of the sample. pulses are deployed the
The spectral emission of the analogous term is prebursts.
plasmas generated with the
prepulses is not used for the
analysis
Measuring Nmp Number of laser pulses
pulses irradiating the sample to
generate plasmas, whose
spectral emission is used to
determine the composition
of the sample
Beam waist s Position of the beam waist of In the description of the
position the focused laser beam experimental setup, the
relative to the surface of the procedure to determine the
sample. s > 0 corresponds beam waist position shall be
to a beam waist position described. Unit: m
inside the sample, see
Fig. 2.2
Delay time tdelay Time delay between the laser Several delay times for a set of
pulse and the start of the integration windows for one
integration of the burst are used, e.g., for the
spectroscopic signal. This experiments described in
delay time denotes the true Sects. 13.1.3 and 16.2.
temporal delay considering Unit: s
pulse propagation times and
electronic signal transit
times. If bursts with two
pulses and more pulses are
used, an index is introduced
so that tdelay j denotes the
delay time after the j -th
pulse, see Fig. 2.4
(continued)
11.1 Measuring Method 211
Table 11.1 (continued)

parameter
Integration tint Width of time gate for the Unit: s
time gate integration of the spectral
signal. The time of
integration starts at the end
of tdelay . If bursts with two
and more pulses are used, an
index is introduced so that
tint j denotes the integration
time after the j -th pulse, see
Fig. 2.4
Gas pressure pg The gas type and the ambient Unit: bar
gas pressure or the gas
pressure inside the
measurement chamber
should be given
Flow rate VPmc Gas flow rate in the measuring A gas flow in the interaction
chamber region assures that ablated
or recondensated
particulates are removed
before the subsequent pulse
irradiates the sample, cf.
Sect. 3.1. Unit: m3 =s
where Iia is the line intensity of analyte line i of species a and Irb is the line intensity
of reference line r of species b.
This ratio is called also internal standardization. The reference line r is usually
an emission line of a dominant matrix element, e.g., iron for a steel sample.
In the case of an optically thin and homogeneous plasma in a nonabsorbing
ambient atmosphere, the measured line intensity is proportional to the line emission
coefficient [cf. (9.3)]. Hence for the ratio Qi;r holds:
Naz .T / uzb .T / Ei Er

Qi;r / e kT ; (11.3)
Nbz .T / uza .T /
where Naz .T /; Nbz .T / is the density of species a; b in charge state z, uza .T /; uzb .T /
are the partition functions of species a; b with charge z, and Ei ; Er is the energy
of the upper level of line i; r.
Relation (11.3) shows that the ratio Qi;r depends on the temperature T of the
plasma. In case of local temperature equilibrium, the temperature dependence of the
densities is determined by a system of Saha equations, see Chap. 10, relation (10.5).
The temperature influence on the last term of (11.3) can be reduced by choosing
emission lines with a small difference in their upper energy levels. Such homologous
line pairs are less sensitive to temperature variations of the laser-induced plasma.
In general, a series of laser pulses is irradiated onto one location of the specimen
and averages are taken over these Nmp measuring pulses for a single location k on
the sample after Nwu warming-up pulses and Npp pre-pulses:
D E
i;r
Qm;j .k/ ; (11.4)
Nwu ;Npp ;Nmp
where the index m denotes the m-th laser burst, and index j the pulse j within
that laser burst. The average is taken from m D 1 to m D Nmp . If K locations are
measured, then the following average is taken:
D E
i;r
Qm;j .k/ : (11.5)
Nwu ;Npp ;Nmp K
The average ratios (11.4) or (11.5) are used as measuring quantities for the
calibration.
11.2 Calibration
Calibration is a standard topic for chemical analysis and well-established procedures

exist which are described in the literature as well as in national and international
standards [11.1, 11.2, 11.3]. These procedures can be deployed for LIBS as well.
Hence, in the following paragraphs, the description is restricted to a short overview
of the basic concepts and characteristic numbers used to characterize the perfor-
mance of a calibration.
A set of reference samples with known composition is measured applying the
defined LIBS measuring method (cf. Sect. 11.1). For a reference sample i having
a known analyte concentration xi the measurand – e.g., (11.4) or (11.5) – is
determined, which is for simplicity denoted as yi in the following [11.3]. Based
on the obtained set of pairs of variates .xi ; yi /, a calibration curve is determined
by regression analysis, examples for calibration functions describing the functional
relationship between the response variable and the analyte amount are:
yO D a0 C a1 x; (11.6)
2
yO D a0 C a1 x C a2 x ; (11.7)
q
1 2
yO D 4a2 .x a0 / C a1 a1 ; (11.8)
2a2
where the coefficients a0 –a2 are determined by the regression analysis. The type
of calibration function to be chosen can be decided by statistical means, e.g., the
Mandel test [11.4].
11.2 Calibration 213
For a quantitative assessment of a calibration function, the working range has to

be defined firstly, i.e., the lowest and highest concentration of the analyte considered.
This concentration range has to be represented by a set of reference samples. If
necessary, this concentration range can be further reduced by means of statistical
methods or visual inspection of the calibration function to further improve the
features of the calibration in sections.
Common assessment criteria for a calibration function are: (a) sensitivity, (b)
standard deviation of procedure, (c) limit of detection, identification limit, and
quantification limit.
The sensitivity is defined as the slope of the calibration function in the middle of
the working range [11.5]:
ˇ
@f ˇˇ
yO 0 D f 0 .x/
N D ; (11.9)
@x ˇxDxN
where N
1 X
xN D xk (11.10)
N
kD1
is the average concentration of the N reference samples used to represent the

working range. The average sensitivity yO 0 can be easily calculated for the functions
(11.6)–(11.8).
The standard deviation of procedure is defined as follows:
sy
sx0 D : (11.11)
yO 0
The quantity sy is the residual standard deviation, in case of a linear calibration

function, cf. (11.6) sy is given by:
v
u N
uP
u O k /2
Œy y.x
t kD1 k
sy D : (11.12)
N 2
For the calibration functions (11.7) and (11.8), the denominator in the radicand has
to be set to .N 3/. The relative standard deviation of procedure is:
sx0
Vx0 D 100%: (11.13)
xN
Figure 11.2 shows an example of a LIBS calibration to determine Si in an aluminum
matrix where the calibration function is a root function. The coefficients of the
calibration function, the standard deviation of procedure, and the calibration range
are given in the diagram.
2.0
I Si 288.16 nm / I Al 305.71 nm
1.5
1.0
regression model: root function
a0 = 0.0076, a1 = 0.764, a2 = 2.523
0.5 sx0 = 0.104 m.- %, Vx0 = 6.93 %, N = 37,
cmin = 0.003 m.- %, cmax = 11.8 m.- %
0.0
0 2 4 6 8 10 12
c(Si) [m.- %]
Fig. 11.2 Example of a calibration function for silicon in aluminum, for details see text
For the detection of traces, the limit of detection is a characteristic parameter

often used in quantitative LIBS studies. The criterion for the detection of an analyte
is that the measurand is greater than a critical value given by the sum of the average
blank value of the measurand and the uncertainty of the blank value. The decision
“detected” or “not detected” is taken on the y-axis only. The uncertainty of the blank
measurement can be determined in two ways: (a) direct method, and (b) indirect
method via the calibration straight line.
For method (a), the uncertainty is determined by a multiple measurement (n10)
of the blank sample. The measured values have to be checked carefully to assess
whether they fulfill the requirement of a normal distribution. If no blank sample
is available, then method (b) has to be applied. In this case, the uncertainty of the
blank value is determined in an indirect way by extrapolation of regression data of a
calibration. This calibration is carried out in the proximity of the limit of detection,
i.e., in the analyte concentration range from zero to about ten times the limit of
detection.
The following simplifying assumptions are taken: there is no significant differ-
ence between the range of blank values and the range of measuring values of the
sample containing the analyte; the measuring values of the blank sample follow
approximately a normal distribution. The critical value is given by:
r
1
yc D yNb C sb tf;˛ 1C ; (11.14)
n
where yNb is the arithmetic average of n measurements of the blank sample, sb is the
standard deviation of the measuring values of the blank sample, tf;˛ is the t-value
of a one-sided test, f is the degrees of freedom, and ˛ is the level of significance
corresponding to the probability of a type I error (the null hypothesis “the sample
11.2 Calibration 215
is a blank sample” is rejected when the null hypothesis is true, so called ˛-error or
false positive).
The limit of detection (LOD) for a linear calibration function [cf. (11.6)] is then:
r
yc yNb sb 1
xLOD D D tf;˛ 1C : (11.15)
a1 a1 n
For a level of significance of ˛ D 0:01 and n D 10 holds:
sb 3sb
xLOD D 2:959 : (11.16)
a1 a1
The limit of detection determined in this way will be denoted with LOD3s . The
limit of detection should not be mixed-up with the smallest concentration that can
be determined reliably. This limit of identification (sometimes also called detection
limit) is the smallest concentration where the measurand y fulfills y yc with a
probability .1 ˇ/, where ˇ is the risk of error to interpret the signal as a blank
value (type II error, false negative). For the limit of identification holds:

tf;ˇ
xID D 1 C xLOD : (11.17)
tf;˛
For ˛ D ˇ relation (11.17) yields: xID D 2xLOD . The limit of quantification (LOQ)
describes an analyte concentration that can be determined with a given uncertainty.
The LOQ can be estimated as follows:
sb
xLOQ k nI˛ ; (11.18)
a1
where 1=k is the relative measuring uncertainty, k D 3 corresponds to 33.33%

uncertainty, nI˛ tabulated value, see, e.g., [11.3], for n D 10 and ˛ D 0:01:
10I0:01 D 3:0.
For the given example, we have: xLOQ D 3xLOD .
Figure 11.3a illustrates graphically the determination of the limit of detection
xLOD . A further quantity used to describe calibration curves is the background equiv-
alent concentration (BEC) shown in Fig. 11.3b. For a linear calibration function [cf.
(11.6)], the BEC is simply defined by:
a0
BEC D : (11.19)
a1
With (11.16) and (11.19), the limit of detection can be expressed as follows:
xLOD D 3 BEC RSDb ; (11.20)
where RSDb D sb =a0 is relative standard deviation of the measurand for the blank
sample.
Fig. 11.3 Graphical illustration of (a) the detection limit xLOD and (b) the background equivalent
concentration BEC
11.3 Measurement and Determination of Analytes
The quantitative measurement is conducted according to the defined measuring

method and the selected process parameters, cf. Sect. 11.1.
For a quantitative measurement, the inverse functions of the calibration functions
[see (11.6)–(11.8)] are determined. The detected radiant flux I .; t/ has to be
corrected first to take into account effects such as continuum background radiation
and line interferences. Then the line intensity is calculated by use of (11.1). With
internal standardization and averaging procedures [cf. (11.2)–(11.5)], the input
quantities for the analysis function are obtained. Finally, the analyte concentrations
are determined.
The quality of an analysis can be assessed by the mean residual defined as
follows: v
u N
uP 2
u
t sD1 .ci;reference;s ci;measured;s /
R.i / D ; (11.21)
N
where i is the element, s is the sample, N is the total number of measured
samples, ci;measured;s and ci;reference;s are the concentrations of element i and sample s
determined with LIBS and the reference concentrations. The precision is defined as
the mean relative standard deviations of the intensity ratios of, e.g., three replicate
measurements:
PN ˝ ˛
RSD Qi;r s
˝ ˛ sD1
RSD. Qi;r / D : (11.22)
N
11.4 Recalibration
The term recalibration is used here for a correction procedure of potential drifts of
a LIBS instrumentation used for quantitative measurements. Such an instrumental
drift can, e.g., be caused by a progressing contamination of optical windows in
11.4 Recalibration 217
the laser or measuring radiation beam path or temperature influences on instru-

mental components. The concept of recalibration is illustrated with the help of
Fig. 11.4.
The aim of recalibration is to reduce the effort of a complete calibration
procedure after a certain time of operation of a LIBS equipment which may be
necessary due to instrumental drifts. Two samples are selected from the set of
calibration samples representing analyte concentrations at the lower end of the
working range and at the upper end to consider the sensitivity of the instrument
for the recalibration. The intensity ratios measured during calibration for these
two samples are documented and compared with those obtained with these two
samples at a later time. For an effective conception of a recalibration procedure,
the selection of recalibration samples should be taken in such a way to minimize
the total number of samples for the correction of all analyte intensity ratios.
Comparing the intensity ratios for an analyte line between calibration and a later
recalibration action – cf. Fig. 11.4 right – yields a linear regression by use of which
the new intensity ratios can be corrected to the intensity ratio scale of the original
calibration. These corrected ratios are then the input quantities for the analysis
functions.
calibration recalibration
5 5
intensity ratio of calibration Ia / Ir
calibration
4 recalibration 4
intensity ratio Ia / Ir
3 3
2 2
1 1
0 0
0 1 2 3 4 5
concentration c a [m.-%] new intensity ratio Ia / Ir
Fig. 11.4 Left: Line intensity ratio of an analyte line to a reference line as a function of the
concentration. The straight line indicates the calibration function. Right: Line intensity ratio of
the calibration as a function of the new line intensity ratio. The arrows indicate the shift of the
ratios in relation to the original values obtained at the time of the calibration due to a drift of the
instrument. The dashed line describes this relation if there is no drift; the solid line is used to
correct the intensity ratios
0.6
concentration ratio c Cr / c Fe [1]

0.5 double pulse
2 × 50 mJ
0.4 2 × 80 mJ
0.3
0.2 single pulse

1 × 100 mJ
0.1 1 × 160 mJ
0.0
0.5 1.0 1.5 2.0 2.5
intensity ratio I Cr 286.26 nm / I Fe 271.44 nm [1]
Fig. 11.5 Analysis functions for chromium for single (squares) and double pulses (triangle)
11.5 Interelement Correction
By use of iterative interelement corrections, the mean residual deviations [cf.

relation (11.21)] can be reduced [11.6]. Figure 11.5 shows as an example four
analysis functions for the element Cr in high-alloy steel. Two burst energies each for
single and double pulses are used. The calibrations with double pulses show greater
saturation effects than the corresponding single pulse calibrations. For concentration
ratios cCr =cFe > 0:3, the deviations of the data points from the calibration curve
increase. Two of the samples with a high Cr concentration have for example a
relatively high Ni concentration. This is an indication that high concentrations of
other elements have an influence on the evaporation and excitation processes in the
laser-induced plasma. Interelement correction is a method to take into account such
effects with the aim to improve the residuals.
Interelement correction is described here as an iterative method. The residuals
(11.21) R.i / can be corrected additively, see (11.23), and multiplicatively, see
(11.24), considering spectral interferences and matrix effects. The additive factors
correct the superposition of line intensities. The multiplicative factors take into
account different excitation conditions in the plasma consisting of species ablated
from the high-alloy steel samples. The following equations describe the analysis
functions for concentrations ratios:
X
cOi D b0;i C b1;i Qi C b2;i Qi2 C kadd;j cOcorr;j ; (11.23)
j;j ¤i
Y
cOi D b0;i C b1;i Qi C b2;i Qi2 1 C kmult;j cOcorr;j ; (11.24)
j;j ¤i
11.5 Interelement Correction 219
3
(c Cr, LIBS – c Cr, reference) [m.-%]
-1
-2 Cr 267.72 nm
Cr 267.72 nm, with interelement correction
-3
0 5 10 15 20 25 30
concentration c Cr, reference [m.- %]
Fig. 11.6 Differences of Cr concentrations measured with LIBS to the reference concentrations
as a function of the Cr concentrations with and without interelement correction. Single pulses,
Eb D 160 mJ
where cOcorr;j is the concentration ratio of element j to correct the residuals additively
or multiplicatively, Qi is the intensity ratio of Ii to Ir [cf. (11.2)], kadd;j is an
additive correction factor for the element j , and kmult;j is a multiplicative correction
factor for the element j .
For example, the correction for Cr can be started multiplicatively with nickel.
Thereby a regression of the Cr residuals of all calibrated samples is made with
the corresponding Ni concentration ratios. The aim is to get a correction factor
to calculate the residuals. Then the reduced residuals are taken for making a
regression with the measured concentrations of another element considered as a
disturbing factor. These corrections are continued with other elements until no
further improvement is achieved. Then the absolute Cr concentrations can be
calculated with the elemental concentration of the analysis functions of those
elements which are contained in the samples.
Figure 11.6 shows the differences of the measured Cr concentration to the
reference concentrations versus the reference concentration, where the filled squares
represent uncorrected and open squares the corrected values.
The Cr 267.72 nm residual is reduced by a factor of 3 taking into account
multiplicative corrections with the elements Ni and Ti for single pulses and 160 mJ.
The coefficient of determination can be improved from r 2 D 0:992 to r 2 D 0:999.
The effect of interelement correction for the quantitative analysis of other analytes
of high-alloy steel will be presented in more detail in Sect. 13.2.
References
11.1. IUPAC Compendium of Chemical Terminology, 2nd edn. (1997). http://www.chemsoc.org/

chbytes/goldbook/
11.2. V. Thomsen, D. Schatzlein, D. Mercuro, Limits of detection in spectroscopy. Spectroscopy
18, 112–114 (2003)
11.3. DIN 32 645, Chemische Analytik – Nachweis-, Erfassungs- und Bestimmungsgrenze –
Ermittlung unter Wiederholbedingungen; Begriffe, Verfahren, Auswertung (Beuth Verlag,
Berlin, 2006)
11.4. DIN 38 402, Teil 51, Deutsche Einheitsverfahren zur Wasser-, Abwasser- und Schlam-
muntersuchung – Allgemeine Angaben (Gruppe A) – Kalibrierung von Analyseverfahren,
Auswertung von Analyseergebnissen und lineare Kalibrierfunktionen für die Bestimmung
von Verfahrenskenngrößen (A51), (Beuth Verlag, Berlin, 1986)
11.5. DIN ISO 8466–2, Wasserbeschaffenheit – Kalibrierung und Auswertung analytischer Ver-
fahren und Beurteilung von Verfahrenskenndaten – Teil 2: Kalibrierstrategie für nichtlineare
Kalibrierfunktionen zweiten Grades (Beuth Verlag, Berlin, 2004)
11.6. J. Vrenegor, R. Noll, V. Sturm, Investigation of matrix effects in LIBS plasmas of high-alloy
steel for matrix and minor elements. Spectrochim. Acta B, 60, 1083–1091 (2005)
Chapter 12
Combination of LIBS and LIF
An approach to further increase the sensitivity of LIBS for the determination of

traces is the combination of LIBS and laser-induced fluorescence (LIF, also called
laser-excited atomic fluorescence spectrometry, LEAFS) [12.1]. Limit of detections
for, e.g., heavy metals in soil according to regulatory demands are required to be
significantly below 1 g=g, which is still difficult to achieve with LIBS only.
LIBS–LIF is a so-called hyphenated technique, which was first proposed by
Measures and Kwong [12.2, 12.3]; the pulse of a second laser source, which is
tuned to an optical transition wavelength of the analyte of interest, is directed
into the plasma plume generated by laser-induced breakdown. The fluorescence
radiation induced by this second laser pulse is then evaluated for chemical analysis.
In addition to improved detection limits, a large linear calibration range and reduced
matrix effects are reported for this technique with respect to LIBS [12.3, 12.4]. The
hyphenated technique was used for basic investigations on the physical properties
of laser-induced plasmas [12.5, 12.6] as well as for studies on the applicability
for the inline detection of heavy metals in aerosols [12.7]. Yet, in most cases,
the measurements were performed at ambient gas pressures reduced to values
in the range of 100 mbar. Although this may be helpful to achieve a maximum
signal-to-noise ratio and to provide matrix-independent measurements [12.4, 12.6],
such experimental conditions are not acceptable as far as the applicability of
the technique for in situ measurements is concerned. So far, only Gornushkin
et al. used the LIBS–LIF technique for the determination of heavy metals in soils
and demonstrated an LOD of 1g=g for cobalt measured in air at atmospheric
pressure [12.8].
While LIF offers a high sensitivity, it is basically to be considered as a single
species detection scheme. The combination of LIBS and LIF allows to link the
multispecies capability of LIBS for a broad range of analytes with the high
sensitivity of LIF for individual selected species.
A schematic drawing of the experimental arrangement used is shown in Fig. 12.1
[12.9]. Most of the measurements were performed in a vacuum-tight measurement
chamber that was filled with air or with argon as a buffer gas. This chamber,
shown in a perspective view in Fig. 12.2, was designed to allow investigations

222 12 Combination of LIBS and LIF
Fig. 12.1 Schematic setup for LIBS–LIF, for details see text. SHG D second harmonic generation,
SFM D sum frequency mixing
Fig. 12.2 Perspective view

of the measurement chamber
with the LIBS and LIF
crossing laser beams
on the influence of the ambient gas pressure on the performance of the LIBS–
LIF technique, because the importance of this experimental parameter has been
demonstrated in other publications [12.4–12.6]. Although the gas pressure can be
adjusted between 1 and 1,000 mbar within the chamber, the results presented were
measured at atmospheric pressure. In the measurement chamber, the sample is
positioned on a sample holder and the LIBS laser beam is focused onto the sample
surface from the top. Depending on the mechanical properties of the samples to
be studied, the LIBS measurements may lead to deep craters. In the case of, e.g.,
granular soil samples – cf. Sect. 14.3 – the craters that were generated in pressed
soil pellets by 100 laser pulses with 80 mJ pulse energy were about 2 mm deep.
Hence, the sample holder is mounted on a computer-controlled two-axis-translation
stage within the chamber, which allows to perform studies on the effects of sample
motion during the measurements.
A Q-switched Nd:YAG laser operating at 1,064 nm with a repetition rate of 10 Hz
(Continuum, model Surelite I) delivered laser pulses with pulse energies of up to
450 mJ and a pulse duration of 5–7 ns, which were used for the ablation of sample
material and the generation of the LIBS plasma (Nd:YAG laser 1 in Fig. 12.1).
12 Combination of LIBS and LIF 223
Fig. 12.3 Sectional view of the measurement chamber and the crossing laser beams. C D
measurement chamber, A D translation stage, G D gas inlet and gas outlet, SH D sample holder,
S D sample, F D flange to load the chamber, OD1–4 D observation directions for the fiber optics,
W 1 D window for the LIBS laser beam, W 2 D detection window, FO D fiber optics to guide the
radiation to the spectrometer, L D focusing lens, LB1 D LIBS laser beam, LB2 D LIF laser beam,
dLIF D diameter of LIF laser beam, aLIFS D distance between the axis of the LIF laser beam and
the surface of the sample
The pulse energy was adjusted using a Glan-laser prism in the optical path and the
LIBS pulses were focused onto the sample at normal incidence with a borosilicate
lens (f D 200 mm).
A second, tunable laser source is needed for LIBS–LIF measurements to
selectively excite analyte atoms or ions within the plasma volume. In this case,
a frequency-doubled Q-switched Nd:YAG laser (Spectron, model SL 852 GSLM,
650 mJ at 532 nm, 10 Hz, 6–9 ns) was used to pump a dye laser (Lumonics, model
HD-300) operating at 583 nm (a mixture of Rhodamin 6G and Rhodamin B was
used as a dye) or 554 nm (Fluorescin 27 was used as a dye), respectively. The
dye laser pulses were frequency-doubled by second harmonic generation (SHG)
in a frequency-doubling and -mixing unit (Spectron, model SL 4000 EMX), thus
generating laser radiation in the UV range at 291.5 or 276.8 nm.
For the resonant excitation of Cd at 228.80 nm, the laser pulses at 291.5 nm were
converted by sum frequency mixing (SFM) with the 1,064-nm laser radiation from
the Nd:YAG laser (laser 2 in Fig. 12.1) that is also used for pumping the dye laser.
These LIF-laser pulses were guided into the measurement chamber via mirror optics
and directed into the plasma volume parallel to the sample surface. Figure 12.3
shows a detailed sectional view of the measurement chamber and the crossing laser
beams.
The distance between the LIF-laser beam axis (LB2 in Fig. 12.3) and the surface
of the sample was adjusted to about half of the beam diameter, i.e., aLIFS
dLIF =2. This distance aLIFS was around 4 mm for 228.80 nm and around 8 mm for
276.78 nm. The LIF-laser pulse energy at the plasma plume amounted to 16 J for
228.80 nm and 40 J for 276.78 nm.
The radiation of the LIBS-plasma as well as the LIF light were guided to a
Paschen–Runge spectrometer via fiber optics. A single silica/silica fiber with a core
diameter of 1;000 m was therefore positioned at a distance of 10 cm from the
plasma at an angle of 14ı to the sample surface and 90ı to the LIF-laser beam.
No lenses were used between the plasma and the optical fiber, thus making the
detection insensitive against small fluctuations of the plasma position and allowing
the observation of the whole interaction volume probed by the LIF laser beam.
This simple detection scheme was used, because it was intended to evaluate the
applicability of the technique for fast in situ measurements. For this application, a
minimum number of optical components should be used, which would increase the
risk of misalignment otherwise. Furthermore, the distance between the detection
optics and the plasma should not be too small to avoid the pollution of the optics by
ablated sample material. A further improvement of the sensitivity of the technique is
feasible, if, e.g., a collecting lens of 10 mm aperture is used, which would increase
the solid angle of observation by two orders of magnitude.
A Paschen–Runge spectrometer equipped with exits slits and photomultipliers
(PMTs) for the simultaneous detection of 29 spectral lines was used for the
spectrally and temporally resolved detection of the plasma radiation (cf. Sect. 4.2).
The diameter of the Rowland circle is 750 mm. The spectral bandwidth detected
by each PMT can be estimated from the dispersion of the spectrometer and the
width of the exit slits and amounts to less than 20 pm. The signals from the PMTs
corresponding to the different spectral lines were processed with a fast gateable
multichannel integrator (cf. Sect. 4.5) [12.10]. This multichannel integrator was
controlled by a computer system that also triggered the two laser systems via a
Stanford DG535 pulse/delay generator. A signal synchronized to the laser Q-switch
was used for the triggering of the integrators to minimize timing jitter. The delay
generator triggering the two laser systems allowed to adjust the time difference
between the LIBS-pulse and the LIF-pulse (tLIBSLIF / between 0 and 250 s.
Figure 12.4 shows as an example simplified level schemes for (a) Cd [12.11]
and (b) Tl illustrating the energy levels and optical transitions that are used for
the detection of traces of these elements in soil, cf. Sect. 14.3 [12.9] (The term
symbols indicate the total spin quantum number (first superscript), the total orbital
quantum number, the total angular momentum quantum number (index), the second
superscript “o” stands for odd parity of the level.)
Cadmium is directly excited from its ground level 5 1 S0 to the 51 P1o level at
5.418 eV by UV laser pulses with the corresponding wavelength of 228.80 nm.
The LIF was also observed at 228.80 nm. This excitation/detection scheme was
already widely used for atomic fluorescence measurements with different atomizer
units like graphite furnaces (GF) and excitation sources like electrodeless discharge
Fig. 12.4 Reduced level scheme for (a) cadmium and (b) thallium. Only those levels and
transitions are shown, which are involved in the LIBS–LIF experiment. The dashed arrow indicates
an energy transfer attributed to processes within the buffer gas plasma
lamps (EDL) [12.12] and also tunable lasers [12.13], showing very good limits of
detection, especially for the latter case. A disadvantage of this excitation/detection
scheme is the possibility of a poor signal-to-noise ratio due to laser light scattered
by the ablated material. Assuming that a portion of the scattered laser light shows
a dipole scattering behavior and that multiple scattering can be neglected, the scat-
tered light intensity has a minimum in the direction parallel to the polarization of the
laser radiation. In the experiments, the angle between the direction of observation
and the horizontal polarization of the laser beam amounts to 14ı , so that the intensity
of this portion of scattered light is reduced to about sin2 .14ı / D 6% of its maximum.
In this respect, the detection geometry was designed to reduce the negative effect of
laser light scattering into the observation direction.
The excitation and detection scheme for Tl is given in Fig. 12.4b. The Paschen–
Runge spectrometer used for the detection of the fluorescence light is equipped with
a spectral line for Tl at 351.92 nm, because the strongest emission was observed in
the UV range with spark excitation for this spectral line [12.14]. As there is no
optical transition from the ground state 62 P1=2 o
to the corresponding upper level
62 D5=2 at 4.489 eV, the atoms were excited from the ground state to the 6 2 D3=2
at 4.478 eV level with laser pulses at a wavelength of 276.79 nm. Consequently,
a transition from the 62 D3=2 state to the 62 D5=2 state is necessary to observe a
fluorescence emission at 351.92 nm, either thermally assisted or by collisions with
the buffer gas [12.5]. The spectral difference between the excitation wavelength
and the detection wavelength of more than 75 nm makes sure that scattering of
the laser light does not deteriorate the signal-to-noise ratio in this case. This
excitation/detection scheme was already used for LIF measurements of Tl in a GF
t 0 : LIBS-laser pulse t 1 : LIF-laser pulse
–10 0 10 20 30 40 50 60 70 80 90 100
time [µs]
0.00 0.0000
-0.0025
-0.02
- 0.0050
PMT-signal [a.u.]
-0.04 -0.0075
-0.0100
-0.06
silicon, 390.55 nm -0.0125 c(Si) = 410 mg / g
cadmium, 228.80 nm c(Cd) = 300 µg / g
-0.08 -0.0150
LIBS-signal LIF-signal
-0.0175
-0.10
0 5 10 15 20 25 –1 0 1 2 3 4 5
time with respect to t 0 [µs] time with respect to t1 [µs]
Fig. 12.5 Photomultiplier signals of a silicon (solid curve) and a cadmium (dashed curve)
emission line as a function of time for a combined LIBS–LIF excitation. The species concentrations
of Si and Cd in the soil sample are given in the right graph. Ambient atmosphere:argon at 950 mbar
[12.15]. Cadmium was also studied in the GF using a double-resonance excitation

scheme to eliminate scattering problems, at the expense of increased complexity of
the experimental apparatus [12.15].
Figure 12.5 shows the PMT signals of a silicon and a cadmium line as a function
of time for a combined LIBS–LIF excitation of a Cd contaminated soil sample.
At the time t0 , the LIBS laser beam having the fixed wavelength 1,064 nm is
irradiated on the specimen and excites for both element channels a first peak
attributed to plasma continuum emission (cf. Sect. 2.1, Fig. 2.4). After that a clear
Si response is observed attaining a maximum at about 2 s and decaying in the
following 15 s (LIBS-signal). For Cd, no significant element-specific signal was
observed. At the time t1 , the LIF laser pulse is irradiated tuned to the Cd transition
228.80 nm (cf. Fig. 12.4, a). The result is a distinct Cd fluorescence response,
whereas at the same time the Si signal rests around the base line.
Calibration curves were recorded by integration of the whole LIF signal
(tLIF delay D 0:5 s, tint D 5 s) for different time distances tLIBSLIF between
the ignition of the LIBS plasma and LIF excitation [12.16]. The negative value
tLIF delay D 0:5 s indicates that the integration of the PMT signal was started
before the exciting LIF laser pulse was directed into the relaxing plasma. Thus,
the whole fluorescence signal was recorded, including also the signal background
due to residual laser radiation scattering. This integration scheme was chosen
300
Del100
250
Polynomial Fit of Offkorrkurz_Del100
LIF-signal [a.u.] 200
150
Cd, 228.80 nm
100
LOD3s = 0.3 µg/g
50 BEC = 9.5 µg/g
0
0 20 40 60 80 100 120
Cd concentration [µg /g]
Fig. 12.6 LIBS–LIF calibration curve for cadmium. The analyte signal measured is plotted against
the analyte concentration measured with ICP–OES. The calibration curve is fitted by a second order
polynomial
taking into account the short fluorescence life time of the observed transition of
fl D 1:7 ns [12.13]. The best detection limit was achieved for a large delay between
the LIBS and the LIF laser pulse of tLIBSLIF D 100 s, which is attributed to
the fact that a maximum of cadmium atoms must have relaxed to their ground state
and that the plasma must have expanded to the volume probed by the LIF laser
beam. Additionally, the spectral background due to the broadband emission of the
plasma is negligible for these delay times, because the decay time of this continuum
emission, which can be estimated from the time behavior of the PMT signal after
the LIBS laser pulse, is only in the order of several 100 ns.
The corresponding calibration curve is shown in Fig. 12.6. A strong saturation
behavior is observed for analyte concentrations higher than 50 g=g. This is
attributed to limitations of the signal processing electronics, because the gain of
the electronics was tuned to a maximum in order to achieve a maximum slope of
the calibration curve at low analyte concentrations. As a disadvantage, this led to a
truncation of the LIF signals for high analyte concentrations. Thus, the saturation
should not be confused with optical saturation in this case and an increased range
of linearity is feasible if the amplification of the signal electronics is reduced.
Nevertheless, the range of linearity might be restricted due to self-absorption, which
may occur for the optical transition investigated for cadmium.
Notably, the background equivalent concentration (BEC), calculated as the ratio
of the axis intercept of the calibration curve with the ordinate (cf. Sect. 11.2,
Fig. 11.3 b) – on which the LIF-signal is plotted – divided by the slope of the
calibration curve, amounted only to BEC D 9:5 g=g, showing that scattering of
laser light gave only minor contributions to the LIF signal observed. The detection
limit (cf. (11.16)) for Cd in soil determined under these experimental conditions
resulted in LOD3s D 0:3 g=g, which is the lowest value achieved with this
technique so far.
LIBS–LIF analysis of thallium in soil samples was also studied and will be
described in more detail in Sect. 14.3.
References
12.1. K. Niemax, W. Sdorra, Optical emission spectrometry and laser-induced fluorescence of

laser produced plasma plumes. Appl. Optics 29, 5000–5006 (1990)
12.2. R. Measures, H. Kwong, TABLASER: trace (element) analyzer based on laser ablation and
selectively excited radiation. Appl. Optics 18, 281–286 (1979)
12.3. H. Kwong, R. Measures, Trace element laser microanalyzer with freedom from chemical
matrix effect. Anal. Chem. 51, 428–432 (1979)
12.4. I. Gornushkin, S. Baker, B. Smith, J. Winefordner, Determination of lead in metallic
reference materials by laser ablation combined with laser excited atomic fluorescence.
Spectrochim. Acta 52B, 1653–1662 (1997)
12.5. W. Sdorra, A. Quentmeier, K. Niemax, Basic investigations for laser microanalysis: II.
Laser-induced fluorescence in laser-produced sample plumes. Microchim. Acta II 98, 201–
218 (1989)
12.6. W. Sdorra, K. Niemax, Temporal and spatial distribution of analyte atoms and ions in
microplasmas produced by laser ablation of solid samples. Spectrochim. Acta 45B, 917–
926 (1990)
12.7. R. Neuhauser, U. Panne, R. Niessner, G. Petrucci, P. Cavalli, N. Omenetto, On-line
and in situ detection of lead in ultrafine aerosols by laser-excited atomic fluorescence
spectroscopy. Sensor Actuator B 39, 344–348 (1997)
12.8. I. Gornushkin, J. Kim, B. Smith, S. Baker, J. Winefordner, Determination of cobalt in soil,
steel and graphite using excited-state laser fluorescence induced in a laser spark. Appl.
Spectrosc. 51, 1055–1059 (1997)
12.9. F. Hilbk-Kortenbruck, R. Noll, P. Wintjens, H. Falk, C. Becker, Analysis of heavy
metals in soils using laser-induced breakdown spectrometry combined with laser-induced
fluorescence. Spectrochim. Acta B 56, 933–945 (2001)
12.10. V. Sturm, L. Peter, R. Noll, Steel analysis with laser-induced breakdown spectrometry in
the vacuum ultraviolet. Appl. Spectrosc. 54, 1275–1278 (2000)
12.11. H. Didra, M. Baumann, Cadmium (51 P1 ! 53 PJ / excitation transfer and 53 PJ intramul-
tiplett relaxation by collisions with molecular gases. Z. Phys. D 37, 29–37 (1996)
12.12. B. Smith, M. Glick, K. Spears, J. Winefordner, A comprehensive table of atomic fluores-
cence detection limits and experimental conditions. Appl. Spectrosc. 43, 376–414 (1989)
12.13. M. Bolshov, S. Rudnev, B. Hütsch, Determination of trace amounts of cadmium by laser
excited atomic fluorescence spectrometry. J. Anal. At. Spectrom. 7, 1–6 (1992)
12.14. A. Zaidel, V. Prokofev, S. Raiskii, V. Slavnyi, E. Shreider, Tables of Spectral Lines
(IFI–Plenum, New York, 1970)
12.15. N. Omenetto, P. Cavalli, M. Broglia, P. Qi, G. Rossi, Laser-induced single-resonance and
double-resonance atomic fluorescence spectrometry in a graphite tube atomizer. J. Anal.
At. Spectrom. 3, 231–235 (1988)
12.16. F. Hilbk-Kortenbruck, C. Rieck, R. Noll, Multi-elemental analysis of heavy metals in
soil samples using laser-based analytical techniques, Abstracts of the lecture groups
“Environmental Technology and Fundamentals of Laser-aided In-situ Soil Analysis”, Int.
Meeting on Chem. Eng., Environmental Protection and Biotechnology, ACHEMA 2000,
Conf. Proc. 2000, DECHEMA e.V., Frankfurt a. M. 386–388 (2000)
Chapter 13
Bulk Analysis of Metallic Alloys
This chapter describes LIBS investigations for a quantitative bulk analysis of

metallic alloys such as steel in solid and liquid state, high-alloy steel, and aluminum.
In case of aluminum, the focus is on LIBS analysis of moving Al specimens, which
is of interest for inline identification of scrap pieces for material specific recycling
(see also cf. Sect. 18.4).
13.1 Steel
The chemical analysis of steel is a demanding task for process control in steelmak-
ing and for quality assessment of preproducts. The dominant analytical tools applied
nowadays are based on physical methods such as spark discharge-optical emission
spectrometry (SD-OES), X-ray fluorescence (XRF), or absorption spectroscopy of
analytes in flames. These techniques are mostly used in off-site laboratories and
they require preparational steps to present the steel sample in a physical state
necessary for the respective method. The sample preparation includes steps such
as cutting, milling, grinding, or drilling, which are time consuming and require
high maintenance efforts even in automated container laboratories. Hence, a strong
need is coming up to simplify or even avoid those steps of chemical analysis such
as sample taking, sample transport, and sample preparation. Laser-based analytical
methods will play a key role for the development of inline methods in metallurgical
processing due to the following features: noncontact measurement at distances of
centimeters to meters, high measuring speed, sample preparation, or conditioning
by the laser beam itself.
13.1.1 Solid Steel Samples
For an inline analytical method the multielement capability is an important

issue. Laser-induced breakdown spectrometry (LIBS) is a method enabling the

230 13 Bulk Analysis of Metallic Alloys
Fig. 13.1 Schematic setup for LIBS of steel samples in the VUV range. M D mirror, L D lens,
W1 D entrance window of the measurement chamber, C D measurement chamber, S D sample,
W2 D entrance window of the spectrometer made of MgF2 , ES D entrance slit of the spectrometer,
width 15 m
simultaneous analysis of various elements in a short time. However, the analytical

performance in terms of limits of detection (LOD) reported for steel constituents so
far does not achieve those of conventional methods such as SD-OES [13.1–13.3].
Hence, the main goal presented in the following was to improve the plasma
excitation by multiple laser pulses and to make use of emission lines in the vacuum
ultraviolet (VUV) spectral range [13.4]. The motivation of this work was to establish
the basis for future inline applications of LIBS in the steel industry requiring
LOD for light elements such as carbon, phosphorus, and sulfur in the range below
10 g=g.
The experimental setup is shown in Fig. 13.1 (cf. also Sect. 4.6). A Q-switched
Nd:YAG laser operating at 1,064 nm was used to excite the plasma. The Q-switch
electronics of this laser was modified to generate up to three separated laser pulses
within a single flashlamp pulse instead of a single laser pulse (SP; cf. Sect. 3.2,
Fig. 3.9 and Sect. 6.2). The flashlamp repetition rate is 10 Hz. The influence of
multiple laser pulses on the dynamics and physical state of the induced plasma was
described in Chap. 6. The laser emits three equal-energy pulses of 16 ns duration
(FWHM) and with a time separation of 25 s between the first and the second pulse,
and 40 s between the second and the third, respectively. The burst energy amounts
to 300 mJ. The laser beam is guided to a plano-convex lens with a focal length of
80 mm. The converging laser radiation is then guided via a mirror and an optical
window into a measurement chamber. This chamber has a cover plate on top with
a circular opening having a diameter of 12 mm. The sample to be analyzed is put
onto this cover plate and located in such a way that the opening is closed. The focus
of the laser beam was adjusted to position the beam waist inside the sample with a
distance of s D 6 mm from the sample surface. The diameter of the craters formed
on the sample surface is about 300 m.
The optical axis of the VUV vacuum spectrometer describes the line between
the center of the diffraction grating and the middle of the entrance slit. This optical
axis intersects the propagation axis of the laser beam on the surface of the sample.
The angle between the surface of the sample and the optical axis of the spectrometer
amounts to ˛D 18ı (cf. Fig. 4.18). With this orientation, it is assured that radiation
13.1 Steel 231
Table 13.1 List of elements, Element Ionization Wavelength En (eV)

wavelengths, and energies En stage (nm)
of the upper levels of the
observed lines [13.30] P I 178.28 6.95
S I 180.73 6.86
Fe II 187.75 9.13
C I 193.09 7.68
Ni II 231.60 6.39
Cr II 267.72 6.18
Fe II 271.44 5.55
Si I 288.16 5.08
Mn II 293.31 5.40
from the whole laser-induced plasma can be received by the spectrometer aperture.
The chamber has an internal volume of about 42 cm3 . This volume is flushed with
Argon 5.0 at 1 atm and a flow rate of about 8 l/min. The entrance window of the
spectrometer is made of VUV transparent magnesium fluoride.
The spectrometer has a Rowland circle diameter of 750 mm and is equipped
with a holographic diffraction grating with 2,400 lines/mm. The optical system
of the spectrometer is installed inside a vacuum vessel which is evacuated to
<103 mbar. The spectral resolution amounts to 25 pm in first order. Table 13.1
lists the emission lines where exit slits are positioned along the Rowland circle.
The lines Fe 187.75 and Fe 271.44 nm are used as internal standards. The dispersed
radiation passing the exit slits is detected by photomultiplier tubes (PMT).
The signals of 16 PMTs are processed by a multichannel integrator electronics
(MCI; see Fig. 4.16, Sect. 4.5). The anode currents of the PMT are fed into
current/voltage-converters that are positioned close to the sockets of the PMT inside
the vacuum vessel to reduce as far as possible a capacitive load of the current source.
The transformed signals are then transmitted via coaxial cables and BNC-feed
through out of the vessel to the MCI. The bandwidth of the i/u-converters and the
integrators is in the range of 10 MHz to enable an undistorted signal processing of
the transient PMT currents, which are characterized by significant signal variations
on a timescale down to 25 ns.
The gate for the integrators is generated by a delay/pulse generator which is
triggered by the Pockels cell synchronization monitor of the laser. The outputs of
the integrators are transmitted to a multiplexer and an A/D-converter. The digitized
charges of the integrators for each laser pulse are stored and processed on a
computer.
For the calibration of the LIBS measurements, a set of 14 reference samples
with certified concentrations of the investigated elements was used. The samples
are not prepared mechanically unless the surface is covered by craters from former
measurements. In that latter case, the sample is simply ground manually with an
Al2 O3 -grinding paper of grit size 60 for about 5 s. The laser beam itself is used to
ablate additionally residual surface layers or contaminations. For the quantitative
0.6
1st pulse 2nd pulse

0.5
PMT signal of C channel [V]
0.4
BAS 456
0.3 c(C) = 1010 μg / g
BR 53
0.2 c(C) < 10 μg / g
0.1
0.0
0 2 4 6 8 25 27 29 31 33
time [μs]
Fig. 13.2 Carbon emission as a function of time after the first and the second pulse of a laser
burst. The solid line shows the signal obtained for the sample BAS 456 with a carbon content of
1;010 g=g, and the dotted line belongs to the sample BR 53 with a carbon content below 10 g=g
analysis, the number of prebursts (cf. Chap. 11, Table 11.1) applied to prepare the
sample surface amounts to Npp D 100.
Figure 13.2 shows time-resolved PMT signals of the carbon channel at 193.09 nm
picked up at the current/voltage-converters with a Tektronix TDS 784D oscillo-
scope. The dotted line was gained using a sample with a carbon content <10 g=g;
the solid line belongs to a sample with a carbon content of 1;010 g=g. With the
beginning of the first laser pulse, the carbon signal shows a steep increase with a
rise time of less than 10 ns. After achieving a maximum, the PMT signal decays
within about 500 ns. In the decaying phase, only a small difference between the
signals for the different samples is observed. After the second pulse of the burst,
the situation is very different. Whereas the rise time is comparable the decay time
differs significantly for the two samples. Moreover, the signal level is much higher
compared to the signals after the first laser pulse. After the third laser pulse of the
burst about the same signal levels as after the second pulse are achieved.
The signal evolution observed for different elements, element concentrations,
and/or experimental parameters, e.g., laser burst energy, pulse separation, and
focusing is qualitatively the same as described above. However, the decay times
and thus the optimum position and width of the integration gate depend on these
parameters. The best results with respect to the LOD were gained with a delay time
of tdelay D 650 ns and an integration gate width of tint D 12 s.
The experimental setup was calibrated for seven elements for low-alloy steel.
Figures 13.3–13.5 show calibration curves for carbon, phosphorus, and sulfur.
The LOD was calculated on the basis of the 3s-criterion, where s is the
empirical standard deviation of the intensity ratio for the sample with the lowest
analyte content [cf. Chap. 11, relation (11.16)]. Each sample was measured five
13.1 Steel 233
I C 193.09 nm / I Fe 187.75 nm 1.4 BAS 453
1.2
1.0 BAS 456

NIST 1166
0.8
NIST 1165
BAS 451
0.6
NIST 1766
0.4 Hu 87D, Br 53, Ov Fe-130
0 200 400 600 800 1000 1200 1400 1600 1800

carbon concentration [μg / g]
Fig. 13.3 Calibration curve of carbon for the emission line 193.09 nm. Sample denominations:
NIST, National Institute of Standards & Technology; BAS, Bureau of Analysed Samples Ltd.; Hu,
Hanau; Br, Breitländer; Ov, Ovako Steel
0.7
BAS 454
I P 178.28 nm / I Fe 187.75 nm
0.6 BAS 453
BAS 452
0.5 BAS 460
Br 53 BAS 456 BAS 458
0.4 BAS 451 NIST 1166

NIST 1165
Hu 87 D
RROY 1, NIST 1766, Ov Fe 130
0.3
0 100 200 300 400 500 600
phosphor concentration [μg /g]
Fig. 13.4 Calibration curve of phosphorus for the emission line 178.28 nm. For abbreviations see
Fig. 13.3. RROY, Rautaruukki Oy
times. Between the measurements, the sample was shifted on the top cover of
the measurement chamber to expose another surface part to the laser beam. Each
measurement comprises 600 laser bursts, where Npp D 100 are applied as prebursts
to locally condition the sample surface and Nmp D 500 measurement bursts to
gain the spectral signals to be evaluated. The data points plotted in the Figs. 13.3–
13.5 denote the mean values from five measurements. Numerical corrections such
0.7
BAS 454
0.6
I S 180.73 nm / I Fe 187.75 nm
BAS 453
0.5
BAS 452 BAS 458

0.4 BAS 456
RROY 1 BAS 451

BAS 460
0.3 Ov Fe 130
NIST 1165, NIST 1165
Hu 87 D
0.2 NIST 1766
Br 53
0 100 200 300 400 500

sulfur concentration [μg / g]
Fig. 13.5 Calibration curve of sulfur for the emission line 180.73 nm. For abbreviations, see
Figs. 13.3 and 13.4
Table 13.2 Results of the calibration measurements: limit of detection (LOD3s /, background
equivalent concentration (BEC), concentration range of the calibration, and coefficient of deter-
mination (r 2 /
Element LOD3s .g=g/ BEC (g=g) Concentration r2
range (g=g)
C 7 530 0–1,800 0.999
P 9 640 0–600 0.998
S 8 210 0–500 0.987
Mn 9 320 0–14,000 0.999
Ni 6 500 0–700 0.991
Cr 7 330 0–1,000 0.991
Si 11 680 0–6,000 0.990
as interelement or matrix corrections were not applied. The horizontal error bars
indicate the concentration errors stated in the certification sheets of the reference
samples; the vertical error bars show the standard deviation calculated from the five
measurements. The curve plotted in Fig. 13.3 is a quadratic fit using a least squares
algorithm describing a saturation behavior of the intensity ratio C/Fe with increasing
carbon concentration.
Figures 13.4 and 13.5 show the calibration curves for phosphorus and sulfur.
The calibration curves are linear functions fitted to the data points. In the setup
of Fig. 13.1, the location of the laser irradiation is fixed; hence the only spatial
averaging is given by the crater diameter of about 300 m. Conventional methods
such as SD-OES have an inherent spatial averaging over areas with a diameter of
about 5 mm (cf. Fig. 3.17). For a comparison of results gained with these methods
and LIBS, a spatial averaging, e.g., by a relative motion between sample and laser
13.1 Steel 235
beam, has to be performed (cf. Sect. 3.6). Table 13.2 summarizes the results of the
calibration for seven elements using as internal standard the Fe 187.75 nm line. The
LODs are <10 g=g for most of these elements. The coefficients of determination
are always greater than 0.987. In general, using the line Fe187 instead of Fe271 as
the internal standard resulted in a better analytical performance in terms of the LOD.
For the elements C, P, S, Si, and Ni, the achieved LOD with Fe187 are approximately
20–50% smaller than with Fe271. In the case of Cr and Mn, the improvement is
less significant. For a further improvement of the LOD, the background equivalent
concentration (BEC) has to be reduced. The origin of the rather high BEC values
observed is presumably due to the high signal level of the PMT immediately after
the beginning of the laser pulse irradiation.
The applied gating of the integrators suppresses this signal part, but there is still
the possibility that the electrical charge generated in the beginning of the PMT signal
distorts the signal behavior at later times.
Multipulse excitation and observation of VUV emission lines allow to push down
the LOD for light elements in low-alloy steel samples in the range below 10 g=g.
No or at least a reduced effort in sample preparation is required in comparison to
SD-OES. The laser itself prepares the sample by a localized ablation of surface
layers or contaminations. With these results, the capability of LIBS to be applied
for inline analytical tasks in the steel industry is demonstrated.
13.1.2 Solid Steel Samples with Scale Layer
Production control samples are used in steel works to control steel composition.
In certain phases of the steelmaking process, the chemical composition of the
steel melt has to be determined with a certain accuracy in a short time of several
minutes. The further treatment and alloying of the heat are regulated dependent
on the determined composition. The interest in steel works analysis is a cost-
effective analytical method, which reduces the required alloying material, shortens
the process control time, and reduces maintenance costs of the analytical equipment.
For process control, the state of the art is to take samples with immersion
sampling probes which are dipped into the melt in order to be filled with the
liquid steel. During the solidification and cooling of the samples, a scale layer is
formed at the surface having a thickness in the range of 0:1–1 mm. In general, the
composition of the scale layer deviates from the composition of the bulk material.
In addition, the thickness of the scale layer depends on the type of the alloy steel.
Especially, carbon shows an enhanced concentration in the inhomogeneous scale
layer. Hence, the analysis of the bulk material requires the removal of the scale
layer. The conventional approach is to remove the scale layers by automatic grinding
or milling machines in the steel works. The subsequent analysis is carried out
predominantly by OES using arc/spark excitation, XRF analysis, and/or combustion
analysis (for the light elements). The analysis of the elements C, P, and S is of
primary interest in steelmaking since these elements are important for the process
Fig. 13.6 Schematic view of the experimental setup. P D laser-induced plasma plume; S D
sample; OL D outlet; Ar D argon gas; W1, W2 D optical windows; L D lens; M1, M2 D mirrors;
BC D beam combiner; ES D entrance slit; OA D optical axis; D D detectors, signal electronics;
SH D shutter
as well as are difficult to analyze spectrometrically. In the investigations presented

in this section, further elements of interest are included such as Al, Cr, Cu, Mn,
and Mo.
Elemental analysis of steel by LIBS is reported for liquid steel [13.5–13.8],
as well as solid steel alloys, such as low-alloy steel [13.4, 13.9], or stainless steel
[13.10, 13.11]. Some applications in steel analysis are summarized in [13.12]. LOD
of less than 10 g=g have been reported [13.4, 13.9, 13.13]. For bulk analysis of
solid steel, samples with thin surface layers are measured or the samples are ground
similar to sample preparation for arc/spark spectrometry. Nevertheless, before
starting the LIBS signal acquisition, laser pulses are irradiated on the sample to
ablate thin surface layers or contaminations (cf. Sect. 3.8). The ablated thicknesses
using a typical LIBS setup with SP are in the range of 0.05 mm, which is not
sufficient for the removal of thicker scale layers regarded here. In the following
triple pulse laser bursts are applied for an enhanced local laser ablation of the scale
layer and subsequent elemental analysis of the bulk material is carried out by LIBS
using double pulse (DP) laser bursts (cf. Chap. 6). The sample preparation and the
analysis are combined in a single setup avoiding a mechanical sample preparation
and reducing sample transportation times [13.14].
To enhance the laser ablation of the scale layer during the preparation step,
10-Hz repetitive bursts consisting of three laser pulses (triple pulse (TP) bursts) are
generated by the superposition of two Q-switched Nd:YAG laser beams (wavelength
1,064 nm); see Fig. 13.6. Laser 1 emits SP and laser 2 produces bursts of two laser
13.1 Steel 237
Fig. 13.7 Temporal

sequence of laser pulses and
signal integration. The single
pulse output of laser 1 is
superposed with the double
pulse output of laser 2 for
scale ablation. PL D laser
power; SP , DP;1; DP;2 D
pulse widths of single pulse
or double pulses,
respectively; t , tl;d , tdel D
time delays; tint D integration
time interval of PMT signal
pulses (DP bursts, interpulse separation t D 50 s); see Fig. 13.7. The widths of
the laser pulses are 6–10 ns depending on the operating parameters of the laser.
The time delay of the second laser pulse to the third laser pulse within the TP burst
of the superposed beams is set to tl;d D 1 s. The subsequent LIBS analysis is
carried out with DP bursts of laser 2.
The laser bursts are focused on the sample positioned at the sample stand, which
is flushed by argon gas. The emitted light of the laser-induced plasma plume (P in
Fig. 13.6) is detected directly without further optical imaging by a vacuum Paschen–
Runge spectrometer (Model RS 1000 from OBLF, Witten, Germany, modified for
this application) equipped with PMT for detection of the spectral line signals and
the zeroth order signal. The distance of the sample surface to the entrance slit
of the spectrometer is 300 mm. The detection angle between the optical axis
of the spectrometer to the sample surface is ˛D 10ı (cf. Sect. 4.6, Fig. 4.18).
The diameter of the Rowland circle of the spectrometer is 0.5 m and the spectral
resolution amounts to 0:02 nm (grating, 2,700 grooves per millimeter). The PMT
signals are directly fed into an MCI electronics (cf. Sect. 4.5) for integration within
a preset time interval, tint (in the following tint D 10 s) after the laser pulses. The
time delay of the integration is set to tdel D 4 s after the second pulse of the DP.
The spectral lines installed are listed in Table 13.3.
The superposed laser beams are focused on the sample by a plano-convex lens
with a focal length f D 200 mm. The diameters of the laser beams amount to
6 mm. The sample surface is positioned perpendicular to the laser beam. The
energy of an SP is ESP D 200 mJ. The energies of the DP are EDP;1 D EDP;2 D
200 mJ; i.e., the energy of the DP bursts is Eb D EDP;1 C EDP;2 D 400 mJ.
Analogously, the energy of a TP burst is 600 mJ. The energy Etotal denotes the burst
energy times the number of bursts irradiated on the sample.
Table 13.3 Element, Element (nm) Ej (eV) log.gi fij /

wavelength , excitation
energy Ej , log.gi fij /, Al 396.152 3.143 0.323
cf. (9.8), of the spectral C 193.091 7.685 0.211
lines installed in the Cr 267.715 6.180 0.36
Paschen-Runge spectrometer Cr 286.257 5.856 0.21
Cu 324.754 3.817 0.062
Fe 271.441 5.553 0.44
Fe 187.747 9.126 0.148
Mn 293.306 5.401 0.12
Mo 281.615 6.071 0.51
Ni 225.385 6.822 0.043
P 178.283 6.955 0.389
S 180.731 6.861 0.22
Si 251.611 4.954 0.24
Some of the spectral lines are selected primarily for high-
alloy steel analysis which is not regarded in this section
Each side of the production control sample is measured three times at different
sample positions. If applied, the number of laser pulse bursts for ablation is set to
Nabl D 1;000. The subsequent LIBS analysis at one sample position is performed
using Npp .DP/ D 100 for additional preparation of the surface and Nmp .DP/ D 500
for analysis.
Production control samples of low-alloy steel, so-called lollypop samples, were
provided by IRSID including a spark OES analysis [13.15]. The samples were
prepared by grinding one side, while the other side is left with the original scale
layer. The samples are measured by spark OES on the ground side and with the
laser method on both sides. The results are compared to evaluate the analytical
performance of the laser method including preparation. The concentration ranges
of the analyzed production samples and the reference samples for determination
of the LOD are listed in Table 13.4. The reference samples used for calibration are
standard certified reference materials of carbon steel from BAS (Bureau of Analysed
Samples, Middlesbrough, UK) and low-alloy steel from NIST (National Institute of
Standards and Technology, Gaithersburg, MD, USA). In addition, two high purity
iron samples: BR53 (setting up sample, not certified) from Breitländer referensma-
terial, Sweden, and Fe 130–35 from Ovako Steel, Sweden, with certificate of Ovako
Steel, are included. The samples used for each specific calibration curve are listed
in Table 13.6. It should be mentioned that the homogeneity of the BAS and NIST
samples is not certified at a scale relevant to LIBS analysis, although the NIST
samples are tested for homogeneity by atomic emission and X-ray spectrometry
at NIST.
A high ablation rate is essential to remove the scale layer fast and efficiently.
To illustrate the enhancement of material ablation by using multipulse bursts, the
depths of the laser-induced craters on the ground side of production samples are
measured for SP, DP, and TP laser irradiation. The total energy of the pulses
irradiated on the sample is held constant at Etotal D 600 J. The craters are generated
13.1 Steel 239
Table 13.4 Concentration ranges of the reference samples and the production control samples of
low-alloy steel
Element NLOD Concentration Concentration
range of reference range of production
samples (m.-%) samples (m.-%)
C 9 0.001–1.03 (0.25) 0.004–0.80
S 18 0.0005–0.06 0.004–0.02
Mn 13 0.0005–1.58 (1.00) 0.040–1.60
P 18 0.0005–0.06 0.002–0.20
Si 14 0.0005–0.63 (0.04) 0.002–0.40
Cu 9 0.0005–0.51 (0.10) 0.005–0.03
Cr 10 0.0015–1.48 (0.26) 0.001–0.04
Al 12 0.0005–0.11 0.001–0.20
Ni 9 0.002–1.99 (0.20) 0.010–0.05
Mo 9 0.001–0.14 (0.5) 0.001–0.05
NLOD , number of reference samples used to determine the LOD by a linear calibration
curve. The maximum concentration of this subset of samples is given in brackets
Table 13.5 Crater depth hc and crater diameter dc for different laser pulse sequences applying the
same total laser energy of Etotal D Nabl Eb D 600 J
Nabl Eb (J) hc (mm) dc (mm)
SP 3,000 0.2 0.10 1.20
DP 1,500 0.4 0.19 1.21
TP 1,000 0.6 0.27 1.19
with Nabl .SP/ D 3;000, Nabl .DP/ D 1;500, and Nabl .TP/ D 1;000. The presented
values are the average of three different sample positions for each pulse sequence
(see Table 13.5). The crater depth of hc D 0:27 mm achieved by TPs is greater than
the depth achieved by SP by a factor of 2.7 and greater than the depth achieved by
DP by a factor of 1.4. The crater diameters do not change significantly for these
pulse sequences.
The calibration with reference samples has been evaluated by the LOD3s (cf.
Sect. 11.2). For the assessment of the combined preparation and analysis of the
production control samples, the congruence Ri and the coefficients of determination
ri2 are evaluated. Ri is defined as the mean quadratic difference of the determined
concentrations on the ground and on the scale side of element i [in analogy to the
mean residual, see Sect. 11.3, relation (11.19)]. The aim of the analysis is to improve
the congruence, i.e., to reduce Ri :
v
u N
u1 X
Ri D t .cscale;a;i cground;a;i /2 ; (13.1)
N aD1
Table 13.6 LOD3s values without (1) and with (2) laser preparation
Refer. Line LOD.1/ .g=g) LOD.2/ .g=g) Reference samples
Al Fe 187 7 7 BR53; Ovako; BAS: 456/1–460/1;
NIST: 1761–1764, 1767
C Fe 187 17 9 BR53; Ovako; BAS: 451/1, 453/1,
456/1, 458/1; NIST: 1763,
1765, 1767
Cr Fe 187 7 27 BR53; Ovako; BAS: 451/1–455/1;
NIST: 1761, 1765, 1767
Cu Fe 187 6 10 BR53; Ovako; BAS: 453/1–455/1;
NIST: 1762, 1763, 1765, 1767
Mn Fe 271 4 10 BR53; Ovako; BAS: 451/1,
454/1–460/1; NIST: 1761,
1765, 1767
Mo Fe 187 3 9 BR53; Ovako; BAS: 451/1–453/1,
455/1; NIST: 1761, 1765, 1767
Ni Fe 187 31 35 BR53; Ovako; BAS: 451/1–454/1;
NIST: 1764, 1765, 1767
P 0th order 4 22 BR53; Ovako; BAS: 451/1–460/1;
NIST: 1761–1765, 1767
S 0th order 9 19 BR53; Ovako; BAS: 451/1 - 460/1;
NIST: 1761 – 1765, 1767
Npp D 100, Nmp D 500. The reference samples used for each calibration are listed in the last
column. The notation, e.g., 451/1–453/1, means samples No. 451/1, 452/1, and 453/1 from BAS
(1) Nabl D 0; (2) Nabl D 1;000
where i is the element, N is the total number of production control samples, a is

the index of the sample, and cscale , cground are the concentrations (g=g) determined
with LIBS on the scale or the ground side, respectively.
To determine the element concentrations of the samples, a suitable calibration
method has to be found. Therefore, it is necessary to investigate the influence
of the crater geometry, e.g., crater depth and crater diameter, on the analysis
step. For the analysis, the analytical performance is compared with and without
the ablation step concerning the sensitivity and precision of the calibration. The
calibration measurements have been carried out on 18 low-alloy reference samples
(see Tables 13.4 and 13.6) with two different pulse sequences: (a) Nabl .TP/ D 0,
Npp .DP/ D 100, Nmp .DP/ D 500, for normal calibration without preablation,
and (b) Nabl .TP/ D 1;000, Npp .DP/ D 100, Nmp .DP/ D 500, including the
laser preparation. After the ablation step, the analysis starts in a crater with a
depth of 0:27 mm. The crater depth was measured using a white-light interfer-
ometric microscope (NewView 200, Zygo Corporation). The calibration curves are
evaluated concerning the LOD, sensitivity, and precision. For every analytical line
and pulse sequence, calibration curves are taken in intensities each referenced to
the reference lines Fe 271.441, Fe 187.747 nm, and the zeroth order signal. The
best reference line for each element is listed in Table 13.6. Figure 13.8 shows the
calibration curves for carbon and molybdenum with and without the ablation step.
Comparing the precision and sensitivity of calibration for both pulse sequences,
13.1 Steel 241
a
3.5


Nabl = 0
3.0 Nabl = 1000
2.5
2.0
1.5
1.0
0.5
0.0
0.0 0.2 0.4 0.6 0.8 1.0
carbon concentration [m.- %]
b
0.7

Nabl = 0
0.6 Nabl = 1000
0.5
0.4
0.3
0.2
0.1
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
molybdenum concentration [m.- %]
Fig. 13.8 Calibration curves without (Nabl D 0) and with (Nabl D 1;000) laser preparation for (a)
carbon and (b) molybdenum
there is a general slight deterioration when the ablation step is included. All listed
element lines show this behavior. The smallest difference is observed for the element
carbon.
The LOD3s for most of the observed elements are below 10 g=g calibrating
without laser preparation; see Table 13.6. For Ni and C, the LOD values amount 31
and 17 g=g, respectively. Calibration with laser preparation, see above, resulted
in higher LOD. For most of the considered elements, there is an increase of the
LOD3s by factors between 2.1 (sulfur) and 5.5 (phosphorus). There is no significant
change for Al and Ni. The LOD3s value for carbon decreases from 17 to 9 g=g. The
obtained accuracy and precision of analysis depend on the reference line. The best
analytical performance for Mn is attained for the referencing to the iron line Fe 271.
For phosphorus and sulfur, the zeroth order signal is found to be the better reference
signal. For the other six listed elements, the Fe 187 line is the best reference line.
The calibrations described above showed that there is an influence of the crater
depth and diameter on the sensitivity and precision of the analytical results. The
calibration without the laser ablation step is more precise and more sensitive than
, scale side 10 Nabl = 0

Nabl = 1000
8
0
0.0 0.5 1.0 1.5 2.0 2.5
, ground side
Fig. 13.9 Analysis on the scale and ground side of ten production control samples. I (element,
wavelength of spectral line) are the integrated PMT signals. The carbon intensities divided by the
reference Fe intensities are measured on the scale side and the ground side without (Nabl D 0) and
with (Nabl D 1;000) the laser preparation
the corresponding one with laser ablation. A conclusion is that the analysis of
production samples has to be carried out with the same parameters including the
same pulse sequences as the calibration. Results are presented concerning the
analysis of production control samples that are measured both on the scale and on
the ground side. The basis for the determination of the element concentrations of the
production control samples is the calibration with five reference samples that cover
almost the concentration range of the production control samples. The analysis is
carried out with ten production control samples for the elements C, Si, Mn, Cr, Mo,
Ni, and Cu. Ten repetitions at different positions of each ground and each scale side
of the samples are performed during the measurements. Calibrations with the pulse
sequences Nabl .TP/ D 0, Npp .DP/ D 100, Nmp .DP/ D 500, and Nabl .TP/ D 1;000,
Npp .DP/ D 100, Nmp .DP/ D 500 are conducted. The determined concentrations on
the ground side are defined as the reference values.
Figure 13.9 illustrates the congruence of the analysis on the scale and the
ground side by comparing the spectrometric intensity ratios of carbon and iron.
The bisecting line represents the ideal location of the referenced intensities or the
determined concentrations (cf. Figs. 13.10 and 13.11). Carbon is the element with
the greatest deviation concerning the concentration in the scale layer and the bulk
material. The congruence for carbon is decisively improved using the ablation step;
see Fig. 13.9.
In Figs. 13.10 and 13.11a–c, only the measurements with the ablation step are
shown. Figure 13.10 gives the determined concentrations of a LIBS measurement
on the scale side versus the arc/spark measurement on the ground side of the same
sample. In Fig. 13.11a–c, the determined concentrations of the LIBS measurements
13.1 Steel 243
C concentration [m.- %], scale side, LIBS

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
C concentration [m.- %], ground side, arc / spark
Fig. 13.10 Congruence between the determined concentrations measured by LIBS on the scale
side and the determined concentrations measured by arc/spark spectrometry on the ground side for
C. Coefficient of determination is r 2 D 0:944. Analysis with laser preparation of ten production
samples. The calibration of the LIBS spectrometer was carried out with only five reference samples,
while the arc/spark spectrometer was calibrated with a greater set of reference samples
of the scale side are compared with those of the ground side. The diagrams of
Figs. 13.10 and 13.11a for C do not differ significantly and the coefficient of
determination is approximately equal, which means that the comparison of the LIBS
concentrations in Fig. 13.11 is suitable.
The coefficient of determination averaged over the observed elements amounts
to r 2 > 0:99. The corresponding values for the congruence and precision are
summarized in Table 13.7. For most elements such as C, Si, Cr, Cu, and Ni, the
congruence Ri is reduced. The carbon analysis is strongly improved by a factor
of nearly 25 to an absolute value of about R D 730 g=g applying the ablation
step. This factor rises to 50 when two samples assessed as outliers are neglected.
This corresponds to a congruence of R D 289 g=g. For the elements Cr, Si, Ni,
and Cu, the congruence values are decreased by factors between 1.2 for Ni and
4.1 for Cu. The absolute Ri values vary relatively strongly because of the different
concentration ranges of the seven elements studied.
The results demonstrate the potential of the laser-based method for a fast
preparation and analysis of production control samples. In an industrial application,
the conventional steps of grinding the samples, their transport to a spark OES
spectrometer, and analysis can be replaced and combined in a single setup for laser
preparation and analysis by LIBS.
For a fast preparation of the sample, the scale removal was strongly improved
by applying TP bursts. The laser-induced craters are deeper than the ones generated
by SP using the same total laser energy by a factor of 3. LOD below 10 g=g are
achieved for most of the elements analyzed without the laser preparation step. The
Fig. 13.11 Congruence a

between the determined
C concentration [m.- %], scale side

concentrations on the scale
0.7
and ground side for the
elements: (a) C, r 2 D 0:946, 0.6
(b) Si, r 2 D 0:992, and 0.5
(c) Mn, r 2 D 0:991. Analysis
0.4
with laser preparation of ten
production samples 0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
C concentration [m.- %], ground side
b
Si concentration [m.- %], scale side
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Si concentration [m.- %], ground side
c
Mn concentration [m.- %], scale side
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Mn concentration [m.- %], ground side
analytical performance deteriorates slightly when the ablation step is applied prior
to the analysis step under the chosen experimental conditions.
The analysis on the scale side attains good congruences for all elements and high
coefficients of determination concerning the analytical results on the conventionally
ground side. Especially, the analysis of carbon is decisively improved.
13.1 Steel 245
Table 13.7 List of the congruence values Ri of seven elements including the analysis of ten
production samples without (1) and with (2) the ablation step
.1/ .2/ .1/ .2/
Element Ri .g=g) Ri .g=g) Ri =Ri
C 17,940 730 25
Ca 14,511 289 50
Cr 110 50 2.1
Cu 160 38 4.1
Mn 190 190 1.0
Mo 19 19 1.0
Ni 48 39 1.2
Si 350 280 1.3
In the last column, the factor of improvement of analysis with laser preparation
is listed
a
Without two samples assessed as outliers. (1) Nabl D 0, (2) Nabl D 1;000
13.1.3 Liquid Steel
The increasing demands on high purity steel with lower impurity contents and a
closely specified composition is a challenge to secondary metallurgy as well as to
chemical analysis in steel industries. One major task is to adjust the multielement
steel composition during ladle processing more precisely and to reduce the costs by
minimizing the time effort, energy, and material consumption. It is essential for the
competitiveness of steelmaking industry to develop new techniques, which enable
the steel works to meet these requirements for ladle metallurgy and to produce more
efficiently high purity steel within narrow tolerance limits.
Fast cycle times for primary melting units and the use of extended casting
sequences result in net available residence time in a ladle, which can be as short as
about 30 min. Conventionally, monitoring of the refining operations by a calibrated
multielement analysis plays an important role to optimize the process control in
ladle metallurgy [13.16, 13.17]. At present, monitoring of the melt composition
takes place off-line and discontinuous by taking samples. The solidified samples
are transported for subsequent analysis to the laboratory. Considerable progress
has been made with regard to analysis duration and analytical performance of fast
analysis of solid steel samples. With optimum local conditions, analysis durations
of about 5 min including sample taking are achieved using conventional analyzing
methods such as XRF spectrometry and SD-OES.
Nevertheless, the samples are only spot checks of the melt composition with
regard to time and space, and reactions to irregularities of the process have a delay
of at least 5 min. An instantaneous control of the process is not possible.
In order to overcome this limitation in future, it is necessary to get an increased
data acquisition rate for the chemical analysis of the actual state of the steel melt,
which can be provided to the steel worker for an efficient control of the refining
processes. The homogenizing of the melt can be observed directly as well as
irregularities and the end of the process are recognized earlier.
The fast multielement laser-based analysis performed directly by the immersion

of a moveable lance into the melt is a methodical approach, which has the
potential to carry out this task during ladle processing. Therefore, the goal of
the investigations is to improve the sensitivity of LIBS and to demonstrate its
capability for a fast quantitative multielement inline analysis for ladle metallurgy.
The emphasis lies on the analysis of the light elements carbon, phosphorus, and
sulfur in liquid steel.
Several papers were published presenting investigations of LIBS for the analysis
of liquid steel. LIBS for inline analysis in a steel converter was studied using an
access of the laser beam to the melt via a gas-flushed hole in the sidewall of the
converter. In plant trials, an LOD for carbon of about 200 g=g was demonstrated
[13.18]. The analysis of molten steel in a small crucible heated with an induction
furnace was investigated with a laser beam access irradiating a free melt surface
from the top. LOD for carbon of 250 g=g were reported [13.6]. First tests of
an inline LIBS-based analyzer in secondary steelmaking were performed in the
Scunthorpe Works in UK on an argon stir station where compositional trimming was
carried out prior to continuous casting [13.19]. Changes of the carbon content in the
steel melt of about 300 g=g could be monitored inline. In a laboratory induction
furnace, the changes of the melt composition by successive additions of the alloying
elements Cr, Cu, Mn, and Ni were studied using LIBS [13.7]. The detection limits
reported lie in the range between 530 and 2;070 g=g for these elements. Light
elements were not investigated. In conclusion, the analytical performance of LIBS
fulfills for some elements the requirements at the converter, but is not sufficient
for a quantitative analysis, especially of the crucial so-called light elements in ladle
metallurgy. Typical concentrations of carbon, phosphorus, and sulfur to be measured
at the ladle are within the range of 5–100 g=g. Spectrometrical analysis of these
elements is more challenging, because the most sensitive emission lines are in the
deep UV, or even the vacuum UV wavelength range, e.g., C 193.1, P 178.3, and
S 180.7 nm (cf. Sect. 13.1.1). An efficient transmission of this radiation from the
laser-induced plasma on the melt surface to the spectrometer is a key issue to achieve
high sensitivity.
Investigations of DP generated within a single flashlamp discharge of a Nd:YAG
laser for plasma excitation showed a significant improvement of the detection limits
[13.20]. With solid steel samples, an LOD for carbon of 90 g=g and for boron
of 6 g=g was obtained. The plasma emission was observed with a Paschen–
Runge spectrometer enabling the simultaneous detection of the line emissions of
15 elements covering a spectral range from 193.09 to 400.87 nm [13.2].
A further improvement of the LOD for light elements in solid steel samples was
achieved by a TP excitation and a VUV Paschen–Runge spectrometer observing
the plasma emission directly [13.4]. With this approach, detection limits for the
elements C, P, S, Mn, Ni, and Cr in low-alloy solid steel samples of less than
10 g=g were achieved for the first time using LIBS. An LOD for carbon in solid
steel samples observing an emission line at 97.70 nm of 1:2 g=g was reported
[13.13]; however, in this investigation an ambient pressure in the target–laser
13.1 Steel 247
Fig. 13.12 Schematic setup for LIBS of liquid steel. M D mirror; L D lens; W1, W2, W3 D
windows; MH D mirror with a hole; OP D optical probe; PL D permanent part of measuring
lance; T D tip of lance; MT D steel melt; P D platform; H D housing; ES D entrance slit of
spectrometer; A D vertical axis; V D vacuum pump; BT D beam tube
interaction region of 0.3 mbar was applied, which is not feasible for an inline
analysis of liquid steel.
Figure 13.12 shows the experimental setup [13.8]. Laser source, spectrometer,
and signal electronics are installed in a thermally stabilized housing. A Q-switched
Nd:YAG laser operating at 1,064 nm with a repetition rate of 10 Hz was used to
excite the plasma. The Q-switching electronics was modified to generate up to three
separated laser pulses within a single flashlamp pulse. The laser emits three equal-
energy pulses with a burst energy of 110–125 mJ [13.4]. The pulse width (FWHM)
of each pulse amounts to about 16 ns. The interpulse separation between the first
and the second pulse amounts to t1 D 25 s, and between the second and the third
pulse t2 D 42 s. The laser beam is guided via a mirror to an optical probe (OP
in Fig. 13.12) which contains the optical components to guide the laser beam and
the plasma radiation. The OP is a gas-tight tube which is evacuated by a vacuum
pump to a pressure of less than 0.1 mbar in order to achieve a high transmittance
for the UV emission lines below 200 nm (cf. Fig. 3.15). The laser beam passes an
entrance window (W3 in Fig. 13.12) of the OP and propagates from the rear side
through a hole in an aluminum-coated mirror (MH) to the focusing lens. The focal
length amounts to 216 mm (at 1,064 nm) and 165 mm (at 150 nm). The laser beam is
focused on the surface of the steel melt inside the tip of the lance. The light emitted
by the laser-induced plasma is collected by the focusing lens and propagates via
the mirror MH and a concave aluminum-coated mirror (not shown in Fig. 13.12) to
the entrance of the Paschen–Runge VUV spectrometer. The windows W1, W2, and
the lens L in Fig. 13.12 were made of MgF2 for UV transmittance (cf. Fig. 4.11).
The overall path length between the spectrometer entrance slit and the melt surface
amounts to about 2 m.
with a holographic diffraction grating with 2,400 lines/mm. Further details of the
spectrometer, the observed emission lines, and the signal electronics are described
in Sect. 13.1.1. Each photomultiplier signal corresponding to the irradiation of a
triple laser pulse burst is integrated using three integration time gates each with a
width of tint D 12 s. Each gate is delayed by tdelay D 900 ns to the respective laser
pulse of the burst. The sum of the three integrated parts is digitized for each laser
burst yielding the intensity value of the respective emission line for one laser burst.
For each measurement, the integrated signals of 600 laser bursts are accumulated
corresponding to a measuring time of 1 min.
The OP forms the inner part of a lance, which protects the optical components
from heat radiation originating from the induction furnace and the steel melt. The
lance was developed at the Institute of Ferrous Metallurgy (IEHK), University
Aachen, Germany within the frame of the SELA project [13.21]. The lance consists
of a permanent lance (PL in Fig. 13.12) manufactured of copper and a lance tip
(T in Fig. 13.12) made of an Al2 O3 refractory, which is immersed into the steel
melt. The refractory material was optimized to reduce any disturbing outgassing
during immersion of the tip in the melt especially with respect to the interesting
analytes carbon, phosphorus, and sulfur. The permanent part is water cooled to
keep the temperature inside the lance below 150ıC to avoid damage of the optical
components.
The whole measuring equipment is installed on a platform which can be lifted
by a hydraulic system in the vertical direction (see P and A in Fig. 13.12). For a
measurement, the lance tip is lowered step by step with the lifting apparatus toward
the steel melt produced in an induction furnace. The liquid runs into the tip of the
lance and forms a melt level. The vertical position of the lance is adjusted while
observing the iron signal of the laser-induced plasma with the VUV spectrometer.
At the optimum vertical position and thus the position of the melt level with respect
to the focal position of the laser beam, the signal height of the iron emission attains
a maximum. The estimated position of the beam waist is about 5–10 mm below
the melt surface. The immersion depth of the tip in the steel melt amounts to
about 20 cm.
Figure 13.13 shows a detailed view of the lower part of the lance and the tip
with the flow directions of cooling water (WF in Fig. 13.13) and argon gas. Argon
flushing of the volume inside the lance tip is necessary to protect the window of
the OP from dust and melt splashes and to maintain a defined UV transmitting
atmosphere in the interaction region. The selected argon purity was 4.8 (purity
99.998%); the flushing rate after immersion of the tip into the melt amounts to
18 l/min.
Based on the setup shown in Fig. 13.12, an analyzing device was designed and
operated in a melting shop for evaluation. The laser analyzer with laser source,
spectrometer, OP, and signal evaluation has been installed on the platform and the
lifting device. Assessment of the analyzer was performed by calibrating the setup
13.1 Steel 249
Fig. 13.13 Schematic view

of the lower part of the lance.
LB D laser beam; G D gas
flow. Further abbreviations:
see Fig. 13.12
for the elements carbon, phosphorus, sulfur, nickel, and chromium at a 100-kg
induction furnace. The initial material Armco ingot iron was melted under an inert
gas atmosphere. The Armco ingot iron which is practically the purest steel mill
produced iron has a carbon content of about 80 g=g.
Before starting the measurements in liquid steel, the adjustment of the laser
analyzer was verified by the measurement of solid steel samples. The samples were
positioned inside the lance probe in such a manner that the sample surface was
approximately at the same position as the melt surface during melt analysis. A pure
iron sample (e.g., Hu 85) and a low-alloy sample (e.g., BAS 456) were measured.
Precipitations on the surface of the entrance window or misalignment of the optical
setup due to vibrations during craning and lifting can be determined by this method.
The entrance window (W2 in Figs. 13.12 and 13.13) was cleaned approximately
every 5 h and replaced after approximately 20 h of accumulated immersion time.
The optical setup had to be realigned after approximately 15 h of operation. Before
the immersion of the lance the steel melt is deoxidized. Due to the enhanced
formation of fumes, the gas-flow rate of Ar 4.8 is increased to 30 l/min during the
immersion process. After reaching the required immersion depth the gas-flow rate
is reduced to 18 l/min. After melting, alloying, and homogenization steps reference
samples are taken from the melt and analyzed after solidification by conventional
methods.
During melt analysis, one or more different materials were alloyed in steps of
approximately 200 g=g. After the alloying process, the steel melt is homogenized
and the melt volume inside the lance tip is exchanged by lifting and lowering of
the lance. The measuring time was 1 min at 10 Hz repetition rate for all elements.
Altogether, measurements of 19 different steel melts were carried out at the
100-kg induction furnace. The average immersion time of a typical data collection
experiment amounts to approximately 4 h.
Fig. 13.14 Calibration curve 10

of carbon in liquid steel
Fe 187.75 nm
8
/ C 193.09 nm / /
4
0
0 50 100 150 200 250 300
carbon concentration [μg /g]
Figure 13.14 shows the calibration curve for carbon in liquid steel. The intensity
ratio of the carbon line at 193.09 nm and the iron line at 187.75 nm is plotted as
a function of the carbon concentration. Alloying of carbon to the melt volume
of 100 kg was difficult to achieve since oxidization occurred and homogenization
was difficult to establish. Hence, only four data points could be taken. The element
concentration given on the abscissa is taken from the conventional analysis of the
samples taken after the alloying steps. The error bars are given by the empirical
standard deviation calculated from three to five measurements made after each
alloying step. The average relative standard deviation of the intensity ratio amounts
to approximately 2%. The LOD is estimated by fitting a linear function to the
data points using the relation 3s=a1 , where s is the standard deviation of the
intensity ratio measured at the lowest carbon concentration and a1 is the slope of
the linear function [13.12,13.22,13.23] (cf. (11.16)). Since the preconditions for the
determination of the LOD are not exactly fulfilled and this approach is based on an
extrapolation well below the lowest measured carbon concentration, the calculated
value for the LOD has to be considered as a rough estimate.
Calibrations were performed for the elements phosphorus, sulfur, nickel, and
chromium in liquid steel all measured simultaneously within the same series of
measurements. Figures 13.15 and 13.16 show exemplarily the calibration curves
for phosphorus and nickel. The LOD are estimated by a fit of a linear function to
the lower concentration range. These linear functions are intersecting the vertical
axis at a positive value of the respective intensity ratio. This intensity ratio can
be interpreted as the expected measuring value for a blank sample, where the
intensity of the analyte line approaches a background value. However, the blank
sample could not be realized experimentally in the given experimental setup for
the liquid steel measurements. The results in terms of the estimated LOD, BEC,
(cf. (11.19)), and the coefficient of determination r 2 are summarized in Table 13.8.
For the first time, it was demonstrated that LIBS is capable to analyze light elements
in liquid steel with estimated LODs in the range of 5–21 g=g [13.8]. This is
an improvement against the results reported before by more than one order of
magnitude [13.18, 13.6, 13.19, 13.7]. The results obtained for liquid steel are similar
to those gained at solid samples [13.4].
13.1 Steel 251
1.1
Fe 187.75 nm
1.0
/ P 178.28 nm / /
0.9
0.8
0.7
0 200 400 600 800 1000
phosphorus concentration [μg /g]
Fig. 13.15 Calibration curve of phosphorus in liquid steel
8
I Ni 231.60 nm / I Fe 187.75 nm
2
0 200 400 600 800 1000 1200 1400
nickel concentration [µg/g]
Fig. 13.16 Calibration curve of nickel in liquid steel
Table 13.8 Analyzed elements, estimated limits of detection, background equivalent concen-
tration, concentration range of calibration, concentration range of linear fit, and coefficient of
determination for the linear fitting curve for LIBS inline analysis of liquid steel
Element Estimated BEC (g=g) Concentration Concentration r2
LOD3s .g=g) range of calibration range of linear fit
(g=g) (g=g)
C 5 45 70–300 70–300 0.950
P 21 1,550 60–890 60–890 0.990
S 11 870 50–1,100 250–1,100 0.993
Ni 9 460 310–1,500 310–1,010 0.989
Cr 9 310 90–900 90–610 0.984
The results show that LIBS is capable to quantitatively analyze liquid steel with
high sensitivity. Measuring times can be further reduced down to about 6 s, e.g., by
the use of a Nd:YAG laser operating at 100 Hz. The results clearly demonstrate the
potential of LIBS to fulfill the requirements for the process integrated inline analysis
of light elements in secondary metallurgy.
13.2 High-Alloy Steel
High-alloy steel grades offer a wide range of material properties with respect to,
e.g., mechanical strength, corrosion resistance, or thermal and electric conductivity
tailored to the needs of usage in industrial applications, home appliances, and
medical technology. The constituents of high-alloy steel are mainly the elements
Fe, Cr, and Ni with a total content of about 95 m.-%. For process control during
steelmaking and for identification purposes, the chemical composition has to be
analyzed. The influence of the composition of high-alloy steel grades on the material
property is illustrated by the following example. There are nickel alloys, which
consist of nickel and iron and have melting temperatures of about 900 ıC, whereas
the melting temperature of iron and nickel alone is about 1;500 ı C.
For production control of high-alloy steelmaking processes, it is important to
analyze the composition of the steel melt, e.g., by taking samples. LIBS is a suitable
tool to determine the composition of high-alloy steel samples. Examples for LIBS
to analyze steel are given in [13.4, 13.6–13.14, 13.18–13.21, 13.24, 13.26].
LIBS investigations of high-alloy steel grades were carried out by several groups.
For example, the influence of different laser wavelengths on the accuracy and
precision of analysis has been studied in air at atmospheric pressure for the elements
Si, Ti, Nb, and Mo [13.11]. Palanco et al. have developed an instrument for a fast
quality assessment in the steel industry with sample handling, surface preparation,
and quantitative analysis [13.10]. The appearance of so-called matrix effects could
be reduced by a multivariate calibration. A system identifying high-alloy steel
grades of pipe fittings in routine industrial operation was developed [13.12]. LIBS
is able to analyze stainless steel samples at high temperatures in air at atmospheric
pressure [13.26]. In a remote LIBS approach, stainless steel has been analyzed at
distances between the plasma and the detection system of more than 10 m [13.25].
Bassiotis et al. obtained calibration curves for the elements Cr, Ni, and Mn in
stainless steel samples at atmospheric conditions using internal standardization
[13.27]. Experimental parameters have been optimized to improve the linearities
of the calibration curves over a broad range of concentrations.
In the following, a different experimental setup is described consisting of an
Ar-flushed sample stand and a Paschen–Runge spectrometer having a high spectral
resolution in a greater spectral range [13.28]. The influence of single and DP
excitation and different laser burst energies on the mean residual deviation for the
quantitative determination of concentrations is investigated.
The experimental setup is shown in Fig. 13.17. The laser beam generated by a
Q-switched Nd:YAG laser working at the fundamental wavelength is guided via
13.2 High-Alloy Steel 253
Fig. 13.17 Schematic view of the experimental setup. VS D vacuum spectrometer; SE D signal
electronics; L D laser; LB D laser beam; PR D Glan-laser prism; M D mirror; FL D focusing
lens; ST D argon-flushed sample stand; Ar D argon inlet; W1 D window of ST; P D laser-induced
plasma; FO D fiber optics; ES D echelle spectrometer; Ar OL D Ar outlet; ˛D D plasma detection
angle; SM D sample movement; OA D optical axis of VS; W2 D entrance window of VS;
E D entrance slit of VS
a Glan-laser prism and a mirror to a lens focusing the laser beam into an Ar-
flushed sample stand onto high-alloy steel samples. The samples are moved during
the measurement so that the laser beam propagates over the sample surface along
a circle with a diameter of 3 mm and a frequency of about 1 Hz (cf. Sect. 3.6,
Fig. 3.20). This leads to reproducible crater geometries over the whole measuring
process. The stationary case – i.e., no relative movement between laser beam and
sample surface – has been compared with the moving case yielding the result that
the progression of the element intensities with increasing laser pulse number is more
uniform.
The plasma light is detected via a direct light channel with a vacuum Paschen–
Runge spectrometer (Model RS 1000 from OBLF) without further optical imaging
and with a fiber optics and an echelle spectrometer (ESA 3000, LLA, Berlin,
Germany). In the Paschen–Runge spectrometer, the plasma light is dispersed
spectrally covering a spectral range from 178 to 406 nm (grating, 2,700 l/mm) with
a spectral resolution of 20 pm. The distance of the sample surface to the entrance
slit of the spectrometer is 300 mm. The plasma signals are detected under the
detection angle of ˛D 10ı between the optical axis of the spectrometer and
the sample surface. Twenty photomultipliers are located on the Rowland circle
(diameter 0.5 m) to detect 19 element lines and the zeroth order signal. The spectral
lines installed for the elements Al, C, Cr, Cu, Fe, Mn, Mo, Ni, and Si are those listed
already in Table 13.3. Additionally, the Ti line 337.280 nm with Ej D 3:687 eV,
log.gf / D 0:18 is detected. A signal electronics (MCI; cf. Sect. 4.5) integrates the
signals in programmable time windows for each generated plasma.
The parameters kept constant for the calibrations and the determination of the
plasma parameters are as follows: number of prepulses Npp D 50, number of
measuring pulses Nmp D 300, and flow rate of argon through the sample stand
20 l/min. The focal length of the plano-convex lens amounts to f D 200 mm and
the focal position is s D 5 mm within the sample to avoid a gas breakdown in
front of the sample surface.
The calibrations are carried out with the Paschen–Runge spectrometer in the SP
and DP mode each with the laser burst energies Eb D 100 and 160 mJ; i.e., Eb D
2 50 and 2 80 mJ for the DP (with pulse widths at FWHM of D 6 ns for SP and
D 10 ns for DP). In the DP mode, the interpulse separation is set to t D 30 s.
The signals are integrated with a time delay of tdelay D 2 s and an integration width
of tint D 10 s. Three repetitions on different sample locations are carried out on
each sample.
For determination of the plasma parameters, the echelle spectrometer is used
in combination with the Ar-flushed sample stand having an optical fiber port. The
echelle spectrometer can detect quasi-continuous spectra in the range from 200 to
780 nm with a resolution of about 20 pm. The distance of the plasma to the optical
fiber is about 60 mm. The following parameters are taken: SP with a burst energy
of Eb D 160 mJ, tdelay D 6, 10, 20, 30 s, and tint D 400 ns. Three high-alloy
samples and a pure iron sample are measured covering iron concentrations in the
range between 43 and 99 m.-%. Two repetitions on different sample locations are
evaluated for each sample.
Twenty-one certified high-alloy reference steel samples were analyzed.
Table 13.9 gives the reference concentrations of ten elements. The reference
samples are certified reference materials of high-alloy steel from BAS, National
Bureau of Standards (NBS, Washington D.C., USA), China National Analysis
Center for Iron and Steel (NCS, Beijing, China), Carpenter Technology (Crawley,
England), Metimex Reference Material (Pyskowice, Poland), MBH Analytical
LTD (Barnet, England), Brammer Standard Company (BSC, Houston, USA), and
Research Institute for Ferrous Metallurgy (Budapest, Hungary). Most of these
samples represent Cr-/Ni-steel grades. For all samples, the iron concentration is
greater than 42 m.-%, the chromium concentration is less than 30 m.-%, and the
nickel concentration is less than 35 m.-%. The concentrations of the minor elements
lie below 6 m.-% for Cu and for the other elements below 3 m.-%. The samples
were prepared by grinding with Al2 O3 -grinding paper, grain size 60. The high
matrix variation complicates the LIBS analysis as reported in [13.10]. Possible
reasons can be spectral interferences from the elements nickel, iron, and chromium
and the wide range of material properties, causing so-called matrix effects.
The calibrations are evaluated both for absolute and standardized intensities. For
the latter case, the analyte intensities are divided by the intensity of the Fe 271.4 nm
line, used as internal standard (cf. Sect. 11.1). The concentration c or concentration
ratios cO D canalyte =creference are functions of the corresponding intensities I or
intensity ratios Q D Ianalyte =Ireference . For the analysis functions, the mean intensity
values hI i; hQi averaged over three repetitions and second-order regressions are
evaluated; see (13.2) and (13.3):
ci D a0;i C a1;i hIi i C a2;i hIi i2 ; (13.2)
cOi D b0;i C b1;i hQi i C b2;i hQi i2 ; (13.3)

Table 13.9 Composition of the reference samples of high-alloy steel in m.-%

Reference Fe Cr Ni C Si Mn Mo Cu Al Ti
sample
CRM S 15 74.0 16.70 3.90 0.043 0.260 0.380 2.460 1.540 n.s. n.s.
CRM S 19 76.1 7.00 12.80 0.260 2.320 0.320 0.110 0.190 n.s. 0.048
NBS SRM C 42.8 19.55 29.30 0.056 0.410 0.830 2.830 3.720 0.003 0.012
1288
NCS HS 71.6 15.36 10.35 0.061 0.663 0.644 0.620 0.152 0.044 0.072
23702–1
NCS HS 72.7 12.46 11.52 0.164 0.413 0.981 0.350 0.363 0.074 0.210
23702–2
NCS HS 71.9 9.32 13.12 0.241 0.960 1.980 0.690 0.276 0.038 0.510
23702–3
NCS HS 70.7 7.40 16.41 0.184 0.920 1.570 0.500 0.303 0.240 1.030
23702–4
NCS HS 70.7 17.16 9.05 0.096 0.654 0.730 0.260 0.126 0.260 0.570
23702–5
NCS HS 71.4 20.76 6.93 0.017 0.171 0.156 0.079 0.049 0.034 0.310
23702–6
NCS HS 70.3 18.90 7.82 0.101 1.080 0.356 0.170 0.098 0.200 0.710
23702–7
ECISS CRM 55.4 14.64 24.68 0.049 0.531 1.016 1.102 0.000 0.199 2.010
289–1
Metimex RM 45.2 15.57 35.23 0.230 1.950 1.750 n.s. 0.000 n.s. n.s.
MW 27
Metimex RM 58.9 25.00 12.26 0.250 1.100 1.550 0.340 0.050 n.s. n.s.
MW 37
BSC RM B.S. 75.4 12.66 8.34 0.035 0.080 0.060 2.200 0.041 1.000 0.051
184A
MBH RM PH 1 73.3 16.10 5.10 0.094 0.180 1.750 0.160 3.120 n.s. n.s.
MBH RM PH 2 72.8 16.70 3.79 0.070 0.510 0.890 0.940 4.110 n.s. n.s.
MBH RM PH 3 71.6 15.13 3.16 0.150 1.940 0.440 0.780 6.280 n.s. n.s.
BSC RM B.S. 67.7 17.09 10.03 0.049 0.550 1.690 0.590 0.450 0.130 0.480
85D
BSC RM B.S. 61.1 25.45 5.84 0.017 0.440 1.040 3.240 1.940 0.009 0.006
179A
Carpenter RM 58.3 29.80 8.93 0.126 0.460 1.600 0.220 0.150 0.010 0.110
CT 312
BAS CRM 93.6 3.03 0.53 0.490 0.660 0.195 0.830 0.397 0.040 n.s.
407/2
Mean 67.9 16.0 11.4 0.1 0.8 0.9 0.9 1.1 0.2 0.4
n.s., concentration value is not specified
where a0;i ; a1;i ; a2;i ; b0;i ; b1;i , and b2;i are the regression coefficients of the second-
order regressions of element i and hIi i; hQi i is the mean intensity value, mean
intensity ratio of element i averaged over three measurements at one sample.
A measurement on one sample location comprises Nmp D 300 laser bursts.
concentration c Cu [m.- %]
6 single pulse, Eb = 160 mJ
5
4
3
2
1 r ² = 0.99996
0
0 500 1000 1500 2000
intensity I Cu 324.75 nm [counts]
0.02
(c Cu, LIBS - c Cu, reference) [m.- %]
single pulse, Eb = 160 mJ
0.01
0.00
-0.01
-0.02
0 1 2 3 4 5 6 7
concentration c Cu, reference [m.- %]
Fig. 13.18 Top: Analysis function for copper in the single pulse mode. RSD.I / D 0:9%.
R.Cu/ D 0:011 m.-%. Bottom: Absolute difference between the concentration measured with
LIBS and the reference concentration versus the reference concentration
As a figure of merit for the accuracy, the mean residual deviations of ten elements
R.i / are determined; see relation (11.21) in Sect. 11.3. The precision is defined
as the mean relative standard deviations of the intensity ratios of three replicate
measurements; see relation (11.22).
Analysis functions for the element chromium were already shown in Fig. 11.5.
Obviously, the differences between the DP and SP calibrations are greater than the
differences concerning the different laser burst energies. Whereas the calibrations
with DP show greater saturation effects than the corresponding SP calibrations, the
sensitivity in the low concentration range is higher than the one for SP. A similar
result was found for low-alloy steel samples, where detection limits can be improved
by use of DP [13.12]. The ablation rates are higher for DP than for SP of the
same laser energy [13.20]. The same behavior can be seen for the element lines of
nickel, copper, manganese, silicon, and titanium. These elements are characterized
by concentrations higher than 1 m.-%.
Figure 13.18 shows the copper analysis function using absolute intensities.
Copper is an element with a small residual value of R.Cu/ D 0:011 m.-% in the SP
Table 13.10 Mean residual deviations R.i / of ten element lines for two different laser burst
modes and two different laser burst energies and ratio of residuals (DP to SP) using absolute
intensities
Element Samplesa c (m.-%) R.i /SP (m.-%) R.i /DP (m.-%) hR.i /DP i=hR.i /SP i
100 mJ 160 mJ 100 mJ 160 mJ
Ni 21 34.7 0.39 0.37 0.87 1.59 3.2
Cr 267 21 29.7 0.63 0.62 1.45 2.09 2.8
Cr 286 21 29.7 0.50 0.57 1.12 1.63 2.6
Cu 19 6.2 0.038 0.011 0.143 0.127 6.0
Mo 20 3.2 0.055 0.050 0.086 0.127 2.0
Si 21 2.2 0.129 0.101 0.051 0.070 0.5
Ti 14 2.0 0.042 0.048 0.019 0.031 0.6
Mn 21 1.9 0.13 0.12 0.07 0.11 0.7
Al 14 1.0 0.019 0.029 0.018 0.023 0.8
C 21 0.5 0.020 0.023 0.014 0.018 0.7
c, concentration range; SP, single pulse; DP, double pulses
a
For some elements not all of the 21 available reference samples are used, if the concentration is
not specified for these elements
mode and for Eb D 160 mJ. Concerning the mean copper concentration of 1.1 m.-%,
this corresponds to a relative accuracy of 1.0%.
Table 13.10 lists the mean residual deviations R.i / of ten element lines. The
elements are sorted by their concentration range c. For chromium two element
lines, Cr 267.72 and Cr 286.25 nm are evaluated. Because not all of the reference
samples have for all elements specified concentration values, the number of regarded
samples varies. In the SP mode for most of the elements except for Cu, there is
no significant difference between the calibrations for the two laser burst energies.
The residuals in the DP mode are significantly greater especially for the matrix
elements Cr and Ni but also for the minor elements Cu and Mo. The mean residual
values for each element in the SP mode are compared with those in the DP mode.
Considering the four different parameter sets SP lead to smaller residuals for the
element lines Ni 225.38, Cr 267.72, Cr 286.26, Cu 324.75, and Mo 281.62 nm. The
ratio hR.i /DP i=hR.i /SP i is 2. The chosen DP parameters show better results for
the elements Si, Ti, Mn, Al, and C by a factor of about 0.5–0.8. Comparing the four
calibration parameter sets, there is no parameter set that is the best for all elements.
Besides the evaluation with absolute intensities, calibrations are carried out
by dividing the absolute analyte intensities to the Fe 271.44 nm-line intensity as
an internal standard (cf. Sect. 11.1). By measuring all the elements that can be
contained in the steel samples, the real content of the elements can be calculated
in m.-% [13.29]. Figures 13.19, top, and 13.20 show analysis functions for the
elements Ni and Si. The coefficients of determination are close to one which results
in good accuracies; see Table 13.11.
Figure 13.19, top, represents the analysis function for nickel in the SP mode for
Eb D 160 mJ, which yields the smallest R(Ni)-value and the smallest RSD(hQNi i/-
concentration ratio Ni / Fe [1]

0.8 single pulse, Eb = 160 mJ
0.6
0.4
0.2
r 2 = 0.9999
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
intensity ratio I Ni 225.39 nm /I Fe 271.44 nm [1]
(c Ni, LIBS - c Ni, reference) [m.- %]
0.3 single pulse, Eb = 160 mJ

0.2 Δc (Ni) = 35 m.- %
0.1
0.0
-0.1
-0.2
-0.3
0 5 10 15 20 25 30 35
concentration c Ni, reference [m.- %]
Fig. 13.19 Top: Analysis function for nickel in the single pulse mode Bottom: Absolute difference
between the LIBS concentration and the reference concentration versus the reference concentration
0.05
concentration ratio cSi / cFe [1]
double pulse, E b = 100 mJ

0.04 Dc(Si) = 2.2 m.-%
0.03
0.02
0.01
r ² = 0.997
0.00
0.2 0.4 0.6 0.8 1.0 1.2
intensity ratioISi 251.61 nm / IFe 271.44 nm [1]
Fig. 13.20 Analysis function for silicon in the double pulse mode
value. The mean residual deviation amounts to 0.13 m.-%. Relating to the mean
nickel concentration value of 11.1 m.-% this means a relative accuracy of 1.2%.
The precision of three replicate measurements averaged over all samples is 0.3%.
Figure 13.19, bottom, illustrates the R(Ni)-values in a different plot to visualize the
Table 13.11 Mean residual deviations R.i / of ten element lines for two different laser burst
modes and two different laser burst energies using internal standardization
Element R.i /SP (m.-%) R.i /DP (m.-%) hR.i /DP i=hR.i /SP i
100 mJ 160 mJ 100 mJ 160 mJ
Ni 0.17 0.13 0.37 0.38 2.5
Cr 267 1.04 0.91 1.04 1.64 1.4
Cr 286 0.50 0.48 0.61 0.66 1.3
Cu 0.060 0.042 0.251 0.337 5.7
Mo 0.040 0.038 0.055 0.038 1.2
Si 0.150 0.112 0.041 0.058 0.4
Ti 0.063 0.065 0.018 0.020 0.3
Mn 0.144 0.134 0.085 0.050 0.5
Al 0.031 0.042 0.031 0.050 1.1
C 0.020 0.025 0.012 0.016 0.6
Same sample set as in Table 13.10
differences of the measured LIBS concentrations to the reference concentrations.

Most of the residuals amount to less than 0.1 m.-%, but all are less than 0.3 m.-%.
In contrast to nickel silicon is an element for which DP of 100 mJ burst energy
have the smallest R(Si)-values comparing the four parameter sets; see Fig. 13.20.
The residual amounts to R.Si/ D 0:041 m.-% with r 2 D 0:997. The precision is
RSD(hQSi i/ D 1:4%.
Table 13.11 summarizes the results of the calibrations using internal standard-
ization. The ratios hR.i /DP i=hR.i /SP i for all regarded elements except for Al
exhibit the same tendencies (>1:0 or <1:0) as for absolute intensities. There is an
improvement of R.i / for Ni by a factor of about 3 achieving 0.13 m.-% for SP with
Eb D 160 mJ. In the DP mode, the internal standardization leads to a significant
reduction of the R.i /-values of Ni, Cr, and Mo for both laser burst energies by
factors between 1.4 and 4.2. The hR.i /DP i-values of the elements Si, Ti, Mn, and C
are in the same order of magnitude. In comparison to the calibrations with absolute
intensities, the residuals are improved for six element lines (Ni, Cr 286, Mo, Si, Mn,
and C) by about 50% comparing the smallest R.i /-values for the four parameter
sets studied.
Table 13.12 lists the precision of the intensity ratios of three replicate measure-
ments in terms of RSD(hQi i/ for ten element lines. The precision for the elements
Ni, Cr, Cu, and Mo is in the range from 0.3% to 1.2%. In contrast to that the
RSD values of the other elements cover a range from 1.1% to 6.2%. Possible
reasons for this behavior are the different properties of the analyte element lines
compared to the internal standard line Fe 271.4 nm. The greatest difference between
the upper energy levels of the analyte line and the reference line occurs for the
element Al with En D 2:4 eV (cf. Sect. 11.1, relation (11.3) and discussion
of homologous line pairs). Furthermore, the spectral distance is high with œ D
396:152 nm 271:441 nm D 124:711 nm. These facts lead probably to the highest
RSD values for Al.
Table 13.12 Relative RSD.hQi i/SP (%) RSD.hQi i/DP (%)

standard deviations
RSD(hQi i) in % for nine 100 mJ 160 mJ 100 mJ 160 mJ
elements using ten element Ni 0.3 0.3 0.4 0.4
lines and internal Cr 267 0.7 0.4 0.7 0.4
standardization Cr 286 0.6 0.5 0.4 0.3
Cu 0.7 0.7 1.2 1.1
Mo 0.7 0.6 0.8 0.9
Si 3.8 2.5 1.4 1.1
Ti 3.5 3.0 2.6 2.1
Mn 2.3 1.6 0.8 0.6
Al 5.8 5.3 6.2 2.4
C 3.3 2.1 4.1 1.9
Bassiotis et al. determined the RSD values for Mn for eight high-alloy steel
samples [13.27]. Five sample positions each irradiated with 150 laser pulses were
measured. With the setup described here averaging over the RSD values of 20
samples, the mean values amount to 1.6% for SP and for Eb D 160 mJ and 0.6%
for DP and for Eb D 160 mJ. Compared to the 3.1% reported in [13.27], this is a
reduction by a factor of 2 for SP and by a factor of 5 for DP.
As already indicated in Figs. 11.5 and 11.6, the chromium analysis function
shows increasing residual values for high Cr concentrations. Deviations observed at
high concentrations are significant because of the high precision of the measured
intensity ratios (0:7% cf.; Table 13.12). In other studies, it has been shown
that multivariate calibrations can reduce the residual values [13.10]. In contrast
to multivariate calibrations interelement correction is an iterative method. The
procedure of interelement correction is described in Sect. 11.5. Table 13.13 lists all
multiplicatively and additively corrected residual values R.i / for all four parameter
sets and for ten element lines. In the column “disturbing elements” those elements
are listed which have significant influence to reduce the residuals. The sign “*”
means that this element is used for a multiplicative correction, whereas the sign “C”
means an additive correction. In the last four columns, the factors of improvement
are listed. By interelement corrections, the mean residual deviations R.i / can be
reduced on average for all ten element lines by a factor of 1.7–2.1. The greatest
improvement is achieved for Cu by a factor of about 4 in comparison to the
uncorrected residuals (Eb D 160 mJ, SP).
The question is whether the residuals can be explained with differences in the
plasma parameters due to the wide range of material properties of the samples.
Hence, the plasma parameters electron density and electron temperature are deter-
mined measuring samples with a high variation of the iron, chromium, and nickel
content. The plasma parameters are determined for the plasmas of four samples
with a high difference concerning the iron concentration. The plasma emission is
detected with an echelle spectrometer. Four time delays tdelay D 6, 10, 20, and 30 s
are studied. The spectra for the burst energy Eb D 160 mJ in the SP mode are
evaluated. The electron densities are calculated by the Stark broadening of the Fe I-
line 492.5 nm [cf. Sect. 9.4, (9.19)]. Three high-alloy steel samples and one pure
Table 13.13 Additive and multiplicative interelement corrections for ten elements
Element Disturbing R.i /SP;corr (m.-%) R.i /DP;corr (m.-%) Factor of improvement a
elements SP DP
100 mJ 160 mJ 100 mJ 160 mJ 100 mJ 160 mJ 100 mJ 160 mJ

Ni Cr, Mo, 0.07 0.08 0.35 0.31 2.7 1.6 1.1 1.2

Ti, C

Cr 267 Ni, Ti 0.31 0.30 0.29 0.81 3.4 3.0 3.6 2.0

Cr 286 Ni, Ti 0.33 0.32 0.34 0.37 1.5 1.5 1.8 1.8

Cu Si, Mo 0.017 0.010 0.086 0.142 3.5 4.2 2.9 2.4
Mo CCr, Ni, 0.021 0.020 0.041 0.026 1.9 1.9 1.3 1.5

Cu,

Ti
Si CCr, Ti 0.112 0.082 0.043 0.053 1.3 1.4 0.9 1.1

Ti Si 0.051 0.051 0.018 0.020 1.2 1.3 1.0 1.0

Mn Ti 0.085 0.085 0.084 0.042 1.7 1.6 1.0 1.2

Al Ni, CMo, 0.011 0.011 0.011 0.017 2.8 3.9 2.8 2.9

Si

C Si 0.020 0.024 0.007 0.009 1.0 1.0 1.8 1.8
Mean 2.1 2.1 1.8 1.7
Corrected R.i / in m.-% and factors of improvement. Same sample set as in Table 13.9
a
Factor of improvement D R.i /uncorr =R.i /corr , where R.i /uncorr are the values from Table 13.11
6
single pulse,
Eb = 160 mJ
5 43 m.-% Fe
59 m.-% Fe
N e [1016 cm- 3 ]
75 m.-% Fe
4 99 m.-% Fe
2
5 10 15 20 25 30
tdelay [μs]
Fig. 13.21 Electron densities Ne for four time delays and four different samples showing a high
matrix variation, Fe I-line 492.5 nm. Samples: C 1,288 (43 m.-% Fe), MW 37 (59 m.-% Fe),
B.S. 184 (75 m.-%Fe), pure iron (99.99 m.-% Fe)
iron sample are measured covering iron concentrations in the range between 43 and
99 m.-%. Figure 13.21 shows the determined electron density Ne as a function of
the delay time. The errors of electron densities Ne are between 4% and 10%. No
significant difference of electron densities between the samples is observed.
Table 13.14 Line selection for the Boltzmann plot

No. (nm) log.gm fmn / En (eV) No. (nm) log.gm fmn / En (eV)
1 273.36 0.06 5.39 12 407.17 0.022 4.65
2 278.81 0.02 5.30 13 411.85 0.28 6.58
3 296.69 0.404 4.18 14 413.47 0.49 5.83
4 299.95 0.47 4.99 15 419.91 0.25 6.00
5 356.54 0.19 4.43 16 421.94 0.12 6.51
6 363.15 0.036 4.37 17 427.18 0.164 4.39
7 364.78 0.194 4.31 18 430.79 0.07 4.43
8 371.99 0.431 3.33 19 432.58 0.01 4.47
9 374.95 0.161 4.22 20 438.35 0.2 4.31
10 404.58 0.28 4.55 21 522.72 0.969 3.93
11 406.36 0.07 4.61
21 atomic iron lines from 273 to 523 nm
Besides the electron densities, the electron temperatures are determined for
atomic lines of Fe. The line intensities are measured assuming that local thermo-
dynamic equilibrium (LTE) conditions are satisfied in the plasma (cf. Sect. 8.5).
The echelle spectrometer offers a wide spectral range for the line selection of the
Boltzmann plot; 21 Fe I-lines are evaluated; see Table 13.14 (cf. also Sect. 19.6). As
intensities for the Boltzmann plot, the line shapes are fitted by Gaussian and Lorentz
profiles and from this fit the peak heights are used. Figure 13.22, top, shows a typical
Boltzmann plot for the time delay tdelay D 20 s and for three different high-alloy
steel samples.
The transition properties of the lines are taken from the Kurucz data base [13.30].
Despite the high variation of the iron concentration the slopes are similar. The errors
of the slopes are taken as the error of the electron temperatures. The significance for
the evaluation is enlarged by regarding many lines. These lines are chosen because
in the concentration range between 43 and 100 m.-% there is a linear relationship of
the intensities versus the concentration. So self-absorption effects can be neglected.
The upper energy levels are nearly uniformly distributed from En D 3:4 eV to about
6.5 eV.
The electron temperatures are evaluated at three time delays: tdelay D 6; 20,
and 30 s. The Fe I electron temperature decays from Te D 10;500 K after
6 s to Te D 8;000 K after 30s. The value for tdelay D 6 s is comparable
with temperatures reported so far [13.31]. But the values at later times lie about
1,000 K higher. The errors of the electron temperatures are between 3% and
5%. No significant difference between the samples is observed. So the origin for
matrix effects – as, e.g., observed for Cr – does not show up in a significant way
looking at the electron densities and electron temperatures. The reason for this
might be that the effect of improvement by interelement corrections is in the same
order of magnitude as the error of determination of the plasma parameters. For
example, the Cr 267.72 nm residual can be reduced by 0.6 m.-% from 0.91 m.-%
(see Table 3.11) down to 0.30 m.-% (see Table 13.11). Relating the value of the
absolute improvement 0.6 m.-% to the mean Cr concentration of 16.0 m.-% of the
measured sample set, this corresponds to an improvement by about 4%.
25
Fe I
24 t delay = 20 μs
43 m.-% Fe
23 59 m.-% Fe
ln(Il3) / (gf )
75 m.-% Fe
22
21
20
19
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
E n [eV]
11000 Fe I
43 m.-% Fe
10000 59 m.-% Fe
75 m.-% Fe
9000
Te [K]
8000
7000
6000
5 10 15 20 25 30
t delay [μs]
Fig. 13.22 Top: Boltzmann plot with 21 Fe I-lines. Relative error of the slopes about 3–5% [The
argument of the ln-function is the same as in Fig. 8.15; however, instead of Anm gn the term gm fmn
is taken, see (9.8)]. Bottom: Electron temperatures Te for three time delays and three high-alloy
steel samples
Calibrations were carried out for nine elements with single and double laser
bursts and two different laser burst energies. The residuals are dependent on the
laser burst mode. SP calibrations have the smallest residuals for the matrix elements
Cr and Ni and for the minor elements Cu and Mo, and DP for the elements Si, Ti,
Mn, and C. With the presented setup, precisions of intensity ratios are achieved in
terms of RSD values lower than 1.0% for most analytes. Interelement corrections
can improve the LIBS analysis on high-alloy steel samples on average by a factor of
two for matrix and minor elements. As a trend interelement corrections reduce the
residuals stronger for SP than for DP. Plasma parameters such as electron density
and electron temperature show no significant difference for a strong variation of
the iron concentration in the range of 40–75 m.-%. Maybe changes of the plasma
state exist but cannot be detected experimentally taking into account that the relative
errors of determination of the plasma parameters are of the same order of magnitude
as the interelement corrections.
13.3 Aluminum
Aluminum is the most important metal after steel [13.32]. Aluminum production
and use has increased worldwide by more than a factor of 10 between the years 1957
and 2003 achieving a level of 32 million tons per year [13.33–13.37]. About 25% is
secondary aluminum, i.e., recycled aluminum [13.38]. In Germany, since 2003 the
secondary aluminum production is even greater than the primary production. The
reason for this development is the high energy consumption necessary to produce
primary aluminum. For the production of 1 ton of aluminum about 170 GJ primary
energy is required. Nearly 80% or this energy is needed for the electrolytic process
[13.32, 13.39, 13.40]. In contrast to this secondary aluminum production allows to
save between 88% and 95% of this primary energy per ton [13.32, 13.39]. This is
an economic and ecological advantage since resources are saved; the charging of
dumping grounds for residual materials and environmental pollution by emissions
and wastes can be reduced. Due to the growing demand for aluminum products a
further increase of recyclable aluminum scrap is expected in a medium to long-term
perspective and hence secondary aluminum production will become more and more
important.
A precondition for high-quality secondary aluminum production by remelters
and refiners is the availability of shredded aluminum scrap sorted into fractions of
wrought and cast aluminum alloys. A mixture of wrought and cast alloys would
result in higher contents of alloying elements such as silicon and copper, which
restricts the reuse to cast alloys only, since the separation of these interfering
elements is barely possible or requires elaborate metallurgical processes. Hence for
the production of wrought alloys from secondary aluminum, the aluminum scrap
has to be melted with primary aluminum or has to be sorted [13.41].
Conventional analyzing and sorting techniques do not permit an economic
separation of light metal alloys and their alloy groups. The task is to separate
particularly aluminum alloys in fractions of cast and wrought alloys as well as
in their different alloy subgroups under industrial conditions with a throughput of
2–4 t/h.
The methodical approach for this task is based on a combination of image
processing, laser-based geometry detection, and LIBS. By means of the optical
and geometrical features of the individual scrap particles, moving at a speed of
3 m/s on a conveyor belt, a pulsed laser beam is guided by a 3D-scan unit to the
scrap particles and the chemical composition is determined by LIBS. Based on all
13.3 Aluminum 265
measured quantities, the individual scrap particles are classified and automatically
separated into two or more fractions. The pilot system developed for single particle
aluminum scrap sorting based on this approach will be presented in Sect. 18.4. In
the following, the combination of LIBS with a scanner and the analysis of static and
moving aluminum samples is described.
The experimental setup is shown in Fig. 13.23 [13.42]. The laser beam can be
directed onto selected surface positions by an xy-scanning unit with a field size of
about 400 400 mm2 . An autofocus unit consisting of a fast movable diverging and
a fixed focusing lens with a focal length of f D 1 m in neutral position allows
rapid changes of the focal point within a total range of 100 mm along the optical
axis, so that the focal plane can be adjusted to different heights of the samples. In
order to realize an economic process, the throughput of identified and sorted scrap
particles has to be at least 2–4 t/h. Considering the speed of the conveyor belt (2–
4 m/s) and the average mass of the particles, about 50 particles per second have to
be identified. For this reason, a high-repetition Nd:YAG laser system with 50 Hz
repetition rate and a Paschen-Runge spectrometer equipped with photomultipliers
are used. The measuring rate allows for one to seven laser bursts per scrap particle
for material identification. The plasma emission counterpropagating to the incoming
optical axis is directed onto a dichroic mirror located on the laser beam axis between
the xy-scanning unit and the autofocus telescope at an angle of 45 degrees. The
emission is deflected by this mirror perpendicularly to the laser beam axis to a
quartz lens .f value 2/ and then imaged onto the input aperture of an optical fiber
coupled to the spectrometer. For the determination of spectral lines, spectroscopic
investigations on different kinds of light metal alloys were carried out. A selection
of 20 lines has been fixed to detect the elements Al, Si, Cu, Fe, Mg, Mn, Sn, Ni,
Cr, Zn, Ti, Zr, Na, Ca, C, Cd, and Pb simultaneously. The signals are digitized and
evaluated in real time. Reference samples of certified aluminum alloys with known
compositions are analyzed statically and while moving on a belt conveyor at 3 m/s
by LIBS.
With this setup, calibration measurements were carried out with static and
moving certified reference samples. In the static case, the samples were measured
at the center of the scan field (pos. A in Fig. 13.23). At each of four different
sample positions, 25 measuring laser bursts were irradiated (cf. Sect. 11.1, K D 4,
Nmp D 25). Each burst consisted of a DP with a total burst energy of Eb D 170 mJ,
a pulse energy ratio of E1 W E2 D 1 W 2, and a temporal interpulse separation of
t D 15 s. In order to measure at moving particles, the samples are put centrally
onto the conveyor belt moving along the y-direction in Fig. 13.23. A 3D-geometry
detection unit acquires the temporal (y D y.t/) as well as the lateral sample position
x and transmits the calculated target position for the laser focus to the scan head
in real time. In contrast to the static case, the measurements at moving particles
comprise Nmp D 7 measuring laser bursts at one sample position tracked by the
scanning mirrors at a conveyor belt velocity of 3 m/s. The burst energy was set to
Eb D 190 mJ with E1 W E2 D 1 W 1 and t D 40 s.
The measurements were carried out with 36 samples of certified aluminum
alloys. This set of samples covers various concentration ranges of the major alloying
Fig. 13.23 Experimental setup for LIBS via an xy-scanner. L1L3 D lenses; DM D dichroic
mirror; SM1, SM2 D xy-scanning mirrors; FO D fiber optic to spectrometer; SF D scan field;
S D sample. The conveyor belt moves along the negative y-direction
Table 13.15 List of analyte Element (nm) cmin (m.-%) cmax (m.-%)
emission lines and
concentration ranges covered Si I 288.16/390.55 0.019 13.03
by the reference samples Fe II/Fe I 259.94/371.99 0.0167 4.93
Cu I 296.12/327.40 0.0006 12.2
Mn II 293.31 0.000004 10.8
Mg II 279.08 0.000062 5.1
Cr II 267.72 0.000002 0.37
Ni I 341.48 0.0007 2.03
Ti II 337.28 0.00034 0.26
Zn I 334.50 0.000008 11.6
Al I 305.71 80:0642a 99:99918a
a
Concentration values for AlPare not certified and calculated
with: c.Al/ D 100 m:-% ccertified elements
elements Si, Fe, Cu, Mn, Mg, Cr, Ni, Ti, and Zn. Table 13.15 shows a list of these
concentration ranges together with their corresponding analyte lines.
The goal of a calibration for sorting applications is to achieve the best possible
accuracy and selectivity over a wide concentration range up to 15 or 20 m.-% with
a few laser pulses. Hence as characteristic numbers of the analytical performance
besides the LOD, the relative standard deviations of procedure (RSDP) will be
considered [cf. Sect. 11.2, (11.13)].
13.3 Aluminum 267
The LOD is the threshold concentration at which the existence of a component

can be confirmed with a defined level of significance ˛ (e.g., ˛ D 0:01). According
to the linear calibration method, LOD values can be estimated by [13.23]:
LODestim: D 1:2ˆK;˛ sco ; (13.4)
where ˆK;˛ denotes a factor depending on the number of measurements K (sample

positions) and sco the standard deviation of procedure (in (11.16) the so-called
blank value method was applied, whereas the relation given here refers to the
linear calibration method). The linear calibration function I.c/ D a0 C a1 c is
determined by pairs of mean values of the measured Intensity INi D hhIi iNmp iK and
certified concentrations ci .i D 1; : : :; Ns;i I Ns;i D number of samples used for the
calibration function). With the residue standard deviation sI;c the standard deviation
of procedure sco is given by
v
uN
uP s;i
u .I.ci / INi /2
sI;c u
1 t i D1
sco D D : (13.5)
a1 a1 Ns;i 2
In case of a quadratic regression function, the denominator of the root is replaced by

Ns;i 3. For K D 4 measurements per sample, the LODestim: can be derived by use
of ˆ4;0:01 D 5:1. Table 13.16 summarizes estimated values for LODs in comparison
to other publications [13.43–13.46]. LODmin = max;ref denotes the lowest/highest
detection limit given in those references. The LODestim: were determined by the
use of an individual subset Ns;i for each analyte. In order to eliminate outliers, the
intensity values used for the calibration functions were twice tested with a 3s-outlier
criterion; i.e., INi D hhILIBS;i iNmp ;3s iK;3s . To each calibration function, reference
samples were added in ascending order of concentration until the coefficient of
determination exceeded r 2 D 0:9.
On the whole, the estimated LODestim: are higher than the maximum given
LODmax;ref values except for Cu. The reason for this difference lies primarily in the
different experimental setup and measuring parameters used. In [13.43], the focal
Table 13.16 Estimated detection limits of major alloying elements for certified aluminum alloys
in the static case in comparison to LOD values published elsewhere
Element (nm) Ns;i ci;min ci;max r2 sco LODestim: LODmin;ref LODmax;ref
(m.-%) (m.-%) (m.-%) .g=g/ .g=g/ .g=g/
Si I 288.16 14 0.019 0.24 0.93 0.0056 341 14 [13.43] 284 [13.45]
Fe II 259.94 10 0.0167 0.1634 0.96 0.0039 241 50 [13.46] –
Cu I 327.40 11 0.0006 0.042 0.91 0.0011 67 10 [13.43] 80 [13.46]
Mn II 293.31 17 0.0008 0.2 0.93 0.0022 135 2 [13.43] 90 [13.46]
Mg II 279.08 9 0.0001 0.028 0.90 0.0007 44 0.5 [13.43] 28 [13.46]
Cr II 267.72 26 0.00047 0.14 0.90 0.0039 242 10 [13.46] –
Ni I 341.48 29 0.0007 1.23 0.96 0.0148 903 100 [13.46] –
Ti II 337.28 10 0.00034 0.0285 0.91 0.0019 119 10 [13.46] –
length for the laser beam focusing amounts to 25 cm, the laser repetition frequency
is 1 Hz, and the plasma light was collected by a f D 10 cm quartz lens. In [13.45],
the laser repetition frequency was 0.2 Hz, i.e., by two orders of magnitude smaller
than the one used here. In [13.46], the repetition rate was 1 Hz, the lens to focus the
laser beam had a focal length of 10 cm and the plasma light was observed by a fiber
at a distance of 3 cm from the plasma.
Hence in [13.43–13.46], the influence of a breakdown on the particles ablated by
previous pulses is greatly reduced by choosing a low laser repetition frequency and
the light collection is maximized by keeping the distance of the collection optics
close to the plasma (10 cm). Thus, the reproducibility of the measurements is
increased. However, such low laser repetition rates are not suited for the desired
sorting task. Furthermore, the small focal length and observation distances used in
these references are not adequate to cover a measuring field of 400 mm 400 mm.
Furthermore, the factor ˆ4;0:01 used in (13.4) differs from ˆ10;0:01 D 3 used
in the stated references in Table 13.16. Finally, the LODestim: are increased by the
factor 1.2 in (13.4) in contrast to the reference values and are determined by raw
intensities.
In order to characterize the experimental setup and its performance over large
concentration ranges, calibrations for the analyte elements Si, Fe, Cu, Mn, Mg,
Cr, Ni, and Ti were carried out. In the following, the relation between outlier
corrected intensities (3s) and concentrations is approximated by the quadratic
analysis function (cf. Chap. 11, Fig. 11.1) c.I / D b0 C b1 I C b2 I 2 , which is the
inverse function of the calibration function. The ratio method is applied to improve
the accuracy of the approximated function by internal standardization [cf. Chap. 11,
(11.2)]. The intensity values of an analyte Ianalyte are divided by the corresponding
intensities of the reference element Ireference , which is in most cases the matrix
element. For each sample, the concentration values are referenced as well. The
analysis function is the approximation curve to Ns;i pairs of .ci;ref ; QN i;ref / with
ci;analyte N ˝ ˛ Ii;analyte
ci;ref ; Qi;ref hQi;ref iNmp ;3s K;3s ; Qi;ref : (13.6)
ci;reference Ii;reference
The diagrams in Fig. 13.24 show the analysis functions for the Si I 288.16 nm
analyte line with and without internal standardization to the Al I 305.71 nm
reference line. The improvement of the calibration by use of the ratio method can be
determined comparing the residue standard deviations sc;I and scref ;Qref . To be correct
in units, the nondimensional residue standard deviation scref ;Qref has to be multiplied
by the average Al-concentration of the used sample subset c.Al/N hci .Al/iNs;i . In
Table 13.17, the residue standard deviations of standardized and nonstandardized
analysis functions for the different alloying elements are listed as well as their
quotient as a factor of improvement. The column at the very right is an estimation
of the RSDP relating to the average analyte concentration in the particular subsets
of samples.
13.3 Aluminum
Table 13.17 Residue standard deviations without and with internal standardization and estimations of RSDP (last column)
sc;I scref ;Qref cN.Al/
Analyte element (nm) Ns;i c.Al/
N (m.-%) sc;I (m.-%) N
scref ;Qref c.Al/ (m.-%) scref ;Qref cN.Al/
cNanalyte (m.-%) cNanalyte
(%)
Si I 288.16 31 91.1871 0.9652 0.7613 1.27 2.5358 30.0
Si I 390.55 31 91.1871 0.6991 0.2632 2.66 2.5358 10.4
Fe II 259.94 30 91.0898 0.1478 0.1189 1.24 0.4909 24.2
Fe I 371.99 30 91.0898 0.2129 0.1423 1.50 0.4909 29.0
Cu I 296.12 29 91.5514 1.2772 1.0337 1.24 0.8987 115.0
Cu I 327.40 29 91.5514 0.5385 0.2115 2.55 0.8987 23.5
Mn II 293.31 30 91.3330 0.0536 0.0483 1.11 0.2746 17.6
Mg II 279.08 29 91.1774 0.2793 0.2536 1.10 0.5284 48.0
Cr II 267.72 31 91.1871 0.0322 0.0328 0.98 0.0603 54.4
Ni I 341.48 31 91.1871 0.0519 0.0418 1.24 0.2357 17.7
Ti II 337.28 31 91.1871 0.0310 0.0174 1.78 0.0802 21.7
269
14
0.16
12
10 0.12
c(Si) / c(Al) [1]

c (Si) [m.–%]
8
6 0.08
4
0.04
2 r 2 = 0.95 r 2 = 0.97
0 0.00
0 2 4 6 8 10 12 14 16 18 0.0 0.5 1.0 1.5 2.0
ISi 288.16 nm [a.u.] ISi 288.16 nm / IAl 305.71 nm [1]
Fig. 13.24 Analysis functions of the Si I 288.16 nm emission line without (left) and with internal
standardization to the Al I 305.71 nm emission line (right)
Table 13.18 Possible perturbing emission lines according to [13.30] with respect to the used
analyte lines, interelement corrected residue standard deviations, and estimations of the RSDP
(last column) for standardized analysis functions
sc;I sccorr:;Q cN.Al/
Analyte (nm) Possible log Aji sccorr:
ref ;Qref
c.Al/
N sccorr: cN.Al/
ref ref
cNanalyte
ref ;Qref
element perturbing .gi fij / (106 s1 / (m.-%) (%)
emission
line (nm)
Si I 288.16 Ni II 288.15 1.177 6.676 0.4495 2.15 17.7
Cu I 296.12 Mn II 296.12 1.280 7.979 0.7029 1.82 78.2
Mn II 293.31 Cu I 293.31 1.900 1.219 0.0442 1.21 16.1
Cr II 267.72 Mn II 267.73 0.082 85.56 0.0074 4.34 12.3
Ti II 337.28 (Mg) – 0.0125 2.48 15.6
For perturbing element in brackets no possible emission lines were found
For all analysis functions, the improvement factor by internal standardization is

greater than 1 and attains more than 2.6 except for Cr (0.98). The best achieved
RSDP ranges from 10.4% (Si) to 54.4% (Cr).
Besides the improvement of a calibration or analysis function by the ratio
method, it is possible to find corrections of these functions due to matrix or
interelement effects (cf. Sect. 11.6). The standardized analysis function of an analyte
a for additive perturbing elements k is described by
X
2
cref;a D d0 C d1 Qref;a C d2 Qref;a C ık cref;k : (13.7)
k;k¤a
This approach allows to improve the residue standard deviation of five emission
lines listed in Table 13.18 by multilinear regression.
By combining internal standardization and interelement corrections for all
analysis functions, improvement factors greater 1 and up to 4.34 were achieved.
To verify the general feasibility of LIBS measurements for samples moving
on a conveyor belt velocity with 3 m/s, calibration measurements with slightly
optimized excitation and detection parameters were carried out. In Figs. 3.25 (see
References 271
0.020 static samples

moving samples r 2 = 0.936
0.015
c(Fe) / c(Al) [1]
0.010
r 2 = 0.997
0.005
0.000
0.0 0.5 1.0 1.5 2.0 2.5
I Fe 259.94 nm / I Al 305.71 nm [1]
Fig. 13.25 Analysis functions of iron in aluminium for static and moving samples at 3 m/s with
internal standardization for Fe II 259.94 nm
Sect. 3.9) and 13.25 standardized analysis functions for static and moving samples
are shown for Cu I 327.40 nm and Fe II 259.94 nm, exemplarily. Although the
moving samples were only analyzed once with seven laser bursts at a tracked
position, the coefficient of determination of the quadratic regression is better than
those for static samples. The improvement factor of the residue standard deviation
is 3.2 for Cu and 1.5 for Fe.
References
13.1. J. Aguilera, C. Aragon, J. Campos, Determination of carbon content in steel using laser-
induced breakdown spectrometry. Appl. Spectrosc. 46, 1382–1387 (1992)
13.2. R. Noll, R. Sattmann, V. Sturm, S. Lüngen, H.-J. von Wachtendonk, Schnelle
Multielementanalyse in der Stahlschmelze mit laserinduzierter Emissionsspektrometrie.
Stahl u. Eisen 117, 57–62 (1997)
13.3. T. Sugihara, Y. Funahashi, I. Fukui, T. Miyama, Rapid determination of ultra low carbon
and nitrogen in steel with a modern simultaneous optical emission spectrometer, Progress
in Anal. Chem. in the Steel and Metals Industry, Ed. Nauche, European Commission
(1996), 229–235
13.4. V. Sturm, L. Peter, R. Noll, Steel analysis with laser-induced breakdown spectrometry in
the vacuum ultraviolet. Appl. Spectrosc. 54, 1275–1278 (2000)
13.5. E. Runge, S. Bonfiglio, F. Bryan, Spectrochemical analysis of molten metal using a pulsed
laser source. Spectrochim. Acta 22, 1678–1680 (1965)
13.6. C. Aragon, J. Aguilera, J. Campus, Improvements in quantitative analysis of steel
composition by laser-induced breakdown spectroscopy at atmospheric pressure using an
infrared Nd:YAG laser. Appl. Spectrosc. 47, 606–608 (1993)
13.7. J. Gruber, J. Heitz, H. Strasser, D. Bäuerle, N. Ramaseder, Rapid in-situ analysis of liquid
steel by laser-induced breakdown spectroscopy. Spectrochim. Acta Part B 56, 685–693
(2001)
13.8. L. Peter, V. Sturm, R. Noll, Liquid steel analysis with laser-induced breakdown spectrom-
etry in the vacuum ultraviolet. Appl. Opt. 42, 6199–6204 (2003)
13.9. M. Hemmerlin, R. Meilland, H. Falk, P. Wintjens, L. Paulard, Application of vacuum
ultraviolet laser-induced breakdown spectrometry for steel analysis – comparison with
spark-optical emission spectrometry figures of merit. Spectrochim. Acta, Part B 56, 661–
669 (2001)
13.10. S. Palanco, J.J. Laserna, Full automation of a laser-induced breakdown spectrometer for
quality assessment in the steel industry with sample handling, surface preparation and
quantitative analysis capabilities. J. Anal. At. Spectrom. 15, 1321–1327 (2000)
13.11. L. Cabalin, D. Romero, C. Garcia, J. Baena, J. Laserna, Time-resolved laser-induced
plasma spectrometry for determination of minor elements in steelmaking process samples.
Anal. Bioanal. Chem. 372, 352–359 (2002)
13.12. R. Noll, H. Bette, A. Brysch, M. Kraushaar, I. Mönch, L. Peter, V. Sturm, Laser-induced
breakdown spectrometry – applications for production control and quality assurance in
steel industry. Spectrochim. Acta, Part B 56, 637–649 (2001)
13.13. M. Khater, J. Costello, E. Kennedy, Optimization of the emission characteristics of laser-
produced steel plasmas in the vacuum ultraviolet: significant improvements in carbon
detection limits. Appl. Spectrosc. 56, 970–984 (2002)
13.14. V. Sturm, J. Vrenegor, R. Noll, M. Hemmerlin, Bulk analysis of steel samples with surface
scale layers by enhanced laser ablation and LIBS analysis of C, P, S, Al, Cr, Cu, Mn and
Mo. J. Anal. At. Spectrom. 19, 451–456 (2004)
13.15. IRSID, Voie Romaine, 57283 Maizières les Metz Cedex, France
13.16. H. Richter, H. Rzepczyk, D. Tembergen, Prozeßsteuerung sekundärmetallurgischer
Verfahren. Stahl und Eisen 114, 110–113 (1994)
13.17. H. Richter, H. Rzepczyk, D. Tembergen, Qualitätsbezogene Prozeßsteuerung in der
Sekundärmetallurgie eines Oxygenstahlwerkes. Stahl und Eisen 115, 83–87 (1995)
13.18. C. Carlhoff, Laserinduzierte Emissionsspektroskopie für die Direktanalyse von flüssigem
Stahl im Konverter. Laser und Optoelektronik 23, 50–52 (1991)
13.19. R. Noll, V. Sturm, L. Peter, I. Whiteside, Analysis using lasers, in Proc. of the 49th
Chemists’ Conference, 1997, ed. by R. Jowitt (Research and Development Department
British Steel, Middlesbrough) pp. 22–27
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D: Appl. Phys. 28, 2181–2187
(1995)
13.21. R. Noll, V. Sturm, L. Peter, R. Hakala, J. Viirret, H.W. Gudenau, K. Mavrommatis
L. Ernenputsch, Sensitivity-enhanced laser analysis (SELA) of steel melts for a fast
multi-element online analysis during ladle processing in secondary metallurgy, EUR
19411 - Analytical techniques for processes, products and the environment, Technical steel
research series; European Commission (Office for Official Publications of the European
Communities), Luxembourg, 2001
13.22. V. Rai, F. Yueh, J. Singh, Study of laser-induced breakdown emission from liquid under
double pulse excitation. Appl. Optics 42, 2094–2101 (2003)
13.23. DIN 32 645, Chemische Analytik – Nachweis-, Erfassungs- und Bestimmungsgrenze –
Ermittlung unter Wiederholbedingungen; Begriffe, Verfahren, Auswertung (Beuth Verlag,
Berlin, 2006)
13.24. M. Corsi, G. Cristoforetti, M. Hidalgo, D. Iriarte, S. Legnaioli, V. Palleschi, A. Salvetti,
E. Tognoni, Temporal and spatial evolution of a laser-induced plasma from a steel target.
Appl. Spectrosc. 57, 715–721 (2003)
13.25. C. Lopez-Moreno, S. Palanco, J. Laserna, Remote laser-induced plasma spectrometry for
elemental analysis of samples of environmental interest. J. Anal. At. Spectrom. 19, 1479–
1484 (2004)
13.26. S. Palanco, L.M. Cabalin, D. Romero, J. Laserna, Infrared laser ablation and atomic
emission spectrometry of stainless steel at high temperatures. J. Anal. At. Spectrom. 14,
1883–1887 (1999)
References 273
13.27. I. Bassiotis, A. Diamantopoulou, A. Giannoudakos, F. Roubani-Kalantzopoulou,

M. Kompitsas, Effects of experimental parameters in quantitative analysis of steel alloy
by laser-induced breakdown spectrometry. Spectrochim. Acta Part B 56, 671–683 (2001)
13.28. J. Vrenegor, R. Noll, V. Sturm, Investigation of matrix effects in LIBS plasmas of high-
alloy steel for matrix and minor elements. Spectrochim. Acta B 60, 1083–1091 (2005)
13.29. K. Slickers, Automatic Atomic Emission Spectroscopy, Brühlsche Universitätsdruckerei,
Giessen, 2nd edn. (1992)
13.30. R. Kurucz, B. Bell, Atomic Line Data. Kurucz CD-ROM No. 23 (Smithsonian Astrophys-
ical Observatory, Cambridge,, 1995)
13.31. J. Aguilera, C. Aragon, A comparison of the temperatures and electron densities of laser-
A 69 [Suppl.], S475–S478 (1999)
13.32. K. Krone, Aluminiumrecycling: Vom Vorstoff bis zur fertigen Legierung, (Aluminium-
Verlag, Düsseldorf, 2000)
13.33. Metallstatistik 1957–1966, Ed. Metallgesellschaft AG, vol. 54, Frankfurt am Main, 1967
13.36. Metallstatistik, Metal Statistics 1989–1999, ed. by World Bureau of Metal Statistics,
vol. 87, Ware, England (2000)
13.37. Metallstatistik, Metal Statistics 1993–2003, ed. by World Bureau of Metal Statistics,
vol. 91, Ware, England (2004)
13.38. Aluminium-Taschenbuch, ed. by Aluminium-Zentrale (Aluminium-Verlag, Düsseldorf,
1983)
13.39. C. Schmitz, Handbook of Aluminium Recycling – Fundamentals, Mechanical Preparation,
Metallurgical Processing, Plant Design (Vulkan-Verlag, Essen, 2006)
13.40. K. Krone, J. Krüger, H. Orbon, H. Sommer, H. Vest, Ökologische Aspekte der Primär- und
Sekundärerzeugung in der Bundesrepublik Deutschland. Metall 44, 559–568 (1990)
13.41. G. Rombach, B. Friedrich, Aluminum recycling in Germany. Light Metal Age 59, 66–75
(2001)
13.42. Ü. Aydin, R. Noll, J. Makowe, Automatic sorting of aluminium alloys by fast LIBS
identification, in 7th Int. Workshop Progress in Analytica Chemistry in the Steel and Metal
Industries, ed. by J. Angeli (Glückauf GmbH, Essen, 2006) pp. 309–314
13.43. M. Sabsabi, P. Cielo, Quantitative analysis of aluminum alloys by laser-induced breakdown
13.44. G. Rieger, M. Taschuk, Y. Tsui, R. Fedosejevs, Laser-induced breakdown spectroscopy for
microanalysis using submillijoule UV laser pulses. Appl. Spectrosc. 56, 689–698 (2002)
13.45. M. Ismail, H. Imam, A. Elhassan, W. Youniss, M. Harith, LIBS limit of detection and
plasma parameters of some elements in two different matrices. J. Anal. At. Spectrosc. 19,
489–494 (2004)
13.46. M. Ismail, G. Cristoforetti, S. Legnaioli, L. Pardini, V. Palleschi, A. Salvetti, E. Tognoni,
M. Harith, Comparison of detection limits, for two metallic matrices, of laser-induced
breakdown spectroscopy in the single and double-pulse configurations. Anal. Bioanal.
Chem. 385, 316–325 (2006)
Chapter 14
Bulk Analysis of Nonconducting Materials
This chapter describes LIBS investigations for a bulk characterization or analysis of

nonconducting materials, such as polymers, waste electric and electronic equipment,
slag, soil, cement, droplets, gases, particulates, and aerosoles.
14.1 Polymers
In this section, chemical analysis of polymers with LIBS refers to the identification
of plastic materials itself and to additives in technical polymers.
14.1.1 Identification of Plastic Materials
Polymers belong to the most wide spread packaging materials for consumer
products as food and cosmetics. One approach to reduce the domestic waste streams
of used polymers is a high-grade material recycling. Use of mixtures of polymers
is restricted to the production of low-grade plastics, and the chemical recycling by,
e.g., hydrogenation or energy recovery [14.1]. In the European Union, the share
of recycled polymers (without energy recovery) is about 17% of the total annual
polymer waste mass, so there is still potential to further increase this amount [14.2].
For a high-grade material recycling, the mixtures have to be separated into
fractions of different polymers. For an economically interesting throughput of a
sorting machine, a high-speed identification method with identification rates of
several pieces per second is necessary.
The requirements for the purity of the separated fractions depend on the
polymer type and the kind of contaminants. For the PE (polyethylene) fraction,
e.g., a contamination of 5 m.-% of PP (polypropylene), PS (polystyrene), or PET
(polyethylene terephthalate) may be tolerable, whereas PVC (polyvinyl chloride),

276 14 Bulk Analysis of Nonconducting Materials
should be close to 0% since corrosion problems will arise in the subsequent

processing steps from decomposition of PVC forming hydrochloric acid.
It is well known that the molecular structure of the polymers influences the
reflection and transmission spectra in the near infrared (NIR). Several approaches
have been undertaken to use this fact for a high-speed identification of post-
consumer plastics [14.3–14.7].
Surface contaminations of the polymers may deteriorate the measuring signals,
thus reducing the identification accuracy. This is especially the case if contaminating
surface layers are not transparent in the NIR and exceed a thickness of about 20 m
[14.7]. Furthermore, NIR is not applicable for black-colored samples. X-rays can be
used to detect the chlorine content to identify PVC [14.8].
In the last years, NIR technologies became the state-of-the-art technology to
identify bottles made of PVC, PET, or PP for automatic sorting lines [14.9, 14.10].
Medium infrared (MIR) techniques are capable to identify technical polymers,
which are used as components of electric appliances [14.11]. Typical identification
times for MIR are several seconds, which excludes such an approach for high-
throughput applications.
The principal feasibility of LIBS to identify polymers and to detect additives
in polymers will be presented in this and the following section. Compared with
NIR, MIR, and X-ray methods, LIBS offers an extended broad spectral signature
to identify polymers and additives in polymers at the same time for high-speed
applications. In a first step, emission spectra of polymers in the near UV and visible
spectral region using LIBS in combination with time-resolved spectroscopy were
studied.
The experimental setup is shown in Fig. 14.1 [14.12]. A Q-switched Nd:YAG
laser operating at 1,064 nm was used in the double pulse configuration mode, see,
e.g., [14.13] (cf. Chap. 6). The laser source emits two 30 ns equal energy pulses with
a time separation of 6 s between them. The use of multiple pulses can significantly
improve the analytical performance in terms of reproducibility of the signal and
detection limits, cf. Chap. 13. The laser beam was focused by means of a 200 mm
plano-convex lens onto the sample. The position of the focal plane was s D 5 mm
below the surface of the sample to avoid an air breakdown, which can shield off
a part of the laser energy. A bifurcated quartz fiber bundle collects the plasma
radiation. One of the arms of the fiber bundle was mounted at the entrance slit
of a spectrometer (Jobin Yvon HR 320 with a 600 l/mm grating) equipped with
an intensified photodiode array (IPDA, 1,024 diodes) covering an effective spectral
bandwidth of about 70 nm and gateable in the 10 ns–100 s time range. The settings
for the signal integration were chosen as follows: tdelay D 1 s between the second
laser pulse and the start of exposure of the photodiode array, tint D 10 s. The
second arm of the fiber bundle was linked to a monochromator (Jobin Yvon HR 250
with a 2,400 l/mm grating) equipped with a photomultiplier tube (PMT, Hamamatsu
R1547) and centered at the chlorine line Cl 725.66 nm. Due to the presence of
strong continuum radiation in that spectral region and for normalization purposes,
the continuum emission was also recorded by means of a positive-intrinsic-negative-
photodiode (PIN) and an interference filter (IF, peak transmission 0.47 at 739.5 nm,
14.1 Polymers 277
Fig. 14.1 Experimental setup. M D mirror, L D lens, P D plasma, S D sample, IF D interference

filter, PIN D positive-intrinsic-negative photodiode, FO D fiber optics, IPDA D intensified
photodiode array, CT D controller, PMT D photomultiplier tube, MCI D multichannel integrator
electronics, PC D personal computer
bandwidth 4 nm). The outputs of the photomultiplier tube and the PIN-photodiode
were fed to gated integrators (MCI) and their signals were integrated from 1 to
10 s after the laser pulse and after the second laser pulse, respectively, when double
pulses were applied.
Plastic bottles have great variations of colors and additives, which influence, e.g.,
the stiffness of the polymer. Within the frame of the investigations described in the
following, it was not possible to measure all different bottles available. To restrict
the number of samples for the experiments to 20 while keeping material variety as
wide as possible, several bottles of a given material with similar color were mixed
to produce for example a “medium green HDPE sample,” which contains colorants
and additives typical for green HDPE bottles. To generate these samples, the bottles
have been washed and dryed for 3–4 h at 80–90ı C to achieve a low moisture content.
Then they were shredded to 3–5 mm flakes. These flakes were mixed to achieve
a homogeneous melt and pressed into a mold. In this way, the samples listed
in Table 14.1 were manufactured. Hence, most of these samples represent plastic
mixtures of used bottles. Only the samples “HDPE blue 2” and “HDPE yellow 2”
are processed by virgin material to obtain TiO2 -free samples. Since bottles made of
LDPE with blue, red, or yellow color are rare, the corresponding samples are made
from virgin material.
An additional pure reference sample of high-density polyethylene was used
(delivered by Goodfellow GmbH, Germany, article no. ET323100/2). All samples
had flat surfaces and were oriented perpendicularly to the laser beam. Finally,
post-consumer plastic bottles of the polymers PE, PET, and PVC were used to
Table 14.1 Polymer samples used for the experiments

No. Material Color No. Material Color
1 HDPE Blue 1 12 PP Blue
2 HDPE Blue 2 13 PP Green
3 HDPE Brown 14 PP Not dyed
4 HDPE Not dyed 15 PP Red
5 HDPE Yellow 1 16 PVC Blue
6 HDPE Yellow 2 17 PVC Not dyed
7 LDPE Blue 18 PVC Red 1
8 LDPE Not dyed 19 PVC Red 2
9 LDPE Red 20 PVC White
10 LDPE Yellow 21 HDPE GF
11 PET Not dyed
Samples 2, 6, 7, 9, and 10 are made from virgin material, and all other samples are made from
recycled bottles
HDPE D high density polyethylene, LDPE D low density polyethylene, PET D polyethylene
terephthalate, PP D polypropylene, PVC D polyvinyl chloride, GF D Goodfellow reference
sample, not dyed
determine identification accuracies on unprepared samples. They were not washed.

Measurements were performed on spots free of labels.
Artificial neural networks were applied for a classification of the spectra.
Several networks like feedforward, dynamic learning vector quantization (DLVQ),
or radial base function (RBF) networks were tested [14.14]. The best results
were achieved with modularized feedforward networks trained by backpropagation.
Modularization means that two or more subnetworks share the same input layer, but
each subnetwork has its own hidden and output layer. Each output layer may have
one or more output cells. The total number of output cells is the same as the number
of different polymer classes. Besides the better results of the modularized networks
compared with all other tested networks, the modularization has the advantage that
the subnetworks can be retrained to adopt to new samples without influencing the
other subnetworks.
Subnetworks with different numbers of hidden layers were compared, and it was
found that one hidden layer is sufficient. All cells of the hidden layer of a subnetwork
are connected with all input cells, and all output cells are connected with all cells in
the hidden layer of its subnetwork. Details of the networks will be given below.
If input data of an unknown or test sample are supplied to the input cells, the
sample is attributed to the class which has the highest activation at the output cell,
regardless of the activation of the other output cells. The activation of an output cell
corresponds to the similarity between actual input data and that input data for which
this output cell was trained. Hence, the amount of incorrect classifications can be
reduced, if a higher activation limit is defined which has to be exceeded. Otherwise
the classification is assessed as ambiguous and the sample is not classified. It is
obvious that there is a strong correlation between the activation limit, the amount
of correct classifications, and undone classifications: the higher the activation limit
is, the higher is the amount of not-classified samples and the amount of correct
14.1 Polymers 279
classifications. For practical reasons, the amount of not-classified samples has to

be limited. The activation limit was fixed to that value, where 20% of the samples,
averaged over all polymer classes, are not classified.
Molecular materials like polymers are partly atomized, when they are exposed to
laser radiation, which is intense enough to create a plasma. This initially implies
a limitation concerning LIBS application on polymers because of the loss of
molecular information in the plasma. Table 14.2 shows the semi-structural formula
of the investigated polymers. Some polymers differ in their stoichiometric C/H ratio.
It is obvious to use this information to identify the polymers.
To simplify the experimental effort, it is advantageous to find a spectral region
which permits a simultaneous detection of carbon and hydrogen lines. Figure 14.2
shows two spectra of a HDPE sample (sample 1 in Table 14.1) in the spectral region
from 460 to 530 nm. In the upper spectrum, two intense clearly distinct lines appear
at 486.1 and 495.7 nm. They are assigned to the H“ line of the Balmer series of
hydrogen at 486.13 nm and to a carbon C I line at 247.86 nm, respectively. The
carbon line arises in second order at 495.7 nm, as it is depicted in the lower spectrum
of Fig. 14.2, which has been recorded adding a low pass filter in front of the fiber
optic to avoid second-order contributions (Schott filter, type GG 375). The hydrogen
line and the C/H ratio will be discussed below.
Table 14.2 Semi-structural formula and stoichiometric ratio of carbon and hydrogen atoms
Polymer Formula C/H ratio
PE .CH2 /n 0.5
PP .CH.CH3 / CH2 /n 0.5
PET .O CH2 CH2 O CO C6 H4 CO/n 1.25
PVC (-CHCl-CH2 -)n 0.66
15000
HDPE blue 1
Hβ 486.13 nm C 247.86 nm
intensity [counts]
10000
(2nd order)
without filter
5000
with filter
0
460 470 480 490 500 510 520 530
λ[nm]
Fig. 14.2 Spectrum of HDPE blue 1 (sample 1 of Table 14.1) with and without the 247.86 nm
carbon line in second order (double pulses 2 100 mJ). For better presentation the unfiltered
spectrum is shifted in intensity by 4,000 counts
HDPE blue 1
12000
1 × 200 mJ
2 × 100 mJ
intensity [counts]
8000
4000
0
450 460 470 480 490 500 510 520 530
λ [nm]
Fig. 14.3 Spectrum of HDPE blue 1 (sample 1 of Table 14.1) for single pulse and double pulse
excitation
Figure 14.3 shows two spectra of a HDPE sample (sample 1 in Table 14.1) gained
with single and double laser pulse excitation. In the spectrum generated with double
pulse excitation, more spectral features appear and the line-to-continuum ratio is
significantly enhanced. Hence, for the following measurements, mostly double pulse
excitation is used.
Figure 14.4 presents spectra of different polymer samples in the spectral region
between 440 and 530 nm. The experimental conditions for all these spectra are
identical, i.e., 2 150 mJ double pulses, tdelay D 1 s and tint D 10 s. Within the
spectra, distinct structures are observed. These structures arise from the excitation
of the molecular species in the plasma plumes which constitute the polymer, their
products of decomposition and additives as well. For example, the bands appearing
at 470 and 516 nm can be attributed to the C2 -molecule (Swan bands) [14.15].
Concerning the atomic species present in the plasma emission spectra, various
atomic lines of carbon, hydrogen, chlorine, titanium, and other metals used as
additives in the polymeric products have been identified.
In the spectral range between 440 and 530 nm, eight “regions of interest” (ROI)
have been defined, among these are the H“ line and the carbon line. The evaluation
of these ROIs using neural networks will be discussed below.
The widths of the Stark broadened hydrogen Balmer series lines have been
extensively studied and used to determine the electron density of plasmas (cf.
Sect. 9.4). In the present case, the relations given by Griem have been used for an
approximate determination of the averaged electron number density ne in the laser
produced plasma [14.16]. Figure 14.5 shows spectra of PET with the hydrogen and
carbon line for different time delays.
With increasing time delay, the electron density decreases and hence the width
of the H“ line. The electron density calculated from the line width of the H“ line
decreases from 1:351017 cm3 at 1 s time delay to 0:651017 cm3 at 3 s.
14.1 Polymers 281
15000
LDPE red PP green
intensity [cts]
10000
5000
0
15000
PET not dyed PVC blue

intensity [cts]
10000
5000
0
15000
HDPE blue 1 HDPE GF
intensity [cts]
10000
5000
0
440 470 500 530 440 470 500 530
λ [nm] λ [nm]
Fig. 14.4 Spectra of different polymers in the spectral region from 440 to 535 nm. HDPE GF
denotes a pure high density polyethylene sample (Goodfellow GmbH), double pulses 2 150 mJ,
tdel D 1 s, tint D 10 s
At the beginning of the plasma formation, and because of the high electron
densities, there is considerable Stark broadening which leads to FWHM of the H“
line in the order of 6 nm. At high electron densities, the red wing of the H“ line can
interfere with the neighboring second-order carbon line. For a time delay longer than
3 s, the width of the hydrogen line is so small that the carbon and the hydrogen
lines are well separated.
The C/H intensity ratios were determined for different samples. The background
contribution to the H“ and C I line intensities has been taken into account to
determine the integrated line intensities. The carbon/hydrogen ratio depends on the
time delay, probably due to the decreasing electron temperature and the different
excitation energies, which amounts to 7.68 eV for the carbon line C 247.86 nm
and 12.74 eV for the hydrogen line H“ 486.13 nm. With increasing time delay, the
C/H ratio increases for the studied polymers LDPE, HDPE, PP, PVC, and PET. As
expected from the stoichiometric C/H ratio, see Table 14.2, the PET sample yields
20000 C line
PET not dyed
t del = 1 μs
15000
intensity [cts]
t del = 2 μs
t del = 3 μs
10000 Hβ line
5000
0
450 460 470 480 490 500
λ [nm]
Fig. 14.5 Spectra of PET not dyed (sample 11 of Table 14.1) for different time delays (single
pulses 1 100 mJ, tint D 5 s)
the highest measured C/H ratio for a given time delay. However, these results are
not sufficient for an unambiguous identification of all different polymers, since the
values of the C/H ratio for some samples partly overlap.
However, it is important to notice that, despite the intrinsic difficulties in
obtaining the appropriate C/H ratios for all categories of polymers, the PET samples
exhibited the highest C/H values enabling a 100% correct classification of this
category which will be described later.
Table 14.2 presents the basic chemical composition of the polymers studied. The
only polymer which contains chlorine is PVC. Therefore, chlorine emission lines in
the LIBS spectra were studied in order to identify PVC and to discriminate it from
the other polymers.
Figure 14.6 shows two spectra of not dyed PVC (sample 17 in Table 14.1) in
the spectral region from 710 to 760 nm. The distinct spectral features at 725.7 and
754.7 nm have been identified as chlorine lines. Both spectra have been obtained
using the same laser energy of 300 mJ. The upper one corresponds to a single 300 mJ
laser pulse, whereas the lower one was obtained with a double pulse having the same
total energy distributed equally into two pulses separated by a time interval of 6 s.
Both spectra have been measured with a time delay of 0:5 s after the laser pulse
(respectively, the second pulse of the double pulse) and integrated over a time period
of 10 s. The 725.66 nm chlorine line in the double pulse spectrum is more intense
by a factor of 4 in comparison with the single pulse excitation.
In Fig. 14.7, the ratio of the intensity of the Cl 725.66 nm line to the intensity of
the nearby background is shown for various polymer samples. As a result, the PVC
samples can be unambiguously identified by measuring that ratio.
Artificial neural networks were applied to classify the polymers to polymer
categories by their LIBS spectra. Nine spectral ROI were evaluated to determine
in arbitrary units: the intensity of the chlorine line Cl I 725.66 nm, the intensity
14.1 Polymers 283
intensity [counts] 8000 PVC no
6000
Cl 754.71 nm
1 × 300 mJ, slit 100 μm
4000
2 × 150 mJ, slit 50 μm
2000 Cl 725.66 nm
0
710 720 730 740 750 760
λ [nm]
Fig. 14.6 Spectra of PVC not dyed (sample 17 of Table 14.1) with chlorine lines. For better
presentation the single pulse spectrum is shifted in intensity by 2,000 counts
1.0
intensity ratioICl / Ibackgr. [a.u.]
0.8
0.6
0.4 PVC samples
0.2
0.0
1 5 10 15 20
sample no.
Fig. 14.7 Intensity ratio of the 725.66 nm chlorine line and the nearby background for polymer
samples 1–20 listed in Table 14.1
of the background emission at 739.5 nm, the background corrected intensity of

the hydrogen line H“ 486.13 nm, the background corrected intensity of the carbon
line C I 247.86 nm, the background corrected intensity of the C2 -band emission
at 516:5 nm, the background corrected intensities of four further lines in the
spectral region between 440 and 535 nm. With the exception of Cl 725.66 nm and
the background emission at 739.5 nm, all intensities were evaluated by integrating
the values of pixels under the lines and applying a linear background correction. The
nine intensities were further evaluated to achieve eight input variables for the neural
networks. The variables are defined as follows: intensity ratio of H“ 486.13 nm

to C I 247.86 nm, intensity ratio of Cl I 725.66 nm to background emission at
739.5 nm.
The other six variables are the intensities of C I 247.86 nm, C2 -band emission
at 516:5 nm, and the four yet unidentified spectral lines. All variables were
normalized by appropriate constants to limit their values to the range 0–1. At first,
a modularized feedforward network for the identification of samples no. 1–20 listed
in Table 14.1 was tested. Two different sets of data were recorded for training and
validation of the neural network. Each sample was measured 100 times at 100
different positions on the sample surface. 50 datasets per sample were stored for
training, the remaining 50 datasets are used for validation. By comparing results on
training and validation data, over-training can be avoided [14.17].
A feedforward network, modularized into two subnetworks, with eight input cells
corresponding to the eight variables was used. One subnetwork has one output cell
for PVC and three cells in the hidden layer, and the other subnetwork has three
output cells for PE, PP, and PET and eight cells in the hidden layer.
Table 14.3 summarizes the results. The column entitled “Not classified” gives the
percentage of samples that could not be identified unambiguously by the network
and thus were rejected (e.g., 28.6% of all PE samples were not classified, in contrast
to only 0.4% of the PVC samples). The column entitled “Fraction pureness” gives,
e.g., the number of real PE samples classified as PE with respect to all samples
classified as PE. Any incorrect classification of a sample as PET could be avoided,
and the amount of samples misclassified as PVC is 0.4%. The mean rate of 20% not
classified samples was adjusted by the activation limit.
The classification was also tested for unprepared post-consumer bottles of PE,
PET, and PVC. PP is barely used for bottles and therefore left out for the following
experiment. A modularized network with a separate subnetwork for PE, PET,
and PVC was used. Hence, each subnetwork has only one output cell. The PE
subnetwork has four, the PET subnetwork five, and the PVC subnetwork two cells
in the hidden layer. The following datasets were used for training and testing: for
training, five bottles of PE, five of PET, and five of PVC were measured 20 times
each, resulting in 300 datasets; for testing, similar 300 datasets were used, but
measured on different bottles.
Table 14.3 Results of the automatic classification of the polymer samples using a feedforward
neural network (values in %, the columns denoted with (1) refer to 100 samples, e.g., of PE;
column (2) refers to 100 samples classified as, e.g., PE)
Sample Correctly Incorrectly Not classified, Fraction
material classified, (1) classified, (1) (1) pureness, (2)
PE 65.0 6.4 28.6 93.6
PP 72.5 4.5 23.0 95.5
PET 72.0 0.0 28.0 100.0
PVC 99.2 0.4 0.4 99.6
Mean 77.2 2.8 20.0 97.2
14.1 Polymers 285
Table 14.4 Results of the automatic classification of post-consumer polymer bottles using
feedforward neural network (values in %, the columns denoted with (1) refer to 100 samples,
e.g., of PE; column (2) refers to 100 samples classified as, e.g., PE)
Bottle Correctly Incorrectly Not classified, Fraction
material classified, (1) classified, (1) (1) pureness, (2)
PE 76.0 6.0 18.0 100.0
PET 67.0 0.0 33.0 84.4
PVC 85.0 6.0 9.0 100.0
Mean 76.0 4.0 20.0 94.9
The results are summarized in Table 14.4. As for the previous results, the
percentage of datasets not classified is adjusted to 20%. The mean fraction pureness
is 94.9%. The PET fraction pureness of 84.4% is due to the fact that 6.0% of the PE
bottles and 6.0% of the PVC bottles are classified as PET.
The applicability of LIBS for certain cases, like a high-speed separation of
washed PET and PVC bottles, seems feasible with high accuracy. LIBS may also
be used complementary to near-infrared spectroscopy, which can also be applied for
the identification of polymers, but is not suited for dark colored samples.
14.1.2 Detection of Heavy Metals and Flame Retardants

in Technical Polymers
During the recycling process of end-of-life waste electric and electronic equipment
(EOL-WEEE), downgrading of valuable technical polymers has to be avoided by
separating the collected material in fractions of high purity. Most EOL-WEEE
pieces are doped with several additives to improve their mechanical, electrical,
and chemical properties. Common additives are flame retardants, antioxidants,
light stabilizers, fillers, dyes, and pigments. Their concentration varies typically
from traces to several percent [14.18]. Waste pieces containing brominated flame
retardants (BFR) and additives with heavy metals have to be automatically identified
and removed from the waste stream to be recycled. This is due to the significant
environmental problems during the waste management phase caused by these sub-
stances. To establish an economically feasible recycling process which meets these
requirements, high-speed automatic sorting systems performing the identification
of both the polymer and the detection of the critical additives at a rate of several
parts per second are required. A prototype automatic identification and sorting line
has been setup and evaluated for material-specific sorting of EOL-WEEE pieces
[14.19]. The detection unit of this automatic sorting line is a multisensor system for
the rapid identification of the polymer matrix and the detection of the contained
additives. It comprises three spectroscopic modules based on LIBS, NIR (near
infrared spectrometry) and MIR (mid-infrared spectrometry). This section presents
investigations of LIBS for the analysis of heavy metals and brominated flame
retardants in EOL-WEEE pieces before recycling, the setup of a LIBS analyzer

for an automatic sorting line, and the evaluation of the LIBS analyzer for the inline
detection of heavy metals and brominated flame retardants in moving EOL-WEEE
pieces [14.20].
Frequently applied methods for the determination of metals and flame retardants
are atomic absorption spectrometry (AAS), inductively coupled plasma atomic
emission spectrometry and mass spectrometry (ICP-AES, ICP-MS), or electro-
chemical techniques [14.21]. However, these methods require the time-consuming
step of sample dissolution prior to the analysis and are therefore not applicable
for high-speed applications. Therefore, direct sampling analytical methods such as
laser ablation mass spectrometry (LA-MS) [14.22], glow discharge spectrometry
(GD-OES) [14.23, 14.24], sliding spark discharge optical emission spectrometry
(sliding SD-OES) [14.25], and X-ray fluorescence spectrometry (XRF) have gained
more interest. However, these methods are not feasible for inline applications with
aspired detection rates of several parts per seconds. On the other hand, LIBS as a
rapid analysis method without sample contact has the potential for inline high-speed
applications. The identification of the polymer matrix using LIBS was described
in Sect. 14.1.1 and studied by several groups [14.12, 14.26]. Bromine has been
detected with LIBS in pure solid organic compounds and a helium atmosphere in the
laser–target interaction zone has been found to significantly improve the detection
[14.27]. The LIBS detection of heavy metals in polymers was reported for single
pulse plasma excitation under laboratory conditions and applied to experiments at a
double chain extruder [14.28,14.29]. However, to sort EOL-WEEE waste according
to the content of hazardous elements before recycling, waste pieces have to be
analyzed while moving on belt conveyors at a speed of up to 1 m/s or higher (cf.
Sects. 13.3 and 18.4). This requires a LIBS setup capable to compensate distance
variations between the focusing optics and the surface of the waste piece.
The heavy metal detection with LIBS described in the following is focused on
lead, chromium, cadmium, and mercury found in pigments, nucleating agents, and
heat stabilizers. Brominated flame retardants are detected via Br and additionally
via Sb as a component of the widely used synergist antimony trioxide Sb2 O3 .
Due to the requirement to provide a LIBS analysis for inline measurement under
industrial conditions, all measurements are constrained to be conducted in air under
normal atmospheric conditions. Signal-to-noise enhancing detection in low-pressure
or noble gas atmospheres is not applicable in this case.
A set of reference samples were used for the LIBS investigations and for calibra-
tion purposes [14.19]. The polymer matrix is acrylonitrile butadiene styrene (ABS)
and the reference samples contain the elements Cd, Cr, Hg, Pb in concentrations
of 50–10;000 g=g, Sb in concentrations of 50–100;000 g=g, and Br in con-
centrations of 50–300;000 g=g. Furthermore, virgin samples with concentrations
c < 1 g=g for all these elements were provided for calibration.
The experimental setup to study the influence of detection parameters and
excitation conditions is shown in Fig. 14.8 [14.20]. The pulses of a Q-switched
Nd:YAG laser, which can be operated in single pulse and in double pulse mode,
are focused onto the surface of the specimen by a single lens with a focal
14.1 Polymers 287
Fig. 14.8 Laboratory setup for LIBS experiments. M D mirror, L D lens, P D laser-induced
plasma, S D sample, W D quartz window, F D fiber optic, ESP D echelle spectrometer, ES D
entrance slits, CLM D collimator mirror, PR D prism, EG D echelle grating, CMM D camera
mirror, FP D focal plane, PC D personal computer
length of f D 220 mm. The parameters of the laser are: wavelength D 1;064 nm,
typical pulse energy EL D 350 mJ, pulse duration (FWHM) L D 7 ns, repetition
rate D 30 Hz. For single pulse plasma excitation, the beam waist of the focus is
positioned inside the sample volume at s D 12 mm underneath the surface of the
specimen. The emission of the laser-induced plasma is received by a fiber optics
with a core diameter of 600 m and a length of 2 m. The receiving aperture of
the fiber optics is protected by a quartz window and positioned at a distance of
d D 60 mm to the laser-induced plasma. The received plasma radiation is guided
to an echelle spectrometer equipped with a microchannel plate (MCP) electro-optic
gateable image intensifier and a megapixel full-frame CCD camera (Kodak KAF
1001), cf. Sects. 4.2, 4.4. The plasma emission is detected during a defined gating
interval at wavelengths from D 200 nm to D 780 nm with a spectral resolution of
D 5 pm to D 19 pm. The integrated intensities are digitized and transferred
to a personal computer for data processing.
The detection of bromium in air under atmospheric pressure is difficult since
the most sensitive Br emission lines are in the vacuum ultraviolet (VUV) spectral
region at wavelengths < 170 nm, e.g., 148.85, 153.17, or 163.34 nm [14.30]. Due
to the strong absorption of these spectral emissions by the ambient atmosphere (cf.
Sect. 3.5, Fig. 3.14), the application of these VUV lines is not feasible for an inline
LIBS analyzer. Hence, bromium is detected using the D 827:24 nm emission line
as the most sensitive Br emission line in the spectral range 200 nm < < 900 nm.
With the available echelle spectrometer, it was not possible to detect the Br emission
at 827.24 nm.
To achieve high sensitivity and also to enable high-speed measurements, the
echelle spectrometer is replaced by a Paschen–Runge polychromator. In this
polychromator, the spectral emission lines Cd 228.80 nm, Cr 425.43 nm,

Hg 253.65 nm, Pb 405.78 nm, Sb 259.80 nm, Br 827.24 nm, and the 0th-order signal
are detected by photomultiplier tubes, cf. Fig. 4.7, Table 4.4, and Sect. 4.4. For each
laser-induced plasma, the photomultiplier signals are processed by a multichannel
integrator electronics (MCI, cf. Fig. 4.16), which integrates the intensities of all
spectral lines simultaneously. The time-integrated line intensities are digitized and
transmitted to the control PC for further data processing.
The target of the investigations was to minimize the limit of detection (LOD)
for the elements Cd, Cr, Hg, Pb, Sb, and Br under the constraints of the inline
application. This can be achieved by maximizing the signal-to-noise ratio (SNR)
of the respective background-corrected integrated line intensities. If spectra are
taken with the echelle spectrometer, both the line intensity and the SNR of an
investigated emission line can be determined from one recorded spectrum. The
spectral background and its noise within the wavelength range of the emission line
are determined by the mean signal level and its standard deviation in wavelength
regions adjacent to the emission line. The background corrected line intensity is then
calculated by standard background correction procedures. The expectation value
and its variation are gained by repeated measurements.
The features of the bromium line at 827.24 nm require the use of the Paschen–
Runge polychromator which was also used for all Sb 259.80 nm measurements since
Sb is contained in the same set of reference samples as Br. Using the Paschen–
Runge polychromator, it is not possible to determine the spectral background and
the background corrected line intensity within one measurement. Here, repeated
measurements with one specimen of high- and with one specimen of low-analyte
concentration are performed. Then, the analytical resolving power A of each analyte
line is calculated, defined here as:
INhigh INlow INanalyte

AD 1=2 D N ; (14.1)
s 2 Ihigh C s 2 .Ilow / s.Ianalyte /
where INhigh and INlow are the mean intensities of the analyte line measured with
the specimens containing the analyte at high (chigh ) and low concentration (clow /
and s.Ihigh ) and s.Ilow ) are the respective empirical standard deviations. For a low
analyte concentration, clow close to or lower than the detection limit, the analytical
resolving power A is similar to the expected mean SNR.
The influence of the laser pulse energy on the SNR was investigated for
single pulse plasma excitation. For the elements Cd, Cr, Hg, and Pb, the echelle
spectrometer and for Sb and Br the Paschen–Runge polychromator was used for
signal detection. The plasma emission was detected in a fixed integration time gate
tint D 10 s starting with a time delay tdelay D 1 s after the laser-induced plasma
ignition. The results for all investigated elements are summarized in Table 14.5. For
Cd 228.80 nm, increasing the laser pulse energy from EL D 50 mJ to EL D 300 mJ
by a factor of 6 improves the SNR by 94%. At the same time, the background-
corrected line intensity increases by a factor of 20. The significantly smaller
14.1 Polymers 289
Table 14.5 Relative change in the signal-to-noise ratio (SNR) or analytical resolving power (A)
with increasing laser pulse energy
Element (nm) clow .g=g/ chigh .g=g/ EL1 (mJ) EL2 (mJ) Relative change
of SNR or A(%)
Cd 228.80 – 10,000 50 300 94
Cr 425.43 – 10,000 50 300 40
Hg 253.65 – 10,000 50 300 100
Pb 405.78 – 10,000 50 300 40
Sb 259.80 50 100,000 100 350 0
Br 827.24 50 100,000 100 350 633
, emission wavelength of analyte; clow , chigh , analyte concentrations in the ABS reference samples;
EL1 , EL2 , laser pulse energy; A, analytical resolving power, see (14.1)
improvement of the SNR in comparison to the line intensity is due to the increasing
electron density in the plasma yielding a higher spectral background level. Still,
the expected detection limit is almost lowered by a factor of two. For mercury
Hg 253.65 nm, the SNR could also be approximately doubled increasing the pulse
energy from EL D 50 mJ to EL D 300 mJ. The most distinct improvement could be
achieved for the Br 827.24 nm emission line, where detection was not possible with
a pulse energy of EL D 50 mJ. Increasing the pulse energy from EL D 100 mJ to
EL D 350 mJ, the SNR could be improved by more than a factor of 6.
In contrast, for Sb 259.80 nm, no significant variation of the analytical resolving
power A was found with increasing laser pulse energy. For the emission lines of
Cr 425.43 nm and Pb 405.78 nm, even a decrease in SNR of about 40% was observed
for higher pulse energies. Hence, the temporal evolution of the SNR has to be
considered for these emission lines and suitable integration time gates for signal
detection have to be determined.
Figure 14.9 shows the lead Pb 405.78 nm emission line for a width of the
integration time gate of tint D 0:5 s at the time delays tdelay D 1, 3, and 20 s.
Figure 14.10 shows the corresponding SNR as a function of the time delay
ranging from 1 to 20 s. While the overall signal intensity is decreasing as a function
of time, the SNR is rising at early time delays of tdelay < 1 s to tdelay D 3 s and
decreases slowly for time delays up to 20 s.
Regarding signal integration, the temporal evolution of both the signal-to-noise
ratio and the line intensity have to be considered. High line intensities with low
SNR at early delay times contribute much more background to the integrated line
intensity than the low line intensities at larger time delays which show a higher SNR.
The noise contribution during the decay of the analyte signal with low SNR at very
large delay times can be neglected due to the very low overall intensity of the line
emission.
The target is to maximize the SNRs of the integrated background-corrected line
intensities Ianalyte i .tdelay ; tint / for each analyte line i by setting the optimum time
delay tdelay and integration time gate tint . Using a Paschen–Runge polychromator,
the optimum parameters can be determined by sampling the photomultiplier signals
directly with an oscilloscope and calculate the analytical resolving power A for any
12000
10000
tdelay = 1 μs
intensity [a.u.]
8000
6000
4000
3 μs
2000 20 μs
0
405.25 405.50 405.75 406.00 406.25
wavelength [nm]
Fig. 14.9 Temporal evolution of the Pb 405.78 nm emission line for single pulse plasma excitation
with laser pulse energy of EL D 250 mJ. Detection with echelle spectrometer using an integration
time gate of tint D 0:5 s at the stated time delays. ABS reference sample, cPb D 10;000 g=g
35
30
25
SNRPb 405.78 nm
20
15
10
0
0 2 4 6 8 10 12 14 16 18 20
t delay [μs]
Fig. 14.10 Signal-to-noise ratio of the background-corrected Pb 405.78 nm emission line as a

function of the time delay of the integration window for single pulse plasma excitation with a
laser pulse energy of EL D 250 mJ. Detection with echelle spectrometer using an integration time
gate of tint D 0:5 s. ABS reference sample, cPb D 10;000 g=g
possible time delay and integration time gate .tdelay ; tint / within the measured time
range. This approach has been already demonstrated for high analyte concentrations
[14.31]. However, for element concentrations lower than 1;000 g=g, the photo-
multiplier signals are too weak to be recorded. Hence, the MCI electronics was
used to sample the temporal evolution of the background-corrected analyte signal
Ianalyte .tdelay ; ıtint / using a sampling interval ıtint for a series of time delays tdelay .
14.1 Polymers 291
For each spectral emission line, repeated measurements were performed at different
time delays tdelay varied in time delay steps of ıtdelay D 1 s with a fixed sampling
interval ıtint D 1 s. The analyte signals Ianalyte .tdelay ; tint / and their empirical
standard deviations sanalyte .tdelay ; tint / are used to calculate the expected analytical
resolving power A [see (14.1)] for those .tdelay ; tint /–combinations, which can
be formed by consecutive sampling intervals. For these .tdelay ; tint / combinations,
the analyte intensity Ianalyte is calculated by summing up the sampled intensities
Ianalyte .tdelay ; ıtint /:
tdelay Ctint ıtint
X
Ianalyte
acc
.tdelay ; tint / D Ianalyte .td ; ıtint /; (14.2)
td Dtdelay
where Ianalyte
acc
.tdelay ; tint / is the analyte signal calculated for the time delay tdelay
and integration time gate tint [this intensity is called accumulated intensity in the
following to clarify the difference to the measured intensity in the sampling interval
Ianalyte .tdelay ; ıtint /], Ianalyte .tdelay ; ıtint / is the measured mean intensity in the fixed
sampling interval ıtint at delay td D tdelay . The empirical standard deviation of the
accumulated analyte signal is determined by
8 91=2
<tdelay Ct
X int ıtint
=
sanalyte
acc
.tdelay ; tint / D s 2 Ianalyte .td ; ıtint / : (14.3)
: td Dtdelay
;
In the calculation of the standard deviation, the covariance of measurements in

consecutive sampling intervals is neglected. This assumption is based on the
experimental observation that a temporally resolved photomultiplier signal of a
single plasma emission shows a mean curve which is distorted by a strong white
noise. This noise is due to statistical processes in the plasma emission, variations of
the laser peak power, variations in the absorption of the laser pulses in the specimen
and to photomultiplier noise. For single shot evaluation, the variations caused by
this noise dominate the variations in the mean signal curve which is determined
by averaging a high number of plasma emissions. For single shot evaluation, the
covariance of the measured intensities in adjacent detection windows is therefore
assumed to be very small. With Ianalyte acc
and sanalyte
acc
, the analytical resolving power A
for the parameters .tdelay ; tint / is then calculated by:
Ianalyte
acc
.tdelay ; tint / 1
A.tdelay ; tint / D D : (14.4)
sanalyte
acc
.tdelay ; tint / acc
RSDanalyte
acc
It is equivalent to the inverse relative standard deviation RSDanalyte of the accumu-
lated analyte signal. The lowest LOD is expected at maximum analytical resolving
power A.tdelay ; tint / or the lowest RSDanalyte
acc
.
The method described above allows to determine the optimum integration inter-
val for raw LIBS signals. However, using internal standardization, i.e., evaluating
the ratios of integrated line intensities of analyte and reference lines, yields higher
stability for quantitative LIBS measurements (cf. Chap. 11). Hence, it is of interest,
whether the optimum detection intervals for “raw” line intensities can also be
applied to internal standardization. In this case, the parameters .tdelay ; tint / can be
optimized individually for the analyte and the reference emission line. Internal
standardization is described by
Ianalyte
acc
.tdelay
a
; tanalyte
a
/
IOanalyte
acc
D ; (14.5)
Ireference
acc
.tdelay
r
; tanalyte
r
/
where the index “a” in .tdelay

a
; tint
a
/ denotes that these parameters belong to the analyte
line and the index “r” refers to the reference line. For the relative variance holds:
!2
sOanalyte
acc
.sanalyte
acc
/2 .sreference
acc
/2 s 2 .Ianalyte
acc
; Ireference
acc
/
O acc
RSD
2
D D C 2
IOanalyte 2 2 Ianalyte Ireference
analyte acc acc acc
Ianalyte
acc Ireference
acc
(14.6)
The first two terms on the right side of (14.6) describe the relative variance of the
integrated values for the analyte and the reference line, as they are obtained by
individual optimization of the integration time gate. The last term is the normalized
covariance of analyte and reference line, both integrated in individual integration
time gates. In case of parameter fluctuations such as laser pulse energy, absorption of
laser radiation or plasma emission, the relative standard deviation RSD O acc
analyte of the
standardized signal will be significantly lower than the relative standard deviation
acc
RSDanalyte of the raw analyte signal. This is due to the covariance of analyte and
reference signal when both reflect the influence of the parameter fluctuations in
a similar way. The dependence of RSD O acc analyte on the applied detection intervals
.tdelay ; tint / and .tdelay ; tint / is not so obvious. The first two terms on the right side
a a r r
of (14.6) vary significantly with the setting of the time delay and integration time
gate. This is evident from Figs. 14.9 and 14.10 and described in other publications
[14.32]. On the other hand, the temporal evolution of the analyte and reference line
intensity are comparable and differ mostly in parameters of the same principal,
nonnegative intensity curve. Hence, the positive normalized covariance between
analyte and reference signal is expected to vary much less with the choice of the
individual integration time gates .tdelay a
; tint
a
/ and .tdelay
r
; tint
r
/ than the sum of the
individual RSDs of analyte and reference signal. Consequently, the optimized signal
integration for internal standardization is expected to be achieved by the individual
optimized detection intervals for the analyte and the reference line.
In the experiments, the temporal evolution of the line emission was sampled for
time delays tdelay D 0:5 s up to 80 s with a fixed sampling interval ıtint D 1 s.
Five measurements with 50 repetitions each were conducted at different sample
spots using laser pulses with a pulse energy of EL D 350 mJ on high and low
concentration samples. For the investigation of the emission lines of Cd 228.80 nm,
Cr 425.43 nm, Hg 253.65 nm, and Pb 405.78 nm, the high concentration specimen
contained chigh D 1;000 g=g and the low concentration specimen contained
14.1 Polymers 293
clow D 50 g=g of the respective analyte. For the emission lines of antimony Sb
259.80 nm and bromium Br 827.24 nm, the analyte concentration in the high
concentration specimen was c D 100;000 g=g and in the low concentration
sample c D 300 g=g. Figure 14.11 shows the calculated, background corrected,
accumulated intensity Ianalyte
acc
.tdelay ; tint / for Pb 405.78 nm [see relation (14.2)] as a
function of the end of the integration time gate .tdelay C tint / for different time delays
tdelay . For simplification, this intensity is denoted in the graphs with IPb 405:78 nm . As
expected, a higher accumulated, background corrected intensity is achievable setting
earlier time delays tdelay . Increasing the integration time gate tint for a fixed time
delay tdelay yields a monotonous increase of the signal with a saturation behavior for
large .tdelay C tint /.
Figure 14.12 shows the calculated relative standard deviation of the accumulated
signal for different delays tdelay as a function of the end of the integration time gate
.tdelay Ctint /. Increasing .tdelay Ctint / reduces RSDPb 405:78 nm for every measured time
delay tdelay and hence the detection limit is improved. The lowest RSDPb 405:78 nm
for the Pb 405.78 nm emission line can be achieved for tdelay D 2 s. For earlier
signal integration with tdelay D 0:5 or 1 s, the lowest achievable RSDPb 405:78 nm
is significantly larger. This reflects the high amplitude of noise introduced by early
signal integration, which cannot be compensated for by later integral contributions
with lower amplitude. Figure 14.13 visualizes the minimum of RSDPb 405:78 nm as a
function of tdelay for a fixed end of the integration time gate of .tdelay C tint / D 80 s.
A minimum of RSDPb 405:78 nm occurs at tdelay D 2 s.
The described method was used to determine the optimum parameters .tdelay ; tint /
for all investigated emission lines and the results are listed in the column for single
pulse excitation in Table 14.6.
2000
0.5
1500 1
I Pb I 405.78 nm [a.u.]
2
3
1000 4
5
6
8
500
t delay [μs]
0
0 20 40 60 80 100
end of signal integration t delay + t int [μs]
Fig. 14.11 Calculated, background corrected, accumulated line intensity of lead Pb 405.78 nm as
a function of the end time of the integration time gate .tdelay C tint / for different time delays tdelay
5.5
tdelay [μs]
RSD acc, Pb 405.78 nm [%] 5.0 0.5
12
4.5
10
4.0
8
1
3.5
6
4
3.0 5
3
2
2.5
0 20 40 60 80 100
end of signal integration t delay+ t int [μs]
Fig. 14.12 Calculated relative standard deviation RSDPb 405:78 nm of the Pb 405.78 nm integrated
line intensity as a function of the end of the integration time gate .tdelay C tint / for different time
delays tdelay
7
RSD Pb 405.78 nm [%]
3
t delay + t int = 80 μs
2
0 4 8 12 16 20 24
time delay t delay [μs]
Fig. 14.13 Calculated RSDPb 405:78 nm of the integrated line intensity of Pb 405.78 nm as a function
of the time delay tdelay for fixed end of the integration time gate .tdelay C tint / D 80 s
For a comparative assessment of the achievable detection limits, double pulse

excitation was studied for the LIBS analysis of the elements Cd, Cr, Hg, Pb, Sb, and
Br in technical polymers (cf. Chaps. 6, 7, and 8) [14.20].
In the experiments, two separate driver units for the Pockels cell in the laser head
of the Nd:YAG laser generate the two laser pulses with an adjustable interpulse
separation of t D 18–30 s. Some authors have demonstrated that for metal
14.1 Polymers 295
Table 14.6 Detection limits according to the 3s-criterion for single and double pulse plasma
excitation and the respective parameters .tdelay ; tint / under laboratory conditions
Analyte Wavelength Excitation Single pulse Double pulse
(nm) energy
(eV)
tdelay tint LOD tdelay tint LOD
(s) (s) (g=g) (s) (s) (g=g)
Cd 228.80 5.42 1 10 19 0.8 20 11
Cr 425.43 2.91 2 98 2 0.8 20 4
Hg 253.65 4.88 1.5 9:5 18 0.8 20 22
Pb 405.78 4.38 2 48 16 0.8 20 24
Sb 259.80 5.82 2 4 50 0.8 9:2 17
Br 827.24 9.36 0.8 5 15,000 0.8 20 11,000
matrices the line intensity and the detection sensitivity could be significantly
improved for interpulse separations in the range of 0:5–80 s depending on the
matrix and the ambient gas conditions [14.13,14.33,14.34]. The signal enhancement
of the line emission observed for double pulse excitation of steel samples is only a
weak function of the interpulse separation in the interval between 2.5 and 80 s
[14.13]. If this is also valid for polymer samples is not known.
The typical pulse widths (FWHM) of the laser pulses amounts to 1 D 12 ns and
2 D 15 ns for the first and second pulse, respectively. The pulse energy ratio of
the first to the second laser pulse is denoted with E1 =E2 . The burst energy is given
by Eb D E1 C E2 . Time delays tdelay given for double pulse excitation refer to
the second pulse. Maximum plasma emission has been found for focusing the laser
s D 5 mm below the sample surface.
The temporal evolution of the emission lines was found to differ significantly
from the one in single pulse laser-induced plasmas. The spectral background of the
Pb 405.78 nm emission line at time delays tdelay < 3 s is reduced significantly using
double pulse excitation, see Fig. 14.14 in comparison with Fig. 14.9. No significant
emission could be detected for time delays tdelay 25 s, and it is evident that
the background-corrected line intensity is decreasing faster for double pulse plasma
excitation than for single pulse plasma excitation. The SNR of the Pb 405.78 nm line
as a function of the time delay is shown in Fig. 14.15. As an important difference to
single pulse plasma excitation (see Fig. 14.10), the SNR is not rising for increasing
delay times between 1 and 4 s. Both the intensity and the SNR of the Pb 405.78 nm
emission are decreasing with the time delay. This was also observed for the emission
lines Cd 228.80 nm, Cr 425.43 nm, and Hg 253.65 nm. According to the discussion
above, it is clear that since both the line intensity and the signal-to-noise ratio SNR
are decreasing with the time delay tdelay , the lowest RSDPb 405:78 nm is achieved for
the shortest time delay within the sampled time interval. Hence, the lowest detection
limit is expected for the shortest time delay. As the integration time gate tint was
found to be of lower importance for the expected detection limit, it is adjusted to
detect the significant parts of the line emission. The resulting settings for the time
delay tdelay and the integration time gate tint are listed in Table 14.6.
16000
12000
intensity [a.u.]
8000
tdelay = 1 μs
3 μs
4000
20 μs
0
405.25 405.50 405.75 406.00 406.25
wavelength [nm]
Fig. 14.14 Temporal evolution of the Pb 405.78 nm emission line for double pulse plasma
excitation with the following parameters: Eb D 250 mJ, E1 =E2 D 1, t D 30 s. Detection
with echelle spectrometer using an integration time gate of tint D 0:5 s at the stated time delays
35
30
25
SNRPb 405.78 nm
20
15
10
0
0 2 4 6 8 10 12 14 16 18 20
t delay [μs]
Fig. 14.15 Signal-to-noise ratio (SNR) of Pb 405.78 nm for double pulse plasma excitation as a
function of the time delay tdelay . Laser burst parameters as for Fig. 14.14. Detection with echelle
spectrometer using an integration time gate of tint D 0:5 s and time delays between tdelay D 1
and 20 s
The detection limits for Cd, Cr, Hg, Pb, Sb, and Br in ABS polymer matrix
with LIBS were determined using the Paschen–Runge polychromator. Fifty laser
bursts onto one spot of the specimen with a frequency of f D 6 Hz were considered
as one measurement. For each specimen, the measurement was repeated at five
different spots. For calibration internally standardized line intensities were applied.
14.1 Polymers 297
This means that for each single laser-induced plasma, the line intensities during
the detection intervals discussed above were digitized individually and internally
standardized on the 0th order which was integrated with a time delay tdelay D 1 s
and an integration time gate of tint D 10 s. From the resulting 50 internally
standardized line intensities of one measurement, outliers were removed if meeting
the Grubbs outlier criteria [14.35]. Then, the average line intensities were calculated
for this measurement. From the five mean values of the measurements taken at one
specimen, the average and the standard deviation were used for calibration.
The spatial inhomogeneity of heavy metal concentrations in polymer matrix
is a well-known phenomenon [14.36]. In order to lower the impact of spatial
inhomogeneity on the calibration, a maximum of one mean value was removed
before calibration in case it met the Nalimov outlier criteria, which is suitable for
statistical populations with N > 3 [14.37].
For each emission line, the detection limit was determined according to the
3s-criterion (cf. Chap. 11) [14.38]. Table 14.6 lists the determined detection limits
(LODs) with the detection interval .tdelay ; tint / for each investigated emission line
using a laser pulse energy of EL D 350 mJ for single pulses and a burst energy of
Eb D 350 mJ for double pulses. Example calibration curves for single pulse plasma
excitation, the respective detection limits, background-equivalent concentrations
(BEC), and mean coefficients of determination r 2 are shown in Fig. 14.16 for
cadmium and in Fig. 14.17 for bromine. The corresponding calibration curves for
double pulse plasma excitation are shown in Figs. 14.18 and 14.19.
The detection limits for Cd, Cr, Hg, and Pb are significantly lower than the
maximum allowed concentration in plastic waste (see Table 14.7, last column) and
therefore meet the requirements for EOL-WEEE recycling. The detection limits of
350
300
I Cd I 228.80 nm / I 0 th order
250
200
150 Cd I 228.80 nm
LOD3s = 19 μg / g
100
BEC = 81 μg / g
50 r2 = 0.992
0 200 400 600 800 1000

Cd concentration [μg / g]
Fig. 14.16 Calibration curves for Cd 228.80 nm referenced to 0th order using single pulse plasma
excitation with optimized time delay and time integration gate, see Table 14.6. Laser pulse energy:
EL D 350 mJ
700
600
I Br I 827.24 nm / I 0th order
500
400
Br I 827.24 nm
LOD3s = 15 mg / g
300 BEC = 172 mg / g
r2 = 0.961
200
0 50 100 150 200 250 300
Br concentration [mg / g]
Fig. 14.17 Calibration curve for bromine Br 827.24 nm referenced to 0th order using single pulse
plasma excitation with optimized time delay and time integration gate, see Table 14.6. Laser pulse
energy: EL D 350 mJ
900
800
I Cd I 228.80 nm / I 0th order
700
600
500
400 Cd I 228.80 nm
300 LOD3s = 11 μg / g
BEC = 45 μg / g
200 r2 = 0.995
100
0
0 200 400 600 800 1000
Cd concentration [μg / g]
Fig. 14.18 Calibration curve for cadmium Cd 228.80 nm referenced to 0th order using double
pulse plasma excitation with optimized time delay and time integration gate, see Table 14.6. Laser
burst parameters: Eb D 350 mJ, E1 =E2 D 1:2, t D 18 s
single and double pulse plasma excitation are comparable. This could be due to
the fact that interpulse separations lower than 18 s were not achievable with the
applied laser system. Further investigations with a two laser setup might give more
insight into this topic. For Br 827.24 nm, a detection limit of around 1.5 m.-% was
achieved in ambient air. Although the detection limit is significantly higher than
for other elements, it is sufficient to detect common concentrations of brominated
flame retardants (BFR) in plastics which typically exceed 3–5 m.-%. Furthermore,
14.1 Polymers 299
1000
900
800
I Br I 827.24 nm / I 0th order
700
600
500
Br I 827.24 nm
400 LOD3s = 11 mg / g
300 BEC = 78 mg / g
r2 = 0.998
200
100
0 50 100 150 200 250 300
Br concentration [mg / g]
Fig. 14.19 Calibration curves for Br 827.24 nm referenced to 0th order using double pulse
plasma excitation with optimized time delay and time integration gate, see Table 14.6. Laser burst
parameters: Eb D 350 mJ, E1 =E2 D 1:2, t D 18 s
Table 14.7 Limits of Element (nm) Dynamic focusing Required LOD

detection (LOD) of the LIBS LOD (g=g) (g=g)
analyzer for the investigated
elements in moving ABS Cd 228.80 96 100
samples Cr 425.43 73 100
Hg 253.65 60 100
Pb 405.78 140 100
Sb 259.80 80 100
Br 827.24 n.d. 5,000
LOD calculated by the 3s-criterion
n:d: D not determinable
the detection limit of 17 g=g for Sb I 259.80 nm with double pulse laser excitation
allows to detect very low concentrations of the BFR synergist Sb2 O3 which is
commonly added along in concentrations of several mass percent to enhance the
flame retarding effect.
On the basis of the laboratory results, an inline LIBS analyzer was setup to
provide the required sensitivity for heavy metal detection in polymers. The task
of the LIBS analyzer is the rapid quantification of heavy metal and halogen
concentrations in EOL-WEEE pieces moving at a speed of 0.5–1.0 m/s on a belt
conveyor. The major problem for the application of LIBS to inline analysis of
real waste EOL-WEEE pieces such as monitors, telephones, and keyboards is the
large variation in size and shape. In contrast to laboratory experiments, no well-
defined measuring distance to the surface of the specimen is guaranteed and large
variations of the measuring distance for successive samples must be compensated
by the LIBS analyzer. For this task, an autofocusing unit for LIBS was setup to keep
the analyzing laser beam focused on the sample surface, see Fig. 14.20.
Fig. 14.20 Setup of the inline LIBS analyzer. PC D personal computer, PR D Paschen–Runge
polychromator, G D grating, PM D photomultiplier, PM-ZO D photomultiplier at 0th-order,
MCI D multichannel integrator electronics, F D fiber optics, P D plasma, S D sample, MR D
measuring range, DM D dichroic mirror, M D mirror, AF D autofocusing unit, AF-TS D
triangulation sensor, AF-C D controller, AF-FO D focusing optics
In the autofocusing unit, a laser triangulation sensor (Cyclop 220, nokra GmbH,
Germany) measures the distance between the focusing optics and the sample surface
with an accuracy of ˙70 m and a frequency of 50 Hz. The autofocus control
processes the triangulation sensor output and generates the control signal for the
variable focal length optical system (VarioScan, SCANLAB, Germany with an addi-
tional external f D 200 mm lens) with a mean focal length of about f D 220 mm.
An example for the signal stabilization achieved with the autofocusing unit is
shown in Fig. 14.21. For rising distances of the sample surface to the detecting fiber,
both the Pb 405.78 nm and the 0th-order signal show the expected intensity decrease.
However, the Pb 405.78 nm signal internally standardized to the 0th-order signal
is constant within 5% over a range of 50 mm. This result could not be achieved
with static focusing. In Fig. 14.22, the internally standardized Pb 405.78 nm line
intensity acquired with static focusing is compared with dynamic focusing for
varying distances between the focusing optics and the sample.
Using static focusing, the LIBS signal shows a strong dependency on the
deviation from the optimum distance to the sample surface. This dependency cannot
be compensated by internal standardization which complies to earlier publications
[14.39]. In contrast, dynamic focusing can provide standardized line intensities
which are constant within 5% over a range of 50 mm.
To further reduce the measuring distance variations due to the shapes of the
waste pieces, the waste pieces pass the LIBS analyzer on a tilted belt conveyor
sliding on the side panel. The LIBS measurement is carried out through a gap in the
conveyor side panel, see Fig. 14.23. However, while the waste pieces pass the LIBS
analyzer on the belt conveyor, the focusing optics has to compensate for the distance
variations due to the individual surface topology of the waste pieces.
14.1 Polymers 301
8000 0.30
I Pb 405.78 nm , I 0th order [a.u.] SD x 3
0.25
6000
0.20
4000 I 0th order 0.15
I Pb 405.78 nm
0.10
2000
0.05
0 0.00
190 200 210 220 230 240 250
distance between focusing optics and sample [mm]
Fig. 14.21 Intensity of Pb 405.78 nm, 0th-order and ratio hPb I 405:78 nm=0th orderi as a func-
tion of the sample distance for a fixed fiber position using the autofocusing unit. Plasma excitation
with double pulse laser in air on ABS polymer sample containing 2;000 g=g lead. The error bars
given by the standard deviation (SD) are shown enlarged by a factor of 3
0.30
dynamic focusing
0.25
I Pb 405.78 nm / I 0th order
0.20
0.15
static focusing
0.10
190 200 210 220 230 240 250

distance between focusing optics and sample [mm]
Fig. 14.22 Comparison of Pb 405.78 nm LIBS signal internally standardized to 0th-order for static
and dynamic focusing
During inline measurements, double pulse laser bursts with a total burst energy
of Eb D 300 mJ, a pulse energy ratio of E1 =E2 D 1:2, a pulse separation of t D
30 s, and a repetition rate of 30 Hz are irradiated onto the EOL-WEEE pieces. The
emission of the laser-induced plasma is picked up by a fiber optics at a distance of
130 mm to the far end of the measuring range and is guided to a Paschen–Runge
polychromator (cf. Sect. 4.2). The photomultiplier signals are integrated during an
integration time gate of tint D 20 s after a delay of tdelay D 0:8 s with respect to
the second pulse of the laser burst.
Fig. 14.23 Schematic setup of the sensor head (right) at the belt conveyor (cross section shown
on the left). R D roll of conveyor system, S D sample, W D window, M D mirror, WD D wedge
to compensate the beam offset, LB D laser beam for LIBS, AF D autofocussing unit, TS D
triangulation sensor
The calibration of the inline LIBS analyzer was performed with reference
samples moving on the belt conveyor at three different measuring distances.
Three repeated measurements comprising 20 laser bursts were performed at each
measuring distance. The limits of detection (LOD) for the investigated heavy metals
and halogens determined with the inline LIBS analyzer operating at the pilot sorting
plant are shown in Table 14.7.
At the pilot sorting plant, the required detection limit of 100 g=g is achieved
for the elements Cd, Cr, Hg, and Sb. The LOD of 140 g=g for Pb 405.78 nm was
slightly higher than the target. In general, it is evident that the inline LIBS analyzer
shows a lower sensitivity for all investigated spectral lines during the inline tests
compared with the laboratory experiments. This is attributed to the additional dis-
turbing influences during operation at the pilot sorting plant. The detection limit for
Br 827.24 nm was higher than the available concentrations of the reference samples
used for calibration and therefore the respective LOD could not be determined.
Hence, the detection of brominated flame retardants had to be accomplished by the
detection of Sb in the BFR synergist Sb2 O3 .
The inline classification performance of the LIBS sensor was evaluated with
160 measurements on 40 real waste monitors. The surfaces of the monitor hous-
ings did not show any lacquer and were not cleaned before measurements. The
reference analysis for the concentration of heavy metals and halogens in these
monitor housings was provided by inductively coupled plasma atomic emission
spectrometry (ICP-OES) [14.19]. In the real waste EOL-WEEE monitor housings,
no mercury concentration >1 g=g was detected with the ICP-OES reference
analysis. Hence, the classification performance of the LIBS analyzer cannot be
quantified for mercury. For the sorting of the monitor housings, the determined
element concentrations were classified into four categories: (a) 0–100 g=g, (b)
100–1;000 g=g, (c) 1;000–2;000 g=g, (d) >2;000 g=g. For lead, the upper
concentration limit of category (a) had to be increased to 140 g=g due to the fact
that the LOD of Pb 405.78 nm was 140 g=g. The results of the inline classification
obtained for monitor housings without surface preparation moving on the belt
conveyor with a velocity of about 0.5 m/s are summarized in Table 14.8. For Cd,
14.2 Slag 303
Table 14.8 Results of the inline LIBS analyzer at the pilot sorting plant with real waste EOL-
WEEE monitor housings moving at a velocity of 0.5 m/s
Classifications Cd Cr Hga Pb Sb
Number of correct classifications 135 152 (160) 152 152
Number of total classifications 160 160 160 160 160
Percentage of correct classifications (%) 84 95 (100) 95 95
a
no mercury concentration > 1 g=g was found in the real waste EOL-WEEE test samples with
the ICP-OES reference analysis
Cr, Pb, and Sb, high classification ratios of up to 95% were achieved during the
inline measurements at the pilot sorting plant.
14.2 Slag
Improvement of steel qualities is one object of the steel producing industry. For
example, concentrations below 10 g=g of sulfur and 100 g=g of phosphorus are
required to achieve the desired material properties of the pre-products. Hence,
the goal is to continuously improve the metallurgical processing steps from the
pre-product to the final product to achieve well-defined material qualities within
narrow tolerances for the specified element concentrations. The composition of a
steel melt is influenced by chemical reactions of the melt with slag components.
Therefore, the chemical analysis of the slag provides essential information for a
metallurgical process control. Thus, one important issue concerning steel production
is the analysis of slag from the converter and the ladle. For the future, an increasing
demand for slag analysis is expected to further optimize the metallurgical process
guiding. For an efficient process control, the total time needed for slag analysis has
to be reduced to less than 3 min.
XRF spectroscopy is a state-of-the-art measuring technique for slag analysis
[14.40]. The slag taken in the plant is transported to a laboratory by, e.g., a
pneumatic dispatch system. The time necessary for this procedure in an off-line
laboratory is given in Table 14.9. Therein, it can be seen that nearly 40% of the total
time is necessary for sample preparation and sample transportation. However, the
analytical results during steel production are in most cases required within a short
time scale of a few minutes, since the next processing step must be controlled on
the basis of this analytical information. Thus, there is an increasingly strong need to
simplify or even avoid those steps of sample transportation or sample preparation.
Laser-based analytical methods will play a key role for the development of inline
methods in metallurgical processing, since they offer noncontact measurements
regardless of the material to be analyzed, high measuring speed, and sample
preparation by the laser beam itself (cf. Sect. 13.1.2).
Several papers deal with basic LIBS investigations of oxidic materials such as
inorganic minerals, concrete, soil, synthetic silicates [14.41–14.44]. First studies of
slag analysis from blast furnace and liquid steel processes using LIBS demonstrated
the principal suitability of this approach for process control purposes [14.45]. In
Table 14.9 Comparison of times needed for sampling, sample transportation, sample preparation,
and analysis for the methods XRF and LIBS
Time needed for Time needed for LIBS
XRF-analysis (min) analysis (min)
Sampling 2.0 2.0
Sample transportation 0.5 –
Sample preparation 2.0 –
Analysis time (including sample 2.0 0.5–2.0
conditioning for LIBS analysis)
Total time 6.5 2.5–4.5
Since for LIBS a mechanical sample preparation is not necessary there is the option to position
a laser analyzing system directly in the neighborhood of the metallurgical aggregate in the plant,
thus avoiding a sample transportation to a laboratory
the following the analysis of converter slags with LIBS will be presented where a
multivariate calibration model was applied for the first time [14.31].
14.2.1 Converter Slags

Before a measurement starts, the sample surface is irradiated by a number of pre-
pulses to ablate a small amount of material from the surface, which may be not
representative for the bulk composition (cf. Sect. 11.1, Table 11.1). The following
laser bursts and the laser-induced plasma emission produced thereby are used
for analysis. The necessary time for the surface preparation by the prepulses and
the subsequent analysis is denoted with the term “analysis time” in Table 14.9.
The comparison with the XRF standard analysis method shows that a significant
improvement of the total time can be achieved.
The goal of the investigations described in the following was to provide a method
based on LIBS for the quantitative analysis of slag samples from steel production.
The tasks to be solved include the issue of sample inhomogeneity, as well as the
issue of matrix effects due to the wide ranges of analyte concentrations in slag
samples.
Figure 14.24 shows the experimental setup [14.31]. A Q-switched Nd:YAG laser
operating at 1,064 nm was used to excite the plasma. The Q-switch electronics was
equipped with a double-pulse option to generate within a single flashlamp pulse up
to two separate laser pulses instead of a single laser pulse (cf. Sect. 3.2, Chaps. 7 and
8). The flashlamp repetition rate is 20 Hz. The laser emits within a single flashlamp
discharge two laser pulses of approximately the same energy. The width of each
pulse amounts to about 16 ns (FWHM). The time separation between the first and
the second pulse is t D 31 s. The burst energy (cf. Sect. 3.2) amounts to Eb D
210 mJ. The laser beam is guided via mirrors to a plano-convex lens with a focal
length of 200 mm.
The converging laser radiation propagates via an optical window into a measure-
ment chamber (cf. Sect. 4.6). This chamber has a cover plate on top with a circular
14.2 Slag 305
Fig. 14.24 Schematic setup for LIBS of slag samples. M D mirror, L D lens, W D window,
C D measurement chamber, S D sample, EM D eccentric mechanism, LC D laser control, IE D
integrator electronics, SY D signal cable for synchronization, SP D spectrometer, PC D personal
computer
opening having a diameter of 12 mm. The sample is put onto this cover plate to
perform a measurement. During the measurement, the sample is fixed to the cover
plate by a holder. The cover plate including the holder and the sample can be moved
relative to the measurement chamber in a plane perpendicular to the optical axis of
the laser excitation by an eccentric mechanism (cf. Sect. 3.6, Figs. 3.20 and 3.21). In
this way, the sample is shifted relative to the laser beam axis on a circular trajectory
with a diameter of typically 2 mm. Thus, a spatial averaging can be performed. The
focus of the laser beam was adjusted in that way that the focal plane lies inside the
sample with a distance of s D 12 mm from the sample surface.
The optical axis of the VUV vacuum spectrometer describes the line between the
center of the diffraction grating and the middle of the entrance slit. This optical axis
intersects the propagation axis of the laser beam on the surface of the sample. The
angle between the surface of the sample and the optical axis of the spectrometer
amounts to ˛D 11ı (detection angle, cf. Fig. 4.18). With this orientation, the
radiation from the laser-induced plasma can be coupled directly into the entrance
aperture of the spectrometer. The internal volume of the chamber is flushed with
argon 4.8 at 1.4 atm. and a flow rate [cf. Sect. 4.6, (4.7)] of about VPin D 520 l=h.
with a holographic diffraction grating with 2,400 lines/mm (cf. Sect. 4.2, Fig. 4.7
and Table 4.4). The optical system of the spectrometer is installed inside a vacuum
vessel which is evacuated to <102 mbar. The dispersed radiation passing the exit
slits aligned along the Rowland circle is detected by photomultiplier tubes (PMT).
A signal electronics to simultaneously process up to 64 PMT signals in a pulse-by-
pulse-mode (cf. Sect. 4.5) is triggered by the Pockels cell synchronization monitor
of the laser, see connection SY in Fig. 14.24. The digitized charges of the integrators
for each laser pulse are stored and processed in a personal computer.
Table 14.10 Concentration Analyte Slag from converter Slag from ladle
ranges of analytes of slag concentration concentration
samples from converter and (m.-%) (m.-%)
ladle
CaO 43–64 30–64
SiO2 4–20 0–17
Fetot 6–34 0–5
Mn 2–6 0–0.5
MgO 0.7–11 1–10
Al2 O3 0.5–2.0 15–42
TiO2 0.1–0.6 0–2.5
P2 O5 1.0–4.0 0.0–0.1
S 0.05–0.2 0.5–1.5
For the quantitative analysis, the number of pre-bursts applied to prepare the
sample surface amounts to Npp D 200 and the number of measuring bursts is Nmp D
600. Based on the setup shown in Fig. 14.24, an analyzing device was designed and
operated in the laboratory of a steel plant for industrial evaluation.
Two different types of slag samples were investigated. The liquid slag is taken
by a special sample probe where the liquid slag flows in a steel ring and solidifies
in a glassy state. This sample can be directly analyzed with LIBS. This type of
samples is called in the following “samples directly from converter.” In order to
suppress the influence of any inhomogeneities of these samples for the LIBS method
development, a second type of samples were studied. In this case, solidified slag
material is crushed and milled to obtain a powder, which is then pressed into a steel
ring. The samples are manufactured by the use of an automatic press. Thereby, a
constant pressure is applied for all samples. Thus, even though not experimentally
verified, a similar sample density should be obtained for all samples. This type of
sample is called “homogenized sample.”
Table 14.10 shows typical concentrations of the analytes of slag samples from
the converter and the ladle. All elements vary in a range of at least 10% and there
is no single element whose concentration remains approximately constant. These
variations of analyte concentrations result in matrix effects.
The influence of the varying composition of the slag samples on the properties
of the laser-induced plasma was studied in terms of the radiation emitted by the
plasma. The radiation reflected into the zeroth order of the grating was detected
as a measure of the total radiation emitted from the plasma within the spectral
sensitivity range of the spectrometer. Figure 14.25 shows the intensity of the zeroth
order signal for different converter slag samples S1–S7 taken from a production
line. Each data point corresponds to the average of ten measurements performed on
the respective sample. If not otherwise stated, a single measurement comprises 200
pre-bursts and 600 measurement bursts. The error bars correspond to the standard
deviation of the ten measurements. Obviously there are significant variations of the
intensity of the zeroth order for the different samples. The standard deviation within
the ten measurements on every sample is by a factor of 3.4 smaller than the standard
deviation of the seven mean intensities from the seven samples.
14.2 Slag 307
1000
900
intensity I 0th order [a.u.]
800
700
600
500
S1 S2 S3 S4 S5 S6 S7
sample no.
Fig. 14.25 Intensity of the zeroth order for seven different homogenized samples taken from
the converter. Each data point represents the average of ten measurements under repeatability
conditions. The error bars show the standard deviation of the ten measurements taken at one sample
Comparisons with experimental parameters, e.g., the laser burst energy for every
measurement did not show any correlation to the observed variation of the plasma
emission from sample to sample. Since these results are gained with homogenized
samples – i.e., the slag material was chrushed, milled, and pressed – a variation of
the density of the pressed samples, which may cause the observed variability in the
intensity of the zeroth order, seems not to be probable. Therefore, it is assumed that
the variations are caused mainly by matrix effects. In order to find out which element
concentrations are favorable for an increasing intensity of the zeroth order, the data
of Fig. 14.25 are analyzed with a partial least square algorithm (PLS) using the con-
centrations of the analytes in the samples determined with conventional methods.
Figure 14.26 shows the determined loadings for the different analytes of the
measurement data of Fig. 14.25. Thereby, a positive loading p1 indicates an increas-
ing intensity of the zeroth order with an increasing concentration of the respective
analyte. A negative loading p1 corresponds to a decreasing intensity of the zeroth
order with an increasing concentration of the respective analyte. No correlation
was found between the behavior shown in Fig. 14.26 and physical properties of the
analytes such as ionization energy, boiling temperature, and evaporation enthalpy.
Obviously, the variations in the intensity of the plasma emission in terms of the
zeroth order radiation are caused by matrix effects.
A series of experiments were performed with varying measuring parameters to
minimize the influence of matrix effects and to optimize the sensitivity as well
as the stability of the measurement for the quantitative determination of analyte
concentrations.
The investigated measuring parameters are: the laser burst energy, the position of
the focal plane with respect to the sample surface s, the time delay and gate width
0.4
0.2
loading p1
0.0
-0.2
-0.4
analytes
Fe CaO SiO2 Mn TiO2 Al2O3 MgO P2O5 S
Fig. 14.26 Influence of the different analyte concentrations on the intensity of the zeroth order
intensity described as loadings p1 deduced from a data analysis by a partial least square algorithm
taking the known analyte concentrations and the measured zeroth order intensity. Analytes
showing positive p1 values cause an increase of the zeroth order intensity with increasing analyte
concentration, whereas negative loadings p1 describe the opposite behavior
of the integration window for the individual element signals, the spatial averaging
and the gas flushing rate of the measurement chamber. In the following, the influence
of the time delay and gate width will be presented.
Figure 14.27 shows the intensity of the silicon line at 390.55 nm as a function
of time recorded with a fast oscilloscope connected directly to the output of the
corresponding photomultiplier tube. The time zero coincides with the irradiation
of the second laser pulse of the double pulse burst. Three different time domains
can be distinguished, denoted with the numbers 1, 2, and 3 in Fig. 14.27. By
laterally shifting the entrance slit of the spectrometer while performing a series of
measurements the spectral profile of this emission in the time domains 1, 2, and
3 was investigated. The intensity of the element emission was integrated in a time
window of length tint starting at tdel with respect to point zero. In the time domains 1
and 2 no distinct line profile is visible indicating that this emission is dominated by
a superposition of continuum radiation and strongly broadened line emissions (cf.
Figs. 2.4 and 15.6), whereas in the time domain 3 with tdel 5 s and tint D 3 s a
clear line profile is detected with a spectral width (FWHM) of about 70 pm.
To find an optimum integration window an approach analogous to the one
described in Sect. 14.1.2 was applied. The intensity of several emission lines
as a function of time was recorded similar to Fig. 14.27 for two samples with
high and low concentrations of the analytes to be determined. The samples were
selected in such a way to cover the whole range of concentrations occurring in
slag samples taken from a production line. The reference concentrations of each
sample were determined with XRF analysis. Each sample was measured at three
14.2 Slag 309
t del
1 2 3
1.5
intensity I Si 390.55 nm [a.u.]
1.0
tint
0.5
0.0
0 4 8 12 16 20 24
time [μs]
Fig. 14.27 Photomultiplier signal of the Si 390.55 nm channel as a function of time recorded with
a fast oscilloscope (the negative signal voltage of the PMT was converted to a positive signal for
this graph). Three different time domains are denoted with the numbers 1, 2, and 3, see text. tdel D
delay time with respect to the start of the PMT signal, and tint D width of the integration window
different locations. At each location, 50 pre-bursts and 100 measuring bursts were
applied. The signals were digitized and numerically integrated within defined time
windows. The position and width of the integration windows were varied and
for each parameter combination the mean and the standard deviation of the three
measurements per sample were calculated. The results were assessed with respect
to the value of the following target function:
Ii =ci
Ti D ; (14.7)
hs.Ii /i
where Ii is the mean intensity of an emission line i for the sample containing the
analyte at high concentration minus the mean intensity of the same emission line
for the sample containing the analyte at low concentration, ci is the difference
of concentrations of the analyte i in the two samples, and hs.Ii /i is the average
standard deviation of the intensity of analyte i for the two samples.
The target function (14.7) increases with higher sensitivity and lower standard
deviation. This corresponds to a reduction of the uncertainty in the determination
of a concentration using the analysis function. As an example, Fig. 14.28 shows
the calculated target function TSi 390:55 of the silicon line at 390.55 nm for the
numerically varied parameters: (a) start of integration tdel and (b) end of integration
(tdel C tint ). At tdel D 3:8 s and tdel C tint D 5:1 s, the target function TSi 390:55
attains a maximum.
In order to further take into account the uncertainty of the analytical results,
the target function was extended in the case of linear calibration functions, by
20 target function
TSi 390.55 [a.u.]
18
16
-6.5
start of integration tdel [μs]
14 not defined -5.3

-4.2
12
-3.0
10 -1.9
8 -0.75
0.40
6
1.6
4 2.7
3.9
2
5 10 15 20 25 30
end of integration tdel + tint [μs]
Fig. 14.28 Target function TSi390:55 for the silicon line 390.55 nm in dependence on start of
integration tdel and end of integration (tdel C tint )
multiplying the numerator of the right side of equation (14.7) with the square of the
correlation coefficient, i.e., the coefficient of determination. A series of parameter
studies were performed and the results assessed using this target function yield
the settings used in the following. Thereby, if possible, mutual interferences of the
different parameters were taken into account.
With this set of parameters the experimental setup was calibrated for the analytes
listed in Table 14.10 using different approaches: (a) univariate calibration with
no internal standardization, i.e., only the measured mean line intensities of the
analytes are used (denoted with “no internal standard” in Fig. 14.29); (b) univariate
calibration with internal standardization: the measured line intensity is divided by
the measured intensity of a single emission signal, e.g., the zeroth order signal, a
background signal or another emission line (denoted with “internal standard.” in
Fig. 14.29, cf. Sect. 11.1); (c) univariate calibration where a sum of emission signals
is used for internal standardization (denoted with “sum of internal standards” in
Fig. 14.29); and (d) multivariate calibration (denoted with “multivariate calibration”
in Fig. 14.29). The theory and the implementation of multivariate calibration models
is described elsewhere [14.46]. In general, all integrated signals are chosen as
input channels. The input channels span a vector space. The calibration model
determines the direction in the vector space, which correlates best with the reference
concentrations of the calibration samples. This direction is used for a linear
14.2 Slag 311
3.5
standard deviation
3.0
<RSD(c)> [rel.-%]
mean relative
2.5
2.0
1.5
1.0
1.00
0.95
determination <r 2>
mean coefficient of
0.90
0.85
0.80
0.75
0.70
0.65
no internal internal sum of internal multivariate

standard standard standards calibration
Fig. 14.29 Mean relative standard deviation <RSD.c/> (top) and mean coefficient of determina-
tion r 2 (bottom) for the analytes listed in Table 14.10 and homogenized samples from converter
slag for different calibration procedures, see text
calibration function. The residuum, namely the variations in the dataset not included
in this first linear calibration of the model, can be used as new input data for a further
linear calibration. The further calibration constitutes a so-called further factor of the
calibration model. Thereby, a model with several factors is implemented.
Figure 14.29 summarizes the results showing the mean relative standard devi-
ation of the determined concentrations and the mean coefficient of determination
for the above-described different calibration procedures for homogenized sam-
ples from converter slag. The mean values refer to the average for all analytes
investigated. Figure 14.29 shows that for the univariate calibration without internal
standardization the relative standard deviation amounts to 3.4% and the coefficient
of determination is about 0.65. With internal standardization, the results could be
improved significantly. However, the mean relative standard deviation could not be
pushed down to less than 2%. Obviously, the matrix effects could not be suppressed
completely by the selection of the measuring parameters using the optimization
procedure described before. Figure 14.29 shows that a multivariate calibration is
capable to reduce the relative standard deviation of the concentration as a mean
value of all elements to 1.8%. Furthermore, the coefficient of determination can be

increased to 0.95 as a mean value for all elements.
Figures 14.30 and 14.31 show measurements from a period of over 2 weeks,
whereby the analytes Fe and CaO were determined with LIBS and XRF during shift
operation with the LIBS analyzing device in the laboratory of the steel plant. The
concentration determined with LIBS is plotted as a function of the concentration of
the same sample measured with XRF. For the graphs shown on the left of Figs. 14.30
and 14.31, an univariate calibration model is used for LIBS, whereas for the graphs
on the right a multivariate calibration model was applied. Besides the improved
results, as shown in Fig. 14.29 for one set of calibration measurements, it can be
seen that also an improvement in long-term stability can be achieved by the use
of multivariate calibrations. For multivariate models, the repeatability for Fetot and
CaO is 0.28 m.-% and 0.40 m.-%, respectively, for a routine analysis on a long-
term basis. The corresponding residual standard deviation of the fit – which is a
measure for the trueness of the determined concentration – are 0.85 m.-% for Fetot
and 1.05 m.-% for CaO.
The results shown were obtained with the homogenized samples in contrast to
the samples taken directly from the converter. This was done to determine the
univariate multivariate
26
26
24
24
22
concentration Fe LIBS [m.-%]
concentration Fe LIBS [m.-%]
20 22
18
20
16
18
14
16
12
10 14
10 12 14 16 18 20 22 24 26 14 16 18 20 22 24 26
concentration Fe XRF [m.-%] concentration Fe XRF [m.-%]
Fig. 14.30 Concentration of Fe determined with LIBS as a function of the concentration

determined with XRF for homogenized slag samples from the converter. The measurements were
performed in a period of over 2 weeks during shift operation. Left: univariate calibration model,
right: multivariate calibration model
14.2 Slag 313
univariate multivariate
60 60
55
55
concentration CaO LIBS [m.-%]
concentration CaO LIBS [m.-%]

50
50
45
40 45
35
40
35 40 45 50 55 60 40 45 50 55 60
concentration CaO XRF [m.-%] concentration CaO XRF [m.-%]
Fig. 14.31 Concentration of CaO determined with LIBS as a function of the concentration
determined with XRF for homogenized slag samples from the converter. Left: univariate calibration
model, right: multivariate calibration model
performance of the investigated LIBS procedure and setup without an additional

influence originating from the inhomogeneity of the samples.
Table 14.11 summarizes the results obtained for samples taken directly from the
converter and for the homogenized samples. In general, the standard deviation of
the measured concentrations s.c/ is higher for the samples taken directly from
the converter. With the exceptions of MgO and P2 O5 , the trueness is a factor
3.4 to 7.2 poorer than the repeatability for the homogenized samples. This is, if
compared to standard XRF results, within the expectations of a routine analysis. The
repeatabilities for the main elements of slag from the converter, namely CaO, SiO2
and Fetot , on the homogenized samples in terms of the relative standard deviation
amounts to 0.6%, 1.5%, and 0.7%, respectively.
In order to further improve these results, different measuring parameters could
be used for the different analytes. In this case, a more specific optimization for the
different elements is possible. With an extended spatial averaging of the sampled
material, the repeatability of the results for the samples taken directly from the
converter can be further improved (cf. Sect. 14.2.2).
By optimizing the measurement parameters as well as using multivariate calibra-
tion models, the relative standard deviation of the concentrations of all elements or
analytes could be improved from 6.1% to 1.8%. The average RSD(c) for these main
Table 14.11 Summary of results gained with LIBS on converter slag samples
Element/ Standard Standard Relative Trueness, Coefficient of
analyte deviation, deviation, standard homoge- determination,
samples homoge- deviation, nized homogenized
directly from nized homoge- samples samples r 2
converter samples nized R.c/
s.c/ (m.-%) s.c/ (m.-%) samples (m.-%)
RSD.c/ (%)
CaO 0.34 0.31 0.6 1.05 0.995
SiO2 0.51 0.18 1.5 0.84 0.997
Fetot 0.8 0.14 0.7 0.85 0.998
Mn 0.08 0.036 0.9 0.24 0.81
MgO 0.07 0.027 0.5 0.48 0.88
Al2 O3 0.03 0.029 2.3 0.21 0.89
TiO2 0.05 0.007 2.0 0.029 0.88
P2 O5 No measurements 0.013 0.5 0.12 0.75
S No measurements 0.009 7.2 0.048 0.87
Average 0.27 0.08 1.8 0.43
values
The relative standard deviation RSD(c) in column 4 is calculated with the values given in column
3 and the mean concentration of the respective analyte given in Table 14.10 for slag samples from
the converter
analytes CaO, SiO2 , and Fetot is less than 1% (see column 4 in Table 14.11). The
average RSD(c) for all analytes amounts to 1.8%.
Additionally, further investigations have to be performed to find out the factors
influencing the homogeneity of the slag samples and to improve the sample taking
procedure to obtain more homogeneous slag samples from the converter and from
the ladle.
The developed LIBS-based analyzing device was installed and set into operation
in a steel plant for industrial evaluation. It allows to increase the number of samples
to be analyzed within a given metallurgical processing step. This analyzing device
is the first approach to bring LIBS into routine use for the analysis of slags from
steel production.
14.2.2 Vacuum Slag
In steel works commonly, aluminum and/or silicon are used for an efficient
deoxidation of steel melts where solid alumina and silica are formed. In order
to obtain a liquid ladle slag, usually lime will be added as a slag forming agent.
Furthermore, lime addition enhances the ability of ladle slag to absorb sulfur from
the steel melt by the reaction [14.47]:
ŒS C .CaO/ C 2=3 ŒAl D .CaS/ C 1=3.Al2 O3 /: (14.8)

14.2 Slag 315
Long experience with steel desulfurization indicates that a maximum removal of

sulfur from liquid steel can be obtained only by the use of lime saturated slags
[14.48]. Therefore, a fast analysis of the major slag components calcium oxide
CaO, silicon dioxide SiO2 and aluminum oxide Al2 O3 can improve the control of
the steel melt and reduce the treatment times in the vacuum degasser. The mass
fractions of the major components of the vacuum slags are in the range of 50–60%
for CaO, 0.5–12% for SiO2 , and 20–40% for Al2 O3 . Samples are taken from the
liquid slag layer at the surface of the steel melt by an automatic immersion slag
sampler. Although great improvements are achieved in the reproducibility of the
sampling from the moving liquid slag layer (thickness 5–12 cm, slag mass 2–4 t)
in the vacuum degasser, the slag samples exhibit a significant heterogeneity in the
species distribution as well as varying color, cracks, and ferrous oxide inclusions.
After solidification, the slag samples are dismantled from the immersion sampler.
Laboratory analysis requires at least 7 min including dismantling, pneumatic deliv-
ery of the sample from the steel works to the laboratory, crushing of the sample, and
XRF analysis of a pressed pellet. A more precise analysis by a borate bead analysis
takes additional time.
The goal of the investigations presented in the following is to demonstrate the
feasibility of an onsite slag analysis by LIBS in the steel works without additional
sample preparation and to reduce the analysis time significantly [14.49].
LIBS has been applied for slag analysis under laboratory conditions and/or for
crushed and pressed samples [14.31, 14.45, 14.50]. The spatial averaging of the
LIBS measurement by an optical microlens array is described in Sect. 3.6 (cf.
Fig. 3.22) [14.51]. This is a key item due to the heterogeneity of the production
slag samples without mechanical preparation. Another topic is the installation of a
LIBS equipment in the harsh environment of a steel works (e.g., dust, temperature
variation, stability of electricity, argon gas, and water supply) and its stable long-
term operation (cf. Sect. 17.2).
The slag samples are taken by immersion samplers from the vacuum degasser
during steel production. The color of the slag samples varies between dark gray,
anthracite to pale gray and mixtures of brown-yellow-green-white. A majority of
the sample surfaces are matt, but some samples are glossy. A variety of cracks and
holes are present in the samples. Both sides of the samples can be used for analysis.
In order to find a representative average value for a sample, the standard analysis
in the laboratory of the steel works is crushing and milling the slag material
of the samples and pressing it to pellets for XRF analysis. In special cases, the
samples are prepared as fused borate beads. For LIBS, a laser beam is focused to a
diameter in the range of 20–500 m. In the focal region, a laser-induced plasma
is generated, which is used for spectrochemical analysis. Therefore, an efficient
spatial averaging has to be applied to measure areas in the range of 1 cm2 . In
this study, the spatial shaping of a Nd:YAG laser beam (wavelength 1,064 nm) by
an optical microlens array (hexagonal packed microlenses of diameter of 0.9 mm,
f D 37 mm) as described in Sect. 3.6 is used. Approximately 40 high-intensity
spots are generated at the sample surface within the Nd:YAG laser beam diameter of
6 mm, see Fig. 14.32. In addition, the sample is moved eccentrically with the sample
Fig. 14.32 Schematic setup of the laser beam optics, sample stand, and Paschen–Runge vacuum
spectrometer (SP). ES, OA D entrance slit and optical axis of the spectrometer, LB D laser beam;
M D mirror; LA D microlens array; W1 D window; W2 D window, AR coated at 1,064 nm;
W3 D vacuum window of spectrometer, MgF2 ; P D laser-induced plasma, S D slag sample;
Ar 1, 2, 3 D inlets and outlet for argon gas flushing of the sample stand. The optical components
are uncoated unless otherwise noted
plate of the argon flushed sample stand within the focal plane of the microlens array
resulting in a surface area of 12 mm diameter covered by the LIBS measurement.
The LIBS instrument includes a vacuum Paschen–Runge spectrometer with
photomultiplier tubes mounted at different positions for simultaneous spectral line
detection and a Nd:YAG laser (pulse repetition rate rep D 20 Hz, pulse duration
L 8 ns, pulse energy 170 mJ). The LIBS plasma of the sample is generated in an
argon flushed sample stand and the emitted light is directly coupled to the spectrom-
eter observing the plasma light under an angle of about 12ı to the sample surface.
The calibration is carried out by production samples of vacuum slag. For
reference mass fractions w (in percent or equivalent 102 g/g D cg/g), the samples
are analyzed directly by X-ray fluorescence as well as by borate bead analysis
(BBA). Both analysis are carried out at the certified laboratory of the steel plant.
For the borate bead analysis, the samples are crushed, and iron granules are removed
by a magnetic separator or sieve. The powder material is then melted with borate
at a furnace temperature of about 1;250ıC. Finally, fused beads were obtained by
casting on Ni-plates and distinct cooling.
In Figs. 14.33 and 14.34, analysis functions for silicon dioxide SiO2 are shown.
The radiant intensity of the spectral line Si I 251.4 nm, ISi 251:4 nm , detected for
each laser pulse by the corresponding PMT, is divided by the radiant intensity
of the zeroth order, I0th order , detected by a PMT mounted at the zeroth order
position of the spectrometer grating (Dposition of reflected beam, cf. Fig. 4.7).
Thus, small variations in the parameters (e.g., of laser, argon flush, sample and
optical component positions) which are proportional to both signals are eliminated
for the ratio ISi 251:4 nm =I0th order of the signals. For each sample, the average
hISi 251:4 nm =I0th order iNmp D800 is determined [cf. (11.4)].
The coefficient of determination is greater than 0.97 which is considered as
good taking into account the sample variation and heterogeneity as well as the
14.2 Slag 317
10
mass fraction BBA wSiO2 [cg / g]
2
r 2 = 0.973
0
0.2 0.4 0.6 0.8 1.0 1.2 1.4
 LIBS
Fig. 14.33 Analysis function of SiO2 with reference mass fractions by borate bead analysis (BBA)
with a calibration set of 26 vacuum slag samples. The LIBS analysis is taken at one side of each
sample. r 2 D coefficient of determination of the fitting curve (second-order polynomial)
8
mass fraction XRF wSiO2 [cg / g]
2
r 2 = 0.991
0
0.2 0.4 0.6 0.8 1.0 1.2
 LIBS
Fig. 14.34 Analysis function of SiO2 with reference mass fractions by XRF with a calibration set
of 37 vacuum slag samples. Both measurements are taken at the same sample side
measurement environment. The slightly better correlation to the XRF reference

values in Fig. 14.34 compared with the borate bead analysis of Fig. 14.33 are
possibly due to the fact that the BBA analysis gives a mean value of the total sample
while the LIBS analysis is taken at one side of the sample. The relative difference of
the mass fractions between side 1 (mass fraction w1 / and side 2 (mass fraction w2 / of
the samples has been measured by XRF and it is in average h.w1 w2 /=w1 i D 1:9%
8
mass fraction XRF wSiO2 [cg / g]
2
r 2 = 0.982
0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
 LIBS
Fig. 14.35 Analysis function of SiO2 with reference mass fractions by XRF with a calibration set
of 37 vacuum slag samples. Same measurement as Fig. 14.34, except that the spectral line Si I
390.6 nm is evaluated. r 2 D coefficient of determination of the linear fitting curve
(rel.) for SiO2 with a maximum difference of 9% (relative; determined for a set of
37 samples).
The spectral line Si I 251.4 nm exhibits a stronger saturation behavior than
the spectral line Si I 390.6 nm as shown by comparison of Figs. 14.34 and 14.35
which are taken during the same calibration measurements. The stronger saturation
of the spectral line Si I 251.4 nm compared with Si I 390.6 nm is expected by
the approximately ten times higher kt coefficient according to (9.33) of Sect. 9.4.
Assuming a plasma temperature of 9,000 K, the kt coefficients are calculated with
spectral line data to 8:9 1030 m3 for Si I 251.4 nm and 0:85 1030 m3 for Si I
390.6 nm, respectively.
The analysis functions of the ratios w.CaO/=w.SiO2 / and w.Al2 O3 /=w.SiO2 /
are given in Figs. 14.36 and 14.37 versus the intensities of the spectral lines Ca I
364.4 nm and Al I 309.3 nm referenced to ISi251:4 nm [cf. (11.2)]. The coefficient
of determination is greater than 0.97 which is considered as good under these
conditions.
The analysis of unknown samples is carried out with a multivariate analysis
[14.52]. This approach as well as the industrial application of the LIBS analysis
of vacuum slags will be described in Sect. 18.2.
14.3 Soil
One of the main advantages of LIBS for chemical analysis purposes is the possibility
to perform contactless measurements at solid substances without causing high
demands on the sample preparation [14.53, 14.54]. This is especially promising
14.3 Soil 319
60
50
w(CaO) / w (SiO2) BBA
40
30
20
10 r 2 = 0.987
0
1 2 3 4 5 6
 LIBS
Fig. 14.36 Analysis function of CaO=SiO2 with reference mass fractions by borate bead analysis
(BBA) with a calibration set of 26 vacuum slag samples
50
w(Al2O3) / w(SiO2) BBA
40
30
20
10
r 2 = 0.972
0
1 2 3 4 5 6 7 8 9
 LIBS
Fig. 14.37 Analysis function of Al2 O3 =SiO2 . Same parameters as Fig. 14.36
for the analysis of heavy metals in soils, because no chemical dissolution of

sample material is required [14.55, 12.9], which in turn is the case for conventional
techniques such as inductively coupled plasma optical emission spectrometry
(ICP-OES) or atomic absorption spectrometry (AAS). Consequently, numerous
investigations were performed in this field, ranging from studies on the influence
of basic experimental parameters [14.56] to the evaluation of LIBS probes for the in
situ subsurface detection of metals in soils using a cone penetrometer [14.57,14.58].
For most practical applications of analytical techniques, it is required that the
analytical figures of merit are sufficient with respect to regulatory demands. In the
case of the analysis of heavy metals in soils, the limits of detection (LOD) for the
various elements should at least fall below the corresponding maximum content
that is allowed for unpolluted soils. In Germany, these values are listed as so-called
Z0-values in [14.59]. The limits of detection reported in the literature for LIBS
measurements of heavy metals in soils are typically in the range of some g/g
to some 10 g/g [14.53, 14.56, 14.60–14.62]. This is sufficient for most of the
elements, but for some substances such as Cd (Z0 D 0:6 g=g), Hg (0:3 g=g)
and Tl (0:5 g=g) detection limits below 1 g=g are required.
An approach to achieve improved limits of detection using laser-based analytical
techniques is the combination of LIBS with laser-induced fluorescence (LIF,
also called laser-excited atomic fluorescence spectrometry, LEAFS) [12.1]), see
Chap. 12.
The aim is to develop a technique that is suited for the fast in situ analysis
of heavy metals in soils. Consequently, as many elements as possible should be
measured simultaneously, a requirement which can be fulfilled using LIBS. For
two of those elements that cannot be determined with limits of detection below
the regulatory limits using LIBS, namely Cd and Tl, the LIBS-LIF approach was
studied. The experimental setup is described in Chap. 12.
For the LIBS measurements, a compact sample stand was used instead of
a measurement chamber, cf. Sect. 4.6, Fig. 4.18. The soil samples have to be
positioned on the top of this sample stand, and the LIBS laser beam is focused
onto the sample surface from the bottom. The geometrical arrangement for the
LIBS excitation and the observation of the plasma radiation is the same as for the
measurement chamber, see Fig. 12.2. As a drawback, the compact sample stand does
not allow to direct a LIF excitation laser beam into the plasma, nor does it allow to
move the samples using a motorized translation stage. On the other hand, the volume
of the gas atmosphere inside the sample stand is significantly smaller than the gas
volume of the measurement chamber. This allows to have a fast exchange of the
buffer gas in the compact sample stand during the measurements by a continuous gas
flow, thus reducing the danger of accumulative effects [see also Sect. 4.6, relation
(4.7) and Fig. 4.17].
For the calibration of the LIBS-LIF technique with respect to the analysis of
heavy metals in solid soil samples, a set of samples containing the substances
of interest over a broad range of concentrations was needed. As it is difficult
to find certified reference material (CRM) with these characteristics, such a set
of calibration samples was artificially produced by spiking standard soils from
the Landwirtschaftliche Untersuchungs- und Forschungsanstalt (LUFA) in Speyer,
Germany, with the heavy metals. The standard soil was dried, ground to a grain
size < 0:1 mm, and suspended with metal–salt solutions. Afterwards, the sample
was dried and homogenized again. The operating cycle for the preparation of the
samples as well as the metal salts used for spiking are described in [14.63]. For
reference, a portion of the samples was analyzed using the standard techniques
ICP-OES or AAS after dissolution in aqua regia. The metal contents that were
determined with this analysis were used as reference values for the calibration
curves shown in the following. Finally, the sample material was mixed with a small
14.3 Soil 321
amount of wax and pressed to pellets, so that an additional reference analysis using
EDXRF spectroscopy (Spectro X-Lab 2000) was made possible with typical limits
of detection in the low ppm-range for the heavy metals and proven accuracy by
standard reference materials.
The applicability of LIBS in its own for the fast in situ analysis of heavy metals
in soils was investigated. For this application, the Paschen–Runge spectrometer with
photomultiplier (PMT) detection was deployed (cf. Sect. 4.2, Fig. 4.8).
The influence of experimental parameters such as the laser pulse energy EL and
the position of the laser focus in relation to the sample surface s was studied.
These experiments were performed on a compact sample stand, on which the
soil samples had to be positioned, instead of the measurement chamber shown in
Figs. 12.2 and 12.3. The experiments were carried out in argon as buffer gas at
atmospheric pressure.
The dependency of the analytical figures of merit of the LIBS measurements
on the time delay tdelay between plasma formation and the start of the integration
of the LIBS signals with the multichannel integrator electronics was investigated.
The duration of the integration window tint was adjusted to tint D 20 s according
to the overall lifetime of the LIBS signal observed for the strong emission of the
matrix element silicon at 288.2 nm. The gateable multichannel integrator electronics
used for these experiments allows to integrate 16 channels with one integration
window and 4 channels with another. As one of the integration windows was
intended to be used for the integration of LIF signals in LIBS-LIF measurements,
all elements that were to be analyzed using LIBS had to be integrated over the
same time window. Thus, a compromise had to be found for the combination
of the parameters: laser pulse energy EL , focus position s, and time delay
tdelay , which resulted in acceptable limits of detection for as many heavy metals
as possible. For the chosen experimental setup, an optimum with respect to this
demand was found for EL D 80 mJ, s D 5 mm (i.e., the laser focus lies inside
the sample), and tdelay D 1 s. The optimization of the conditions to a particular
element allowed the improvement with respect to the limit of detection by about a
factor of 2 or more. For example, an integration delay of tdelay D 2 s resulted in an
improved limit of detection of LOD3s D 3 g=g for cadmium instead of 6 g=g with
tdelay D 1 s.
Calibration curves were recorded by measuring the LIBS signals for a set of
soil samples prepared as described above. For each sample, 100 laser pulses were
measured at three different positions on the sample surface. The number of laser
pulses applied to one position on the sample surface was limited to 100 to make
sure that the crater depth, which is approximately 2 mm for 100 laser pulses with a
pulse energy of 80 mJ, does not reach the backside of the sample. As a result of the
granular structure of the samples, soil particles can be found in the vicinity of the
crater after a measurement, which are supposed to be ejected out of the interaction
volume by the pressure of the laser-induced plasma. In consequence, only an upper
limit for the amount of sample material which is atomized and therefore actually
provided for the analysis can be estimated roughly. This estimate for the ablated
mass per laser pulse – calculated from the crater volume generated by 100 laser
pulses applied to the same sample position, the average mass density of the sample
and the number of laser shots – amounts to 9 g per laser pulse.
Two of the calibration curves obtained in this way are depicted in Fig. 14.38.
As artificially spiked soil samples were used for the experiments, matrix effects
were minimized. Thus, it was possible to directly plot the LIBS signal without
any internal standardization against the analyte concentrations determined by
conventional analytical techniques for calibration purposes. The error bars shown in
the curves are given by the standard deviation of the three measurements performed
under reproducible conditions.
a
300
250
I Cu 324.8 nm [a.u.]
200
150
100
Cu 324.8 nm
50 LOD3s = 3.3 μg / g
0 r 2 = 0.9977
0 100 200 300 400 500 600 700 800

copper concentration [μg / g]
b
80
60
I Zn 334.5 nm [a.u.]
40
20 Zn 334.5 nm
LOD3s = 98 μg / g
0 r 2 = 0.9887
0 250 500 750 1000 1250 1500 1750 2000 2250

zinc concentration [μg / g]
Fig. 14.38 LIBS calibration curves for (a) copper and (b) zinc. The analyte signal measured for
the corresponding spectral line is plotted against the analyte concentration without using internal
standardization. The ordinate is not corrected for a negative offset caused by the signal processing
electronics. The calibration curves are second-order polynomials fitted to the experimental data
14.3 Soil 323
A slight saturation behavior was observed for the LIBS signals for many spectral
lines with increasing analyte concentration, which was covered by the soil samples.
This can be seen in Fig. 14.38a for Cu as well as for Zn in Fig. 14.38b. This
behavior can be attributed to self-absorption of the element-specific line emission
from the plasma at high analyte concentrations, which is supported by the fact
that the nonlinearity is higher for resonance lines than for nonresonance lines,
but additional effects such as limitations of the dynamic range by the detection
and signal processing scheme cannot be excluded [cf. Sect. 9.4, relation (9.33)].
In consequence, one would have to select different spectral lines or to adjust the
gain of the signal electronics, if an increased range of sensitivity is necessary to
determine high analyte concentrations in certain applications. However, the goal of
the investigations described is to achieve a high sensitivity for trace concentrations
[14.44]. A second-order polynomial was fitted to the experimental data as a
calibration curve. From this curve, the limit of detection was calculated by LOD3s D
3s0= m, where s0 is the standard deviation of the three replicate measurements for
the least concentrated sample and m is the slope of the calibration curve for low
analyte concentrations (cf. Sect. 11.2). Calibration curves for all relevant heavy
metals were evaluated in this way, resulting in the limits of detection listed in
Table 14.12. Additionally, the spectral lines that were used for the calibration and
the coefficients of determination of the calibration curves are given in this table. The
results are illustrated in Fig. 14.39, where the regulatory limits for unpolluted soils
are shown for comparison. The detection limits achieved with LIBS are sufficient
for six elements, namely As, Cr, Cu, Ni, Pb and Zn, but have to be improved for Cd,
Hg, and Tl. Experiments using air as ambient gas instead of argon have been carried
out, indicating that similar limits of detection can be achieved.
For two elements, Cd and Tl, investigations on the applicability of the hyphenated
LIBS-LIF technique were performed. A description of the setup and result for
cadmium is given in Chap. 12. A LOD3s of 0:3 g=g was achieved for Cd which
lies below the regulatory limit of 0:6 g=g, cf. Fig. 14.39.
Table 14.12 Limits of detection (LOD3s / for the analysis of heavy metals in soils using LIBS
Element Wavelength of the spectral Limit of detection Coefficient of determination
line (nm) LOD3s (g=g) r2
As 235.0 3.3 0.9912
Cd 228.8 6 0.9993
Cr 425.4 2.5 0.9936
Cu 324.8 3.3 0.9977
Hga 253.7 84 0.9858
Ni 231.6 6.8 0.9838
Pb 405.8 17 0.9863
Tl 351.9 48 0.9905
Zn 334.5 98 0.9887
The wavelength of the spectral line () used for the calibration is given as well as the coefficient
of determination of the calibration curve (r 2 /
a
Linear calibration based on three soil samples
arsenic
cadmium
chromium
copper
lead
mercury
nickel
thallium
zinc
range of calibration samples
limit for unpolluted soils (LAGA-Z0, Germany)
LOD3s - LIBS
LOD3s - LIBS-LIF
0.1 1 10 100 1000

analyte concentration [μg / g]
Fig. 14.39 Illustration of the detection limits (LOD3s ) achieved with LIBS and LIBS-LIF
for the analysis of soils at atmospheric pressure in comparison with German regulatory limits
for unpolluted soils (Z0). The bars depict the range of analyte concentrations covered by the
calibration samples
LIBS-LIF measurements for Tl were performed in an air atmosphere at 950 mbar,

for the practical reason that this experimental condition is even more promising
for an implementation of the technique in an in situ measurement system than an
analytical technique depending on argon as buffer gas. A variation of the LIBS-LIF
delay tLIBS-LIF showed that distinct LIF signals can be observed for at least 250 s.
An example for the time behavior of the photomultiplier signal at 351.9 nm is given
in Fig. 14.40. The signal was averaged over 30 laser shots. In the upper part (a), the
signal for a soil sample containing 827 g=g Tl is shown. The LIBS signal following
the continuum emission after the plasma ignition can be observed, lasting for about
10 s, as well as the strong LIF peak after 20 s. The overshoot of the LIF signal
to negative values after the excitation of laser-induced fluorescence is attributed
to bandwidth limitations of the signal processing electronics. For comparison, the
signal observed at the same spectral line is shown for a soil sample containing less
than 0:3 g=g of thallium Tl, which was the limit of detection of the analytical
technique, which was used for the reference analysis. Neither an element-specific
LIBS signal, nor any LIF signal can be observed in this case (note the different
scaling of the y-axis in Fig. 14.40).
Figure 14.41 shows the calibration curve that was recorded for a LIBS-LIF delay
of tLIBS-LIF D 20 s. The detection parameters were the same as for Cd (tdelay LIF
D
0:5 s, tint D 5 s). Similarly, a strong saturation behavior is observed. As the
analyte concentration range was essentially larger than for Cd and thus the saturation
behavior much more evident, an exponential saturation function was fitted to the
experimental data instead of a second-order polynomial. The slope of this curve for
14.3 Soil 325
a
1.0
823 μg / g Tl
0.8
PMT-signal [a.u.]
0.6
0.4
plasma ignition LIF-signal
0.2
0.0
-0.2
-0.4
b
0.4
< 0.3 μg / g Tl
0.3
PMT-signal [a.u.]
0.2
0.1
0.0
-5 0 5 10 15 20 25 30 35
time [μs]
Fig. 14.40 Photomultiplier (PMT) signal of the thallium channel at 351.9 nm for LIBS-LIF
measurements in air at atmospheric pressure, averaged over 30 laser pulses. The LIF excitation
wavelength was tuned to the Tl transition at 276.8 nm. While distinct LIBS- and LIF-signals can
be observed for a soil sample containing 823 g/g of Tl (a), only the LIBS continuum peak is
found for a sample containing less than 0:3 g=g of Tl (b)
low analyte concentrations was calculated to determine the detection limit. In this
case, a value of LOD3s D 0:5 g=g was achieved, matching the regulatory limit for
unpolluted soils.
For further investigations on the influence of different experimental parameters
such as the LIBS-LIF delay tLIBS-LIF on the analytical figures of merit of the
technique with respect to the analysis of Tl in soils, additional calibration samples
are required to expand the experimental data in the range of analyte concentrations
below 100 ng/g.
200
150
LIF-signal [a.u.]
100
Tl, 351.92 nm
50
LOD3s = 0.5 μg / g
BEC = 15.7 μg / g
0
0 250 500 750 1000 1250 1500 1750 2000 2250

Tl concentration [μg / g]
Fig. 14.41 LIBS-LIF calibration curve for thallium. The analyte signal measured is plotted against
the analyte concentration measured with ICP-OES. The calibration curve is a fit of an exponential
saturation function to the experimental data
14.4 Cement
The analysis of building materials is of vital importance to ensure structural safety

and durability of existing constructions. By using reliable analytical methods,
damages can be detected and their origins can be determined. The analytical results
are the basis for the restoration planning. Currently concentrations for different
salts, in particular chlorides, that have a negative influence on the stability of
concrete constructions, are determined by wet chemical methods. These methods
require an extensive sample preparation and are therefore time-consuming and
expensive [14.64, 14.65]. Because of their complexity, this procedure is prone to
faults. Additionally only one element within a pulverized drill core can be analyzed
and the result is an average value representing the whole sample. Hence only a small
number of wet chemical analysis are carried out in practice, resulting in higher risk
of undetected stability and durability problems.
LIBS was studied for the determination of trace elements in building materials
such as concrete and mortar by several groups [14.66, 14.67]. Weritz et al. studied
the influence of the heterogeneity of concrete consisting of cement, fine and coarse
aggregates and additives on the quantitative analysis of sulfur [14.66]. They could
show that the uncertainty in the sulfur calibration can be improved by selecting
data with a high Ca/O ratio indicating less heterogeneous constituents such as
cement and cement mortar. Burakov et al. studied chlorine analysis in cement mortar
(cement, fine-grained sand, gypsum, 1:2:1) using the IR chlorine line 837.59 nm
[14.67]. They compared a LIBS setup with a combination of LIBS and an electrical
discharge. The energy stored for the latter is in the order of 3 J; however, it seems
14.4 Cement 327
that only a part of this energy is dissipated in the laser-induced plasma indicated by
the shown voltage and current oscillations (cf. Fig. 1 in [14.67]). For the combined
excitation, an increase in the chlorine signal by a factor of 11 on average was
observed. A chlorine detection limit of 0.05% in cement mortar is stated for the
combined excitation scheme using a 2s criterion. A calibration curve for laser
excitation only required the use of internal standardization with the oxygen line
O I 844.54 nm and averaging over 2,500 laser pulses per data point.
The aim of the investigations described in the following was to improve the limit
of detection of chlorine in hydrated cement via the application of LIBS that is stated
with 0.5 m.-% using He at a pressure of p D 1;000 mbar and the chlorine line
at D 837:59 nm [14.68]. To increase the line emission of chlorine in the laser-
induced plasma, several methods were studied by various authors. One of these
methods is the use of an additional process gas like argon, air, nitrogen, oxygen, or
helium [14.69,14.70]. Tran et al. showed that the signal-to-noise ratio of the chlorine
spectral line at D 837:59 nm can be improved in comparison with normal air at
atmospheric pressure by a factor of 12 if a helium gas flow is used at atmospheric
pressure to analyze solid organic compounds [14.71]. A further parameter studied
to increase the chlorine emission is the ambient pressure. Asimellis et al. studied
helium at pressures between p D 0 and 600 mbar inside a chamber to analyze
NaCl-powder mixed up with CH3 COONa and Na2 SO4 using a laser at a wavelength
of D 355 nm [14.72]. At a pressure of about p D 50 mbar, they observed a
maximum of the chlorine line emission at D 837:59 nm. In a further paper of
Asimellis et al., the detection of bromine in pressed powder samples consisting of a
mixture of KBr and K2 CO3 was studied in a helium atmosphere with laser excitation
at 355 nm [14.70]. A calibration curve for bromium ranging from 1 to 33.6 m.-%
was determined at a helium pressure of 30 mbar. A limit of detection was not stated.
The authors declare that an optimum He pressure appears to be in the range of 60–
100 mbar.
Weritz et al. conducted measurements with different types of building materials
under a 1 bar He atmosphere [14.68]. They estimated the limit of detection
for chlorine in cement and mortar moving the sample continuously during the
measurements to achieve an average value. The best limit of detection for chlorine
in hydrated cement is stated with LOD D 0.5 m.-% using the 3s criterion.
LIBS-based detection of chlorine in artificial soil and geological samples was
studied using the Cl line at 133.57 nm in a Mars atmosphere, i.e., in CO2 gas at
pressures below 10 torr [14.73]. The stated detection limit of Cl at 0.25 torr CO2 is
7.5 m.-%, which is by far too great for the purpose stated in the introduction.
In the following investigations are described where the chlorine lines at D
837:59 nm – and for the first time – D 134:72 nm were studied to assess their
analytical properties in a comparative way [14.74]. To detect the chlorine line
in the UV, a direct light-channel is used to guide the plasma emission from
the measurement chamber into the spectrometer. The total pressure inside the
measurement chamber is varied between 10 and 1,000 mbar while the total gas inlet
flow rate amounts 1.5–50 l/min. To reduce the LOD for chlorine in hydrated cement
different settings for the integration time gate of the photomultiplier (PMT) signals
tdelay and tint are investigated and compared between the two chlorine lines.
Cement samples with different chlorine concentrations are used as calibration
set for LIBS measurements. The set comprises seven cement samples with different
contents of chlorine: 0.05, 0.2, 0.5, 0.8, 1, 2, and 4 m.-%.
The experiments are carried out with a Nd:YAG laser system operating at a
wavelength of L D 1;064 nm. The pulse duration amounts to L D 8:5 ns and the
repetition rate is L D 50 Hz. The laser pulse energy in the experiments ranges
between Eb D 450 mJ and Eb D 480 mJ. Figure 14.42 shows a schematic view of
the experimental setup. The laser beam is guided into a measuring chamber passing
an antireflecting coated window and is focused by a f D 500 mm lens. The chosen
pulse energy level range and focal length are significantly greater than those values
used in [14.67, 14.71, 14.72] (these authors used a focal length <150 mm and pulse
energies <200 mJ) due to the following reason. The intended application is the
analysis of concrete drill cores. These cores have to be split to obtain a fracture
surface unaffected by the water flushing used during drilling. The water flushing
may lead to wash out effects of the species to be analyzed. However, the fracture
surface is uneven due to the included aggregates. Hence, the depth of focus in terms
of the Rayleigh length (about 12.5 mm, cf. Sect. 3.4) of the focused laser beam has
to be sufficiently large to assure a nearly constant irradiance. For the experiments
described in the following plane samples were used.
The angle between the incident laser beam and the normal of the sample surface
is ˛ D 12ı (cf. Sect. 3.9, Fig. 3.26). The measuring chamber has a net volume of
Fig. 14.42 A schematic view of the experimental setup. L D laser system, M1–M2 D mirrors,
BE D beam expander, L1 D lens (f D 50 cm), W1–W2 D windows, GI1–GI3 D gas intake,
GO D gas outlet, P D plasma, S D sample, SH D specimen holder, LS D linear stages, MC D
measuring chamber, L2–L3 D MgF2 lenses (each f D 20 cm), SP D spectrometer, VP D vacuum
pump
14.4 Cement 329
50 l. Pressures between p D 10 and 1,000 mbar can be set. Inside the chamber
two linear stages are mounted to move the samples without venting and opening
the chamber. The emission of the laser-induced plasma propagates through a direct-
light channel to a Paschen–Runge spectrometer (cf. Sect. 4.2). Inside this channel
two MgF2 lenses (L2 and L3 in Fig. 14.42) each with a focal length of f D 201 mm
at D 525 nm collimate and couple the plasma light into the spectrometer. An
additional magnesium fluoride MgF2 window (W2 in Fig. 14.42) is used inside the
channel between the chamber and the lenses to seal the chamber and to protect
the MgF2 -optics of getting contaminated by ablated particles. The space inside the
direct light channel between the lens L3 and the MgF2 -window (W2) is flushed
with argon to avoid absorption in the UV due to residual oxygen gas or water vapor
in the channel (cf. Sect. 3.5, Fig. 3.14). Between D 121:5 and 921.5 nm a total
number of 55 spectral lines, including background channels can be detected by the
Paschen–Runge spectrometer and a multichannel integrator electronics (MCI, cf.
Sect. 4.5) being capable to integrate the channeltron signals in programmable time
gates. Windows W1 and W2 are flushed during a measurement by a total helium
gas flow rate of 1.5–50 l/min depending on the chosen pressure inside the chamber.
The gas passes the surfaces of the optics and is guided onto the sample surface.
By this setup, a contamination of the optical path by ablated particles between the
plasma at the surface of the samples and the spectrometer is kept at a sufficient
low level. A vacuum pump is connected to the measurement chamber to achieve
a constant pressure inside the chamber while the optics are flushed. The pump
evacuates the chamber and the pressure inside the chamber is set by the gas flow
that can be adjusted by two valves. The gas flows for the entrance window and
the MgF2 window can be controlled separately. For all measurements, the ratio
of gas flow rates at W1, W2 was set to 1:10 because of the greater aperture of
the direct-light channel. The position of the beam waist was set underneath the
sample surface s > C1:5 cm (cf. Chap. 2, Fig. 2.2 and Chap. 11, Table 11.1). The
diameter of the laser beam at the sample surface amounts to Db > 800 m. During
all measurements, helium is used as process gas.
To be able to estimate the detection limit for chlorine in cement, samples with
different chlorine concentrations are needed. Cement powder is mixed with water
that contains a specific concentration of NaCl to achieve the wanted chlorine
concentration in the sample. The cement water mixture is filled into plastic
bottles. During the hardening process, the bottles are rotated constantly to ensure
a homogenous distribution of chlorine in the sample. After some days, the plastic
bottle is removed and the hydrated cement is divided into smaller pieces by a saw.
The samples have a diameter of Ds D 4:5 cm and a thickness of Ts D 1:5 cm,
see Fig. 14.43. The samples have many pores and by this they can take up to 8%
of moisture due to the interaction with the ambient air, which was determined by
weighing of the samples before and after a drying cycle in an oven heated to 60ı C.
Single and collinear double pulse configurations are studied at different ambient
pressures. The burst energy ranges from Eb D 450 mJ to Eb D 480 mJ and the
pulse energy ratio is set to E1 W E2 D 1 W 1 in case of double pulses (cf. Sect. 3.2).
The interpulse separation is varied between t D 1 and 8 s. In a first step, the
Fig. 14.43 Photograph of a

hydrated cement sample that
is used for the measurements.
There are several craters
visible on the sample surface
that were generated by the
laser bursts
delay time tdelay and integration time gate tint have to be optimized for the detection
of chlorine lines. Integration time gates reported so far vary between tint D 10 and
20 s to detect the emission of the chlorine line D 837:59 nm using helium as inert
gas to analyze organic compounds or pressed disks containing pure NaCl powder
[14.71, 14.72]. Lancelin et al. used a delay time of tdelay D 2 s and an integration
time gate of tint D 20 s to study the signal-to-noise ratio of the chlorine 837.59 nm
line of different gas mixtures containing chlorine compounds at ambient pressure
[14.75]. No studies were published so far concerning the temporal evolution of the
chlorine 134.72 nm line.
To determine the optimal delay time tdelay and the integration time gate tint ,
the signals of the channeltrons inside the spectrometer are detected by an oscil-
loscope with an input impedance of 50 . Figure 14.44 shows the difference of
the channeltron signals for the chlorine line 134.72 nm between a sample with a
natural basic concentration of about 0.05 m.-% .Ul / and a sample with a high
chlorine concentration .Uh / in cement as a function of time. About 100 ns after
the interaction of the laser pulse with the sample surface, a distinct maximum
of the difference is observed. The corresponding signal difference of the chlorine
spectral line in the IR shows a similar behavior. In the following the starting point
of the integration time gate of the MCI is set to tdelay D 100 ns. An integration
time gate of tint D 10 s was chosen, where the signals of both channeltrons
for the IR and UV chlorine line reach the base level. To reduce the influence of
a possible inhomogeneous species distribution inside the calibration samples, the
mean intensity of 20 subsequent spectra at 9 different positions are averaged to one
value for each element detected by the spectrometer. The measuring positions are
arranged in a grid of 10 mm 10 mm, where each measuring point has a distance of
d D 5mm to the adjacent point.
To investigate the influence of the gas pressure on the plasma emission, the
pressure is varied between p D 10 and 320 mbar where the line intensity enhance-
ment effect for the spectral line is expected [14.72]. Figures 14.45 and 14.46
show the signal-to-noise ratio of the chlorine lines in the IR and UV as a function
of the helium pressure. Regarding the intensity for the Cl line at D 837:59 nm
the ratio ICl 837 nm =IBg 861 nm is increasing for both samples with growing pHe from
1.6 at pHe D 10 mbar to 1.9 for the 2 m.-% sample and to a value of 2.3 for the
14.4 Cement 331
0.01
1E-3
Ul - Uh [V]
1E-4
1E-5
1E-6
0 200 400 600 800
t [ns]
Fig. 14.44 The difference of the channeltron signals for a sample with a high chlorine concen-
tration (4 m.-%) and a low concentration (0.05 m.-%) for the spectral line at D 134:72 nm as a
function of time (t D 0 ns corresponds to the time of the laser pulse irradiation) at pHe D 100 mbar
2.2
I Cl 837 nm / I Bg 861 nm
2.0
1.8
2 m.-%
1.6 4 m.-%
0 50 100 150 200 250 300 350

p He [mbar]
Fig. 14.45 Referenced spectral line intensity of the chlorine line D 837:59 nm as a function of
the helium pressure for two cement samples with different chlorine concentrations
4 m.-% sample at a pressure of pHe D 75 mbar. A further increase in the pressure

up to p D 320 mbar does not significantly change the signal ratio. Regarding
the pressure regime between pHe D 75 and 320 mbar, the signal ratio seems to
level off at a plateau. A different development is observed for the chlorine line
at D 134:72 nm, see Fig. 14.46. At pHe D 10 mbar for both Cl concentrations
different signal-to-noise ratios exist. By increasing the pressure both ratios increase.
For the 4 m.-% sample, a maximum is attained at pHe D 40 mbar with a ratio of
ICl 134 nm= IBg 861 nm D 2:5: For the 2 m.-% sample, the maximum is located at about
2.6
2 m.-%
I Cl 134 nm / I Bg 861 nm 2.4 4 m.-%
2.2
2.0
1.8
1.6
1.4
0 50 100 150 200 250 300 350

p He [mbar]
Fig. 14.46 Referenced spectral line intensity of the chlorine line D 134:72 nm as a function of
the helium pressure for two cement samples with different chlorine concentrations
pHe D 80 mbar. For pressures higher than 100 mbar both ratios decrease reaching
1.7 for the 4 m.-% sample and 1.4 for 2 m.-%. The found maxima for the 134.72 nm
line in the He pressure range 40–80 mbar coincide with those reported by Asimellis
et al. detecting the Cl-line at D 837:59 nm. However, for the chlorine line in the
IR a significant decrease for pressures >75 mbar was not observed as found for
the chlorine line in the UV. For a sufficient signal even for lower Cl concentrations
in cement .<0:5 m.-%) a pressure of pHe D 60 mbar is chosen for the following
measurements. This pressure is located between the observed maxima of the IR and
UV chlorine lines.
The electron density of the plasma was measured using an echelle spectrometer
to detect the wavelength range between D 200 and 780 nm. The Stark broadened
Ca II line at D 315:89 nm is used to estimate the electron density ne (cf. Sect. 9.4).
The profile of the spectral line is fitted by a Voigt profile and the Lorentzian and
Gaussian parts are determined by deconvolution. The Lorentzian full width at half
maximum (FWHM) of the line width is used to calculate the electron density
[14.76]:
ne
1=2 D 2w 17 ; (14.9)
10 cm3
where w is the half width at half maximum (HWHM) electron Stark parameter for
the Ca II spectral line (w D 29:2 pm), 1=2 the FWHM of the line profile, and ne
is the spatially integrated electron density of the plasma [14.77].
Figure 14.47 shows the electron density as a function of the pressure pHe . At
pHe D 5 mbar, the electron density amounts to ne D 4:2 1016cm3 . In the range of
pHe D 5–60 mbar, the increase in the electron density can be described by a linear
regression curve with a coefficient of determination of r 2 D 0:99. In the regime
between 120 and 1,000 mbar, the slope of the electron density increase is reduced
14.4 Cement 333
3.0 × 1017
2.5 × 1017
electron density ne [cm-3]
2.0 × 1017
1.5 × 1017
1.0 × 1017
5.0 × 1016
error bars × 5
0 200 400 600 800 1000

pressure p He [mbar]
Fig. 14.47 Electron density for single pulses as a function of the helium pressure pHe inside the
chamber
and at 1,000 mbar the electron density reaches a maximum of ne D 2:3 1017 cm3 .
At the observed maximum of the chlorine lines for an ambient pressure of about
pHe D 60 mbar – see Figs. 14.45 and 14.46 – the electron density attains a value of
ne > 1 1017 cm3 .
After a suitable pressure regime for the chlorine signal was found, the influence
of the integration time gate on the limit of detection is studied by varying the
integration time gate tint between 1.25 and 40 s. Figure 14.48 shows the chlorine
line intensity 837.59 nm for different integration time gates as a function of the
chlorine concentrations. The slope is not as high as in the case of the chlorine line
in the UV, cf. Fig. 14.49. For an integration time gate of tint D 1:25 s, the dynamic
range of the counts between 0.05 and 4 m.-% amounts only 900 counts. An
increase in the integration time gate up to 10 s results in an increased background
that is indicated by a shift of the spectral line intensity for 0.05 m.-% from 8,200
to 10,000 counts. The dynamic range for the lowest and highest concentration
varies between 9,500 and 17,000 counts. The increase in the detection time by a
factor of 8 results in an increase in the dynamic range by a factor of 8. A similar
case can be seen for the atomic chlorine line at D 134:27 nm in Fig. 14.49.
For an integration time gate tint D 1:25 s, the signal varies between 4,600 and
6,600 counts, corresponding to a dynamic range of 2,000 counts for chlorine
concentrations between 0.05 and 4 m.-% in cement. An increase in the integration
time gate by a factor of 8 up to tint D 10 s results in a range of 5,500–15,000
counts. By this the dynamic range rises nearly by a factor of 5. A further increase in
the detection time up to 20 and 40 s does not significantly improve the maximum
intensity neither the dynamic range.
The influence of the integration time gate on the limit of detection is studied
applying the calibration curve method of the standard DIN 32 645 to estimate
20000
parameter t int
18000
1.25 μs
2.5 μs
16000
I Cl 837 nm [a.u.]
5 μs
10 μs
14000 20 μs
40 μs
12000
10000
8000
0 1 2 3 4
c(Cl) [m.-%]
Fig. 14.48 Intensity of the Cl-line at D 837:59 nm as a function of the chlorine concentration
for various integration times tint , pHe D 60 mbar
18000
parameter t int
16000
1.25 μs
14000 2.5 μs
I Cl 134nm [a.u.]
5 μs
12000 10 μs
20 μs
10000 40 μs
8000
6000
4000
0 1 2 3 4
c (Cl) [m.-%]
Fig. 14.49 Intensity of the Cl-line at D 134:72 nm as a function of the chlorine concentration
for various integration times tint , pHe D 60 mbar
the detection limit [11.3] (cf. Chap. 11). The 3s-criterion to estimate the limit of
detection for chlorine in cement is not used here because of the natural content of
chlorine in cement generally. Figure 14.50 shows the detection limit for the Cl lines
D 134:72 nm and D 837:59 nm as a function of the integration time gate tint
using the ratios ICl 837 nm =IBg 861 nm and ICl 134 nm =IBg 861 nm . Starting with a detection
limit of 0.9 m.-% for tint D 1:25 s, a minimized detection limit of 0:1 m.-% is
achieved for tint D 5 and 10 s for the chlorine line in the UV. A further increase of
tint results in an increased detection limit of up to 0.28 m.-% for tint D 40 s. The
limit of detection for the chlorine line in the IR at an integration time tint D 1:25 s
14.4 Cement 335
1.0
134.72 nm
0.8 837.59 nm
limit of detection [m.-%]
0.6
a
0.4 b
0.2
0.0
0 5 10 15 20 25 30 35 40 45
t int [μs]
Fig. 14.50 The estimated limit of detection for single pulses and the Cl lines D 134:72 nm ,
D 837:59 nm as a function of the integration time gate tint , tdelay D 100 ns. The horizontal lines
indicate the achieved LODs for chlorine using the 837:59 nm line reported so far, a D [14.68],
b D [14.72]
is not included into the diagram because the value is above 3 m.-%. Similar to the
chlorine line in the UV, the limit of detection can be decreased by increasing the
integration time tint . For tint D 5 s, a minimum is achieved with 0.2 m.-%. A further
increase in tint results in a limit of detection of 0.5 m.-%. At a maximum integration
time of tint D 40 s, the limit of detection for the chlorine line in the UV and the
IR are nearly the same. Therefore, an integration time gate between tint D 5 s
and tint D 10 s should be chosen for the further experiments. The 95% confidence
interval of the limit of detection data shown in Fig. 14.50 (and also in Fig. 14.51)
is given by multiplying the respective values by a factor of 0.57 and a factor of 3.5
(see Table 2 in [11.3] for n 2 D 3 degrees of freedom).
To further improve the LOD, the influence of collinear double pulses was studied.
The pulse ratio is set to E1 W E2 D 1 W 1 and interpulse separations between
t D 1 s and 8 s are used in the measurements. Peter et al. reported that the use
of collinear double pulses increases the spectral line intensity at ambient conditions
[14.78] (cf. Sects. 6.1, 7.2, and 7.4). Delay time tdelay and integration time gate tint
are set to 100 ns and 10 s, respectively. Because of the laser system, the energy is
reduced to Eb D 450 mJ in the double pulse configuration. Process gas and pressure
inside the chamber are the same like in the case of single pulses. Figure 14.51 shows
the limits of detection achieved for the ratio ICl 134 nm =IBg 861 nm using double pulses.
All values are greater than those achieved with single pulses. The LOD ranges
between 0.23 and 0.31 m.-%. The limit of detection for t D 8 s is not shown
in the diagram because the value is greater than 1 m.-%.
Figure 14.52 shows the detected spectral line intensities of the Cl 134.27 nm
line as a function of the chlorine concentration for different interpulse separations
0.32
0.30
limit of detection [m.-%]
0.28
0.26
0.24
0.22
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Δt [μs]
Fig. 14.51 The estimated limit of detection for double pulses and the Cl line D 134:72 nm as a
function of the interpulse separation t
12000 parameter Δt
1 μs
2 μs
4 μs
I Cl 134 nm [a.u.]
10000
8 μs
8000
6000
0.0 0.2 0.4 0.6 0.8 1.0

c (Cl) [m.-%]
Fig. 14.52 The detected spectral line intensity of the chlorine spectral line a D 134 nm as a
function of the chlorine concentration for different interpulse separations t , pHe D 60 mbar,
tdelay D 100 ns, tint D 10 s
t. Regarding the detected intensity for a chlorine concentration of 0.05 m.-% the
line intensity drops from 6,500 counts for t D 1 s down to 5,500 counts for
t D 8 s. Additionally the dynamic range of the line intensity for concentrations
between 0.05 and 4 m.-% is reduced for greater interpulse separations t. For
t D 1 s, a range of ICl 134 nm D 4;700 counts exists whereas for an interpulse
separation of t D 8 s this value is reduced to ICl134 nm D 550 counts. For
t 2 s, an increase of the spectral line intensity can be seen compared with
14.5 Droplets 337
single pulses. On the other hand, the dynamic range is reduced. At a He pressure
of 60 mbar, the preconditioning effect of the first pulse of a double pulse burst does
obviously not result in an increased spectral line intensity for interpulse separations
t 4 s.
This effect could be attributed to a further reduced electron density. In compari-
son with single pulses, the electron density reaches values that exist for single pulses
at pressures below pHe < 60 mbar where a significant decrease in the spectral line
intensities is seen for both lines.
The described investigations show that the estimated limit of detection for
chlorine deploying the calibration curve criterion of DIN 32 645 can be reduced to
LOD D 0.1 m.-% for the UV chlorine line, single pulse excitation, and an integration
time gate of tint D 5 s corresponding to an improvement of a factor of >5
compared with the best values achieved so far.
14.5 Droplets
Under special conditions, LIBS allows to analyze the elemental compounds in

liquids with high sensitivity [14.79]. This suggests the use of LIBS for a field of
growing interest in the analytical sciences and speciation analysis [14.80, 14.81].
The chemical, biological, and toxicological properties of many elements are
critically dependent on their chemical form, its oxidation number, and the compound
the element is occurring in. This has a great impact on environmental and biological
chemistry, occupational health, nutrition, and medicine. Typical targets of speciation
analysis are metal ions in different oxidation states, metal complexes with inorganic
and organic ligands, metalloid oxoanions, organometallic compounds, and ionic
nonmetal species. An analytical technique yielding only the total amount of the
analyzed element does not consider different possible forms. Coupled techniques
combining chromatographic separation with element-specific detection have there-
fore become fundamental tools for speciation analysis. These so-called “hyphenated
techniques” enable the elemental analysis of separated fractions of complex sam-
ples. Typical detection techniques are mass spectrometry or atomic absorption
and atomic emission spectrometry and the most common separation techniques
are gas chromatography (GC) for volatile compounds and high-pressure liquid
chromatography (HPLC) for less volatile compounds. In recent years, capillary
electrophoresis has also become more and more important in this field [14.82].
Table 14.13 gives an overview of the most important hyphenated techniques and
their status of applicability.
The detection procedure should give maximum information with minimum
sample consumption. The LIBS process as a detection technique offers several
significant advantages: LIBS signals yield the elemental composition of complex
samples, the plasma breaks up any molecular structure, and the individual elements
lead to specific radiation. The information content is therefore higher than with the
UV–VIS or fluorescence detectors that are frequently used in HPLC chromatogra-
phy. LIBS is a true multielement technique. Depending on the spectrometer used,
Table 14.13 Different hyphenated techniques

GC HPLC SFC HPTLC CE
MS }}} }}} } } }}
IR }}} }} } }} }
OE }} } }
AAS } }}
NMR }} }
Fluorescence }} }}} }}} }}}
UV-VIS } }}} }}} }}}
Separation: GC (gas chromatography), HPLC (high-pressure liquid chromatography), SFC
(supercritical fluid chromatography), HPTLC (high-performance thin layer chromatography), CE
(capillary electrophoresis). Detection: MS (mass spectrometry), IR (infrared spectrometry), OE
(optical emission), AAS (atomic absorption spectrometry), NMR (nuclear magnetic resonance),
fluorescence spectrometry, UV–VIS spectrometry
˙˙˙ Standard technique
˙˙ Commercially available, less spread
˙ Experimental status, not commercially available
several elements can be monitored at the same time without loss of sensitivity. ICP-
MS detectors (inductively coupled plasma mass spectrometry) are used to achieve
highest sensitivity, making sample concentrations as low as pg/g (ppt) possible
[14.83, 14.84]. Plasma generation is achieved with the aid of an inductively heated
argon torch, and the sample is diluted in a large amount of gas needed for the
plasma generation (typically several liters per minute). ICP-MS detectors achieve
their highest sensitivity working with flow rates in the range of ml/min, which fit
well to HPLC. Significantly lower flow rates lead to a loss in sensitivity.
With LIBS the sample is transferred directly into the plasma state, without the
need for an additional gas flow. No dilution of the sample occurs and the method is
capable to analyze sample amounts as low as a few nanoliters. This will turn out to
be valuable for the coupling with low flow separation techniques such as capillary
electrophoresis or micro-HPLC. In these separation techniques, only a few l=min
are needed.
Water is one of the common solvents in HPLC chromatography and therefore
the process of plasma generation using aqueous samples must be well understood.
The mechanisms of plasma formation in water have been thoroughly examined
and different theoretical models to describe the breakdown have been published
(multiphoton model, cascade model) [14.85].
Different setups to record laser-induced breakdown spectra in a water matrix have
been described in the literature: bulk analysis in cuvettes [14.86] or on the liquid
surface [14.87], the analysis of liquid jets [14.88], aerosols [14.89] or the analysis
of single, isolated droplets [14.90]. Plasma formation in bulk water generally suffers
from short lifetimes such that the emission intensity falls steeply after approximately
1 s, making sensitive measurements impossible [14.91]. Cremers et al. [14.92]
compared single and double pulse LIBS spectra from metal samples placed under
water and showed that only the double pulse arrangement provides sufficient exci-
tation to identify the sample compounds. With a single pulse, the plasma formed is
14.5 Droplets 339
surrounded by an aqueous medium that leads to a fast decay of the plasma radiation.
In the double pulse experiment, the first pulse produces a gaseous atmosphere in
which the second pulse, reaching the target some 10 s later, produces a plasma
with a significantly longer lifetime. The excitation of the plasma in air or noble
gas atmospheres yields larger signal intensities than plasma generation inside bulk
water; this can also be seen in single pulse experiments [14.93]. Thin water jets,
aerosols or single, isolated droplets are superior sample forms compared with bulk
water, because the plasma can expand in a gaseous atmosphere.
Using a water film and a dual-beam dual-pulse setup, Kuwako et al. could
demonstrate a limit of detection as low as 0.1 ppb for sodium in water [14.94]. An
alternative approach to obtain a very high sensitivity was made by Cheung et al.,
who used an ArF-Laser at 193 nm for excitation instead of the commonly used
YAG Laser (1,064 nm). Limits of detection in the low ppb region were achieved
for sodium, calcium, and barium as a result of the efficient photoionization of hot
water molecules by the UV photons [14.95, 14.96].
Figure 14.53 shows a diagram of the hyphenated technique HPLC–LIBS [14.97].
The HPLC output is fed into the droplet generator and transformed into a continuous
stream of uniform droplets. In order to achieve the best possible transfer of
sample material into the plasma, a single droplet generator based on a piezoelectric
nozzle was applied. Unlike electro-spray or aerosol generators, a piezoelectric
Fig. 14.53 Setup for HPLC–LIBS. The eluate of the chromatographic separation (CS) is fed into
a droplet generator (DG) that produces a continuous stream of droplets (D). Individual droplets
are detected by a unit consisting of a laser diode (LD), an achromatic lens, and a photodiode
for droplet detection (DD). The detection of a droplet triggers a Q-switched Nd-YAG laser (LA),
the laser pulse transforms a single droplet into the plasma state, the radiation is guided into a
Paschen–Runge spectrometer (PR) and analyzed by a multichannel integrator electronics (MCI).
Measurements are stored and processed in a computer (PC). A gas guiding system (GG) leads
argon into the measurement chamber (MeC)
nozzle is a “drop on demand” device that produces single, isolated droplets with
volumes typically between 100 pl and 1 nl and an initial velocity of 1–2 m/s.
Figure 14.54 shows a stroboscopic picture of the generated sub-nanoliter droplets.
By synchronizing laser pulses to single droplets plasmas are generated in a gaseous
environment. The process of droplet formation is highly reproducible, leading to a
uniform droplet size. Phase Doppler anemometry measurements performed with a
piezoelectric droplet generator similar to the one used in this work revealed a relative
standard deviation of the droplet volume of 3% for droplets of 86 m diameter
[14.98]. After detection of an individual droplet, the laser is triggered and emits a
single Q-switched pulse that is focused onto the droplet and generates a plasma. The
plasma radiation is collected and guided to the spectrometer.
Figure 14.55 shows the optical setup for the analysis of single picoliter droplets.
For a better control of the experimental parameters, the HPLC was replaced by a
stock of sample solution, having a constant analyte concentration. Prior to analysis
all samples were filtered through 0:2 m pore size membrane filters. The plasma
radiation is resolved spectrally with a Paschen–Runge spectrometer equipped with
33 channel photomultipliers (type C 932, C 933, C 911 from Perkin Elmer), enabling
the simultaneous detection of 31 different elements and the zeroth order signal.
Among these elements are nonmetals such as Ar, H, C, O, N, P, S, alkali and
alkaline earth metals (Na, Ca, Mg) and various metals that play a significant role
in the fields of environmental chemistry, clinical biochemistry, industrial chemistry,
nutrition and toxicology, e.g., As, Hg, Cr, Pb, and Pt.
Most of these elements can form different metalloid species. The choice of
elements for the Paschen–Runge spectrometer was made to guarantee a broad
spectrum of possible chromatographic applications for the new LIBS detector. The
nonmetal elements could be used for the analysis of organic substances. Because all
elements have multiple transitions a careful choice of the lines that are best suited
for the analysis is important. Most of the stated elements were analyzed by recording
Fig. 14.54 Image of 270 pl droplets. The picture was taken with stroboscopic illumination, and the
droplet formation frequency is approximately 800 Hz. A part of the glass capillary of the droplet
generator can be seen in the upper part of the picture
14.5 Droplets 341
Fig. 14.55 Optical setup for LIBS experiments on droplets. M D mirror, LA D laser, A D aper-
ture, =2 PL D =2 plate, =4 PL D =4 plate, GP D Glan-laser polarizer, BE D beam expander,
D D droplet, MC D mirror chamber, MeC D measuring chamber, CCD D CCD-camera for
droplet observation, DD D droplet detector (photodiode), SI D stroboscopic illumination, LD D
laser diode (part of the droplet detector), PM D photomultiplier
laser-induced breakdown spectra with an echelle spectrometer. The matrix the ele-
ment is surrounded by might affect the line intensities; therefore, all echelle spectra
were recorded from small water droplets containing diluted samples. Usually the
line showing the strongest intensity was chosen. A complete list of all elemental
transitions detectable with the Paschen–Runge spectrometer is given in Table 14.14.
The plasma emission shows a strong continuum peak directly after the laser pulse
irradiation that decays faster than the element specific line emission (cf. Chap. 2,
Fig. 2.4). After a few microseconds, the element-specific line emissions dominate.
For a better signal-to-noise-ratio, the continuum part of the plasma emission is
excluded from the analytical signal with the aid of gated signal integration. Typical
delay times between the laser pulse and the start of the signal integration are
0:5–1:5 s, the integration gate width is between 20 and 50 s. The photomultiplier
signals are processed by a multichannel integrator electronics (MCI, cf. Sect. 4.5).
The photocurrents from 33 photomultipliers are integrated simultaneously with
individual time gates. For single pulse evaluation, the signal intensities of all
channels are transmitted to a computer.
In order to select the elemental transitions with the best analytical performance,
laser-induced breakdown spectra of droplets were studied between 200 and 780 nm
with an echelle spectrometer (cf. Sect. 4.2, type ESA 3000 from Laserlabor Adlers-
hof, Berlin, Germany) for each element. The experimental setup for the selection of
the transitions was similar to the one described above except that the spectrometer
was connected to the experiment with a fiber optics.
Table 14.14 List of element lines installed in the Paschen–Runge spectrometer. Ei D energy of
lower level, Ej D energy of upper level
Element Wavelength (nm) Ei (eV) Ej (eV) Grating order
P 178.28 0.0000 6.9548 1
S 180.73 0.0000 6.8606 1
Se 196.09 0.0000 6.3232 1
Zn 213.86 0.0000 5.7961 1
Sb 217.59 0.0000 5.6969 1
Cd 228.80 0.0000 5.4175 1
As 234.98 1.3134 6.5884 1
H 121.57 0.0000 10.1995 2
B 249.77 0.0019 4.9646 1
Si 251.61 0.0277 4.9541 1
Hg 253.65 0.0000 4.8868 1
O 130.22 0.0000 9.5220 2
Pt 265.94 0.0000 4.6600 1
Mg 279.55 0.0000 4.4341 1
N 149.26 2.3837 10.6900 2
V 310.23 0.3680 4.3637 1
Sn 317.50 0.4250 4.3291 1
Cu 324.75 0.0000 3.8169 1
C 165.70 0.0054 7.4883 2
Ti 334.94 0.0488 3.7496 1
Ni 341.48 0.0254 3.6554 1
Co 345.35 0.4318 4.0212 1
Cr 357.87 0.0000 3.4638 1
Fe 371.99 0.0000 3.3322 1
Al 394.40 0.0000 3.1429 1
Ca 396.85 0.0000 3.1236 1
Ce 399.92 0.2954 3.3949 1
Pb 405.78 1.3206 4.3754 1
Ar 420.07 11.5491 14.5000 1
La 433.37 0.1729 3.0332 1
H 486.13 10.1995 12.7494 1
Na 589.00 0.0000 2.1024 1
0th order 0
The choice of the respective transitions was made on the basis of measurements with an echelle
spectrometer
In order to achieve higher signal intensities, the plasma emission is collected

with a spherical mirror and focused on the entrance slit of the Paschen–Runge
spectrometer. The mirror is placed inside the measurement chamber and can be
adjusted without opening the chamber. The droplet generator is also placed in the
measurement chamber and produces a free-falling stream of droplets. It is made of
a glass capillary with a small orifice (50 m diameter) on one side that is enclosed
by a tube-shaped piezoelectric element. The piezoelectric element is activated with
a voltage pulse that induces a pressure wave in the liquid leading to the formation
14.5 Droplets 343
of a single isolated droplet with a volume between 100 pl and 1 nl and an initial
velocity of 1–2 m/s. A CCD camera with a stroboscopic illumination is used to
control and adjust the droplet generation (cf. Fig. 14.54). The piezoelectric nozzle is
capable of delivering up to 2,000 droplets per second; however, droplet frequencies
as low as 10–20 Hz are also possible. The droplet size can only be varied in a narrow
range by adjusting the voltage pulse. A higher peak voltage leads to a higher droplet
speed, and larger pulse length leads to larger droplets. Every individual piezoelectric
nozzle has a set of parameters it is giving the best performance with (best stability
and droplet uniformity) and the nozzles show little tolerance for deviations from
these parameters. Therefore, systematic studies with droplets of different sizes were
not accomplished.
A droplet detection unit consisting of a diode laser, an achromatic lens, and
a photodiode is used to detect single droplets with a temporal precision of
approximately 1 s. The photodiode signal shows a characteristic “W” shape, when
the droplet trajectory passes through the focus of the diode laser beam. This can be
exploited for precise adjustment of the droplet in space and time. The photodiode
signal is shown in Fig. 14.56. The shape of the signal can be explained with the
droplet being a small sphere that does not refract light when the beam focus lies
in the center of the sphere, so the loss of photodiode signal due to the droplet is
minimized in this moment since all rays are propagating perpendicularly through
the surface of the sphere.
After detection of a droplet, a synchronized laser pulse is emitted that transforms
the droplet into a plasma. Temporal and spatial overlap between laser pulse and
droplet are essential for a stable analytical signal. Temporal misalignment of
a few microseconds or spatial misalignment of a few micrometers leads to a
1 + 40 μm
accurate
adjustment
photodiode signal [V]
0 - 40 μm
-1
-2
-3
-40 -20 0 20 40 60 80
time [μs]
Fig. 14.56 Photodiode signal for different positions of the droplet. When the droplet passes
through the focus of the diode laser beam, the characteristic “W” shape can be seen. A lateral
displacement of ˙40 m leads to the two signals shown with dashed lines
significant decrease in signal intensity. A Q-switched Nd-YAG laser (10 ns pulse

width) was used for the experiments. The repetition rate of the laser restricts the
measurement frequency to 10 Hz, although the other components of the experiment
(droplet generator, spectrometer, signal integration, and recording) could permit
measurements at a rate of 1,000 Hz. The laser pulses were attenuated with the aid of
a /2-plate and a Glan laser polarizer to 50–140 mJ. The laser beam is expanded to a
diameter of approximately 3 cm and then focused with an achromatic lens having a
focal length of 10 cm. For adjustment, the lens can be translated perpendicularly
to the laser beam and in the direction of the laser beam. For the timing of the
signal detection and the laser trigger relative to the droplet detection, a digital delay
generator was used.
For a better understanding of the interaction between laser pulse and droplet, the
plasma dynamics was studied with an electro-optical high-speed camera enabling
exposure times down to 200 ps (Stanford Computer Optics, 4 Picos). A 500
ppm sodium solution was used to produce intense element-specific radiation. An
interference filter (central wavelength 589 nm, FWHM 10 nm) placed in front of the
camera transmits the radiation coming from the sodium transitions at 589.00 and
589.59 nm. A series of images recorded with a shutter time of 80 ns with varying
Fig. 14.57 Development of the plasma recorded with an electro-optic camera. Sample: 500 ppm
sodium solution. Shutter time: 80 ns. Interference filter: center at 589 nm, FWHM 10 nm. Laser
pulse: 50 mJ, 10 ns, focus 5 mm behind droplet. The upper left picture shows the relative positions
of laser focus and droplet, and the initial droplet position is marked with a reticle in all pictures.
Delay times relative to the start of the laser pulse are given for each picture. LB D laser beam,
D D droplet
14.5 Droplets 345
delay times is shown in Fig. 14.57; the hair cross indicates the initial position of the
droplet. The laser beam (50 mJ pulses, 10 ns pulse width) enters from the left side
and the focus position is a few millimeters behind the droplet. Every picture has a
size of 5 mm 3:75 mm.
The first breakdown can be detected at the rear side of the droplet because
the beam is focused by the droplet itself leading to the highest laser irradiance at
the rear side of the droplet. During the first 500 nanoseconds of expansion (not
shown here), the breakdown moves in the direction of the laser beam, because
evaporated material is further heated by the laser pulse. The intensity distribution
of the plasma radiation is asymmetric with the side facing the laser emitting most
of the radiation. The plasma grows within 500 ns to a diameter of approximately
2.5 mm. No fragmentation of the droplets was observed on the images. The
diameter of the luminous plasma was measured as a function of time and the
speed of propagation was calculated which decreases from 40 km/s during the first
nanoseconds of the expansion to less than 1 km/s after 500 ns, see Fig. 14.58. Two
regions with a different dynamic behavior can be distinguished. Between 15 and
30 ns the expansion can be fitted by a linear regression described by the function
d t 1:76 .d D plasma diameter), between 100 and 500 ns the linear fit yields
d t 0:34 . It must be pointed out that this is not the analysis of the expansion of
the shock wave which can be described by the Sedov model [cf. Sect. 8.1, relation
(8.1)] but the expansion of the luminous plasma front. Normally the plasma front
is retarded relative to the shock wave. An example for the analysis of shock waves
produced during a LIBS experiment is described in Sect. 8.1.
In consideration of the heat capacity of liquid water (4.18 J/gK) and the enthalpy
of vaporization (2,258 J/g at 100ıC) it takes approximately 0.7 mJ to vaporize 270 pl
of liquid water. With the laser pulse having an energy of 50 mJ, sufficient energy is
2
plasma diameter d [mm]
d ~t 0.34
d ~t 1.76
0.3
10 100 1000
time t [ns]
Fig. 14.58 Plot of the plasma expansion for the first 500 ns. The speed of the expansion drops
from 40 km/s during the first nanoseconds to less than 1 km/s after 500 ns. Two different expansion
regimes can be distinguished that follow different power laws
available for complete vaporization. From the fast expansion and the homogeneity
of the plasma no hints are observed for any fragmentation into smaller subdroplets
or parts of the droplet which are not transferred to the plasma state.
In a series of experiments the influence of the laser pulse energy and wavelength
on the analytical sensitivity was studied. The fourth harmonic of a Nd-YAG Laser
(266 nm) was applied to perform LIBS experiments with small droplets and the
results were compared to the excitation with 1,064 nm. For both wavelengths
calibration curves with different sodium concentrations between 0 and 10 g=g
were measured and the limits of detection were determined (see Sect. 11.2). With
60 mJ pulses at 1,064 nm, a limit of detection of 0.75 g=g was calculated using the
3s-criterion [11.3]. The same measurements were made with 25 mJ pulses at 266 nm
and yielded a limit of detection of 2 g=g. When using the same pulse energy (25
mJ) for both wavelengths, the signal intensities of the sodium 589.00 and 589.59 nm
lines are approximately twice as high with the excitation at 266 nm compared with
the 1,064 nm excitation. Since the laser could not deliver more than 25 mJ of the
UV wavelength per pulse, the fundamental wavelength was used in all subsequent
experiments.
The signal intensities show a strong correlation to the laser pulse energy.
Figure 14.59 shows the rise of the line intensities of sodium lines for 500 ppm
sodium droplets with the laser pulse energy. For more than 60 mJ/pulse, a
saturation of the line emission is observed. Since very high pulse energies can
lead to instabilities in the operation of the piezoelectric droplet generator, most
measurements were performed in the range between 60 and 100 mJ/pulse.
In order to determine the analytical sensitivity, echelle spectra from different
samples were recorded containing between 0 and 0:7 g=g calcium. Samples were
diluted from a 100 g=g stock solution containing Ca stabilized with 5% nitric
I Na 589.00 nm, I Na 589.59 nm [a.u.]
3 × 104
2 × 104
1 × 104 589.00 nm
589.59 nm
0
0 50 100 150 200
laser pulse energy [mJ]
Fig. 14.59 Line intensities INa 589:00 nm and INa 589:59 nm versus laser pulse energy. Sample: droplets
containing 500 g=g sodium. Fifty laser pulses were accumulated. A maximum is observed at
approximately 60 mJ per pulse
14.5 Droplets 347
acid. The average integral intensity of 20 measurements, with each measurement

consisting of 100 droplets, is plotted as a function of the concentration in Fig. 14.60.
Integral intensities were calculated by fitting a Gaussian function to the calcium
peak at 393.37 nm and subtracting the background intensity (baseline subtraction).
The concentration with the same signal intensity as three times the standard
deviation of the blank sample, the limit of detection is calculated to be 20 ng/g
[11.3] (cf. Sect. 11.2).
This detection limit is higher than those obtained with ICP-MS systems. Many
elements can be detected in the range of low ppt values with modern ICP-MS
systems [14.83, 14.84]. Nevertheless, the absolute sensitivity of the experimental
setup described here is very high; for elements showing strong emission intensities
it is comparable with ICP-MS systems, since 20 ng/g calcium contained in 100
droplets of 280 pl volume correspond to an absolute amount of only 560 fg calcium.
The analytical sensitivity is strongly dependent on the excitation and emission
characteristics of the corresponding element. The excitation energy and the Einstein
coefficient determine together with the plasma properties the emission intensity
of the respective line of any element. Calcium belongs to the group of elements
showing strong line emission yielding a low limit of detection. Similar elements are
aluminum, boron, cadmium, chromium, copper, magnesium, sodium, and zinc. Here
limits of detection below 1 g=g can be expected. Weak emission intensities were
detected for the elements arsenic, cobalt, mercury, lead, platinum, selenium, and tin.
Here the corresponding limits of detection are higher, with the setup studied values
between 20 and 100 g=g are expected. Between these two groups, the elements
iron, nickel, silicon, titanium, and vanadium can be found with limits of detection
1.0
ICa 393.37 nm - I BG [a.u.]
0.8
0.6
0.4
Ca II 393.37 nm
0.2 LOD3s = 20 ng / g
r = 0.988
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Ca concentration [μg / g]
Fig. 14.60 Calibration curve for calcium. Experimental parameters: delay time tdelay D 1:5 s,
integration gate width tint: D 50 s, laser pulse energy 140 mJ, pulse width 10 ns, 100 droplets
contribute to one measurement; 20 measurements were accumulated for each concentration. Limit
of detection (LOD) D 20 ng/g, correlation coefficient r D 0:988: The error bar shown was
multiplied by a factor of 5
in the low g=g region. The elements were classified by comparing signal-to-noise
values from echelle spectra recorded under identical conditions.
In the normal mode of operation the Paschen–Runge spectrometer can only
detect the intensity of radiation in very narrow spectral windows defined by the
resolution of the instrument and the positions of the photomultipliers respectively
the exit slits. Scanning over a narrow spectral region (less than one nanometer) is
possible by tilting a refractory plate positioned directly behind the entrance slit. The
tilting moves the incoming beam and tunes all spectrometer channels to a shifted
wavelength. The Paschen–Runge spectrometer is equipped with a stepper motor that
allows the precise tilting of this plate. Plotting the photomultiplier intensity against
the stepper motor position gives the spectral shape of the respective transition.
Figure 14.61 shows the line shape for different concentrations of copper samples
between 10 and 1,000 g=g.
In order to simulate a chromatographic separation, continuous flow measure-
ments were performed by connecting the droplet generator to a HPLC pump that
forces a constant liquid flow (typically 5–15 l=min) through the glass capillary.
For the stability of the droplet production, it is important to adapt the volume flow
of the droplet generator to the volume flow delivered by the pump. Otherwise the
droplet generator may either run dry or droplet formation may fail because too much
liquid is delivered to the glass capillary. In order to match the volume flows of the
HPLC pump and the droplet generator, an active closed loop controlled the droplet
frequency. A liquid flow of 13 l=min is equivalent to a droplet frequency of 800 Hz
(droplet volume 270 pl). Because the laser was operated at 10 Hz, a pulse divider
was used to reduce the measurement frequency to 10 Hz. A six-way switching valve
(Rheodyne) integrated into the flow system in front of the droplet generator allowed
to inject small amounts of sample (20 l) without interrupting the droplet stream.
1000
intensity I Cu 324.75 nm [a.u.]
800 1000 μg / g
600
10 μg / g
200 μg / g
400 0 μg / g
50 μg / g
200
150 200 250 300 350

entrance slit position [a.u.]
Fig. 14.61 Shape of the Cu I 324.75 nm line, recorded with the Paschen–Runge spectrometer by
tilting the refractor plate behind the entrance slit. The concentrations of the different samples are
given at the curves
14.6 Gases 349
14000
12000
I Mg 279.55 nm [a.u.]
10000
8000
6000
4000
2000
0
0 100 200 300 400 500 600 700 800 900
time [s]
Fig. 14.62 Continuous flow measurement of 20 l of 18 g/g magnesium solution. The sample
was injected in the continuous solvent flow coming from the HPLC-pump with the aid of a
Rheodyne six-way valve. Temporal broadening of the peak can be seen as a result of dead volumes
in the system. The droplet frequency was 850 Hz, and the measurement frequency was 10 Hz. The
analyzed amount of sample is 4.2 ng. The data are smoothed with a ten point moving average
function
Monitoring the signal intensity of the sample element as a function of time delivered
a signal peak comparable with signals detected in chromatographic separations.
Because of dead volumes in the droplet generator and in the adapter tubes used
for connecting the capillary and the droplet generator a temporal peak broadening is
detected. Figure 14.62 shows a continuous flow measurement of 20 l of a sample
containing 18 g/g magnesium. The total amount of analyzed magnesium that gives
rise to the signal shown is 4.2 ng because only one out of 85 droplets was analyzed
due to the mismatch in droplet and laser frequency.
The results show the potential of the piezoelectric droplet principle for the
development of new detectors for ultra low volume flow separation techniques such
as micro-HPLC or capillary electrophoresis. Volume flows in the sub l/min region
will be accessible without dilution of the sample.
14.6 Gases
LIBS measurements of gas mixtures such as chlorine in air [14.99], fluorine in

air [14.100], and NH3 vapor in N2 [14.101] have been described. Laser-induced
breakdown in molecular gases initiates the fragmentation of molecules into elemen-
tal components, which are identified by detection of their atomic emission lines
[14.102]. The detection of elements in gas mixtures is of interest for monitoring
of gaseous components in process control, atmospheric pollution, and combustion
systems [14.103]. LIBS detection of aerosols as solid particulates in gaseous
components will be presented in Sects. 14.7 and 14.8. In combustion, hydrocarbon
mixtures and the fuel–air ratio or the equivalence ratio are of particular interest
[14.104, 14.105]. For example, the atomic abundance ratios of O/C, N/C, and H/O
are measures of the fuel/air ratio [14.106,14.107]. Hydrocarbon mixtures have been
investigated in LIBS experiments in which TEA CO2 lasers emitting at a wavelength
of 10:6 m were used, but calibration curves were not presented [14.102, 14.106].
Fuel/air ratios of CH4 –air mixtures of a jet diffusion flame were measured by
LIBS [14.107]. The study described in the following was carried out to determine
calibration curves of LIBS detection through a single, narrow aperture (diameter 7
mm), which would facilitate access to smaller combustion engines [14.108]. Five
bursts of two closely spaced laser pulses at a wavelength of 1,064 nm were applied
for the measurement.
The emissions of C, N, and O were measured simultaneously with an echelle
spectrometer (cf. Sect. 4.2). In the following propane (C3 H8 / gas mixtures that are
not combustible were used to find the analytical performance of LIBS without
interference from ignition of combustion. However, the calibration curves cover
atomic abundance ratios, which are given for the combustion of heptane (C7 H16 /
and air near an equivalence ratio of ˆ D 1. Assuming an ideal homogeneous
distribution of fine droplets, heptane was taken as an example with which to estimate
the atomic abundance ratios for ˆ D 1 because heptane is a primary reference
fuel and is often used as an engine model fuel [14.109]. Measurement of the
equivalence ratio through a single narrow optical window is of interest, e.g., in
automotive engineering for in-cylinder diagnostics before ignition at engine test
stations [14.110] as well as in other combustion systems [14.105, 14.107].
Double pulse bursts with an interpulse separation of t D 250 ns are generated
at a 10 Hz repetition rate by polarization coupling of orthogonally polarized output
beams (beam diameter Ø6 mm) of two Nd:YAG lasers (Continuum Surelite I-10 and
I-20). The energy of the double pulse bursts amounts to 230 mJ, with a first pulse of
130 mJ (pulse duration 8 ns) and a second pulse of 100 mJ (8.8 ns). The interpulse
separation t of the two laser pulses is set by a trigger delay between the two lasers.
The lasers are operated with reduced flashlamp power to permit the pulse energy to
be adjusted.
The Q-switching Pockels cell of laser I-20 with a 20 Hz flashlamp rate is
triggered at each second flashlamp pulse for a 10 Hz laser output. The laser beam
is focused into the gas mixture in a sample chamber by a setup that comprises
two 140-mm distant fused-silica lenses: lens 1 (f1 D 300 mm; diameter 50 mm;
plano–convex) and lens 2 (f2 D 50 mm; diameter 25 mm; biconvex). Lens 2 is
positioned in front of the fused-silica entrance window of the sample chamber.
The sample chamber is an aluminum vacuum cross (material, AlCuMgPbF38; inner
diameter 40 mm). The optical aperture at the entrance window is 7 mm. The spectra
are recorded with an echelle spectrometer (Model ESA 3000EV/i, LLA GmbH,
Berlin) covering a spectral range from 200 to 780 nm with some gaps without
detection. A single UV-enhanced fiber of 0.8 mm core diameter and 0.8 m length is
positioned beside the incoming laser beam at lens 1. With this configuration, only
a small fraction of the emitted plasma light is captured and guided to the echelle
spectrometer. This configuration is rather insensitive to misalignment. Nevertheless
14.6 Gases 351
the optical signal can be increased further by imaging of the plasma light onto the
fiber.
The CCD camera of the echelle spectrometer is equipped with a gated
microchannel-plate (MCP) image intensifier (cf. Sect. 4.4, Table 4.8, Fig. 4.14).
The exposure time is set to 500 ms, yielding an acquisition of five pulse bursts at a
repetition rate of 10 Hz. The MCP gate is set to 10 s per laser pulse at a delay of
0:5 s to the second laser pulse of the burst. The Balmer H˛ line ( D 656:2 nm) is
within a wavelength gap of 653–658 nm, which cannot be detected with the echelle
spectrometer used. Therefore, the H˛ line is measured with a monochromator (Jobin
Yvon Model HR320; 600 grove/mm grating; slit, 50 m; photodiode array detector)
triggered simultaneously with the echelle spectrometer. An additional fiber bundle
(cross section diameter 3 mm; length 3 m) is positioned beside the single fiber
of the echelle spectrometer at lens 1 to guide the plasma light to the HR 320
monochromator. The MCP gate of the HR 320 monochromator is 2:8 s at a delay
of 0:3 s.
Before filling, the sample chamber is evacuated to 3 Pa by a two-stage rotary vane
vacuum pump via a fine vacuum adsorption trap with zeolite filling. The gases are
carbon dioxide, nitrogen, and propane GA312. For gas mixtures with air, ambient
air is used without any purification.
The gas pressure is measured by a test pressure gauge. The gas inlet is set by a
metering valve. Dilution series were taken by starting with step i D 0 with a single
gas component A in the sample chamber at a pressure p1 D 100 kPa. Gas mixtures
were set for the measuring points in successive steps numbered i by removing a
portion of the gas or the gas mixture to a second pressure, p2 D 60 kPa, and refilling
the chamber with the second gas component B to the former gas pressure p1 . The
dilution series results in partial pressure pA for component A of pA =p1 D .p2 =p1 /i
and for component B of pB =p1 D 1 .p2 =p1 /i . LIBS measurements are always
taken at a total pressure of the gas mixtures of p1 D 100 kPa and a room temperature
of 293 K.
One spectrum of a laser-induced plasma in a carbon dioxide–air mixture near
the spectral lines of C, N, and O included is shown in Fig. 14.63. For evaluation,
the spectral peak area is determined by numerical integration at 247.54–248.22nm
for the C I line at 247.86 nm (denoted IC247 ; excitation energy of the upper level,
Eu D 7:68 eV), at 741.7–748.3 nm for the N I lines at 742.36, 744.23, and 746.83 nm
(denoted IN7428 ; Eu D 11:99 eV), and 776.3–778.5 nm for the O I triplet at 777.19,
777.42, and 777.54 nm (denoted IO777 ; Eu D 10:74 eV). The mean values and
standard deviations of the spectral peak areas (or the spectral peak area ratios) from
three spectra acquisitions are calculated and given in the calibration curves, whereas
each spectrum results from the accumulation of five laser bursts at the CCD camera
of the echelle spectrometer. Interference of other emission lines in a spectral region
of at least 5 nm adjacent to these lines was not observed. This absence of interference
permits the use of low-resolution spectrometers or spectral filters for detection. The
composition of air is assumed to be the same as the standard atmosphere, i.e.,
0.78084 N2 , 0.209476 O2 , 0.00934 Ar, and 0.000314 CO2 as fractional volumes
of the major gas species [14.111].
a b
4000 4000
O I 777.42
CO2 + air CO2 + air
O I 777.19
p1 = 100 kPa p1 = 100 kPa
3000 3000
intensity [counts]
intensity [counts]
C I 247.86
O I 777.54
N I 746.83
N I 744.23
2000 2000
N I 742.36
1000 1000
0 0
246 247 248 249 250 742 744 746 748 777 778 779
wavelength [nm] wavelength [nm]
Fig. 14.63 Spectral emission of (a) carbon and (b) nitrogen and oxygen for partial pressures of
0.2177 kPa CO2 and 99.782 kPa air, respectively. This mixture results from step i D 12 of the
dilution series that starts with pure CO2 to which air is added
2.5
CO2 + air
vertical error bars x5
spectral peak area I C247 [a.u.]
pure CO2
pure air
0.1
* = (pCO2 + 0.000314 pair ) / p1
0.02
1E-3 0.01 0.1 1
fractional atomic abundance of C*
Fig. 14.64 Spectral peak area of the carbon C I 247.9 nm emission line versus the fractional
atomic abundance of C. The horizontal error bars are estimated from the added relative errors
(times step number i ) of the measured pressures p1 and p2 ; with an uncertainty of the manometer
of 1% of the final value (200 kPa) which means 2% at 100 kPa for each pressure value, taken into
account
Figure 14.64 gives the spectral peak area versus the fractional atomic abundances
of C that result from a dilution series starting with pure CO2 .i D 0) and with air
added until step i D 12. The fractional atomic abundance of C is calculated as noted
in Fig. 14.64 by the partial pressures of the CO2 gas as well as the CO2 fraction of
the air filled in during the dilution series. It is assumed that nearly 100% of the CO2
molecules are fragmented at the laser pulse energies irradiated. Additionally, pure
air is measured after evacuation of the sample chamber.
A function I / x b , where x denotes the fractional atomic abundance of C, is
fitted to the data points of pure air and in the middle of the curve of Fig. 14.64. For
low C abundances of the dilution series, probably systematic errors caused by the
14.6 Gases 353
stepwise gas pumping and filling actions during a dilution series are the reasons
for the deviations. For high C abundances, an overflow of the CCD camera occurs
because the amplification of the MCP has been adjusted to produce sufficient
sensitivity for low C abundances. Nevertheless, it is shown that the LIBS signal
varies over more than three decades, approximately according to IC247 / x 0:62˙0:01 .
In addition, absorption and plasma conditions may change when the mixture
approaches a pure CO2 mixture. A saturation behavior similar to that of C is
observed for the O I 777 nm line, where no saturation of the camera occurs.
This saturation leads to an O/C calibration curve versus the atomic abundance
ratio of O and C – see Fig. 14.65 – that varies approximately according to IO777 =
IC247 / x 0:63˙0:03 (x is the atomic abundance ratio of O/C), with smaller deviations
at high CO2 fractions.
From this observation, the presumption is that the absolute atomic abundances or
the number densities might also affect the calibration curves in such a way that the
spectral peak area will saturate at higher number densities. Such an influence of the
absolute atomic abundances would also result in an influence of gas temperature on
the calibration curves, which should be considered when the gas temperatures vary.
Dilution series with propane gas mixtures were conducted by starting with pure
propane and adding CO2 until step number i D 8 was reached; see Figs. 14.66–
14.68. The spectral peak area ratio of O/C in Fig. 14.66 approaches a constant value
at high CO2 fractions that is consistent with the ratio of 0:94 ˙ 0:03 measured
for pure CO2 , cf. Fig. 14.65. The measured values are fitted by IO777 =IC247 D
axO=C =.d C xO=C /, where xO=C denotes the partial pressure ratio of O and C,
a D 0:97 ˙ 0:02, and d D 0:85 ˙ 0:07.
Figures 14.67 and 14.68 show the calibration curves for the hydrogen to carbon
(H/C) and the oxygen to hydrogen (O/H) ratios, respectively. For the H˛ line, the
peak value Ipk;H656 was taken instead of the peak area.
80
spectral peak area ratio I O777 / I C247
CO2 + air
pure air
10
pure CO2
* = 2(pCO2+ 0.20979 pair) / (pCO2+ 0.000314pair)

1
10 100 1000
atomic abundance ratio O / C*
Fig. 14.65 Ratio of the spectral peak area of the O I 777 nm and the C I 247.9 nm emission lines
versus the atomic abundance ratio O/C from the same dilution series as in Fig. 14.64
spectral peak area ratio I O777 / I C247

0.8 C3H8 + CO2
0.6
0.4
0.2
pure C3H8
0.0
0 5 10 15 20 25 30 35
partial pressure ratio p CO2 / p C 3 H 8
Fig. 14.66 Ratio of the spectral peak area of the O I 777 nm and the C I 247.9 nm emission lines
versus the partial pressure ratio of CO2 to C3 H8 for a dilution series that starts with pure propane
to which CO2 is added
C3H8 + CO2
15000
I pk, H656 / I C247 [a.u.]
10000
5000
* = 8p C3H8 / (p CO2 + 3p C3H8)
0
0.0 0.5 1.0 1.5 2.0 2.5
atomic abundance ratio H / C*
Fig. 14.67 Ratio of spectral peak value Ipk;H656 of the BALMER H˛ line at 656.2 nm and spectral
peak area IC247 of C I 247.9 nm versus the atomic abundance ratio H/C. Same dilution series as in
Fig. 14.66
The curves are fitted in Fig. 14.67 by Ipk; H656 =IC247 / x 1:47˙0:03 and in Fig. 14.68
by IO777 =Ipk; H656 / x 1:29˙0:03 , where x is the corresponding abundance ratio.
The nitrogen to carbon (N/C) signal when N2 is added to pure propane in
a dilution series is similar to the former measurement. The curve is fitted by
IN742-8 =IC247 / .xN=C xo /1:13˙0:03 , where xN=C is the atomic abundance ratio of
14.7 Particulates 355
C3H8 + CO2
0.002
I O777 / I pk, H656 [a.u.]
pure C3H8
0.001
* = 2pCO2 / 8pC3H8
0.000
0 2 4 6 8
atomic abundance ratio O / H*
Fig. 14.68 Ratio of the spectral peak area of the O I 777 nm line and the spectral peak value
Ipk;H656 of the Balmer H˛ line at 656.2 nm versus the atomic abundance ratio O/H. Same dilution
series as in Fig. 14.66
N/C and xo D 0:92 ˙ 0:4 is an offset value. The addition of N2 to hydrocarbon

mixtures results in the occurrence of nitrile (CN) emission bands in the LIBS
spectra. The CN signal varies with the atomic abundance ratio of N/C and attains a
peak for an N/C atomic abundance ratio of about 2.5.
The combustion of 1 kg of heptane (C7 H16 , 100.2 g/mol) with 15.2 kg of air
(28.96 g/mol) according to the chemical reaction C7 H16 C 11O2 D 7CO2 C 8H2 O
results in an equivalence ratio of ˆ D 1, which is the desired value in the ignition
region of a combustion engine. Therefore, the atomic abundance of C can be
estimated to be 7 NA 1 kg/(0.1002 kg/mol) from propane and 0:000314 NA
15:2 kg/(0.02896 kg/mol) from CO2 of air (NA D 6:02 1023 mol1 , the Avogadro
constant). Similar calculations for the other elements result in atomic abundance
ratios of 2.3 for H/C, 1.38 for O/H, 11.7 for N/C, and 3.1 for O/C, which correspond
to ˆ D 1. The measurements of Figs.14.67 and 14.68 and the calibration curve for
N/C demonstrate that LIBS through a single narrow aperture allows to monitor the
abundance ratios in regions covering these values.
14.7 Particulates
Alkalis, zinc and lead, are well known for their negative impact on the blast furnace
process. As accompanying elements of the charged burden and coke, they affect
its degradation behavior and can lead to higher coke consumption, accretions, and
scaffoldings. These elements are partly removed from the blast furnace by the slag
and the top gas, while another part forms inner circuits and remains in the blast
furnace. The control and removal of alkalis, zinc and lead, can be improved by
certain measures in blast furnace operation. Therefore, the build-up of inner circuits
has to be detected as early as possible. The inline measurement of the alkalis, zinc
and lead, in the top gas of the blast furnace will significantly help in closing the
metallurgical mass balances. For the inline multielement analysis, the method of
laser-induced breakdown spectroscopy (LIBS) is investigated [14.112].
Figure 14.69 shows the schematics of the optical setup. The output pulses of two
Nd:YAG lasers (wavelength 1,064 nm, repetition rate 10 Hz, pulse duration 8 ns)
are superposed to collinear double pulses with a short interpulse separation of
t D 64 ns (cf. Chap. 6 and [14.108]). The combined laser beam is guided through
the measuring probe (length 2.7 m, diameter 76 mm) and is focused by lens L1 into
the top gas stream with top gas velocities in the range of 10–20 m/s. The emitted
light of the plasma is coupled into a fiber optic cable and guided to a Paschen–
Runge spectrometer with photomultipliers (cf. Sect. 4.2). The channels installed
are (element and wavelength of spectral line in nm): Na 588.99, K 766.49, Pb
405.78, Zn 334.50, Zn 213.86, C 247.86, N 746.83, O 777.19, H 656.28, Ca 393.37,
Fe 271.44, and zeroth order. The signals of the photomultipliers are integrated by a
multichannel integrator electronics (MCI, cf. Sect. 4.5). The closely spaced double
pulses of the laser beam are used to increase the spectral signal for top gas pressures
of about 3–4 bar [14.113]. The sensitivity of the measuring system was proofed
with a reference gas cell for the gaseous components and with solid dust samples
for other channels.
The access to the top gas is achieved via an optical measuring probe (MP in
Fig. 14.69) inserted through a flange of the top gas tube before the dust separator of
the blast furnace, see Fig. 14.70. The top gas tube diameter is approximately 3 m.
Fig. 14.69 Schematics of the optical setup. M D mirror; PBS1 D beam combiner; BS D beam
splitter; PD D photodiode; FO D fiber optics; W1; W2 D optical windows; L1; L2; L3 D lenses;
TT D top gas tube; P D laser-induced plasma; MP D measuring probe; LH D lens holder;
RC D reference chamber; PR D Paschen–Runge spectrometer; MCI D multichannel integrating
electronics; PC D computer
Fig. 14.70 Schematics of the top of the blast furnace with the measurement location. The cabin
(3 2:4 2:9 m3 ) with the laser measuring equipment is installed at the top gas tube before
the first dust separator
The measurement cabin includes the measuring probe handling, the optical setup
with two lasers, the Paschen–Runge spectrometer, a computer and further measure-
ment equipment for gas flow, pressure, and temperature. The CO concentration is
monitored in the cabin in order to ensure that it is below the safety limit. The gas
tightness of the measuring equipment was verified before the first top gas access for
pressures up to 4 bar.
Table 14.15 summarizes key data of the blast furnaces involved in the onsite
trials. The measuring cabin with the LIBS instrumentation is fixed at a platform
at about 42–46 m height above ground.
The front side of the measuring probe and the measurement cabin installed at the
top gas tube are shown in Fig. 14.71.
The measuring equipment was tested in onsite measuring campaigns at each blast
furnace. The measuring system was able to monitor simultaneously and in situ the
relative concentrations of sodium, potassium, zinc, lead (Na, K, Zn, Pb), as well
as carbon, nitrogen, oxygen, hydrogen, calcium (C, N, O, H, Ca), and iron (Fe) in
the top gas stream. The average concentration of the circulating materials is in the
range of about 1–20 mg=m3 top gas. For every 10 s a measuring value was taken
being the mean value of 100 laser bursts. A quantitative measurement would require
calibration by reference values of the top gas composition or a reference top gas
stream at 3–4 bar with defined element concentrations which was not available at
the time of the study.
The precision of the laser measurement is estimated by the relative standard
deviation of the mean line intensity observed in a time interval not influenced by the
charging process. The element-specific relative standard deviation was determined
Table 14.15 Summary of key data of the blast furnaces chosen for the onsite trials
ThyssenKrupp Rautaruukki Oyj Aceralia Corporation
Steel AG (TKS) (RR) Siderurgia (ACER)
Blast furnace Schwelgern 2 Masuuni 2 Blast furnace A
Height of platform above 46 42 42
ground (m)
Top gas composition 48:24:24:4 49:23:23:5 44.9:22.4:22.2:4.1
N2 WCOWCO2 WH2 in %
Average top gas pressure 3.4 2.4 3.0
(absolute) (bar)
Average top gas temperature 120 180 178
(ı C)
Average dust concentration 8 5 10
(g=m3 )
Fig. 14.71 Left: measuring probe before insertion into the top gas tube at TKS. Right: measure-
ment cabin installed at the blast furnace of RR
4.0
Zn I 334.50 nm 3.0 Pb I 405.78 nm
3.5
I Zn I 334.50 nm [a.u.]
I Pb I 405.78 nm [a.u.]
2.5
3.0
2.5 2.0
2.0 1.5
1.5
1.0
1.0
10:20 10:40 11:00 11:20 11:40 10:20 10:40 11:00 11:20 11:40
time [hh:mm] time [hh:mm]
Fig. 14.72 Monitoring of Zn (left) and Pb (right) at the blast furnace of TKS
for ten consecutive measuring values, i.e., 1,000 laser bursts, and ranges between
2.3% and 7% for the circulating materials. Figure 14.72 gives an example of the
time behavior of the spectral signals for Zn and Pb during smooth operation of the
blast furnace. The peaks coincide with sinter charging having a period of 14 min.
6.0 3.0
11:52 13:47 11:52 13:47
5.5
2.5
5.0
intensity I Hα656.28 [a.u.]

4.5
2.0
burden level [m]
4.0
3.5 1.5
3.0
1.0
2.5
2.0
0.5
12:00 13:00 14:00 15:00 12:00 13:00 14:00 15:00
local time [hh:mm] local time [hh:mm]
Fig. 14.73 The left diagram shows the burden level during several hours. Two slippings of the
blast furnace occur. After the slipping, a high output of the dust and the gas amount was monitored
as can be seen in the hydrogen signal (right diagram) measured with LIBS at the top gas tube
of RR
All gained laser data were correlated to data of the blast furnace operation. The
monitoring of the dust revealed information about the charging process and the
correlation of the element concentrations to operating parameters like hot blast
or differential pressure (pressure difference between the top and the hot blast).
Moreover, it enabled the observance of differences in the dust discharge evoked
by effects like the shutdown or the slipping of a blast furnace.
Figure 14.73 shows such a slipping behavior monitored at the blast furnace of
RR: a sudden change in the burden level is followed by a high output of dust and gas
which can be monitored by the LIBS signal of hydrogen (right graph of Fig. 14.73).
Figure 14.74 shows the relative concentrations of Na and Zn during operation of
the blast furnace with decreasing hot blast which results in a smaller output of dust
and a corresponding decrease in the Na and Zn signals.
The periodical changes in the circulating materials result from the charging
process of the blast furnace with coke and burden. Sodium and potassium are
coming mainly from the coke, while zinc is carried into the blast furnace by the
ore and sinter. It was observed that the signals of Na and K as well as the signals of
Zn and Pb are running parallel.
4.6 × 105
2.7 Na 2.4 Zn cold blast
4.5 × 105
2.4 4.4 × 105

2.1
amount of cold blast [m3n / h]

4.3 × 105
intensity [a.u.]
intensity [a.u.]
2.1
1.8 4.2 × 105
1.8 4.1 × 105
1.5
4.0 × 105
1.5
3.9 × 105
1.2
3.8 × 105
1.2
12:36 13:04 13:32 14:00 14:28 12:36 13:04 13:32 14:00 14:28 12:40 13:08 13:36 14:04 14:32
local time [hh:mm] local time [hh:mm] local time [hh:mm]
Fig. 14.74 LIBS signal of sodium Na and zinc Zn during decreasing hot blast operation of the
blast furnace at TKS
sinter coke Fe II 271.44 nm
14 6:00 m 6:30 m 8:00 m 9:00 m

4100
6:15 m
12
6:05 m
charged material [t]
I Fe 271.44 nm [a.u.]
8:00 m
10 8:55 m
4000
8
6 3900
2 3800
11:30 11:40 11:50 12:00 12:10 12:20

local time [hh:mm]
Fig. 14.75 Monitoring of iron at the top gas tube of RR. The time of putting sinter or coke into
the blast furnace is marked by symbols. The periodical fluctuation of the LIBS intensity of Fe is
mainly caused by charging of sinter. The time between two maxima is consistent with the time
between two charging maxima of sinter
Comparing the monitored data of the blast furnace at RR with the data taken
at the blast furnace of TKS, similarities and differences can be observed. Similar
to TKS, the progression of the circulating materials is influenced by the charging
process. But different from TKS, the blast furnace at RR is working with open
charging with a charging process which runs irregular and much faster than at TKS.
Figure 14.75 gives an example of a measurement of Fe. The periodical changes
result from charging of sinter which is indicated by the maxima of the iron intensity.
14.8 Aerosoles 361
The in situ monitoring of relative concentrations of the elements Na, K, Zn, Pb, as
well as C, N, O, H, Ca, and Fe in the top gas has been demonstrated successfully at
three blast furnaces. Continuous measuring periods of up to 10 h a day were carried
out in the top gas stream of a blast furnace showing the capability of the LIBS
instrumentation to measure in a harsh industrial environment. The optical window of
the measuring probe is kept free from dust for more than 10 h continuous operation
in the top gas stream due to an improved nitrogen flow. The LIBS instrumentation
delivers for all ten elements every 10 s a measuring value which is the mean of
100 laser bursts. The standard deviation is between 2.3% and 7% for the circulating
materials. Various safety tests of the measuring equipment had been carried out to
ensure that the safety concept (valves, stuffing boxes, etc.) was suitable.
14.8 Aerosoles
The risk of fine (0:1–2:5 m) and ultrafine (0:1 m) particulates to affect the
organism and human health depends on particle size [14.114, 14.115]. Ultrafine
particulates can enter alveoli and, through them, penetrate the blood vessels and
cells of the human body. Since the health risks are governed by the composition
of the particles – especially when heavy metals are involved [14.116, 14.117] –
there is a need to measure the chemical composition as a function of particulate
size. In addition, the size-dependent composition may serve as a signature for the
particle generating process and source apportionment. This information allows an
identification of particle emitting processes and a focus upon reducing emissions
from the most relevant processes of, e.g., large-scale industrial facilities.
14.8.1 Aerosoles Sampled on Substrates
A study is presented where particle sampling methods are combined with laser-
induced breakdown spectroscopy [14.118]. While standard analytical methods
applied to emission samples are costly and time consuming, the use of LIBS offers
the possibility to gain composition information after sample collection much faster.
Furthermore, LIBS can be applied on-site.
Panne et al. and Yamamoto et al. investigated particulate matter collected on
substrates [14.119, 14.120]. Yamamoto used an electrical low-pressure impactor
(ELPI) with Al foils for size classification, while Panne took quartz fiber as a
substrate to collect the aerosol without any size classification. In the study described
in the following, the laser pulse energies used to excite the plasma are an order
of magnitude smaller than those used in the cited works. While Panne measured
absolute emission intensities related to the mass of the collected material, the ratio
of two intensities related to size classes was investigated here. As a reference for
the line ratios Yamamoto added yttrium to the samples, which is not a constituent of
the collected particulates. Line emissions from the elements iron and calcium were
used as internal reference for those are the dominant constituents of the particulates
studied. Furthermore, the possibility of source apportionment was not discussed by

Panne and Yamamoto.
Arnold and Cremers analyzed thallium particles smaller than 20 m on filters
[14.121]. They did not classify the particles by size and related the intensity to
the mass of the particles collected. To investigate the capabilities of LIBS for this
application, the influence of various measurement parameters were studied in a
laboratory setup. The elements Ag, Ba, Cd, Co, Cr, Cu, Hg, V, Mg, Mn, Na, Ni,
Pb, and Zn were analyzed. Calibration curves were determined for 28 emission line
ratios, where each analyte line was referenced to an iron and a calcium line. Finally,
samples of particulate originating from steelmaking processes were analyzed based
on this set of calibration curves.
For LIBS analysis, the airborne particles are collected using an electrical low-
pressure impactor (ELPI, Dekati Ltd.). This system combines a cascade impactor
with a charger and an electrometer to measure the particle load during the sampling
process. The impactor collects the aerosol particles separately in 13 size fractions
according to their different aerodynamic diameters ranging from 40 nm to 13 m.
As shown in Fig. 14.76, the impactor consists of a series of jet and collection plates.
Large particles are deposited first, while smaller ones can follow the air stream
to the lower stages with the cut-off size depending on the aerodynamic design of
each stage. The particulate matter is collected on Al foils with a thin grease coating
applied as exchangeable substrates on the collection plates. The conductivity of the
substrates is necessary for the electrometer operation, while the grease increases the
adhesion of the surface for an efficient collection of the particles. Since the grease
layer is very thin, the electrical capacity of the ELPI is not influenced.
Figure 14.77 shows an allocation of mass and particle number, plotted as a
function of the aerodynamic diameter of the particles, typical for calibration samples
as obtained from electrometer measurements. The aerodynamic diameter is defined
as the diameter of a sphere with a standardized density of 1 g=cm3 , which has the
same descent rate as the particle itself. The data were taken during the collection
of a calibration sample lasting 30 min. The width of the bars indicates the limits at
which 50% of the particles with the respective diameter are deposited on a particular
stage. On the substrates, the material is deposited not homogeneously, but only at
the locations of the jets, as shown in Fig. 14.78. Depending on the design of the jet
plates, there are up to several dozen depositions with diameters of 0:5 mm on a
substrate area having a diameter of 20 mm.
The area covered by particles depends on the stage and can be of the order of
some percent of the total substrate surface. Furthermore, the diameter of the spots
is independent of the composition of the particles used for sampling. For the larger
size fractions, a lower number of larger jets depositions is found.
Table 14.16 shows the number of jets as well as the respective jet diameter for
each stage of the impactor [14.122]. In order to calibrate the LIBS measurements,
a set of calibration samples were generated. Salt solutions were nebulized by an
aerosol generator, guided through a diffusion dryer and sampled by the ELPI as
described above. The produced aerosol is assumed to represent the salt composition
in the solution, which is supported by the calibration measurements (see below).
14.8 Aerosoles 363
The salt solutions were prepared from chlorides of the metals stated above. The
concentration of these metals was varied in relation to a constant iron and calcium
concentration which was included as reference material in all samples. Since a large
range of concentrations is expected in the industrial samples, the concentrations for
the calibration samples are varied according to a logarithmic scale covering three to
Fig. 14.76 Cascade

impactor. ST1–ST13 D
stages, JP D jet plate,
CP D collection plate,
BF D backup filter,
I D inlet, O D outlet
20 mass ST6
number 15
number of particles [108]
ST7
15
mass M [mg]
ST5 10
10
ST8
ST4
5
5
ST9 ST10
ST3
ST1 ST2
0 0
0.1 0.5 1 2 3
aerodynamic diameter D P [μm]
Fig. 14.77 Typical allocation of mass and particle number for the used impactor plotted versus
the aerodynamic diameter, ST1–ST10 D stages of the impactor, cf. Fig. 14.76
five orders of magnitude. Thus, the molar mixing ratio for calibration ranges from
105 to 0.1.
An experimental LIBS setup was developed to analyze particulate matter on
substrates. Due to the characteristics of the samples, laser pulse energy as well as
the pressure of the ambient gas has to be adapted to achieve an efficient excitation
of a thin layer of deposited particulates into the plasma state. An ablation of the
substrate material has to be kept as small as possible to reduce interference by
high intensity lines of the Al substrate. A schematic drawing of the experimental
setup is shown in Fig. 14.79. The measurements were performed in a vacuum-tight
measuring chamber filled with air, helium, or argon. This chamber allows the study
of the influence of the ambient gas type and pressure on the performance of the
LIBS analysis. In the measurement chamber, the sample is positioned on a sample
holder and the excitation laser beam is focused onto the sample surface from the top
with an achromatic lens system (f D 60 mm) as shown in more detail in Fig. 14.80.
The sample holder is mounted on xy-translation stages within the chamber. A Q-
switched Nd:YLF slab laser operating at 1;047 m with repetition rates of up to
Fig. 14.78 Microscope

image of a substrate with
FeCl3 particulates (ST7,
magnification 100) showing
the areas of deposited
particulates (dark spots)
corresponding to the
geometry of the respective jet
plate
Table 14.16 The number of Stage Number of jets Jet diameter (mm)
jets and the respective jet
diameter for each stage 1 69 0.3
2 58 0.25
3 21 0.3
4 19 0.3
5 27 0.3
6 50 0.3
7 48 0.4
8 20 0.7
9 17 1.0
10 14 1.4
11 3 3.2
12 1 6.3
13 1 8.3
14.8 Aerosoles 365
Fig. 14.79 Experimental setup. AL D alignment laser; CL D stage controller; L D laser;

FO D fiber optic; SP D spectrometer; PG 1; 2 D pulse generator; PD 1; 2 D photodiodes; ST D
threshold switch
Fig. 14.80 Optical arrangement. FO D fiber optic, LB D laser beam, O D focusing objective,
PD D photodiode, S D sample, W D protection window
1 kHz delivers laser pulses with pulse energies of up to 4 mJ, and pulse widths
of 7 ns. It was found that pulse energies of 0.6 mJ were sufficient for the desired
excitation, and higher energies resulted mainly in additional ablation of substrate
material. Based on a crater diameter of approximately 30 m, the fluence can be
estimated to 85 J=cm2 . The results presented were gained with single pulses only,
even though the laser system is capable of generating double pulses. Since the
complete analyte material is already evaporated or ablated by the first laser pulse
irradiating a certain spot, a subsequent laser pulse is likely to mainly excite substrate
material unless its energy can solely be made available for a re-heating of the
plasma. Additionally, the reduction of the local ambient gas density as for collinear
double pulse configurations (cf. Chap. 8) can be realized in this setup by the pressure
chamber. It is observed that plasma ignition on aluminum and grease is achieved
even at the lowest possible energy of the laser at 0.03 mJ. However, at such settings
the laser does not operate very stable; hence, a pulse energy of 0.6 mJ was used in
the following.
The plasma light is collected by an optical fiber. The fiber is positioned at a
distance of 20 mm to the plasma at an angle of about 40ı to the sample surface.
It is protected by a silicate glass window to prevent depositions on the fiber
entrance face by ablated material. The received light is guided to an echelle
spectrometer (ESA3000, LLA Instruments) covering a wavelength range from 200
to 895 nm with a spectral resolution of 5 to 20 pm. For this study 100 spectra,
each taken at a different location, were accumulated following a laser track as
illustrated in Fig. 14.81. Thereby, the tracks were chosen to be half circles covering
the highest possible percentage of areas with particulate depositions. While the
axes are moving, the laser is triggered periodically irrespective whether a particle
accumulation (dark spot in Fig. 14.78) is hit or not. With the experimental setup
chosen, there was no possibility to selectively irradiate the particle accumulations.
For the evaluation, the total intensity of a spectral line is integrated, preventing
influences from other lines nearby. The intensities’ standard error is calculated from
the standard deviation of the signal background on both sides of the spectral line. In
a first step, uniform pure or greased Al foils are measured to study the influence of
the substrate material.
Fig. 14.81 Schematic view

of a measurement track on a
substrate, LT D laser track,
AC D areas with particles
deposited (dark spots in
Fig. 14.78), S D sample foil
14.8 Aerosoles 367
The influence of various ambient gases under different pressures on the line
intensity was investigated. The importance of the ambient gas has been demon-
strated in previous publications (cf. Sect. 7.3) [14.123]. The measurement chamber
was evacuated and then refilled with laboratory air under defined pressures from 1 to
1,000 mbar. The intensity of the aluminum line at 257.509 nm as a function of the
pressure is shown in Fig. 14.82. At a pressure of 5 mbar, the line intensity reaches
a maximum, which is approximately a factor of five greater than at 1,000 mbar.
For helium, the maximum line intensity is achieved at 10 mbar with an intensity
three times higher than for air, whereas argon shows a maximum at 600 mbar
with an intensity two times higher than the maximum for air. Tests performed with
particulates on the substrate showed that the use of air or helium at low pressures
had a negative effect on the line intensities of elements comprised in the particulate
matter. Obviously, the low pressure facilitates a detaching of the particles in the
vaporizing phase, thus dispersing the particles in a larger volume, so that fewer
particles are enclosed in the plasma itself. Therefore, argon at 600 mbar was used
for the further measurements.
Figure 14.83 shows the line intensities of Al 257.509 nm and C 247.856 nm as a
function of the delay between the laser pulse and the start of the signal integration
tdelay measured on greased Al foils. The integration time tint was kept constant
to 50 ns. With rising tdelay , the signal from aluminum decreases monotonically. On
the other hand, the carbon signal is constant for delays up to about 300 ns. To keep
the substrate signal (aluminum) low while simultaneously maintaining the particle
signal (simulated by the grease coating) as high as possible, the delay time of
tdelay D 180 ns was chosen for calibration and sample measurements. Additionally,
it was observed that the signal-to-noise ratio shows a maximum for a delay of a
air
1500 x 20 He
Ar
intensity I Al 257.509 [a.u.]
1000
500
0
1 10 100 1000
pressure [mbar]
Fig. 14.82 Intensity of the aluminum Al 257.509 nm line as a function of the gas pressure for the
gas types air, helium, argon for a pure, clean aluminum foil. For better visibility, all error bars are
magnified by a factor of 20
t int = 50 ns Al 257.509 nm
1000 C 247.856 nm
intensity [a.u.]
x5
100
0 500 1000 1500 2000

t delay [ns]
Fig. 14.83 Line intensity of Al 257.509 nm and C 247.856 nm as a function of the delay time
tdelay . For better visibility, all error bars are magnified by a factor of 5
few hundred nanoseconds depending on the chosen line. This is due to the decrease
of the continuum background shortly after plasma ignition. For the calibration and
sample measurements, the integration time tint of the spectrometer was set to 3 s.
No significant improvement of signal intensity was observed for longer integration
times.
With the chosen parameters the prepared calibration samples were measured
using concentric circles as shown in Fig. 14.81. One hundred pulses were accumu-
lated to enhance the signal-to-noise ratio. The intensity ratio of an analyte line to a
reference line was plotted versus the molar mixing ratio given by the preparation of
the calibration samples. Figure 14.84 shows as an example the calibration curves for
the line ratio of Zn 213.857 nm to Ca 422.673 nm for the stages three to eight. It can
be observed that the curve progression is nearly the same for all stages. Obviously
there is no significant dependence between particle size and the slope of the
calibration curves, although the relative error for small molar mixing ratios can be
up to 25%. This supports the assumption that the particle composition represents the
salt composition of the solutions from which the aerosol was generated. Therefore,
it is possible to use a mean curve for all stages for calibration (see dotted line in
Fig. 14.84). Due to its independence on stage numbers, the mean curve was also
taken for calibration of the stages other than ST3 to ST8. Potential matrix effects
within the calibration curves caused by the substrate material could not be observed,
since only one substrate material was studied.
An example for a mean calibration curve is shown in Fig. 14.85 for lead
referenced to an iron line. The intensity ratio increases with the molar mixing ratio
following a power law with an exponent of 0.7. Calibration curves were gained for
all elements of interest – Ag, Ba, Cd, Co, Cr, Cu, Hg, V, Mg, Mn, Na, Ni, Pb, and
Zn – referenced to iron or calcium lines.
14.8 Aerosoles 369
intensity ratio I Zn 213.857 / I Ca 422.673 [1] 1
0.1
ST3
ST4
ST5
0.01 ST6
ST7
ST8
mean
1E-3
fitting of mean
1E-3 0.01 0.1 1

molar mixing ratio [1]
Fig. 14.84 Calibration curves of zinc and stages 3–8 showing monotonously increasing relations
6
intensity ratio I Pb 368.347 / I Fe 381.584 [1]
0.1
0.01 0.1 1
molar mixing ratio [1]
Fig. 14.85 Calibration curve of lead referenced to iron as a function of the molar mixing ratio
The limit of detection (LOD) is the lowest quantity of an element that can be
distinguished from the absence of that substance. It is estimated from the mean
of the blank samples (samples without the respective element, cf. Sect. 11.2), the
standard deviation of the blank samples, and a confidence factor depending on the
amount of blank samples. The LOD is usually estimated by the 3scriterion [11.3].
A linear relation between intensity ratio and molar mixing ratio is a precondition
for this approach. In the study presented here, only a small set of molar mixing
ratios could be realized which was not sufficient to fit a linear curve to the data
points at low concentrations. For that reason, a fit curve was used described by the
expression y D ˛ x d and the LOD is determined with xLOD D ˆ.sb =˛/1=d , where
the confidence factor ˆ D 2:9 equals the factor given in [11.3] for the respective
amount of blank samples and sb is the standard deviation of the measuring values of
the blank sample. For lead referenced to iron, the LOD amounts to 2.7%.
Industrial emission samples were analyzed based on the determined calibra-
tion curves using the same measurement parameters. The results are shown in
Figs. 14.86–14.88. Figure 14.86 shows the molar concentration of lead in relation
to iron measured with LIBS as a function of the particle diameter. Three samples
were taken from direct exhausted air of a sintering process, while two samples were
taken from steel production. The sinter samples show an enhanced concentration for
particle diameters between 50 and 500 nm. A peak can be observed at approximately
200 nm. Particles with diameters greater than 1 m have concentrations that
are two orders of magnitude lower than for smaller particles. The samples from
steel making, on the other hand, have a relatively flat distribution. The maximum
concentration is attained for particles at about 500 nm, but the concentration level is
three orders of magnitude lower than the concentrations determined for the sintering
process. The size-dependent curves are characteristic for the process where the
particles are emitted. Hence, the curves can be used as a signature to determine
the source of the measured particulates.
Figure 14.87 shows the molar concentration ratio of copper to iron as a function
of the particle diameter revealing a similar result like in Fig. 14.86, as for a particle
diameter of about 0:2 m the molar concentration of copper to iron for sintered
samples is about two orders of magnitude higher than for steel-making samples.
Therefore, the sample source can be determined as in the example described before.
101
sinter plant A; sp1
molar concentration ratio Pb / Fe
sinter plant A; sp2

100
sinter plant B
BOS; sp1
10-1 BOS; sp2
10-2
10-3
10-4
0.1 1 10
particle diameter [μm]
Fig. 14.86 Results of the LIBS analysis of collected samples showing the molar concentration
ratio of lead to iron as a function of the particle diameter. sp D sample
14.8 Aerosoles 371
molar concentration ratio Cu / Fe sinter plant A; sp1

sinter plant A; sp2
10-1 sinter plant B
BOS; sp1
BOS; sp2
10-2
10-3
0.1 1 10
particle diameter [μm]
ratio of copper to iron as a function of the particle diameter
sint. pl. A; sp 1
102
sint. pl. A; sp 2
molar concentration ratio Zn / Ca
sint. pl. B
101
BOS; sp 1
BOS; sp 2
100
10-1
10-2
10-3
0.1 1 10
particle diameter [µm]
ratio of zinc to calcium as a function of the particle diameter
Figure 14.88 shows the molar concentration ratio of zinc to calcium as a function
of the particle diameter. For particle diameters between 0.3 and 5 m, the molar
concentration ratio for samples from steel making is on average one to two orders
of magnitude higher than for sintered samples. Thus, the sample source can also be
distinguished with a high level of significance.
In the experiments described, the majority of laser-induced plasmas were
generated on the substrate only, because it was not possible to selectively irradiate
the particle accumulations. To increase the quality of the results, the amount of
irrelevant data – i.e., the data resulting from plasma ignition on grease and sub-
strate – can be minimized. With, e.g., a microscope coupled with a camera particle
accumulations can be detected and distinguished from substrate material. With the
use of photodiodes – cf. PD 1 and PD 2 in Fig. 14.79 – the light of the alignment
laser (632 nm) scattered by the deposited particulates can be detected. This signal
triggers a threshold switch which enables the laser operation for LIBS analysis of
particle accumulations. With such an approach a significant enhancement of the
signal-to-noise ratio is expected.
14.8.2 Aerosoles in an Air Stream
A number of studies have successfully demonstrated the use of laser-induced

breakdown spectroscopy (LIBS) for the analysis of aerosol particles [14.118,
14.119, 14.124–14.132].
While standard analytical methods like mass-spectrometry, e.g., ICP-MS, are
time consuming, LIBS is a fast method that can be automated and applied inline
and onsite. As described in Sect. 14.8.1 some of the works concern the analysis of
particles sampled on filter substrates fractionated in size classes by an impactor, and
a subsequent LIBS analysis is undertaken in the lab [14.118, 14.119, 14.128]. For
inline applications, this approach is not adequate; hence, there is a need to analyze
aerosol particles directly in an air stream. In this regard, D. Hahn and coworkers
have investigated the use of LIBS for aerosol particle analysis in a number of studies
dealing with aerosol sampling statistics, conditional data analysis, and plasma–
particle interaction [14.129–14.132].
In this section, fundamental studies are described to develop a system for size-
dependent aerosol analysis directly in an air stream with regard to industrial appli-
cations [14.133]. For inline process control, there is a need for a fast multielement
analysis with a fast response-time of the system. In the following the development
of a measurement system comprising a Paschen–Runge spectrometer with rapid
multichannel electronics is presented enabling fast reactions to changes in the
chemical composition of particulate matter. The focus lies on basic investigations
of the system behavior, which are necessary for its calibration. Compared with the
analysis of solid samples, the calibration is more complex in this case since it can
be influenced by both the element concentrations in the particles and the total mass
of particulate matter in the aerosol stream.
Additionally, the measurement process is influenced by the size of the individual
particles, the carrier gas, and the particle location within the plasma [14.134–
14.136]. In the following investigations focused on the effect of the particle mass
concentration and the relative elemental concentration are presented. The temporal
plasma evolution of the analyte response of particle matter and gas phase species is
compared. Electron density and temperature are derived as a function of time. The
system behavior is investigated for the analysis of monodisperse and polydisperse
aerosols under different measurement conditions.
14.8 Aerosoles 373
Artificial aerosol is nebulized by an aerosol generator (AGK-2000, Palas,

Germany) into droplets of up to 2 m diameter using N2 . The aerosol is then guided
via a diffusion dryer by using of N2 as a carrier gas into a measurement chamber
to provide a stream of solid particles ranging between 20 and 800 nm in particle
diameter. The generator can produce aerosol of different elements by nebulizing
different salt solutions. In the following the analysis is restricted to the elements
calcium and sodium. It can be assumed that the composition of the dried particles
equals the composition of the dissolved material.
To provide size-resolved measurements, a commercially available SMPS (Scan-
ning Mobility Particle Sizer, TSI, USA) is used to fractionate particles by size. The
SMPS system comprises a DMA (differential mobility analyzer) model 3080L and
a CPC (condensation particle counter) model 3785. The DMA separates particles
in a gas stream in a size range from 10 to 1,000 nm aerodynamic diameter. Due to
their electrical mobility, which is inversely proportional to their size, particles are
fractionated by size, depending on the electrostatic field strength of the DMA and
the sheath gas flow rate, and counted using the CPC. By scanning the electrostatic
field strength of the DMA and simultaneously counting the particles, an aerosol size
distribution is obtained. An example of a particle size distribution given in number
and mass concentration is shown in Fig. 14.89. The particle size was scanned from
19 to 800 nm in 27 logarithmic steps. The line intensity for calcium measured with
LIBS is also given.
1000 18
Ca II 393.366 nm
16
22 error bars x 10
line intensity I Ca II 393.366 nm [a.u.]
800 particle number concentration

14 particle concentration [%]
1 particle mass concentration
21 23
12
600
20 10
8
400 24
19 6
151617 25
200 14 18 4
13
12 2
11 26
0 8 910
12 34567 27 0
10 100 1000
particle diameter DP [nm]
Fig. 14.89 Particle size distribution given in number- and mass distribution. The size distribution
is scaled in percentage of the total particle number and the total mass, respectively (right ordinate).
Additionally the obtained size-dependent LIBS-signal for Ca is plotted and scaled to the left
ordinate. The measurement points of the mass concentration are marked with numbers in ascending
order
Fig. 14.90 Setup for the

LIBS experiments.
SP D spectrometer,
ES D entrance slit,
AF D aerosol flow,
CF D co-flow, MC D
measuring chamber, LB D
laser beam propagating
perpendicularly to the plane
of projection, SM D
spherical mirror,
SA D symmetry axis,
VP D vacuum pump,
a D 50 mm; and b D 10 mm
The outlet of the DMA is connected via a conductive silicon tube of 2 m length
and 5 mm diameter to the measurement chamber of the LIBS setup, cf. Fig. 14.90.
The gas pressure in the measuring chamber is set to 980 mbar to generate a
continuous aerosol flow of 1 l pm through the DMA for all experiments.
The number distribution shows a maximum at 70 nm particle diameter and the
mass distribution at about 400 nm. The total number and mass of particles as a sum
of all size classes were 4:5 106 particles/cm3 and 10 mg=m3 , respectively, in this
measurement. The dips occurring at 34 and 200 nm particle diameter (marked with
arrows in Fig. 14.89) in the number distribution and in the mass distribution are due
to two different measurement modes of the particle counter. Aerosols of low particle
concentration (<2 104 particles/cm3 / are counted particle by particle (single
particle only). In the second mode for particle number concentrations exceeding
2 104 particles/cm3 , the particle number is measured by photometry. The dips
appearing in the size distributions are indications of the switching between these
measurement modes, i.e., single particle counting and photometric mode.
For the analysis of the particulate matter in the gas stream, a LIBS setup was
developed which is shown in Fig. 14.90. All experiments are carried out with
a flashlamp pumped Q-switched Nd:YAG laser at the fundamental wavelength
D 1;064 nm, a repetition rate of vrep D 10 Hz and a pulse duration of L D 10 ns.
Pulse energies of up to 300 mJ can be used. Analyses are undertaken in a vacuum
chamber enabling the use of a co-flow in addition to the aerosol stream (CF in
Fig. 14.90). This co-flow is guided into the chamber via two diffuser inlets beside
the aerosol inlet which are arranged 50 mm above the aerosol inlet (distance a in
Fig. 14.90).
The expanded laser beam with a diameter of 18 mm is focused into the flow
stream perpendicular to the aerosol flow at a distance b D 10 mm downstream of
the aerosol exit aperture (2 mm diameter) using an achromatic lens having a focal
length of f D 110 mm. Under these conditions, breakdown is initiated by each
laser pulse. The generated plasma light emission is collected by a spherical mirror
and focused directly onto the entrance slit of a spectrometer in Paschen–Runge
14.8 Aerosoles 375
configuration (cf. Sect. 4.2) in which channel electron multipliers are installed for
each spectral line of interest. For multichannel measurements, the fast multichannel
integrator electronics (MCI) is used (cf. Sect. 4.5). To detect spectral line profiles
and the whole emission spectrum, an echelle spectrometer (cf. Sect. 4.2) is used in
addition. In this case, the plasma light emission is collected by a quartz lens into an
optical fiber.
To investigate the temporal resolution of the LIBS setup, aerosol generated
from sodium and calcium chloride solutions was injected into the measuring
chamber. Therefore, the pressure in the measuring chamber was held constant to
980 mbar to provide a continuous gas flow of 1 l pm as described above. NaCl and
CaCl2 particles were injected into the gas flow for nearly 10 s in succession. This
measurement was carried out without and with the use of an additional argon co-
flow. While using the argon co-flow with a flow rate of 5 l pm the pressure in the
measuring chamber was still 980 mbar. The time in which the complete gas volume
in the chamber is exchanged is estimated to be 8 min without the co-flow. Through
the use of the co-flow with a flow rate of 5 l pm, the gas volume is exchanged within
approximately 1.3 min.
The plasma emission was observed using the echelle spectrometer with a delay
time of 800 ns and an integration time gate of 30 s. Two pulses were averaged
which indicates a temporal resolution of 0.2 s. Figure 14.91 shows the time-
dependent line intensity for the two elements Na and Ca.
Without co-flow the time in which the signal rises between 10% and 90% of
the maximum signal amounts to 3.5 s (Fig. 14.91, top). In the case of an argon co-
flow this value can be reduced to 2.5 s. The 1=e decay time of the signal is in the
order of 20 s without the co-flow, and the duration until the signal can no longer be
distinguished from the background is longer than 90 s. In contrast to that, the use of
a co-flow reduces this time to approximately 20 s: see Fig. 14.91, bottom. The time
when the signal has reached 1=e of its maximum amounts to 10 s.
Using no additional co-flow, the nonaveraged line intensity of the calcium line is
very noisy whereas with the Ar co-flow this noise level is reduced.
The fast response time is very important for industrial applications to be able to
react to changes in the elemental composition. A fast decay time of the system
is also important for the temporal resolution and the size resolution concerning
SMPS measurements. Assuming a SMPS scanning time of 135 s, which is used
here, approximately seven size classes can be swept consecutively.
The emission response to varying concentrations of Na within the salt solution
was investigated. Figure 14.92 shows the obtained calibration curve. The laser
energy was 110 mJ in this case, and the plasma emission was observed using the
echelle spectrometer. Spectra were recorded using a delay time of 800 ns and an
integration time gate of 20 s. A series of 100 (10 10 pulses) spectra were
recorded and averaged. Error bars are equal to the standard deviation of the ten-pulse
averages. The fit curve of the double-logarithmic plot shows a linear behavior with
a slope of 0.9 up to a concentration of Na of 0.2 m.-%. Beyond this concentration,
the slope of the curve decreases to 0.5.
without co-flow
350 3500
Na I 588.995 Ca II 393.366
300 3000
line intensity I Ca II 393.366 [a.u.]

line intensity I Na I 588.995 [a.u.]
2 s mean 2 s mean
250 2500
200 Na Ca 2000
150 1500
100 1000
50 500
Na injection Ca injection
0 0
0 20 40 60 80 100 120
t [s]
with co-flow
350 3500
Na I 588.995 Ca II 393.366
line intensity I Na I 588.995 [a.u.]
300 3000
line intensity I Ca II 393.366 [a.u.]

2 s mean 2 s mean
250 Na Ca 2500
200 2000
150 1500
100 1000
50 500
0 0
0 20 40 60 80 100 120
t [s]
Fig. 14.91 Intensity of a sodium and a calcium line as a function of time without (top) and with
(bottom) the use of an additional co-flow into the measurement chamber. Additionally the moving
intensity average over 2 s is plotted. The arrows mark the time at which NaCl and CaCl2 particles
were injected into the measurement chamber via a tube of 2 m using N2 as a carrier gas
There are two possible explanations for this behavior. On the one hand, it can be
explained by the concept of the curve of growth (COG). According to Gornushkin
et al., who applied the COG method to laser-induced breakdown spectroscopy,
nonlinearity in the calibration function of solid samples for concentrations in the
range between 0.1 and 1 m.-% is observed depending on the delay time and the
laser energy used [14.137]. The double logarithmic plot of the theoretical curve
of growth has an asymptote with slope 1 for concentrations up to 0.1 m.-%, and
14.8 Aerosoles 377
1000
line intensity I Na I 588.995 nm [a.u.]
100
error bars × 10
10
0.01 0.1 1
Na concentration in solution [m.-%]
Fig. 14.92 Double-logarithmic calibration curve for particles of all size classes obtained from
sodium solution using N2 as carrier gas without co-flow. Error bars are multiplied by a factor of
10 for concentrations 0:05 m.-%
another with slope 0.5 for higher values. The intersection between these asymptotes
marks the point of transition between low and high optical density of the plasma.
Because the conditions differ from those used here no quantitative comparison can
be made. Gornushkin et al. studied the curves of growth of Cr in a steel matrix for
solid samples.
The emission of the two lines Na lines, i.e., Na I 589.592 nm and Na I
588.995 nm was tested for self-absorption. Saturation effects can be observed above
a concentration of 0.1 m.-% supported by the line intensity ratio of these two lines,
but the accuracy of the data is not high enough for quantification. For the Na I
589.592 line, which is expected to show a weaker self-absorption, a slope of 0.7
was estimated for concentrations 0:2 m.-%.
On the other hand, the observed behavior in Fig. 14.92 could be due to growing
particle size with increasing concentration of the solution. The increase of the
concentration results in the growth of the size of solid particles remaining when
the liquid droplets dry out, which leads to a rise in the total mass concentration
of the aerosol. Thus, on the one hand, the reduced increase in the signal above a
concentration of 0.2 m.-% can be due to incomplete dissociation of larger particles
or, on the other hand, due to stronger losses of the larger particles during the
transport to the measuring chamber.
However, both effects, i.e., plasma properties and particle size effects, may play
a role here. To investigate the influence of the argon co-flow on the LIBS signal,
the signal response to a variation of sodium and calcium with and without the use
of the co-flow was measured. In this experiment, solutions of sodium and calcium
chloride with an elemental mass concentration of 1 m.-% each were generated.
For the LIBS measurements, the solutions were mixed such that the content of
0.30
intensity ratio Na / Ca with argon co-flow
I N aI589.0nm / I C aII393.4nm [1]
0.25 intensity ratio without argon co-flow
0.20
0.15
0.10
0.05
error bars × 5
0.00
0 1 2 3 4 5 6 7 8 9
mixing ratio c Na / cCa [1]
Fig. 14.93 Line intensity ratio as a function of the mass concentration ratio of sodium and calcium
within the solution. Error bars are multiplied by a factor of 5
Na was successively increased and an aerosol stream was generated as described

before. Spectra were recorded with the echelle spectrometer integrated with a delay
time of 800 ns after the laser pulse and a gate width of about 30 s. The laser
energy was 200 mJ and a series of 200 (10 20 pulses) spectra were averaged.
The measurements were undertaken with and without an additional argon co-flow.
In Fig. 14.93, the obtained line intensity ratio of Na and Ca for both measurements
is plotted versus the mixing ratio of Na and Ca cations of the solution. Figure 14.93
shows that the variation of the mixing ratio is well represented by the LIBS signal.
The elemental mixing ratio can be clearly derived from the ratio of the measured
line intensities. The measurement with an argon co-flow reduces the ratio of the line
intensities by an offset. This can be explained by different plasma conditions and
excitation conditions for the lines. Because of the enhanced plasma confinement for
the argon co-flow, the degree of ionization is expected to increase, which causes a
reduced Na I to Ca II signal ratio. By approximating a polynomial function to the
data points, the standard deviation of procedure was estimated to be 0.16 for the
measurement without argon according to [11.4]. In case of the argon co-flow, this
value is reduced by a factor of 5 down to 0.032.
Size-resolved LIBS analysis of calcium particles were undertaken for different
measurement conditions and compared with the aerosol size distribution and the
particle mass concentration for each size class.
To provide a size-resolved analysis of particle composition, particles generated
from CaCl2 solution as described above were fractionated by size using the DMA.
The CaCl2 solution had a calcium elemental mass concentration of 0.23 m.-%.
The particle size was scanned from 19 up to 800 nm in 27 logarithmic steps, and
the fractionated particles were subsequently analyzed by LIBS for each size class.
The characteristic of the obtained emission line intensity for Ca II at 393.366 nm
14.8 Aerosoles 379
1000
23
line intensity I Ca II 393.366 nm [a.u.]
error bars ×10

22
800 290 nm 21
20
220 nm 24 390 nm
600 18 520 nm
19
25
1716
400 15 600 nm
14
13
200
12
11
910
0 16234578
0 2 4 6 8 10 12 14 16 18
particle mass concentration [%]
Fig. 14.94 Line intensity of Ca II 393.366 nm as a function of size-dependent particle mass

concentration. Particle diameters are represented by numbers in ascending order as in Fig. 14.89.
All error bars are multiplied by a factor of 10
with increasing particle size is shown in Fig. 14.89 together with the particle size
distribution. The plasma emission was observed using the echelle spectrometer as
described above and the laser pulse energy was set to 200 mJ.
Figure 14.89 allows a comparison of the line intensity of Ca with the particle size
distribution. The Ca line emission intensity follows the particle mass distribution
and attains a maximum in the particle diameter range between 350 and 450 nm as
the mass distribution does.
In Fig. 14.94, the calcium line intensity is plotted as a function of the particle
mass-concentration. Here, a correlation between the particle mass concentration and
the obtained LIBS signal can clearly be seen. For small particles, the line intensity
increases nearly linearly with mass up to a particle diameter of approximately
140 nm (no. 15) and a corresponding mass concentration of 4.5%.
From a particle mass concentration of 6% upwards, corresponding to a particle
diameter of 220 nm (no. 18) the signal rises more slowly. The offset of measurement
point no.18 is due to the two different measurement modes of the CPC (cf.
discussion of Fig. 14.89). A possible reason for the slower increase may be the
influence of the particle size in the laser-induced plasma. Laser-particle interaction
of larger particles is different from that of smaller particles. In this regard, one
aspect could be a different location within the plasma [14.134]. Furthermore, the
interpretation given for the Na calibration curve (Fig. 14.92) may also be applied
here, i.e., the particle size.
Figure 14.95 shows the line intensity of calcium Ca II 396.847 nm as a function of
particle diameter with and without the use of a co-flow. The LIBS signal is analyzed
by the Paschen–Runge spectrometer with a delay time of 2 s and an integration
line intensity I Ca II 396.847 nm [a.u.] 20000
16000
12000
8000 Ca II 396.847 nm without argon co-flow

Ca II 396.847 nm with argon co-flow
60 100 400 900

particle diameter D P [nm]
Fig. 14.95 Size-resolved LIBS signal of Ca particles with and without the use of an argon co-flow,
measured with the Paschen–Runge spectrometer
time gate of 25 s. The laser energy was 265 mJ in this case and 1,000 consecutive
single spectra were accumulated. By the use of the Ar co-flow, the LIBS signal is
enhanced by a factor of about 1.5. The use of the Ar co-flow does not influence the
characteristic of the size-dependent curve significantly.
References
14.1. J. Brandrup, M. Bittner, W. Michaeli, G. Menges (eds.), Die Wiederverwertung von

Kunststoffen (Carl Hanser Verlag, München, 1995)
14.2. C. Delgado, L. Barruetabeña, O. Salas, in Assessment of the Environmental Advantages
and Drawbacks of Existing and Emerging Polymers Recovery Processes, ed. by O. Wolf,
EC Joint Research Center, 22939 EN, 2007
14.3. H. Lucht, U. Plauschin, H. Dürr, Kunststoffe mit Infrarot-Messung sortenrein trennen.
Umwelt 23, 443–444 (1993)
14.4. D. Scott, A two-colour near-infrared sensor for sorting recycled plastic waste. Meas. Sci.
Technol. 6, 156–159 (1995)
14.5. M. Alam, S. Stanton, Sorting of waste plastics using near-infrared spectroscopy and neural
networks. Process Contr. Qual. 4, 245–252 (1993)
14.6. R. Feldhoff, D. Wienke, K. Cammann, H. Fuchs, On-line post consumer package
identification by NIR spectroscopy combined with a fuzzy ARTMAP classifier in an
industrial environment. Appl. Spectrosc. 51, 362–368 (1997)
14.7. N. Eisenreich, T. Rhode, Identifizieren von Kunststoffen. Kunststoffe 86, 222–224 (1996)
14.8. N. Basta, K. Fouhy, K. Gilges, A. Shanley, S. Hshio, Chem. Eng. 11, 37–43 (1990)
14.9. see e.g., products of the following companies: LLA Instruments GmbH, Justus-von-
Liebig-Str. 9, 12489 Berlin, www.lla.de; National Recovery Technologies Inc., 566
Mainstream Dr., Suite 300, Nashville, TN, http://www.nrtsorters.com, 2010
14.10. http://www.dradio.de/dlf/sendungen/firment/1020957, 2009/8/21
References 381
14.11. J. Jansen, C. van Hastenberg, Identifikation von Kunststoffteilen. Kunststoffe 84, (1994)
51–54
14.12. R. Sattmann, I. Mönch, H. Krause, R. Noll, S. Couris, A. Hatziapostolou, A. Mavro-
manolakis, C. Fotakis, E. Larrauri, R. Miguel, Laser-induced breakdown spectroscopy
for polymer identification. Appl. Spectr. 52, 456–461 (1998)
using multiple Q-switch Nd:YAG laser pulses. J. Phys. D Appl. Phys. 28, 2181–2187
(1995)
14.14. A. Zell, Simulation Neuronaler Netze (Addison Wesley, Berlin, 1994)
14.15. R. Pearse, A. Gaydon, The Identification of Molecular Spectra (Chapman and Hall,
London, 1976), p. 83
14.16. H. Griem, Plasma Spectroscopy (McGraw-Hill, New York, 1964), p. 305, 538
14.17. R. Hecht-Nielsen, Neurocomputing (Addison Wesley, Berlin, 1990)
14.18. Ullmann’s Encyclopedia of Industrial Chemistry, vol. A 20 (VCH Publishers, Weinheim,
1992)
14.19. European Project, Development of multipurpose industrial units for recycling of plastic
wastes by on-line pattern recognition of polymer features (Sure-Plast), contract no.
BRPR-CT98–0783, project no. BE97–4890, 2001, samples prepared by Gaiker, Bilbao,
Spain
14.20. M. Stepputat, R. Noll, On-line detection of heavy metals and brominated flame retardants
in technical polymers with laser-induced breakdown spectrometry. Appl. Optics 42,
6210–6220 (2003)
14.21. J. Marshall, J. Carrol, J.S. Crighton, C.L.R. Barnard, Industrial analysis: metals, chemi-
cals and advanced materials. J. Anal. At. Spectrom. 9, 319–356 (1994)
14.22. R. Zenobi, Modern laser mass spectrometry, Fresen. J. Anal. Chem. 348, 506–509 (1994)
14.23. C. Lazik, R.K. Marcus, Electrical and optical characteristics of a radio frequency glow
discharge atomic emission source with dielectric sample atomization. Spectrochim. Acta
B 48, 1673–1689 (1993)
14.24. Z. Weiss, New method of calibration for glow discharge optical emission spectrometry.
14.25. A. Golloch, D. Siegmund, Sliding spark spectroscopy – rapid survey analysis of flame
retardants and other additives in polymers. Fresen. J. Anal. Chem. 358, 804–811 (1997)
14.26. J. Anzano, I. Gornushkin, B. Smith, J. Winefordner, Laser-induced plasma spectroscopy
for plastic identification. Polym. Eng. Sci. 40, 2423–2429 (2000)
14.27. M. Tran, Q. Sun, B. Smith, J. Winefordner, Determination of F, Cl and Br in solid organic
compounds by laser-induced plasma spectroscopy. Appl. Spectrosc. 55, 739–744 (2001)
14.28. H. Fink, U. Panne, R. Niessner, Analysis of recycled thermoplasts from consumer
electronics by laser-induced plasma spectroscopy. Anal. Chim. Acta 440, 17–25 (2001)
14.29. H. Fink, U. Panne, R. Niessner, Process analysis of recycled thermoplasts from consumer
electronics by laser-induced plasma spectroscopy. Anal. Chem. 74, 4334–4342 (2002)
14.30. NIST Atomic spectra data base, http://physics.nist.gov/cgi-bin/AtData/main asd
14.31. M. Kraushaar, R. Noll, H.-U. Schmitz, Slag analysis with laser-induced breakdown
spectrometry. Appl. Spectrosc. 57, 1282–1287 (2003)
14.32. R. Wisbrun, I. Schechter, R. Niessner, H. Schroder, K. Kompa, Detector for trace
elemental analysis of solid environmental samples by laser plasma spectroscopy. Anal.
Chem. 66, 2964–2975 (1994)
14.33. L. St-Onge, M. Sabsabi, P. Cielo, Analysis of solids using laser-induced plasma spec-
troscopy in double-pulse mode. Spectrochim. Acta B 53, 407–415 (1998)
14.34. F. Colao, V. Lazic, R. Fantoni, S. Pershin, A comparison of single and double pulse laser-
induced breakdown spectroscopy of aluminium samples. Spectrochim. Acta B 57, 1167–
1179 (2002)
14.35. National Standard of Germany, DIN 53 804–1, Statistische Auswertung, Teil 1: Kon-
tinuierliche Merkmale (Beuth, Berlin, April 2002)
14.36. A. Dobney, A. Mank, K. Grobecker, P. Conneely, C. de Koster, Laser ablation inductively

coupled plasma mass spectrometry as a tool for studying heterogeneity within polymers.
Anal. Chim. Acta 423, 9–19 (2000)
14.37. R. Kaiser, W. Gottschalk, Elementare Tests zur Beurteilung von Messdaten (Biblio-
graphisches Institut, Mannheim, 1984)
14.38. National Standard of Germany, DIN 32 645, Chemische Analytik – Nachweis-,
Erfassungs- und Bestimmungsgrenze (Beuth Verlag, Berlin, September 2006)
14.39. A. Cremers, D. Romero, An evaluation of factors affecting the analysis of metals using
laser-induced breakdown spectroscopy (LIBS), in Remote Sensing, ed. by R.T. Menzies,
Proc. SPIE 644, 7–12 (1986)
14.40. H. Bennet, G. Oliver, XRF Analysis of Ceramics, Minerals and Allied Materials (Wiley,
New York, 1992)
14.41. J. Vadillo, J. Laserna, Laser-induced breakdown spectroscopy of silicate, vanadate and
sulfide rocks. Talanta 43, 1149–1154 (1996)
14.42. H. Wiggenhauser, D. Schaurich, G. Wilsch, LIBS for non-destructive testing of element
distributions on surfaces. NDT&E Int. 31, 307–313 (1998)
14.43. A. Knight, N. Scherbarth, D. Cremers, M. Ferris, Characterisation of laser-induced
breakdown spectroscopy (LIBS) for application to space exploration. Appl. Spectrosc.
54, 331–340 (2000)
14.44. F. Hilbk-Kortenbruck, R. Noll, P. Wintjens, H. Falk, C. Becker, Analysis of heavy
metals in soils using laser-induced breakdown spectrometry combined with laser-induced
fluorescence. Spectrochim. Acta B 56, 933–945 (2001)
14.45. G. Doujak, R. Mertens, W. Ramb, J. Flock, J. Geyer, S. Lüngen, Slag analysis by laser-
induced breakdown spectroscopy. Stahl Eisen 121, 53–58 (2001)
14.46. H. Martens, T. Næs, Multivariate Calibration (Wiley, New York, 1998)
14.47. B. Bergmann, N. Bannenberg, Schlackenführung und Schlackenoptimierung in der
Sekundärmetallurgie. Stahl Eisen 111, 125–131 (1991)
14.48. H. Lachmund, Y. Xie, High purity steels: a challenge to improved steelmaking processes.
Ironmak. Steelmak. 30, 125–129 (2003)
14.49. V. Sturm, H.-U. Schmitz, T. Reuter, R. Fleige, R. Noll, Fast vacuum slag analysis in a
steel works by laser-induced breakdown spectroscopy. Spectrochim. Acta B 63, 1167–
1170 (2008)
14.50. C. López-Moreno, S. Palanco, J.J. Laserna, Quantitative analysis of samples at high tem-
perature with remote laser-induced breakdown spectrometry using a room-temperature
calibration plot. Spectrochim. Acta B 60, 1034–1039 (2005)
14.51. V. Sturm, Optical micro-lens array for laser plasma generation in spectrochemical
analysis. J. Anal. At. Spectrom. 22, 1495–1500 (2007)
14.52. K. Esbensen, Multivariate Data Analysis, 5th edn. (Camo Process, Oslo, 2002), p. 159
14.53. R. Wisbrun, I. Schechter, R. Niessner, H. Schröder, K.L. Kompa, Detector for trace
elemental analysis of solid environmental samples by laser plasma spectroscopy. Anal.
Chem. 66, 2964–2975 (1994)
14.54. D. Cremers, J. Barefield Il, A. Koskelo, Remote elemental analysis by laser- induced
breakdown spectroscopy using a fiber-optic cable. Appl. Spectrosc. 49, 857–860 (1995)
14.55. F. Hilbk-Kortenbruck, R. Noll, P. Wintjens, H. Falk, C. Becker, Spectrochemical analysis
of heavy metals in soils with laser-based techniques, Abstracts of the lecture groups
“Environmental Technology and Fundamentals of Laser-aided In-situ Soil Analysis”, Int.
Meeting on Chem. Eng., Environmental Protection and Biotechnology, ACHEMA 2000,
Conf. Proc. 2000, DECHEMA e.V., Frankfurt a. M., 449–451 (2000)
14.56. R. Multari, L. Foster, D. Cremers, M. Ferris, Effect of sampling geometry on elemental
emissions in laser-induced breakdown spectroscopy. Appl. Spectrosc. 50, 1483–1499
(1996)
14.57. G. Theriault, S. Bodensteiner, S. Lieberman, A real-time fiber-optic LIBS probe for the
in situ delineation of metals in soils. Field Anal. Chem. Technol. 2, 117–125 (1998)
References 383
14.58. B. Miles, J. Cortes, Subsurface heavy-metal detection with the use of a laser-induced
breakdown spectroscopy (LIBS) penetrometer system. Field Anal. Chem. Technol. 2,
75–87 (1998)
14.59. Mitteilungen der Länderarbeitsgemeinschaft Abfall (LAGA) Nr. 20: Anforderungen an
die stoffliche Verwertung von mineralischen Reststoffen/Abfällen – Technische Regeln
(Erich Schmidt Verlag, Berlin, 1995)
222–232 (1996)
14.61. A. Eppler, D. Cremers, D. Hickmott, M. Ferris, A.C. Koskelo, Matrix-effects in the
detection of Pb and Ba in soils using laser-induced breakdown spectroscopy. Appl.
Spectrosc. 50, 1175–1181 (1996)
14.62. A. Ciucci, V. Palleschi, S. Rastelli, R. Barbini, F. Colao, R. Fantoni, A. Palucci, S.
Ribezzo, H. van der Steen, Trace pollutants analysis in soil by a time-resolved laser-
induced breakdown spectroscopy technique. Appl. Phys. B 63, 185–190 (1996)
14.63. C. Becker, F. Hilbk-Kortenbruck, R. Noll, Preparation of soil samples containing heavy
metals for the calibration of laser-based analytical techniques, Abstracts of the lecture
groups “Environmental Technology and Fundamentals of Laser-aided In-situ Soil Analy-
sis”, Int. Meeting on Chem. Eng., Environmental Protection and Biotechnology, ACHEMA
2000, Conf. Proc. 2000, DECHEMA e.V., Frankfurt a.M., 394–396 (2000)
14.64. R. Springenschmidt, Arbeitskreis “Prüfverfahren Chlorideindringtiefe” des Deutschen
Ausschusses für Stahlbeton: Anleitung zur Bestimmung des Chloridgehaltes von Beton,
Beuth Verlag GmbH, Berlin; Research group “Test methods for chloride penetration
depth” of Deutscher Ausschuss für Stahlbeton: instruction for the determination of
chloride content in concrete (Beuth Verlag GmbH, Berlin, 1989)
14.65. American Society for Testing and Materials (ASTM). Standard Test Method for Electrical
Indication of Concrete’s Ability to Resist Chloride Ion Penetration, ASTM C 1202–97
(ASTM, West Conshohocken, 1997)
14.66. F. Weritz, D. Schaurich, A. Taffe, G. Wilsch, Effect of heterogeneity on the quantitative
determination of trace elements in concrete. Anal. Bioanal. Chem. 385, 248–255 (2006)
14.67. V. Burakov, V. Kiris, S. Raikov, Optimization of conditions for spectral determination of
chlorine content in cement-based materials. J. Appl. Spectrom. 74, 321–327 (2007)
14.68. F. Weritz, D. Schaurich, G. Wilsch, Detector comparison for sulfur and chlorine detection
with laser induced breakdown spectroscopy in the near infrared. Spectrochim. Acta B 62,
1504–1511 (2007)
14.69. J. Aguilera, C. Aragon, A comparison of the temperatures and electron densities of laser-
A 69, S475-S478 (1999)
14.70. G. Asimellis, A. Giannoudakos, M. Kompitsas, Near-IR bromine LIBS detection and
ambient gas effects on emission line asymmetric Stark broadening and shift. Spectrochim.
Acta B 61, 1270–1278 (2006)
14.71. M. Tran, Q Sun, B.W. Smith, J. Wineforder, Determination of F, Cl and Br in solid organic
compounds by laser-induced plasma spectroscopy. Appl. Spectrosc. 52, 739–744 (1999)
14.72. G. Asimellis, S. Hamilton, A. Giannoudakos, M. Kompitsas, Controlled inert gas
environment for enhanced chlorine and fluorine detection in the visible and near-infrared
by laser-induced breakdown spectroscopy. Spectrochim. Acta B 60, 1132–1139 (2005)
14.73. L. Radziemski, D. Cremers, K. Benelli, C. Khoo, R. Harris, LIBS-based Detection of As,
Br, C, Cl, P and S in the VUV spectral region in a Mars atmosphere. Lunar and Planetary
Science XXXVI, 1747 (2005)
14.74. C. Gehlen, E. Wiens, R. Noll, G. Wilsch, K. Reichling, Chlorine detection in cement
with laser-induced breakdown spectroscopy in the infrared and ultraviolet spectral range.
14.75. H. Lancelin, L. Dudragne, P. Adam, J. Amouroux, Time resolved laser induced break-
down spectroscopy for fluorine; chlorine and sulfur detection using an optical fibre probe.
High Temp. Mater. Proc. 4, 109–126 (2000)
14.76. N. Konjevic, Plasma broadening and shifting of non-hydrogenic spectral lines: present
status and applications. Phys. Rep. 316, 339–401 (1999)
14.77. M. Dimitrijevic, S. Sahal-Brechot, Stark broadening of Ca II spectral lines. J. Quant.
Spectrosc. Radiat. Transfer 49, 157–164 (1993)
interacting with iron samples at ambient gas pressures below 1 bar. Appl. Phys. B 86,
159–167 (2007)
analytical tool for the detection of metal ions in aqueous solutions. Fresen. J. Anal. Chem.
355, 16–20 (1996)
14.80. B. Welz, Speciation analysis: where is it going? An attempt at a forecast. Spectrochim.
Acta B 53, 169–175 (1998)
14.81. R. Lobinski, J. Szpunar, Biochemical speciation analysis by hyphenated techniques. Anal.
Chim. Acta 400, 321–332 (1999)
14.82. A. Timerbraev, Element speciation analysis by capillary electrophoresis. Talanta 52, 573–
606 (2000)
14.83. http://www.agilent.com
14.84. D. Beauchemin, Inductively coupled plasma mass spectrometry. Anal. Chem.74, 2873–
2894 (2002)
14.85. P. Kenedy, D. Hammer, B. Rockwell, Laser-induced breakdown in aqueous media. Prog.
Quant. Electr. 21, 155–248 (1997)
analytical tool for the detection of metal ions in aqueous solutions. Fresen. J. Anal. Chem.
355, 16–20 (1996)
14.87. P. Fichet, P. Mauchien, J. Wagner, C. Moulin, Quantitative elemental determination in
water and oil by laser-induced breakdown spectrometry. Anal. Chim. Acta 429, 269–278
(2001)
14.88. W. Ho, C. Ng, N. Cheung, Spectrochemical analysis of liquids using laser induced plasma
emission: effects of laser wavelength. Appl. Spectrosc. 51, 87–91 (1997)
14.89. J. Carranza, B. Fisher, G. Yoder, D. Hahn, On-line analysis of ambient air aerosols using
laser-induced breakdown spectroscopy. Spectrochim. Acta B 56, 851–864 (2001)
14.90. H. Archontaki, S. Crouch, Evaluation of an isolated droplet sample introduction system
for laser-induced breakdown spectroscopy. Appl. Spectrosc. 42, 741–746 (1988)
14.91. G. Arca, A. Ciucci, V. Palleschi, S. Rastelli, E. Tognoni, Trace element analysis in water
by laser-induced breakdown spectroscopy technique. Appl. Spectrosc. 51, 1102–1105
(1997)
14.92. A. Pichahchy, D. Cremers, M. Ferris, Elemental analysis of metals under water using
14.93. R. Wal, T. Ticich, J. West, P. Householder, Trace metal detection by laser-induced
breakdown spectroscopy. Appl. Spectrosc. 53, 1226–1236 (1999)
14.94. A. Kuwako, Y. Uchida, K. Maeda, Supersensitive detection of sodium in water with use
of dual-pulse laser-induced breakdown spectroscopy. Appl. Optics 42, 6052–6056 (2003)
14.95. K. Lo, N. Cheung, ArF laser-induced plasma spectroscopy for part-per-billion analysis of
metal ions in aqueous solution. Appl. Spectrosc. 56, 682–688 (2002)
14.96. X. Pu, W. Ma, N. Cheung, Sensitive elemental analysis of aqueous colloids by laser-
induced plasma spectroscopy. Appl. Phys. Lett. 83, 3416–3418 (2003)
14.97. C. Janzen, R. Fleige, R. Noll, H. Schwenke, W. Lahmann, J. Knoth, P. Beaven,
E. Jantzen, A. Oest, P. Koke, Analysis of small droplets with a new detector for liquid
chromatography based on laser-induced breakdown spectroscopy. Spectrochim. Acta B
60, 993–1001 (2005)
References 385
14.98. H. Ulmke, T. Wriedt, H. Lohner, K. Bauckhage, The piezoelectric droplet generator – a

versatile tool for dispensing applications and calibration of particle sizing instruments,
Precis. Eng. – Nanotechnology, Proc. of the 1st Int. Euspen Conf., vol. 2 (Shaker Verlag,
Aachen, 1999), pp. 290–293
14.99. C. Haisch, R. Niessner, O. Matveev, U. Panne, N. Omenetto, Element-specific determina-
tion of chlorine in gases by laser-induced-breakdown-spectroscopy LIBS. Fresen. J. Anal.
Chem. 356, 21–26 (1996)
14.100. M. Tran, B.W. Smith, D. Hahn, J. Winefordner, Detection of gaseous and particulate
fluorides by laser-induced breakdown spectroscopy. Appl. Spectrosc. 55, 1455–1461
(2001)
14.101. D. Plemmons, C. Parigger, J. Lewis, J. Hornkohl, Analysis of combined spectra of NH
and N2 . Appl. Optics 37, 2493–2498 (1998)
14.102. A. Schwebel, A. Ronn, Spectroscopy of laser-induced dielectric breakdown in gas
mixtures. Chem. Phys. Lett. 100, 178–182 (1983)
14.103. M. Casini, M. Harith, V. Palleschi, A. Salvetti, D. Singh, M. Vaselli, Time-resolved LIBS
experiment for quantitative determination of pollutant concentrations in air. Laser Part.
Beams 9, 633–639 (1991)
14.104. I. Glassman, Combustion (Academic, San Diego, 1996), pp. 20–29
14.105. A. Sandrowitz, J. Cooke, N. Glumac, Flame emission spectroscopy for equivalence ratio
monitoring. Appl. Spectrosc. 52, 658–662 (1998)
14.106. R. Schmieder, Combustion applications of laser-induced breakdown spectroscopy, in
Proceedings of the Electro-optics Laser Conference, Cahners, Chicago, 1981, pp. 17–27
14.107. X. Phuoc, F. White, Laser-induced spark for measurements of the fuel-to-air ratio of a
combustible mixture. Fuel 81, 1761–1765 (2002)
14.108. V. Sturm, R. Noll, Laser-induced breakdown spectroscopy of gas mixtures of air, CO2 ,
N2 , and C3 H8 for simultaneous C, H, O, and N measurement. Appl. Optics 42, 6221–
6225 (2003)
14.109. K. Fieweger, R. Blumenthal, G. Adomeit, Self-ignition of S.I. engine model fuels: shock
tube investigation at high pressure. Combust. Flame 109, 599–619 (1997)
14.110. K. Kuwahara, H. Ando, Diagnostics of in-cylinder flow, mixing and combustion in
gasoline engines. Meas. Sci. Technol. 11, R95–R111 (2000)
14.111. D. Lide (ed.), CRC Handbook of Chemistry and Physics, 73rd edn. (CRC Press, Boca
Raton, 1992), pp. 14–11
14.112. V. Sturm, A. Brysch, R. Noll, H. Brinkmann, R. Schwalbe, K. Mülheims, P. Luoto,
P. Mannila, K. Heinänen, D. Carrascal, L. Sancho, A. Opfermann, K. Mavrommatis, H.W.
Gudenau, A. Hatziapostolou, S. Couris, Online multi-element analysis of the top gas of
a blast furnace, in 7th Int. Workshop Progress in Analytical Chemistry in the Steel and
Metal Industries, ed. by J. Angeli (Glückauf GmbH, Essen, 2006), pp. 183–188
14.113. A. Brysch, V. Sturm, R. Noll, H. Denecke-Arnold, H. Brinkmann, K. Mülheims,
A. Opfermann, K. Mavrommatis, H.-W. Gudenau, K. Heinänen, Laser-based elemental
analysis of the top gas of a blast furnace, Int. Symp. on Photonics in Measurement,
11–12 June 2002, Aachen, Germany, VDI-Berichte Nr. 1694, 2002, 117–123, ISBN
3-18-091694-X
14.114. European commission, Directorate C: information note: airborne particles and their health
effects in Europe, ENV.C1/SAZr, March 2003
14.115. World Health Organisation, World Health Report 2002 (WHO, Geneva, 2002)
14.116. D. Adriano, Biogeochemistry of Trace Metals (Lewis, Boca Raton, 1992)
14.117. J. Fergusson, The Heavy Elements: Chemistry, Environmental Impact and Health Effects
(Pergamon, Oxford, 1991)
14.118. T. Kuhlen, C. Fricke-Begemann, N. Strauss, R. Noll, Analysis of size-classified fine
and ultrafine particulate matter on substrates with laser-induced breakdown spectroscopy.
14.119. U. Panne, R. Neuhauser, M. Theisen, H. Fink, R. Niessner, Analysis of heavy metal

aerosols on filters by laser-induced plasma spectroscopy. Spectrochim. Acta B 56, 839–
850 (2001)
14.120. Y. Yamamoto, R. Yoshiie, S. Uemiya, Simple and rapid analysis of heavy metals in sub-
micron particulates in flue gas, 6th Int. Symp. and Exhibition on Gas Cleaning at High
Temperature, Osaka, Japan, 2005
14.121. S. Arnold, D. Cremers, Rapid determination of metal particles on air sampling filters using
laser-induced breakdown spectroscopy. Am. Ind. Hyg. Assoc. J. 56, 1180–1186 (1995)
14.122. M. Marjamäki, J. Keskinen, D. Chen, D. Pui, Performance evaluation of the electrical
low-pressure impactor (ELPI). J. Aerosol Sci. 31, 249–261 (2000)
14.123. G. Cristoforetti, S. Legnaioli, V. Palleschi, A. Salvetti, E. Tognoni, Influence of ambient
gas pressure on laser-induced breakdown spectroscopy technique in the parallel double-
pulse configuration. Spectrochim. Acta B 59, 1907–1917 (2004)
14.124. L. Radziemski, T. Loree, D. Cremers, N. Hoffman, Time-resolved laser-induced break-
down spectrometry of aerosols. Anal. Chem. 55, 1246–1252 (1983)
14.125. R. Neuhauser, U. Panne, R. Niessner, G. Petrucci, P. Cavalli, N. Omenetto, On-line
and in-situ detection of lead aerosols by plasma-spectroscopy and laser-excited atomic
fluorescence spectroscopy. Anal. Chim. Acta 346, 37–48 (1997)
14.126. S. Yalcin, D. Crosley, G. Smith, G. Faris, Spectroscopic characterization of laser-produced
plasma for in situ toxic metal monitoring. Hazard. Waste Hazard. Mater. 13, 51–61 (1996)
14.127. M. Nunez, P. Cavalli, G. Petrucci, N. Omenetto, Analysis of sulfuric acid aerosols by
laser-induced breakdown spectroscopy and laser-induced fragmentation. Appl. Spectrosc.
54, 1805–1816 (2000)
14.128. R. Yoshiie, Y. Yamamoto, S. Uemiya, S. Kambara, H. Moritoni, Simple and rapid analysis
of heavy metals in sub-micron particulates in flue gas. Powder Technol. 180, 135–139
(2008)
14.129. D. Hahn, W. Flower, K. Hencken, Discrete particle detection and metal emission
monitoring using laser-induced breakdown spectroscopy. Appl. Spectrosc. 51, 1836–1844
(1997)
14.130. D. Hahn, M. Lunden, Detection and analysis of aerosol particles by laser-induced
breakdown spectroscopy. Aerosol Sci. Tech. 33, 30–48 (2000)
14.131. J. Carranza, D. Hahn, Sampling statistics and considerations for single-shot analysis using
14.132. J. Carranza, D. Hahn, Plasma volume considerations for analysis of gaseous and aerosol
samples using laser-induced breakdown spectroscopy. J. Anal. At. Spectrom. 17, 1534–
1539 (2002)
14.133. N. Strauss, C. Fricke-Begemann, R. Noll, Size-resolved analysis of fine and ultrafine
particulate matter by laser-induced breakdown spectroscopy. J. Anal. At. Spectrosc. 25,
867–874 (2010)
14.134. V. Hohreiter, D. Hahn, Calibration effects for laser-induced breakdown spectroscopy of
gaseous sample streams: analyte response of gas-phase species versus solid-phase species.
Anal. Chem. 77, 1118–1124 (2005)
14.135. G. Lithgow, S. Buckley, Influence of particle location within plasma and focal volume
on precision of single-particle laser-induced breakdown spectroscopy measurements.
14.136. V. Hohreiter, D. Hahn, Plasma-particle interactions in a laser-induced plasma: implica-
tions for laser-induced breakdown spectroscopy. Anal. Chem. 78, 1509–1514 (2006)
14.137. I. Gornushkin, J. Anzano, L. King, B. Smith, N. Omenetto, J. Winefordner, Curve
of growth methodology applied to laser-induced plasma emission spectroscopy. Spec-
trochim. Acta B 54, 491–503 (1999)
Chapter 15
Spatially Resolved Analysis
This chapter describes LIBS investigations for a spatially resolved microanalysis

of samples using the capability to focus laser radiation to spot sizes below 10 m.
Applications of the developed method will be presented in Sect. 18.3.
15.1 High-Speed LIBS for Microanalysis
The parameter range of laser pulse energies used for bulk analysis of solid, liquid, or
gaseous substances typically lies between 10 mJ and 1 J, the repetition frequency at
10–50 Hz [15.1–15.6], see Fig. 3.1. Several authors have studied spatially resolved
measurements using LIBS [15.7-15.12]. In this case, the gained spectral information
is linked to the position of irradiation of the laser pulse onto the sample surface.
A relative motion between the focused laser beam and the sample enables a scanning
across the surface. The result is for example, presented as an element line intensity
of the laser-induced plasma versus one or two spatial coordinates. The latter situa-
tion yields maps of element-specific intensities of a sample surface. Two kinds of
spatial resolution have to be distinguished: the lateral resolution and the longitudinal
or depth resolution, see Sect. 3.6. For scanning LIBS, the lateral resolution is
generally influenced by: (a) the beam diameter on the sample surface, (b) the laser
pulse energy, (c) the laser pulse duration, (d) the material properties of the sample,
and (e) the type, pressure and flow conditions of the ambient gas atmosphere.
A lateral spatial resolution of 6 m was reported for a sharp aluminum–copper
transition using a frequency quadrupled Nd:YAG laser at 266 nm and pulse energies
below 100 J [15.8]. The repetition rate of the laser was 2 Hz, and the spectral range
recorded simultaneously was 50 nm. Surface analysis of photovoltaic cells using a
pulsed N2 laser at 337.1 nm and 1 mJ pulse energy was performed with a lateral
spatial resolution of 30 m and a repetition rate of 10 Hz [15.9]. Single shot spectra
were detected in the spectral range from 542 to 657 nm. Mapping of platinum group
metals in automotive exhaust catalysts has been performed with a 20 Hz laser at
a wavelength of 532 nm [15.10]. The spectral range detected simultaneously was

388 15 Spatially Resolved Analysis
15 nm. With a line-shaped laser focus and an imaging spectrometer an inclusion in a

stainless steel sample was studied by scanning the laser line across a sample surface
of 1:6 mm 4:4 mm. The estimated spatial resolution achieved along the laser
line is 17 and 15 m in the direction perpendicular to the microline focus [15.11].
The acquisition rate was 2 Hz and the spectral range detected simultaneously was
approximately 15 nm. LIBS-based microanalysis of tool steel and glass was studied
using an echelle spectrometer allowing to observe a spectral range from 200 to
800 nm simultaneously [15.12]. However, in this case, the measuring frequency is
limited by the read-out time of the applied megapixel CCD-sensor array to well
below 1 Hz.
The repetition rates used so far for scanning LIBS lie in the range of 2–20 Hz,
with the consequence that the scanning of macroscopic samples with small step
sizes takes a long time. The use of a line-focus offers an opportunity to speed
up scanning; however, in this approach an imaging spectrometer has to be used,
limiting the spectral range detected simultaneously to typically less than 20 nm.
This is a significant restriction for the selection of emission lines of analytes
and reference elements most suitable for quantitative spectral analysis and for
the number of elements to be determined simultaneously. Figure 15.1 shows an
overview of the LIBS parameters measuring frequency and spectral resolution.
The measuring frequency is strongly linked to the type of detector used for the
acquisition of the spectral information. Common configurations of spectrometers
and detector types are as follows (cf. Sect. 4.2): (a) Czerny–Turner spectrometer
and intensified PDA or CCD, (b) echelle spectrometer and intensified CCD-camera,
and (c) Paschen–Runge spectrometer and PMT. The requirement concerning the
spectral resolution = depends on the expected density of emission lines in the
spectrum of the laser-induced plasma to clearly separate analyte emissions from
interfering lines. A quantitative analysis of iron-based samples generally requires a
spectral resolution of about 5 103 and more. Most of the existing LIBS setups are
presented by the data points 1–10, where the measuring frequency is mostly limited
by the detector read-out time or the laser repetition frequency to less than 60 Hz.
The data points 11–14 illustrate the potential to push the measuring frequency into
the range of 100 Hz to 1 kHz. The combination of a Paschen–Runge spectrometer
with PMT detectors achieves the highest ever so far achieved measuring frequencies
for LIBS, see data point 14 in Fig. 15.1.
In this chapter, R&D activities will be presented enabling to push the measuring
speed of scanning LIBS to the 1 kHz range [15.24]. The spectral range detected
simultaneously was extended to cover the vacuum ultraviolet, visible, and near-
infrared spectral region.
A diode-pumped solid-state laser (DPSSL) was used for the excitation of the
plasma, see Sect. 4.1, Figs. 4.2 and 4.3, and Table 4.3. Oscillator and amplifier are
pumped longitudinally using a fiber optic link to two diode pumping modules. The
oscillator is equipped with an electro-optic Q-switch. The beam is expanded using a
Galilean telescope and has a beam diameter at the exit of about 10 mm. The laser is
operated at a wavelength of 1,064 nm. The other laser parameters are: repetition rate
up to 1,000 Hz, pulse energy 2 mJ, pulse width (FWHM) 5–6 ns, beam propagation
ratio M 2 < 1:3 [cf. Sect. 3.4, relation (3.5)].
15.1 High-Speed LIBS for Microanalysis 389
spectral resolution λ / Δλ 10000 2 4 5
8000 14
3
6000
4000
6
9 11 12 13
1
2000
10
7 8
0
0.1 1 10 100 1000

νm [Hz]
Fig. 15.1 Measuring frequency m versus spectral resolution = for different detector, spec-
trometer, and laser combinations used for LIBS are plotted with filled bullets. The numbers close
to the data points in the graph refer to the following references: 1 D [15.13], 2 D [15.14], 3 D
[15.15], 4 D [15.16], 5 D [15.17], 6 D [15.18], 7 D [15.19], 8 D [15.20], 9 D [15.21],
10 D [15.22]. Data point 14 is the high-speed LIBS system described in this chapter. The open
squares for data points 11–13 represent commercially available detectors which can be used for
LIBS [15.23], the assumed resolution of 3,000 can be achieved with standard spectrometers and
monochromators
The laser beam is coupled to a measuring chamber shown schematically in

Fig. 15.2. The measuring chamber is a gas-tight vessel made of high-alloyed steel.
Inside the measuring chamber, an argon gas atmosphere is maintained at a pressure
of about 900 mbar. The use of the inert gas allows to transmit the ultraviolet (UV)
emission of the laser-induced plasma to the spectrometer, cf. Sect. 3.5. The sample
to be analyzed is brought into the chamber through a transfer port (V). The sample
is clamped in a sample cassette which is fixed to the x-, y-translation stages (A)
installed inside the chamber (C). These translation stages allow to shift the sample
with respect to the position of the laser focus in two perpendicular directions. The
positioning accuracy of the axes is 1 m, and the minimum step size amounts
to 5 m.
The laser beam is focused by an objective consisting of three lenses to correct
the spherical aberration. The focal length is 60 mm, and the aperture 48 mm. The
beam diameter on the focusing lens amounts to 26 mm. The calculated 1=e 2 -
focus diameter taking into account the beam quality of the DPSSL of M 2 D 1:3 is
4 m. The waist of the beam focus is positioned onto the surface of the sample
.s D 0 mm/.
The emission of the laser-induced plasma is collimated by a spherical mirror to
the entrance slit of a Paschen–Runge vacuum spectrometer having a Rowland circle
diameter of 1 m and covering a spectral range of 130–777 nm. Along the Rowland
circle, exit slits and photomultipliers are positioned at predefined spectral positions
Fig. 15.2 Schematic setup for the high-speed scanning microanalysis. DPSSL D beam of the
diode-pumped solid-state laser; BS D beam splitter; W1 D window for the laser beam; L D
focusing lens; S D sample; A, x-, y D translation stages; P D laser-induced plasma; M D spherical
mirror to collect the plasma emission; OA D optical axis of the spectrometer; W2 D window to
observe the sample surface; MO D microscope zoom objective (10); CM 1, 2 D CCD camera to
observe the craters; G D gas flushing of the measurement chamber; C D measuring chamber; and
V D transfer port
to detect the radiation of up to 48 emission lines, cf. Sect. 4.2, Fig. 4.7. The signals
of the photomultiplier tubes detecting the transient radiation are transmitted to a
multichannel integrator electronics (MCI, cf. Sect. 4.5, Fig. 4.16).
The sample surface can be observed by an optical system consisting of a
microscope zoom objective and a CCD-camera. The optical axis of this microscope
is oriented parallel to the optical axis of the analyzing laser beam. For an inspection
of the produced craters, the sample can be moved by the translation stages toward
the field of view of the microscope and the CCD-camera, see MO and CM2 in
Fig. 15.2.
DPSSL, translation stages, and MCI electronics are synchronized to allow for
single-shot data evaluation and a defined allocation of the spectral signals to the
respective measuring position, i.e., the x- and y-position. At 1 kHz measuring
frequency and a scanned area of 11 cm2 with a step size of 20 m, a data volume of
about 48 MB is collected which is then processed to generate the element mappings
(cf. Sect. 15.2).
Sample translation stages, laser source, pumping unit, Pockels cell, and integrator
electronics have to be synchronized to achieve a correct measuring point position of
better than 1 m and a high signal stability. Figure 15.3 shows the timing diagram
for high-speed scanning LIBS. The counter signal of the linear stages is counted
by a synchronization electronics to determine the true position of the stages, see
graph a in Fig. 15.3 (the nonequidistant pulse sequence of the counter signal is a
consequence of the interpolation procedure applied in connection with the measures
Fig. 15.3 Timing diagram for high-speed scanning LIBS. a–f: see captions on the left side. tit ,
internal transit time; tLT
ET
, time set in the event table (ET) for the laser trigger (LT); tbt , built-up time
(bt) for the laser pulse after activation of the Q-switch including transit times; tdelay
ET
, delay time set
in the event table for the start of the integration window; tint , time set in the event table for the
ET
width of the integration window; and tdelay , true delay time between the start of the laser pulse and
the start of the integration window
integrated in both translation axes). After a programmable number of counts, the

pumping diodes of the DPSSL are triggered and pump the laser crystal for 140 s
(graphs b and c in Fig. 15.3). After the falling edge of the pumping current and
a fixed internal transit time tit , the integrator electronics is triggered and an event
table – containing the temporal sequence for various trigger signals and the start
and widths of the integration windows for the different signal channels – is started
(graphs c and d ). The Pockels cell is triggered by a signal initiated by the event
table (graph d ). This procedure ensures that the temporal jitter between the Pockels
cell switching and the start of the integration windows is kept to less than 1 ns.
After the laser pulse built-up time tbt D 400 ns (including transit times), the
laser pulse is emitted and the plasma is generated on the sample surface (graph e).
The vertical dashed arrow in Fig. 15.3 running from the laser pulse peak in graphs
e–a illustrates that the laser ablation takes place at a defined position of the axes.
The event table is processed step by step and at the end the measured values are
digitized and transferred to a computer. During the whole time, the synchronization
800
600
y [µm]
400
200
0
0 200 400 600 800
x [µm]
Fig. 15.4 Optical microscope image taken with a white light confocal microscope of a polished
steel surface with line scans of laser generated craters. The left row is a line scan with a single laser
pulse per position. The middle and right row is a tenfold overlay of lines scans. Step size: 100 m,
laser pulse energy 2 mJ
electronics tracks the counter signal of the linear stages. When the next measuring
position is reached, the cycle will begin again. With this timing principle, a spatial
repeatability of the position of the laser-produced craters of better than 1 m is
achieved within a scanning area of up to 45 mm 110 mm.
This capability was proofed by comparing the optical microscopic image of a line
scan, where only a single laser pulse is irradiated per position with a neighboring
line scan where ten single pulse scans are overlaid, see Fig. 15.4 [15.25]. As can
be seen from the microscope image, the ablation craters lie exactly on top of each
other. By successive two dimensional scans, a three dimensional analysis can be
performed. For each further laser pulse irradiation at the same position, the crater
depth increases successively thus enabling to yield chemical information underneath
the sample surface.
For a more detailed determination of the geometry of the craters produced by the
pulses of the DPSSL, a ground steel surface was prepared using a grinding paper
with a grain size of 80. Figure 15.5 shows images of craters generated by 1, 2, 4, and
8 laser pulses taken with an optical microscope. A central tiny crater is surrounded
by a ring-shaped structure originating from residues of the melt pushed out of the
interaction region by the pressure of the laser-induced plasma and the accompanying
shock wave (cf. Sect. 7.1). The central crater diameter for a single shot can be
estimated to be less than 15 m. The crater depth is measured using an optical
microscope determining the difference in the focusing positions of the microscope
objective while sequentially focusing on the sample surface in the neighborhood of
the crater and the bottom of the crater. However, this yields only a rough estimate
of the crater depth, since the resolution of the microscopic images is limited by
diffraction of the illuminating light at the crater structure itself. The crater depth
amounts to about 2–8 m for a single laser pulse depending on the pulse energy and
1 laser pulse 2 laser pulses
4 laser pulses 8 laser pulses
Fig. 15.5 Images of the impact area of 1, 2, 4, and 8 laser pulses of the DPSSL with a pulse energy
of 2 mJ on a ground steel surface taken with an optical microscope
the number of pulses irradiated at one position. The diameter of the region enclosing
the debris around the crater is about 50 m.
The diameter of the central crater will mainly determine the lateral analytical
resolution. The debris area will affect consecutive measurements in the sense of a
crosstalk. However, an analysis of the element maps gained with this set-up and a
step size for scanning of 20 m has not revealed any significant crosstalk, which
may be attributed to the fact that the ratio of the debris mass – contributing to the
plasma signal of the subsequent laser pulse – to the mass originally present in the
central crater volume is very small.
Stereo SEM images and white light confocal microscopy images of a single
pulse crater generated with a 2 mJ laser pulse on a steel target show crater depths of
about 2:3 m [15.25]. With a scanning measurement comprising 3:75 106 laser
pulses having each 2 mJ at a step size of 20 m an average mass loss of 0.77 ng
per pulse was determined. Taking the mass density of steel this corresponds to an
effective crater volume of 98 m3. Assuming that the central crater geometry can
be described by a right circular cone, this volume implies a diameter of the base of
the cone of 13 m, which is consistent with the observations of Fig. 15.5.
0.0
-0.1
u pmt [a.u.]
-0.2
-0.3
-0.4
Fe 371.99 nm average over 10,000 pulses
-0.5
0 2 4 6 8 10
time [μs]
Fig. 15.6 Photomultiplier signal of the element channel for Fe at 371.99 nm as a function of time
Figure 15.6 shows the photomultiplier signal of the iron line at 371.99 nm as a
function of time. The signal was transmitted from the socket of the multiplier by
a coaxial cable RG 174 to an oscilloscope terminated with 50 . At the beginning a
narrow emission peak is observed corresponding to the continuum emission in the
early phase of the plasma generation (cf. Chap. 2, Fig. 2.4). This emission starts in
coincidence with the laser pulse irradiation. The temporal behavior of this peak is
similar for all element channels. The width of this peak amounts to about 25 ns.
The subsequent second emission maximum occurring at about 600 ns after the
laser pulse irradiation belongs to the element-specific radiation component. For the
element maps shown in the following, the element signals are integrated within
defined time gates using the following parameters, if not stated otherwise: delay
time with respect to the laser pulse 2 s, gate width 8 s.
The principal limitation of the plasma generation frequency for high-speed LIBS
with single-pulse evaluation is among others defined by the temporal behavior of the
line emission of the laser-induced plasma. The element-specific emission extends
over a period of about 20 s at maximum. Hence, the principal limit is estimated to
be at about 50 kHz, which is nearly two orders of magnitude beyond the achieved
maximum frequency for LIBS of 1 kHz described in this chapter. However, further
limits are imposed by plasma residuals and gas exchange processes [cf. Sect. 3.1,
Fig. 3.1, relation (3.1)]. The gas exchange is limited by turbulences, which may
cause an enlarged laser beam focus due to local gradients of the refractive index of
the neighboring ambient atmosphere, thus reducing the laser irradiance in the focus
and decreasing the pulse-to-pulse stability of the intensity distribution irradiating the
target. Therefore, a gas filtering system was integrated in the measuring chamber
to remove ablated and recondensated particles from the argon atmosphere. The
gas is exhausted via a nozzle positioned close to the interaction region. To study
the efficacy of this approach, a steel sample was scanned with 100 and 1,000 Hz
1.2
intensity ratio I i1000 Hz / I i100 Hz Ar lines
1.1
1.0
0.9
0.8
100 200 300 400 500 600 700 800

wavelength λi [nm]
Fig. 15.7 Ratio of element intensities acquired at a measuring frequency of 1,000 Hz to those
gained at 100 Hz at the same steel sample as a function of the wavelength i of 37 atomic and ionic
lines. The error bars are the standard deviation of the 250,000 measuring data for each plotted data
point. The four data points between 430 and 480 nm are argon lines representing the ambient gas
atmosphere in the measuring chamber
laser repetition rate. The overall number of measuring points in each case amounts
to 250,000. The ratio of the average intensities of the 1,000 and the 100 Hz
measurement was calculated for each element channel i and plotted in a diagram
versus the wavelength i of the respective emission line i , see Fig. 15.7. From
all detected element lines, only the four argon lines – representing the ambient
atmosphere – showed a significant change in the average signal intensity ratio.
The upper energy level of the transitions of the atomic and ionic lines of the
species originating from the steel sample lies in the range between 3.12 eV (Ca II at
396.85 nm) and 10.74 eV (O I at 777.19 nm). Figure 15.7 shows that for these lines
no significant difference in the intensity is observed within the relative standard
deviation of the intensity ratio. This is an indication that the conditions of the
laser-induced plasma with respect to the excitation of the sample species do not
significantly change while increasing the repetition rate from 100 to 1,000 Hz. As
the excitation of the sample species does not change, it is assumed that the exhaust
gas is equally effective at both repetition rates.
In contrast to this, the argon lines show an enhancement by about 15–20%. The
physical reason is not yet clear. All argon lines plotted in Fig. 15.7 are argon ion
lines with upper energy levels between 19.26 and 21.14 eV. A slight temperature
rise in the argon atmosphere adjacent to the sample surface caused by the increased
energy dissipation while irradiating a ten times higher average laser power (from
100 Hz 2 mJ D 200 mW to 1; 000 Hz 2 mJ D 2 W) may cause a layer of reduced
argon particle density close to the surface. This density reduction favors a faster
shock wave expansion after the irradiation of the laser pulse which in turn excites
800
Si 212.4 nm
C 193.1 nm
intensity [a.u.]
600
400
200
5000 5020 5040 5060 5080 5100
measuring steps
Fig. 15.8 Silicon and carbon signal for a line scan across the surface of a steel sample as a function
of the measuring step. The step size amounts to 20 m, so the total length of the line scan shown
is 2 mm. The peaks at the measuring step 5,046 indicate a SiC inclusion
the ambient argon stronger leading to an increase in the argon line intensities (cf.
Sect. 8.2, Figs. 8.24 and 8.26).
15.2 Element Mappings
With the SML system (cf. Sect. 18.3) so-called element mappings can be generated
via one- or two-dimensional scanning. Element mappings are presentations of
intensities of element channels as a function of one- or two-dimensional spatial
coordinates. Since all element channels are measured simultaneously, a set of
element mappings is generated for all channels. An example for a scanning along
a single line – one-dimensional scanning – is shown in Fig. 15.8. The intensities of
the elements, silicon and carbon, are plotted as a function of the measuring steps for
a scan across a steel sample. In this figure, a step size of 20 m was chosen. Hence,
the length of the scan shown in Fig. 15.8 amounts to 2 mm. Both signal value are
scattering around a base level. However, at measuring step 5,046 both signals attain
a significantly higher level which returns to the base level again for the next step.
Within a localized length of 20 m, a clear spatial correlation of enhanced silicon
and carbon signal is observed indicating that the laser beam has irradiated a silicon
carbide (SiC) inclusion.
An example of a two-dimensional scan is shown in Fig. 15.9. The scanning area
is 10 10 mm2 of a low-alloyed steel sample. At a step size of the translation
axis of 20 m, the shown element maps for sulfur (top) and manganese (middle)
comprise 250,000 measuring points each generated by a single laser pulse of 2 mJ
15.2 Element Mappings 397
Fig. 15.9 Sulfur and manganese map of a steel sample in a scan field of 10 10 mm2 (upper and
middle plane). The lower plane shows a magnification of a detail of the manganese map with the
dimensions 1 1 mm2 . The wavelengths given adjacent to the element denomination state the used
emission line of the spectrometer
and a measuring frequency of 1 kHz. The maps show intensities of the sulfur
line at 180.73 nm and of the manganese line at 263.82 nm. Only those intensities
are plotted, which are greater than the average intensity value plus five times the
standard deviation of the intensity distribution of all measuring points. By this those
positions show up where a significantly higher element signal persists. Such signals
are caused by, e.g., inclusions in the steel matrix, cf. Sect. 18.3.
A close look on the sulfur and manganese map shows that the dot patterns show
a clear spatial correlation. This is exemplified for a pair of dots by the arrow shown
in Fig. 15.9. This means that at positions of enhanced manganese concentration also
an enhanced sulfur concentration exists. At these measuring positions, manganese
sulfide (MnS) inclusions are identified. The element map shown in the lower plane
of Fig. 15.9 shows a magnified part of the manganese map for a subarea of 11 mm2 .
Single as well as groups of inclusions can be seen with lateral dimensions in the
range of 50–100 m. Within a cluster of manganese inclusions, different manganese
concentrations exist, which is discernible by the different gray scales. The detection
sensitivity for analytes of inclusions reaches down to the sub-femtogram range.
Depending on the type of inclusion, smallest inclusions with volumes in the range
Si 212.41 nm Mn 293.31 nm Fe 371.99 nm
Fig. 15.10 Element maps of the cross section of the weld seam of a T-joint of two aluminum
sheets
of 105 to 4 102 m3 are detected corresponding to inclusion sizes between 30

and 430 nm [15.25].
Figure 15.10 shows three element maps of a cross section of the weld seam of a
T-joint of two aluminum sheets (horizontally and vertically oriented in Fig. 15.10).
The scanned area is 10 mm 10 mm, step size 20 m, and measurement frequency
1 kHz. For welding, a welding wire containing silicon was used. The silicon element
map shows how the additional silicon was mixed in the weld bath. The manganese
and iron maps reveal that the two sheets consisted of two different aluminum grades.
The application of element mappings for segregation and inclusion analysis, as
well as quantification methods, will be described in Sect. 18.3.
References
15.1. D. Cremer, Overview of applications of laser-induced breakdown spectroscopy (LIBS),

in Proceedings of the 6th International Congress on Applications of Lasers and Electro-
optics, ICALEO 87, San Diego, 1987, ed. by G. Kychakoff, pp. 45–50
15.2. J. Winefordner, I. Gornushkin, D. Pappas, O. Matveev, B. Smith, Novel uses of lasers in
atomic spectroscopy – plenary lecture. J. Anal. At. Spectrom. 15, 1161–1189 (2000)
15.3. M. Milan, P. Lucena, L. Cabalan, J. Laserna, Depth profiling of phosphorus in photonic-
grade silicon using laser-induced breakdown spectrosmetry. Appl. Spectrsc. 52, 444–448
(1998)
15.4. L. Berman, P. Wolf, Laser-induced breakdown spectroscopy of liquids: aqueous solutions
of nickel and chlorinated hydrocarbons. Appl. Spectrosc. 52, 438–443 (1998)
15.5. H. Zhang, F. Yueh, J. Singh, Laser-induced breakdown spectrometry as a multimetal
continuous-emission monitor. Appl. Optics 38, 1459–1466 (1999)
15.6. R. Adrain, J. Watson, Laser microspectral analysis: a review of principles and applications.
J. Phys. D Appl. Phys. 17, 1915–1940 (1984)
15.7. L. Moenke-Blankenburg, Laser Microanalysis (Wiley, New York, 1989), p. 93
15.8. C. Geertsen, J. Lacour, P. Mauchien, L. Pierrard, Evaluation of laser ablation optical
emission spectrometry for microanalysis in aluminium samples. Spectrochim. Acta B 51,
1403–1416 (1996)
References 399
15.9. J. Vadillo, S. Palanco, M. Romero, J. Laserna, Applications of laser-induced breakdown

spectrometry (LIBS) in surface analysis. Fresenius J. Anal. Chem. 355, 909–912 (1996)
15.10. P. Lucena, J. Vadillo, J. Laserna, Mapping of platinum group metals in automotive exhaust
three-way catalysts using laser-induced breakdown spectrometry. Anal. Chem. 71, 4385–
4391 (1999)
15.11. M. Mateo, L. Cabalin, J. Baena, J. Laserna, Surface interaction and chemical imaging in
plasma spectrometry induced with a line-focused laser beam. Spectrochim. Acta B 57,
601–608 (2002)
15.12. K. Loebe, A. Uhl, H. Lucht, Microanalysis of tool steel and glass with laser-induced
breakdown spectroscopy. Appl. Optics 42, 6166–6173 (2003)
15.13. D. Cremers, R. Wiens, M. Ferris, R. Brennetot, S. Maurice, Capabilities of LIBS for analy-
sis of geological samples at stand-off distances in a Mars atmosphere, in Technical Digest,
Conference: Laser-Induced Plasma Spectroscopy and Applications, Orlando (September
25–28, 2002), pp. 5–7
15.14. U. Panne, R. Neuhauser, M. Theisen, H. Fink, R. Niessner, Analysis of heavy metal
aerosols on filters by laser-induced plasma spectroscopy. Spectrochim. Acta B 56, 839–
850 (2001)
15.15. B. Charfi, M.A. Harith, Panoramic laser-induced breakdown spectrometry of water.
15.16. O. Samek, D. Beddows, H. Telle, J. Kaiser, M. Liska, J. Caceres, A. Gonzales Urena,
Quantitative laser-induced breakdown spectroscopy analysis of calcified tissue samples.
15.17. Thermo-ARL 4460 with SparkDat. http://www.thermoarl.com
15.18. J. Carranza, B. Fisher, G. Yoder, D. Hahn, On-line analysis of ambient air aerosols using
15.19. D. Body, B. Chadwick, Optimization of the spectral data processing in a LIBS simultaneous
elemental analysis system. Spectrochim. Acta B 56, 725–736 (2001)
15.20. A. Whitehouse, J. Young, I. Botheroyd, S. Lawson, C. Evans, J. Wright, Remote material
analysis of nuclear power station steam generator tubes by laser-induced breakdown
spectroscopy. Spectrochim. Acta B 56, 821–830 (2001)
15.21. K. Lo, N. Cheung, ArF laser-induced plasma spectroscopy for part-per-billion analysis of
metal ions in aqueous solutions. Appl. Spectrosc. 56, 682–688 (2002)
15.22. S. Palanco, M. Klassen, J. Skupin, K. Hansen, E. Schubert, G. Sepold, J. Laserna, Spectro-
scopic diagnostics on CW-laser welding plasmas of aluminium alloys. Spectrochim. Acta
B 56, 651–659 (2001)
15.23. HR 2000 from http://www.oceanoptics.com, DU 401 from http://www.andor-tech.com,
Spec-10:100B from http://www.prinst.com
15.25. G. Müller, Lokale Elementverteilungsanalyse mittels rasternder laser-induzierter optischer
Emissionsspektrometrie zur Charakterisierung von Einschlüssen und Seigerungen in
Stählen, Dissertation, Shaker Verlag, Aachen (2004)
Chapter 16
Depth Profiling
This chapter describes laser-induced breakdown spectroscopy (LIBS) investigations

for the determination of the thickness of coatings or the depth profile of an element
in a surface layer.
16.1 Measurement of the Thickness of Coatings
In the production process of galvanized and galvannealed sheet steel, inline

monitoring of the coating thickness and the chemical depth profile of the zinc
coating is of great interest. To assure reliable corrosion protection, the coating
thickness and composition, e.g., the Zn, Fe, and Al content in hot-dip galvanized
coatings, have to be kept constant over complete coils with a sheet length of up to
3 km. A small amount of Al ( 0:15–0:25 m.-%) leads to improved adhesion of the
coating and prevents iron diffusion from the substrate due to the formation of an
Fe2 Al5 inhibition layer at the interface between substrate and zinc coating [16.1].
A state-of-the-art method for coating thickness analysis in galvanizing plants is
X-ray fluorescence (XRF) [16.2]. The X-ray line emission of iron from the substrate
is measured, which is partially absorbed by the zinc coating. Depending on the
coating thickness, the iron line is more or less absorbed. Online monitoring of the
Al content is currently not realized as XRF is less sensitive to light elements such as
Al. Depth profiles of the coatings are measured offline only, e.g., by glow discharge
optical emission spectroscopy (GD-OES) [16.3].
Depth profiling of coated samples by LIBS was studied by different research
groups mostly with the aim to achieve high depth resolution [16.4–16.12]. There-
fore, the experiments were performed on stationary samples by irradiating a single
spot or a microline on the sample with a series of laser pulses (see also the
description in Sect. 3.6, Fig. 3.19) [16.10].
On a moving sheet steel in a production line, the irradiation of a series of laser
pulses on the same surface position is not applicable because of the following
reasons. First, the diameter of LIBS craters caused by a series of single pulses on

402 16 Depth Profiling
one position with a beam profile optimized for high depth resolution is rather large
(1 mm) and would reduce the corrosion protection. The technical effort to irradiate
one position on a moving sheet steel with several laser pulses would be tremendous.
The laser focus would have to move with the sheet steel with velocities in the range
of 1 m/s while keeping a spatial precision in the micrometer range. Therefore, a
new method was developed to characterize the coating with a stationary laser beam
and a series of single laser bursts each irradiated on a different laterally displaced
sheet steel position (for the term “burst” see Sect. 3.2, Fig. 3.9). Depth information
is obtained by tuning the ablation depth by variation of the burst energy. As the
coating thickness is in the range of about 10 m, the ablation depth per burst has
to be of the same order, see Fig. 3.19 for n D 1 [16.13]. This increased ablation
depth is achieved using bursts consisting of collinear double pulses (cf. Sect. 3.2
and Chap. 6). The burst parameters (i.e., interpulse separation, energy distribution
within the burst) of the analyzing laser were optimized to minimize the impact of
the LIBS craters on the sheet steel quality. These parameters led to craters with
diameters smaller than 100 m, which did not influence the corrosion behavior and
were invisible for the naked eye on the typical surface texture of hot-dip galvanized
sheet steel. The thickness resolution was determined by experiments performed on
a set of reference samples of well-characterized electrolytic-coated sheet steel with
different coating thicknesses in the range of 3:2–11:2 m. The performance to gain
depth information was evaluated on industrial hot-dip galvanized sheet steel.
Figure 16.1 shows schematically the experimental setup [16.13]. A flashlamp
pumped Q-switched Nd:YAG-laser operating at 1,064 nm and modified for the
emission of up to six pulses at 10 Hz repetition rate (>2 s interpulse separation) is
used. The widths of the individual laser pulses in the burst are 15–20 ns depending
on the number of pulses in the burst. In the experiments described here, the laser is
operated in double pulse mode (interpulse separation of t1 D 4 s, pulse energy
ratio E1 W E2 D 1 W 1, burst energy Eb D E1 C E2 D 0:2–3 mJ, see Fig. 3.9) to
achieve an ablation depth with a single burst in the range of the coating thickness
(ca. 10 m).
The laser light is guided via a dichroic mirror (high reflectivity for 1,064 nm,
transparent for the plasma light) and focused by an achromatic lens L1 (f D
60 mm) onto the surface of the sheet steel, where a small amount of the sample
is evaporated and the plasma is generated. The ambient atmosphere was air. The
plasma light is collected and collimated with the same lens, passes the dichroic
mirror, and is reimaged by a quartz lens (f D 100 mm) onto the core of the
optical fiber of the spectrometer. An echelle spectrometer (cf. Sect. 4.2, Table 4.4)
offering a broad spectral detection range (200–780 nm) with high spectral resolution
(5–20 pm) detects the plasma emission. The time delays used for the signal
integration were 0.5 and 2:0 s, respectively, and the integration window was set
to 10 s. For energy control, the laser setup has been modified by inserting an
additional Pockels cell as electro-optical attenuator between the laser resonator
and the amplifier stages (not shown in Fig. 16.1). The Pockels cell attenuates the
transmitted laser pulse dependent on the applied high voltage (0 to 8 kV) by up
to a factor of 10. The typical burst energy was in the range of 0.2–3 mJ. The chosen
16.1 Measurement of the Thickness of Coatings 403
Fig. 16.1 Schematic view of the experimental setup. P D laser-induced plasma plume, C D
coating, ST D steel substrate, L1 D focusing lens, L2 D imaging lens, M D mirror, OF D fiber
optics, EC D energy control, LH D laser head, LC D laser control, SP D echelle spectrometer,
GN D gas nozzle
laser parameters allow to penetrate the zinc coating with a single laser burst, which
is a necessary precondition for the approach studied here. The crater diameters
were <100 m, which is small enough not to reduce the corrosion protection of
the coating.
The new method is developed for coating thickness measurement and depth
profiling of moving sheet steel in a production line, i.e., under conditions where only
single laser bursts can be applied to one distinct sheet steel position. The approach to
obtain depth information is to control the crater depth by changing the burst energy
in small increments and to measure the LIBS signal for every single laser burst, see
Fig. 16.2, top. The signal gained for one distinct burst energy is a weighted mean of
the chemical composition over the whole ablation volume and is evaluated further
as described in the following.
The measuring conditions were simulated by moving the sheet steel samples
(dimensions: 10 cm 15 cm) with motorized xy-translation stages. For different
burst energies, rows with equally spaced LIBS craters are generated as the sheet
steel moves under the static optical setup. In each row in the x-direction, the burst
energy was kept constant and in the y-direction the burst energy was attenuated.
For each burst energy Ebi , a number of k (k D 20 30) replicates is applied to
the sheet steel samples. Figure 16.3 shows a schematic diagram of a part of the
resulting crater pattern. The distance s between adjacent craters represented by
circles was set to about 0.5 mm. The number of craters per row was in the range
of 20 30 depending on the experiment. Each crater was generated with a single
laser burst, and the complete spectrum was measured for each laser burst. From row
Fig. 16.2 Top: Procedure to gain depth information of the coating (thickness, composition) on
moving sheet steel. The measuring depth is adjusted by variation of the laser burst energy Ebi ,
i D 1; 2; : : :. For Eb3 , the ablated depth reaches the steel substrate. Abbreviations as in Fig. 16.1.
Bottom: Sequence of laser bursts. For each burst energy Ebi a number of k.k D 2030/ replicates
is applied to the sheet steel samples
Fig. 16.3 Schematic view of

the resulting crater pattern on
the sheet steel. The burst
energy is kept constant in
x-direction and attenuated in
y-direction. The separation s
between adjacent craters (CR)
represented by circles
amounts to about 0.5 mm
to row the burst energy is varied in small steps (step size 0:05 mJ, depending on
the experiment) within the range of 0.2–2.6 mJ using the laser energy control. The
whole setup is automated to synchronize the sheet steel motion, the laser operation,
and the spectrometer.
A set of reference samples of electrolytic galvanized sheet steel with well-
defined coating thicknesses provided by ThyssenKrupp Steel AG (TKS, Dortmund,
Germany) was used for the experiments on coating thickness determination. The set
consists of sheet steel plates with different coating thicknesses ranging from 3.1 to
11:2 m. Industrial hot-dip galvanized samples with an average coating thickness
in the range of 10 m were used for the experiments to gain depth information of
the Al content of the coating. These samples were also provided by TKS.
The method for coating thickness determination was studied for the electrolytic
galvanized sheet steel samples taking the same set of laser parameters and spec-
trometer settings to assure comparability of the results. The laser burst energy was
Fig. 16.4 Contour plot of

a LIBS crater measured
by white light interferometry.
The crater was generated with
a 2-mJ laser burst on an
electrolytic galvanized
sheet steel
Table 16.1 Crater depths of Coating Average crater Standard

LIBS carters generated with a thickness depth deviation of
burst energy of 2 mJ on (m) hc (Eb D 2 mJ) hc (Eb D 2 mJ)
electrolytic galvanized sheet (m) (m)
steel with different
thicknesses of the zinc 11.2 12.6 2.0
coating 9.2 11.7 2.0
6 11.7 0.7
4.7 9.8 1.6
3.2 9.7 1.9
The average crater depth is the mean value of five
crater depths determined by white light interferometry
tuned in 18 steps from 0.22 to 2 mJ, and for each energy step 20 measurements were
carried out. Figure 16.4 shows exemplarily a contour plot of a LIBS crater generated
with 2 mJ on the sheet steel with 9:2 m coating thickness. The topography
was determined by white light interferometry. Obviously, the surface roughness
complicates the determination of the crater depth and leads to large uncertainties.
In Table 16.1, the average crater depths of LIBS craters generated with a burst
energy of 2 mJ on the sheet steel with different coating thicknesses are compared.
For 5 of the 20 craters of each coating thickness, the crater depth was determined and
the mean values were calculated. Due to the rough surfaces, the relative errors are
up to 20%. There seems to be the tendency that the crater depth is lower for thinner
coatings, which may be due to a lower ablation rate for the hard steel substrate
compared to the soft zinc coating.
In the LIBS experiment, the measurement window starts at a delay time of 2:0 s
after the laser burst and has a width of 10 s. For each laser burst, a complete
spectrum was acquired, allowing to study various analyte lines for data evaluation.
As an elemental line of the coating, the zinc line at 472.22 nm (3 S1 ! 3 P1 / and
for the substrate the iron line at 438.35 nm (4 F5 ! 4 F4 / were selected, as these
lines have strong line intensities and are not interfered by neighboring lines. The
a
5 TKS electrolytic galvanized
3.2 µm
4 4.1 µm
I Fe438 / I Zn472 [1]
4.7 µm
5.2 µm
3 6.0 µm
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
laser burst energy E b [mJ]
b
0.6
TKS electrolytic galvanized
0.5
8.2 µm
9.2 µm
I Fe438 / I Zn472 [1]
0.4
10.2 µm
0.3 11.2 µm
0.2
0.1
0.0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
laser burst energy E b [mJ]
Fig. 16.5 Intensity ratio IFe438 =IZn472 as a function of the burst energy Eb for different coating
thicknesses. Samples: electrolytic galvanized sheet steel; (a) 3:2–6:0 m; (b) 8:2–11:2 m. tdelay D
0:5 s, tint D 10 s
line intensities IFe438 and IZn472 were computed by integrating the line profile
and subtracting the background. For data evaluation, the ratio IFe438 =IZn472 was
calculated for each laser burst and averaged over the 20 measurements of the same
burst energy. In Fig. 16.5a, b, this ratio is shown with the standard deviations as a
function of the burst energy.
The IFe438 =IZn472 value of the thinner coatings (thickness d 6:0 m, Fig. 16.5a)
is even at the lowest burst energy greater than zero, i.e., the coating is always
TKS electrolytic galvanized
0.66 mJ
3
0.99 mJ
I Fe438 / I Zn472 [1]
1.33 mJ
1.57 mJ
2
1.78 mJ
2.00 mJ
0
3 4 5 6 7 8 9 10 11
coating thickness d [µm]
Fig. 16.6 Intensity ratio IFe438 =IZn472 as a function of the coating thickness d for different burst
energies Eb . To prevent overlapping of the graphs, the graphs for 0.66 and 0.99 mJ are shifted by
1.0 and the graphs for 1.33 and 1.57 mJ are shifted by 0.5 in the positive direction of the ordinate.
Samples: electrolytic galvanized sheet steel (TKS); coating thickness: 3:2–11:2 m
penetrated. The decline of the IFe438 =IZn472 value with increasing burst energy
seems to be in conflict with the expected higher relative number of Fe atoms but
can be explained by the different temperature behavior of the population density
of the excited states corresponding to the IFe438 and IZn472;22 lines, according to
the Boltzmann distribution. At low burst energies, the amount of ablated material
and hence the plasma size is smaller and cools down more rapidly. This means
that at a given time delay tdelay , the plasma temperature is lower for lower burst
energies. As the excitation energy of the IFe438 line (Ej D 4:31 eV) is much lower
than that for the IZn472:22 line (Ej D 6:65 eV), the relative number of Fe atoms
excited in the upper state of the Fe438 transition is higher in cooler plasmas than
in hotter plasmas expected for higher burst energies. At small coating thicknesses,
this effect seems to have a stronger impact than the increase due to the higher
number of ablated Fe atoms. For the thicker coatings (d 8:2 m; Fig. 16.5b),
the intensity ratio starts at zero and rises for higher burst energies. For both groups
it is obvious that the intensity ratio at a fixed burst energy is larger for thinner than
for thicker coatings. Hence, the intensity ratio at a fixed burst energy is a measure
for the coating thickness, and the crater depths for different thick coatings vary only
slightly, see Table 16.1.
To determine a proper burst energy, the intensity ratio is plotted as a function
of the coating thickness for different burst energies, see Fig. 16.6. Coatings that are
thinner than 6 m can be distinguished for all burst energies, and coatings thicker
than 8:2 m can only be distinguished for burst energies >1 mJ as at a lower burst
energy the coating is not penetrated. The standard deviation becomes smaller with
2.0
linear fit
1.6
EDP = 2 mJ
I Fe438 / I Zn472 [1]
1.2
0.8
0.4
0.0
2 3 4 5 6 7 8 9 10 11 12 13
coating thickness d [µm]
Fig. 16.7 Intensity ratio IFe438 =IZn472 as a function of the coating thickness d for a fixed
burst energy of 2 mJ and linear fit. Samples: electrolytic galvanized sheet steel (TKS); coating
thicknesses: 3:2–11:2 m
increasing burst energy. Regarding the laser burst energy range studied, a burst
energy of 2 mJ should be used to determine thickness variations.
For an estimation of the achieved depth-resolving power, the measured data for a
burst energy of Eb D 2 mJ are fitted, and an approximately linear relation is found
between the intensity ratio IFe438 =IZn472 and the coating thickness d , see Fig. 16.7,
which can be described as follows:
IFe438
.d / D A C Bd; (16.1)
IZn472
where B is the slope and A the axis intercept. Calculating the inverse function and
using standard error calculations with the known errors A, B, .IFe438 =IZn472 /,
the error of d can be estimated to about 400 nm.
The depth resolution can be further improved based on a systematic line
selection [16.14] using multivariate data analysis (MVDA) [16.15–16.18], correc-
tion of plasma temperature [16.19, 16.20], and selection of ambient atmosphere
[16.21–16.23].
To find optimum experimental conditions and the most sensitive IFe =IZn ratio, 21
iron lines and 6 zinc lines were selected in the spectral range from 300 to 500 nm,
and all possible IFe =IZn ratios were evaluated in terms of thickness resolution using
an MVDA software and applying a temperature correction.
The experimental parameters are those described above for Fig. 16.1. In the
experimental setup a gas nozzle was added, mounted at a distance to the sample
of 12 mm directing the argon (Ar) and nitrogen gas flows under an angle of 45ı
onto the sample surface. The estimated gas flows are approximately 170 l/min for
Ar and 215 l/min for N2 . The experiments in air were performed at normal pressure
without a gas flow.
The time delays used for the signal integration were tdelay D 1:0 s in air
and N2 and tdelay D 1:0 and 2:0 s in Ar, and the integration window was in all
measurements set to tint D 1 s. The times tdelay and tint were not optimized. The
burst energy was varied in the range of 0.5–2.6 mJ.
In the experiments, the different sheet steels were mounted on an xy-translation
stage and were moved under the static optical setup. For each detector setting and
ambient atmosphere, the burst energy is varied from 0.5 to 2.6 mJ in 22 equidistant
steps, and for each energy step 40 measurements are performed. This was repeated
for each of the different coating thicknesses, so for each parameter set 40 repetition
measurements are available for each of the eight coating thickness giving in total a
set of 3 (air, N2 ; Ar/ 22 40 8 D 21; 120 spectra.
In the first step, the integrated line intensities of 21 iron lines and 6 zinc lines
were calculated by fitting Voigt profiles to each of the lines of each echelle spectrum
and subsequent numerical integration. The lines with their atomic data are listed in
Table 16.2.
Then the average values of the 40 repetitions for all possible IFe =IZn ratios
(6 21 D 126) for the 22 burst energies and the eight coating thicknesses are
calculated. For coating thickness determination, the IFe =IZn ratio at a fixed burst
energy is plotted as a function of the coating thickness and a fit curve is determined.
The root mean square error (RMSE) value of the fits is defined as follows:
v
u N
uP i 2
u d d i
t i D1 Zn;m Zn;c:t:v:
RMSE D 2 ; (16.2)
N
i
where dZn;m is the thickness determined with the measured IFe =IZn ratio and the
i
calibration function, dZn;c:t:v: is the conventional true value of thickness determined
by GD-OES, and N is the number of different coating thicknesses (eight in this
case). The RMSE value is a measure for the accuracy of the thickness determination
related to a specific IFe =IZn ratio for a given parameter set, i.e., burst energy, delay
time, and ambient atmosphere.
The parameter combinations studied are: 22 burst energies, 3 ambient atmo-
spheres with 2 delay times for Ar, and 1 delay time for N2 and air, and 126 IFe =IZn
ratios with and without temperature correction. From this follows a total number of
22,176 calibration curves, which has to be handled. This large number of calibration
curves was split in eight groups, for each ambient atmosphere and the two delay
times for Ar, with and without temperature correction. Each group comprises 2,772
different IFe =IZn ratios (=126 line ratios 22 burst energies) and was evaluated by a
commercial software for MVDA (Unscrambler 9.2 from CAMO Technologies), see
Table 16.3.
Unscrambler calculates a linear model for the prediction of the coating thickness
including all IFe =IZn ratios using the partial least squares (PLS) method and returns
Table 16.2 List of studied iron and zinc lines with atomic parameters in the spectral range from
300 to 500 nm
Element (nm) log.gi fij / Ej (eV) kt (1030 m3 /
Group A: iron lines for temperature determination
Fe I 385.26 1:24 5.39 0.08
Fe I 392.29 1:65 3.21 0.61
Fe I 410.75 0:73 5.85 0.12
Fe I 410.98 0:91 5.86 0.08
Fe I 411.85 0.28 6.58 0.46
Fe I 428.24 0:81 5.07 0.27
Fe I 436.98 0:73 5.88 0.1
Fe I 441.51 0:62 4.42 0.99
Fe I 449.46 1:14 4.96 0.14
Fe I 452.86 0:82 4.91 0.3
Group B: additional iron lines for coating thickness measurement
Fe I 355.85 1:12 4.47 0.48
Fe I 382.78 0.06 4.80 3.81
Fe I 384.33 0:14 6.27 5.3
Fe I 404.58 0.28 4.55 7.74
Fe I 407.17 0:02 4.65 3.31
Fe I 420.20 0:71 4.43 0.86
Fe I 430.79 0:07 4.43 3.54
Fe I 432.58 0:01 4.47 3.82
Fe I 438.35 0.20 4.31 7.52
Fe I 440.48 0:14 4.37 3.13
Fe I 495.76 0.13 5.31 1.32
Group C: zinc lines
Zn I 328.23 0:38 7.78 1.61
Zn I 330.26 0:06 7.78 3.3
Zn I 334.50 0.25 7.78 6.36
Zn I 468.01 0:82 6.65 1.17
Zn I 472.22 0:34 6.65 3.45
Zn I 481.05 0:14 6.65 5.32
The lines in group A with low self-absorption (kt < 1 1030 m3 at Te D 8,500 K) were used
to determine the plasma temperatures. In the multivariate data analysis, all combinations of iron
lines from group A and group B with the zinc lines from group C were used. The quantity kt is a
measure for the optical thickness calculated for 8,500 K [cf. Sect. 9.4, relation (9.33)]. log.gi fij / D
logarithm of statistical weight of lower level and oscillator strength (cf. relation (9.8)), Ej D
energy of upper level
the regression coefficient for each of the IFe =IZn ratios for the 22 different burst
energies. In the next step all the 2,772 regression coefficient are sorted in ascending
order, and an algorithm is applied to find the IFe =IZn ratio with the highest regression
coefficients within five neighboring energy steps. Hence, IFe =IZn ratios with higher
regression coefficients at only one burst energy are neglected. In the final step, the
model is recalculated with the IFe =IZn ratio with the five neighboring energies and
the RMSEs are calculated.
Table 16.3 Comparison of the accuracies obtained with the standard IFe438 =IZn472 ratio and the
best IFe =IZn ratio determined by MVDA
Ambient Delay time Temperature Intensity ratio RMSE Eb range
atmosphere tdelay (s) correction IFe =IZn (m) (mJ)
IFe438 =IZn472
Air 1 No 438.35/472.22 0.51 0.7–1.1
Air 1 Yes 438.35/472.22 0.50 0.6–1.0
Ar 1 No 438.35/472.22 0.41 0.5–0.9
Ar 1 Yes 438.35/472.22 0.41 0.5–0.9
Ar 2 No 438.35/472.22 0.30 0.6–1.0
Ar 2 Yes 438.35/472.22 0.36 0.6–1.0
N2 1 No 438.35/472.22 0.61 1.0–1.4
N2 1 Yes 438.35/472.22 0.80 1.2–1.6
IFe =IZn ratios with the highest accuracy
Air 1 No 440.48/468.01 0.37 0.5–0.9
Air 1 Yes 410.75/468.01 0.37 1.5–1.9
Ar 1 No 428.24/468.01 0.19 2.2–2.6
Ar 1 Yes 411.85/472.22 0.32 0.5–0.9
Ar 2 No 410:98=328:23a 0:14a 1:5–1:9
Ar 2 Yes 411.85/328.23 0.20 1.2–1.6
N2 1 No 495.76/334.50 0.27 2.2–2.6
N2 1 Yes 382.78/481.05 0.68 1.3–1.7
a
The highest accuracy is obtained in an Ar atmosphere at tdelay D 2 s without temperature
correction using the ratio IFe410:98 =IZn328:23
An issue to be considered arises because the upper levels of an iron line and a
zinc line are different. Then the corresponding IFe =IZn ratio is a function of the ratio
of the number of iron atoms NFe and zinc atoms NZn in the ablated crater volume
and the temperature, since the line intensity of species i depends on the temperature
assuming negligible optical thickness (cf. Sect. 9.2, relations (9.3) and (9.8)):

gm fmn En
Ii / N i exp ; (16.3)
3 u.T / kB Te
where Ni is the number of atoms of species i , gm the degeneracy of the lower level,
fmn the oscillator strength, the transition wavelength, u.T / the partition function,
En the energy of the upper energy level, kB the Boltzmann constant, and Te the
electron temperature.
To take care of this effect and also to compensate temperature fluctuations due
to fluctuations of the laser power or different absorbtivity of the sample surface, a
temperature correction of the IFe =IZn ratios was applied as follows:

En; Fe
exp
IFe .Te / uFe .Te / k T NFe
B e / : (16.4)
IZn .Te / uZn .Te / En; Zn NZn
exp
kB Te
For each spectrum Te was determined by a Boltzmann plot using iron lines with low
self-absorption, i.e., kt < 1 1030 m3 (at Te D 8;500 K), see Table 16.2, group
A. Te is determined separately for each spectrum yielding values in the range of
6,000–9,000 K for the various parameter settings.
A set of reference samples of electrolytic galvanized sheet steel with well-defined
coating thicknesses provided by ThyssenKrupp Steel AG (TKS) was used for the
experiments. The set consists of sheet steel plates with different coating thicknesses
ranging from 3.1 to 11:2 m. In the following experiments on samples with coating
thicknesses from 4.1 to 11:2 m were carried out.
Figure 16.8 shows the line intensity ratio IFe438 =IZn472 as a function of the burst
energy for the coating thicknesses 4.1, 8.2, and 11:2 m.
The Zn line at 472.22 nm (3 S1 ! 3 P1 , Ej D 6:65 eV) and the Fe line at
438.35 nm (4 F5 ! 4 F4 , Ej D 4:31 eV) are both strong lines, and the difference in
the upper energy levels E is 2.34 eV. Figure 16.8a shows the ratio IFe438 =IZn472 as
a function of the burst energy without a temperature correction (for the interpretation
of the shown curve progression see discussion given above in connection with
Fig. 16.5). Figure 16.8b shows the same data applying a temperature correction
as described above [cf. relation (16.4)]. Now the IFe438 =IZn472 values of all three
coating thicknesses increase with the burst energy as expected. But it can be seen
in both figures that at any burst energy the IFe438 =IZn472 ratio is higher for thinner
coatings. Hence, the IFe438 =IZn472 ratio at a fixed burst energy can be used as a
measure for the coating thickness, cf. Fig. 16.7. Table 16.3 lists the IFe =IZn ratios
with the lowest RMSE values determined at a specific parameter set by MVDA
together with the RMSE values of the standard IFe438 /IZn472 ratio used, e.g., in
Fig. 16.7.
In each case, the RMSE value is given for the burst energy range leading to the
lowest RMSE values. Table 16.3 shows the effects of the ambient atmosphere, the
delay time, and the temperature correction on the RMSE values of the standard
IFe438 =IZn472 ratio. For a delay time of 1 s, and without temperature correction, the
thickness resolution is best in Ar (0:41 m) and worst in N2 (0:61 m). Physically,
reasonable temperature correction does not enhance thickness resolution for an
N2 atmosphere – it makes it even worse. In Ar atmosphere, a delay time of 2 s
enhances the resolution from 0.41 to 0:3 m.
The smallest RMSE values for a specific parameter set are found for the IFe =IZn
ratios listed in the second part of Table 16.3. For each of the parameter sets, another
IFe =IZn ratio was actually optimal, but most of the lines are lines with low self-
absorption (kt < 1:5 1030 m3 , Te D 8;500 K), cf. Table 16.2. In Ar, these values
are more than a factor of 2 smaller than the RMSE values of the IFe438 =IZn472 ratios.
Again, temperature correction does not improve the resolution. The best thickness
resolution of 140 nm was achieved in Ar atmosphere at a delay time of 2 s with
the IFe410 /IZn328 ratio. Figure 16.9 visualizes the increase in thickness resolution in
comparison to the IFe438 =IZn472 ratio. The plot shows the difference between the
thicknesses dZn;m determined with the calibration model and the measured IFe =IZn
ratio and the conventional true coating thickness dZn; c:t:v: as a function of dZn; c:t:v:
a
2.0
coating thickness
4.1 µm
without temperature correction
1.5 8.2 µm
I Fe438 / I Zn472 [1]
11.2 µm
1.0
0.5
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
laser burst energy Eb [mJ]
b
2.5
with temperature correction
2.0
I Fe438 / I Zn472 [1]
1.5
1.0
0.5
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Fig. 16.8 IFe438 =IZn472 ratio as a function of the burst energy for different coating thicknesses:
(a) without temperature correction, (b) with temperature correction. In both cases, the ratio is
higher for thinner coatings and can be used as a measure for the coating thickness
The thickness resolution is further enhanced from 510 nm down to 140 nm, i.e., by
more than a factor of 3.
Applying the temperature correction did not appear to lead to any improvement
in the thickness resolution. An explanation for this is that the plasma temperature
also depends on the coating thickness. For a particular burst energy, the plasma
temperature was found to be higher for thinner coatings (Te 500–1;000ı K
higher for dZn D 4:1 m than for dZn D 11:2 m). Hence, Te contains also depth
1.0
standard conditions RMSE = 510 nm

optimized conditions 140 nm
d Zn, m - d Zn, c.t.v. [µm]
0.5
0.0
-0.5
-1.0
2 4 6 8 10 12
coating thickness d Zn, c.t.v. [µm]
Fig. 16.9 Comparison of the resolving power for the IFe438 =IZn472 ratio in air with tdelay D 1 s,
tint D 1 s (standard conditions) and for the ratio IFe410:9 =IZn328 in Ar with tdelay D 2 s, tint D
1 s (optimized conditions). The optimized conditions were found by multivariate data analysis.
The plot shows the deviations of the thicknesses determined with the calibration functions and
the measured IFe =IZn ratios as a function of the conventional true coating thickness determined
by GD-OES and wet-chemistry. The RMSE value at optimum conditions is 140 nm compared to
510 nm at standard conditions
information which is eliminated by the temperature correction. Since a calibration

method was used, the IFe /IZn ratio does not need to reflect the true ratio of ablated
Fe and Zn atoms.
A similar experiment was performed using a hot-dip galvanized sheet steel
sample to study which depth information in terms of the aluminum concentration
is accessible with the studied approach. The laser burst energy was varied between
0.2 and 2.6 mJ in 54 steps, and for each energy step 30 single acquisitions were
taken. The measurement window starts at 0:5 s after the laser burst and has a
width of 10 s. As elemental lines, the zinc line at 472.22 nm (3 S1 ! 3 P1 / and the
aluminum line at 396.15 nm (2 S1=2 ! 2 P3=2 / were selected. Figure 16.10a shows
the line intensities IAl396 and IZn472 as a function of the burst energy. Both element
lines can be seen from the beginning as there is an Al enrichment mainly in the
form of Al2 O3 at the surface typical for hot-dip galvanized coatings [16.24]. The
error bars indicate the standard deviation of 30 LIBS signals. In the referenced
signal IAl396 =IZn472 (Fig. 16.10b), a finger print of the Al depth profile can be seen.
The decline in the ratio at low burst energies may be attributed to the penetration
of an Al2 O3 surface layer with a typical thickness of less than 10 nm [16.24].
A high Al concentration at the surface was also found in the LIBS depth profiles
of galvannealed (annealed hot-dip galvanized) sheet steel by St-Onge et al. [16.5].
The IAl396 =IZn472 ratio attains a minimum right before the ablation depth reaches the
depth of the inhibition layer. Then the ratio increases as more and more Al from the
inhibition layer and the steel substrate (Al is alloyed in the steel substrate, 2 m.-%)
a
2000
I Zn472
I Al396
1600
I Zn472, I Al396 [a.u.]
error bars:
1200 SD of 30 bursts
800
400
0
0.5 1.0 1.5 2.0 2.5
b
0.7
IAl396 / IZn472
0.6
I Al396 / I Zn472 [1]
0.5
0.4
0.3
0.2
0.5 1.0 1.5 2.0 2.5
Fig. 16.10 Raw intensities of zinc and aluminum (a) and referenced line intensity IAl396 =IZn472
(b) as a function of the laser burst energy Eb . Sample: hot-dip galvanized sheet steel (TKS); coating
thickness: 10 m. tdelay D 0:5 s, tint D 10 s
contributes to the signal with increasing ablation depths. After a maximum at about
1.5 mJ the ratio decreases slightly, which may be attributed to influences of the crater
geometry on the plasma dynamics and the plasma state. More details concerning the
characterization of elemental depth profiles will be described in Sect. 16.2.
The results demonstrate the potential of LIBS for the inline analysis of galvanized
sheet steel. The estimated thickness resolution of less than 200 nm is comparable to
the resolution limit achievable with inline XRF gauges. In addition, LIBS offers the
opportunity to get simultaneously information about the Al depth profile and the Zn
coating thickness.
For an inline application, the measurement time is an important point. The
measurement time for a single thickness value depends on the repetition rate of
the laser, the speed of the detector electronics, and the number of averages. The
readout time of the ICCD detector (2 s) used in the echelle spectrometer limits
the measurement time to about 40 s (for an average of 20 single measurements).
Using a diode-pumped Nd:YAG laser with a repetition rate in the kilohertz range
(cf. Sect. 4.1, Table 4.3, Fig. 4.3) and a spectrometer of the Paschen–Runge type
(cf. Sect. 4.2, Fig. 4.7) equipped with photomultiplier (PMT) detectors, a measuring
frequency of 1 kHz (for single readouts) can be obtained and the measurement time
is reduced to less than 100 ms (cf. Sect. 15.1).
16.2 Characterization of Depth Profiles
Coating depth profiles of galvanized and galvannealed coatings are measured only
offline, e.g., by GD-OES [16.25]. For an enhanced quality control, also a depth
analysis of the chemical composition of the coating is of great interest. In particular,
light elements such as Al and Mg, which are hard to measure with XRF, are of major
interest.
In the following, a new approach for an inline monitoring of the depth profile of
Al in hot-dip galvanized coatings based on LIBS is described [16.26]. LIBS has been
widely used for depth profiling of coatings on static samples [16.5–16.12, 16.27,
16.28], mostly with the aim to achieve high depth resolution. Typically, a series of
laser pulses with a large spot diameter is applied on the same sample position. From
the spectra of the successive laser pulses, information about the depth profile of the
elements was obtained. On a moving sheet steel in a production line this technique
is not applicable. In Sect. 16.1, a method is described to characterize the coating
with a stationary laser beam and a series of single laser bursts each irradiated on a
different laterally displaced sheet steel position to determine the coating thickness.
Depth information is obtained by tuning the ablation depth by variation of the burst
energy.
In the following, the method is applied to characterize the aluminum depth profile
of different hot-dip galvanized coatings, and a method of inline monitoring the
depth profile is presented. The method is further extended using triple pulses and
measuring the LIBS signals after each laser pulse to monitor three depths for each
triple-pulse burst. The motion of the sheet steel in a galvanizing plant is simulated
using sheet steel disks mounted on a rotary table. LIBS measurements on sheet
steel samples moving with up to 1 m/s were performed and compared to results
obtained on slowly moving sheet steel.
Figure 16.11 shows the experimental setup [16.26]. The Q-switch pulses of a
flashlamp pumped Nd:YAG-laser operating at 1,064 nm and modified for the emis-
sion of up to six pulses at 10 Hz repetition rate (interpulse separation t 2 s,
16.2 Characterization of Depth Profiles 417
Fig. 16.11 Optical setup with autofocus system
cf. Sects. 13.1.1 and 13.1.3) are guided via a dichroic mirror (high reflectivity for
1,064 nm, transparent for the plasma light) and focused by an achromatic lens
(f D 60 mm) onto the surface of the sheet steel, where a small amount of the sample
is evaporated and the microplasma is generated. The laser is operated in double
pulse mode (interpulse separation of t1 D 4 s, energy ratio E1 W E2 D 1 W 1,
burst energy Eb D E1 C E2 D 0:1–2:1 mJ, see Fig. 16.12a) or triple pulse mode
(interpulse separations of t1 D 4 s, t2 D 4 s, energy ratio E1 W E2 W
E3 D 1 W 1 W 0:7, burst energy Eb D E1 C E2 C E3 D 0:165–1:65 mJ, see
Fig. 16.12b) to achieve an ablation depth with a single burst exceeding the range
of the coating thickness (ca. 10 m) to be measured. The attenuation of the burst
energy is achieved by an external Pockels cell. Attenuation of up to a factor of 10
is possible. With the focus diameter of approximately 23 m and the widths of the
individual laser pulses in the burst of 15–20 ns (depending on the number of pulses
in the burst) power densities in the range of 1–20 GW=cm2 are reached.
The plasma light is collected and collimated with the same lens, passes the
dichroic mirror, is reflected by a second dichroic mirror, and is reimaged by a
quartz lens (f D 100 mm) onto the core of an optical fiber which guides the light
to a Paschen–Runge spectrometer (diameter of Rowland circle 750 mm, resolution
30 pm) equipped with 16 PMTs placed on the Rowland circle at the positions
of specific elemental lines. The elemental lines used are the Al line at 396.15 nm
(2 S1=2 ! 2 P3=2 /, the Zn line at 472.22 nm (3 S1 ! 3 P1 /, and the Fe line at
438.35 nm (4 F5 ! 4 F4 /. The photocurrent of the PMTs is integrated and digitized
by a multichannel integrator electronics (cf. Sect. 4.5), which allows for sampling
rates of up to 1 kHz and programmable time delays tdelay and width of the integration
window tint with a resolution of 50 ns for each pair of channels. In addition, it is
possible to set the integration windows after each pulse of one laser burst. The time
delays and integration windows used are different for the various elemental lines
and for the pulses within the laser burst. The specific settings are given below, but
Fig. 16.12 Temporal structure of the laser bursts and the settings of the integration windows used
in this study: (a) Double pulses with interpulse separation t1 D 4 s and burst energy distributed
equally, signal integration after pulse 2; (b) Triple pulses with interpulse separation t1 D 4 s,
t2 D 4 s, and E1 W E2 W E3 D 1 W 1 W 0:7, signal integration after pulse 1, 2, and 3
generally the delay times tdelay were in the range 0:3 0:5 s and the widths of
the integration windows were in the range tint D 0:5 3 s. Short delay times
are recommended to maximize the signals observed, as only small plasmas are
generated. According to investigations on microplasmas performed by Detalle et al.
with comparable focusing conditions, also LTE can be assumed at such delay times
[16.29]. The optical setup contains also an autofocus system for the compensation of
sheet steel vibrations of the rotating sheet steel disks. It is designed to compensate
vibrations of the measuring object with amplitudes of up to 1.5 mm which is in
the range of the vibrations occurring typically at S-rolls in galvanizing plants. The
autofocus system is based on a laser triangulation sensor and a telescope with a
motorized dispersing lens, see Fig. 16.11. The laser beam of the triangulation sensor
passes the two dichroitic mirrors and is aligned orthogonal and central through the
focusing lens. The distance of the sensor to the sample surface is measured and
transferred to a control PC. The second component is a telescope with a motorized
dispersing lens, whereby the beam divergence can be adjusted to shift the focus
position of the analyzing laser beam. A control software calculates the shift of
the dispersing lens to keep the focus position of the analyzing laser beam constant
relative to the sample surface.
The new method is developed for coating thickness measurement and depth
profiling of moving sheet steel in a production line, i.e., under conditions where only
single laser bursts can be applied to one distinct sheet steel position. The approach to
obtain depth information is to control the crater depth by changing the burst energy
in small increments and to measure the LIBS signal for every single laser burst.
Figure 16.2 (bottom) shows schematically the applied sequence of laser bursts. The
number of craters per energy step is 80–200 and the energy is tuned in more than
50 steps. The signal gained for one distinct burst energy is a weighted mean of the
chemical composition over the whole ablation volume. A set of measurements is
performed with the same burst energy to determine an average value of the line
intensities for the corresponding ablation depth. For the next set, the burst energy is
tuned to the next value. The averaging area is the length of the sheet steel traversing
during one measurement cycle and is determined by the speed of the sheet steel,
the laser repetition rate, and the number of measuring bursts. By referencing the Al
line intensity and the Fe line intensity to the Zn line intensity of the main element
zinc, the influence of fluctuations of the measuring process on the spectroscopic
information is reduced.
The measuring conditions were simulated by moving the sheet steel samples
under the focusing objective. Two different translation stages were used. For
quasi-static experiments and small sheet steel pieces (dimensions: 10 cm 15 cm),
a motorized xy-translation stage was used as described in Sect. 16.1. For the
experiments on moving sheet steel disks (diameter 490 mm), a rotating table was
used. The table rotates with a frequency of up to 1.66 Hz leading to velocities of
more than 2 m/s in the outer part of the disk. In order to move to different measuring
radii, the rotating table is mounted on a linear translation stage.
The resulting crater patterns are rectangular arrays (xy-translation stage) or
spirals (rotating table). The whole setup is automated to synchronize sheet steel
motion, laser operation, and PMT readout.
Different industrial hot-dip galvanized samples provided by ThyssenKrupp Steel
AG (TKS) and by OCAS N.V. (OCAS, Zelzate, Belgium) were used as samples
[16.30]. The samples from OCAS (OCAS II) had a coating thickness of about
10 m. Rectangular pieces (10 cm 15 cm) and sheet steel disks were available.
A GD-OES analysis shows that 50% of the maximum Al concentration is attained
at a depth of about 6:2 m. The TKS samples had coating thicknesses of 11 m
(TKS I), 10 m (TKS IV), and 7 m (TKS III). The TKS samples were cut in
rectangular pieces with the dimensions 10 cm 15 cm (TKS I) and 5 cm 5 cm
(TKS III, TKS IV). The GD-OES depth profile of TKS I shows that also Al is
present in the steel substrate, whereas the samples TKS III, TKS IV, and OCAS II
0.10
ZnAl bulk IAl396 II Zn472

0.08 t int(Al) = 3 µs, t int(Zn) = 1 µs
error bars: standard dev. of 40 bursts
I Al396 / I Zn472 [1]
0.06
0.04
0.02
0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
laser burst energy [mJ]
Fig. 16.13 IAl396 =IZn472 versus burst energy measured on a ZnAl binary bulk sample (cAl D
0:14 m.-%), tint D 3 s for the Al396-channel and tint D 1 s for the Zn472-channel, delay time
tdelay D 0:5 s. The IAl396 =IZn472 ratio is nearly constant in the investigated burst energy range up
to 1 mJ
had no Al in the substrate. For the TKS I sample, the Al concentration reaches 50%
of its maximum value at a depth of 10 m.
For comparative measurements, a certified binary reference ZnAl-bulk sample
(Breitländer, MBH 41XZ5) with an Al concentration of cAl D 0:14 m.-% compara-
ble to hot-dip galvanized coatings was used.
In the measurement procedure, the delay time tdelay and the width of the
measurement window tint are kept constant while changing the burst energy over
more than one order of magnitude. As the plasma parameters – like temperature or
size – are changing with the burst energy, it is important to know how this influences
the measured IAl396 =IZn472 ratio. Also fractionation can have an influence, as found
for zinc in brass for power densities lower than 0:3 GW=cm2 [16.31]. Therefore, the
procedure was initially applied to the ZnAl binary reference sample using double
pulse bursts with burst energies of 0.1–2.1 mJ (burst energy equally distributed on
double pulses with interpulse separation t1 D 4 s), see Fig. 16.12a. The delay
time tdelay was set to 0:5 s and the integration windows tint were 3 s for the
Al channel and 1 s for the Zn channel. The resulting referenced line intensity
IAl396 =IZn472 shown in Fig. 16.13 is nearly independent of the burst energy for
energies up to 1 mJ, i.e., the ratio of the number of Al and Zn atoms in the crater
volume is constant, and so it is not necessary to correct the measurements made on
the sheet steel samples. The slight increase at Eb > 1 mJ is due to saturation of the
Zn472 channel, whereas the signal of the Al396 channel continues to increase.
In the experiments on the hot-dip galvanized sheets, it was checked what depth
information for the aluminum concentration is made available with the studied
approach. The procedure described above was applied to the hot-dip galvanized
sheet steel samples OCAS II and TKS I (coating thicknesses: 10 and 11 m,
respectively). For both samples, the laser burst energy was varied between 0.1 and
2.1 mJ as for the ZnAl-bulk sample. For each energy step, 200 single acquisitions
were performed. The delay time tdelay was also set to 0:5 s, the width of the
integration window was 3:0 s for the Al line at 396.15 nm and 0:5 s for the zinc
line at 472:22 nm, which is shorter as for the ZnAl-bulk sample to avoid a saturation
of this channel. For the different sheet steels, the intensity ratio IAl396 =IZn472 as
a function of the burst energy is shown in Fig. 16.14. The data points represent
the mean values measured at each energy step, and the black lines are interpolated
curves calculated using a polynomial fit. In this representation, the samples OCAS
II and TKS I can be distinguished clearly. Both curves start at a ratio of about 0.05,
but the OCAS II curve increases much more quickly and reaches a slightly higher
end value at 2.1 mJ. This behavior can be explained by the different Al depth profiles
of the sheet steels. The Al concentration starts to increase at about 6 m for OCAS
II and at about 10 m for TKS I. Assuming comparable ablation behavior at each
particular burst energy for both sheet steels, the region of high Al concentration is
reached at a lower burst energy for OCAS II. Therefore, the relative number of Al
atoms in the ablated crater volume is also higher for OCAS II, which leads to a
higher intensity ratio IAl396 =IZn472 . Thus, the depth profile obviously influences the
ratio IAl396 =IZn472 measured as a function of the burst energy.
Based on these results, a method for a continuous inline detection of variations
in the aluminum depth profile can be deduced. The target Al profile is represented
by the ratio
IAl396
Rt .Eb / D .Eb /t : (16.5)
IZn472
The actual Al profiles are represented by the continuously measured ratio
IAl396
Rm .Eb / D .Eb /m : (16.6)
IZn472
Deviations from the target profile can then be determined by calculating the ratio
Rm .Eb /=Rt .Eb /. To avoid artifacts arising from scatter in the data, it is best to
determine polynomial fits to the data points Rm .Ebi / and Rt .Ebi / in a first step,
and then to divide the two polynomial functions. In the ideal case of an unchanged
profile, the ratio of the ratios would be constant and equals 1, see Fig. 16.15
(dotted line).
To simulate the case of changing profiles, Al depth profile of the OCAS II sample
is defined as the target profile and the Al depth profile of TKS I as the profile to be
monitored. The ratio of the two fitting functions (black lines in Fig. 16.14) is shown
in Fig. 16.15. This curve shows the variation in the Al depth profile for the sample
TKS I relative to that for sample OCAS II even better than the graphs in Fig. 16.14.
At low burst energy, the curve falls due to the higher Al concentration of TKS I
(0.15 m.-%) compared to OCAS II (0.1 m.-%) up to a depth of about 5–6 m.
Then it increases as the region of high Al concentration is reached for the OCAS
a
0.35
OCAS II IAl396 / I Zn472
0.30
0.25
I Al396 / I Zn472 [1]
0.20
0.15
0.10
0.05
0.00
0.0 0.4 0.8 1.2 1.6 2.0
b
0.35
TKS I IAl396 / I Zn472
0.30
0.25
I Al396 / I Zn472 [1]
0.20
0.15
0.10
0.05
0.00
0.0 0.4 0.8 1.2 1.6 2.0
Fig. 16.14 Referenced line intensity IAl396 =IZn472 versus laser burst energy. Samples: (a) Hot-dip
galvanized sheet steel (OCAS II) without Al in substrate; (b) Hot-dip galvanized sheet steel (TKS I)
with Al in substrate. Measurement parameters: tdelay D 0:5 s, tint D 3 s (IAl396 ), tint D 0:5 s
(IZn472 /
II sample at a lower depth (> 6 m) compared to a depth > 10 m for TKS I. It
increases further and saturates at burst energies > 1:8 mJ.
Since a laser burst consists of several pulses, one approach to enhancing the depth
information is to measure also between the pulses of one laser burst as well. The
first pulse ablates to an initial depth 1 and the LIBS signal contains the integrated
depth information up to this depth. The following pulses of the burst ablate deeper
and the LIBS signals correspond to larger depths. This approach was tested with
hot-dip galvanized samples of different coating thickness (TKS III, dZn D 7 m;
TKS IV, dZn D 10 m) using triple pulses. The burst parameters are shown in
Fig. 16.12b. The delay times after a pulse was set to 0:3 s. The widths of the
2.0
1.5
Rm(Eb) / Rt (Eb)
1.0
0.5 TKS I / OCAS II
0.0
0.0 0.4 0.8 1.2 1.6 2.0
Fig. 16.15 Ratio of the intensity ratios IAl396:15 =IZn472:22 of the samples TKS I and OCAS II of
Fig. 16.14 as a function of the burst energy. The curves visualize the deviation of the Al depth
profiles of the TKS I sample from that of sample OCAS II
integration windows were varied; they were shorter for Zn472 and Fe438 to avoid
saturation at higher burst energies, and longer for Al396 to get a sufficient signal
at low burst energies. The integration windows were 0:5 s for Zn472 and Fe438
and 2:0 s for Al396 after pulse 1; 1:0 s for Zn472, Fe438, and Al396 after pulse
2, and 0:7 s for Zn472, and Fe438 and 2:0 s for Al396 after pulse 3. The burst
energy was tuned in 51 steps from 0.165 to 1.65 mJ, and 80 measurements were
made at each energy step.
Figure 16.16 shows the corresponding IAl396 =IZn472 versus Eb graphs and
IFe438 =IZn472 versus Eb graphs of the 7m (Fig. 16.16, left) and the 10m
(Fig. 16.16, right) samples measured after each pulse of the triple burst. The burst
energy axis displayed for pulse 1 (depth 1) is the energy of the first pulse, for
pulse 2 (depth 2) it is the total energy of the first pulse and second pulse, and for
pulse 3 (depth 3) it is the complete burst energy. Figure 16.16a shows that both
coatings are already penetrated with the first pulse, as the IFe438 /IZn472 ratio starts to
increase at about 0.25 mJ for the 7m coating and at about 0.5 mJ for the 10m
coating. As the Fe2 Al5 -inhibition layer is located at the interface between coating
and substrate, it is approximately penetrated at the same energies. This is reflected
in the trends seen in the IAl396 =IZn472 ratios. For both coatings, the ratio starts at
about 0.04 and increases with increasing burst energy, but the increase is faster
for the thinner coating because the Al-rich inhibition layer is reached at 0.25 mJ.
The corresponding ratio IAl396 =IZn472 is 0.06, whereas for the thicker coating the
value of 0.06 is reached at a significantly higher burst energy of approximately
0.35 mJ. The IAl396 =IZn472 ratios and IFe438 =IZn472 ratios after pulse 2 are shown
in Fig. 16.16b. Both ratios increase with increasing burst energy, but the increase
a 0.12 0.5 0.12 0.5

hot dip galvanized hot dip galvanized
0.10 dZn ~ 7.0 µm 0.4 0.10 dZn ~ 10 µm 0.4
IAl396 /IZn472 IAl396 /IZn472
IFe438 /IZn472 [1]

IFe438 /IZn472 [1]
IAl396 /IZn472 [1]

IAl396 /IZn472 [1]
0.08 IFe438 /IZn472 0.08 IFe438 /IZn472

0.3 0.3
0.06 0.06
0.2 0.2
0.04 0.04
0.1 0.1
0.02 0.02
0.0 0.0
0.00 0.00
0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6
laser burst energy E1 [mJ] laser burst energy E1 [mJ]
b 0.9 2.0 0.9 2.0
0.8 hot dip galvanized IAl396 /IZn472 0.8 hot dip galvanized IAl396 /IZn472
dZn ~ 7.0 µm IFe438 /IZn472 1.6 dZn ~ 10 µm
0.7 0.7 IFe438 /IZn472 1.6
IFe438 /IZn472 [1]
IFe438 /IZn472 [1]

IAl396 /IZn472 [1]
IAl396 /IZn472 [1]

0.6 0.6
1.2 1.2
0.5 0.5
0.4 0.8 0.4 0.8
0.3 0.3
0.2 0.4 0.2 0.4
0.1 0.1
0.0 0.0
0.0 0.0
0.2 0.4 0.6 0.8 1.0 1.2 0.2 0.4 0.6 0.8 1.0 1.2
laser burst energy E1+ E2 [mJ] laser burst energy E1 + E2 [mJ]
c 0.9 0.9
2.8 IAl396 /IZn472 2.8
0.8 hot dip galvanized, dZn ~7.0 µm IAl396 /IZn472 0.8 hot dip galvanized
dZn ~10 µm IFe438 /IZn472 2.4
0.7 IFe438 /IZn472 2.4 0.7
IFe438 /IZn472 [1]

IAl396 /IZn472 [1]
IFe438 /IZn472 [1]
IAl396 /IZn472 [1]
0.6 2.0 0.6 2.0

0.5 1.6 0.5 1.6
0.4 0.4 1.2
1.2
0.3 0.3
0.8 0.8
0.2 0.2
0.4 0.4
0.1 0.1
0.0 0.0 0.0 0.0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
laser burst energy E1 + E2 + E3 [mJ] laser burst energy E1 + E2 + E3 [mJ]
Fig. 16.16 IAl396 =IZn472 ratios and IFe438 =IZn472 ratios as a function of the burst energy for coating
thicknesses of dZn D 7:0 m (TKS III) and dZn D 10:0 m (TKS IV) for different ablation depths.
(a) Pulse 1 (depth 1), tdelay D 0:3 s, tint D 2:0 s (IAl396 /, tint D 0:5 s (IZn472 , IFe438 ). (b) Pulse
2 (depth 2), tdelay D 0:3 s, tint D 1:0 s (IAl396 /, tint D 1:0 s (IZn472 , IFe438 ). (c) Pulse 3 (depth
3), tdelay D 0:3 s, tint D 2:0 s (IAl396 ), tint D 0:7 s (IZn472 , IFe438 /
is faster and the absolute values are higher for the thinner coating. This is due
to the higher number of ablated Zn atoms in the case of the thicker coating. The
coating is penetrated even at the lowest energy. After the third pulse (depth 3),
the IAl396 =IZn472 ratios are approximately constant, i.e., the coating is completely
penetrated. Again the IAl396 =IZn472 ratio and IFe438 =IZn472 ratio are higher for the
thinner coating.
The results discussed so far were all measured on a slow moving xy-translation
stage. In order to check the applicability of the technique for an inline monitoring on
moving sheet steel in a galvanizing plant and to study the influence of the sheet steel
speed on the LIBS signals, experiments were performed on rotating sheet steels of
the type OCAS II using the setup with the rotary stage. As the rotary axis is not
perfectly orthogonal to the rotary plane, a slight variation in the distance of 300 m
between the focusing lens and the sheet steel surface occurs within a full rotation.
Four different sheet steel velocities were tested. In the first step, a measurement
a 0.8 1.4
b 0.8 1.4
0.7 v0 = 0.008 m /s 0.7 v1 = 0.22 m /s 1.2
1.2
0.6 1.0 0.6 1.0
0.5 0.8 0.5 0.8
0.4 0.4 0.6
0.6
0.3 0.3
0.4 0.4
0.2 0.2
0.2 0.2
0.1 0.1
0.0 0.0
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
laser burst energy [mJ] laser burst energy [mJ]
c 0.8 1.4
d 0.8 1.4
0.7 v2 = 0.46 m/s 1.2 0.7 v3 = 0.96 m /s 1.2
0.6 1.0 0.6 1.0
0.5 0.8 0.5 0.8
0.4 0.4 0.6
0.6
0.3 0.3
0.4 0.4
0.2 0.2
0.2 0.2
0.1 0.1
0.0 0.0
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
laser burst energy [mJ] laser burst energy [mJ]
Fig. 16.17 Referenced line intensity IAl396 =IZn472 and IFe438 =IZn472 versus laser burst energy for
different sheet steel velocities. Sample: OCAS II. Measurement starts after pulse 2 of double pulse.
Parameters: tdelay D 0:25 s, tint D 3 s (IAl396 ), tint D 1:0 s (IZn472 , IFe438 ). (a) v0 D 0:008 m=s,
(b) v1 D 0:22 m=s, (c) v2 D 0:46 m=s, (d) v3 D 0:96 m=s
was taken on a slowly moving sheet steel (v0 0.008 m/s), which is the reference
measurement. Double pulses were used, and the burst energy was varied between
0.13 and 1.3 mJ in 51 steps; the intensities of the lines Al396, Zn472, and Fe438
were measured after the second pulse of the burst. In Fig. 16.17a, the IAl396 =IZn472
ratio and IFe438 =IZn472 ratio are shown as a function of the burst energy. The result
for the IAl396 =IZn472 ratio is different to that obtained on the xy-translation stage,
see Fig. 16.14a, due to the different settings of the integration delay and window.
Figure 16.17b–d shows the results for sheet steel velocities of v1 D 0:22, v2 D 0:46,
and v3 D 0:96 m=s. All the graphs look very similar, although both ratios are slightly
lower at 0.96 m/s. To see the influence of the sheet steel speed in more detail, the
IAl396 =IZn472 ratios measured at v1 D 0:22, v2 D 0:46, and v3 D 0.96 m/s are
referenced to the IAl396 =IZn472 ratio measured at v0 D 0.008 m/s, see Fig. 16.18.
The graphs were generated similar to Fig. 16.15, i.e., first a polynomial fit is applied
to the curves in Fig. 16.17 and after that the fit curves are divided. For 0.22 and
0.46 m/s, and burst energies 0.2 mJ the ratio is 1.
The variations from this – in particular at low burst energies for 0.46 m/s – are due
to the uncertainty associated with the fits. However, clear deviations can be seen for
0.96 m/s. On average the IAl396 =IZn472 ratio measured at 0.96 m/s is approximately
15% lower than that seen at the other sheet steel speeds. This effect may be due to
the limited speed of the autofocus system, which means that the sample surface is
on average slightly out of focus, meaning that the penetration depth at a given burst
energy and therefore the relative number of Fe and Al atoms is reduced.
1.4
v1 = 0.22 m / s / v0 = 0.008 m / s
IAl396 /IZn472(vi) / IAl396 /IZn472 (v0) 1.3
v2 = 0.46 m / s / v0 = 0.008 m / s
1.2 v3 = 0.96 m / s / v0 = 0.008 m / s
1.1
1.0
0.9
0.8
0.7
0.6
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Fig. 16.18 Ratio of the intensity ratios IAl396 =IZn472 of the sample OCAS II measured at v1 D
0:22, v2 D 0:46, and v3 D 0:96 m=s to the intensity ratio IAl396 =IZn4;722 measured at v0 D
0:008 m=s as a function of the burst energy. The graphs were generated similar to Fig. 16.15 and
visualize the effect of the sheet steel speed on the intensity ratio IAl396 =IZn472 , see text
The results demonstrate the potential of LIBS for inline monitoring of the
aluminum depth profile of galvanized coatings and offers the possibility to further
improve the quality control in galvanizing plants.
For an inline monitoring, the measurement time is an important issue. For the
described experiments, this time is limited by the repetition rate of the applied
flashlamp pumped Nd:YAG laser of 10 Hz. By replacing this laser by a diode-
pumped solid state laser (DPSSL, cf. Sect. 4.1), the acquisition rate can be increased
up to 1 kHz. Due to the better beam quality of the DPSSL of M 2 1 also the
Rayleigh length of the focus is increased, and the method is expected to become
less sensitive to sheet steel vibrations. The measurement time for an Al depth profile
will be in the range of 5 s (50 burst energy steps and 100 averages). In combination
with the method for inline measurement of the Zn coating thickness, the Al depth
profile can be monitored simultaneously [16.13]. As LIBS is sensitive to other light
elements such as Mg, this method is as well of interest for the inline analysis of
magnesium zinc coatings and may become an alternative method to XRF for inline
monitoring of metallic coatings [16.32].
References
16.1. The role of aluminum in continuous hot-dip galvanizing. http://www.galvinfo.com/

GINotes/G Note10.PDF (2004)
16.2. Metallic coatings – measurement of coating thickness – X-ray spectrometric methods, ISO-
Norm 3497: (E) (2000)
References 427
16.3. Ed. R. Payling, D. Jones, A. Bengtson, Glow Discharge Optical Emission Spectrometry
(Wiley, New York, 1997)
16.4. D. Anderson, C. McLeod, T. English, A. Smith, Depth profile studies using laser-induced
16.5. L. St-Onge, M. Sabsabi, Towards quantitative depth-profile analysis using laser-induced
plasma spectroscopy: investigation of galvannealed coatings on steel. Spectrochim. Acta B
55, 299–308 (2000)
16.6. V. Margetic, M. Bolshov, A. Stockhaus, K. Niemax, R. Hergenröder, Depth profiling of
multi-layer samples using femtosecond laser ablation. J. Anal. At. Spectrom. 16, 616–621
(2001)
16.7. M. Mateo, J. Vadillo, J. Laserna, Irradiance-dependent depth profiling of layered materials
using laser-induced plasma spectrometry. J. Anal. At. Spectrom. 16, 1317–1321 (2001)
16.8. J. Vadillo, C. Garcia, S. Palanco, J. Laserna, Nanometric range depth-resolved analysis of
coated-steels using laser-induced breakdown spectrometry with a 308 nm collimated beam.
16.9. C. Garcia, M. Corral, J. Vadillo, J. Laserna, Angle-resolved laser-induced breakdown
spectrometry for depth profiling of coated materials. Appl. Spectrosc. 54, 1027–1031
(2000)
16.10. M. Mateo, L. Cabalı́n, J. Laserna, Line-focused laser ablation for depth-profiling analysis
of coated and layered materials. Appl. Optics 42, 6057–6062 (2003)
16.11. D. Papazoglou, V. Papadakis, D. Anglos, In situ depth and topography monitoring in LIBS
elemental profiling of multi-layer structures. J. Anal. At. Spectrom. 19, 483–488 (2004)
16.12. M. Mowery, R. Sing, J. Kirsch, A. Razaghi, S. Bechard, R. Reed, Rapid at-line analysis of
coating thickness and uniformity on tablets using laser induced breakdown spectroscopy.
J. Pharm. Biomed. Anal. 28, 935–943 (2002)
16.13. H. Balzer, M. Hoehne, V. Sturm, R. Noll, Online coating thickness measurement and depth
Acta B 60, 1172–1178 (2005)
16.14. H. Balzer, S. Hölters, V. Sturm, R. Noll, Systematic line selection for online coating
thickness measurements of galvanised sheet steel using LIBS. Anal. Bioanal. Chem. 385,
234–239 (2006)
16.15. M. Martin, N. Labbe, T. Rials, S. Wullschleger, Developing laser-induced breakdown
spectroscopy as a high throughput technique for quantifying the elemental composition
of preservatives-treated wood. Spectrochim. Acta B 60, 1179–1185 (2005)
16.16. M. Kraushaar, R. Noll, H.U. Schmitz, Slag analysis with laser-induced breakdown
spectrometry. Appl. Spectrosc. 57, 1282–1287 (2003)
16.17. H. Fink, U. Panne, R. Niessner, Process analysis of recycled thermoplasts from consumer
electronics by laser-induced breakdown plasma spectroscopy. Anal. Chem. 74, 4334–4342
(2002)
16.18. S. Palanco, J. Laserna, Full automation of a laser-induced breakdown spectrometer for
quality assessment in the steel industry with sample handling, surface preparation and
quantitative analysis capabilities. J. Anal. At. Spectrom. 15, 1321–1327 (2000)
16.19. H.S. Huang, K.C. Lin, Laser-induced breakdown spectroscopy of liquid droplets: corre-
lation analysis with plasma-induced current versus continuum background. J. Anal. At.
Spectrom. 20, 53–59 (2005)
16.20. V. Lazic, R. Fantoni, F. Colao, A. Santagata, A. Morone, V. Spizzichino, Quantitative
laser induced breakdown spectroscopy analysis of ancient marbles and corrections for the
variability of plasma parameters and of ablation rate. J. Anal. At. Spectrom. 19, 429–436
(2004)
16.21. S. Lui, N. Cheung, Resonance-enhanced laser-induced plasma spectroscopy: ambient gas
effects. Spectrochim. Acta B 58, 1613–1623 (2003)
16.22. V. Detalle, M. Sabsabi, L. St-Onge, A. Hamel, R. Heon, Influence of Er:YAG and Nd:YAG
wavelengths on laser-induced breakdown spectroscopy measurements under air or helium
atmosphere. Appl. Optics 42, 5971–5977 (2003)
16.23. L. Peter, V. Sturm, R. Noll, Liquid steel analysis with laser-induced breakdown spectrom-
etry in the vacuum ultraviolet. Appl. Optics 42, 6199–6204 (2003)
16.24. S. Feliu Jr., V. Barranco, XPS study of the surface chemistry of conventional hot-
dip galvanised pure Zn, galvanneal, and Zn-Al alloy coatings on steel. Acta Materialia
51, 5413–5424 (2003)
16.25. Ed.R. Payling, D. Jones, A. Bengtson, Glow Discharge Optical Emission Spectrometry,
(Wiley, New York, 1997)
16.26. H. Balzer, M. Höhne, R. Noll, V. Sturm, New approach to monitoring the Al depth profile
of hot-dip galvanised sheet steel online using laser-induced breakdown spectroscopy. Anal.
Bioanal. Chem. 385, 225–233 (2006)
16.27. D. Anderson, C. McLeod, W. English, A. Smith, Depth profile studies using laser-induced
16.28. P. Pouli, K. Melessanaki, A. Giakoumaki, V. Argyropoulos, D. Anglos, Measuring the
thickness of protective coatings on historic metal objects using nanosecond and fem-
tosecond laser induced breakdown spectroscopy depth profiling, Spectrochim. Acta B 60,
1163–1171 (2005)
16.29. V. Detalle, J. Lacour, P. Mauchien, A. Semerok, Investigation of laser plasma for solid
element composition microanalysis. Appl. Surf. Sci. 138–139, 299–301 (1999)
16.30. H. Balzer, M. Hoehne, S. Hoelters, V. Sturm, R. Noll, E. Leunis, S. Janssen, M. Raulf,
P. Sanchez, M. Hemmerlin, Online depth profiling of zinc coated sheet steel by laser-
induced breakdown spectroscopy, in 7th International Workshop Progress in Anal. Chem.
in the Steel and Metal Industries, ed. by J. Angeli, (Glückauf GmbH, Essen, 2006),
pp. 237–242
16.31. X. Mao, A. Ciocan, R. Russo, Preferential vaporization during laser ablation inductively
coupled plasma atomic emission spectroscopy. Appl. Spectrosc. 52, 913–918 (1998)
16.32. ZE-Mg Zink-Magnesium-Oberflächen, Die neue Generation von Zinklegierungsüber-
zügen: korrosionsbeständiger, verarbeitungsfreundlicher. http://www.thyssenkrupp-steel.
com/upload/binarydata tkscsauto/3695/ze mg de.pdf (2003)
Chapter 17
LIBS Instruments
This chapter describes LIBS instruments designed for different application fields,
the requirements, setup, and performance. The topics chosen for the following
sections refer mainly to industrial applications in R&D, inline process control, and
quality inspection.
A comprehensive presentation of LIBS instruments for applications such as, e.g.,
security, space exploration, or in pharmaceutical industry is outside the scope of
this chapter. The different types of LIBS instruments for industrial applications are
described in the following with the help of typical examples.
17.1 Laboratory Systems
Laboratory systems are defined here as LIBS instruments designed for use in a
laboratory of an industrial company. As an example, a LIBS system is presented
for the analysis of steel samples having a scale layer. The industrial application
case and the methodology to ablate the scale layer and to subsequently analyze
the bulk steel sample with LIBS is described in detail in Sect. 13.1.2. In the
following, the requirements, concept, and design of the LIBS system called ATLAS
is presented [17.1].
In a first step, the requirements are defined with the end user of the ATLAS
system, in this case a steel plant. Key points are the times for combined sample
preparation and analysis to be attained, correctness and precision of the analysis, the
procedure of production control in the steel plant (interval operation), the number of
samples to be analyzed per day, sample types, thicknesses of scale layers, elements
to be analyzed, and their concentration ranges.
Table 17.1 lists the times of the different process steps for a conventional sample
analysis and those aspired for the ATLAS system. The laser method has the potential
to reduce the total time by up to 60 s, which allows to shorten the feedback time for
process guiding. The limit of detection to be achieved especially for the elements

430 17 LIBS Instruments
Table 17.1 Comparison of the times of the process steps needed for conventional sample analysis
and those aimed for the LIBS-based ATLAS system
Process step Conventional tmin tmax (s) ATLAS tmin tmax (s)
Pneumatic dispatch from converter 20–35 20–35
to laboratory of steel plant
Receipt and preparation by 70 n/a
grinding
Transport from grinding machine 12–15 n/a
to SD-OES system
SD-OES analysis (two to four 40–80 –
sample positions)
Laser ablation and LIBS (two to – 60–120
four sample positions)
Total time (s) 142–200 80–140
n=a, not applicable; SD-OES, spark discharge optical emission spectroscopy
C, P, and S is < 10 g=g and the relative standard deviation of procedure should be
about 2 rel.% [cf. Sect. 11.2, relation (11.13)].
The LIBS system has to be able to analyze on average 420 samples per day.
This number may vary between 320 and 550. On average, the time between two
consecutive samples arriving at the laboratory of the steel plant amounts to 3.5 min.
After taking of the production control sample in the plant, the sample is transported
via pneumatic dispatch to the steel works laboratory, which takes 20–35 s. Within
this time, the LIBS system has to be stabilized and set into the state “ready for
measurement.” A sample is measured at two to four sample positions [K D 2–4,
cf. Sect. 11.1, relation (11.5)] depending on the mutual deviations of the measuring
results. A measurement is divided into three steps:
P
24
1. Ablation of the scale layer at K D 2–4 sample positions, tabl;K D tabl;k ,
kD1
P
24
2. Analysis of the bulk material at K D 2–4 sample positions, tLIBS;K D tLIBS;k ,
kD1
P
24
3. Change of sample position for K D 2–4 sample positions, TSP;K D TSP;k1 .
kD2
The total response time of the ATLAS system is then: tr D tabl;K C tLIBS;K C TSP;K ,
which should be in the range between 60 and 120 s, cf. Table 17.1.
The samples to be analyzed are production control samples, see Fig. 17.1,
left, with a scale layer on the surface formed during the cool down phase of the
sample. For comparison Fig. 17.1, right shows a production control sample after
grinding. Conventional production control requires the removal of the scale layer at
one side of the sample by grinding or milling prior to SD-OES analysis.
The scale layer covers the sample surface completely. It has a thickness of several
tenths of millimeters. The chemical composition of the scale layer differs from the
composition of the bulk material. The carbon content in the scale layer, e.g., is not
representative for the carbon content of the bulk material. A standard procedure
is to irradiate a series of laser pulses onto the same position on the sample surface
17.1 Laboratory Systems 431
Fig. 17.1 Production control sample, left: original appearance with scale layer, right: after
grinding; width of the sample: about 35 mm
Fig. 17.2 Crater and plasma geometry of a sample covered with a surface layer of different
composition. LB D laser beam, S D bulk sample, SL D surface layer, MT D melt phase of
transition region, MS D melt phase of surface layer, P D plasma
(cf. Sect. 11.1, prepulses and measuring pulses). The laser beam ablates material and
a crater is formed penetrating the surface layer and extending into the bulk material
to be analyzed. In general, the result of the interaction with the prepulses is residues
at the crater walls, such as, e.g., solidified melt phases of a metallic bulk sample.
This situation is illustrated schematically in Fig. 17.2 [17.2]. These residues (MS in
Fig. 17.2) contain constituents from the surface layer (SL) or from both the surface
layer and the bulk material (MT). Similar issues have been discussed in various
papers [17.3,17.4,17.5]. Since the plasma expands very rapidly and strikes the crater
walls, there is a principal danger of crosstalk, i.e., the plasma emission originates
from material from both the bulk material and the surface layer. This is caused
by the mixing of melt phases in the transition region and subsequent excitation by
the plasma. Apart from that, the plasma conditions are changing when penetrating
pulse-by-pulse into a material forming a deep crater (depth > width), as shown
schematically in Fig. 17.3 and studied, e.g., by [17.6].
Penetration of hard surface layers – such as scale layers on steel samples
discussed here – involves focusing of a series of laser pulses on one spot for a
stepwise ablation of the surface layer to access the bulk material to be analyzed.
For the first few laser pulses, the sample surface can be considered as being flat
Fig. 17.3 Change of plasma parameters depending on the crater aspect ratio, case 1: crater depth <
crater width, case 2: crater depth > crater width. ne D electron density, Te D electron temperature
Fig. 17.4 Crater geometry generated by a step-by-step reduction of the laser beam width to
minimize cross talk and to stabilize plasma conditions. The crater width is reduced successively
with increasing depth (no. 1–3). LB, P, SL, S D see Fig. 17.2. 2wi , diameter of the laser beam; dCi ,
diameter of crater; hCi , depth of crater
and the laser-induced plasma expands into the free half space (case 1 in Fig. 17.3)
[17.2]. After a series of laser pulses, a deep crater is formed and the interaction
of the plasma with the crater walls becomes significant (case 2). The plasma state
described by the temporal evolution of electron temperature and electron density is
shown qualitatively for the two cases considered in Fig. 17.3 on the right.
In the crater, the induced plasma initially has both higher electron density and
temperature due to the confinement by the crater. On the other hand, it is quenched
by the contact with the crater wall and thus relaxes more rapidly. If the integration
window parameters tdelay and tint have been optimized for the flat surface, this
corresponds to a plasma state (ne1 , Te1 /. As the crater forms, the plasma state during
the fixed integration window changes from (ne1 , Te1 / to (ne2 , Te2 /. The integration
window parameters are not optimal for the new plasma state; hence, this effect
limits the precision of LIBS measurements if not taken into account adequately.
The approach chosen here to overcome the limitations of cross talk and change of
plasma conditions is the step by step generation of a defined crater geometry as
illustrated in Fig. 17.4 [17.2, 17.7]. The deposition of the laser energy is controlled
spatially by adjusting the beam width and focal position. This is performed by an
x-, y-, z-scanner steering the laser beam onto the sample.
3
C 2
1
4
5
B 6 7
8
12
A
11
9
10
Fig. 17.5 Components of the ATLAS system for the analysis of steel samples with scale layers.
The components are mounted in three levels AC. 1, multichannel integrator electronics (MCI; cf.
Sect. 4.5); 2, laser source; 3, shutter box; 4, Glan laser prism to attenuate the laser pulse energy; 5,
beam shaping and beam diagnostic module; 6, x-, y-, z-scanner; 7, spectrometer; 8, sample stand
(cf. Sect. 4.6); 9, control computer; 10, vacuum pump; 11, electric wiring board; 12, fan
Figure 17.5 illustrates the components of the ATLAS system comprising an x-,
y-, z-scanner. The setup is structured in three levels, AC. The laser source (number
2 in Fig. 17.5) is mounted in level C to assure good accessibility for, e.g., the
exchange of the laser flashlamp. The laser pulse energy is attenuated externally by
a Glan laser prism (4) to the pulse energy necessary for the ablation and analysis
step. The attenuated beam enters a beam shaping and beam diagnostic module (5),
which allows to adjust the spatial beam profile and an inline monitoring of the beam
parameters. The laser beam propagates through the scanner (6), which deflects the
laser beam for the ablation step prior to the analysis in the center of the ablation
crater as illustrated in Fig. 17.4.
After the scanner, the beam passes the laser channel in the sample stand (8) and
is focused onto the production control sample with the scale layer. This sample is
positioned by a sample handling system that is an integral part of the sample stand.
The measuring radiation is guided to a Paschen–Runge spectrometer (7), spectrally
dispersed and detected by photomultiplier (PMT) tubes. The signals are processed
by a multichannel integrator electronics (1), which also synchronizes the whole
measuring process.
The dust-proof housing is equipped with an air conditioning to reduce the
influence of changing ambient conditions on the stability of the measuring system
Fig. 17.6 Photograph of the

ATLAS system
such as temperature, humidity, and dust load. Figure 17.6 shows a photograph of the
LIBS system ATLAS for the analysis of scaled steel samples.
Figure 17.7 shows a photograph of the sample stand and its components, see
figure caption. Figure 17.8 illustrates the operation in four steps. The first photo
shows the sample stand with the pivoting plate (cf. Fig. 17.7) in horizontal position
ready to take over the production-control sample. In the second photograph, the
sample is put manually or by a robot onto the sample support and fixed by activation
of an underpressure in the sample support. In the third photo, the pivoting plate is
moving into an upright position driven by a pneumatic cylinder. Finally, the fourth
photo shows the pivoting plate in vertical position. The sample is now fixed onto the
rotating disk and ready to start the ablation and measurement phases.
After a first measurement, the rotating disk allows to rotate the sample by a
defined angle to a further measurement position as illustrated in Fig. 17.9. This
procedure is repeated several times depending on the number of measuring positions
required. The radius of the white circle shown in Fig. 17.9 amounts to about 12 mm.
The angular positions differ by 20ı .
After the measurement, the pivoting plate moves again to the horizontal position,
the underpressure is deactivated and the production control sample can be taken
away.
The composition range of the production control samples to be analyzed by
ATLAS is given in Table 17.2.
Fig. 17.7 Sample stand with

sample handling system. The
1
sample handling system is
shown in the acceptance
position. 1, connecting plate; 2
2, rotating disk; 3, production
control sample; 4, sample
support; 5, pivoting plate; 6,
housing for pneumatic
cylinder 3
4 5
Fig. 17.8 Operation of sample stand shown in a sequence of four steps. From left to right: (a) The
pivoting plate is in horizontal position ready to take over the production control sample. (b) The
sample is put onto the sample support. (c) The pivoting plate moves into an upright position. (d)
The pivot plate has reached the vertical position and is in contact with the connecting plate. The
sample is fixed onto the rotating disk and the measurement starts
To detect the elements listed in Table 17.2, a Paschen–Runge spectrometer

(cf. Sect. 4.2) was designed equipped with 27 detectors in the spectral range from
178 to 438 nm. Figure 17.10 shows a photograph of the spectrometer.
The signals of the PMTs are transferred to the multichannel integrator electronics
(MCI; cf. Sect. 4.5), see Fig. 17.11.
The MCI has a modular structure comprising a set of A/D-converter electronic
boards where each board has 16 inputs for PMT signals. Up to four A/D-converter
boards are linked to a signal electronics having the following features: ethernet
interface for communication with a PC, input for an incremental encoder, three
trigger inputs, and eight trigger outputs with a resolution of 10 ns to control process
sequences, digital input and outputs likewise to control process sequences.
Fig. 17.9 Production control

sample with scale layer. By
rotation of the sample in the
rotating disk (no. 2 in
Fig. 17.7), a series of
measuring positions can be
defined lying on a circle
Table 17.2 Ranges of chemical composition of production control samples to be analyzed with
ATLAS
Element cmin (m.-%) cmax (m.-%) Element cmin (m.-%) cmax (m.-%)
Fe 85 100 Mo 0 1:7
C 0 1:5 Ni 0 12
Si 0 3:5 Cr 0 4
Mn 0 2:5 V 0 0:6
P 0 0:1 Ti 0 0:45
S 0 0:1 B 0 0:13
Cu 0 1:5 Co 0 0:35
Al 0 0:5 Ca 0 0:015
N 0 0:03
For each PMT signal, two independent integration channels are available with
different sensitivity. Each of the 32 integrator channels of a single A/D-converter
board is digitized by a 24-bit A/D converter. The position and width of the
integration windows are freely programmable for each PMT signal.
The ATLAS system was tested with 40 reference samples to determine limits
of detection (LOD) in different crater depths, see Table 17.3 (cf. Sect. 11.2). The
number of prepulses applied is kept constant: Npp D 200 (cf. Sect. 11.1). The
number of measuring pulses is Nmp D 3; 000. Table 17.3 shows the achieved
LODs for different pulse ranges within this series of 3,000 pulses corresponding
to different crater depths. The crater depths are in the range of 0–50 m for 1–500
pulses, 50–100 m for 501–1,000 pulses, and so on up to 250–300 m for 2,501–
3,000 pulses. No systematic change of LODs with increasing crater depths up to
300 m is observed. For nine elements, LODs less than 10 g=g are achieved for a
crater depths range of 250–300 m.
17.2 Inspection Systems for Inline Process Control 437
Fig. 17.10 Paschen–Runge

spectrometer (no. 7 in
Fig. 17.5) of the ATLAS
system. Above the middle:
feed-throughs of the
photomultiplier signals,
bottom: interface to the
sample stand
Fig. 17.11 Multichannel

integrator electronics to
process the signals of the
photomultipliers
The limit of detection depends on the number of measuring pulses used to

average the measuring signal, see Table 17.4. For most elements, the best LODs
were achieved for an averaging over 3,000 measuring pulses. For 14 elements, the
LODs are better than 10 g=g.
The ATLAS system was tested extensively on-site in a steel plant [17.2].
Figure 17.12 shows the concentration of manganese of scaled process control sam-
ples determined with ATLAS compared to the reference value from conventional
analysis. For the conventional analysis, the samples have to be ground to remove the
scale layer and are then analyzed by a spark optical emission spectrometer (OES).
This graph shows the results of LIBS measurements of 276 scaled samples. The
residual deviation as a measure for the accuracy is 2.3 rel.-% [cf. Sect. 11.3, relation
(11.21)].
17.2 Inspection Systems for Inline Process Control
Inspection systems for inline process control are defined here as LIBS instruments
designed for use in an industrial plant. They are integrated in a production line. The
measuring results gained inline are evaluated to control preceding or subsequent
Table 17.3 Limits of detection for different crater depths up to 300 m

Element LOD3 s .g=g), pulse range
1–500 501–1000 1,001–1,500 1,501–2,000 2,001–2,500 2501–3,000
Al 4 6 6 5 8 6
As 7 5 7 9 13 14
B 1 4 2 2 1 3
C 7 9 11 9 7 12
C 6 17 18 7 19 12
Ca 9 7 3 2 1 6
Co 16 19 29 31 30 13
Cr 26 12 19 25 31 23
Cu 3 1 4 6 5 3
Mn 5 12 13 9 9 6
Mo 16 23 21 15 24 21
Nb 23 10 18 23 19 15
Ni 9 20 27 15 16 14
P 9 20 18 10 10 8
S 9 8 13 7 9 7
Si 45 14 19 11 15 24
Ti 2 2 2 3 3 1
V 11 7 12 14 18 8
The blank sample is measured ten times. Data are evaluated for a linear calibration curve in the
lower concentration range. For carbon two lines are installed: C 193.1 nm and C 165.8 nm. s D
8 mm
process steps. The requirements on such machines can be summarized as follows:

automated operation, capability to work under harsh industrial conditions, long-term
stability and reliability, user-friendly operation and maintenance.
As an example for an inline inspection machine based on LIBS, a system for
the identification of material mix-ups in a production line of pipe fittings and tubes
made of high-alloy steel grades will be presented [17.8]. Especially in the oil and gas
industry, pipe lines and process piping are increasingly exposed to corrosion as the
fields become more sour, wells are deeper with higher pressures and hotter products.
In order to meet the requirements with regard to corrosion resistance and mechanical
stability, more than 30 different steel grades are used for the production of pipe
fittings, in particular alloy steels, stainless steels, duplex, superduplex, 6Mo grades,
high nickel alloys, titanium, and clad steels. The concentration range of Fe and Ni
extends to 100 m.-%, the range of Cr, Mo, and Cu to 30 m.-%. The use of wrong steel
grades can lead to corrosion with severe consequential damages. Growing quality
requirements, particularly in the nuclear industry and environmental responsibilities
demand a material identification of each produced pipe fitting. The developed
inspection system is called LIFT, which stands for Laser I dentification of F ittings
and T ubes.
Before LIFT was installed, the fittings were inspected by Spark-OES. The fittings
had to be cleaned and a considerable percentage of the fittings had to be measured
two or three times before they had been correctly identified. The duration of one
Table 17.4 Limits of detection for averages of 1,000, 2,000, and 3,000 measuring pulses for each
sample position
Element LOD3 s .g=g), number of pulses
1,000 2,000 3,000
Al 4 2 3
As 5 5 6
B 2 2 1
C 4 5 4
C 10 9 7
Ca 7 3 2
Co 16 20 19
Cr 6 12 15
Cu 2 2 2
Mn 7 7 5
Mo 13 13 9
Nb 14 14 12
Ni 10 11 8
P 14 10 7
S 7 7 5
Si 26 17 10
Ti 1 1 1
V 6 7 7
The blank sample is measured ten times
2.0
Mn 293.3 nm / Fe 273.1 nm
1.8
recalibrated values
1.6 Rc, rel = 2.3 rel.- %
c LIBS [m.- %]
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
c Reference [m.- %]
Fig. 17.12 Concentration of manganese in scaled steel samples measured with LIBS without any
mechanical preparation compared to the reference concentration values gained with conventional
methods requiring grinding of the sample to remove the scale layer
inspection was 4 s and the electrode of the Spark-OES had to be cleaned every
three measurements. The throughput was 60 fittings per hour. The surface of the
fittings had to be abraded after the spark discharge. In a third step, the fittings
had to be marked with an ink jet printer or by electrolytic etching. In view of a
Fig. 17.13 Schematic setup of the inspection machine LIFT for the automated identification of
steel grades of pipe fittings. PF D pipe fitting, CC D table type circular conveyor, P D protection
cover, MC D machine cabinet, EC D electric cabinet, MO D monitor, operation panel, LS D laser,
LTS D laser triangulation sensor, SP D spectrometer, IJ D ink jet marker, G D gripper, BC D belt
conveyor, PU D packaging unit. The numbers one to five indicate the sequence of actions. 1 D
input of pipe fitting to be inspected, 2 D inspection with LIBS, 3 D marking of the specimen, if
result of inspection is okay, 4 D the gripper takes the specimen and puts it on the belt conveyor,
5 D packaging of the specimens
material inspection of each produced pipe fitting, the economic efficiency had to be
improved.
The first generation of LIFT systems was installed in a plant in the year 1998
and is in operation since then with only minor interrupts. Within a period of 4 years,
more than 108 LIBS measurements have been performed and monitoring data was
stored to assess the system status. This is the largest number of data collected with
a single LIBS setup so far offering the possibility to study the long-term behavior of
a LIBS system in routine industrial use.
Figure 17.13 illustrates the setup of the inspection machine, and Fig. 17.14 shows
a photograph of the first installed system in the production line. The fittings are
inserted by hand into a table-type circular conveyor with four positions (numbered
with 1–4 in Fig. 17.13). After the handling system is activated with a switch for
two hands, the inserted fitting is positioned in front of the laser inspection machine
(position 2). A protection cover encloses the fitting during the inspection and assures
the laser safety according to IEC 825–1. The duration of the inspection is 2 s. If the
result of the inspection is “OK,” the fitting will be marked at the third position
of the handling system with an ink jet printer. At the fourth position, the fitting
is transferred by a gripper to a belt conveyor and transported to the packaging unit.
When a material mixing is identified by LIFT, the fitting will not be marked. Instead,
it will be sorted out reversing the direction of movement of the belt conveyor. The
handling system can be used for sizes of the pipe fittings ranging from 0.5 to 800 and
the maximum throughput amounts to 450 fittings per hour.
Fig. 17.14 First LIFT inspection system installed in the production plant. Right: inspection
system, middle: table type circular conveyor to feed the pipe fittings, left: belt conveyor to transport
the inspected parts to the packaging line, cf. Fig. 17.13
In order to operate LIFT in dusty surroundings at a humidity of up to 90% and

at an ambient temperature of up to 40ı C, all components of the laser inspection
machine are installed in a closed air-conditioned cabinet. For maintenance purposes,
the cabinet can be entered through a door that is usually locked.
A TCP/IP interface enables to backup measurement data and records regularly
by the host computer. An interface for the remote maintenance by modem allows for
an operational support, e.g., to set parameters of LIFT or for diagnosis and recovery
of errors during operation.
Figure 17.15 shows the optical setup of LIFT. The emitted light of the Nd:YAG
laser (50 Hz repetition rate, 100 mJ) passes a beam splitter where a small part is
reflected to a photo diode to monitor the laser pulse energy. The laser is guided
via two mirrors, shutter, quarter wave plate, focusing lens, beam combiner, and
window to the specimen. The quarter wave plate protects the laser from back
reflections originating from the target, which may cause damage to the laser system.
A mechanical shutter blocks the laser beam when no measurement is activated. The
position of the workpiece in relation to the beam waist of the Nd:YAG laser beam
determines the irradiance in the interaction region. A triangulation sensor measures
the sample position d between a reference plane and the surface of the specimen
using a wavelength of 670 nm generated by a diode laser. The beam combiner
transmits the laser beam of the triangulation sensor propagating collinearly to the
Nd:YAG laser beam to the interaction region. Window W1 in Fig. 17.15 is part of a
wall of the machine cabinet that separates the outer region from the internal volume
of the machine. This window is heated in order to avoid condensation of water from
the outer humid atmosphere.
Fig. 17.15 Optical setup of LIFT. LA D Nd:YAG laser, BS D beam splitter, PD D photo diode,
M D mirror, LB D laser beam, SH D shutter, QW, quarter wave plate, L D focusing lens, BC D
beam combiner; W1, W2 D window; S D specimen, TS D triangulation sensor, FO D fiber optics,
d D working distance
The emitted light of the plasma is guided with fiber optics (length 2 m, core
diameter 1 mm, numerical aperture 0.22) – positioned at a distance of about 65 mm
from the plasma – to a Paschen–Runge spectrometer equipped with 12 PMTs to
detect the element-specific radiation of the elements Fe, Ni, Cr, Mo, Ti, Cu, Nb,
Al, and W. The diameter of the Rowland circle amounts to 500 mm, the grating
has 2,700 l/mm. Table 17.5 shows the list of lines installed in the spectrometer. The
entrance slit of the spectrometer can be shifted laterally by a servo motor. Detecting
measuring signals of the photomultipliers as a function of the entrance slit position
yields line profiles used to define the correct position of the entrance slit in order
to move the line maxima to the positions of the respective exit slits (cf. Fig. 14.61).
By this, drifts of the adjustment of the spectrometer can be compensated. The PMT
signals are transmitted via I/U-converters to a multichannel integrator electronics.
The signals are integrated in a defined time window with respect to the laser pulse.
Typical parameters for the delay and integration time are tdelay D 250 ns, tint D
10 s. The same integration window is used for all PMT channels.
The measurement time amounts to 2 s comprising Npp D 30 laser pulses as
prepulses to ablate surface contaminations – as, e.g., oil residues from preceding
processing steps – and the spectral signals generated by the subsequent Nmp D 70
laser pulses are evaluated for material identification. During the irradiation of the
laser pulses, the interaction area is flushed with filtered pressurized air. For the
classification of the measured data set concerned, an expert system was developed
based on a hierarchical structure of rules to identify the steel grade. According
to this hierarchy, the data set is tested step by step. Based on the outcome of
each test, the next rule to be applied is selected. In this way, only a subset of
the rules has to be evaluated and the consultation time is minimized to a few
milliseconds. A calibration is necessary for the training of the rule set. During the
Table 17.5 List of lines installed in the spectrometer of LIFT

Element (nm) Elow (eV) Eupp (eV)
Fe I 371.99 0.00 3.33
Fe II 273.07 1.08 5.62
Ni I 351.50 0.11 3.64
Ti II 337.28 0.01 3.69
Cr II 267.72 1.55 6.18
Cr II 298.92 3.74 7.89
Al I 396.15 0.01 3.14
Nb II 319.50 0.33 4.21
Cu I 324.75 0.00 3.82
WI 400.87 0.37 3.46
Mo II 281.62 1.67 6.07
calibration, several samples of each steel grade are measured. The calibration data
set establishes the knowledge base of the expert system. In addition to the normal
samples, five reference samples are measured during the calibration. The data of
these measurements can later be used for the recalibration of the algorithm (cf.
Sect. 11.4). The recalibration can be performed within 5 min. During the automatic
operation, all data sets and their classification are logged.
For the identification of different steel grades, it is necessary to detect the specific
differences of the element concentrations. It is well known that internal standardiza-
tion of an analyte line reduces the standard deviation of the spectral signals [cf.
Sect. 11.1, relation (11.2)]. To assess the effect of internal standardization of analyte
lines and the capability of a clear discrimination of different material grades, an
analytical resolution A is defined as follows [cf. relation (14.1)]:
ˇ ˇ
Ian ˇ Ian ˇ
Iref ˇhigh
Iref ˇlow
AD s (17.1)
ˇ ˇ
Ian ˇ Ian ˇ
s2 Iref ˇhigh C s2 Iref ˇlow
where Ian , Iref are the intensities of the analyte line and the reference line; s is
the empirical standard deviation. The indices “high” and “low” refer to a sample
containing the respective analyte at high and low concentration, respectively. In
case of A 1, the chosen analyte line and internal standardization represents a
significant spectral measure for the difference of analyte concentrations, whereas
for A 1 no clear discrimination is possible. For the analyte lines of Mo, Ni,
and Ti, 8–10 different reference lines were studied using the analytical resolution
defined in (17.1). To calculate A, five different combinations of material grades are
taken with the restriction, that the “high” samples always have approximately the
same analyte concentration and the internal standard has the same concentration
for the “high” and the “low” sample. Within an operational period of LIFT of 12
months, 50,000 representative data sets were selected from the steel grades 304L,
Table 17.6 Average analytical resolution hAi and standard deviation A for the molybdenum
line 281.62 nm for different internal standards stated in the first column
Internal standard (nm) hAi A
Fe II 273.07 5.11 0.23
Cr II 298.92 5.03 0.23
Cr II 267.72 4.60 0.65
Nb II 319.50 3.86 0.23
W I 400.87 2.86 0.29
Fe I 371.99 2.71 0.51
Ni I 351.50 1.90 0.22
Ti II 337.28 1.83 1.07
Al I 396.15 1.79 0.78
Cu I 324.75 1.33 0.38
316L, Duplex, 1.4541, and 1.4571. Each data set corresponds to one inspected pipe
fitting. In the considered period of 12 months, the laser power dropped by up to 65%.
Table 17.6 lists the determined average analytical resolution hAi for the five material
combinations and the standard deviation A. Depending on the chosen reference
line, the differences in the analytical resolution vary by more than a factor of three.
The highest values of hAi are attained for reference lines having an upper energy
level close to the one of the analyte line. Those intensity ratios are less sensitive to
changes of the induced plasma state as caused, e.g., by long-term variations of the
laser pulse energy or the focusing conditions [cf. Sect. 11.1, (11.3)].
The average throughput achieved by the LIFT systems amounts to 300 specimens
per hour, which is an increase by a factor of 5 in comparison to the former applied
inspection by spark discharge OES. A measure of the reliability of the material
identification is the percentage of workpieces, which are not the correct material
grade but are classified as being okay. To study this case, specimens of the wrong
material are let into the lot of pipe fittings to be inspected in regular time intervals.
Since the start of the operation, every wrong steel grade introduced intentionally
was identified correctly. The second possible type of error refers to the case that
the material inspected is okay but it is identified to be not okay. This means the
specimen is rejected although it is okay. This case occurs in less than 0.1% of all
inspected specimens.
For the first time, more than 1.4 million inspections corresponding to 108 LIBS
measurements have become available giving an insight into the long-term behavior
of the system performance. Figure 17.16 shows the average laser pulse energy of 1
day and the intensity of the Fe I 371.99 nm line of a pure iron sample as a function of
time. The pure iron sample (99.99 m.-% iron) is used for monitoring purposes. This
sample is measured nearly every day to check the adjustment of the spectrometer
(so-called profiling of the spectrometer). The laser pulse energy is measured for
each inspected pipe fitting. The data points shown in Fig. 17.16 for the pulse energy
are averages over an operation period of 1 day. The data reveal strong variations of
the laser pulse energy and the spectral line intensity over the 5 years of operation.
In a first period (denoted with “I” in Fig. 17.16) the laser pulse energy is rather
average laser pulse energy EL [a.u.]

2500 A B I Fe 371.99 600
EL
[a.u.]
500
2000
Fe 371.99
400
1500
300
intensity I
1000
200
500 100
Ι ΙΙ ΙΙΙ
0 0
1999 2000 2001 2002 2003 2004
time
Fig. 17.16 Intensity of the Fe I 371.99 nm line of a pure iron sample and laser pulse energy as a
function of time. A, new adjustment of laser; B, new laser crystal
low, which is attributed to a misalignment of an oscillator mirror of the laser head.

A new adjustment of the laser head in the second half of the year 1999 (marked with
“A” in Fig. 17.16) has led to a significant increase of the laser energy and of the line
emission. In the following 2 years, the laser pulse energy drops continuously leading
to a corresponding decrease of the line intensity (period II). After installation of a
new laser crystal (marked with “B” in Fig. 17.16), the laser pulse energy rises again
followed by a decay over 2 years (period III). The strong variation of the laser pulse
energy is attributed to various degradation processes of the laser system (flash lamp,
pump cavity, laser crystal, optical components, adjustment) occurring during the
observed operation time. Figure 17.16 shows that the ratio of the iron intensity to
the laser pulse energy is successively decreasing over 5 years. Comparing, e.g., the
last quarter of phase III with the last quarter of phase II shows that the same level
of iron intensity is achieved with a significantly higher pulse energy. This may be
attributed to a gradual decrease of the transmittance of the beam path between the
exit of the laser source and the specimen (cf. Fig. 17.15) and/or a gradual reduction
of the spectral transmittance of the beam path of the measuring radiation from
the specimen to the detectors of the spectrometer. From the available experimental
data, it is not possible to decide which is the dominant underlying process. Further
investigations are necessary to clarify in detail the reasons for this behavior.
Figure 17.17 shows the intensity of the iron line Fe I 371.99 nm as a function of
the laser pulse energy using the same data set as in Fig. 17.16. The general trend is
that higher laser pulse energies lead to higher line intensities of the iron line. Three
clusters can be identified indicating different states of the measurement system in
the three periods I–III (cf. ellipses marked with I–III in Fig. 17.17). Actions at the
laser system as, e.g., readjustment or change of the laser crystal have changed
the response characteristics of the measurement system. The laser pulse energy
measured close to the laser system (cf. Fig. 17.15) is obviously not a sufficient
monitoring information. The influence of the changing laser pulse energy on the
2500 ΙΙ
intensity I Fe 371.99 [a.u.]
2000
Ι
1500
ΙΙΙ
1000
500
0
0 100 200 300 400 500 600
average laser pulse energy EL [a.u.]
Fig. 17.17 Line intensity of the Fe I 371.99 nm line as a function of the laser pulse energy (same
data set as in Fig. 17.16)
line intensity ratios used to identify the steel grades is less pronounced with relative
variation of about 50%. However, such variations can only be handled by periodic
recalibration of the measuring system to assure the correct identification of the steel
grades to be inspected (cf. Sect. 11.4).
An approach for further improvement is to monitor additionally the laser pulse
energy close to the interaction region of the laser beam with the specimen. By this,
the transmittance of the laser pulse energy from the laser to the specimen can be
assessed to recognize, e.g., misalignments or contaminations of the optical path
between laser and specimen. Furthermore, monitor samples should be measured
in regular time intervals to survey the long-term behavior of the line intensity of
several emission lines having different excitation energies to detect changes of the
plasma state or changes of the spectral characteristics of the receiving optics (e.g.,
windows, fiber optics), spectrometer and detectors (e.g., photocathodes, CCD).
The triangulation sensor measures the sample position for each inspected pipe
fitting, cf. Fig. 17.15. Figure 17.18 shows the measured position d averaged for
each day of operation over a period of 5 years. In principle, this distance should
be constant, since it is defined by the applied fixture for the pipe fittings installed
on the circular-type conveyor. Figure 17.18 shows that most of the specimens are
clustered around a sample position of 232 mm.
Whereas the standard deviation for one data point – shown as error bars in
Fig. 17.18 – is of the order of 1–2 mm, the overall spread of positions is up to
20 mm. This is largely due to the adjustments of the fixtures for the specimens on the
circular conveyor. For different pipe-fitting geometries to be inspected, the operator
has to change and to mount the respective fixtures for the pipe fitting on the circular
conveyor (depending on the size and diameter of the pipe fittings). Because these
actions were done manually, it is possible that the wrong fixture is taken or the
17.3 Mobile Systems 447
260
250
sample position d [mm]
240
230
220
210
200
1999 2000 2001 2002 2003 2004
time
Fig. 17.18 Sample position d as a function of time
correct fixture is not adjusted sufficiently. In this case, different sample positions
occur. Potential solutions for this problem is the use of an automated positioning
system for the specimens and/or an autofocus system for the laser focusing.
The temperature measured inside the cabinet of LIFT and the position of the
entrance slit of the spectrometer after the profiling is shown for a time of 5 years in
Fig. 17.19. The figure shows clearly that the temperature correlates with the position
of the entrance slit. Changing temperatures inside the cabinet have an influence
on the mechanical structure of the spectrometer. They cause slight changes of the
spatial positions of entrance slit and exit slits. By the profiling action, the optimum
entrance slit position is readjusted. Figure 17.19 shows as well that the installation
of a new air condition for the cabinet has led to a shift of the temperature level with a
corresponding change in the position of the entrance slit. Outliers are attributed to an
opening of the door of the cabinet of LIFT, which is necessary for maintenance and
adjustment actions. The scattering of entrance slit positions can be further reduced
by installation of a temperature stabilization for the spectrometer.
17.3 Mobile Systems
Mobile systems are defined here as LIBS instruments designed for use at different
locations in an industrial environment. The equipment can be easily transported
to different places of use. To this category of LIBS systems belong the so-called
portable systems that can be ported by a user to the place of application [17.9,17.10,
17.11, 17.12, 17.13, 17.14, 17.15]. There are several companies offering portable
LIBS systems [17.16]. The requirements on such systems can be summarized as
26
average cabinet temperature T C [°C]

position of entrance slit x ES [a.u.] 950 TC
C xES
900 24
850
22
800
750 20
700
18
650
1999 2000 2001 2002 2003 2004
time
Fig. 17.19 Position of the entrance slit of the spectrometer and the temperature inside the cabinet
as a function of time. C, installation of a new air condition
Fig. 17.20 Analysis of

metallic samples with the
compact measuring system
FML
follows: compact design, capability to work under harsh industrial conditions, long-
term stability and reliability, user-friendly operation and maintenance.
As an example of a mobile system designed for industrial use the instrument
FML (f lexible measuring system based on LIBS) will be described [17.17].
Figure 17.20 shows a photograph of FML. The spectrometer that is integrated into
FML is a Paschen–Runge system with a focal length of 150 mm. It is equipped with
a holographic grating and three CCD detectors that allow to measure the whole
spectral range from 278 to 560 nm simultaneously with a resolution of approx.
45 pm. The signal electronic of the spectrometer is modified in such a way that
it can be used to trigger the laser system and allows to record single pulse spectra.
A compact and robust Nd:YAG laser is used for the setup of FML. This laser is
a flashlamp excited, Q-switched solid state laser system delivering laser pulses with
< 8 ns pulse duration and a maximum pulse energy of 50 mJ at a repetition rate of
up to 20 Hz. The weight of the laser head itself amounts to only about 1 kg, which
makes it possible to integrate the laser directly into the measuring probe.
FML is equipped with a handheld measuring pistol, which is connected to the
spectrometer and the laser power supply via a 2-m long umbilical. The plasma
radiation that is generated by the laser pulses is guided to the spectrometer through
a fused silica fiber within this umbilical. Additionally, argon can be delivered to the
measuring probe in order to purge the measuring volume. The flexible connection
between the probe and the other components of the instrument allows for a versatile
use of the system even for measuring objects which are not easily accessible.
Typical measuring times of the FML range from less than a second to several
10 s, depending on the desired accuracy and precision of the measurement. For the
investigations described in the following, FML was operated in a mode in which for
each laser shot, the whole spectrum of the first CCD detector – ranging from 278 to
384 nm – was recorded.
One potential application for a portable measuring system is the sort identifica-
tion of workpieces, half-finished products or of individual scrap particles (cf. also
Sects. 13.3, 18.1, and 18.4). Metallic materials can also be analyzed by spark
emission spectrometry, provided that the sample surface is suitably prepared. As the
laser is capable to penetrate surface coatings, the sample preparation and the analysis
can both be performed with a laser emission spectrometry system (cf. Sects. 13.1.2
and 17.1). In this case, acceptable measuring times and analytical results are only
achieved if the surface layers are suitably thin and if the sample composition close
to the surface is representative for the bulk.
In order to evaluate the capabilities of FML for such an application, calibration
curves were recorded for alloyed aluminum samples. For aluminum, the distinction
between cast aluminum and the different types of wrought aluminum is a precondi-
tion for a high-quality recycling of scrap particles.
Figure 17.21 shows an excerpt of the LIBS spectra that were measured with FML
for a pure aluminum sample and for an alloyed aluminum sample containing a small
amount of copper. From the comparison of the two spectra, the spectral lines of Cu at
324.8 and 327.4 nm can be clearly identified. The spectral background, given by the
signal intensity at a spectral position at which no element-specific line emission can
be observed, differs for the two samples. The reason for this different background
was not yet investigated in detail.
However, it is important to correct the line intensities that should be correlated
to the analyte concentrations for the differing background. Therefore, a spectral
background position was defined for each spectral line, and the signal intensity at
this background position is subtracted from the signal intensity at the peak position
of the spectral line in order to calculate a background-corrected line intensity.
For the calibration, the ratio of this line intensity and the line intensity of a
spectral line of the main matrix element is used (cf. Sect.11.1). In this case, the
aluminum line at 309.4 nm was selected for internal standardization.
28000 c(Cu) = 0.1043 m.-%

Cu 324.8 nm
c(Cu) < 0.0000 m.-%
24000
background
intensity [cts]
position Cu 327.4 nm
20000
16000
12000
320 321 322 323 324 325 326 327 328 329 330
wavelength [nm]
Fig. 17.21 LIBS spectra of a pure and an alloyed aluminum sample measured with FML
0.4 0.3
relative intensity [cts / cts]
relative intensity [cts / cts]
0.3
0.2
0.2 analyte line Cu 324.8 nm analyte line Si 288.2 nm

reference line Al 309.3 nm reference line Al 309.3 nm
0.1
pulse energy 50 mJ pulse energy 50 mJ
0.1 Npp 1
Npp 1
Nmp 99 Nmp 99
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0 2 4 6 8 10 12 14
relative analyte content [(m.-% / m.-%)*100 %] relative analyte content [(m.-% / m.-%)*100 %]
Fig. 17.22 Calibration curves for the alloying elements copper (left) and silicon (right). Npp ,
number of prepulses; Nmp , number of measuring pulses
In Fig. 17.22, calibration curves for the alloying elements copper and silicon in
aluminum are shown. The intensity ratio is plotted versus the analyte concentration
ratio. This method is used because the range of analyte concentrations partly exceeds
10 m.-%. In consequence, this means that the concentration of the matrix element
aluminum is significantly reduced and, thus, the measured intensity of the spectral
line used as the internal standard decreases as well.
The calibration curves shown in Fig. 17.22 show a saturation behavior. This
means that for the high concentration range, it might be better to use other, less-
sensitive lines for the calibration. The measured data is fitted by an analytical
function, and it was found that the saturation behavior can be described well by
a power function of the type Qi;r D a C bc˛rel with the exponent ˛ in the interval
0 < ˛ 1.
Additionally, a linear function was fitted to the data of the four samples with low
analyte concentrations. The slope of this function gives a reasonable approximation
of the slope of the calibration curve for small concentrations and was used to
Table 17.7 Analytical figures of merit for calibration curves of aluminum alloying elements
measured with FML in the wavelength range from 278 to 384 nm
Element Spectral line (nm) LOD3 s (%) r2
Cu 324.8 0.013 0.9962
Si 288.2 0.015 0.9975
Mg 279.6 0.001 0.9943
Mg 383.8 0.057 0.9921
Fe 373.8 0.024 0.9982
Mn 294.9 0.019 0.9827
Zn 334.5 0.040 0.9859
Zr 357.6 0.005 0.9876
Ti 334.9 0.003 0.9947
calculate the limit of detection LOD3 s according to the 3 s-criterion, where s is

the empirical standard deviation of the signals of five measurements with the pure
aluminum sample performed under repeatability conditions (cf. Sect. 11.2).
Table 17.7 summarizes the LOD and the coefficient of determination r 2 of nine
spectral lines that were evaluated for the determination of the eight most important
alloying elements in aluminum. The data is based upon five measurements of 10
samples of certified reference material (CRM), with prepulse and 100 measuring
pulses at one sample position for each of the measurements.
The average limit of detection in terms of an analyte concentration ratio amounts
to 0:020 .m:-%=m:-% 100%/ with an average coefficient of determination of
0.9921. These mean values give an impression of the system performance.
For many applications, especially for sort identification or even sorting of scrap
particles, the rapid availability of the measurement result is essential. For LIBS, this
measuring time is directly connected to the number of laser pulses that has to be
applied to the measuring object in order to achieve a significant result.
In order to investigate this influence of the number of laser pulses, different
numbers of single shot spectra were accumulated numerically and the standard
deviation of the signal intensity ratios was calculated. Firstly, this calculation was
performed for all laser pulses applied to the sample from the first one on, but
secondly it was also performed for the spectra of all laser pulses but the first.
The latter case practically means that the first laser pulse is regarded as a surface-
cleaning prepulse and that its spectrum is discarded. The mean RSD averaged over
the 10 CRM samples and the nine investigated spectral lines is shown in Fig. 17.23
as a function of the number of measuring pulses and shows – as expected – the more
measuring pulses are used, the smaller the RSD is.
Most notably, the RSD is improved by more than a factor of 10, if a single pre-
pulse (Npp D 1) is used and up to ten measuring pulses are evaluated. Furthermore,
the improvement of the RSD by using more than ten measuring pulses is almost
negligible if a single prepulse is taken into account. Due to these results, a single
prepulse was always used for the data evaluation for the calibration curves that had
been shown above.
1000
mean relative standard deviaiton Npp = 0
Npp = 1
of intensity ratio [%]
100
10
1
1 10 100
number of laser pulses per measurement Nmp
Fig. 17.23 Influence of the number of measuring pulses Nmp on the RSD of the intensity ratio.
Pulse energy: 50 mJ; number of samples: 10; measurements per sample: 5; evaluated spectral
lines: 9
In order to validate the accuracy of measurements of aluminum pieces based

on the measured calibration curves, a number of aluminum-based test workpieces
was measured and their analyte concentrations were calculated from the measured
spectra and analysis functions.
Exemplarily, the result of these measurements is shown in Fig. 17.24 for copper.
In this diagram, the analyte concentration ratio of copper to aluminum determined
by LIBS is plotted versus the analyte concentration ratio of the reference measure-
ment. Whereas the error of the measurements with FML is also shown, it has to be
noted that the error of the reference analysis is unknown.
The calibration that was used for the calculation is based upon 100 measuring
pulses per sample position. Only ten . D Npp C Nmp / laser pulses were applied to
the test workpieces, and the analyte concentration was calculated for different
numbers of numerically accumulated spectra. Figure 17.24 shows the result for a
single prepulse and nine measuring pulses and it can be seen that the analyte content
determined with FML matches the reference analyte content within the experimental
errors. The accuracy, calculated as the mean of the deviation of the measured analyte
content from the reference analyte content, amounts to 0:226 .m:-%=m:-%100%/.
If three measuring pulses are evaluated, the measured Cu content matches the ref-
erence content negligibly worse, with an average accuracy of 0:244.m:-%=m:-%
100%/. Even if only two laser pulses – one prepulse and one measuring pulse – are
applied to the sample at five different positions on the sample surface, the accuracy
increases only slightly to 0:338.m:-%=m:-% 100%/. These values show that,
depending on the specific demands of the application, the FML system can be used
for a rapid assessment of alloyed aluminum measuring objects.
analyte concentration ratio determined 8 Npp = 1

by LIBS [(m.-% / m.-%)*100 %] 7 Nmp = 9
6
5
4
3
analyte line Cu 324.8 nm
2
reference line Al 309.4 nm
1 pulse energy 50 mJ
0
0 1 2 3 4 5 6 7 8 9
analyte concentration ratio of reference
measurement [(m.-% / m.-%)*100 %]
Fig. 17.24 Analyte concentration ratio determined by LIBS versus the analyte concentration ratio
of the reference measurement for copper in aluminum
Another potential application for a LIBS measuring system with a flexible and
compact measuring probe is the rapid depth profile analysis of coatings on sheet
products. Due to the demands of the customers of sheet metal manufacturers, these
surface-refined products represent an increasing fraction of the sheet product mar-
ket. On the other hand, the production and the processing of these coated products
requires fast measuring systems to monitor the surface coating characteristics for
process control and quality assurance purposes.
Due to the restricted pulse energy of the laser system, the investigations are based
on repetitive measurements at one position of a metallic sample. In a first step, a hot-
dip zinc-coating with a thickness of 20 m was analyzed, and the single-shot spectra
for 200 laser pulses of 50 mJ pulse energy were recorded.
Figure 17.25a shows the signal intensity that was measured for the Fe line
373.8 nm and the Zn line 307.2 nm as a function of the laser pulse number. Even
from the raw intensities, the Zn/Fe-transition can be clearly identified.
In order to determine the penetration depth per laser pulse, the derivative of the
signal intensity was calculated numerically. This first derivative, which is depicted
in Fig. 17.25b, shows a maximum for the Fe line and a minimum for the Zn line at
approximately 50 laser pulses.
From the zinc-coating thickness of 20 m, which had been measured using glow-
discharge optical emission spectrometry (GD-OES), the average ablation rate for Zn
can be estimated as 0:4 m/pulse.
Assuming that the Zn/Fe-transition of the hot-dip coated sheet metal is very sharp
with respect to the thickness of the coating, the width of the maximum of the first
derivative in Fig. 17.25b can be taken as a measure for the depth resolution that
was achieved with these measuring parameters. In this case, the restricted depth
a b 8
140 7
Zn 307.2 nm
6
intensity [cts / pulse]

120
Fe 373.8 nm
derivative of raw
Fe 373.8 nm 5
intensity [cts]
100 4
80 3
2
60
1
40 Zn 307.2 nm
0
20 -1
-2
0 -3
0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200
pulse number pulse number
Fig. 17.25 Investigation of hot-dip zinc-coated sheet metal with FML. (a) signal intensity of zinc
and iron, and (b) change of signal intensity of zinc and iron versus the number of laser pulses
resolution can be attributed to the laser beam parameters. Especially, the beam
profile and the pulse energy result in cone-shaped craters on the sample surface,
which in turn means that with each laser pulse, sample material from different depth
regions contribute to the LIBS signal (cf. Sects. 7.1, 16.1).
The mobile system FML is a first step toward compact-measuring probes that
can be integrated into automated measuring systems for industrial applications. The
flexible coupling of a measuring probe to a spectrometer might be extended to a
multiprobe for parallel inline inspection tasks at moving products.
17.4 Remote Systems
Several laser spectroscopic methods and systems are described in the literature for
remote analysis in terms of measurements over distances of several meters and
kilometers [17.18]. The applied methods are laser fluorometers, LIDAR, resonance
Raman spectroscopy, and LIBS. One of the first remote LIBS systems was studied
for space exploration [17.19]. A Galilean telescope was used to focus the laser beam
onto soil samples at a distance of 19 m in a low pressure CO2 atmosphere. Stainless
steel samples were measured at distances of up to 45 m showing the capability
to identify steel grades [17.20]. A rover-based LIBS system was described for
mineralogical and elemental identification of rock and soil targets at distances of
2–6 m [17.21]. Aluminum and steel samples at high temperatures were studied using
a Newton telescope of 200 mm aperture to collect the plasma emission [17.22].
A remote LIBS system was tested for real-time monitoring of high-temperature
corrosion in stainless steels at temperatures up to 1; 200ıC [17.23]. A coaxial optical
design was used for open-path remote LIBS measurements up to distances of 300 m
[17.24]. However, no quantitative analytical data were presented. Laboratory-scale
melts of stainless steel were measured over a distance of 7.5 m with a Czerny–
Turner spectrometer having a spectral range of 30 nm and an estimated resolution of
about 160 pm [17.25]. Analyte signal levels as a function of temperature increase by
17.4 Remote Systems 455
factors of 5–10 for temperatures up to 1; 420ıC. LOD were determined for Cr and
Ni: 1,190 and 540 g=g. Compared to measurements performed at smaller distances
in an inert gas atmosphere as described in Sect. 13.1.3 (cf. Table 13.8), these values
are worse by more than an order of magnitude, which can be attributed to the
longer measuring distance and the less-defined measuring conditions as caused, e.g.,
by the oxidizing atmosphere. A remote LIBS system was described for elemental
analysis of samples of environmental interest such as bark, leaves, soil rock, and
stones [17.26]. The measuring distance was 12 m. A LIBS system operating with
a frequency tripled Nd:YAG laser and measuring distances of 60 m was tested
for remote imaging and remote cultural heritage ablative cleaning [17.27]. The
primary mirror of the receiving Newton telescope amounts 400 mm, the spectral
range detected covers 360–800 nm with a resolution of 2.2 nm. Steel slag samples
at temperatures of 850ı C were studied with a remote LIBS system at distances
of 12 m demonstrating the capability to obtain basicity values in the range 0.9–2
with good correlation to XRF-based basicity indices [17.28]. Detection of explosive
residues at distances of 45 m were studied with a remote LIBS system equipped
with a Herschelian telescope [17.29].
In the following, the remote LIBS system TeleLis will be described offering a
broad spectral range of more than 340 nm with a high spectral resolution of 25 pm
and single as well as double pulse operation [17.30, 17.31]. The system design is
tailored to industrial applications in the metal industry allowing for quantitative
measurements under harsh conditions.
Figure 17.26 shows the optical setup. TeleLis comprises a pulsed Nd:YAG laser
source with up to 20 Hz repetition rate and a pulse energy of up to 300 mJ. The laser
system has double pulse capability with a freely choosable interpulse separation
between 2 and 200 s. The laser beam diameter is enlarged to approx. 80 mm by
a Galilean telescope before being focused by an achromatic lens onto the sample.
The focusing system is designed for measuring distances between 3 and 12 m. An
integrated range finder – LD in Fig. 17.26 – allows to measure the distance to the
sample and to adjust the focus automatically onto the sample. The plasma radiation
is collected by a Newton telescope with a main mirror diameter of 300 mm. The
plasma is imaged onto a fiber optic guiding the measuring light to the spectrometer
entrance slit. The spectrometer is a multi-CCD Paschen–Runge system with 16 CCD
detectors. It covers a wavelength range of 176–520 nm with an average spectral
resolution of approximately 25 pm. The design of the laser beam guidance, the
plasma radiation collection system, and the range finder optics is collinear and freely
turnable within an angular range of ˙45ı vertically and ˙90ı horizontally. This
allows to target a new sample within seconds without a readjustment of the optics.
TeleLis is designed as a transportable device. Figure 17.27, left shows the system
ready for measurement. The laser source is mounted on the base platform together
with the beam-shaping optics. The beam is then guided within the vertical arm and
coupled into the telescope. The Paschen–Runge spectrometer is mounted on top of
the vertical arm just above the telescope allowing for short fiber optic connection.
Fig. 17.26 Optical setup of the remote LIBS system TeleLis. L D laser source, GT D Galilean
telescope, M1–M5 D plane mirrors, NT D Newton telescope, M6 D primary mirror of NT,
T D measuring object, s D distance to T, LD D laser distance measurement, FO D fiber optics,
SP D spectrometer
Fig. 17.27 Left: Remote LIBS system TeleLis with the optics in the position ready for measure-
ment. Right: TeleLis system ready for transport
For transport purposes, the vertical arm with telescope and spectrometer can be
folded into the transport container, which is mounted on wheels, see Fig. 17.27,
right. It is not necessary to disassemble any components to transport the system.
After arriving at the new measuring position, it is sufficient to unfold the optics, to
repower and to reinitialize the system to be ready for measurement.
The spectrometer performance was tested by guiding the light of a mercury
lamp by a fiber optic to the spectrometer. Figures 17.28 and 17.29 show the Hg
I line at 289.359 nm and the Hg II line at 194.230 nm, respectively. The spectral
resolution is determined to be 25 pm in both spectral regions. This resolution is a
15000
Hg I 289.360 nm
10000
intensity [cts]
25 pm
5000
0
289.0 289.4 289.8 290.2
wavelength [nm]
Fig. 17.28 Hg I line of a mercury lamp at 289.360 nm
15000
10000 Hg II 194.23 nm
intensity [cts]
25 pm
5000
0
193.4 193.8 194.2 194.6 195.0
wavelength [nm]
Fig. 17.29 Hg II line of a mercury lamp at 194.230 nm
necessary prerequisite for quantitative measurements of metal samples, since those

samples show very line rich spectra. As an example, Figs. 17.30 and 17.31 show
LIBS spectra of a steel and an aluminum sample, respectively, taken at a measuring
distance of 5 m. The laser is running in double pulse mode with a pulse separation
of t D 3 s and a burst energy of 100 mJ. The pulse energy ratio amounts to
E1 W E2 D 1 W 5. The spectra shown are gained by averaging of 50 pulses.
Figure 17.30 shows well separated closely spaced iron lines allowing to reduce
interferences of analyte lines to a large extent. The spectrometer is capable to resolve
self-reversal features of resonance lines as shown exemplarily in Fig. 17.31 for two
aluminum resonance lines. The spectral resolution achieved is comparable to the
one of a high-performance echelle system, cf. Sect. 10.3, Fig. 10.6.
Fe I 373.486
7 × 105 Fe I 376.554
Fe I 376.719
intensity [cts]
5 × 105
2 × 105
0
372 374 376 378
wavelength [nm]
Fig. 17.30 LIBS spectrum of a steel sample, taken at a measuring distance of 5 m. For measuring
parameters see text
60000
Al I 394.40 nm Al I 396.15 nm
intensity [cts]
40000
20000
0
394.0 394.5 395.0 395.5 396.0 396.5
wavelength [nm]
Fig. 17.31 LIBS spectrum of an aluminum sample, taken at a measuring distance of 5 m. The
laser parameters are the same as those used for Fig. 17.30
The effect of double pulse mode versus single pulse excitation for remote
measurements was studied for a set of analytes of the high-alloy steel sample NCS
HS 23702–1 (the composition of this sample is given in Sect. 13.2, Table 13.9).
Figure 17.32 shows the intensities and standard deviations of analyte lines of 12
species on a logarithmic scale. All species are measured simultaneously at a distance
of 12 m. The laser is running in single and double pulse mode, respectively. The
settings for double pulses are the same as stated above. The pulse energy is 190 mJ in
single pulse mode. For all 13 species, the double pulse mode gives higher intensities
than the single pulse mode, although the total burst energy of the double pulse is
106
intensity, standard deviation [cts] double pulses
single pulses
105
104
103
Mn 293.306
Al 305.99
Si 251.6
C 193.091
W 400.80
Al 396.152
Mo 386.40
Ni 341.50
Ti 337.28
Cu 324.754
Si 288.16
Mo 281.615
Cr 267.716
element line [nm]
Fig. 17.32 Intensity of 12 species in single and double pulse operation of the laser source,
measured simultaneously at a distance of 12 m of the high-alloy steel sample NCS HS 23702–1
about half of that used for the single pulse. At the same time, the standard deviations
are significantly reduced in double pulse mode.
A measure of the capability to discriminate different analyte
q concentrations in a
specimen is the analytical resolving power A D .I1 I2 /= s12 C s22 [cf. definition
in Sect. 14.1.2, relation (14.1)], where I1 is the intensity of one particular spectral
line of sample 1, I2 the intensity of the same spectral line of sample 2, s1 and s2 are
the corresponding standard deviations. Figure 17.33 shows the analytical resolving
power for single pulse mode and three different double pulse modes. The double
pulse modes have the same interpulse separation of t D 5 s, whereas the pulse
ratios E1 W E2 (cf. Sect. 3.2, Fig. 3.9) are varied from 5:1, 1:1, and 1:5. The burst
energy is changing slightly with different pulse ratios between Eb D 75 and 100 mJ.
The sample distance is 5 m. The analytical resolving power is calculated from the
intensity of the Si 288.2 nm line measuring two aluminum samples with different Si-
content: c1 (Si) D 25:5 m.-%, c2 (Si) D 1:57 m.-%, respectively. Fifty measurement
bursts are averaged (Nmp D 50) after five prebursts (Npp D 5). In single pulse
mode, the analytical resolving power reaches ASP D 7:3. In double pulse mode,
the analytical resolving power increases with increasing pulse energy of the second
pulse in the burst, starting at ADP .5 W 1/ D 2:3 for a pulse ratio of 5:1, via ADP .1 W
1/ D 5:4 reaching ADP .1 W 5/ D 19:2 which is about 2.5 times higher than the
corresponding value for single pulse mode. These investigations clearly show that
the double pulse excitation improves the analytical performance not only at focusing
20
analytical resolving power A [1]
16
12
0
5:1 1:1 1:5
single pulse double pulse, pulse ratio E1 : E2 [1]
Fig. 17.33 Analytical resolving power for single pulses and three double pulse modes with
different pulse ratios, measured at a distance of 5 m; t D 5 s
24
analytical resolving power A [1]
20
16
12
30 5 3
single pulse double pulse, interpulse separation Δt [μs]
Fig. 17.34 Analytical resolving power at a measuring distance of 5 m for single and three double
pulse modes with different interpulse separations. E1 W E2 D 1 W 5
distances of a few centimeters (cf. Sects. 6.1, 7.2, and 8.1) but also at distances that
are orders of magnitude larger.
Figure 17.34 shows the analytical resolving power A for three different interpulse
separations t of 30, 5, and 3 s at a constant pulse ratio of 1:5. The total burst
energy varies for the three double pulse modes in the range Eb D 100–140 mJ.
The pulse energy for the single pulse amounts to 190 mJ. Compared to the results
shown in Fig. 17.33, the analytical resolving power A is increased to 23.8 when the
interpulse separation t is reduced to 3 s.
14
Mg
12
10
r 2 = 0.999
c [m.-%]
0
0 5 10 15 20
I Mg / I Al [1]
Fig. 17.35 Analysis curve for magnesium in aluminum matrix. Measuring distance 5 m, double
pulse mode, interpulse separation t D 3 s, burst energy Eb D 100 mJ, pulse ratio 1:5, Npp D 5,
Nmp D 50
4 Ni
r 2 = 0.996
c [m.-%]
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
I Ni / I Fe [1]
Fig. 17.36 Analysis curve for nickel in iron matrix. Measuring parameters are those given in
Fig. 17.35
To study the capability for a remote quantitative analysis of analyte concen-

trations, analysis functions were determined using a set of aluminum and steel
reference samples.
Figures 17.35 and 17.36 exemplify analysis curves for Mg in aluminum matrix
and for Ni in iron matrix.
Table 17.8 summarizes the results obtained for the investigated analytes. Coeffi-
cients of determination greater than 0.99 were found for all analytes except for Mn
Table 17.8 Coefficients of determination and limits of detection (LOD) for different analytes in
aluminum and iron matrix measured with remote LIBS system TeleLis at a measuring distance
of 5 m
Matrix Analyte Concentration r2 LOD (m.-%)
range (m.-%)
Al Cu 0–14 0.996 0.13
Mg 0–20 0.999 0.12
Mn 0–2.5 0.984
Ni 0–4 0.996 0.07
Si 0–7 0.996
Zn 0–1.8 0.998
Fe Cr 0–6 0.994 0.19
Ni 0–3 0.996 0.16
Mn 0–8 0.990 0.15
in aluminum. As expected, LOD are worse than those found for smaller measuring
distances, see, e.g., Sect. 13.3, Table 13.16 for aluminum and Sect. 13.1, Tables 13.2
and 13.6 for steel. However, for the remote measurement no defined inert gas
atmosphere can be applied as it was the case for the measurements described in
Sect. 13.1. Thus, the influence of the ambient atmosphere along the beam path
of the laser and the measuring radiation (5 m each) is expected to deteriorate
the measurement. Furthermore, the gas exchange close to the interaction volume
(cf. Sect. 4.6) is not controlled in a defined way as it is possible by the use of a
sample stand (cf. Fig. 4.18) or a measuring lance (cf. Fig. 13.12).
The described remote LIBS system TeleLis was studied for quantitative inline
analysis of the composition of liquid pig iron and liquid slag at an iron runner
of a blast furnace [17.30]. The temperature of the melt is 1;500ıC. The temporal
evolution of the silicon, manganese, and carbon concentration was monitored
inline for 40 min and the obtained results were compared to the results gained by
conventional analysis of so-called lollipop samples taken during the test period.
A high degree of agreement was found and the feasibility of liquid pig iron and
liquid slag analysis was proven successfully.
The remote system was also tested to discriminate organic materials that are of
relevance, e.g., as explosive residues [17.31]. Figure 17.37 shows line intensities
of several species for the detection of ammonium nitrate on a substrate consisting
of greased aluminum foil. The measurements were performed using an external
echelle spectrometer (cf. Sect. 4.2, Table 4.4) instead of the built-in spectrometer.
Single pulses with 190 mJ pulse energy were irradiated at a distance of 5 m.
The measurement parameters used are delay time tdelay D 1 s, integration time
tint D 10 s, Nmp D 3 for substrate, Nmp D 15 for NH4 NO3 . Ammonium nitrate can
clearly be detected even at low mass density levels of approx. 0:15 g=mm2 by the
enhanced intensity of the hydrogen and cyanide emission.
References 463
1.2
line intensity relative to Al line [1] substrate: greased Al foil
0.15 μg / mm2 ammonium nitrate, NH4NO3
1.0
0.8
0.6
0.4
0.2
0.0
C N O H CN CN
247.9 nm 742 nm 777 nm 656.3 nm 358 nm 387.1 nm
Fig. 17.37 Intensities of several species for the detection of ammonium nitrate (NH4 NO3 ),
measured with an echelle spectrometer at a distance of 5 m using single pulses with 190 mJ pulse
energy
References
17.1. J. Vrenegor, V. Sturm, D. Eilers, R. Noll, Neuartiges Laserverfahren für die schnelle
Präparation und Multielement-Analyse metallischer Prozessproben, Final Report of the
Joint Research Project ATLAS funded by the Federal Ministry of Economics and Technol-
ogy of Germany within the InnoNet Programm, support code IN4051, 1.1.2004–31.3.2007,
74 p
17.2. R. Noll, V. Sturm, Ü. Aydin, D. Eilers, C. Gehlen, M. Höhne, A. Lamott, J. Makowe,
J. Vrenegor, Laser-induced breakdown spectroscopy – from research to industry, new
frontiers for process control. Spectrochim. Acta Part B 63, 1159–1166 (2008)
17.3. H. Balzer, M. Höhne, V. Sturm, R. Noll, Online coating thickness measurement and depth
Acta B 60, 1172–1178 (2005)
17.4. G. Asimellis, A. Giannoudakos, M. Kompitsas, Rapid, automated measurement of layer
thickness on steel coin blanks using LIBS depth-profiling. Appl. Optics 46, 935–942 (2007)
17.5. D. Papazoglou, V. Papadakis, D. Anglos, In situ interferometric depth and topography
monitoring in LIBS elemental profiling of multi-layer structures. J. Anal. At. Spectrom.
19, 483–488 (2004)
17.6. M. Corsi, G. Cristoforetti, M. Hidalgo, D. Iriarte, S. Legnaioli, V. Palleschi, A. Salvetti,
E. Tognoni, Effect of laser-induced crater depth in laser-induced breakdown spectroscopy
emission features. Appl. Spectrosc. 59, 853–860 (2005)
17.7. Patent, DE 103 61 727, Verfahren und Vorrichtung zur Analyse fester Materialien mit der
Laser-Emissionsspektrometrie, 21.7.2005
17.8. R. Noll, I. Mönch, O. Klein, A. Lamott, Concept and performance of inspection machines
for industrial use based on LIBS. Spectrochim. Acta B 60, 1070–1075 (2005)
222–233 (1996)
17.10. B. Castle, A. Knight, Battery powered laser-induced plasma spectrometer for elemental
determinations. J. Anal. At. Spectrom. 13, 589–595 (1998)
17.11. R. Harmon, F. De Lucia, A. Miziolek, K. McNesby, R. Walters, P. French, Laser-induced
breakdown spectroscopy (LIBS) – an emerging field-portable sensor technology for real-
time, in-situ geochemical and environmental analysis. Geochem. Explor. Environ. Anal. 5,
21–28 (2005)
17.12. W. Pierce, S. Christian, Portable LIBS Instrumentation Can Identify Trace Levels of
Environmental Pollutants, (Photonik International, Fellbach, 2006), 92–94
17.13. J. Goujon, O. Musset, A. Giakoumaki, V. Pinon, D. Anglos, E. Georgiou, A new
compact laser source for portable LIBS applications. Proc. SPIE 6871, 68712Q (2008).
doi:10.1117/12.777953
17.14. A. Taffe, D. Schaurich, G. Wilsch, Development of a portable LIBS-device for quality
assurance in concrete repair, in Concrete Repair, Rehabilitation and Retrofitting ed. by
G. Alexander (Taylor & Francis, London, 2009), pp. 547–549
17.15. M. Myers, J. Myers, J. Sarracino, C. Hardy, B. Guo, S. Christian, J. Myers, F. Roth,
A. Myers, LIBS system with compact fiber spectrometer, head mounted spectra display
and hand held eye-safe erbium glass laser gun, SPIE Photonics West (2010) Solid State
Lasers XIX: Technology and Devices Conference LA101, # 7578–87, 20 p
17.16. StellarNet Inc., FL; Avantes, CO; Ocean Optics, FL; Applied Photonics, UK
17.17. F. Hilbk-Kortenbruck, M. Höhne, R. Noll, M. Freit, J. Joosten, H. Falk, Compact
measuring system for laser emission spectrometry of coated and uncoated metals. Proc. 9.
Anwendertreffen Röntgenfluoreszenz- und Funkenemissionsspektrometrie, 59–69 (2002)
17.18. F. Hoge, Recent advances in laser remote sensing. Proc. SPIE 3707, 2–9 (1999), Laser
Radar Technology and Applications IV, ed. by G. Kamerman, C. Werner
17.19. A. Knight, N. Scherbarth, D. Cremers, M. Ferri, Characterization of laser-induced
breakdown spectroscopy (LIBS) for application to space exploration. Appl. Spectrosc. 54,
331–340 (2000)
17.20. S. Palanco, J. Baena, J. Laserna, Open-path laser-induced plasma spectroscopy for remote
analytical measurements on solid surfaces. Spectrochim. Acta B 57, 591–599 (2002)
17.21. R. Wiens, R. Arvidson, D. Cremers, M. Ferris, J. Blacic, F. Seelos, K. Deal, Combined
remote mineralogical and elemental identification from rovers: field and laboratory tests
using reflectance and laser-induced breakdown spectroscopy. J. Geophys. Res. 107, 8003
(2002). doi:10.1029/2000JE001439
17.22. J. Vadillo, P. Garcia, S. Palanco, D. Romero, J. Baena, J. Laserna, Remote, real-time,
on-line monitoring of high-temperature samples by noninvasive open-path laser plasma
spectrometry. Anal. Bioanal. Chem. 375, 1144–1147 (2003)
17.23. P. Garcı́a, J. Vadillo, J. Laserna, Real-time monitoring of high-temperature corrosion in
stainless steels by open-path laser-induced plasma spectrometry. Appl. Spectrosc. 58,
1347–1352 (2004)
17.24. S. Palanco, J. Laserna, Remote sensing instrument for solid samples based on open-path
atomic emission spectrometry. Rev. Sci. Instr. 75, 2068–2074 (2004)
17.25. S. Palanco, S. Conesa, J. Laserna, Analytical control of liquid steel in an induction melting
furnace using a remote laser-induced plasma spectrometer. J. Anal. At. Spectrom. 19, 462–
467 (2004)
17.26. C. López-Moreno, S. Palanco, J. Laserna, Remote laser-induced plasma spectrometry for
elemental analysis of samples of environmental interest. J. Anal. At. Spectrom. 19, 1479–
1484 (2004)
17.27. R. Grönlund, M. Lundqvist, S. Svanberg, Remote imaging laser-induced breaddown
spectroscopy and remote cultural heritage ablative cleaning. Opt. Lett. 30, 2882–2884
(2005)
17.28. C. López-Moreno, S. Palanco, J. Laserna, Quantitative analysis of samples at high
temperature with remote laser-induced breakdown spectrometry using a room-temperature
calibration plot. Spectrochim. Acta B 60, 1034–1039 (2005)
References 465
17.29. C. López-Moreno, S. Palanco, J. Laserna, F. DeLucia, J. Miziolek, J. Rose, R. Walters,

A. Whitehouse, Test of a stand-off laser-induced breakdown spectroscopy sensor for the
detection of explosive residues on solid surfaces. J. Anal. At. Spectrom. 21, 55–60 (2006)
17.30. G. Mathy, B. Monfort, B. Vanderheyden, V. Tusset, Liquid steel process: advanced on line
sensors under development at CRM. Metall. Anal. 30(Suppl. 1), 6–14 (2010)
17.31. R. Noll, C. Fricke-Begemann, Stand-off detection of surface contaminations with explo-
sives residues using laser-spectroscopic methods, in Stand-off Detection of Suicide
Bombers and Mobile Subjects, ed. by H. Schubert, A. Rimski-Korsakov (Springer,
New York, 2006), pp. 89–99
Chapter 18
Industrial Applications
This chapter describes a selection of industrial applications of LIBS systems

based on methodical approaches and instrumental system designs presented in the
previous chapters.
18.1 Identification Testing of High-Alloy Pipe Fittings
This application deals with the detection of material mix-ups in production lines
of pipe fittings and tubes made of high-alloy steel grades. The motivation and the
system design of the first generation LIBS system – called LIFT – for this task were
described in Sect. 17.2. The second generation inspection system uses a flexible
laser beam steering based on the optical setup shown in Fig. 13.23, where the laser
beam is deflected by an xy-scanner to irradiate various inspection spots inside of a
measuring volume of 400 mm 400 mm 100 mm [18.1]. Figure 18.1 shows the
modules of the LIFT II system.
The spectrometer is linked via a fiber optics with the optic module. The three
modules are designed in such a way that they can be integrated easily into an existing
production line. Figure 18.2 shows the specimens with the laser-induced plasmas.
Each specimen is measured at up to three positions to inspect the steel grades of the
main body of the fitting and of the flanges.
In the production line, the test objects are placed on carriers to assure a defined
orientation of the fittings. These carriers are transferred to the inspection module.
After closing of a protection gate, the laser inspection is initiated. The scanner
directs the laser beam sequentially to the predefined inspection positions which are
transmitted automatically by the host computer where the coordinates of inspection
positions for all types of pipe fittings are stored. Number and position of the
inspection points can be chosen freely inside the measuring volume stated above.
As soon as the scanner has reached the preset coordinates, the measurement starts.
For each measurement position, the time needed amounts to about 1 s, within which
up to 25 elements are detected simultaneously and the concentrations of the alloying

468 18 Industrial Applications
Fig. 18.1 Modules of the second generation inspection system LIFT II: top, left: optic module
with laser source and scanner; bottom, left: spectrometer; right: electrical cabinet
Fig. 18.2 Pipe fittings with

laser-induced plasmas for
mix-up detection
elements are determined automatically. If these concentrations are within the ranges
specified by the manufacturer for all test objects and measurement positions, then
the inspected products on the carrier are cleared for subsequent processing steps and
transported onward. If a mix-up is detected, the test object concerned is identified
and discharged in the next step. All measuring results for each manufactured product
are documented electronically.
The inspection system LIFT II fulfills the requirements for laser class 1 according
to the IEC standard [18.2]. No further protection measures are needed for the user.
18.2 Analysis of Slags 469
number of inspected products 5,000,000
4,000,000
3,000,000
2,000,000
1,000,000
-31 -04 -11 -05 -09 -21 -19 -09 -09 -31 -08
7-10 7-12 8-02 8-04 8-06 8-08 8-10 8-12 9-02 9-03 9-06
200 20
0
20
0
20
0
20
0
20
0
20
0
20
0
20
0
20
0
20
0
date
Fig. 18.3 Number of inspected products as a function of time for a second generation LIFT II
system
Inspection equipment monitoring according to ISO 9001 and recalibrations are

performed automatically. LIFT is equipped with a remote maintenance interface.
The described system is in use in three shift operation, 7 days a week. Figure 18.3
shows the number of inspected products over a period of about 1.5 years. More than
five million products were inspected automatically in routine industrial operation.
Within the shown time period more than 300 different product types and geometries
were measured. The data acquired comprise more than 250 million LIBS measure-
ments.
Figure 18.4 shows the laser power measured during each inspection over the
same time period. The relative standard deviation amounts to 4.86%. A measure
of the stability of the excitation process is the intensity of the zeroth order signal
acquired with a separate detector installed inside the Paschen–Runge spectrometer.
The zeroth order comprises all spectral components of the laser-induced plasma
reflected by the grating. Figure 18.5 shows this signal over a time period of 1.5
years. The relative standard deviation is 6.51%.
The technical availability of the LIFT II system was tested according to estab-
lished standards [18.3]. The availability is 99.94% demonstrating the high degree of
system performance achieved.
18.2 Analysis of Slags
Motivation and LIBS setup for slag analysis were described in Sect. 14.2.2.
Figure 18.6 shows the LIBS instrument with a closed cabinet .1:0 1:8 1:7 m3 /
installed 10 m from the vacuum degasser in the steel works. The instrument is
1,000,000
800,000
laser power [cts]
600,000
400,000
200,000
0
-31 -04 -11 -05 -09 -21 -19 -09 -09 -31 -08
7 -10 7-12 8-02 8-04 8-06 8-08 8-10 8-12 9-02 9-03 9-06
0 0 0 0 0 0 0 0 0 0 0
20 20 20 20 20 20 20 20 20 20 20
date
Fig. 18.4 Laser power over a time period of 1.5 years
400,000
350,000
zeroth order signal [cts]
300,000
250,000
200,000
150,000
100,000
50,000
0
-28 -22 -04 -06 -24 -19 -03 -07 -18
7 -10 7-11 8-02 8-04 8-06 8-09 8-12 9-02 9-04
0 0 0 0 0 0 0 0 0
20 20 20 20 20 20 20 20 20
date
Fig. 18.5 Zeroth order signal over a time period of 1.5 years
24 h/7 d ready for operation. The routine operation is via an operator panel at the
instrument. No computer keyboard or monitor is required at site in the steel works.
For analysis, a button at the operator panel is pushed and the sample is put onto
the sample stand. After closing the lid of the sample stand, the measurement runs
automatically. The result of the analysis is transferred to the host computer (TCP/IP)
of the steel works after 80 s.
18.2 Analysis of Slags 471
Fig. 18.6 LIBS instrument

installed in a steel plant (in
front on the right)
The analysis of unknown samples is carried out by a multivariate data analysis

of the measured LIBS signals with a partial least square regression (PLS1). The
PLS1 calibration model is constructed by the measured intensities of the spectral
lines I for a set of 31 production samples with BBA (beta barium borate) reference
analysis. The calibration ranges given by the variation of the production samples is
w.SiO2 / 0:8–10 cg=g; w.CaO/ 50–56 cg=g, and w.Al2 O3 / 24–41 cg=g.
Afterward the PLS1 calibration model is validated by the measurement of another
set of 32 production samples with XRF reference mass fractions. In Fig. 18.7, the
mass fractions of SiO2 predicted by the multivariate analysis of the measured LIBS
data versus the reference values are given together with the linear fitting curve
[14.49]. The ideal case, i.e., identity of LIBS and reference value, is the bisecting
line in this graph. The root mean square error of prediction, RMSEP, for a species i
is defined as [cf. (11.19) in Chap. 11]:
" N
#1=2
1 X
RMSEP.i / D .wi; ref; s wi; LIBS; s /2 ; (18.1)
N sD1
where wi; LIBS; s is the mass fraction predicted by multivariate analysis of the LIBS
signals for analyte i and sample s, wi; ref; s is the mass fraction determined by the
reference analysis, and N is the number of samples of the validation set. The
RMSEP value is a measure of the average prediction error, expressed in the same
units as the original values.
The corresponding values for CaO=SiO2 are: slope 1.12, r 2 D 0:973, RMSEP
.CaO=SiO2 / D 1:87 and for Al2 O3 =SiO2 : slope 1.06, r 2 D 0:979, RMSEP.Al2 O3 =
SiO2 / D 1:11. The RMSEP values for CaO and Al2 O3 directly are RMSEP.CaO/ D
1:0 cg=g and RMSEP.Al2 O3 / D 1:5 cg=g, respectively. For comparison, RMSEP
values determined by XRF analysis of homogenized vacuum slag sample (homo-
genized by crushing and milling to powder and afterward pressing to a solid sample)
10
mass fraction LIBS w(SiO2) [cg/g]
6 linear fit
slope 0.996
r2 0.979
4
2
RMSEP (SiO2) = 0.20 cg /g
0
0 2 4 6 8 10
mass fraction XRF w (SiO2) [cg /g]
Fig. 18.7 Mass fraction of the species SiO2 of production samples by LIBS analysis versus
reference mass fraction by XRF
amount to RMSEP.CaO/ D 0:85 cg=g, RMSEP.SiO2 / D 0:66 cg=g, and RMSEP

.Al2 O3 / D 1:8 cg=g, respectively.
The ambient temperature at the site of the LIBS instrument in the steel works
varies from 0 to 40ı during the year. The temperature of the instrument cabinet
and especially the temperature of the Paschen–Runge spectrometer is actively
controlled. The resulting stability of the spectrometer temperature is T D .33:1 ˙
0:25/ı C, see Fig. 18.8. The stable operation of the LIBS instrument in the harsh
steel works environment is verified by repeated measurements of a monitor sample
which is a glass-like sample remelted from slag powder of the vacuum degasser, see
Fig. 18.9.
The fast analysis of vacuum slag samples by LIBS was demonstrated success-
fully in the steel works nearby a vacuum degasser station. The LIBS analysis takes
about 80 s and a further reduction is feasible. By spatial averaging of the LIBS
measurement with an optical microlens array, a representative analysis of SiO2 ,
CaO, and Al2 O3 of production samples of the vacuum degasser without any further
preparation was performed in spite of the sample heterogeneity and variations in
color, cracks, and holes.
18.3 Characterization of Inclusions and Segregations
The concept and LIBS setup for high-speed spatially resolved microanalysis were
described in Chap. 15, demonstrating the capability to gain element mappings.
Figure 18.10 shows a photograph of the second generation high-speed scanning
18.3 Characterization of Inclusions and Segregations 473
35
spectrometer temperature [° C]
30
25
20
* power supply from steel

works switched off
15
Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
time
Fig. 18.8 Stability of the spectrometer temperature for a period of 10 months after installation in
the steel works. The temperature is logged every 10 min (24 h/7 d)
1.4
1.2

1.0
0.8
0.6
0.4
RSD = 1.48 %
0.2
0.0
Dec 2 Dec 16 Dec 30 Jan 13 Jan 27
time of measurement
Fig. 18.9 Stability test of the LIBS instrument in the steel works by repeated measurements of a
monitor sample
LIBS system SML (Scanning Microanalysis with Laser Spectrometry) developed

at Fraunhofer ILT, Aachen [18.4]. The technical data of SML is given in Table 18.1.
The SML system is used to characterize inclusions and segregations in steel
samples [18.5, 18.6, 18.7]. The production of high-value steel products, such as thin
wires for energy-saving steel belt tires or thin sheets for lightweight packaging,
requires high-quality steel grades. Inclusions reduce the steel quality by limiting
Fig. 18.10 High-speed scanning LIBS microanalysis system SML to measure steel samples for
the analysis of segregations, inclusions, and decarburization zones
Table 18.1 Data of the high-speed scanning LIBS system SML

Sample positioning accuracy <1 m
Minimal point to point distance 5 m
Spatial resolutiona <20 m
Maximum measuring frequency 1,000 Hz
Number of photomultipliers 41
Number of different detected elements 24
Wavelength range 130 777 nm
Maximum size of scan field 45 mm 110 mm
Measuring atmosphere Argon
a
Refers to the spectroscopic analysis of a metallic sample
the minimum diameter of wires for steel belt tires or the thickness of sheets
for beverage cans. Figure 18.11 illustrates these products and shows electron-
microscopical images of inclusions in steel [18.8].
The development of high-quality steels requires an analyzing method to detect
inclusions like Al2 O3 , aluminum nitride (AlN), TiC, SiC, CaO, ZrO2 , and others
fast, reliably, and cost efficient. State of the art analyzing methods are wet-chemical
analysis, spark optical emission spectrometry, micro-X-ray fluorescence analysis
(-XRF), and scanning electron microscopical energy dispersive X-ray analysis
(SEM-EDX). However, these methods have specific drawbacks. For wet-chemical
Fig. 18.11 Left: high-value steel products like wires for steel belt tires and thin sheets; right:
scanning electron microscopical images of inclusions in steel; top: aluminum nitrite AlN; bottom:
aluminum oxide Al2 O3 (1), titanium nitride TiN (2), and Ti4 C2 S2 (3). The scale given corresponds
to 5 m [18.8]
analysis, the inclusions are extracted chemically or electrolytically and made

accessible to a subsequent analysis. The particular advantage of this approach is the
high confidence level to gather all inclusions. The personnel and chemical costs are
high [18.8,18.9]. SEM-EDX requires an extensive sample preparation with diamond
polishing and measuring times of a few hours. Spark optical emission spectrometry
(SD-OES) has only a limited capability for a spatially resolved analysis due to
the fact that the position of impact of the spark on the sample surface cannot be
determined exactly (cf. Sect. 3.6, Fig. 3.17). Hence, with SD-OES, it is not possible
to gain a spatially resolved analysis of inclusions with resolutions in the micrometer
range. However, a statistical evaluation of single spark signals allows to estimate the
steel cleanness with constraints. Especially, the light elements, oxygen and nitrogen,
are partially difficult to be detected.
The methodical approach for a spatially resolved analysis with LIBS was
described in Sect. 15.1. For each measuring point, only one tightly focused laser
pulse is irradiated onto the sample surface. The advantages of LIBS microanalysis
compared with SD-OES are electrode contamination is excluded, and the interaction
zone of the laser radiation with the sample material is well defined. Compared with
wet-chemical methods, consumables and time are saved. In comparison with SEM-
EDX, elaborate sample preparation and long measuring times are cut down.
Fig. 18.12 Manganese maps of two steel samples (each 10 mm10 mm) of different quality show-
ing significantly different numbers of detected manganese enrichments. Measuring parameters are
those given in Sect. 15.2 for Fig. 15.8
Inclusions in a steel matrix consist of, e.g., carbides, sulfides, nitrides, and oxides
of the elements Al, Ca, Si, Mg, Mn, and Ti. Typical analyte lines for optical emission
spectrometry lie in the wavelength range from 130 nm (O) to 777 nm (O), the highest
excitation energies are 12 eV (e.g., for the nitrogen emission line at 746 nm). Since
oxygen and nitrogen shall be detected, it is necessary to analyze the samples in a
pure inert gas atmosphere (cf. Fig. 15.2).
Element mappings are used to identify inclusion types such as shown for
manganese sulfide (MnS) in Sect. 15.2 and Fig. 15.8. An approach to obtain a
measure of the cleanness of a steel sample is to count the number of measuring
locations where a clear correlation of a set of elements is observed and where the
intensities are greater than a predefined threshold. Figure 18.12 shows manganese
maps of two steel samples of different quality [15.24]. All measuring parameters
were kept constant. The number of measuring points, where the manganese intensity
is greater than the average plus five times the standard deviation, amounts to 968
for the left image and 492 for the right image. An SEM-EDX measurement of
these samples yields approximately the same relation for the cleanliness as the
measurement with scanning LIBS.
Oxides and nitrides are critical inclusions in the steel matrix since they have a
great hardness and cannot be deformed by rolling or drawing processes. Thyssen
Krupp Stahl (TKS) prepared a sample of pure iron with embedded Al2 O3 inclusions
[18.10]. The sample was prepared with a grinding paper having a grain size of 80.
The grains consist of 80% ZrO2 and 20% Al2 O3 . The grinding time amounts to
less than 1 min, yielding a surface roughness of below 15 m. This simple kind
of sample preparation is sufficient for scanning LIBS in contrast to SEM-EDX,
Fig. 18.13 Left: Aluminum map of the 308.22 nm line. Right: Aluminum map of the 396.15 nm
line taken at the same measurement. The measuring parameters are measuring frequency 1,000 Hz,
atmosphere Ar 5.0, total measuring time 11 min
Fig. 18.14 Left: Oxygen map of the O I 130.22 nm line. Right: Oxygen map of the O I 777.19 nm
line taken at the same measurement. The measuring parameters are the same as for Fig. 18.13
which requires a surface roughness of well below 1 m and therefore a polishing

of the surface for more than 20 min. Figure 18.13 shows two aluminum maps
using two aluminum lines: Al I 308.22 nm and Al I 396.15 nm, having different
excitation energies of 4.02 eV and 3.14 eV, respectively. Both lines were investigated
simultaneously to minimize the chance of error. Most of the observed signal peaks
are observed with both lines; hence, a random character of the signal peaks can be
excluded.
Fig. 18.15 Left: Aluminum map of the 308.22 nm line, middle: nitrogen map of the 149.26 nm
line, right: oxygen map of the 777.19 nm line taken at the same measurement. The measuring
parameters are the same as for Fig. 18.13
Figure 18.14 shows two oxygen maps using the emission lines O I 130.21 and
O I 777.19 nm measured at the same time as the data for Fig. 18.13. Many of the
signal peaks coincide in both maps, but not all. The signal level is for both lines
comparatively small. The 130.21 nm line lies in the deep VUV and may be affected
by absorption due to impurities in the atmosphere of the measuring chamber or the
transmission and reflectivity features of the used optics. For the 777.19 nm line,
the sensitivity of the photo cathode of the photomultiplier is quite low. Comparing
the maps of Fig. 18.13 with those of Fig. 18.14 shows that at most of the positions
where an enhanced aluminum intensity occurs also an enhanced oxygen emission is
observed. This is an indication of the detection of inclusions consisting of Al2 O3 .
Another sample prepared by TKS contained AlN inclusions in a pure iron matrix.
The sample was ground in the same way as described above. Figure 18.15 shows
three maps of aluminum, nitrogen, and oxygen. In this case, two different kinds of
coincidences can be observed: (a) between the Al and the N map and (b) between
the Al and the O map. Obviously, case (a) belongs to the AlN inclusions brought
into the sample, whereas (b) reveals Al2 O3 grains from the grinding paper which
contaminated the sample surface of the soft pure iron matrix.
Figure 18.16 shows a plot of the nitrogen intensity versus the aluminum intensity
for all 250,000 data points corresponding to Fig. 18.15. Two different branches can
be seen. One corresponds to the AlN inclusions, and the other one is not correlated
to the nitrogen intensity and interpreted as the Al2 O3 originating from the grinding
paper. In Fig. 18.17, the oxygen intensity is plotted versus the aluminum intensity
for all data points corresponding to Figs. 18.15 and 18.16. The difference between
the AlN and the Al2 O3 inclusions is difficult to detect. In order to identify the Al2 O3
inclusions easier, in Fig. 18.18 a plot of the oxygen intensities versus the aluminum
intensities is shown like in Fig. 18.17, but only those points are plotted, which
belong to the lower branch in Fig. 18.16. The signal correlation between oxygen
and aluminum is not very high, but a clear trend can be observed, supporting the
interpretation that the lower branch in Fig. 18.16 belongs to Al2 O3 inclusions.
30000
N 149.26 nm intensity [a.u.]
24000
AlN
18000
12000
Al2O3
6000
0 5000 10000 15000 20000 25000 30000

Al 308.22 nm intensity [a.u.]
Fig. 18.16 Nitrogen intensity at 149.26 nm versus aluminum intensity at 308.22 nm for the whole
measurement of 250,000 points corresponding to Fig. 18.15. The upper branch is attributed to AlN
inclusions and the lower branch to Al2 O3 inclusions (cf. Figs. 18.17 and 18.18)
2500
O 777.19 nm intensity [a.u.]
2000
1500
1000
500
0 5000 10000 15000 20000 25000 30000
Fig. 18.17 Oxygen intensity at 777.19 nm versus aluminum intensity at 308.22 nm for the
measurement corresponding to Fig. 18.15
Another approach to reveal correlations is the principal components analysis

(PCA) [18.11]. Figure 18.19 shows the score plot of a PCA for the first 50,000 mea-
suring points of the dataset of Fig. 18.15. The data points form distinctive branches.
The corresponding loading plot reveals the dominant elements contributing to these
branches. The AlN, copper sulfide (CuS), and SiMnCaV containing inclusions can
be identified clearly.
2500
O 777.19 nm intensity [a.u.]
2000
1500
1000
500
0 5000 10000 15000 20000 25000 30000
Fig. 18.18 Oxygen intensity at 777.19 nm versus aluminum intensity at 308.22 nm for the
measurement corresponding to Fig. 18.15, but only the points of the lower branch of Fig. 18.16
are plotted
Fig. 18.19 Score plot of the principal components analysis for the first 50,000 data points of
the dataset of Fig. 18.15. The principal components (PC) 2 and 3 are plotted in the x- and
y-direction, respectively. The elements contributing to the distinctive branches have been identified
by the corresponding loading plot
For a quantification of concentrations of inclusions, a statistical approach was

studied using the skewness of the frequency distribution of the measured intensities
[18.5]. Figure 18.20 shows the frequency distribution of the intensities of aluminum
measured in a scan field by the SML system. The frequency distribution has an
asymmetric shape as a result of the superposition of a nearly Gaussian distribution
50000
frequency distribution dN/dI [1]
median
40000
30000
20000
mean
10000
0
0 100 200 300 400 500
Fig. 18.20 Frequency distribution of 250,000 emission signals of an aluminum emission line. The
dashed lines indicate the position of the median and mean value
caused by the analytes dissolved in the iron matrix and the inclusions causing
outliers with significantly higher intensities. The asymmetric frequency distribution
can be characterized by the skewness given by:
3.M A/
D (18.2)
s
with M mean, A median, and s standard deviation. This skewness is shown in

Fig. 18.21 as a function of the aluminum oxide concentration determined conven-
tionally. The different symbols of the data points correspond to different aluminum
lines used for this evaluation. The skewness correlates with the content of aluminum
oxide inclusions in the sample and hence provides a measure to quantify inclusion
concentrations based on scanning microanalysis with LIBS.
Quantification of the composition of inclusions based on calibrations with
Si-killed steel lollipop samples as reference materials was demonstrated success-
fully based on data gained with SML [18.6]. This type of sample has unusually
large nonmetallic inclusions. The average chemical composition of this steel grade
is 0.7% C, 0.4% Si, 1.1% Mn, 0.001% Al, 0.015% S, 0.01% P, and 0.014% Cr.
The chemical composition and size of the inclusions were first characterized by
SEM-EDX analysis and then by LIBS. Calibration curves were constructed using
a set of inclusions identified by both methods at the same positions. Intensity data
related to inclusions were extracted from the LIBS data as outlier points having an
Si intensity value greater than the average of all measured Si intensities plus three
times the associated standard deviation. Table 18.2 shows a comparison between
LIBS and SEM-EDX values. The accuracy of the results found by LIBS is within
20% of the EDX results.
skewness coefficient ζ 0.8
0.6
0.4
Al 396.15 nm
Al 308.22 nm
0.2
0 200 400 600 800 1000
c(Al2O3) [μg g–1]
Fig. 18.21 Skewness coefficient as a function of the concentration of Al2 O3 inclusions [18.5]
Table 18.2 Chemical quantification of nonmetallic inclusions by LIBS compared with SEM-EDX
in the lollipop sample
Ø.m/ SiO2 MgO Al2 O3 MnO CaO TiO2
SEM-EDX 350 40 17 4.5 1 37 0.5
LIBS 40 17 5.7 0 37 0.4
SEM-EDX 33 47 3.9 11 9.4 28 1.1
LIBS 44 4.9 13 8.1 30 1.2
SEM-EDX 21 46 6.6 17 9.6 20 0.9
LIBS 46 6.7 21 7.7 18 1
SEM-EDX 8 47 3.1 18 10 21 1.2
LIBS 54 5.4 19 3.3 18 0.9
Values in columns 3–8 are given in m.-% [18.6]
While inclusions are characterized by high concentration gradients of certain

elements with respect to the embedding material matrix, segregations have only
weak concentration gradients extending over larger areas. In Fig. 18.22 sulfur,
copper, and phosphorus mappings of a segregation zone in steel are shown [18.10].
An area of 40 40 mm2 of the steel sample was scanned with a step size of 100 m
corresponding to 400400 measuring points. The scanning time amounts to 15 min.
In this type of presentation, the element intensities at each measuring point are
plotted in a gray scale representing different signal levels. The sample is a low-
alloyed disc from billet. The segregation zone is clearly visible in all three mappings,
see enriched element concentration along a line parallel to the y-axis at x 15 mm.
In the sulfur and copper map, the segregation shows up as a series of remaining
melt zones. The phosphorus mapping additionally reveals a dendritic microstructure
running to the central segregation line.
Segregation ratios can be determined, defined as ratio of analyte concen-
tration in the segregation over the analyte concentration in the homogeneous,
Fig. 18.22 Segregations and dendrites in steel. All mappings are gained during the same scanning
measurement with SML. The sulfur, copper, and phosphorus channels show a segregation zone
along the y-direction at x 15 mm [18.10]
Table 18.3 Segregation ratios ks determined with ESMA and LIBS for the elements S, P, and Mn
ks (S) ks (P) ks (Mn)
ESMA 2.5 4.6 1.2
LIBS 2.2 4.1 1.3
segregation-free area. Table 18.3 shows a comparison of segregations ratios deter-

mined with scanning LIBS with those gained from electron scanning microanalysis
(ESMA).
The capability of SML to analyze decarburization zones was demonstrated
successfully [18.12]. Monitoring of decarburization layers can help controlling and
improving the quality of the final steel product. Conventional methods are based
on optical observation of a polished and etched surface, but this method lacks
precision and consistency. Decarburization occurs when the steel starts solidifying
during casting. By reaction with the ambient oxygen and moisture carbon is lost in
a layer close to the surface. While pearlite is the crystal structure formed in 0.8%
C solid steel, the loss of C leads to areas of pure ferrite and mixtures of ferrite and
pearlite. A billet sample surface of 4 1 mm2 was studied in a first step by optical
microscopy and subsequently the same area was scanned with SML [18.12]. Both
images – the conventional optical and the chemical image of the carbon gained with
LIBS – correlate largely. The ferritic zones and the ferritic grain boundaries show
up clearly in the carbon element mapping corresponding to low or even vanishing
carbon concentrations. Carbon concentration as a function of the distance from
the surface to depths of up to 10 mm were determined by averaging the carbon
concentration gained in the C mapping over the width of the scan field. The results
were compared with those detected by electron probe microscopy (EPMA) of the
same sample showing a very high degree of correlation. The EPMA measurement
required 2 h, whereas the LIBS measurement took 4 min only.
The achieved results show that scanning LIBS is applicable not only to detect
and quantify inclusions, but also to visualize and quantify extended segregation as
well as decarburization zones.
18.4 Recycling of Aluminum Scrap
Motivation and concept of a LIBS setup with scanner mirrors for the identification
of aluminum alloys were described in Sect. 13.3. Aluminum is – after steel –
one of the most important metals. Production of secondary aluminum saves up to
95% of the amount of energy needed for primary aluminum production. However,
secondary production of aluminum wrought alloys (WA) requires melting of high-
value primary aluminum or a presorting of shredded aluminum pieces. The main
processing route for nonsorted mixtures of wrought and cast alloys (CA) is the
production of CA. Float sink plants used to separate residues from old vehicles yield
only a mixed aluminum shredder scrap consisting of a mixture of CA and WA. Up
to now no large-scale, commercially available method exists, which would allow for
an automatic separation of aluminum alloys.
LIBS offers a new approach for the high-speed characterization and identification
of scrap pieces. Figure 18.23 illustrates the concept (the feeding vibrating chute
is not shown) [18.13]. Scrap pieces are singularized and accelerated on a belt
conveyor up to velocities of 3 m/s passing two sensor systems. In a first step,
the position and size of the parts are detected by a 3D sensor. With this data the
beam of the LIBS module is directed onto the object to determine the composition
(cf. Fig. 13.23). Depending on the measuring result, the part is ejected by an array
of pressurized valves to one of up to four sorting fractions (module ejection in
Fig. 18.23). Table 18.4 shows the technical data of the developed laser sorting
system for aluminum alloys. The volume depicted by dashed lines is the measuring
Fig. 18.23 Setup of a laser sorting system with the three main modules. 3D D laser line section
sensor to detect the position and shape of the pieces, LIBS D LIBS module with xy-scanner (cf.
Fig. 13.23), ejection D array of pressurized valves for automatic ejection. 1–4 D boxes for the
sorted fractions. The volume depicted by dashed lines is the measuring volume within which the
laser beam of the LIBS module irradiates the pieces passing by
18.4 Recycling of Aluminum Scrap 485
Table 18.4 Technical data of laser sorting system for aluminum alloys
Characteristics Data
Laser Nd:YAG
Simultaneously detected elements Al, Si, Fe, Zn, Cu, Mg,
Mn, C, Cd, Pb, Ti, Sn,
Cr, Zr, Ni, Ca
Speed of belt conveyor 3 m/s
Sorting rate 40 pieces per second
Fig. 18.24 System for the identification of scrap pieces on belt conveyors. Left: electric control
cabinet, middle: laser module mounted above the belt conveyor with scrap pieces and laser-induced
plasmas (time exposure showing a multitude of laser-induced plasmas), right: spectrometer module
volume within which the laser beam of the LIBS module irradiates the pieces
passing by.
Figure 18.24 shows a laser sorting system with the laser module, which is
mounted above a belt conveyor [18.1]. With a single laser pulse, the spectral signals
of all elements stated in Table 18.4 are detected simultaneously.
In a first step, operating figures are defined to assess the performance of the laser
sorting system shown in Fig. 18.23. In case of two material classes, there is always
a finite probability of error. The symbol Y states that the material class Y is existent.
The symbol X means that the material is classified – in the case described here by
means of the LIBS measurement – as material class X. Table 18.5 lists, e.g., the
different possible classification scenarios for the two material classes WA and CA
with the corresponding conditional probabilities. The correctness of an identification
result can be described by the conditional probabilities CWA D p.WAjWA/ and
CCA D p.CAjCA/. The identification yields GWA and GCA are defined as ratio of
the number of measuring results from which an allocation to a material class WA
or CA is deduced to the number of all measuring results gained with a pure scrap
charge consisting of WA and CA, respectively.
Table 18.5 Classification scenarios for two possible states and corresponding conditional proba-
bilities
X Y p D (XjY) Result
WA WA 1˛ Correct
CA WA A Type 1 error
WA CA B Type 2 error
CA CA 1ˇ Correct
WA D wrought alloy, CA D cast alloy, for further explana-
tions see text
1000 918 52 SC1 - wrought alloys

187
number of scrap pieces
800
679 607
600
Al / Ti threshold
cast alloy
data provided
wrought alloy
400
no position
classified
number
200
total
72
0
Fig. 18.25 LIBS classification result for the scrap charge SC1 consisting of wrought aluminum
alloys only
The components 3D sensor and LIBS module of the system were tested using
such known scrap charges SC1 and SC2, where SC1 comprises 918 scrap pieces of
WA and SC2 255 scrap pieces of CA [18.15]. In order to exclude the influence of
the feeding and ejection module, the pieces were put by hand on the belt conveyor.
During the measurement, the pieces are moving with a speed of 3 m/s. Analyte
signals of aluminum, titanium, and silicon were used to identify the material classes
wrought and cast aluminum. Test runs were performed to determine the correctness
and yield of the laser measurement. Figure 18.25 shows the classification results
for SC1. For 52 pieces, no position data were provided by the 3D sensor (cf.
Fig. 18.23), which is attributed to complex geometries of the scrap pieces which
could not be measured reliably by the laser line section sensor. Without geometry
data, the scanner system of the laser module cannot be controlled and therefore no
LIBS results are available for these pieces. From the numbers given at the bars
shown in Fig. 18.25, the identification correctness and the yield are determined
(values given in brackets refer to the mass): CWA D 89:4%.91:3 m:-%/, GWA D
78:4%.79:3 m:-%/. The type 1 error amounts to ˛ D 10:6%.
Figure 18.26 shows the classification results for SC2. Analogous to the SC1 case,
the following numbers are deduced for the correctness and yield: CCA D 94:0%
18.4 Recycling of Aluminum Scrap 487
300 SC2 - cast alloys

255 21
250 35
number of scrap pieces
199
200 12
187
150
Al / Ti threshold
data provided
wrought alloy
100 no position
cast alloy
classified
number
50
total
Fig. 18.26 LIBS classification result for the scrap charge SC2 consisting of cast aluminum alloys
only
Table 18.6 Results of five repetitive measurements with the scrap charges SC1, wrought alloys
and SC2, cast alloys [18.15]
No. Scrap charge G (m.-%) CWA (m.-%) CCA (m.-%)
1 SC1 74.3 91.2
2 SC1 74.9 91.0
3 SC1 71.3 92.0
4 SC1 77.3 92.1
5 SC1 73.0 92.4
Av. 74.2 91.7
1 SC2 70.6 94.3
2 SC2 72.4 94.4
3 SC2 73.9 93.8
4 SC2 74.5 89.2
5 SC2 77.3 96.5
Av. 73.7 93.6
Yield G and correctness CWA , CCA refer to the whole system including 3D
sensor, LIBS module, and ejection, cf. Fig. 18.23. Av., arithmetic mean
.91:1 m:-%/, GCA D 85:0%.82:0 m:-%/. The type 2 error amounts to ˇ D 6:0%.
With the average mass of the scrap pieces of the charges SC1, SC2 and a laser source
with 40 Hz repetition rate, the extrapolated mass throughput per hour is 2.2 t/h for
WA and 7.3 t/h for CA.
Further test runs were performed including the vibrating chute feeding the scrap
pieces to the belt conveyor and the ejection module (cf. Fig. 18.23). Table 18.6
shows a summary of the results gained for five repetitive measurements with each
scrap charge SC1 and SC2, respectively.
The correctnesses determined are close to those found in the first test scenario.
The yield is reduced which is attributed to a partly too dense occupancy of scrap
pieces on the belt conveyor and/or errors occurring in the ejection step.
100
Ccast
identification correctness [%]
80
Cwrought
60
40
20
0.0 0.1 0.2 0.3 0.4 0.5

threshold value [a.u.]
Fig. 18.27 Identification correctness for cast and wrought aluminum alloys as a function of a
decision threshold. At a threshold value of 0.17 correctnesses above 90% are achieved for cast and
wrought simultaneously
The identification correctness was determined by use of a known charge of

aluminum scrap pieces. Scrap pieces were sorted in two classes: WA with c.Si/ <
3:5 m:-% and CA with c.Si/ > 6:0 m:-%.
Figure 18.27 shows the measuring results. The identification correctness is shown
as a function of a threshold value which describes a decision limit to distinguish
between CA and WA. For a threshold value of 0.17, the identification correctness
for CA and WA is greater than 90%. Next generation laser identification and sorting
systems will allow for mass throughputs of about 5 t/h.
Within a running European project, the laser sorting will be extended to separate
in future different alloy groups of wrought aluminum with sorting rates of 60
and 100 Hz [18.14]. By this technology, the material-specific recycling can be
enlarged significantly compared with the present state of the art. Recently the
identification of WA with a correctness >95% was demonstrated successfully for
the first time [18.16].
Automatic laser-based sorting of alloys improves the sustainability of light metal
cycles by provision of a technology, with counteracts the downward spiral of
recycling processes toward low quality levels (down cycling).
References
18.1. C. Gehlen, J. Makowe, R. Noll, Automatisierte Verwechslungsprüfung von Edelstahl-

halbzeugen in der Produktion. Stahl Eisen 129, S70–S72 (2009)
18.2. International Standard, Safety of laser products – Part 1: Equipment classification and
requirements, IEC 60825–1, Ed. 2.0, 2007–03, 200 p
References 489
18.3. German Standard (2002), Technical availability of machines and production lines, terms,
definitions, determination of time periods and calculation, VDI 3423, January
18.4. R. Noll, V. Sturm, J. Makowe, Laser-Emissionsspektrometrie analysiert zeitnah Stahl-
proben. Stahl Eisen 129, S61–S64 (2009)
18.5. H. Kuss, H. Mittelstädt, G. Müller, Inclusion mapping and estimation of inclusion contents
in ferrous materials by fast scanning laser-induced optical emission spectrometry. J. Anal.
At. Spectrom. 20, 730–735 (2005)
18.6. F. Boué-Bigne, Analysis of oxide inclusions in steel by fast laser-induced breakdown
spectroscopy scanning: an approach to quantification. Appl. Spectrosc. 61, 333–337 (2007)
18.7. F. Boué-Bigne, Laser-induced breakdown spectroscopy applications in the steel industry:
rapid analysis of segregation and decarburization. Spectrochim. Acta B 63, 1122–1129
(2008)
18.8. E. Pappert, Studien zur Bestimmung der Bindungsformen von Al, Ti, Cr und Zn in Fest-
stoffen mittels Röntgenspektrometrie für Anwendungen in der Stahlindustrie, Dissertation,
Dortmund, 1998
18.9. Ç. Nazikkol, J. Flock, H. Mittelstädt, Klassische Methoden zur Analyse von Einschlüssen
in Stählen, Proc. Nichtmetalle in Metallen, ed. by D. Hirschfeld, 2000, p. 77–84
18.10. H. Bette, R. Noll, G. Müller, H. Jansen, Ç. Nazikkol, H. Mittelstädt, High-speed scanning
laser-induced breakdown spectroscopy at 1000 Hz with single pulse evaluation for the
detection of inclusions in steel. J. Laser Appl. 17, 183–190 (2005)
18.11. H. Martens, T. Mæs, Multivariate Calibration (Wiley, New York, 1998)
18.12. F. Boué-Bigne, Laser-induced breakdown spectroscopy applications in the steel industry:
rapid analysis of segregation and decarburization. Spectrochim. Acta B 63, 1122–1129
(2008)
18.13. R. Noll, C. Fricke-Begemann, P. Jander, T. Kuhlen, V. Sturm, P. Werheit, J. Makowe.
Perspektiven der Lasertechnik zur Steigerung der Ressourceneffizienz, Hrsg. U. Teipel,
Rohstoffeffizienz und Rohstoffinnovation, Fraunhofer Verlag, 2010, pp. S287–S298
18.14. European project, Pilot demonstration of high performance aluminium recycling by inline
laser identification - PARILAS, ECO/09/256174/SI2.564797, Start: June 17th, 2010,
project coordination: R. Noll, Fraunhofer ILT
18.15. Ü. Aydin, Materialidentifikation bewegter Einzelkörner aus Aluminium-Legierungen mit
der Laser-Emissionsspektrometrie, Dissertation, Shaker Verlag, Aachen, 2009
18.16. P. Werheit, C. Fricke-Begemann, M. Gesing, R. Noll, Fast single piece identification with a
3D scanning LIBS for aluminium cast and wrought alloys recycling. J. Anal. At. Spectrom.
23 August 2011, DOI: 10.1039/c1ja10096c
Annex A
The annex starts with a section giving an estimate of the necessary energy to displace
the ambient gas atmosphere caused by the expansion of the laser-induced plasma.
The following sections list the abbreviations and symbols used in this book. The
most relevant Stark data for the spectroscopic determination of the electron density
of LIBS plasmas are compiled in Sect. A.4. Lists of wavelengths and elements are
provided in Sect. A.5. Finally, Sect. A.6 describes a method to select spectral lines
for Boltzmann plots.
A.1 Displacement of Ambient Atmosphere
As shown in Sect. 8.1 and Table 8.2 more than 97% of the ambient atmosphere
particles are displaced by the expanding laser-induced plasma. A result which is
consistent with the simulations presented in Sect. 10.3, Fig. 10.5. In the following
the energy necessary for the displacement of the ambient atmosphere will be
estimated.
The irradiated laser beam evaporates target material. The particles flowing off
collide with particles of the ambient atmosphere. By these collisions, the ambient
atmosphere is partially displaced. A part of the ambient gas particles diffuses into
the material vapor (cf. Table 8.2, Fig. 10.5). For simplicity, it is assumed that the
ambient atmosphere is displaced completely by the expanding material vapor.
In case of a complete displacement, the material vapor and plasma has to do
work, which will be calculated taking the following assumptions (a) spherical
expansion, (b) the ambient gas is collected at the front of the expanding plasma.
The latter is a simplification and yields an upper estimate for the displacement
energy (a more realistic description would have to consider the flow of ambient
gas induced by the expanding vapor). Then the expanding plasma has to provide the
energy given by:
Z R Z R
d
Ed D F dr D .ma rP / dr (A.1)
0 0 dt
R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9, 491

492 Annex A
with Ed displacement energy, R radius of the expanding plasma, ma accumulated

mass of ambient gas, rP velocity of the accumulated ambient gas at the expanding
plasma front.
For the accumulated mass of the ambient gas and a half sphere holds:
2
ma .t/ D a r.t/3 ; (A.2)
3
where a is the density of the ambient gas. Insertion of relation (A.2) into (A.1)
leads to:
ZR
2a 3
Ed D r rR C 3r 2 rP 2 dr: (A.3)
3
0
The second integrand can be transformed as follows:

Z Z Z Z
3r 2 rP 2 dr D rP 2 dr 3 D rP 2 r 3 r 3 drP 2 D rP 2 r 3 2 r 3 rR dr (A.4)
which yields for (A.3):

Z ˇˇR
2a 3 2 3 ˇ
Ed D r rP r rR dr ˇ (A.5)
3 ˇ
0
The kinetic energy of the accumulated gas at the front of the material vapor is
given by:
1 2a 3 P 2
Ekin D R R : (A.6)
2 3
Relations (A.5) and (A.6) show that for negligible acceleration of the front of
the expanding material vapor, i.e., the second term in (A.5) can be neglected, the
displacement work is just twice as much as the kinetic energy of the accumulated
gas. This case is equivalent to a completely inelastic collision of the ambient gas
at the front of the expanding vapor. In the rest frame of the expanding front, the
incoming ambient gas loses its kinetic energy being transformed completely into
internal energy.
Since the expanding material vapor has a high temperature, it emits visible
radiation. In the following it is assumed that the position of the expanding front is
identical to the position of the luminous front observed experimentally, cf. Sect. 8.1.
The height of the luminous front is observed with high-speed streak images
in the time interval between 3 and 65 ns after the irradiation of the laser pulse
with an energy of 40 mJ, see Fig. 8.6. Figure A.1 shows this data in a linear–linear
presentation and a second degree polynomial as fit curve.
A.1 Displacement of Ambient Atmosphere 493
1.0
1. laser pulse
0.8
0.6
z [mm]
0.4
r 2 = 0.993
0.2
z / mm = 0.0453 + 0.0246 t / ns -1.873 10-4(t / ns)2
0.0
0 10 20 30 40 50 60 70
t [ns]
Fig. A.1 Height z of the luminous plasma front above the sample surface as a function of time
after the irradiation of the first laser pulse with 40 mJ pulse energy (same data as Fig. 8.6 top, but
linear scaling of x- and y-axis). As fit curve a second degree polynomial is taken
The general form of the polynomial is given by:
r D a0 C a1 t C a2 t 2 (A.7)
with (A.5) follows:

2a 3 P 2 a2 4
Ed D R R R (A.8)
3 2
with the fit parameters given in Fig. A.1: a0 D 0:0453 mm, a1 D 0:0246 mm=ns,
a2 D 1:873 104 mm=ns2 and the density of air 1:29 kg=m3 relation (A.8) yields
Ed D 150 mJ for t D 30 ns. This estimate of the upper limit of the displacement
energy is higher than the laser pulse energy. Hence, the assumption of a complete
displacement and accumulation of the ambient gas at the expanding front is a too
strong simplification. Anyway, this estimate shows that the displacement energy is
of the same order of magnitude as the laser pulse energy itself. An estimate of the
necessary energy to heat the ablated mass of the iron target of about 100 ng (for
a single laser pulse with 40 mJ, cf. Fig. 7.5) from room temperature to the melting
point (average specific heat capacity of iron 0.8 J/(gK), melt temperature 1;535ı C),
provision of the melt enthalpy (270 J/g), heating of the melt to the evaporation
temperature (assuming the same heat capacity as for solid iron [A.8]), provision of
the enthalpy of vaporization (6,370 J/g), heating the vapor to 10,000 K, and ionizing
90% of the Fe particles (ionization energy 7.9 eV) yields 2.1 mJ. Therefore, the
displacement energy is significantly greater than the energy necessary to produce
the plasma state.
494 Annex A
A.2 Abbreviations
Abbreviation Explanation Chapter(s)/Section(s) Page(s)

AAS Atomic absorption 14.1.2, 14.5 286, 338
spectroscopy
ABS Acrylonitrile butadiene 14.1.2 286
styrene
AOM Acousto-optic modulator 8.1 120
BAS Bureau of Analysed 6.1, 13.1.2 86, 238
Samples Ltd.
BBA Borate bead analysis 14.2.2 316
BEC Background equivalent 11.2, 12, 13.1.3 215, 227, 250
concentration
BFR Brominated flame 14.1.2 285
retardants
BNC Bayonet Neill Concelman 13.1.1 231
CARS Coherent Anti-Stokes 1 1
Raman spectroscopy
CCD Charge-coupled device 4.2, 4.4, 15.1 55, 62, 388
CE Capillary electrophoresis 2.4, 14.5 14, 338
COG Curves-of-growths 10, 14.8.2 189, 376
CPC Condensation particle 14.8.2 373
counter
CRM Certified reference 2.1 11
material
DLVQ Dynamic learning vector 14.1.1 278
quantization
DMA Differential mobility 14.8.2 373
analyzer
DP Double pulses 3.2, 7.3, 11.1 23, 104, 209
DPSSL Diode-pumped solid state 3.1, 3.6, 4.1, 15.1 19, 35, 47, 389
laser
EDL Electrodeless discharge 12 225
lamps
EDXRF Energy dispersive X-ray 14.3 321
fluorescence analysis
ELPI Electrical low-pressure 14.8.1 362
impactor
EOL-WEEE End-of-life electric and 14.1.2 285
electronic equipment
EPMA Electron probe 18.3 483
microscopy
ESMA Electron scanning 18.3 482
microanalysis
fb Free–bound 8.4 161
ff Free–free 8.4 161
FHG Fourth harmonic 4.1 48
generation
FML Flexible measuring 17.3 448
system based on LIBS
(continued)
A.2 Abbreviations 495
(continued)
FWHM Full width at half 2.1, 3.1, 4.3, 6.2, 9.4, 10, 19, 59, 92,
maximum 13.1.1, 14.1.1 173, 230,
281
GC Gas chromatography 14.5 338
GD-OES Glow discharge optical 14.1.2, 16.1, 16.2 286, 401, 416
emission spectrometry
GF Graphite furnace 12 224
HDPE High density polyethylene 14.1.1 278
H” Hydrogen Balmer line 8.2 149
HPLC High pressure liquid 2.4, 14.5 14, 338
chromatography
HPTLC High performance thin 14.5 338
layer chromatography
HWHM Half width at half 8.2, 14.4 151, 332
maximum
ICP-AES Inductively coupled 14.1.2 286
plasma atomic emission
spectrometry
ICP-MS Inductively coupled 14.1.2, 14.5 286, 347
plasma mass
spectrometry
ICP-OES Inductively coupled 12, 14.1.2 227, 302
plasma optical emission
spectrometry
IEC International 18.1 468
Electrotechnical
Commission
ILT Fraunhofer-Institut für Preface, 4.5 v, 66
Lasertechnik
IPDA Intensified photodiode 14.1.1 277
array
IR Infrared 14.4, 14.5 334, 338
LA-MS Laser ablation mass 14.1.2 286
spectrometry
LAS Laser absorption 1 3
spectroscopy
LDPE Low density polyethylene 14.1.1 278
LEAFS Laser-excited atomic 12, 14.3 221, 320
fluorescence
spectrometry
LIBS Laser-induced breakdown 1 2
spectroscopy
LIDAR Light detection and 1 1
ranging
LIF Laser-induced 1, 12 1, 3, 221
fluorescence
LIFT Laser identification of 17.2, 18.1 438, 467
fittings and tubes
(continued)
496 Annex A
(continued)
LOD Limit of detection 11.2, 13.1.1, 13.1.2, 215, 234, 240,
13.3, 14.1.2, 14.3, 267, 299,
17.3, 17.4 323, 451,
462
LOQ Limit of quantification 11.2 215
LTE Local temperature 8.4, 10 162, 187
equilibrium
MCI Multichannel integrator 4.5, 13.1.1 66, 231
electronics
MCP Microchannel plate 4.2, 4.4 52, 62, 64
XRF Micro X-ray fluorescence 18.3 474
analysis
MS Mass spectrometry 14.5 338
MVDA Multivariate data analysis 16.1 408
NBS National Bureau of 13.2 254
Standards
NCS National Analysis Center 13.2 254
for Iron and Steel
Nd:YAG Neodymium-doped 3.1 17
yttrium aluminum
garnet
NIR Near infrared 14.1.1 276
NIST National Institute of 13.1.2 238
Standards and
Technology
NMR Nuclear magnetic 14.5 338
resonance
OES Optical emission 13.1.2 235
spectrometry
PA Polyamide 3.3 27
PC Polycarbonate, 3.3 27
Principal component, 18.3 480
Personal computer 17.1 435
PCA Principal component 18.3 478
analysis
PDA Photodiode array 4.2, 15.1 53, 388
PE Polyethylene 14.1.1 275
PET Polyethylene 14.1.1 275
terephthalate
PIN Positive-intrinsic-negative 14.1.1 276
photodiode
PLS Partial least squares 16.1, 18.2 409, 471
PMT Photomultiplier tube 3.5, 4.3, 4.4 30, 53, 61
PP Polypropylene 14.1.1 275
PS Polystyrene 14.1.1 275
PVC Polyvinyl chloride 14.1.1 275
(continued)
A.2 Abbreviations 497
(continued)
Q-switch Variable attenuator to 4.1, 6 47, 83
influence the quality
factor of the laser
resonator
RMSE Root mean square error 16.1 409
RMSEP Root mean square error of 18.2 471
prediction
ROI Region of interest 14.1.1 280
RR Rautaruukki Oyj 14.7 358
RSD Relative standard 3.1, 14.1.2 20, 291
deviation
RSDP Relative standard 13.3 266
deviation of procedure
SBS Stimulated Brillouin 4.3 61
scattering
SD-OES Spark discharge optical 13.1, 13.1.3, 17.1, 229, 245, 430,
emission spectrometry 18.3 475
SEM Scanning electron 1, 7.1, 15.1 3, 98, 393
microscopy
SEM-EDX Scanning electron 1, 18.3 3, 474
microscopy energy
dispersive X-ray
fluorescence analysis
SFC Supercritical fluid 14.5 338
chromatography
SFM Sum frequency mixing 12 223
SHG Second harmonic 4.1, 12 48, 223
generation
SML Scanning microanalysis 15.2, 18.3 396, 473
with laser spectrometry
SMPS Scanning mobility 14.8.2 373
particle sizer
SNR Signal-to-noise ratio 14.1.2 288
SP Single pulses 3.2, 11.1 23, 209
Spark-OES Spark optical emission 17.2 438
spectrometry
TEA Transverse excited 14.5 350
amplifier
TKS ThyssenKrupp Steel 14.7, 16.1 358, 404
TP Triple pulses 3.2, 7.3 25, 104
UV Ultraviolet 3.5 33
VUV Vacuum ultraviolet 3.5, 13.1.1 32, 230
XRF X-ray fluorescence 13.1, 13.1.2, 13.1.3, 229, 235, 245,
analysis 16.1 401
498 Annex A
A.3 List of Symbols
Symbol Description Unit Chapter(s), Section(s), Page(s)

Figure(s), Formula(s)
A Absorptance 1 5.1, (5.1) 75
A Dimensionless coefficient 1 9.4, (9.19) 173
for Stark effect
A Analytical resolving 1 14.1.2, (14.1), (14.4), 288, 291,
power, analytical 17.2, (17.1), 443
resolution Table 17.5
A ion broadening parameter 1 A.4, Table A.1 512, 514
Ao Overlapping area m2 3.6, Fig 3.20 37
AC Effective cross section of m2 4.6, Fig 4.17 67
measuring chamber
Anm ; Aji Einstein coefficient of the s1 8.2, (8.5) 133
transition n ! m,
j !i
9.2, (9.3), (9.8), 10.2, 169, 170,
(10.13) 194
Ap Surface area of a plasma m2 9.2, (9.10) 171
sphere
As Sampling area m2 3.6 38
1
˛; ˛ ./; ˛ ./ Absorption coefficient m 5.3, (5.8) 79
9.3, (9.11) 171
9.4, (9.32) 180
˛ Angle of incidence ı 4.2, Fig 4.4, (4.1) 51, 52
˛ Fine structure constant 1 9.1, (9.1), 10.2, 167, 195
(10.15)
a0 Bohr radius m 9.1, (9.1), 10.2, 167, 195
(10.15)
˛O Angle of observation ı 3.7, Fig 3.23 39
˛D Detection angle ı 4.6, Fig 4.18, 13.1.2 68, 237
˛j Polarizability of species j m3 8.3, (8.12) 154
˛L Angle of incidence of the ı 3.7, Fig 3.23 39
laser beam
3.9, Fig 3.26 42
ˇ Angle of diffraction ı 4.2, Fig 4.4, (4.1) 51, 52
Bnm Einstein’s coefficient of m/kg 10.2, (10.12) 194
stimulated emission
Bij , Bmn Einstein’s transition m/kg 9.3, (9.11) 171
probability for
absorption
10.2, (10.12), (10.13) 194, 195
c Vacuum speed of light m/s 5.3, (5.8) 79
9.2, (9.8) 170
c Specific heat capacity J/(kg K) 5.1, ((5.2) 75
cj ; ca Concentration of analyte g/g 2.1, 10 11, 192
j, a
(continued)
A.3 List of Symbols 499
(continued)
cOi Concentration ratio 1 11.5, (11.23), (11.24); 218, 254
13.2, (13.2)
Cs Stark broadening Å1016 cm3 A.4, Table A.1 512, 514
parameter
Adiabatic coefficient 1 8.3, (8.10) 140
d Diameter of laser beam m 3.3, (3.5) 28
d Stark shift Å A.4, Table A.1 512, 514
ds Single spot diameter m 3.6 38
df Foil thickness m 7.4, (7.1) 114
dg Grating period m 4.2, (4.1) 51
dmc Diameter of material m 3.1, (3.1) 21
cloud
dplasma Diameter of the spherical m 10, (10.1) 191
plasma
dAs Receiving entrance m2 10.2, (10.24) 197
aperture of the
spectrometer
D Diameter of the m 3.4, (3.3); 11.1, 27, 209
illuminated aperture of Table 11.1
the focusing lens
D Stark shift parameter m 9.4, (9.20) 174
Ds Stark shift parameter Å1016 cm3 A.4, Table A.1 512, 514
Dg Effective diameter of the m 4.2, Fig 4.5, (4.5) 52
grating
ı.y; z; t / Fringe displacement 1 8.3, (8.11) 154
ıh Heat penetration depth m 3.3, Fig. 3.10, (3.2) 26
3.6, (3.7) 34
ıopt Optical penetration depth m 3.3, Fig. 3.10 26
5.1, Fig. 5.1, (5.3) 77
Ezion ; Ea;
ion
z Lowering of the J 9.2, (9.7), Fig. 9.2 170
ionization energy, for
species a
10.2, (10.5) 193
Difference between two m 4.2, (4.3) 51
neighboring spectral
lines of equal intensity
a Apparative broadening m 10.1, Table 10.1, 10.2, 192, 197
given as full width at (10.24)
half maximum
(FWHM)
D Doppler line width m 9.4, (9.18), (9.23), 173, 174,
(FWHM) (9.26) 175
m
G Measured Gaussian part m 9.4, (9.27), Fig 9.7 175, 178
of the line width
(continued)
500 Annex A
(continued)
L Lorentz line width m 9.4, (9.26) 175
(FWHM)
aL Apparative broadening m 9.4, (9.28), (9.29) 177, 178
described by a Lorentz
function
m
L Measured Lorentzian part m 9.4, (9.28), Fig 9.6 177
of the line width
N Natural line width m 9.4, (9.26) 175
shift Line shift m 9.4, (9.20) 174
Stark Stark line width (FWHM) m 9.4, (9.19), (9.22) 173, 174
aV Total apparative m 9.4, (9.29) 178
broadening described
by a Voigt profile
s Distance between the m 2.1, Fig 2.2 9
beam waist location
and the surface of the
target
3.4 29
3.9, Fig 3.26; 11.1, 42, 210
Table 11.1
Q Spectral width of the m1 1, Table 1.1; 4.3, 3, 61
laser radiation Table 4.7
H Full frequency width at s1 9.2, (9.5) 169
half maximum
V Frequency shift of the s1 9.2, (9.5) 169
line maximum in
relation to an
unperturbed line
t Interpulse separation s 3.2, Fig 3.9; 11.1, 25, 209
Table 11.1
ti Interpulse separation s 3.2, Fig 3.9 25
between pulse i and
pulse i C 1
tLIBSLIF Time difference between s 12 224
the LIBS-pulse and the
LIF-pulse
e Elementary charge As 5.3, (5.9) 79
8.4, (8.15) 161
9.2, (9.8) 170
Eb Burst energy,
P J 3.2; 11.1, Table 11.1 25, 209
Eb D Ej
j
EH Ionization energy of eV 8.4, (8.17), (8.19); 163, 164, 167,
hydrogen, Rydberg 9.1, (9.1), (9.2); 168, 195
constant 10.2, (10.16),
(10.18)
(continued)
(continued)
Ej Energy of pulse j in a J 3.2, 11.1, Table 11.1 25, 209
pulse burst
EL Energy of the laser pulse J 3.1, 3.4, (3.6) 20, 28
Em Specific melt enthalpy J/kg 10, Fig 10.1 186
Em ; Ei Energy of the lower level eV 9.3, (9.13); 9.4, 172, 180,
m; i (9.33); 14.5, 342
Table 14.14
En ; Ej Energy of the upper level eV 8.2, Table 8.1; 9.2, 133, 169,
n; j (9.3); 13.1.1, 231,
Table 13.1; 13.1.2, 238
Table 13.3
Ev Specific enthalpy of J/kg 10, Fig 10.1 186
vaporization
"0 Vacuum permittivity, As/(Vm) 5.3, (5.9), 8.4, (8.15), 79, 161,
dielectric constant 9.2, (9.8) 170
©a1;2 Weighting factor 1 10, Table 10.1 192
describing the
proportion of an
element a in the
plasma in relation to its
proportion in the
specimen
"d; th Damage threshold radiant J=m2 4.3, Table 4.5 59
exposure
"fb ./ Emission coefficient of W=.m3 sr m) 10.2, (10.10), (10.16) 194, 195
free–bound transitions
"!;d
fb Emission coefficient for W=.m3 sr rad s1 ) 9.1, (9.2) 168
free–bound transitions
"ff ./ Emission coefficient of W/(m3 sr m) 10.2, (10.10), (10.14) 194, 195
free–free transitions
"ff Emission coefficient of W/(m3 m) 9.1 167
"!;d
ff Emission coefficient of W/(m3 sr rad s1 / 9.1, (9.2) 167
"ji;d Line emission coefficient W=.m3 sr s1 / 9.2, (9.3) 169
"l ./ Line emission coefficient W=.m3 sr m) 10.2, (10.10), (10.14) 194
Degree of overlap 1 3.6, (3.8) 37
f Focal length of a lens m 3.4, (3.3) 27
fB Boltzmann factor 1 8.2, Table 8.2 136
fij ; fmn Oscillator strength 1 9.2, (9.8); 9.4, (9.33); 170, 180,
13.1.2, Table 13.3; 238,
13.2, Table 13.14 262
ga;zIn Statistical weight of 1 10.2, (10.8) 193
excited level n of
species a in charge
state z
(continued)
502 Annex A
(continued)
gm ; gi Statistical weight of 1 9.2, (9.8), 9.4, (9.33) 170, 180
lower level m; i
gn ; gj Statistical weight of the 1 8.2, (8.5); 9.2, (9.3), 106, 133,
upper level n; j (9.8) 169,
170
./ Nonnormalized shape 1 9.4, (9.30) 178
function
./ Line profile as a function s 9.2, (9.3), (9.4) 169
of the frequency
nm ./ Line profile of the m1 10.2, (10.12) 194
transition n ! m
L ./ Lorentz line profile s 9.2, (9.5) 169

V ./ Voigt profile m1 9.4, (9.23) 174
(order) Weighting factor between 1 10.1, Table 10.1 192
0 and 1 describing the
fraction of diffracted
power for a given order
of the grating
spectrometer
h Planck constant Js 8.4, (8.14) 161
h, hc Crater depth m 7.1, Fig 7.1; 13.1.2, 98, 239,
Table 13.5; 16.1, 405
Table 16.1
hji Photon energy of the J 9.2, (9.3) 169
transition fromj to i
„! Photon energy J 9.1, (9.1) 167
H Enthalpy of the plasma J 10.4, (10.25) 203
I Irradiance at the location W=m2 1, Table 1.1 3
of interaction
I./ Spectral radiance W=.m2 sr m) 10.2, (10.9) 193
Ianalyte
acc
Accumulated analyte a.u. 14.1.2, (14.2) 291
.tdelay ; tint / signal
Id;th Damage threshold W=m2 4.3, Table 4.5 59
irradiance
If Average irradiance at the W=m2 3.4, (3.6) 28
location of the beam
waist
I. D i ; t / Spectral intensity of the W=.m2 m) 2.1, Fig 2.4 10
plasma emission at the
wavelength i as a
function of time
IL .R; t / Irradiance of the incident W=m2 10, Fig 10.1 186
laser beam as a
function of coordinate
R and time t
Ii L Irradiance of the incident W=m2 2.1, Fig 2.2 9
laser beam
(continued)
(continued)
Ir L Reflected laser irradiance W=m2 2.1, Fig 2.2 9
IsL Scattered laser irradiance W=m2 2.1, Fig 2.2 9
Im ./ Calculated spectral W/m 10.2, (10.24) 197
radiant flux
I3 ./ Calculated spectral W/(m sr) 10.2, (10.24) 197
radiant intensity at
radius r3
kB Boltzmann constant JK1 5.3, (5.13) 80
8.4, (8.14) 161
ks Segregation ratio 1 18.3, Table 18.3 483
kt Quantity proportional to m3 9.4, (9.33), Table 9.1 180, 181
the absorption
coefficient
Thermal diffusivity m2 /s 3.3, (3.2) 26
5.1, (5.2) 75
7.4, (7.1) 114
ln ƒ Coulomb logarithm 1 5.3, (5.13), (5.14) 80
L Distance between the m 3.6, Fig 3.20 37
centers of two adjacent
laser beam positions
L; d Spectral radiance of a W=.m2 sr m/ 9.2, (9.9); 171, 195
black body, Planck’s 10.2, (10.19), (10.20)
law
Lji ; d Spectral radiance of a W=.m2 sr m/ 9.4, (9.31) 179
black body at the
wavelength of the line
transition ji
Wavelength of the laser m 3.4, (3.3); 11.1, 27, 209
radiation Table 11.1
B Thermal de Broglie m 8.4, (8.14) 161
wavelength of
electrons
exc Wavelength of the laser m 1, Table 1.1 3
used for excitation
i Element-specific m 2.1, Fig. 2.4 10
wavelength
ji Wavelength of the m 9.2, (9.8) 170
transition fromj to i
ji 0 Central wavelength of the m 9.4, (9.18) 173
transition from j to i
L Landau length m 8.4, (8.15) 161
ƒe Thermal de Broglie m 10.2, (10.5) 193
wavelength of the
electrons
(continued)
504 Annex A
(continued)
me Electron mass kg 5.3, (5.9) 79
8.4, (8.14), (8.19) 161, 164
M Mass of atoms or ions kg 8.4, (8.19) 164
M2 Beam propagation ratio 1 3.3, (3.5), 11.1, 28, 209
Table 11.1
Ma Mass of emitting atom kg 9.4, (9.18) 173
species a
n Diffraction order 1 4.2, (4.1), (4.3), 51, 52
Fig 4.4
n Principal quantum 1 8.4, (8.17) 163
number
9.1, (9.2) 168
na Refractive index of the 1 8.3, (8.11) 154
ambient atmosphere
ne Electron density m3 5.3, (5.9) 79
8.2, Table 8.2; 9.4, 136, 173,
(9.19), (9.20) 174
nFe Total iron density (atoms m3 8.2, Table 8.2, (8.6), 136, 139
and ions) Table 8.3
nFe I Neutral iron (atoms) m3 8.1 175
density
nFe II Density of singly ionized m3 8.1 136
iron
ni Density of ions m3 5.3, (5.13) 80
nN Nitrogen density m3 8.2, Table 8.2 136
np Refractive index of the 1 8.3, (8.11) 154
plasma
nref Refractive index 1 4.2, (4.4) 52
N Number of illuminated 1 4.2, (4.3) 51
grooves of a grating
N Number of samples 1 11.2, (11.10), (11.12), 213, 216
(11.21), (11.22)
Naz Density of ion species a m3 9.1, (9.1) 167
with charge z
9.2, (9.3) 169
10.2, (10.5), (10.6) 193
Na 1;2 Particle density of m3 10.1, (10.2), 191, 192
species a originating Table 10.1
from the specimen in
sphere 1 or shell 2
(continued)
(continued)
g
Na 1;2 Particle density of m3 10.1, (10.3), 191, 192
species a originating Table 10.1
from the ambient gas
in sphere 1 or shell 2
g
Na .r; t / Density of ambient gas of m3 10, Fig 10.1 186
species a
g
Na 3 Particle density of m3 10.1, (10.4), 191, 192
species a (atoms or Table 10.1
molecules) of the
ambient gas in shell 3
Nabl , Nabl; 3s Number of ablation 1 7.3, 13.1.2 104, 238
bursts
ND Number of electrons in 1 8.4, (8.16) 162
the Debye sphere
9.4, (9.19) 173
Ne Electron density m3 8.3, (8.12) 154
9.1, (9.1), (9.2) 167, 168
10.1, (10.5), (10.7) 193
NFe Number of iron particles 1 8.2, Table 8.2 136
NFe I Number of neutral iron 1 8.2, Table 8.2 136
particles
NFe I n Number of iron atoms in 1 8.2, Table 8.2 136
the upper state n
Nj Densities of neutral m3 8.3, (8.12) 154
atoms and ionic species
Nmp Number of measuring 1 3.8,11.1, Table 11.1, 39, 210,
pulses 13.1.1, 14.2.1 232,
306
Np Number of 1 3.6 38
simultaneously
generated plasmas
Npp Number of prepulses 1 3.8; 11.1, Table 11.1; 39, 210,
13.1.1, 14.2.1 232,
306
Nwu Number of warming-up 1 3.8; 11.1, Table 11.1 39, 210
pulses
NAES Numerical aperture of the 1 4.2, (4.4) 52
spectrometer seen from
the entrance slit
(continued)
506 Annex A
(continued)
Collision frequency of s1 5.3, (5.8), (5.12) 79, 80
electrons with ions and
atoms
0 Frequency of the line s1 9.2, (9.5) 169
maximum for the
unperturbed line
ei Collision frequency of s1 5.3, (5.12), (5.13), 80, 125
electrons with ions 8.1
en Collision frequency of s1 5.3, (5.12) 80
electrons with atoms
8.1 98
rep Repetition rate of the Hz 3.1, Fig. 3.1; 11.1, 18, 209
laser source Table 11.1
vsound Velocity of sound of the m/s 3.1, (3.1) 21
ambient gas
vth Thermal velocity of the m/s 10.1, (10.1) 191
atoms and ions in the
plasma
! Angular frequency rad/s 5.3, (5.8) 79
!p Plasma frequency rad/s 5.3, (5.9) 79
pg ; pg .r; t / Ambient gas pressure N/m2 2.1, Fig. 2.2; 10, 9, 186,
11.1, Table 11.1 211
Qi;r Line ratio of analyte line 1 11.1, (11.2); 11.3, 208, 216
D E i to reference line r (11.22)
i;r
Qm;j .k/ Line ratio averaged over 1 11.1, (11.4) 212
Nwu ;Npp ;Nmp
Nmp measuring pulses
after Nwu warming-up
pulses and Npp
prepulses at a single
D location k
E
i;r
Qm;j .k/ As previous symbol but 1 11.1, (11.5) 212
Nwu ;Npp ;Nmp K
averaged over K
locations
r Degree of ionization 1 8.2, Table 8.2, 136, 139
Table 8.3
r Correlation coefficient 1 14.5, Fig 14.60 347
r2 Coefficient of 1 11.5; 13.1.1, 219, 234,
determination Table 13.2; 251,
13.1.3, 267,
Table 13.8; 13.3, 323,
Table 13.16; 14.3, 451,
Table 14.12; 17.3, 462
Table 17.7; 17.4,
Table 17.8
(continued)
(continued)
r1 Radius of plasma core m 10.1, Table 10.1 192
r2 ; r3 Radius of plasma shells m 10.1, Table 10.1 192
rc Radius of the produced m 3.6, (3.7) 34
laser crater
R Resolving power of 1 4.2, (4.3) 51
spectrometer
R Reflectivity 1 5.1, (5.3) 77
R.i / Mean residual of g/g 11.3, (11.21); 13.2, 216, 257,
element i Tables 13.10, 13.11 259
Ri Congruence of element i g/g 13.1.2, (13.1), 239, 245
Table 13.7
RMSEP.i / Root mean square error g/g 18.2, (18.1) 471
of prediction
RSD(c) Relative standard % 14.2.1, Fig. 14.29, 304, 314
deviation of Table 14.11
concentration c
Mass density kg/m3 5.1, (5.2) 75
0 Density of the kg/m3 8.1, (8.1) 125
homogeneous
undisturbed ambient
gas
sanalyte
acc
.tdelay ; tint / Standard deviation of the a.u. 14.1.2, (14.3) 291
accumulated analyte
signal
sb Standard deviation of the 1 11.2, (11.14) 214
measuring values of
the blank sample
s1 Distance between m 4.2, Fig 4.5, (4.5) 52
entrance slit and mirror
sx0 ; sc0 Standard deviation of g/g 11.1, (11.11); 13.3, 213, 267
procedure (13.4)
sy Residual standard 1 11.1, (11.12) 213
deviation
S./ Spectrum J/m 2.1 11
tdelay Delay time between the s 2.1, Fig. 2.4; 11.1, 10, 210
laser pulse and the start Table 11.1
of the integration
window
texc Pulse duration of the s 1, Table 1.1 3
laser beam
tf; ˛ t -Value of a one-sided 1 11.2, (11.14), (11.15), 214, 215
test, f degrees of (11.17)
freedom, ˛ level of
significance
(continued)
508 Annex A
(continued)
tfq Time between the s 3.1, Fig. 3.3 20
ignition of the
flashlamp and the
Q-switching
tint Width of the integration s 2.1, Fig. 2.2, Fig. 2.4; 9, 10,
window 11.1, Table 11.1 211
trep Repetition time of laser s 3.1, (3.1) 21
pulses
8.4, Fig. 8.35 161
T1; 2 Temperature in sphere 1 K 10.1, Table 10.1 192
and shell 2
Te Electron temperature K 5.3, (5.13), (5.14) 80
8.4, (8.14), (8.15) 161
Tg Temperature of the K 10.1, Table 10.1 192
ambient gas assumed
to be stationary
Tge Gas exchange time of the s 4.6, (4.7) 67
measuring chamber
Decay time of plasma s 10.1, (10.1) 191
e; 0 Time for population of s 8.4, Fig 8.35, (8.21) 161, 165
discrete levels
according to
Boltzmann statistics
for neutrals
ee Electron–electron s 8.4, Fig 8.35, (8.18) 161, 164
relaxation time
e; z1 Time for population of s 8.4, (8.21) 165
discrete levels
according to
Boltzmann statistics
for stage of ionization z
with z D 1 for
neutrals, z D 2 for
singly ionized species
ge Local gas exchange time s 3.1, (3.1) 21
8.4, Fig 8.35 161
k0 Electron-neutral s 8.4, (8.19) 164
relaxation time
k1 Electron-ion relaxation s 8.4, (8.20) 164
time
L Temporal width (FWHM) s 3.1, Fig. 3.3; 3.4, 20, 28,
of the laser pulse (3.6) 22
7.4, (7.1); 8.4, 114, 161,
Fig 8.35; 11.1, 209
Table 11.1
(continued)
(continued)
Lj Temporal width (FWHM) s 3.2, Fig 3.9; 11.1, 25, 209
of the laser pulse j in a Table 11.1
pulse burst
mod Time scale of multiple s 8.4, Fig 8.35 161
pulses
Plasma Life time of plasma s 8.4, Fig 8.35 161
s Optical thickness of a 1 9.3, (9.16) 172
plasma layer
0 Beam divergence angle rad 3.4, (3.4) 28
of a Gaussian beam
‚ Beam divergence angle rad 3.4, (3.5) 28
(full angle) of the laser
beam
uaz .T; Ea;z
ion
/ Partition function of 1 9.2, (9.3), (9.6); 10.2, 169, 193,
species a with charge z (10.5); 11.1, (11.3) 211
as a function of
temperature and
reduction of ionization
energy
uFe0 ; uFe1 ; uN0 ; uN1 Partition function of 1 10.4, (10.25) 203
neutral, singly ionized
iron; neutral, singly
ionized nitrogen
uS Partition function of the 1 8.2, (8.7) 136
plasma state generated
with SP
uD Partition function of the 1 8.2, (8.7) 136
plasma state generated
with DP
Uosc Charging voltage of the V 3.1, Fig 3.3 20
capacitors driving the
flashlamp
Upc1 Bias voltage at V 3.1, Fig 3.3 20
Pockels cell
UPMT Signal of a PMT V 3.5, Fig 3.13 30
3
VC Volume of measuring m 4.6, (4.7) 67
chamber
VPin Inlet gas flow rate m3 =s 4.6, Fig 4.17, (4.7) 67
VPmc Gas flow rate in the m3 =s 11.1, Table 11.1 211
measuring chamber
Vp Plasma volume m3 8.3, (8.11) 154
9.2, (9.10) 171
Vx0 Relative standard % 11.1, (11.13) 213
deviation of procedure
w0 Radius of beam waist m 3.4, (3.3) 27
(continued)
510 Annex A
(continued)
wb Radius of the beam waist m 3.6, (3.7), (3.8) 34, 37
on the surface of the
specimen
wm measured full Stark width Å A.4, Table A.1 512, 514
of the line at the given
wavelength
wth theoretically calculated Å A.4, Table A.1 512, 514
full Stark width of the
line at the given
wavelength
We Electron energy eV 8.4, (8.18) 164
WFWHM Stark broadening m 9.4, (9.19) 173
parameter for the full
width (FWHM)
x Concentration of g/g 11.2, (11.6) 212
reference sample
xID Limit of identification g/g 11.2, (11.17) 215
xLOD Limit of detection g/g 11.2, (11.15), (11.16), 215
(11.17)
xLOQ Limit of quantification g/g 11.2, (11.18) 215
yO Prediction of measurand 1 11.2, (11.6), (11.7), 212
by regression curve (11.8)
yO 0 Sensitivity as first g/g 11.2, (11.9) 213
derivative of the
regression curve at the
middle of the working
range
yNb Arithmetic average of n 1 11.2, (11.14) 214
measurements of the
blank sample
yc Critical value 1 11.2, (11.14) 214
Parameter describing the 1 8.1, (8.1) 125
geometry of the energy
release and the shock
wave expansion
z Ionization stage 1 8.4, (8.17) 163
9.1, (9.1) 167
zR Rayleigh length m 3.4, Fig 3.12 29
3.6 33
Z Ion charge 1 5.3, (5.13) 80
Skewness 1 18.3, (18.2) 481
./ Spectral characteristics of 1 10.1, Table 10.1, 192, 197
the observation (10.24)
channel
A.6 Spectral Line Selection for Boltzmann Plots 511
A.4 Stark Data
Table A.1 lists a selection of Stark data for various atom and ion emission lines
which are useful for diagnostic purposes of laser-induced plasmas (cf. Sect. 9.4). If
several values are stated in a reference, then the average value is taken. In [A.4],
Stark parameters are calculated for an electron density of 1016 cm3 . The calculated
widths were multiplied by 2 to obtain the full widths, such as the experimental
values given in, e.g., [A.1,A.3]. The experimental values taken from [A.2] and [A.9]
were also multiplied by 2 to get the full width. References [A.10–A.20] give further
information on Stark data as well as Table 10.2 in Sect. 10.2.
A.5 Wavelengths and Elements
The following compilation of wavelengths and elements provides an overview of

those lines investigated for LIBS and described in this book. They were used,
e.g., to determine calibration curves, analysis functions, detection limits, standard
deviations of procedure, background equivalent concentrations, coefficients of
determination, plasma temperatures, electron densities, self-absorption, temporal
behavior of analyte emission. Table A.2 is arranged in ascending order of wave-
lengths starting with vacuum ultraviolet lines and ending in the near infrared with
a ratio of the longest to the shortest wavelength of 6.4, i.e., more than two octaves.
Table A.3 lists the studied elements in alphabetic order and allocates the observed
emission wavelengths. In this way, both tables allow for an easy access to identify
lines, their features, and usage. These tables are not intended as a comprehensive
overview of emission lines studied so far for LIBS. The reader is referred to
the respective literature given in the previous chapters and data bases as, e.g.,
[13.30, A.21]. With the help of Table A.2, these wavelengths are linked to the
respective matrices investigated.
A.6 Spectral Line Selection for Boltzmann Plots
An important method to determine the temperature of laser-induced plasmas is

based on the Boltzmann plot, see, e.g., Sect. 8.2, Fig 8.15 and Sect. 13.2, Fig 13.22.
The choice of appropriate spectral lines is often based on experience and reference
values and can even be restricted by spectroscopic or other hardware equipments.
The development of spectrometers with large wavelength ranges, e.g., echelle
spectrometers as described in Sect. 4.2, enables the simultaneous detection and
acquisition of spectral lines with resolutions better than 20 pm, see Table 4.4 [A.22,
A.23]. In the following a basic, quantitative, and objective method for the selection
of spectral lines needed for temperature determinations by Boltzmann plots is
512
Table A.1 List of Stark data

El IS (nm) Cs (Å 1016 cm3 ) wm ; wth .Å/ Ds (Å1016 cm3 / d(Å) T(K) ne .1017 cm3 / A Ref.
Al II 199:053 0:0440 0:044 n.i. 10,500 0:1 n.i. [A.2]
Al II 263:155 0:0640 0:064 n.i. 10,500 0:1 n.i. [A.2]
Al II 266:917 0:0062 0:0062 n.i. 10,500 0:1 n.i. [A.2]
Al II 281:618 0:0424 0:0424 n.i. 10,500 0:1 n.i. [A.2]
Al II 390:068 0:0100 0:01 n.i. 10,500 0:1 n.i. [A.2]
Al I 308:215 0:0398 0:51 0:0148 0.19 13,200 1:28 n.i. [A.1]
Al I 309:271 0:0398 0:51 0:0469 0.6 13,200 1:28 n.i. [A.1]
Al I 394:401 0:0336 0:840 0:0168 0.42 11,700 2:50 n.i. [A.3]
Al I 396:152 0:0336 0:840 0:0168 0.42 11,700 2:50 n.i. [A.3]
Al II 466:305 0:1206 0:121 0:0651 0.07 10,500 0:10 n.i. [A.2]
Al II 559:323 0:3800 0:380 n.i. 10,500 0:10 [A.2]
Al II 704:206 0:1930 1:930 0:0690 0.69 18,000 1:00 [A.3]
Al II 705:656 0:1870 1:870 0:0710 0.71 18,000 1:00 [A.3]
Al II 706:362 0:1900 1:900 0:0680 0.68 18,000 1:00 [A.3]
Ar II 295:539 0:0900 0:900 n.i. 28,500 1:00 n.i. [A.3]
Ar I 415:859 0:1952 1:210 n.i. 11,900 0:62 n.i. [A.3]
Ar I 425:936 0:2323 1:440 n.i. 11,900 0:62 n.i. [A.3]
Ar I 425,900a 0:2600 0:260 0:1320 0.13 10,000 0:10 0.069 [A.4]
Ar I 427:217 0:1903 1:180 n.i. 11,900 0:62 n.i. [A.3]
Ar II 434:81 0:0197 0:197 1:00 [A.3]
Ar II 457:935 0:0304 0:210 0:0087 0.06 12,000 0:69 n.i. [A.3]
Ar I 470:232 0:2136 1:410 n.i. 12,000 0:66 n.i. [A.3]
Ar II 514:531 0:5449 3:760 0:1957 1.35 12,000 0:69 n.i. [A.3]
Ar I 750:386 1:16 0:1 [A.5]
Ar I 751:465 1:01 0:1 [A.5]
C I 165:7a 0:0017 0:0165 n.i. 12,500 1:00 n.i. [A.1]
Annex A
C I 165:7a 0:0035 0:00348 0:0021 0.0021 10,000 0:10 0.031 [A.4]
C I 165:812 0:0036 0:060 0:0020 0.0325 14,000 1:65 n.i. [A.1]
C I 193:027 0:0044 0:0044 0:0027 0.0027 10,000 0:10 0.03 [A.1]
C I 193:091 0:0050 0:0300 0:0028 0.017 11,700 0:60 n.i. [A.1]
0:0051 0:1120 0:0026 0.057 15,500 2:20 n.i. [A.1]
Cl I 725:662 0:1378 0:510 n.i. 10,000 0:37 n.i. [A.3]
n.i. 0:0642 0.17 9,500 0:27 n.i. [A.3]
Cl I 754:707 0:1486 0:550 n.i. 10,000 0:37 n.i. [A.3]
n.i. 0:0642 0.17 9,500 0:27 n.i. [A.3]
Cu I 510:554 0:0430 0:430 0:0067 0.07 10,000 1:00 n.i. [A.3]
Fe I 373:487 0:0090 0:090 8,000 1:00 [A.6]
Fe I 376:554 0:0090 0:090 8,000 1:00 [A.6]
A.6 Spectral Line Selection for Boltzmann Plots
Fe I 376:719 0:0090 0:090 8,000 1:00 [A.6]

Fe I 378:788 0:0090 0:090 8,000 1:00 [A.6]
Fe I 381:584 0:0140 0:140 8,000 1:00 [A.6]
Fe I 492:050 0:1070 0:107 n.i. 8,000 0:10 n.i. [A.1]
Fe I 536:7 0:06–0:14 7,400–9,500 [A.7]
Fe I 538:337 0:2120 0:212 n.i. 9,500 0:10 n.i. [A.8]
Fe I 538:3 0:08–0:2 6,600–10,000 [A.7]
Fe I 540:4 0:07–0:12 7,400–9,500 [A.7]
Fe I 542:407 0:1300 0:130 n.i. 8,000 0:10 n.i. [A.1]
Hg I 435:834 0:0280 0:03 n.i. 6,000 0:1 n.i. [A.3]
K I 766:490 0:1240 1:24 0:0275 0.28 20,000 1:0 n.i. [A.1]
(continued)
513
514
Table A.1 (continued)

El IS (nm) Cs (Å 1016 cm3 ) wm ; wth .Å/ Ds (Å 1016 cm3 / d(Å) T(K) ne .1017 cm3 / A Ref.
Mg I 291.545 0.0500 0.50 n.i. 10,000 1.0 n.i. [A.1]
N I 149.263 0.0003 0.00034 n.i. 12,350 0.1 n.i. [A.1]
0.0003 0.00303 0.0002 0.0016 13,000 1.0 n.i. [A.1]
N I 149:3a 0.0025 0.0025 0.0015 0.0015 10,000 0.1 0.029 [A.4]
N I 790.45 2.4000 2.4 0.9000 0.9 10,000 0.1 [A.9]
O I 130:4a 0.0016 0.00164 0.0010 0.001 10,000 0.1 0.028 [A.4]

O I 777:3a 0.0630 0.0630 0.0143 0.0143 10,000 0.1 0.012 [A.4]
P I 178:0a 0.0097 0.0097 0.0053 0.0053 10,000 0.1 0.043 [A.4]
Pb I 405.781 0.0620 0.620 n.i. 11,600 1.0 n.i. [A.1]
a
Si I 288:2 0.0128 0.0128 0.0079 0.0079 10,000 0.1 0.033 [A.4]
a
Average wavelength of a multiplet
El D element, IS D ionization stage, Cs D Stark broadening parameter, wm D measured full width of the line at the given electron density, wth D theoretically
calculated full width of the line at the given electron density, Ds D Stark shift parameter (calculated with d=ne /, d D Stark shift, positive if the wavelength is
increased, T D temperature of the plasma, A D ion broadening parameter (cf. Sect. 9.4, (9.19), (9.20)), n.i. D no information available
Annex A
Table A.2 Wavelengths and elements studied for LIBS in this book, allocation to chapters,
sections, figures and tables
Wavelength Element Matrix Chapter(s), Section(s), Page(s)
(nm) Figure(s), Table(s)
130.22 OI Inclusions/steel 18.3, Fig. 18.14 477
134.72 Cl I Cement 14.4, Fig. 14.44, 331, 332, 334,
Fig. 14.46, Fig. 14.49, 335, 336
Fig. 14.50, Fig. 14.51,
Fig. 14.52
149.26 NI Inclusions/steel 18.3, Fig. 18.15, 478, 479
Fig. 18.16
165.81 CI Steel 17.1, Table 17.3 438
178.28 PI Steel, 3.5, Fig. 3.15; 13.1.1, 32, 233, 234,
segregation/steel Fig. 13.4, Table 13.2; 240, 251, 483
13.1.2, Table 13.6;
13.1.3, Fig. 13.15,
Table 13.8; 18.3,
Fig. 18.22
180.73 SI Steel, 3.5, Fig. 3.15; 13.1.1, 32, 234, 240,
segregation/steel Fig. 13.5, Table 13.2; 251, 397, 483
13.1.2, Table 13.6;
13.1.3, Table 13.8;
15.1, Fig. 15.9; 18.3,
Fig. 18.22
187.75 Fe II Steel 3.5, Fig. 3.15; 3.8, 32, 40, 233, 234,
Fig. 3.24; 13.1.1, 241, 242, 250,
Fig. 13.3, Fig. 13.4, 251
Fig. 13.5; 13.1.2,
Fig. 13.8, Fig. 13.9;
13.1.3, Fig. 13.14,
Fig. 13.15, Fig. 13.16
193.09 CI Steel, 3.8, Fig. 3.24; 13.1.1, 40, 232, 233,
inclusions/steel Fig. 13.2, Fig. 13.3, 234, 240, 241,
Table 13.2; 13.1.2, 242, 250, 251,
Table 13.6, Fig. 13.8, 396, 438
Fig. 13.9; 13.1.3,
Fig. 13.14, Table 13.8;
15.1, Fig. 15.8; 17.1,
Table 17.3
194.23 Hg II Mercury lamp 17.4, Fig. 17.29 457
212.41 Si I Steel 6.1, Fig. 6.5; 15.1, 91, 396, 398
Fig. 15.8, Fig. 15.10
213.86 Zn I Aerosol 14.8.1, Fig. 14.84, 369, 371
Fig. 14.88
225.39 Ni II Steel, high-alloy 13.1.2, Table 13.6; 13.2, 240, 257, 258
steel Table 13.10, Fig. 13.19
(continued)
516 Annex A

228.80 Cd I Soil, polymer 12, Fig. 12.5, Fig. 12.6; 226, 227, 295,
14.1.2, Table 14.6, 297, 298, 299,
Fig. 14.16, Fig. 14.18, 303, 323
Table 14.7, Table 14.8;
14.3, Table 14.12
231.60 Ni II Steel, soil 13.1.1, Table 13.2; 13.1.3, 234, 251, 323
Fig. 13.16, Table 13.8;
14.3, Table 14.12
234.98 As I Soil 14.3, Table 14.12 323
247.86 CI Polymer, air, 14.1.1, Fig. 14.2; 14.6, 279, 352, 353,
grease/aluminum, Fig. 14.63, Fig. 14.64, 354, 368, 463
particulate/top Fig. 14.65, Fig. 14.66,
gas Fig. 14.67; 14.8.1,
Fig. 14.83; 17.4,
Fig. 17.37
251.43 Si I Slag 14.2.2, Fig. 14.33, 317, 319, 473
Fig. 14.34, Fig. 14.37;
18.2, Fig. 18.9
251.61 Si I High-alloy steel 13.2, Fig. 13.20 258
253.65 Hg I Polymer, soil 14.1.2, Table 14.6, 295, 299, 303,
Table 14.7, Table 14.8; 323
14.3, Table 14.12
257.51 Al I Aluminum, 14.8.1, Fig. 14.82, 367, 368
grease/aluminum Fig. 14.83
259.80 Sb I Polymer 14.1.2, Table 14.6, 295, 299, 303
Table 14.7, Table 14.8
259.94 Fe II Aluminum 13.3, Table 13.15, 266, 267, 269,
Table 13.16, 271
Table 13.17, Fig. 13.25
263.82 Mn II Inclusions/steel 15.2, Fig. 15.9; 18.3, 397, 476
Fig. 18.12
267.72 Cr II High-alloy steel 11.5, Fig. 11.6; 13.1.1, 219, 234, 257,
steel, aluminum Table 13.2; 13.2, 261, 266, 267,
Table 13.10, 269, 270, 443,
Table 13.13; 13.3, 444
Table 13.15,
Table 13.16,
Table 13.17,
Table 13.18; 17.2,
Table 17.5, Table 17.6
271.44 Fe II High-alloy steel 11.5, Fig. 11.5; 13.1.1, 218, 231, 258,
steel, Table 13.1; 13.2, 360
particulate/top Fig. 13.19, Fig. 13.20;
gas 14.7, Fig. 14.75
(continued)

273.07 Fe II Steel 3.5, Fig. 3.13; 6.1, 30, 88, 91, 92,
Fig. 6.3, Fig. 6.5, 439, 444
Fig. 6.6; 17.1,
Fig. 17.12; 17.2,
Table 17.6
273.36 Fe I High-alloy steel 13.2, Table 13.14 262
279.08 Mg II Aluminum 9.4, Fig. 9.10; 13.3, 181, 266, 267,
Table 13.15, 269
Table 13.16,
Table 13.17
279.55 Mg II Aluminum, water 9.4, Fig. 9.10; 14.5, 181, 349, 451
Fig. 14.62; 17.3,
Table 17.7
281.62 Mo II Steel, high-alloy 13.1.2, Table 13.6; 240, 241, 257,
steel Fig. 13.8; 13.2, 444
Table 13.10; 17.2,
Table 17.6
286.26 Cr II High-alloy steel 11.5, Fig. 11.5; 13.2, 218, 257, 259
Table 13.10,
Table 13.11
288.16 Si I Aluminum steel 11.2, Fig. 11.2; 13.1.1, 214, 234, 266,
Table 13.2; 13.3, 267, 269, 270,
Table 13.15, 450, 451
Table 13.16,
Table 13.17, Fig. 13.24,
Table 13.18; 17.3,
Fig. 17.22, Table 17.7
289.36 Hg I Mercury lamp 17.4, Fig. 17.28 457
293.31 Mn II Steel, pure iron, 6.1, Fig. 6.4, Fig. 6.6; 89, 92, 234, 240,
aluminum 13.1.1, Table 13.2; 266, 267, 269,
13.1.2, Table 13.6; 270, 398, 439
13.3, Table 13.15,
Table 13.16,
Table 13.17,
Table 13.18; 15.1,
Fig. 15.10; 17.1,
Fig. 17.12
294.92 Mn II Aluminum 17.3, Table 17.7 451
296.12 Cu I Aluminum 13.3, Table 13.15, 266, 269, 270
Table 13.17,
Table 13.18
298.92 Cr II High-alloy steel 17.2, Table 17.6 444
(continued)
518 Annex A

305.71 Al I Aluminum 11.2, Fig. 11.2; 13.3, 214, 266, 270,
Table 13.15, Fig. 13.24, 271
Fig. 13.25
307.21 Zn I Zinc-coated steel 17.3, Fig. 17.25 454
308.22 Al I Inclusions/steel 18.3, Fig. 18.13, 477, 478, 479,
Fig. 18.15, Fig. 18.16, 480, 481
Fig. 18.17, Fig. 18.18,
Fig. 18.21
309.27 Al I Slag 14.2.2, Fig. 14.37; 17.3, 319, 450
Fig. 17.22
324.75 Cu I Copper, steel, 7.3, Fig. 7.7; 13.1.2, 105, 238, 240,
high-alloy steel, Table 13.3, Table 13.6; 256, 257, 322,
soil, water, 13.2, Fig. 13.18, 323, 348, 371,
aerosol, Table 13.10; 14.3, 450, 451, 483
aluminum, Fig. 14.38, Table 14.12;
segregation/steel 14.5, Fig. 14.61;
14.8.1, Fig. 14.87;
17.3, Fig. 17.21,
Fig. 17.22, Table 17.7;
18.3, Fig. 18.22
327.40 Cu I Aluminum 3.9, Fig. 3.25; 13.3, 42, 266, 267,
Table 13.15, 269, 450
Table 13.16,
Table 13.17; 17.3,
Fig. 17.21
328.23 Zn I Zinc-coated steel 16.1, Table 16.2, 410, 411, 414
Table 16.3, Fig. 16.9
334.50 Zn I Aluminum, soil, 13.3, Table 13.15; 14.3, 266, 322, 323,
particulate/top Fig. 14.38, Table 14.12; 358, 360, 451
gas 14.7, Fig. 14.72,
Fig. 14.74; 17.3,
Table 17.7
334.94 Ti II Aluminum 17.3, Table 17.7 451
337.28 Ti II Aluminum 13.3, Table 13.15, 266, 267, 269,
Table 13.16, 270
Table 13.17,
Table 13.18
341.48 Ni I Aluminum 13.3, Table 13.15, 266, 267, 269
Table 13.16,
Table 13.17
351.92 Tl I Soil 14.3, Table 14.12, 325, 326
Fig. 14.40, Fig. 14.41
356.54 Fe I High-alloy steel 13.2, Table 13.14, 262, 263
Fig. 13.22
357.58 Zr I Aluminum 17.3, Table 17.7 451
358 CN Ammonium nitrate 17.4, Fig. 17.37 463
Fig. 13.22
(continued)

364.44 Ca I Slag 14.2.2, Fig. 14.36 319
Fig. 13.22
368.35 Pb I Aerosol 14.8.1, Fig. 14.85, 369, 370
Fig. 14.86
371.99 Fe I High-alloy steel, 13.2, Table 13.14, 262, 263, 266,
aluminum, steel Fig. 13.22; 13.3, 269, 394, 398,
Table 13.15, 445, 446
Table 13.17; 15.1,
Fig. 15.6, Fig. 15.10;
17.2, Fig. 17.16,
Fig. 17.17
373.83 Fe I Aluminum 17.3, Table 17.7, 451, 454,
Fig. 17.25
Fig. 13.22
381.30 Fe I Steel 9.4, Fig. 9.6, Fig. 9.7 177, 178
381.58 Fe I Steel, aerosol 9.4, Fig. 9.6, Fig. 9.7; 177, 178, 369,
14.8.1, Fig. 14.85, 370
Fig. 14.86
382.78 Fe I Steel 9.4, Fig. 9.6, Fig. 9.7 177, 178
383.83 Mg I Aluminum 17.3, Table 17.7 451
384.99 Fe I Steel 9.4, Fig. 9.6, Fig. 9.7 177
387.1 CN Ammonium nitrate 17.4, Fig. 17.37 463
390.55 Si I Soil, aluminum, 12, Fig. 12.5; 13.3, 226, 266, 269,
slag Table 13.15, 309, 310, 318
Table 13.17; 14.2.1,
Fig. 14.27, Fig. 14.28;
14.2.2, Fig. 14.35
393.37 Ca II Water, aerosol, 14.5, Fig. 14.60; 14.7, 347, 355, 373,
particulate/top 14.8.2, Fig. 14.89, 376, 378, 379
gas Fig. 14.91, Fig. 14.93,
Fig. 14.94
394.40 Al I Aluminum 9.4, Fig. 9.6, Fig. 9.11; 177, 182, 202,
10.2, Fig. 10.6; 17.4, 458
Fig. 17.31
396.15 Al I Steel, zinc-coated 13.1.2, Table 13.6; 16.1, 240, 415, 420,
steel, ZnAl Fig. 16.10; 16.2, 422, 423, 424,
binary alloy, Fig. 16.13, Fig. 16.14, 425, 458, 477,
aluminum, Fig. 16.15, Fig. 16.16, 481
inlcusions/steel Fig. 16.17; 17.4,
Fig. 17.31; 18.3,
Fig. 18.13, Fig. 18.21
396.85 Ca II Aerosol 14.8.2, Fig. 14.95 380
Fig. 13.22
(continued)
520 Annex A

405.78 Pb I Polymer, soil, 14.1.2, Fig. 14.9, 290, 293, 294,
particulate/top Fig. 14.10, Fig. 14.11, 295, 299, 301,
gas Fig. 14.12, Fig. 14.13, 303, 323, 358
Table 14.6, Fig. 14.14,
Fig. 14.15, Table 14.7,
Fig. 14.21, Fig. 14.22,
Table 14.8; 14.3,
Table 14.12; 14.7,
Fig. 14.72
Fig. 13.22
Fig. 13.22
410.98 Fe I Zinc-coated steel 16.1, Table 16.3, Fig. 16.9 411, 414
Fig. 13.22
Fig. 13.22
Fig. 13.22
Fig. 13.22
422.67 Ca I Aerosol 14.8.1, Fig. 14.84 369
425.43 Cr I Polymer, soil 14.1.2, Table 14.6, 295, 299, 303,
Table 14.7, Table 14.8; 323
14.3, Table 14.12
427.18 Fe I Pure iron, 8.2, Fig. 8.24; 13.2, 150, 262, 263
high-alloy steel Table 13.14, Fig. 13.22
430.01 Ar I Gas 8.2, Fig. 8.24 150
430.79 Fe I Pure iron, 8.2, Fig. 8.24; 13.2, 150, 262, 263
high-alloy steel Table 13.14, Fig. 13.22
432.58 Fe I Pure iron, 8.2, Fig. 8.24, Fig. 8.26; 150, 152, 262,
high-alloy steel 13.2, Table 13.14, 263
Fig. 13.22
433.12 Ar II Gas 8.2, Fig. 8.24 150
434.52 Ar I Gas 8.2, Fig. 8.24 150
434.81 Ar II Gas 8.2, Fig. 8.24, Fig. 8.25; 150, 151, 395
15.1, Fig. 15.7
438.35 Fe I Steel, high-alloy 9.4, Fig. 9.5; 13.2, 176, 262, 263,
steel, zinc-coated Table 13.14, Fig. 13.22; 406, 407, 408,
steel 16.1, Fig. 16.5, 410, 411, 413,
Fig. 16.6, Fig. 16.7, 414, 424, 425
Table 16.2, Table 16.3,
Fig. 16.8, Fig. 16.9;
16.2, Fig. 16.16,
Fig. 16.17
(continued)

460.96 Ar II Gas 15.1, Fig. 15.7 395
472.22 Zn I Zinc-coated steel, 16.1, Fig. 16.5, Fig. 16.6, 406, 407, 408,
ZnAl binary alloy Fig. 16.7, Table. 16.2, 410, 411, 413,
Table. 16.3, Fig. 16.8, 414, 415, 420,
Fig. 16.9, Fig. 16.10; 422, 423, 424,
16.2, Fig. 16.13, 425
Fig. 16.14, Fig. 16.15,
Fig. 16.16, Fig. 16.17
473.59 Ar II Gas 15.1, Fig. 15.7 395
476.49 Ar II Gas 15.1, Fig. 15.7 395
486.13 HI Polymer 14.1.1, Fig. 14.2 279
Fig. 13.22
531.66 Fe II Steel 10.2, Fig. 10.5 198
532.42 Fe I Steel 8.2, Fig. 8.13, Fig. 8.16, 134, 138, 139,
Table 8.3; Fig. 8.19, 142, 143, 144,
Fig. 8.20, Fig. 8.21; 198
10.2, Fig. 10.5
532.80 Fe I Steel 6.3, Fig. 6.7; 8.2, 93, 134,
Fig. 8.13, Fig. 8.16; 138, 198
10.2, Fig. 10.5
533.99 Fe I Steel 10.2, Fig. 10.5 198
534.6 Continuum Steel 8.2, Fig. 8.22 145
536.29 Fe II Steel 10.2, Fig. 10.5 198
537.15 Fe I Steel 8.2, Fig. 8.13, Fig. 8.14; 134, 198
10.2, Fig. 10.5
538.34 Fe I Steel 8.2, Fig. 8.13, Fig. 8.14, 134, 145, 158,
Fig. 8.22; 8.3, Fig. 8.33; 168, 176, 179,
9.1, Fig. 9.1; 9.4, 180, 198
Fig. 9.4, Fig. 9.5,
Fig. 9.8, Fig. 9.9; 10.2,
Fig. 10.5
539.71 Fe I Steel 8.2, Fig. 8.13; 10.2, 134, 198
Fig. 10.5
540.58 Fe I Steel 8.2, Fig. 8.13 134
589.00 Na I Water, 14.5, Fig. 14.59; 14.7, 346, 360, 376,
particulate/top Fig. 14.74; 14.8.2, 377, 378
gas, aerosol Fig. 14.91, Fig. 14.92,
Fig. 14.93
589.59 Na I Water 14.5, Fig. 14.59 346
656.28 HI Propane/carbon 14.6, Fig. 14.67, 354, 355, 359,
dioxide, Fig. 14.68; 14.7, 463
particulate/top Fig. 14.73; 17.4,
gas, ammonium Fig. 17.37
nitrate
725.66 Cl I Polymer 14.1.1, Fig. 14.6 283
(continued)
522 Annex A

742.36 NI Air, ammonium 14.6, Fig. 14.63; 17.4, 352, 463
nitrate Fig. 17.37
744.23 NI Air 14.6, Fig. 14.63 352
746.83 NI Air, particulate/top 14.6, Fig. 14.63 352
gas, inclusions
754.71 Cl I Polymer 14.1.1, Fig. 14.6 283
766.49 KI particulate/top gas 14.7 355
777.19 OI Air/carbon dioxide, 14.6, 14.7, Fig. 14.63, 352, 353, 354,
particulate/top Fig. 14.65, Fig. 14.66, 355, 463, 477,
gas, Fig. 14.68; 17.4, 478, 479,480
propane/carbon Fig. 17.37; 18.3,
dioxide, Fig. 18.14, Fig. 18.15,
ammonium Fig. 18.17, Fig. 18.18
nitrate,
inclusions/steel
777.42 OI Air/carbon dioxide, 14.6, Fig. 14.63, 352, 353, 354,
propane/carbon Fig. 14.65, Fig. 14.66, 355
dioxide Fig. 14.68
777.54 OI Air/carbon dioxide, 14.6, Fig. 14.63, 352, 353, 354,
propane carbon Fig. 14.65, Fig. 14.66, 355
dioxide Fig. 14.68
827.24 Br I Polymer 14.1.2, Table 14.6, 295, 298, 299
Fig. 14.17, Fig. 14.19,
Table 14.7
837.59 Cl I Cement 14.4, Fig. 14.45, 331, 334, 335
Fig. 14.48, Fig. 14.50
described [A.24]. This approach implicates advantages especially in those cases

where no spectral line experiences or empirical values are available for a specific
diagnostic task. Additionally, even for comparatively low laser pulse energies, larger
sets of spectral lines can be determined by a systematic approach, which effect a
decrease in the relative error of the electron temperature to be determined. Thus, the
temporal characterization of the LIBS plasma can be improved and well grounded.
Low laser pulse energies are, e.g., used for microanalysis as, e.g., those described in
[A.25–A.27] and Chaps. 15, 16.
The experiments are carried out with a flashlamp pumped Q-switched Nd:YAG
laser at the fundamental wavelength D 1;064 nm and a repetition rate of
rep D 20 Hz. The laser beam diameter amounts to 6 mm. The pulse duration is
about L D 5 ns. The pulse energy is attenuated by a Glan-laser-prism and four
different laser pulse energies, e.g., 0.2, 0.5, 1, and 2 mJ, are studied. This pulse
energy regime is of interest for the design of compact LIBS instrumentation, where
the use of more powerful laser sources is not possible or for LIBS in combination
with laser micromachining or LIBS microanalysis [A.28,A.29]. All experiments are
carried out in air at atmospheric pressure. The samples are certified C75 steel alloy
Table A.3 Elements in alphabetic order and emission wavelengths

Element Wavelength(s) (nm)
Aluminum 257.51, 305.71, 308.22, 309.27, 394.40, 396.15
Antimony 259.80
Argon 430.01, 433.12, 434.52, 434.81, 460.96, 473.59, 476.49
Arsenic 234.98
Bromine 827.24
Cadmium 228.80
Calcium 364.44, 393.37, 396.85
Carbon 165.81, 193.09, 247.86
Chlorine 134.72, 725.66, 754.71, 837.59
Chromium 267.72, 286.26, 298.92
Copper 296.12, 324.75, 327.40
Hydrogen 486.13, 656.28
Iron 187.75, 259.94, 271.44, 273.07, 273.36, 278.81, 296.69, 356.54, 363.15,
364.78, 371.99, 373.83, 374.95, 381.30, 381.58, 382.78, 384.99, 404.58,
406.36, 407.17, 410.98, 411.85, 413.47, 419.91, 421.94, 422.67, 425.43,
427.18, 430.79, 432.58, 438.35, 522.72, 531.66, 532.42, 532.80, 533.99,
536.29, 537.15, 538.34, 539.71, 540.58
Lead 368.35, 405.78
Magnesium 279.08, 279.55, 383.83
Manganese 263.82, 293.31, 294.92
Mercury 194.23, 253.65, 289.36
Molybdenum 281.62
Nickel 225.39, 231.60, 341.48
Nitrogen 149.26, 742.36, 744.23, 746.83
Oxygen 130.22, 777.19, 777.42, 777.54
Phosphor 178.28
Silicon 212.41, 251.43, 251.61, 288.16, 390.55
Sodium 589.00, 589.59
Sulfur 180.73
Thallium 351.92
Titanium 334.94, 337.28
Zinc 213.86, 307.21, 328.23, 334.50, 472.22
Zirconium 357.58
samples (German designation 1.1750) with low concentrations of C (0.78 m.-%), Cr

(0.17 m.-%), Si, (0.28 m.-%), Mn (0.7 m.-%), P (0.009 m.-%), and S (0.005 m.-%).
The sample surface was put in a vertical plane and the laser beam was focused
horizontally at right angle onto the sample surface by a planoconvex achromatic
lens (f D 50 mm). The beam waist diameter is about 2w0 D 23 m and the average
irradiance at the sample surface amounts to If D 1010 ; 2:5 1010 ; 5 1010 , and
1011 W=cm2 for the different pulse energies. The sample is moved by linear axes
perpendicularly to the laser beam axis with a speed of 2 mm/min. The plasma is
imaged with a reduction ratio of 1 : 0.7 using a concave mirror into an optical fiber
with a core diameter of 800 m connected to the spectrometer. The spectrometer
is an echelle system with a spectral resolution of 1–22 pm in the wavelength range
524 Annex A
between 200 and 780 nm, cf. Sect. 4.2, Table 4.4. The detection system consisting
of spectrometer and optical fiber uses an internal response correction function to
achieve a wavelength independent response within its wavelength range. Due to
the small emission intensities at laser pulse energies of only a few millijoule 100
consecutive LIBS spectra are accumulated in this wavelength range. To study the
time evolution of the plasma, the line emission is investigated at nine different delay
times tdelay D 100 900 ns with an integration time of tint D 100 ns each. The upper
limit for the delay time has to be chosen because of the fast decreasing spectral line
intensity of the Fe II species. In the case of Eb D 200 J, the 1/e-decay time of the
intensity is about 260 ns. For times up to tdelay D 900 ns, a sufficient spectral line
intensity can be assured for all studied pulse energies. Thus, for every pulse energy
nine spectra corresponding to the nine different time windows were recorded so that
in total 36 spectra were obtained for the four laser pulse energies.
Due to the comparatively low laser pulse energies used low emission line inten-
sities are expected. In order to assure a minimum error of the determined electron
temperatures, a procedure to find appropriate emission lines for the Boltzmann
plots is developed. In a first step of this procedure, a preselection of spectral
lines based on theoretical calculations is carried out. Afterwards, these preselected
lines are assessed within an iterative optimization step using experimental data
in which improper spectral lines are discarded successively. The total number of
all atomic and ionic emission lines of elements contained in the used samples
within the spectral range of 200–780 nm amounts to 93;000 [A.30]. Assuming
local thermal equilibrium (LTE) – see Sect. 8.4 – the electron temperature derived
by the Boltzmann plot method is independent of the considered species within
the plasma. The emission coefficient is proportional to the number density of the
particular species in the plasma, cf. Sect. 9.2, (9.3). Thus, in the following only
the emission lines of the matrix element, which in this case is Fe (>98 m.-%),
are considered to determine the electron temperature using Boltzmann plots. The
total number of Fe emission lines amounts to 38; 000 in the considered spectral
range. Electron temperatures in LIBS plasmas are typically in the range of 5,000–
16,000 K. The excitation probability decreases with the excitation energy according
to the Boltzmann distribution, cf. (9.3). As on average higher ionization stages have
higher excitation energies and lower occurrences in the plasma due to the Saha
equation [cf. Sect. 10.1, (10.5)], and emissions of Fe lines with ionization stages
higher than I and II are negligible for the expected temperature ranges. Hence, the
further considered spectral lines are limited to the first two ionization stages Fe
I and Fe II. The total number of remaining lines for these two species is about
30,000. To reduce this huge number of spectral lines to only those which are
clearly identifiable in the whole spectrum, it is reasonable to develop a criterion
based on the emission coefficients of the spectral lines. For the remaining Fe I
and Fe II spectral lines, the emission coefficients can be calculated according to
the relationship (9.3). The partition function for each species can be obtained for
example from the Atomic Spectral Database of the National Institute of Standards
and Technology [A.21]. The electron temperature Te can be estimated or derived
from preliminary measurements. Here, the following temperatures were chosen:
10-12
Fe I iron emission coefficient e i [a.u.]

10-16
10-20
10-24
10-28
200 300 400 500 600 700 800
wavelength l i [nm]
Fig. A.2 Calculated emission coefficients for all Fe I spectral lines in the considered wavelength
range. The horizontal line represents the threshold value (mean value of all emission coefficients)
for this species. The index i characterizes the individual Fe I emission line
Te D 8;900 K for Fe I emission lines and Te D 11;300 K for Fe II emission lines,

respectively. In order to extract the lines with the highest emission coefficients,
threshold values are calculated for each species, Fe I and Fe II. As particular
threshold value the mean value of all emission lines of the regarded species is
taken. The number density N can be set to a constant value, since only emission
lines of the same species are compared with each other. Figure A.2 shows the half-
logarithmic plot of calculated emission coefficients for Fe I emission lines versus
their wavelength.
After discarding all emission lines with lower emission coefficients than the
chosen threshold value in total 966 Fe I and 585 Fe II emission lines can be identified
as potential candidates for the Boltzmann plots. In a second step, these remaining
spectral lines have to be assessed quantitatively with respect to superposition and
overlapping with other spectral lines.
Spectral lines emitted by the laser-induced plasma are broadened by a variety
of mechanisms such as Stark, Van-der-Waals and Doppler broadening, cf. Sect. 9.4.
Additionally, the apparative broadening of the spectrometer increases the widths of
all spectral lines as well, see, e.g., Figs. 9.6 and 9.7.
Since the Boltzmann plot method relies on the measured emission line intensity,
it is necessary to identify those lines which do not significantly overlap with
other emission lines, e.g., due to line broadening mechanisms. Therefore, in the
next step each of the remaining Fe I and Fe II lines is investigated with respect
to superpositions with other emission lines which can occur in the plasma. This
investigation has to cover all present species and elements in the sample even if the
concentration is comparatively low. In our case for each of the remaining Fe I and
Fe II emission lines a wavelength interval of 200 pm is centered at the transition
wavelength k . The emission coefficients "i of all other spectral lines within this
526 Annex A
wavelength interval are summed up according to

X
Uk D "i : (A.9)
i
for all i with ji k j 100 pm and i ¤ k.

Afterwards, this sum of all emission coefficients in the 100 pm-neighborhood of
the considered Fe emission line is compared with the emission coefficient at k .
Only those emission lines which fulfill the condition
"k > 3 Uk (A.10)
are retained. With this condition, it is assured that each remaining emission line does
not interfere significantly with other spectral lines within the chosen neighborhood.
Even if there are still interfering emission lines, the condition (A.10) ensures that
the disturbing lines have a cumulative emission coefficient of less than a third of the
emission coefficient "k at k . By this preselection for the above-described case 337
Fe I and 96 Fe II spectral lines are obtained as potential candidates for the electron
temperature determination. It might happen that some of these remaining lines have
the same upper energy level. In this case, only those spectral lines are retained with
the highest "k =Uk ratio. All other lines of the same energy level are discarded, since
otherwise particular energy levels would be preferred in the Boltzmann plot. By
eliminating spectral lines with the same upper energy level 199 Fe I and 64 Fe II
emission lines are obtained as an intermediate result after the successive preselection
processes. The average "k =Uk ratios amount to 191.3 for Fe I and 239.7 for Fe II.
As mentioned above, this result shows that even if overlapped lines are not discarded
by the proposed method and used criteria, the contribution of the disturbing lines
to the emission coefficient would theoretically be less than 1% and can therefore
be neglected. Finally, comparing all initially regarded iron lines (9,922 Fe I, 20,162
Fe II) only 0.87% are further considered after the preselection.
After the preselection procedure, the line intensities of these emission lines
have to be determined from experimentally recorded spectra in order to generate
Boltzmann plots. The line profile is divided in three wavelength intervals marked
with A, B, and C. Interval B contains the peak of the line and ranges in-between
the closest local minima on either side of the peak. Adjacent on both sides of
interval B are the intervals A and C with an extent of 10 pm each. To account for
background noise of the detection system the average intensity in the intervals A
and C is subtracted from the line intensity in interval B. The background corrected
net line intensity can then be determined via trapezoidal integration. However,
some of the preselected lines might not be detectable, visible or identifiable due
to various reasons. In the case described here, the echelle spectrometer has gaps
in its detectable wavelength range. Additionally, some lines may disappear in
the continuum radiation emitted by the plasma which was not considered in the
theoretical part as well. After eliminating emission lines which were predicted as
prominent in the spectrum but not present in one of the 36 experimentally recorded
spectra 167 Fe I and 48 Fe II emission lines were clearly identified and allocated.
With the spectral line intensities determined, the electron temperatures were
deduced via Boltzmann plots for every time window and laser pulse energy. The
relative error of the electron temperature averaged over the 36 spectra amounts to
Te =Te D 4:9% (r 2 D 0:72) for Fe I lines and Te =Te D 61:2% (r 2 D 0:06) for Fe
II lines. The coefficients of determination are rather poor, for iron ion lines the
situation is worse compared with iron atomic lines. Some data points are far away
from the regression curve. To eliminate outliers in the Boltzmann plot, an iterative
method to discard spectral lines which do not match with the overall trend given by
the other data points was developed. The goal of this method is to identify a set of
lines for the specific time domain and pulse energy regime based on an algorithm
which considers the residuum of an individual data point in the Boltzmann plot
from the regression line, i.e., its distance in y-direction. For each spectral line,
the following measure is calculated to ensure that the used spectral line has an
emission intensity level high enough to allow detection within the entire specified
time domain and pulse energy regime:
X
2
Sk D Dk;x;y : (A.11)
x;y
The indices x and y are for the different time windows and pulse energies. The
index k denotes the individual spectral lines and Dk;x;y are the deviations between
a data point and the regression line. Sk corresponds to the total sum of the squared
deviations for all time windows and laser pulse energies. Thus, the spectral line
with the maximum value for Sk has on average, e.g., in the considered case for
all four laser pulse energies and for all nine time windows, the largest deviation
from the regression line. The iterative method is based on discarding this line and
on recalculation of the regression function. This procedure is repeated until the
coefficient of determination exceeds a threshold value. By this algebraic approach,
it is ensured that data points which appear as obvious outliers in every time window
are discarded. As break condition for the iterative Boltzmann plot method, an
average threshold value of r 2 D 0:98 is taken.
This iterative method stopped after discarding 106 Fe I lines and 36 Fe II
lines. By this approach, relative errors for the electron temperature of 1.8% with
the remaining 61 Fe I and 4.4% with 12 Fe II emission lines averaged over
all considered nine time windows and laser pulse energies were achieved. These
remaining emission lines for the temperature determination are summarized in
Tables A.4 and A.5.
The development of the average relative error of the electron temperature for
Fe I emission lines by iteration number is shown in Fig. A.3. As can be seen the
averaged relative errors are decreasing by nearly a factor of three in the case of
Fe I emission lines during the iterative process. Due to a couple of outlier emission
lines in the case of Fe II emission lines which are considered in the beginning of
the iteration process, the averaged relative errors are reduced by even a factor of 14.
528 Annex A
Table A.4 Spectroscopic data of selected Fe I emission lines

No. (nm) Anm (1/s) En (eV) gn
1 244.256 2:14E C 08 7.508 11
2 282.328 1:86E C 07 5.349 7
3 289.942 5:92E C 07 6.554 3
4 295.999 4:81E C 07 6.880 13
5 299.039 3:90E C 07 6.873 11
6 301.148 4:70E C 07 6.875 9
7 301.898 1:26E C 07 5.064 7
8 302.403 4:87E C 06 4.209 5
9 303.015 4:79E C 07 6.524 11
10 305.307 1:53E C 07 6.484 5
11 305.526 8:63E C 06 5.615 5
12 307.572 2:94E C 07 4.988 5
13 308.374 3:01E C 07 5.010 3
14 309.158 5:41E C 07 5.021 1
15 328.026 5:38E C 07 7.080 11
16 331.474 6:89E C 07 7.041 7
17 336.955 2:42E C 07 6.406 9
18 337.078 3:29E C 07 6.370 11
19 338.011 2:35E C 07 6.426 7
20 341.313 3:57E C 07 5.830 7
21 342.266 2:79E C 07 5.845 5
22 342.712 5:49E C 07 5.793 9
23 347.545 7:01E C 06 3.654 5
24 349.057 4:47E C 06 3.603 7
25 349.529 9:46E C 06 6.106 7
26 351.382 3:40E C 06 4.387 11
27 360.668 8:24E C 07 6.129 13
28 364.951 4:22E C 07 6.089 9
29 365.147 6:23E C 07 6.153 9
30 367.763 8:01E C 07 6.129 5
31 368.411 3:36E C 07 6.092 7
32 369.073 2:74E C 07 6.932 11
33 370.557 3:21E C 06 3.397 7
34 372.256 4:97E C 06 3.417 5
35 372.438 1:25E C 07 5.607 7
36 376.554 9:69E C 07 6.529 15
37 376.719 6:39E C 07 4.302 3
38 379.955 7:31E C 06 4.221 9
39 380.535 9:81E C 07 6.559 11
40 382.782 1:05E C 08 4.796 5
41 384.326 4:67E C 07 6.272 7
42 384.997 6:05E C 07 4.231 1
43 385.637 4:64E C 06 3.266 5
44 386.552 1:55E C 07 4.218 3
45 386.722 3:39E C 07 6.223 5
46 387.376 8:01E C 06 5.633 9
47 392.026 2:60E C 06 3.283 3
(continued)

48 392.291 1:08E C 06 3.211 9
49 395.116 3:56E C 07 6.411 5
50 397.774 7:01E C 06 5.314 5
51 401.453 2:37E C 07 6.661 11
52 404.581 8:62E C 07 4.549 9
53 406.359 6:78E C 07 4.608 7
54 407.174 7:65E C 07 4.652 5
55 410.749 2:45E C 07 5.850 3
56 421.936 3:80E C 07 6.511 13
57 430.790 3:40E C 07 4.435 9
58 432.576 4:97E C 07 4.474 7
59 436.977 7:22E C 06 5.884 9
60 438.354 5:00E C 07 4.313 11
61 440.475 2:75E C 07 4.372 9
Table A.5 Spectroscopic data of selected Fe II emission lines

1 224.550 2:61E C 08 10.289 10
2 233.131 2:92E C 07 5.549 8
3 234.534 8:46E C 07 7.920 12
4 236.860 5:93E C 07 5.585 4
5 237.519 9:81E C 07 5.605 2
6 238.076 3:14E C 07 5.289 8
7 257.792 1:26E C 08 5.905 2
8 259.154 5:11E C 07 5.824 6
9 261.383 1:99E C 08 4.849 2
10 262.829 8:56E C 07 4.837 4
11 263.957 1:07E C 08 8.036 2
12 271.441 5:48E C 07 5.553 6
Figure A.4 shows the development of the averaged coefficient of determination for
Fe II emission lines.
The coefficient of determination in iteration step 37 exceeds the threshold value
of r 2 D 0:98 so that in total 36 Fe II emission lines are discarded in this case.
The corresponding Boltzmann plots are shown in Figs. A.5 and A.6. The relative
temperature errors are reduced in this particular case to Te =Te D 1:7% for Fe I
and Te =Te D 5:9% for Fe II emission lines. This effect is due to the fact that on
average a certain coefficient of determination is required which in turn is responsible
for the exclusion of emission lines with comparatively high residue to the regression
line. The higher relative error for Fe II emission lines can be related to the lower
number of remaining spectral lines compared with those for Fe I. This result is a
consequence of the fact that ionized species in the plasma relax continuously to
lower ionization stages.
530 Annex A
5.0
averaged relative error ΔTe / Te [%] 4.5 Fe I
4.0
3.5
3.0
2.5
2.0
1.5
0 20 40 60 80 100
iteration number
Fig. A.3 Development of the averaged relative errors of electron temperature derived with Fe I
emission lines
1.0
coefficient of determination r 2
0.8
0.6
0.4
0.2
Fe II
0.0
0 5 10 15 20 25 30 35 40
iteration number
Fig. A.4 Development of the averaged coefficient of determination for the considered 36 Boltz-
mann plots vs. the iteration number for Fe II emission lines
In Figs. A.7 and A.8, the calculated electron temperatures are shown for the
investigated pulse energies and delay times tdelay . For Fe I and Fe II, two different
electron temperatures exist within the plasma. Starting from tdelay D 100 ns to
tdelay D 900 ns for both ionization stages and most pulse energies the temperatures
are decreasing. Comparing those values with the estimated electron temperatures at
the beginning of the theoretical line preselection, the average electron temperature
Te 8;300 K of the atomic lines for all investigated times and pulse energies is
600 K below the chosen temperature at the beginning of the line selection process.
In the case of the ionic lines, the average temperature Te 12;200 K is about 900 K
-2
-4 Fe I
r 2 = 0.983
ln(el / (A nm g n ) ) [a.u.]
-6
-8
-10
-12
-14
-16
3 4 5 6 7 8
energy of upper level En [eV]
Fig. A.5 Boltzmann plot with 61 Fe I emission lines for an exemplarily chosen time window
tdelay D 900 ns, tint D 100 ns, Eb D 0:5 mJ. The relative temperature errors are Te =Te D 1:7%
for this time window
Fe II
-4 r 2 = 0.967
ln(e l / (A nm g n ) ) [a.u.]
-8
-12
-16
-20
4 6 8 10 12 14
energy of upper level En [eV]
Fig. A.6 Boltzmann plot with 12 Fe II emission lines, other data as in Fig. A.5. The relative
temperature error is Te =Te D 5:9%
higher than the initial one. These average temperatures are in the same regime like
those that were reported by Detalle et al. where electron temperatures between 7,000
and 10,000 K were detected using similar laser irradiances [A.31]. To determine the
electron temperature, six Fe I spectral lines in the wavelength range between 370 and
380 nm were chosen. Only delay times between 100 and 400 ns were investigated,
with an error of 37.5% for tdelay D 400 ns.
It should be noticed that the above-described iterative Boltzmann plot procedure
is a systematic approach to improve the precision of the electron temperature
532 Annex A
10000 Fe I
8000
Eb [mJ]
0.2
6000 0.5
1.0
2.0
4000
0.2 0.4 0.6 0.8
delay time tdelay [μs]
Fig. A.7 Electron temperatures determined by Boltzmann plots with 61 Fe I emission lines at
different delay times for pulse energies between 0.2 and 2 mJ
16000
Fe II
14000
12000
10000 Eb [mJ]
0.2
0.5
8000
1.0
2.0
6000
0.2 0.4 0.6 0.8
delay time tdelay [μs]
Fig. A.8 Electron temperatures determined by Boltzmann plots with 12 Fe II emission lines at
different delay times for pulse energies between 0.2 and 2 mJ
determination. However, the reduction of the relative temperature errors does

not necessarily imply an improvement regarding the accuracy. The assessment
of accuracy would require a comparison between independent techniques for the
determination of the electron temperature.
The results show that even at comparatively low laser pulse energies (Eb 2 mJ)
up to 61 atomic emission lines can be identified within a typical spectral range for
temperature measurements. It should be mentioned that the chosen criteria within
the proposed procedure were applied to achieve pragmatic but quantitatively well-
grounded results for the experiment. In fact, this result improves the precision of
References 533
the electron temperature determination. The three main advantages of the described
method are: the transferability to other problems, e.g., samples with arbitrary
chemical composition; the quantitative validation between theoretical predictions
and experimental data; the systematic approach to find appropriate spectral lines for
LIBS applications.
References
A.1. N. Konjevic, M. Dimitrijevic, W. Wiese, Experimental Stark widths and shifts for spectral
lines of neutral and ionized atoms, J. Phys. Chem. Ref. Data, 13, 619–647 (1984)
A.2. C. Colon, G. Hatem, E. Verdugo, P. Ruiz, J. Campos, Measurement of the Stark broadening
and shift parameters for several ultraviolet lines of singly ionized aluminum, J. Appl. Phys.
73, 4752–4758 (1993)
A.3. N. Konjevic, W. Wiese, Experimental Stark widths and shifts for spectral lines of neutral
and ionized atoms, J. Phys. Chem. Ref. Data, 19, 1307–1385 (1990)
A.4. H. Griem, Spectral Line Broadening by Plasmas, (Academic, London, 1974), Appendix IV,
p. 320
A.5. Z. Szymanski, J. Kurzyna, W. Kalita, The spectroscopy of the plasma plume induced during
welding of stainless steel and titanium, J. Phys. D. Appl. Phys. 30, 3153–3162 (1997)
A.6. J. Aguilera, C: Aragon, Curves of growth of spectral lines emitted by a laser-induced
plasma: influence of the temporal evolution and spatial inhomogeneity of the plasma,
Spectrochim. Acta B 58 221–237 (2003)
A.7. A. Lesage, J. Lebrun, J. Richou, Temperature dependence of Stark parameters for Fe I lines,
Astrophys. J. 360, 737–740 (1990)
A.8. M. Rösner-Kuhn, D. Matson, K. Drewes, U. Thiedemann, G. Kuppermann, M. Flemings,
M. Frohberg, Enthalpies and heat capacities of liquid Fe-Cr-Ni alloys with the focus on pure
liquid chromium, Thermochim. Acta 314, 123–129 (1998)
A.9. A. Bartecka, T. Wujec, J. Halenka, J. Musielok, Experimental Stark-broadening studies of
the N I multiplet at 7904.5 A, Eur. Phys. J. D 29, 265–271 (2004)
A.10. S. Freudenstein, J. Cooper, Stark broadening of Fe I 5383 A, Astron. Astrophys. 71, 283–
288 (1979)
A.11. V. Bakshi, R. Kearney, Measurement of Stark width of some Ar I transitions in a d.c. argon
plasma jet at atmospheric pressure, J. Quant. Spectrosc. Radiat. Transf. 42, 405–413 (1989)
A.12. M. Dimitrijevic, S. Sahal-Brechot, Stark broadening of Ca II spectral lines, J. Quant.
Spectrosc. Radiat. Transf. 49, 157–164 (1993)
A.13. J. Puric, S. Djenize, A. Sreckovic, S. Bukvic, S. Pivalica, J. Labat, Stark widths of singly-
ionized iron spectral lines, Astron. Astrophys. Suppl. Ser. 102, 607–609 (1993)
A.14. M. Dimitrijevic, S. Sahal-Brechot, Stark broadening of Al I spectral lines, Physica Scripta
49, 34–38 (1994)
A.15. K. Dzierzega, K. Musiol, Stark broadening and shift for Ar II lines, J. Quant. Spectrosc.
Radiat. Transf. 52, 747–754 (1994)
A.16. M. Dimitrijevic, S. Sahal-Brechot, Stark broadening of Mg I spectral lines, Physica Scripta
52, 41–51 (1995)
A.17. E. Sarandaev, M. Salakhov, Regularities in the Stark widths and shifts of spectral lines of
singly-ionized aluminium, J. Quant. Spectrosc. Radiat. Transf. 56, 399–407 (1996)
A.18. J. Knauer, M. Kock, Experimental Stark broadening constants for Ar I, Ar and II Kr I
resonance lines in the vacuum ultraviolet region, J. Quant. Spectrosc. Radiat. Transf. 56,
563–572 (1996)
A.19. N. Nessib, Z. Lakhdar, Stark broadening of neutral oxygen lines in the impact and
quasistatic approximations, Physica Scripta 54, 608–613 (1996)
534 Annex A
A.20. L. Popovic, M. Dimitrijevic, Stark broadening of heavy ion lines: As II, Br II, Sb II and
I II, Physica Scripta 53, 325–331 (1996)
A.21. NIST Atomic Spectra Database, Vers. 4, Lines, http://physics.nist.gov/PhysRefData/ASD/
lines form.html
A.22. V. Detalle, R. Héon, M. Sabsabi, L. St-Onge, An evaluation of a commercial echelle
spectrometer with intensified charge-coupled device detector for materials analysis by laser-
induced plasma spectroscopy, Spectrochim. Acta B 56, 1011–1025 (2001)
A.23. M. Sabsabi, V. Detalle, M. A. Harith, W. Tawfik, H. Imam, Comparative study of two
new commercial echelle spectrometers equipped with intensified CCD for analysis of laser-
induced breakdown spectroscopy, Appl. Opt. 42, 6094–6096 (2003)
A.24. Ü. Aydin, P. Roth, C. Gehlen, R. Noll, Spectral line selection for time-resolved investiga-
tions of LIBS plasmas by an iterative Boltzmann plot method, Spectrochim. Acta B 63,
1060–1065 (2008)
A.25. P. Fichet, J.-L. Lacour, D. Menut, P. Mauchien, A. Rivoallan, C. Fabre, J. Dubessy, M.-C.
Boiron, in Micro LIBS Technique, ed. by A.W. Miziolek, V. Palleschi, I. Schechter. Laser-
Induced Breakdown Spectroscopy (LIBS) – Fundamentals and Applications (Cambridge
University Press, Cambridge, 2006), ISBN-13: 9780511243042
A.26. C. Geertsen, J.-L. Lacour, P. Mauchien, L. Pierrard, Evaluation of laser ablation optical
emission spectrometry for microanalysis in aluminium samples, Spectrochim. Acta B 51,
1403–1416 (1996)
A.27. V. Detalle, J.-L. Lacour, P. Mauchien, A. Semerok, Investigation of laser plasma for solid
element composition microanalysis, Appl. Surf. Sci. 138, 299–301 (1999)
A.28. C. Hartmann, A. Gillner, Investigaion on Laser Micro Ablation of Steel Using ps-IR Pulse
Bursts, LIA Conference Proceedings 2007, (LIA, Orlando, 2007), pp. 38–44
A.29. H. Bette, R. Noll, in High-Speed, High-Resolution LIBS Using Diode-Pumped Solid
State Lasers, ed. by A. Miziolek, V. Palleschi, I. Schechter. Laser-Induced Breakdown
Spectroscopy, Chap. 14 (Cambridge University Press, Cambridge, 2006), pp. 490–515
A.30. R. Kurucz, B. Bell, 1995 Atomic Line Data, Kurucz CD-ROM No. 23 (Smithsonian
Astrophysical Observatory, Cambridge, MA)
A.31. V. Detalle, J. Lacour, P. Mauchien, A. Semerok, Investigation of laser plasma for solid
element composition microanalysis, Appl. Surf. Sci. 138–139, 299–301 (1999)
A.32. S. Brym, R. Ciurylo, E. Lisicki, R. Trawinski, Pressure broadening and shift of the 326.1 nm
Cd line perturbed by argon, Physica Scripta 53, 541–544 (1996)
A.33. Z. Szymanski, J. Kurzyna, W. Kalita, The spectroscopy of the plasma plume induced during
welding of stainless steel and titanium, J. Phys. D. Appl. Phys. 30, 3153–3162 (1997)
A.34. S. Pellerin, K. Musiol, J. Chapelle, Measurement of atomic parameters of singly ionized
argon lines – III. STARK broadening parameters, J. Quant. Spectrosc. Radiat. Transf. 57,
377–393 (1997)
A.35. J. Val, J. Aparicio, S. Mar, Experimental Stark widths and shifts of several Ne I spectral
lines, Astrophys. J. 513, 535–541 (1999)
A.36. N. Konjevic, Plasma broadening and shifting of non-hydrogenic spectral lnines: present
status and applications, Phys. Rep. 316, 339–401 (1999)
A.37. N. Konjevic, A. Lesage, J. Fuhr, W. Wiese, Experimental Stark widths and shifts for spectral
lines of neutral and ionized atoms, J. Phys. Chem. Ref. Data 31, 819–927 (2002)
A.38. A. Bartecka, T. Wujec, J. Halenka, J. Musielok, Experimental Stark-broadening studies of
the N I multiplet at 7904.5 A, Eur. Phys. J. D 29, 265–271 (2004)
A.39. J. Hoffman, Z. Szymanski, Time-dependent spectroscopy of plasma plume under laser
welding conditions, J. Phys. D. Appl. Phys. 37, 1792–1799 (2004)
A.40. S. Bukvic, A. Sreckovic, S. Djenize, Mg II h and k lines Stark parameters, New Astron. 9,
629–633 (2004)
Index
ABEL inversion, 154 resonance line, 201

Ablation samples, 449
burst number, 104, 105 scrap, 484
depth, 402 secondary aluminum, 264
efficiency, 105 shredded aluminum scrap, 264, 484
of mass, 97, 102 wrought alloy, 264
rate, 36, 116 Ammonium nitrate, 462
Absorptance, 75 Analysis function, curve, 41, 42, 216
Absorption chromium, 260
coefficient, 26, 79, 124, 171, 180 copper, 256
properties of air, 33 magnesium, 461
surface, 75 nickel, 257, 461
volume, 76 silicon, 258, 270, 316
Abundance ratio, 355 Analyte
Acousto-optic modulator, 120 line, 11
Acrylonitrile butadiene styrene, 286 signal, 291
Additive correction, 260 Analytical
Additive(s), 285 resolution, 443
Adiabatic coefficient, 140 resolving power, 289, 459, 460
Aerodynamic diameter, 362, 373 Angle
Aerosol, 361 of detection, 67, 237, 253
analysis, size-dependent, 372 of observation, 39
generator, 373 Antimon, 293
particle, 372 Apparative broadening, 177, 178
Air Argon
compressed, 154 co-flow, 377
pressure, 111 flushed sample stand, 316
stream, 372 laser, 120
Airborne particles, 362 Arsenic, 324
Alkali, 355 Artificial neural network, 278
Aluminum, 264, 367, 461, 486 ATLAS system, 429
cast alloy, 264, 449, 484 Atomic abundance, 352
depth profile, 416 Attenuation, 193
map, 477, 478 Autofocusing unit, 300
nitride, 474, 475 Availability, technical, 469
oxide, 475 Averaging
pure aluminum sample, 449 spatial, 13, 36, 315, 472
R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9, 535

536 Index
Back illuminated CCD, 63 nitrogen/carbon ratio, 354

Background equivalent concentration, 215, oxygen/hydrogen ratio, 353
227, 250 phosphorus, 234, 250
BALMER line, 149, 351 silicon, 91, 450
Beam sodium, 377
guiding optics, 56 sulfur, 234
GAUSSIAN , 27 thallium, 326
parameters, 19 zinc, 322
propagation ratio, 28 Capillary electrophoresis, 14
quality, 27 Carbide, 476
waist, 28, 113 Carbon, 279, 352, 357, 462
Black body emission, 171 -dioxide, 351
Blast furnace, 356 line, 281
slipping, 359 mapping, 483
Boltzmann Cassegrain optics, 56
distribution, 135 C2-band emission, 283
factor, 135 Cement
formula, 172 hydrated, 327
plot, 133, 144, 262, 531 samples, 328
statistics, 160 Certified reference material, 238
Borate bead analysis, 316 Channeltron signal, 331
Bound-free radiation, 195 Charged-coupled device, 55, 62
Breakdown, 345 Charge detection limit, 66
Bremsstrahlung, 149, 167 Charge neutrality, 193
Brillouin scattering, 61 Chlorides of metals, 363
Bromine, 286, 293, 297 Chlorine, 280
line, 288 line, 327, 331
Building materials, 327 ultraviolet line, 337
Bulk Chromatic aberration, 58
analysis, 18, 41, 275 Chromium, 323
characterization, 275 Circulating materials, 359
Burden level, 359 Classification, 278, 284, 486
Burst performance, 302
energy, 25, 403 CN emission bands, 355
Coating thickness, 406
Coefficient of determination, 219, 243, 250,
310, 451, 462
Cadmium, 224, 227, 297, 323 Co-flow, 374
Calcium, 347, 373, 376 Coherent anti-Stokes Raman spectroscopy, 1
line intensity, 378 Collision, inelastic, 492
particles, 378 Combustion engine, 350
Calibration Congruence, 239
-free LIBS, 207 Continuum
multivariate, 310 emission of the plasma, 10, 324, 394
univariate, 310 intensity, 146
Calibration curve, 12, 212, 350 peak, 31, 341
bromine, 298 radiation, 149
cadmium, 297, 298 Conveyor belt, 266, 270, 484
calcium, 347, 368, 373 Copper, 323, 370
carbon, 241, 250 copper in aluminum, 453
copper, 322, 450 map, 483
hydrogen/carbon ratio, 353 Correlation coefficient, 90
lead, 369 Covariance, 291
nickel, 251 normalized, 292
Index 537
Crater, 8, 34 Droplet, 339, 373

contour plot, 405 generator, 339
cross section, 99 picoliter, 340
depth, 239, 392, 405 Duplex, 438
diameter, 107, 239, 366, 392 Dust separator, 357
dimensions, 98 Dye laser, 223
formation, 97 Dynamic range, 65
geometry, 431
volume, 98
Echelle spectrometer, 53, 287, 290, 351, 366,
Critical value, 214
375, 403
Crosstalk, 431
Einstein
Curves-of-growths, 189, 376
coefficient of spontaneous emission, 133,
Cyanide emission, 462
170, 195
Czerny-Turner spectrometer, 51
coefficient of stimulated emission, 195
transition probability for absorption, 172
Ejection, 484
Damage threshold, 59
Electron
De Broglie wavelength, 162, 193
density, 133, 140, 146, 147, 157, 261, 332
Debris, 393
density, spatially resolved, 152, 158
Debye
free, 154
number, 163
ion collision frequency, 80
sphere, 162
probe microscopy, 483
Decarburization zone, 474, 483
temperature, 132, 140, 146, 147, 262
Decision limit, 488
Element
Decomposition temperature, 78
in alphabetic order allocated to
Delay time, 9, 66
wavelengths, 523
Dendrite, 483
list of wavelengths, 511
Depth
maps, mapping, 393, 396, 476
information, 404, 420
-specific wavelength, 10
profile, 402, 416, 453
Emission
resolving power, resolution, 408, 453
coefficient, 193
Detector types, 62
coefficient of free-free transistions, 167
Diamond polishing, 475
coefficient of recombination radiation, 167
Dichroic mirror, 56, 266, 402
line, 31
Differential mobility analyzer, 373
spectrum of laser-induced plasma, 9
Diffusion dryer, 373
Energy buffer, 148
DIN 32 645, 333
Enthalpy of the plasma, 203
Diode-pumped solid state laser, 19
Equilibration time, 164
Direct light channel, 253, 327
Equivalence ratio, 350
Direction of
Error, type 1, type 2, 486
incidence of the laser beam, 38
Evaporation, 7, 75
observation of the plasma emission, 38
temperature, 76
Displacement
Expansion
energy, 491
adiabatic, 140
of ambient atmosphere, 491
of planar shock front, 126
Dissociation, 7
spherical, 126, 491
Doppler broadening, 173
velocity, 125
Double pulse, 23, 24, 84
Explosion, 125
collinear, 84, 103
Explosive residues, 462
mode, 457
Extinction coefficient, 193
symmetric, 131
unsymmetric, 131
Down cycling, 488 Feedforward network, 284
Drilling depth, 102 Ferrite, 483
538 Index
Fiber conduction, 114

optics, 57, 301, 467 penetration depth, 26, 34
UV-enhanced, 350 stabilizer, 286
Flame retardants, 285 Heavy metal, 286
brominated, 285, 298 concentration in polymer matrix, 297
Flow rate, 103, 253 in soil, 319
Fluence, 366 Helium pressure, 330
FML instrument, 448 Heptane, 350, 355
Focal position, 141 Heuristic modeling, 187, 190, 197
Focusing High-pressure liquid chromatography, 14
dynamic, 300 High-speed
static, 300 camera, 344
Foil thickness, 116 framing images, 119
Fraction pureness, 284 identification of scrap pieces, 484
Framing images, 122, 153 LIBS, 388
Free-bound radiation, 196 photography, 120
Free-free radiation, 196 scanning LIBS, 473
Free parameters, 197 streak images, 119, 492
Frequency distribution of emission signals, 481 Hot-dip
Fringe displacement, 127, 154, 155 coated sheet metal, 454
Front illuminated CCD, 63 galvanized samples, 419
F-value, 52 Hydrogen, 279, 357, 462
BALMER series, 280
Gallilean telescope, 56, 455 line, 283
Galvanized coatings, 401 signal, 359
Galvanizing plant, 424 Hyphenated
Gas HPLC-LIBS, 339
accumulated, 492 LIBS-LIF technique, 323
ambient, 102, 367 technique, 221, 337
ambient atmosphere, 29, 491
carrier, 372
density, 107, 191 Identification
exchange, 21, 67, 394 correctness, 485, 488
exchange time, 67 of material mix-ups, 438
inert, 102 method, 275
influence of gas pressure, 330 of scrap pieces, 41
local gas exchange time, 21 yield, 485
mixtures, 105, 349 Impactor
nozzle, 408 cascade, 362
pressure, 374 low-pressure, 362
reduced ambient pressure, 110 Inclusion
Gauss analysis, 398, 475
distribution, 480 concentration, 480
function, 173 non-metallic, 481
width, 177 size, 481
Glan in steel, 397
laser prism, 433 types, 476
THOMPSON polarizer, 119 In-cylinder diagnostics, 350
Grating, 52 Index of refraction, 152
Inductively coupled plasma atomic emission
spectroscopy, 83
Hand-held measuring pistol, 449 Industrial applications, 429, 467
Heat Inline analysis, 2
affected zone, 34 analysis of galvanized sheet steel, 415
Index 539
analysis of magnesium zinc coatings, 426 spectroscopic methods, 1

LIBS analyzer, 299 triangulation sensor, 300
monitoring of the depth profile of Al in types, 47
hot-dip galvanized coatings, 416, Lead, 289, 324, 355, 357, 370
426 LIBS instruments, 429
process control, 437 Life time of the plasma, 8, 160
In-situ analysis of heavy metals in soils, 320 Light elements, 246
Inspection Light metal
equipment monitoring, 469 alloy, 265
machine, 440 cycles, 488
Integration Limit of detection, 214, 235, 241, 250, 267,
time, 65 288, 294, 299, 323, 334, 347, 369,
window, 9, 10 436, 451, 462
Interaction volume, 21 Limit of identification, 215
Interelement correction, 218, 260 Limit of quantification, 215
Interference fringes, 154 Line
Interferometric measurement, 121 broadening, 173
Internal standardization, 211, 259, 291 broadening, natural, 169
Interpulse separation, 24, 108, 143 broadening, pressure, 169
Ionization emission coefficient, 168, 208
degree, 146, 149 pairs, homologous, 211
stage, 514 radiation, 167
Iron, 357, 361 -ratio, 208
emission line, 133, 527 selection, 408, 511
liquid pig iron, 462 shape, 348
matrix, 461 shape function, profile, 169, 195
neutral iron atoms, 159 Line-to-background emission, 90
pure iron sample, 249, 445, 476 Line-to-continuum intensity, 145
singly ionized iron ions, 159 Linear
spectrum, 198 calibration function, 250
Irradiance, 2, 8 dispersion, 51
of incident laser beam, 185 Loading, 307
threshold, 43 Local
ISO 9001, 469 dilution, 107
temperature equilibrium, 144, 162, 193
transient reduction of particle density, 159
Lollipop samples, 462, 482
Kirchhoff’s law, 172, 195
Long term stability, 312
Lorentz
curve, 133
Laboratory systems, 429 function, 169, 173, 201
Lance, 248 Luminous front, 124
Landau length, 161
Laser
absorption spectroscopy, 3 Mach number, 126
-induced breakdown spectroscopy, 2, 8, 12 Mach-Zehnder interferometer, 120
-induced damage, 58 Magnesium, 349
-induced fluorescence, 3, 14, 320 Magnesium fluoride
-induced fluorescence signal, 324 lens, 328
line section sensor, 484 window, 329
measuring methods, 1 Mandel test, 212
power, 469 Manganese, 437, 462
radiant exposure, 107 map, 397, 476
sorting system, 484 Mass throughput, 488
540 Index
Material Multiplicative correction, 260

ablation, 21 Multivariate analysis, 318, 408, 471
identification, 442
organic, 462
recycling, 275, 488 Natural line width, 175
vapor, 491 Nd:YAG laser, 47
Matrix continously-pumped, 19
effect, 262, 311, 368 diode-pumped, 48, 51
element, 11 flashlamp-pumped, 17
Maxwellian velocity distribution, 160 Near infra-red, 276
Mean, 481 Nickel, 324
Measuring high nickel alloys, 438
bursts, 306 Nitride, 476
frequency, 17, 66 Nitrogen, 352, 357
method, 207 flow, 361
parameters, 14 map, 478
position, 434 Non-conducting materials, 275
probe, 356, 358 Non-metal elements, 340
procedure, 14 Nozzle, 394
pulses, 14, 39, 212, 253, 437 Nucleating agents, 286
volume, 467 Numerical aperture, 52
Measuring chamber, 12, 66, 222, 223, 374, 389 NEWTON
pressure inside the measuring chamber, method, 193
327 telescope, 58, 455
volume, 67
Median, 481 Optical penetration depth, 26, 76
Melt Optical probe, 247
analysis, 249 Optical thickness, 172, 410
composition, 245 optically thick, 196
phase, 97, 245 optically thin, 197
Mercury, 289, 324 Ore, 359
lamp, 456 Oscillator strength, 170, 181
Metal-salt solutions, 320 Outlier points, 481
Microanalysis, 387, 472 Oxide, 476
scanning, 41, 390 Oxygen, 352, 357
Microchannel plate, 55, 62 map, 477, 478
Microlens array, 38, 315, 472
Mix-up
detection of material mix-ups, 467 Partial least square algorithm, 307, 409
Mobile systems, 447 Particle
Molar diameter, 370
concentration ratio, 370, 371 number, 363
mixing ratio, 364 size distribution, 373
Molybdenum, 444 Particulates, 361
Moving Partition function, 169, 193
particles, 265 Partition of laser burst energy, 131
samples, 270 Paschen-Runge spectrometer, 53, 265, 301,
Multi-CCD Paschen-Runge system, 455 342, 356, 372, 379, 388, 389, 417,
Multi-channel integrator electronics, 65, 231, 433, 437, 448, 455, 469, 472
237, 288, 329, 356, 390, 417, 433, Pearlite, 483
435 Pellets for XRF analysis, 315
Multi-element analysis, 2 Penetration depth per laser pulse, 453
Multiple laser pulses, 84 Phosphor
Multiplets, 199 map, 483
Index 541
Photocathode sensitivity, 63 nanosecond, 22

Photodiode picosecond, 35
array, 55 width, 104
signal, 343 Pulse-to-pulse fluctuations, 20
Photomultiplier Pulse width, 22
signal, 30, 88, 309, 394
tube, 53, 62, 231
Physical model, 185 Q-switching
Piezoelectric nozzle, 339 electro-optical, 17
Pigments, 285 Quantitative analysis, 207
Pipe fittings, 438, 467 Quantum efficiency, 63
Planck’s law, 171 Quartz fiber bundle, 276
Plasma Quasi-static broadening by ions, 173
air-breakdown, 87
characteristic time of decay, 190
confinement, 131, 378 Radiance of emission line, 179
core, 190 Radiation processes, 167
dynamics, 119, 123, 131 Radiation transport equation, 193
expansion, 124, 345 Rarefied ambient atmosphere, 131
geometry, 121, 431 Ratio of ablation rates, 116
ignition, 366 Rayleigh
light-emitting, 87 criterion, 52
luminous front, 345 length, 28, 113
parameter, 260 Recalibration, 216, 443, 469
residuals, 394 Recombination
shell, 190 energy, 140
volume, 139 radiation, 167
Plastic bottles, 277 Recycling of aluminum scrap, 484
Pneumatic dispatch, 430 Redeposition, 107
Pockels cell, 23, 111 Reference line, 211
Polarizability, 154 Reference samples, 11, 254
Polarization coupling, 350 certified, 11
Polyamide, 27 certified binary reference ZnAl-bulk
Polycarbonate, 27 sample, 420
Polymer samples, 278 Reflectivity, 77
Polyvinyl chloride, 275 Refractive index, 154
Portable systems, 447 distribution, 154
Positive-intrinsic-negative-photodiode, 276 Refractor plate, 348
Potassium, 357 Regions of interest, 280
Pre-bursts, 306 Regression
Precision, 216 analysis, 212
Preconditioning effect, 337 least square, 471
Prediction error, 471 multilinear regression, 270
Pre-pulses, 14, 40, 212, 442, 451 Relaxation time, 164, 165
Principal component analysis, 479 Remote LIBS system, 454
Profiling of the spectrometer, 444 Repeatability, 313
Propane, 350 Repetition
gas mixtures, 353 rate, 17
Publications on laser-induced breakdown time of laser pulses, 161
spectroscopy, 4 Residual, 216
Pulse, 22 deviation, 256, 257
built-up time, 391 Resolution
energy ratio, 104, 402, 457 depth, 33, 36
femtosecond, 22 lateral spatial, 34, 387
542 Index
spatial, 33 Solid state laser

spectral, 88 diode-pumped, 47, 388
thickness, 402 flashlamp-pumped, 47
Resolving power of a spectrometer, 51 Sort identification, 451
Resonance line, 188 Sorting plant, 302
Response time, 430 Sound velocity, 21
Root mean square error, 409 Source apportionment, 361
of prediction, 471 Space exploration, 454
Rowland circle, 53 Spark optical emission spectrometry, 33, 229
Spatially resolved LIBS analysis, 43
Speciation analysis, 337
Scanner, 432, 468, 484 Spectral background, 295
Scanning, 265 Stability of the spectrometer, 473
LIBS, 388 Standard deviation
mobility particle sizer, 373 of procedure, 266
Score plot, 480 relative standard deviation of procedure,
Scrap particles, pieces, 264, 485 213, 268
Second harmonic generation, 222 residual standard deviation, 213
Segregation Stark
analysis, 398, 474, 482 broadening, 133, 149, 159, 173, 198, 260,
ratio, 482 281
Self-absorption, 92, 180, 188, 377, 410 broadening parameter, 173
Self reversal, 201 coefficient, 134
Sensitivity, 213 data, 511
Separation techniques, 349 quadratic STARK effect, 173, 194
Shell model of the plasma, 189 red shift, 201
Shock wave, 125, 151 shift parameter, 174, 199
concentric, 156 width, 149
propagation in inhomogeneous atmosphere, State of aggregation, 42
130 Steel, 229
Signal-to-noise ratio, 288, 289, 295, 341 electrolytic galvanized sheet, 404
Silicon, 462, 486 high alloyed steel foils, 112
line, 308 high-alloy steel, 252, 254, 467
Simulated plasma emission spectra, 32 liquid, 245, 249, 251
Simulated spectral radiant flux, 197 low-alloy steel, 232
Single pulses, 25, 83 moving sheet, 401, 403
Sinter, 358 rotating sheet steel disks, 418
Size-dependent stainless, 438
composition, 361 thin sheets, 475
particle mass concentration, 379 with scale layer, 235, 433
Skewness, 480 Stoichiometric ratio, 279
coefficient, 482 Streak
Slag, 303 camera, 120
analysis, 303, 469 photographs, 123
components, 315 Sulfide, 476
converter, 304 Sulfur, 314
lime saturated, 315 map, 397, 482
liquid slag, 462 Superduplex, 438
vacuum, 317, 471 Swan bands, 280
SML system, 473
Sodium, 344, 357, 360, 373, 376
Soil, 224 Table-type circular conveyor, 440
contaminated soil sample, 226 Tailored pulse train, 94
Solid sample, 14 Target function, 309
Index 543
TeleLis system, 455 UNS ÖLD ’s formula, 146, 169

Temperature correction, 409, 412
Temporal width of laser pulse, 160
Thallium, 225, 324 Vacuum
Thermal conductivity, 108 degasser, 315, 472
Thermal diffusivity, 26, 34, 114 spectrometer, 389
Thickness ultraviolet, 32, 67, 287
of coatings, 401 Vapor density, 78
resolution, 415 Velocity of measuring object, 41
of scale layers, 429 Vibrating chute, 484
Time scales, 160 Voigt profile, 174, 409
electron-electron relaxation time, 161
electron-ion relaxation time, 161
electron-neutral relaxation time, 161 Warming-up pulses, 39, 212
gas exchange time, 21, 67 Waste electric and electronic equipment, 285
population of levels, 161 Water droplets, 341
Titanium, 438, 486 Weld seam, cross section, 398
nitride, 475 Wet-chemical methods, 475
Top gas, 356 White light interferometry, 405
tube, 358 Wires of steel belt tyres, 475
Transmittance
of CaF2 , MgF2 windows, 58
of silica fiber optics, 58 X-ray fluorescence, 229
Triangulation sensor, 418, 440, 446 analysis, 304
Triple pulses, 24, 236, 416
collinear, 92
Trueness, 313 Zeroth order, 469
Tube, 467 of the grating, 306
Tunable laser source, 223 Zinc, 324, 355, 357, 360, 371
Two-step technique, 83 coating, 403

Reinhard Noll (Auth.) - Laser-Induced Breakdown Spectroscopy - Fundamentals and Applications-Springer-Verlag Berlin Heidelberg (2012)

Uploaded by

Copyright:

Available Formats

Reinhard Noll (Auth.) - Laser-Induced Breakdown Spectroscopy - Fundamentals and Applications-Springer-Verlag Berlin Heidelberg (2012)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Reinhard Noll (Auth.) - Laser-Induced Breakdown Spectroscopy - Fundamentals and Applications-Springer-Verlag Berlin Heidelberg (2012)

Uploaded by

Copyright:

Available Formats

Laser-Induced Breakdown Spectroscopy

ISBN 978-3-642-20667-2 e-ISBN 978-3-642-20668-9

Library of Congress Control Number: 2011940321

c Springer-Verlag Berlin Heidelberg 2012

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

This book is meant as comprehensive source of the fundamentals and applications of

Aachen, Reinhard Noll

5 Evaporation and Plasma Generation .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 75

11.3 Measurement and Determination of Analytes . .. . . . . . . . . . . . . . . . . . . . 216

17.3 Mobile Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 447

Laser radiation is a high-quality form of electromagnetic energy enabling a mul-

R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9 1, 1

monitoring of combustion processes or material-dependent production processes,

on the surface of macroscopic samples. An example of this approach is the analysis

In this chapter, the principle of laser-induced breakdown spectroscopy is presented.

2.1 Principle and Measuring Parameters

The principle of laser-induced breakdown spectroscopy is shown schematically in

R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9 2, 7

Fig. 2.3 Schematic

background, increases significantly. At t3 , the plasma temperature decreases further

Fig. 2.4 Schematic

2.2 Setup for Laser-Induced Breakdown Spectroscopy

Fig. 2.7 Arrangement of components in a LIBS-based analyzing system; T D sample table, S D

Fig. 2.8 Analyzing system

2.3 Measuring Procedure

A measuring procedure comprises the following steps:

2.4 Applications and Methodical Extensions

Laser-induced breakdown spectroscopy is a versatile tool to analyze solid, liquid, or

3.1 Laser Pulse Energy and Repetition Rate

R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9 3, 17

pulse width [ns] – – – –

phenomenologically as formation of a crater. Furthermore, the structural, physical,

3.2 Temporal Characteristics of the Laser Pulse

Fig. 3.7 Temporal shape of a

50 1st pulse 2nd pulse

Fig. 3.8 Double pulses of a flashlamp-pumped Nd:YAG laser

3.3 Laser Wavelength

The absorption properties of a substance depend on the wavelength of the laser

where  is the thermal diffusivity, Œ D cm2 =s.

Fig. 3.11 Optical penetration

3.4 Beam Quality and Focusing

where 0 is the beam divergence angle of a Gaussian beam, is a propagation invariant

3.5 Ambient Gas Atmosphere

-0.04 Fe 273.07 nm, argon

3.6 Spatial Resolution and Spatial Averaging

10000 9 double pulse

Fig. 3.21 Photograph of a

Fig. 3.22 Optical microlens

With a microlens array a simultaneous generation of up to Np D 40 plasmas

3.7 Direction of Incidence of the Laser Beam and Direction

3.8 Warming-up Pulses, Prepulses, and Measuring Pulses

Table 3.2 Examples for moving measuring objects in LIBS setups

3.9 State of the Measuring Object

0.06 moving samples

Fig. 3.26 Laser beam

3.29. B. Le Drogoff, J. Margot, M. Chaker, M. Sabsabi, O. Barthelemy, T.W. Johnston, S. Laville,

R. Noll, Laser-Induced Breakdown Spectroscopy, DOI 10.1007/978-3-642-20668-9 4, 47

Table 4.1 Overview of laser types used for LIBS

where is the thermal diffusivity, Œ D cm2 =s.

NAES D nref sin ; (4.4)

D ei C en ; (5.12)

bi .p/ D mi .p/Œdf df 0 .p/; (7.2)