IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS Related content

- Concepts in Physical Metallurgy: The
Fundamentals of electron energy-loss theory of alloys
A Lavakumar
spectroscopy - Energy-loss contribution for fast He atoms
axially scattered off a silver surface
C A Ríos Rubiano, G A Bocan, J I Juaristi
To cite this article: F Hofer et al 2016 IOP Conf. Ser.: Mater. Sci. Eng. 109 012007 et al.

- DESIGN FUNDAMENTALS OF
ALGORITHMS FOR PHOTON-
COUNTING SYSTEMS
James Dick, Charles Jenkins and Jacek
View the article online for updates and enhancements. Ziabicki

Recent citations
- Optimization of monochromated TEM for
ultimate resolution imaging and ultrahigh
resolution electron energy loss
spectroscopy
Sergei Lopatin et al

This content was downloaded from IP address 111.68.101.182 on 27/10/2017 at 09:32

EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

Fundamentals of electron energy-loss spectroscopy

F Hofer, F P Schmidt, W Grogger and G Kothleitner

Graz University of Technology and Graz Centre for Electron Microscopy, Institute of
Electron Microscopy and Nanoanalysis, Steyrergasse 17, 8010 Graz, Austria

E-mail: ferdinand.hofer@tugraz.at

Abstract. Electron energy-loss spectroscopy (EELS) is an analytical technique that is based

on inelastic scattering of fast electrons in a thin specimen. In a transmission electron
microscope (TEM) it can provide structural and chemical information about a specimen, even
down to atomic resolution. This review provides an overview of the physical basis and new
developments and applications of EELS in scanning transmission electron microscopy. Recent
advances in elemental mapping, spectrum imaging of plasmonic structures and quantitative
analysis of atomically resolved elemental maps are highlighted.

1. Introduction
The need to understand atomic processes in solids has led to an increasing demand for new imaging,
diffraction and spectroscopy methods with high spatial resolution. This need has been reinforced by
the growing interest in nanoscience and nanotechnology. Particularly, transmission electron
microscopy (TEM) is of unrivalled value, because it can provide structural information with excellent
spatial resolution (down to atomic dimensions) via high resolution TEM imaging and electron
diffraction. These structural data can be supplemented by chemical information from the same
specimen region, obtained using analytical techniques, such as energy-dispersive X-ray spectroscopy
(EDS) and electron energy-loss spectroscopy (EELS) [1].
Due to the broad range of inelastic interactions of the high energy electrons with the specimen
atoms, ranging from phonon interactions to ionisation processes, electron energy-loss spectroscopy
offers unique possibilities for advanced materials analysis. It can be used to map the elemental
composition of a specimen, but also for studying the physical and chemical properties of a wide range
of materials and biological matter.
Beginning with the increasing use of EEL spectrometers during the eighties of the last century [1],
the energy-filtering technique was a major step forward for two-dimensional mapping of the spectral
features visible in an EEL spectrum [2, 3]. The wide distribution of energy-filtering TEMs (EFTEM)
led to numerous practical applications both in the materials and biological sciences [4]. In the
meanwhile, EFTEM is a widely used method for both overview and nanoscale characterisation of thin
samples, applicable to most chemical elements and especially sensitive to light elements.
With the introduction of high brightness electron sources and spherical aberration correctors, the
image resolution of the TEM reaches now the 50 picometre level [5]. Especially scanning
transmission electron microscopy (STEM) equipped with a Cs-probe corrector is extremely useful for
high resolution imaging [6] and the parallel chemical analysis by EELS and EDS because of the much
increased probe current in the corrected beam.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

Before discussing how these various methods and their recent developments provide us with
information about the specimen, we will briefly examine the principles and instrumentation involved
in recording energy-loss spectra in the TEM.

2. The electron energy-loss spectrum

A beam electron in a (S)TEM may be inelastically scattered when it interacts with the atomic electrons
in the specimen. The electron beam loses energy and is bent through a small angle
(5 - 100 milliradians). The energy distribution of all the inelastically scattered electrons provides
information about the local environment of the atomic electrons which in turn relates to the physical
and chemical properties of the specimen. This is the basis of electron energy-loss spectroscopy
(EELS) [1]. Much of the information obtainable from EELS is similar to that of X-ray absorption
spectrometry (XAS) in the synchrotron.
Figure 1 shows a typical electron energy-loss spectrum recorded up to 600 electron volts. The first
peak, the most intense for a very thin specimen, occurs at 0 eV and is therefore called the zero-loss
peak. It represents electrons which have not been scattered in the specimen (transmitted electrons) and
which have been elastically scattered via interaction with the atomic nuclei.

Figure 1. EELS spectrum of a 20 nm thin titanium carbide specimen recorded

in a conventional 200 kV TEM equipped with an energy-filtering spectrometer.

The low-loss or valence region of an EEL spectrum (< 50 eV) provides similar information to that
provided by optical spectroscopy, containing valuable information about the band structure and in
particular about the dielectric properties of a material (e.g., band gap, surface plasmons). The most
prominent peak, centred at 24 eV, comes from a plasma resonance of the valence atoms. Signal
intensities in the low-loss region are larger than in the high-loss region of the spectrum.
At higher energy losses (> 50 eV), where the number of inelastically scattered electrons is much
lower, the spectrum shows characteristic features called “ionisation edges” (due to their typical shape,
a rapid rise followed by a more gradual fall). These edges are the exact equivalent of an absorption
edge in XAS and arise from the same process. The edges are formed when an inner-shell electron
absorbs enough energy from a beam electron to be excited to a state above the Fermi level. Not all
ionisation edges are saw-toothed like the carbon K-edge in figure 1, but exhibit more complex edge

2
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

shapes such as the L2,3-edge of titanium in figure 1. This L2,3-edge includes sharp excitations at the
onset of the ionisation edge, called white-lines, which are typical for elements in the first row of
transition elements and for the rare earth elements. The ionisation edges can be used for the analysis
of almost all chemical elements in particular for the lighter elements, the edge onset gives the
ionisation energy and allows the qualitative analysis and in case of very thin samples the edge
intensities are proportional to the concentration of the corresponding elements.

3. Recent advances in EELS methodology

TEM-EELS instrumentation is based on a magnetic prism, in which a uniform magnetic field is
generated by an electromagnet with specially designed pole pieces. The prism bends the inelastically
scattered electrons by about 90o, disperses them according to their different kinetic energies and also
has a focussing action. The spectra are recorded with a charge coupled device (CCD camera). Due to
the enormous dynamic range of an EEL spectrum, spanning many orders of magnitude, and due to
restrictions in dynamic range and sensitivity of CCD based spectrometers, a complete EEL spectrum
normally has to be recorded in several segments by changing illumination conditions and/or
acquisition times.
Energy filters are used to form images from electrons that have suffered a specific energy loss
[energy filtered TEM (EFTEM) or electron specific imaging (ESI)]. In particular, two types of
energy-filtering instruments are most often used: firstly post-column energy filters, with a single prism
geometry and multipole lenses, can be attached and retrofitted to practically any TEM or STEM
instrument [7]. The post-column filter is not only an efficient imaging system, but also a versatile
instrument for acquiring EEL spectra in high quality. Secondly the in-column filter, whose
spectrometer consists of four magnetic prisms that are arranged symmetrically according to the shape
of a Greek Omega, is located below the projector lens system of the TEM. This type of filter can be
also used for spectroscopy, but the main application field is high quality imaging such as large area
elemental mapping and energy-filtered electron diffraction studies [8].
Another important development was the introduction of monochromators for the electron source,
which paved the way for acquiring EEL spectra at high energy resolution, typically in the
100 - 200 meV range (e.g., the Wien filter approach) [9]. The improved energy resolution opens new
possibilities for studying detailed electronic structure and bonding effects evaluated from near edge
fine structures of the ionisation edges, but also accurate band gap and dielectric function
measurements via the low-loss part of the spectrum [1].
Several years ago, the problem of the high dynamic range of the EEL spectrum could be addressed
by acquiring the elastic and inelastic parts at nearly coincident times at identical experimental
conditions, offering the advantage of using the low-loss regime as a reference for quantitative data
analysis. This system works with an additional electrostatic vertical deflector and is now successfully
used by several groups worldwide [10, 11].
Advances in X-ray detection with the advent of large area or four-quadrant solid state silicon drift
detectors [12] and the rapid progress in EDS data processing, pushed the ideas for highly efficient
elemental mapping in combination with EELS. The collaboration between Gatan, Bruker and the
TU Graz enabled the fast recording of multimodal data so that STEM-images, low-loss and high-loss
EELS and EDS data can now be acquired with a speed of 1,000 to 1,500 spectra per second [13].
While the conventional method of EELS mapping combines TEM/STEM images with the local
concentration of the elements, the development of more powerful computers and data reduction
procedures nowadays allows a “holistic” approach: the whole spectral information is gathered for each
point of an image, generating a three-dimensional data set which is often called a spectrum image [14].
The spectrum-imaging technique offers various advantages: the wealth of data allows a certain amount
of “post experiment microscopy” and because of the completeness of data, interpretation mistakes can
be avoided. Additionally, a data-evaluation software can automatically identify and highlight the most
prominent features, e.g., chemical phases [15]. Therefore, spectrum-imaging techniques are the
essential basis for successful EELS mapping at the atomic scale.

3
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

4. Elemental mapping at high spatial resolution

One of the most commonly used applications of STEM-EELS or EFTEM is to derive compositional
information by recording energy-filtered images using the element characteristic ionisation edges.
Although advanced energy-filters or spectrometers now enable the fast acquisition of elemental maps
almost on a routine basis, care must be taken due to several experimental limitations of EELS:
generally, the signal-to-noise ratio of the elemental signal is very low which is mainly caused by the
high uncharacteristic background below the ionisation edges and the low ionisation cross-sections for
heavier elements and for elements occurring at low concentrations.
Another important limitation comes from multiple scattering of the inelastic signal in thicker
specimens, restricting elemental mapping to specimen thicknesses well below the mean free path
lengths of the inelastically scattered electrons (specimen thickness < 70 nm for most materials).
Alternatively, EELS spectra from thicker specimens may be deconvolved by using the low-loss part of
the EELS spectrum, recovering a single scattering distribution EELS spectrum. Typical procedures
involve Richardson-Lucy or Fourier log methods [16, 17], which however decrease the signal-to-noise
ratio of the spectrum. When crystalline specimens are studied, a frequent problem is the preservation
of diffraction contrast in inelastic imaging, which can be reduced by employing advanced illumination
conditions (hollow-cone or rocking-beam illumination) [2].
Firstly, detailed strategies had to be developed to optimize the quality of the experimental
spectrum. Secondly, advanced spectrum imaging techniques, which include a variety of data
correction or processing procedures, are a prerequisite for successful elemental mapping with EELS or
EFTEM [15, 18] (see example in figure 2). Additionally, it is known for a long time that EELS
spectra or elemental maps can be quantified using the standard EELS quantification procedures thus
providing absolute and relative concentration values at the nanometre scale [1, 19].

Figure 2. MgO/Ni core shell nanoparticles investigated with a CS-corrected STEM and
EELS spectrum imaging revealing the Ni core and the MgO shell, the inset shows the
Z-contrast image of the Ni core (specimen courtesy: Gopi Krishnan, University of
Groningen, The Netherlands).

4
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

Although EFTEM elemental mapping is now widely used for nanoscale analysis, it cannot deliver
atomic resolution information even in aberration-corrected TEMs. Since the CS-image corrector
affects only spherical aberrations, but not chromatic aberrations (CC), EFTEM mapping at atomic
resolution is only possible by incorporating a chromatic aberration corrector. This so far is only
realized at the NCEM in Berkeley (USA) and the Ernst-Ruska Center in Jülich (Germany) [20].
Elemental mapping at atomic resolution can be easier achieved via the scanning transmission
electron microscope (STEM) which is now in wide general use, harnessing Z-contrast imaging and
EELS [21]. The first elemental maps at atomic resolution have been recorded by using the
STEM-EELS technique by Kimoto et al. [22] and they have been able to visualize the atomic columns
of La, Mn and O in the layered manganite La1.2Sr1.8Mn2O7. In the meanwhile Allen and co-workers
[23] have shown that it is also possible to record atomically resolved elemental maps by using X-ray
spectroscopy in the STEM.
Two important developments enhanced the potential for successful elemental mapping at atomic
resolution: Firstly, the introduction of collection-efficient solid state X-ray detectors (SDD) [12] and
secondly, the simultaneous acquisition of EELS and EDS spectrum images with fast scanning rates
[13]. Since then many examples of atomically resolved elemental maps have been published, but it
has to be realized that coloured elemental maps only yield qualitative information. However, it is
extremely important to provide quantification data in terms of volumetric densities (i.e., number of
atoms per unit volume), as it allows correlating structural and chemical information with each other
(“unit cell atom counting”). Using a SrTiO3 crystal of known thickness and density Kothleitner et al.
[24] could show how EELS and EDS elemental maps (figure 3) can be quantified by combining both
spectroscopies and including the zeta factor method for X-ray quantification. The results show that it
is possible to obtain quantitative numbers of Sr, Ti and O atoms in the individual atom columns,
however elastic and thermal scattering effects can severely influence accuracy at atomic resolution
[25].

Figure 3. Experimental atomic resolution EELS elemental maps for the {001} direction in SrTiO3 for
the Sr L2,3-edge, the Ti L2,3- and the O K-edge with the simultaneously acquired HAADF image and
the RGB composite of the elemental maps; quantum mechanical simulations, assuming a Gaussian
incoherent source size of 0.08 nm, are inset in each figure; from [25].

5
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

5. Chemical bonding information

Edge fine structures arouse considerable interest in the application of EELS, especially in the field of
materials science, because they can be used to extract information regarding local charge distributions,
coordination numbers and bonding characteristics. The fine structures are divided into the near
edge-fine structures (ELNES) within 50 eV of the edge onset and the extended fine structure
(EXELFS) about some 100 eV above the edge onset. Most of the ionisation edges contain a more
complex structure than can be explained in simple atomic terms. The ionisation edges are often
modified by the solid state environment of the atom undergoing the inner-shell excitation and this
information can be used as a “fingerprint” from elements in similar chemical environments [26, 27].
In the meanwhile the calculation of ELNES structures has reached a high degree of sophistication
[28].
One important development for ELNES studies was the introduction of the monochromated
(S)TEMs which helped to improve the instrumental energy resolution to values as low as 100 meV. In
combination with advanced energy-filters and/or spectrometers it was now possible to fully exploit
informational the spectral details contained in the EEL spectrum. It was not at all obvious, however,
at the time of the development during the late nineties that these tools would gain such importance.
The reduced energy spread of the TEM immediately opened possibilities for detailed studies of the
near-edge fine structures at K- and L2,3-ionisation edges of oxides, perovskites and similar materials
[29]. The knowledge about these ELNES structures at high energy resolution is an important basis for
STEM-EELS imaging of chemical bonding in materials. Figure 4 shows the ELNES structure of the
V L2,3 white lines of V2O5 recorded at different energy resolutions. The L3 white line reveals the
symmetry of the vanadium site being surrounded by six oxygen atoms forming a strongly distorted
octahedron unit [30].

Figure 4. Comparison of the V L2,3-

and O-ionisation edges of V2O5 which
have been recorded with different
energy resolution; a) 200 kV TEM
with LaB6 cathode (0.7 eV);
b) 200 kV TEM with Schottky emitter
(0.6 eV); c) 200 kV TEM with a
monochromator and a HR-energy-
filter (0.3 eV); d) X-ray absorption
spectrum recorded with an energy
resolution of 0.08 eV; from [30].

6. Physical property mapping

The low-loss spectrum contains energy losses to valence or conduction electrons which provide rich
information about the physical properties of a specimen e.g., inter- or intraband transitions, band gaps
and the dielectric properties [31, 32]. One important application of low-loss EELS lies in the study of
optical properties of metallic nanostructures which may drastically change at the nanometre scale as a
function of size, morphology and environment. In particular, collective oscillations of quasi free

6
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

electrons on a metallic/dielectric interface, i.e., surface plasmons, are increasingly studied in the
(S)TEM, taking advantage of its unbeaten high spatial resolution (compared to the scanning optical
near-field microscope). Here the introduction of monochromators, providing an energy resolution of
100 meV or even less [33], was mainly responsible for the big success of the technique [34].
Since the early seventies of the last century EELS has been used for studying surface plasmons
[31, 35], but it was only in 2007 when two independent groups showed how fast electrons can be used
to map localized surface plasmons of single noble metal nanoparticles [36, 37]. In these pioneering
works STEM-EELS was used, but later it was also shown that a monochromated EFTEM can yield
comparable results [38, 39]. It is now widely accepted that EELS or EFTEM are the most advanced
methods to probe plasmonic modes and they are increasingly used to study nanostructures of
increasing complexity. For example STEM-EELS was used for studying dark plasmonic breathing
modes in silver nanodisks [40], for morphing a plasmonic nanodisk into a nanotriangle [41], for
revealing universal dispersions of surface plasmons in flat nanostructures [42], for studying the
3D distribution of surface plasmons around a metal nanoparticle [43] and for exotic nanostructures as
shown in figure 5.

Figure 5. a) High-angle annular dark field (HAADF) image of a 30 nm

thick Au-nanostar on a 15 nm thin silicon nitride membrane (left) and
corresponding electron energy loss (EEL) maps at 0.8 eV (A), 1.35 eV (B)
and 1.70 eV (C) integrated over an energy width of 150 meV. The EELS
maps were generated using the STEM-EELS approach with a
monochromated 200 keV electron beam with 150 meV energy resolution
(FWHM). The sample was prepared by electron beam lithography, raw
data are presented. B) The regions 1-3 in the spectrum image (left) mark
the areas from which EEL spectra were extracted. The peaks labelled by
A, B and C corresponds to the energies of the EELS maps shown in (a).

7
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

Recent progress in the development of monochromators with improved energy resolution in the
range of 10 meV has shown that it is now possible to reveal vibrational modes (phonons) in
EEL spectra recorded in a STEM [44]. This new technique allows to study the spatial variation of
vibrational modes in nanostructures and nanoparticles.

7. Conclusions
Overall, inelastic excitations, probed by electron microscopy, have been a continuous source of
information on composition, chemical and physical properties of nanostructured materials over the last
decades. In this review we briefly describe the physical basis of electron energy-loss spectroscopy and
recent developments in EELS instrumentation. We show how STEM-EELS can be used for elemental
mapping of heterogeneous nanoparticles and discuss the prospects for quantitative elemental analysis
at atomic resolution. We describe the advantages of monochromated STEMs for obtaining chemical
bonding information from the EELS near-edge fine structures. Furthermore, we explain how the
low-loss region of the EEL spectrum can be used for physical property mapping. Here we concentrate
on surface plasmon imaging of noble metal nanoparticles. In summary, EELS imaging and
spectroscopy with aberration corrected STEMs offer many new opportunities to study fundamental
questions in physics, chemistry and materials science at atomic resolution.

Acknowledgements
This work has received funding from the European Union within the 7th Framework Programme under
Grant Agreement no. 312483 (ESTEEM2). This work was financially supported by the FFG in
Vienna, Austria (COIN-project ASTEM).

References
[1] Egerton R F 2011 Electron energy-loss spectroscopy in the electron microscope, 3rd edition.
(New York-Dordrecht-Heidelberg-London: Springer)
[2] Shuman H, Chang C F and Somlyo A P 1986 Ultramicroscopy 19 121
[3] Mayer J 1992 in: Proc. 50th EMSA Meeting. (San Francisco, CA: San Francisco Press)
1198-1199
[4] Hofer F and Warbichler P 2004 in: Transmission electron energy loss spectrometry in
materials science and the EELS atlas. 2nd edition. (Ahn C C; Ed.) (Weinheim:
Wiley-VCH) 159-222
[5] Haider M, Uhlemann S., Schwan E, Kabius B and Urban K 1998 Nature 392 768
[6] Batson P E, Dellby N and Krivanek O L 2002 Nature 418 617
[7] Krivanek O L, Gubbens A J, Dellby N and Meyer C E 1991 Microsc. Microanal. Microstruct.
2 315
[8] Mayer J, Deininger C and Reimer L 1995 in: Energy-filtering transmission electron
microscopy. (Reimer L; Ed.) (Berlin-Heidelberg-New York: Springer Verlag) 291-346
[9] Tiemeijer P C 1999 J. Phys.: Conf. Ser. 161 191-194
[10] Scott J, Thomas P J, MacKenzie M, McFadzean S, Wilbrink J, Craven A J and
Nicholson W A P 2008 Ultramicroscopy 108 1586
[11] Angseryd J, Albu M, Andren H O and Kothleitner G 2011 Micron 42 608
[12] Von Harrach H S, Dona P, Freitag B, Soltau H, Niculae A and Rohde M 2010 J. Phys.: Conf.
Ser. 241 012015
[13] Kothleitner G, Fladischer S, Grogger W, Haber T, Meingast A and Hofer F 2012 Microsc.
Microanal. 18 (Suppl. 2) 972
[14] Jeanguillaume C and Colliex C 1989 Ultramicroscopy 28 252
[15] Schaffer B, Kothleitner G and Grogger W 2004 Ultramicroscopy 106 1129
[16] Gloter A, Douiri, Tence M. Colliex C 2003 Ultramicroscopy 96 385
[17] Thomas P J and Midgley P A 2001 Ultramicroscopy 88 187
[18] Kothleitner G and Hofer F 1998 Micron 29 349

8
EMAS 2015 Workshop IOP Publishing
IOP Conf. Series: Materials Science and Engineering 109 (2016) 012007 doi:10.1088/1757-899X/109/1/012007

[19] Hofer F, Grogger W, Kothleitner G and Warbichler P 1997 Ultramicroscopy 67 83

[20] Urban K W, Mayer J, Jinschek J R, Neish M J, Lugg N R and Allen L J 2013 Phys. Rev. Lett.
110 185507
[21] Pennycook S J and Nellist P D 2011 Scanning transmission electron microscopy. (New York-
Dordrecht-Heidelberg-London: Springer)
[22] Kimoto K, Asaka T, Nagai T, Saito M, Matsui Y and Ishizuka 2007 Nature 450 702
[23] Allen L J, D’Alfonso A J, Freitag B. and Klenov D O 2012 MRS Bull. 37 47
[24] Kothleitner G et al. 2014 Microsc. Microanal. 20 678
[25] Kothleitner G, Neish M J, Lugg N R, Findlay S D, Grogger W, Hofer F and Allen L J 2014
Phys. Rev. Lett. 112 085501
[26] Hofer F and Golob P 1987 Ultramicroscopy 21 379
[27] Keast V J, Scott A J, Brydson R, Williams D B and Bruley J 2001 J. Microscopy 203 135
[28] Rez P, Alvarez J R and Pickard C 1999 Ultramicroscopy 78 175
[29] Mitterbauer C, Kothleitner G, Grogger W, Zandbergen H, Freitag B, Tiemeijer P and Hofer F
2003 Ultramicroscopy 96 469
[30] Su D S, Zandbergen H W, Tiemeijer P C, Kothleitner G., Hävecker M, Hebert C,
Knop-Gericke A, Freitag B H, Hofer F and Schlögl R 2003 Micron 34 235
[31] Geiger J 1967 Phys. Stat. Sol. 24 457
[32] Schattschneider P 1986 Fundamentals of inelastic electron scattering. (Vienna, Austria:
Springer)
[33] Schaffer B, Riegler K, Kothleitner G, Grogger W and Hofer F 2009 Micron 40 269
[34] Grogger W, Hofer F, Kothleitner G and Schaffer B 2008 Top. Catal. 50 200
[35] Batson P E 1980 Solid State Comm. 34 477
[36] Nelayah J, Kociak M, Stephan O, de Abajo G, Javier F, Tence, Henrard L, Taverna D,
Pastoriza-Santos, Liz-Marzan L M and Colliex C 2007 Nature Phys. 3 348
[37] Bosman M, Keast V J, Watanabe M, Maarof A I and Cortie M B 2007 Nanotechnology 18
165505
[38] Schaffer B, Hohenester U, Trügler A and Hofer F 2009 Phys. Rev. B 79 041401
[39] Schaffer B, Grogger W, Kothleitner G and Hofer F 2010 Ultramicroscopy 110 1087
[40] Schmidt F P, Ditlbacher H, Hohenester U, Hohenau A, Hofer, F and Krenn J R 2012 Nano
Lett. 14 5780
[41] Schmidt F P, Ditlbacher H, Hofer F, Krenn J R and Hohenester U 2014 Nano Lett. 14 4810
[42] Schmidt F P, Ditlbacher H, Hohenester U, Hohenau A, Hofer F and Krenn J R 2014 Nature
Comm. 5 3604
[43] Nicoletti O, de la Peña F, Leary R K, Holland D J, Ducati C, Midgley P A 2013 Nature 502 80
[44] Krivanek O L, Lovejoy T C, Dellby N, Aoki T, Carpenter R W, Rez P, Soignard E, Zhu J,
Batson P E, Lagos M J, Egerton R F and Crozier P A 2014 Nature 514 209