Thesis IPZ Final

Electronic Bands of Nanoporous Networks
and One-Dimensional Covalent Polymers

Assembled on Metal Surfaces
PhD Thesis
Ignacio Piquero-Zulaica
Supervisors:
Dr. Jorge Lobo-Checa
Prof. J. Enrique Ortega
E USKAL H ERRIKO U NIBERTSITATEA (EHU-UPV)
Electronic Bands of Nanoporous Networks

and One-Dimensional Covalent Polymers
Assembled on Metal Surfaces
Author: Supervisors:
Ignacio Piquero-Zulaica Dr. Jorge Lobo-Checa
Prof. J. Enrique Ortega
A thesis submitted in fulfillment of the requirements for the degree of

Doctor of Philosophy in Physics of Nanostructures and Advanced Materials
in the
Nanophysics Lab
Centro de Física de Materiales (CFM-CSIC)
October 19, 2018

iii
Abstract
Complex molecular layers self-assembled on surfaces with engineered architectures
and tailored properties, are expected to play an important role in the development
of future devices at the nanoscale [1–3]. The reversibility of non-covalent interac-
tions such as hydrogen bonds or metal ligand interactions, allows error correction
processes in the formed structures. Such elimination of defective structures can
give rise to almost defect-free, long-range ordered formations. Metal-organic net-
works grown on metallic surfaces fall into such self-healing structures and show
novel magnetic properties [4, 5], catalytic effects [6, 7], oxidation states [8], exotic
tesellation patterns [9–12] and even bear the prospect of exhibiting topological band
structures [13, 14].
Nanoporous networks featuring long-range order belong to error corrected non-
covalent structures. The recent finding of electron confinement of the two-dimensional
electron gas (2DEG) within the nanopores of self-assembled supramolecular nanoporous
networks, is an experimental demonstration of a quantum box effect. This is an ef-
fect which may play a crucial role in engineering future molecular devices [15–17].
By using scanning tunneling microscopy and spectroscopy (STM/STS), in a similar
fashion to quantum corrals [18, 19], it is possible to probe such localized electronic
states at the single pore or quantum dot (QD) level [15–17].
However, studies on the long-range ordered and robust 3deh-DPDI metal-organic
network on Cu(111), revealed that nanopores are rather imperfect or leaky confining
entities, leading to significant coupling to neighboring nanopores. The periodicity
of the highly-ordered supramolecular network induces the formation of Bloch-wave
states that result into new electronic bands that can be observed by spatially averag-
ing angle-resolved photoemission spectroscopy (ARPES) [15].
The well-established control of the structures of porous networks, together with
its characteristic degree of coupling between ad-molecules and the surface state, is
our starting point for the fabrication and investigation of coupled electronic systems
with tailored band structures. Based on the concepts of Supramolecular Chemistry
on surfaces, by choosing suitable molecular constituents (functional groups and/or
carbon backbone size) and guided by reversible, non-covalent bonding mechanisms,
we are able to generate six different long-range ordered nanoporous networks on
(111)-terminated coinage metal surfaces in ultra-high vacuum (UHV). Such nanopo-
rous structures are analogous to QD arrays on surfaces, bearing distinct sizes, barrier
separations and scattering strengths. As a result, with each particular nanoporous
system grown, we not only engineer the local confinement properties at each QD,
but also modulate the coherent electronic band structure steming from the overall
array. We observe changes in its fundamental energy, band dispersion, effective
mass, zone boundary gaps and Fermi surface contour.
Our experimental findings are supported by the electron boundary elements
method in combination with the electron plane wave expansion (EBEM/EPWE)
iv
modelling, density functional theory (DFT) calculations, and the phase accumula-
tion model (PAM). In this way, we disentangle the repulsive scattering potential
landscape of each nanoporous network and delve into subtle surface-organic over-
layer interactions, such as hybridization and geometry induced effects, which are
altogether responsible for the confinement effects and distinct electronic band mod-
ulations. Our findings envision the engineering of 2D electronic metamaterials, in
analogy to the well-established optical metamaterials [20].
The studied electronic structure from nanoporous networks correspond to the

modified substrate’s surface state, which is independent of the molecular states.
However, low-dimensional organic electronic states, such as the one obtained in
graphene nanoribbons (GNRs) and oligophenylene chains are currently very at-
tractive to the Scientific Community based on their industrial prospects. These
one-dimensional polymeric structures have been extensively studied as simple, ap-
pealing nanostructures leading to distinct electronic features, such as gap opening
and peculiar edge states. Their quantum confinement origin can be readily tuned
through their width, shape, and edge terminations [21, 22]. The rapidly progressing
on-surface chemistry is a highly versatile bottom-up tool for the controlled-synthesis
of such atomically precise, graphene-based nanostructures [23–25]. This achieve-
ment has paved the way towards the precise mapping of their intriguing electronic
structures with ARPES and STS [26–30], making them promising candidates for the
realization of exotic graphene-based nanodevices [31–33].
In this thesis, we engineer the electronic band structure of the well-known poly-
(para-phenylene) (PPP), namely the Nα = 3 armchair GNR [34], by introducing peri-
odically spaced meta-junctions into its conductive path. We synthesize and macro-
scopically align a saturated film of cross-conjugated oligophenylene zigzag chains
on a vicinal Ag(111) surface. We find that these atomically precise chains, hosting pe-
riodically spaced meta-junctions, remain sufficiently decoupled from each other and
from the substrate. ARPES reveals weakly dispersing one-dimensional electronic
bands along the chain direction, which is reproduced by DFT and EPWE. In addi-
tion, STS shows a significantly larger frontier orbital bandgap than PPP chains and
that straight segments are able to confine electrons. These weakly interacting QDs
confirm that periodically spaced meta-junctions constitute strong scattering centers
for the electrons. These findings corroborate the important effects that the conduc-
tive path topology of a molecular wire has on its electronic states, which are respon-
sible for defining its chemical, optical and electronic properties. Such arrays of semi-
conducting QDs hold potential for designing future oligophenylene-based quantum
devices such as electrically driven, telecom-wavelength, room-temperature single-
photon sources [35].
v
Acknowledgements
Many people have taken part in the realization and fulfillment of this thesis.
First of all, I would like to wholeheartedly thank Jorge and Kike for adopting me
as their student and for guiding me during this thrilling journey. Thank you Jorge
for your patience in showing me how a good scientist should work in a lab and out-
side. Thank you Kike for opening the doors to the world of teaching, I did not know
it could be so enjoyable.
Thank you very much to the current Nanophysics laboratory family, Celia, Max,
Dimas, Martina, Laura, Fred, Sara, Khadiza, Marco and Andrew. It’s been a pleasure
to share my time with you. I wish Fernando, Paul, Sabri and Nestor the best of luck
in your own PhD journey, you are surrounded by great people.
The first steps were not easy, but when accompanied by nice people, it is al-
ways easier. I would like to thank former group members Ana, Rubén, Alejandro,
Pavel, Guillaume, Jens, Mikel and Luciano for their support when the sky was dark.
Mikel, eskerrik asko lankide paregabea izateagatik. Jens, thanks for sharing your
knowledge and pizzas at weekends. I wish you all the best for your future.
This thesis has been possible due to many international collaborations that we
have carried out. It has been a very enriching scienctific activity, but most impor-
tantly, I have had the chance of meeting incredible people from all over the world.
Dzi˛ekuj˛e Sylwia and Olha from the University of Basel for helping us understand
the fascinating world of nanoporous systems. Sylwia, it was a pleasure to have you
with us and show you the beauty of my land. Olha, thanks for your support during
thesis, is it snowing in Basel?
Many thanks to Prof. Gottfried and Claudio from Philipps Universität Marburg
for nice collaboration, it’s been a pleasure to host Claudio with us and perform ex-
periments. Also, it was a pleasure to perform a short stay in Philipps Universität
Marburg at Prof. Höfers group, lovely little town and equally nice research group.
Thanks for hosting us. Many thanks to Prof. Meike Stöhr from the University
of Groningen for fruitful collaboration. I would like to acknowledge group mem-
bers Leonid, Jun and Juan Carlos for the experiments performed together. Jun and
Leonid, it was a pleasure to host you in San Sebastian, I wish you the best. Thank
you very much to Prof. Nian Lin from The Hong Kong University of Science and
Technology for collaborating with us. Guowen, I hope you keep nice memories of
your stay in San Sebastian and hope that your thesis is successful.
I am very grateful to the people in Universidad de Zaragoza, Fernando and
Leyre, for all the beamtimes that we have shared, enjoyed and suffered together.
Leyre, good luck with your thesis, you will do great! Uuuum?
I am very grateful to Shigeki, Nakayama-sensei and Uchihashi-sensei for al-
lowing me to work at your facilities in Tsukuba. Working with you and travel-
ing to Japan has been a life changing experience. Thank you very much Atori-san,
Nomura-san and Li Quiao Quiao for making my stay so enjoyable, domo arigato.
vi
I would like to thank our theory collaborators in San Sebastian, Andrés, Iker and
Aran for helping us understand our experimetal results with their wonderful DFT
calculations. Thanks for your patience when answering all our questions.
Before concluding, I would like to thank Jorge, Kike, Zaka, Khadiza, Afaf, Eli and
Martina for carefully reading this manuscript, suggesting wise changes and helping
during writing. Thank you Eli for such a nice cover, you do have a good taste.
Zakaria, thanks a million for your support, daily scientific conversations and in-
valuable help with my thesis. I wish you the best in your scientific career, you will
do great. Coffee?
Last but not least, I would like to thank my family for giving me the support I
needed during this journey. Thank you, Marc and Ainhoa for the nice pintxopotes
and dinners we shared together and kuadrila members for dragging me out of the
office and enjoying life. Eskerrik asko Egoi for listening, afaita gaur?
vii
Contents
Abstract iii
Acknowledgements v
Contents vii
List of Figures xi
List of Tables xv
List of Abbreviations xvii
1 Part I: Introduction and Motivation 1
2 Experimental Techniques and Theoretical Methods 11

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Ultra-High Vacuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Sample Growth: the Gradient or Wedge Method . . . . . . . . . . . . . 13
2.4 Laboratory Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4.1 STM Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4.2 ARPES Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
ARPES Spectroscopy System . . . . . . . . . . . . . . . . . . . . 16
2.5 Experimental Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.1 Angle-Resolved Photoemission Spectroscopy . . . . . . . . . . . 18
2.5.2 Principles of ARPES . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.3 Theoretical Description of the Photoemission Process . . . . . . 21
Three-Step Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
One-Step Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5.4 Electronic Structure of Noble Metal (111) Surfaces . . . . . . . . 24
2.5.5 Low Energy Electron Diffraction . . . . . . . . . . . . . . . . . . 26
2.5.6 Scanning Tunneling Microscopy and Spectroscopy . . . . . . . 30
2.5.7 Scanning Tunneling Spectroscopy . . . . . . . . . . . . . . . . . 31
2.5.8 Noncontact Atomic Force Microscopy . . . . . . . . . . . . . . . 32
2.6 Theoretical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.6.1 Electron Boundary Elements Method and the Electron Plane
Wave Expansion Method for Periodic Systems . . . . . . . . . . 34
2.6.2 Density Funcitonal Theory . . . . . . . . . . . . . . . . . . . . . 35
viii
The Many-Body Problem . . . . . . . . . . . . . . . . . . . . . . 35

Density Functional Theory: Hohenberg-Kohn Theorems and
Kohn-Sham Equations . . . . . . . . . . . . . . . . . . 36
2.6.3 Phase Accumulation Model . . . . . . . . . . . . . . . . . . . . . 37
3 Part II: An Introduction to Assembly and Electronic Bands of Nanoporous

Networks 39
4 Temperature Dependence of the Partially Localized State in a 2D Molecular

Nanoporous Network 47
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Supplementary Information for This Chapter . . . . . . . . . . . . . . . 55
4.4.1 ARPES Measurements . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4.2 STM Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4.3 Method Used for the ARPES Spectral Deconvolution . . . . . . 56
5 Precise Engineering of Quantum Dot Array Coupling Through Their Bar-

rier Widths 57
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.4.1 ARPES Sample Preparation . . . . . . . . . . . . . . . . . . . . . 68
5.4.2 STM/AFM Measurements . . . . . . . . . . . . . . . . . . . . . . 69
Detection of CO Molecules and Furan Group in DW Network . 70
Defect Concentration in SW and DW Networks . . . . . . . . . 71
From 2D Halogen-Bonded Networks to 1D Covalent Polymers 71
5.4.3 EBEM/EPWE Simulations . . . . . . . . . . . . . . . . . . . . . . 72
Simulation Procedure . . . . . . . . . . . . . . . . . . . . . . . . 72
5.4.4 Ab-Initio Calculations . . . . . . . . . . . . . . . . . . . . . . . . . 74
6 Tunable Energy and Mass Renormalization from Homothetic Quantum Dot

Arrays 77
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.3 Corroboration of 2DEG Renormalization in Other MOCNs . . . . . . . 86
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.5.1 Sample Preparation for ARPES . . . . . . . . . . . . . . . . . . . 89
6.5.2 STM/STS Measurements . . . . . . . . . . . . . . . . . . . . . . 89
6.5.3 Confined State Tunability with Pore Size . . . . . . . . . . . . . 90
ix
6.5.4 EBEM/EPWE Simulations . . . . . . . . . . . . . . . . . . . . . . 91

6.5.5 The Phase Accumulation Model . . . . . . . . . . . . . . . . . . 93
6.5.6 DFT Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7 Effective Determination of Surface Potential Landscapes from Metal-Organic

Nanoporous Overlayers 95
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.4.1 ARPES Measurements . . . . . . . . . . . . . . . . . . . . . . . . 105
Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.4.3 LCPD/AFM Measurements . . . . . . . . . . . . . . . . . . . . . 106
7.4.4 Side Considerations to This Chapter . . . . . . . . . . . . . . . . 107
8 Configuring Electronic States in an Atomically Precise Array of Quantum

Dots 111
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
8.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
8.4.1 Sample Preparation and ARPES Acquisition Details . . . . . . . 120
Sample Preparation for STM/STS Measurements . . . . . . . . 122
Repositioning of Single Xe Atoms . . . . . . . . . . . . . . . . . 122
STM/STS Measurement Details and Data Analysis . . . . . . . 122
9 Part III: Introduction to On-Surface Synthesized One-Dimensional Zigzag

Covalent Polymers 125
10 Electronic Structure Tunability by Periodic meta-Ligand Spacing in One-

Dimensional Organic Semiconductors 129
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
10.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.4.1 VT-STM and LT-STM/STS Measurements . . . . . . . . . . . . . 141
10.4.2 Ab-Initio DFT Calculations . . . . . . . . . . . . . . . . . . . . . . 141
10.4.3 The Electron Plane Wave Expansion Method (EPWE) . . . . . . 142
10.4.4 Sample Preparation and Zigzag Alignment on a Ag(111) Curved
Vicinal Crystal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
x
10.4.5 ARPES Band Structure and Photoemission Intensity Simula-

tions Using EPWE . . . . . . . . . . . . . . . . . . . . . . . . . . 145
10.4.6 ARPES Band Structure Comparison of Zigzag Chains vs PPP
Chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
10.4.7 Band Structure Variations with Straight Segment’s Length of
the Zigzag Chains from DFT Calculations . . . . . . . . . . . . . 148
11 Conclusions and Outlook 151

11.1 Conclusions to Part II: Nanoporous Networks . . . . . . . . . . . . . . 151
11.2 Conclusions to Part III: Zigzag Chains . . . . . . . . . . . . . . . . . . . 155
11.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
A Resumen de las Conclusiones Principales 161

A.1 Estructura Electrónica de Redes Orgánicas Nanoporosas Generadas
en Superficies de Metales Nobles . . . . . . . . . . . . . . . . . . . . . . 161
A.2 Estructura Electrónica de Cadenas Orgánicas Covalentes Unidimen-
sionales Sintetizadas en Superficie . . . . . . . . . . . . . . . . . . . . . 166
B List of Publications 167
Bibliography 169
xi
List of Figures
1.1 Top-Down and Bottom-Up Fabrication Methods . . . . . . . . . . . . . 2

1.2 STM Imaging and Manipulation . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Self-Assembled Molecular Nanoporous Networks . . . . . . . . . . . . 4
1.4 Electronic Coupling of QD Arrays on Surfaces . . . . . . . . . . . . . . 5
1.5 On-Surface Synthesized 1D and 2D Covalent Nanostructures . . . . . . 7
1.6 Ullmann Coupling Reaction and Long-Range Ordering on Surfaces . . 8
2.1 ARPES Gradient Evaporation Method . . . . . . . . . . . . . . . . . . . 14

2.2 STM Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 ARPES Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Geometry and Energetics of ARPES . . . . . . . . . . . . . . . . . . . . 19
2.5 Parallel Momentum Conservation . . . . . . . . . . . . . . . . . . . . . 21
2.6 One-Step and Three-Step Models . . . . . . . . . . . . . . . . . . . . . . 22
2.7 The Universal Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.8 Surface and Bulk Electronic States . . . . . . . . . . . . . . . . . . . . . 24
2.9 Dispersion and Fermi Surface Maps (FSM) of Noble Metal (111) Surfaces 25
2.10 LEED Schematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.11 LEED Pattern Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.12 Schematic Drawing of STM . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.13 STS vs ARPES Cu(111) Surface State . . . . . . . . . . . . . . . . . . . . 31
2.14 NC-AFM Measurements on Pentacene . . . . . . . . . . . . . . . . . . 33
2.15 PAM Solution For (111)-Terminated Noble Metals . . . . . . . . . . . . 38
3.1 A Journey Into Confining Nanostructures . . . . . . . . . . . . . . . . . 41

3.2 Electronic Band Structure of 3deh-DPDI Nanoporous Network . . . . . 43
4.1 Coverage Dependent Formation of 3deh-DPDI Molecular Network . . 49

4.2 Coverage Dependent ARPES Band Structure of 3deh-DPDI Molecular
Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Temperature Dependent Cu SS and PLS Band Structures . . . . . . . . 52
4.4 Full Range Temperature Dependence of PLS and CuSS Fundamental
Energies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5 Cu SS and DPDI PLS Spectra Fittings . . . . . . . . . . . . . . . . . . . . 56
5.1 QD Arrays Generated by Single-Wall (SW) and Double-Wall (DW)

Nanoporous Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
xii
5.2 Bonding Motifs That Stabilize the SW and DW Formations . . . . . . . 59

5.3 Local Electronic Structure of the SW and DW Networks . . . . . . . . . 60
5.4 Modification of the Au(111) 2DEG with Ag Film Thickness . . . . . . . 61
5.5 Electronic Bands from SW and DW Networks Along ΓM . . . . . . . . 62
5.6 Electronic Bands from SW and DW Networks Along ΓK . . . . . . . . 63
5.7 SW and DW Induced SBZ Distortions of the 2DEG . . . . . . . . . . . . 64
5.8 EBEM/EPWE Simulations of the SW and DW Local Electronic Structure 65
5.9 Localization of Bonding and Antibonding States . . . . . . . . . . . . . 66
5.10 Molecular Arrangements of Br-DNT Deposited on Au(111) . . . . . . . 69
5.11 Adsorption of CO Molecule and Imaging of the Oxygen Atom by AFM 70
5.12 Network Imperfections After Array Formation . . . . . . . . . . . . . . 71
5.13 Halogen Bond Stabilized 2D Nanoporous Network Transformation to
1D Covalent Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.14 Influence of the Effective Mass on the EBEM/EPWE Simulated SW
and DW Band Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.15 Electronic Band Structures Simulated by EBEM/EPWE . . . . . . . . . 74
5.16 Models for SW and DW Networks on Ag(111) . . . . . . . . . . . . . . 75
6.1 STM Characterization and EBEM/EPWE Scattering Regions of Ph6Co

and Ph3Co QD Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.2 ARPES Electronic Bands of Ph3Co and Ph6Co QD Arrays . . . . . . . . 80
6.3 EBEM/EPWE Simulated Electronic Bands of Ph6Co and Ph3Co QD
Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.4 Ph6Co Confinement Effects and Downward Shift Observed by STM/STS 82
6.5 DFT and PAM Simulations on Hybridization and Overlayer Height
Effects on the SS Reference . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.6 Au(111) SS Gradual Band Bottom Energy Shift upon Co, Ph3 and Ph6
Gradient Depositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.7 Cu-TPyB QD Array on Cu(111) . . . . . . . . . . . . . . . . . . . . . . . 87
6.8 Ph6Co and Ph3Co Confined State Energy Tunability with QD Size . . . 90
6.9 EBEM/EPWE Band Structure and LDOS Simulations on Ph6Co Using
Repulsive/Attractive Potential Combinations . . . . . . . . . . . . . . . 91
7.1 3deh-DPDI Metal-Organic Network on Cu(111) . . . . . . . . . . . . . 96

7.2 EBEM/EPWE Simulated and Experimental LDOS for the 3deh-DPDI
Metal-Organic Network . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.3 Three-Fold Symmetry Effects on the Confined States in 3deh-DPDI . . 99
7.4 Experimental and Simulated Electronic Band Structures and Isoener-
getic Maps of 3deh-DPDI Network . . . . . . . . . . . . . . . . . . . . . 101
7.5 Reported Renormalization of 2DEGs by Physisorbed Overlayers . . . . 102
7.6 Xe Adsorption on 3deh-DPDI Network . . . . . . . . . . . . . . . . . . 104
7.7 Evolving into 3deh-DPDI Band Structure . . . . . . . . . . . . . . . . . 106
7.8 ARPES and EBEM/EPWE Raw Data of 3deh-DPDI Network . . . . . . 107
xiii
7.9 Characterization of the 3deh-DPDI Band Structure . . . . . . . . . . . . 108

7.10 EBEM/EPWE Simulations of the Confined State Energy Within an In-
finite Potential Pore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.11 EBEM/EPWE Simulated and STS LDOS at the Pore Center of 3deh-
DPDI Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.1 QD Electronic Structure Alteration Through Xe Occupation . . . . . . . 113

8.2 3deh-DPDI Electronic Band Structure upon Xe-Filling . . . . . . . . . . 114
8.3 Effective Mass Renormalization After Xe-Filling of 3deh-DPDI Net-
work Nanopores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.4 Electronic Structure of Isolated Xe-Filled QD or Empty QD . . . . . . . 116
8.5 LDOS for Empty and Xe-Filled Pores Simulated with EBEM/EPWE . . 117
8.6 EBEM/EPWE Simulated LDOS for Different Xe-Filling Configurations 119
8.7 Changes in Electronic Structure of the Cu-Coordinated 3deh-DPDI
Network upon Xe Exposure Monitored by ARPES . . . . . . . . . . . . 121
9.1 On-Surface Synthesis of GNRs and Nanoporous Graphene . . . . . . . 126
10.1 Structural Arrangement of the Zigzag Chains . . . . . . . . . . . . . . . 131

10.2 ARPES Electronic Band Structure of Zigzag Chains . . . . . . . . . . . 132
10.3 Effect of Br on the Electronic Structure of Isolated and Condensed
Zigzag Chains Grown on Ag(111) . . . . . . . . . . . . . . . . . . . . . . 133
10.4 DFT Molecular Orbital Shape and Band Structure Comparison be-
tween Zigzag Chains and Straight PPP Chains . . . . . . . . . . . . . . 135
10.5 STS Experimental Determination of the Zigzag Chain’s Frontier Or-
bitals on Ag(111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.6 Tuning the Electronic Bound States through the Linear Segment’s Length138
10.7 Influence of Phenyl-Phenyl Twisting on the Stability of the Zigzag
Chains and Their Electronic Bandgap . . . . . . . . . . . . . . . . . . . 139
10.8 Atomic and Macroscopic Geometry of the 100% kinked Curved Ag(111)
Crystal and Its Electronic Structure . . . . . . . . . . . . . . . . . . . . . 143
10.9 Uniaxial Zigzag Chain Ordering vs Multi-Domain Structures Deter-
mined by LEED and STM . . . . . . . . . . . . . . . . . . . . . . . . . . 144
10.10Raw Data Comparison to the 2nd Derivative Treatment for the Elec-
tronic Band Structure of Zigzag Chains . . . . . . . . . . . . . . . . . . 145
10.11Experimental ARPES Band Structure Comparison to EPWE Photoe-
mission Intensity Simulations of Zigzag Chains . . . . . . . . . . . . . . 146
10.12N =4 Zigzag Chain ARPES Band Structure Comparison with that of
PPP Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
10.13DFT Band Structure Comparison between PMP, Zigzag and PPP Chains148
10.14Evolution of the Zigzag Chain Frontier Orbital Bandgap with Increas-
ing Number of Phenyl Rings at the Straight Segments . . . . . . . . . . 149
11.1 ARPES Electronic Band Structure of the Studied Nanoporous Networks152

xiv
11.2 Summary of the Zigzag Chain’s Electronic Structure Probed by ARPES

and STS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
11.3 Fe Nanocluster Size and Assembly Control When Grown on Cu-DCA
MOCN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
11.4 Organic Topological Insulators . . . . . . . . . . . . . . . . . . . . . . . 157
11.5 Image Potential States in Cross-Conjugated Oligophenylene Zigzag
Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
A.1 Resumen de la Estructura de Bandas de las Diferentes Redes Nanoporosas

Estudiadas con ARPES . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
A.2 Estructura Electrónica de las Cadenas Zigzag Medidas con ARPES y
STS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
xv
List of Tables
2.1 Noble Metal (111) Shockley State Dispersion Fit Results . . . . . . . . . 26

2.2 Noble Metal (111) Vacuum Level and Bulk Band Gap Edges . . . . . . 38
5.1 Extracted ARPES Experimental Parameters from SW and DW Networks 62
6.1 ARPES Experimental BEs at Γ and Effective Masses of Ph3Co and

Ph6Co QD Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.2 Energy Shift of the Au(111) SS with Co Adatom Concentration and
Array Size Obtained from DFT . . . . . . . . . . . . . . . . . . . . . . . 84
6.3 The Parameters Used in PAM for Modeling Noble Metals . . . . . . . . 93
7.1 EBEM/EPWE Potential Parameters Used in the Models Discussed Through-

out This Chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.2 Scattering Potential Values for Molecules (Vmol ) and Metal Centers
(Vmet ) for Three Different Networks . . . . . . . . . . . . . . . . . . . . 103
8.1 Band Sructure Parameters Extracted From ARPES Measurements on

Empty and Xe-Filled 3deh-DPDI Network . . . . . . . . . . . . . . . . . 114
8.2 EBEM/EPWE Reference and Scattering Parameters Used for Different
Xe-Filling Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . 117
11.1 Summary of ARPES and EBEM/EPWE Values Obtained for Nanoporous

Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
A.1 Resumen de las Propiedades Electrónicas Extraídas de las Medidas de

ARPES y Referencias y Potenciales para EBEM/EPWE . . . . . . . . . 163
xvii
List of Abbreviations
2DEG Two-Dimensional Electron Gas

AGNR Armchair Graphene Nanoribbon
ARPES Angle-Resolved Photoemission Spectroscopy
BE Binding Energy
BEM Boundary Element Method
Br-DNF 3,9-dibromodinaphtho [2,3-b:2’,3’-d]furan
Br-DNT 3,9-dibromodinaphtho [2,3-b:2’,3’-d]thiophene
CB Conduction Band
CCD Charge-Coupled Device
COF Covalent Organic Framework
DBTP 4,4"-dibromo-p-terphenyl
DCA 9, 10-Anthracenedicarbonitrile
DFT Density Funcitonal Theory
DMTP 4,4"-dibromo-meta terphenyl
DOS Density of States
DPDI 4,9-diaminoperylene quinone-3,10-diimine
DW Double Wall
EBEM Electron Boundary Element Method
EDC Energy Distribution Curve
EPWE Electron Plane Wave Expansion
FSM Fermi Surface Map
GNR Graphene Nanoribbon
Gr Graphene
hBN Hexagonal Boron-Nitride
IPS Image Potential State
KPFM Kelvin Probe Force Microscopy
LCPD Local Contact Potential Difference
LDOS Local Density of States
LEED Low Energy Electron Diffraction
LT Low Temperature
ML Monolayer
MOCN Metal Organic Coordination Network
MOF Metal Organic Framework
NC-AFM Noncontact Atomic Force Microscopy
NPG Nanoporous Graphene
xviii
OTIs Organic Topological Insulators

PAM Phase Accumulation Model
PES Photoemission Spectroscopy
Ph3 Para-terphenyl-dicarbonitrile
Ph4 Para-tetraphenyl-dicarbonitrile
Ph5 Para-pentaphenyl-dicarbonitrile
Ph6 Para-hexaphenyl-dicarbonitrile
PLS Partially Localized State
PMP Poly-(meta-phenylene)
PPP Poly-(para-phenylene)
QD Quantum Dot
QMB Quartz Micro Balance
RT Room Temperature
SBZ Surface Brillouin Zone
SOC Spin-Orbit Coupling
SS Surface State
STM Scanning Tunneling Microscopy
STS Scanning Tunneling Spectroscopy
SW Single Wall
TI Topological Insulator
TPyB 1,3,5-tri(4-pyridyl)-benzene
TWD Terrace Width Distribution
UHV Ultra-High Vacuum
UPS Ultraviolet Photoemission Spectroscopy
VB Valence Band
VT Variable Temperature
XPS X-Ray Photoemission Spectroscopy
ZGNR Zigzag Graphene Nanoribbon
xix
To family and friends, and to you.

1
Chapter 1
Part I: Introduction and Motivation
The modern inspiration of nanoscience and nanotechnology, the idea that atoms and
molecules could be used to construct nanoscale devices and machines, dates back to
December 29th 1959 when Professor Richard Feynmann delivered his famous speech
There’s Plenty of Room at the Bottom at the annual meeting of the American Physical
Society at Caltech [36]. In his talk, he predicted exciting new phenomena that might
revolutionize science and technology and affect our everyday lives, if only we were
to gain precise control over matter, down to the atomic level.
The terms nanoscience and nanotechnology were coined, almost 15 years after
Feynmann’s lecture, by Norio Taniguchi in 1974 [37] to refer to the study of the in-
trinsic properties of nanoscale objects. These go from ∼ 100 nm to ∼ 1 Å in size and
encompass the capabilities of observing, using, manipulating, or organizing them
into assemblies in order to perform specific functions [1, 38].
Creating materials at the nanoscale is still nowadays technologically challenging.
The two main basic approaches for generating surface patterns and devices in a con-
trolled and repeatable manner are the top-down and bottom-up techniques [39]. In
the former approach, materials are patterned from larger structures, while in the lat-
ter, materials are made by chemical processes starting from their atomic and molecu-
lar units. More precisely, the top-down approach [Figure 1.1 (a)] is commonly used,
for instance, in manufacturing silicon integrated circuits, where tiny transistors are
built up and connected in complex circuits starting from a bare silicon wafer. Re-
lying on processes like thin film depostion, lithography and etching using masks,
among others, it is possible to fabricate electronic devices with features below 100
nm. The ultimate size is limited by the physical laws governing these techniques,
i.e. the wavelength of light and etch reaction chemistry [1].
An alternative and very promising strategy to exploit science and technology at
the nanometer scale is offered by the bottom-up approach [Figure 1.1 (b)]. It starts
from nano- or subnano-scale objects (namely, atoms or molecules) to build nanos-
tructures [38]. In essence, the bottom-up approach makes use of intermolecular in-
teractions and self-asembly to produce a desired pattern or nanostructure [38]. This
approach became a reality and gained much interest with the invention of the STM in
1981. This extensively used Surface Science tool, developed at IBM Zurich Research
2 Chapter 1. Part I: Introduction and Motivation
a Top-down
b Bottom-up
F IGURE 1.1: Building matter at the nanoscale. (a) Top-down fabrication methods
such as lithography, writing or stamping are used to define the desired patterns
at the nanoscale (adapted from ref. [1]). (b) The bottom-up technique makes use
of self-assembly protocols in order to form ordered materials at the atomic and
molecular level (adapted from ref. [6]).
Laboratory by Nobel Prize winners Binnig and Rohrer, is based on the tunneling ef-
fect [40], and enables imaging surface structures at the atomic scale [Figure 1.2 (a)].
However, STM is more than a tool for imaging surfaces, as its use was soon ap-
plied to atomic manipulation at the nanoscale. In 1990 Eigler et al. [41] reported the
positioning of single Xe atoms on a Ni crystal surface with atomic precision [Fig-
ure 1.2 (b)]. Some years later, Crommie et al. [18] created a quantum corral by using
Fe atoms on a Cu(111) surface, evidencing standing wave patterns of surface state
electrons confined within the Fe ring [Figure 1.2 (c)]. However, such tip manipula-
tion fails in the formation of long-range ordered nanostructures.
Supramolecular Chemistry, the chemistry of the intermolecular non-covalent
bond, emerged as a powerful technique to develop nanoarchitectures with potential
impact in a wide variety of fundamental and applied fields of research [44]. The con-
cept of self-assembly has been recognized as a most elegant tool for creating ordered,
functional materials at all scales [45]. The most widespread use of this autonomous
assembly process is probably found in solution chemistry, where supramolecular
structures are formed from tailored molecular building blocks. For the development
of this strategy, Jean-Marie Lehn, Donald James Cram and Charles J. Pedersen were
honored with the Nobel Prize in chemistry in 1987 [44, 46]. However, this knowl-
edge of Supramolecular Chemistry in solution cannot be directly translated to guide
the assembly of adsorbed molecules on atomically clean surfaces. The influence of
the substrate’s atomic lattice and electronic sturcture on non-covalent bonds strongly
affects the resulting structures. Such a chellenge provides a new frontier in the field
of Chemical Physics as it uniquely combines the versatility of Organic Synthesis and
Chapter 1. Part I: Introduction and Motivation 3
aa aa
a b
aa aa
c d
B A
C
F IGURE 1.2: STM imaging and manipulation of matter at the nanoscale. (a) STM
image of Xe adsorbed on Ni(110) (adapted from ref. [41]). (b) Construction of a
patterned array of xenon atoms on a nickel (110) surface (adapted from ref. [42]).
(c) 48-atom Fe ring constructed on a Cu(111) surface (adapted from ref. [18]). (d)
Three adsorbed CO molecules (A, B and C) on a Cu(111) terrace (adapted from
ref. [43]).
the Physics of Interfaces [1, 2].

This chemist’s concept of molecular self-assembly was transferred to the sur-
face science community, starting first with the adsorption of simple molecules onto
single-crystal surfaces under well-controlled conditions in UHV [Figure 1.2 (d)]. The
main objective of early studies was to learn about the molecules’ physical and chem-
ical interactions with the supporting/underlying substrate [43]. From this single-
molecule perspective, studies moved towards investigating more complex molecu-
lar ensembles, in which intermolecular interactions began playing a role. So began
the study of molecular self-assembly on surfaces [1, 47]. Guided by the advances
and power of chemical synthesis, which provides access to a potentially vast range
of functionally and structurally diverse building blocks (known as tectons), it was
soon recognized that tuning the balance between molecule-surface and intermolec-
ular interactions offered the potential for creating different supramolecular architec-
tures on surfaces with tailored properties [1, 3, 46–48] [Figure 1.3 (a)].
Molecular assemblies can be classified as covalent or non-covalent, depending
on the bonds formed between the molecules. The reversibility of non-covalent inter-
actions (predominantly hydrogen bonds and metal-ligand interactions) allows error
correction through elimination of defective structures. This results in the forma-
tion of almost defect-free, long-range ordered supramolecular architectures such as
nanoporous networks [Figure 1.3 (b, c)]. Contrarily, molecular structures based on
a b
c
d
F IGURE 1.3: Self-assembled molecular nanoporous networks on noble metal sur-

faces varying intermolecular interactions. (a) Schematic representation of molec-
ular self-assembly on surfaces (adapted from ref. [49]). (b) Hydrogen bond sta-
bilized anthraquinone molecular nanoporous network on Cu(111) (adapted from
ref. [50]). (c) Size varying ditopic dicarbonitrile-polyphenyl molecular linkers (NC-
Phn -CN) coordinated with Co adatoms and forming long-range ordered metal-
organic nanoporous networks on Ag(111) (adapted from ref. [51]). (d) Covalent net-
work formed from BDBA and HHTP molecules on Ag(111) (adapted from ref. [52]).
the formation of irreversible covalent bonds do not allow for defect correction. How-
ever, their stability and distinct electronic structure is nowadays driving a strong re-
search effort. On-surface synthesis provides the opportunity to create short-range
ordered covalent organic frameworks (COFs), even if their periodicity is still not yet
satisfactory [Figure 1.3 (d)].
The second part of this thesis is devoted to the growth of single domain, long-
range ordered nanoporous networks on (111)-terminated coinage metal surfaces
in solvent-free UHV. The nanoporous networks have been obtained from differ-
ent molecular precursors and long-range ordering has been achieved through self-
healing halogen bonding and metal-organic coordination of precursors. Such kind
of metal-organic networks grown on metallic surfaces show novel magnetic proper-
ties [4, 5], catalytic effects [6, 7], oxidation states [8], exotic tesellation patterns [9–12]
a b
STM/STS (site specific) ARPES (ensemble information)
F IGURE 1.4: Electronic coupling of QD arrays on surfaces. (a) Conceptual

schematic picture of a STM tip repositioning single Xe atoms on a metal-organic
nanoporous network on Cu(111). This picture highlights the ability of STM not
only to access the electronic structure of nanoporous networks at the single pore
level but also of modifying it with adsorbate manipulation [53]. (b) Conceptual
schematic picture of a halogen bond stabilized nanoporous network on Ag(111).
Each pore confines the 2DEG of Ag, defining a QD. These QDs couple with the
neighboring ones due to intercoupling effects and generate electronic bands which
can be accessed with spatially averaging techniques such as ARPES [54].
and even bear the prospect of exhibiting topological band structures [13, 14].
Moreover, nanoporous structures with periodicities ranging from ∼ 2 to 6 nm
are ideal 2D systems for altering the surface electronic properties of the supporting
metals. The regular nanopores interact with the 2DEG present on the surface, and do
so on length scales of the order of the Fermi wavelength (λF ). This is in fact of a few
2π
nanometers on (111) surfaces of noble metals (λF = kF , where kF is the momentum
of the surface electron at the Fermi energy). As a consequence the 2DEG is confined
inside each nanocavity and changes its energy [Figure 1.4 (a)], in a similar fashion to
quantum corrals [18] [Figure 1.2 (c)]. By using scanning tunneling microscopy and
spectroscopy (STM/STS), it is possible to probe such electronic features at the sin-
gle pore or quantum dot level [15–17]. Interestingly, the resulting array of QDs are
known to couple with each other [Figure 1.4 (b)], giving rise to the formation of new
electronic bands with distinct fundamental energies, dispersion and zone boundary
gaps. These features critically depend on the QD shape, size and lattice constants
of the arrays formed [15]. These electronic bands become accessible when spatially
averaging techniques such as angle-resolved photoemission spectroscopy (ARPES)
are used.
In the second part of this thesis our aim is to study and engineer electronic bands
by constructing different types of nanoporous networks. By subtle changes in the
molecular constituents (functional groups and/or carbon backbone size) and guided
by reversible, non-covalent bonding mechanisms, we are able to form a variety of
QD arrays on noble metal surfaces with tunable QD sizes, barrier separations and
scattering strength. As a result, we engineer the surface band structure by locally

controlling the confinement properties of each nanopore. Supported by semi em-
pirical simulations, we disentangle the repulsive scattering potential landscape that
organic and metal-organic nanoporous networks exert on the 2DEG and also delve
into subtle surface state-organic overlayer interactions that show up as hybridization
effects and geometrical variations. In this way we achieve a general understanding
and predictability of the electronic properties of this family of nanoporous network
overlayers grown on noble metals.
The resulting electronic structure from nanoporous networks that we study is

a consequence of the modification of the substrate’s 2DEG, which is independent of
the existing molecular states. In the third part of this thesis, we investigate possi-
ble confinement effects on surface synthesized molecular structures. Thus, we study
the electronic signal stemming from polymeric organic chains. These structures are
reminiscent units of graphene, which has been studied in depth due to its excep-
tional electronic properties, with its singular Fermi surface with linear dispersions
associated with massless Dirac electrons at the K point of the surface Brillouin zone
(SBZ). Many efforts have been devoted to integrating this 2D material into electronic
devices. However, in electronic devices semiconducting materials play a promi-
nent role, so using graphene as a base material in nanoelectronics would depend on
the ability to engineer its semimetallic π bands in order to open energy gaps at the
Fermy energy. Different attempts have been made: On the one hand, functional-
ization of graphene by doping with external species, such as hydrogen [55], and on
the other hand, laterally reducing the dimensions of the graphene sheet limiting its
size in one-dimension (1D). The latter results in a measureable band gap caused by
quantum confinement of the electrons [21, 56]. A way to limit graphene to 1D with
atomic precision is by on-surface chemical reactions [57, 58]. These have proven to
be a highly versatile bottom-up tool for the creation of stable molecular structures
such as 1D polymers [Figure 1.5 (a)] and GNRs [Figure 1.5 (b)] or 2D COFs [59–
63] [Figure 1.5 (c)]. These graphitic examples bear intrinsic semiconducting proper-
ties with different bandgaps and distinct valence and conduction band dispersions.
They represent interesting examples not only for studying their fundamental elec-
tronic properties [26, 28–30, 64], but also to be used in (opto)electronic and sensing
devices and in catalysis [24, 33].
In 2000, Lha et al. [67] presented a pivotal study carried out in UHV, induc-
ing the classical Ullmann coupling reaction with the STM tip, by creating biphenyl
from iodobenzene precursor molecules on a Cu(111) surface [67, 68]. Similarly, the
demonstration of the thermal initiation of an on-surface covalent linking was pre-
sented by Grill et al. [69] in 2007. In their study, an Ullmann-like reaction was per-
formed using bromine substituted porphyrin molecules on Au(111). Depending on
the number of halogen substituents on the precursor, dimers, linear chains or ex-
tended networks could be obtained [Figure 1.6 (a)]. A decade after Grill’s seminal
a c
2 nm
1 nm 5 nm
F IGURE 1.5: On-surface synthesized 1D and 2D covalent nanostructures. (a)

STM image showing poly-(para-phenylene) oligomers (adapted from ref. [65]). (b)
Constant-height nc-AFM frequency shift image taken with a CO-functionalized tip
of the zigzag edge graphene nanoribbon (ZGNR) (adapted from ref. [28]). (c) STM
image of a covalent network on Ag(111) (adapted from ref. [66]).
work, a multitude of different reactions have been proven possible on various con-
ducting substrates including Ullmann-like coupling, cyclodehydrogenation, Glaser-
like coupling of terminal alkynes and condensation reactions such as boronic acid
based chemistry [57]. Owing to the interesting electronic properties and remarkable
chemical and mechanical stability [70], there have been many attempts to achieve
long-range order in COFs [52, 63, 71–74]. For instance, Figure 1.6 (b) shows a quasi-
regular COF [73], but the irreversibility of covalent bonds, restricting rearrangement
and self-correction processes, prevents the formation of equilibrium structures that
result in an increased amount of geometrical defects [73, 75].
Guided by the asymmetry present at the surface of certain metallic crystals, it
is in fact easier to synthesize and order 1D covalent nanostructures than the afore-
mentioned 2D COFs. In this regard, 1D poly-(para-phenylene) chains [26, 77, 78]
and GNRs with different edge topologies [27, 78, 79] have been successfully synthe-
sized and ordered by using vicinal surfaces as nanotemplates [80–82] [Figure 1.6 (c)].
These vicinal surfaces are made up of monoatomic steps separating homogenously
distributed terraces. These miscut surfaces are used as nucleation sites for the on-
surface reaction, as well as for steering the alignment and packing of polymeric
chains into single-domain films extending long-range on the surface. Such long-
range ordering allows the access to their electronic band structure with ARPES [26,
27, 77–79, 83]. Recently, the formation of highly-regular nanoporous graphene (NPG)
has been achieved by laterally fusing well-aligned GNRs with engineered edge topol-
ogy [24] [Figure 1.6 (d)], proposing a promising strategy for building 2D COFs with
distinct electronic properties.
The frontier orbitals of 1D oligophenyele chains and GNRs, which are respon-
sible for their chemical, optical and electronic properties, strongly depend on the
topology of the conducting pathways, i.e. width, edge-type and conjugation of the
a b
10 nm
c
d
4 nm
F IGURE 1.6: Ullmann coupling reaction and long-range ordering on surfaces.

(a) Ullmann coupling of substituted porphyrins with control over the resulting
structure dimensionality. By using different Br substituents, formation of dimers,
one-dimensional lines or two-dimensional networks are generated (adapted from
ref. [69]). (b) Sequential polymerization of a high quality 2D polymer with hexag-
onal pores accompanied by other irregular geometries (adapted from ref. [73]).
(c) STM image of Chevron-type nanoribbons aligned on Au(788) (adapted from
ref. [76]). (d) Nanoporous graphene (NPG) formation after laterally fusing one-
dimensional GNRs (adapted from ref. [24]).
semiconducting nanostructures. In the third part of this thesis, we introduce period-

ically spaced topological variations in the form of meta-junctions into a straight and
linearly conjugated poly-(para-phenyle) chain and proceed to synthesize and align
cross-conjugated zigzag shaped oligophenylene chains into long-range ordered films
on a vicinal silver crystal. ARPES and STS measurements evidence a radical change
of the electronic structure as compared to its straight counterpart [26, 29], featur-
ing shallow dispersive bands, an enhanced frontier orbtial bandgap, electron con-
finement effects along the interrupted small straight segments and frontier orbital
bandgap tunability.
This thesis is composed by the following parts: the first part will be completed
with Chapter 2 where the experimental and theoretical methods used are briefly
described, highlighting the sample preparation process. This initial step, is an es-
sential prerequisite for the formation of long-range ordered nanostructures and for
probing their electronic structure with non-local techniques. In the second part of
the thesis, we start with a general introduction into nanoporous networks in Chap-
ter 3 and unravel the electronic band structures from six different organic and metal-
organic nanoporous networks grown on metal surfaces (Chapters 4 to 8). Here we
also disentangle the scattering potential landscapes and their interactions with the
2DEG present on the surface. In the third part of the thesis, we move into on-
surface synthesis studies and start with a general introduction in Chapter 9 and then,
in Chapter 10, the electronic band structure of periodically spaced meta-junctioned
oligophenylene chains is reported. Finally, this thesis ends presenting the main con-
clusions, an outook section that includes prospective new lines of research (Chap-
ter 11), as well as a short summary in spanish (Resumen de las principales conclusiones),
the author’s publication list and the bibliography section.
11
Chapter 2
Experimental Techniques and

Theoretical Methods
2.1 Introduction
This chapter gives a brief account of the main experimental details and techniques
and theoretical methods used in this thesis. The chapter begins with an introduction
into ultra-high vacuum and then, the sample preparation and coverage gradient
growth method is described in detail as it is key for performing spatially averaging
studies. Next, the laboratory setup in which the photoemission experiments have
been carried out is shown. During the second half of this chapter the main exper-
imental techniques (ARPES, STM and LEED) are explained, as well as, the experi-
mental techniques provided by our collaborators (STM/STS and NC-AFM). Finally,
the theoretical methods that have been used to support our experimental findings
are briefly introduced: the combined electron plane wave expansion and electron
boundary element method, density functional theory and the phase accumulation
model. Note, that theoretical calculations have been performed through collabora-
tions.
12 Chapter 2. Experimental Techniques and Theoretical Methods
2.2 Ultra-High Vacuum

In surface science experiments, it is important to perform the measurements in a
clean environment, where the surfaces being probed remain clean (without impuri-
ties or particles sticking on the surface) as long as possible. Such time must last at
least as long as the acquisition time, that ranges from 30 min to 24 hours. To achieve
this, samples must be kept in UHV conditions, which means at pressures lower than
10−9 mbar. UHV requires operating different types of pumps: roughing pumps,
turbo pumps and ion pumps. Roughing pumps are used to put the system down to
10−1 mbar. Then, turbo pumps can be activated and pressures as low as 10−11 mbar
can be obtained inside the vacuum chamber. Once such a stable pressure is reached,
this conditions can be kept by using ion-pumps [84].
Another key factor for a successful study of the electronic properties of a mate-
rial is its surface quality (monocrystalline and atomically flat) and cleanliness (non
presence of impurities on the surface). If dirt is present on a surface it alters the mea-
surement, for instance in ARPES, since the impurities can act as scattering centers
for the 2DEG and increase dramatically the background, masking the main spec-
troscopic signal. Therefore, the (111)-terminated coinage metals used in this thesis
require a defined cleaning protocol before any experiment is performed. This clean-
ing process comprises of a combination of sputtering and annealing cycles:
• For ion sputtering a noble gas (generally Ne or Ar) is injected into the vacuum
chamber up to a pressure of ∼ 10−6 mbar through the sputter gun. The gas is
ionized (Ne+ or Ar+ ) and a focused ion beam with an energy of 0.5 to 2 keV
is directed against the sample to remove the outermost atomic layers. Such
inert gases are chosen so that they do not bind to the substrate and avoid any
additional source of sample contamination. The aftermath of this process is
that the ion bombardment creates a large number of defects on the surface.
In order to repair such defects after sputtering, a thermal annealing process is
required.
• The annealing process consists in supplying thermal energy to the sample. For
the noble metal single crystals studied in this thesis, the heat is generated by
a filament that is placed behind the sample. In order to increase the heating
efficiency and temperatures (generally up to ∼ 700 − 800 K), a positive bias
voltage (high-voltage) is applied on the sample so that the emitted electrons
from the filament are accelerated against it. This process is generally done for
10 minutes reaching final pressures in the low 10−9 mbar.
2.3. Sample Growth: the Gradient or Wedge Method 13
2.3 Sample Growth: the Gradient or Wedge Method

In this thesis the in situ thermal evaporation is used to deposit single layer molec-
ular films on surfaces. Molecules are sublimated in UHV conditions by supplying
thermal energy from a radiative filament. For such purposes, a home-built molec-
ular evaporator [Figure 2.1 (a, b)] is used, which is mounted inside the preparation
chamber and focused into the central point of the chamber. The molecules are de-
posited inside a quartz crucible that is surrounded by a W filament. Upon heating
the filament (direct heating evaporator), the crucible containing the organic powder
raises its temperature until it sublimates with a desired rate towards the sample.
One particular advantage of our ARPES preparation setup is that instead of per-
forming a homogeneous, single coverage deposition, we are able to perform a gradi-
ent or wedge evaporation. The gradient method gives us the possibility of altering
the coverage of either molecules or metal depositions on a large sample. This gra-
dient evaporation can be performed both on the vertical axis of the manipulator (z
axis) and/or the horizontal axis of the manipulator (x axis). This is possible due to
the presence of a mask and the possibility to move the manipulator in small steps
(µm/s) using motors. The deposition process is as follows:
• The molecule flux is firstly monitored by using a quartz crystal microbalance

(QMB) [green arrow in Figure 2.1 (c)]. During this process, the freshly clean
sample is kept hidden behind the mask shielding wall (red arrow). Generally
the pressure during evaporation remains below ∼ 2 × 10−9 mbar.
• Once the rate is stable, the QMB is retracted, the mask is placed at the focal
point of the evaporation beam and the sample is fully exposed to the molecules
[Figure 2.1 (d)]. In the case where a clean region is required to be kept on
the sample, a small fraction of the surface can be hidden behind the mask.
Generally, the sample can be further approached towards the mask (y axis) so
that the sample is positioned close to the focal point.
• In order to perform the gradient evaporation in the vertical (z) or horizontal (x)
direction of the sample, the manipulator is moved using motors in a stepwise
fashion (µm/s) during the evaporation, going from a fully-exposed situation
to a completly hidden one behind the mask. A schematic picture of the final
evaporation gradient is shown in Figure 2.1 (e). Note that for molecular depo-
sitions, generally a very shallow gradient is chosen, ranging from 0 to 1 ML,
while the evaporation time ranges from 5 to 20 min.
• To make sure that the gradient evaporation was successful, the post-evaporation
rate is again monitored using the QMB.
a In-built QMB b
Thermocouple
Quartz crucible
W filament
Side view Top view
c Hidden d Fully exposed e

Δz
x y
F IGURE 2.1: ARPES gradient evaporation. (a, b) Home-built direct heating evap-
orator. The evaporator is inserted into the exchangeable port of the preparation
chamber either in ARPES or the STM side. The evaporator is composed of two
quartz crucibles, which are filled with molecules. Each crucible is wound up by
0.5 mm thick W filament. Current passing through the filament (generally ranging
from 1 to 5 A) will heat up the quartz crucible. The temperature can be monitored
by a thermocouple (blue arrow) positioned between the filament and the crucible
cap. A temperature uncertainty of ∼ 50◦ C may exist between the inner and outer
region of the crucible. (c) The molecular deposition rate is monitored by position-
ing a quartz micro balance (QMB) in front of the molecule evaporation beam. A
mask is used in order to hide the already clean sample behind it to avoid any direct
contamination. (d) When the evaporation rate is stable, the QMB is retracted and
the sample is fully exposed to the molecular beam. During the evaporation, the
manipulator is moved by a motor in the z or x axis in small steps until the sample
is completly hidden behind the mask. (e) Schematic representation of a resulting
molecular gradient evaporation (in green), which generally ranges in close proxim-
ity to 1 ML deposition.
2.4. Laboratory Setup 15
2.4 Laboratory Setup

The laboratory setup mainly used for this thesis corresponds to the high-resolution
ARPES chamber of the Nanophysics lab that is divided into two interconnected sec-
tions: the STM setup and the ARPES setup (Figures 2.2 and 2.3). In the following,
both setups will be explained separately.
2.4.1 STM Setup
The STM setup (Figure 2.2) is equipped with a preparation chamber (in orange) to
perform the sample cleaning process and the subsequent molecular deposition. The
molecule depositions are performed using an exchangeable port with a home-built
molecular evaporator (in purple). The base pressure in this chamber is 3 × 10−10 mbar.
In order to characterize the samples, they are brought into the STM chamber (in
blue). The STM chamber is equipped with a LEED (yellow) and a variable tempera-
ture (VT-Omicron) STM (inset). In general, the STM is operated at room temperature
(RT), except for the characterization of the zigzag polymers (Chapter 10) where the
sample was cooled down to ∼ 100 K to improve resolution and stability. The base
pressure in this chamber is 1 × 10−10 mbar.
Preparation
LEED Chamber
STM Chamber
VT-STM Molecule
Evaporator
F IGURE 2.2: The STM laboratory setup used. It is divided into two main cham-
bers, the preparation chamber (orange) and the STM analysis chamber (blue). The
preparation chamber is equipped with all the necessary tools for sample cleaning
and a exchangeable molecular evaporator (in purple) is used for performing the
molecular depositions. The STM analysis chamber (in blue) consists of a LEED (in
yellow) and a variable temperature VT-STM (see inset).
2.4.2 ARPES Setup
The ARPES setup [Figure 2.3 (a)] is divided into two sections: the preparation cham-
ber (in blue) and the ARPES analysis chamber (in green). A vertical cryogenic ma-
nipulator (closed He cycle), which can vary its temperature from 30 to 320 K, con-
nects both chambers. The manipulator is equipped with two large (15 x 15 mm)
Au(111) and Cu(111) single crystals and an exchangeable holder to insert any other
sample [Figure 2.3 (e)]. In the preparation chamber, the samples are cleaned and or-
ganic and inorganic evaporations (in orange) can be performed on a cold substrate, if
desired. This chamber is equipped with a mask that enables us to perform coverage
gradient evaporations into such large crystals, as described in Section 2.3 (see Fig-
ure 2.1). This is very advantageous since it facilitates obtaining the optimal coverage
much quicker than performing single-shot coverage molecular depositions. Note
that photoemission is a spatially averaging technique, thus, long-range ordered, sin-
gle domain nanostructures have to be prepared optimally covering the surface. The
base presseure of the ARPES UHV system is 8 x 10−11 mbar, but increases up to
∼ 2 x 10−10 mbar when operating the discharge lamp that produces the UV light.
ARPES Spectroscopy System
The ARPES spectroscopy system consists of four major components: light source or
discharge lamp [Fig. 2.3 (b)], monochromator [Fig. 2.3 (c)], electron energy analyzer
[Fig. 2.3 (d)] and sample manipulator [Fig. 2.3 (e)]. The photoemission data is mea-
sured with a display-type hemispherical electron analyzer (SPECS Phoibos 150) and
a monochromatized He discharge lamp positioned under an incidence angle of 45◦
to the analyzer direction [Figure 2.3 (e)]. The lab based gas discharge lamp usually
has an energy resolution of 1.2 meV (for He case), a photon flux of 1012 /s, a spot
size of ∼ 1 mm2 and the light is partially polarized (90% p and 10% s light polar-
ization). In this thesis the He Iα line is commonly used providing a photon energy
of 21.2 eV. It has a photoelectron escape depth or bulk sensitivity of ∼ 5 − 10 Å,
which means it is very sensitive to the low dimensional structures formed on the
surface. Such low photon flux and large spot size are optimal for measuring or-
ganic systems (such as nanoporous networks) grown on surfaces since photon in-
duced beam damage is avoided. Synchrotron radiation has a photon flux of 1013 /s
or even larger and a very collimated spot size of less than 0.2 mm in diameter. In
these conditions organic films are strongly damaged by the secondary electrons and
it is specifically the beam damage the reason behind the use of the gas discharge
lamp for the ARPES investigation of the selected systems in this thesis. The main
drawback of the He discharge lamp is the fast aging effect of samples at very low
temperatures due to the flowing helium. That is why ARPES measurements are per-
formed at 150 K, since it provides the best compromise between measurement time
and band structure quality. This temperature is controlled using a closed-cycle He
2.4. Laboratory Setup 17
a b
c
Preparation
Chamber
Molecule
Evaporator
Analyzer d
Analysis
Chamber
Monochromator UV Lamp
e
q
f
Au(111) Analyzer q
e-
Channelplate
Cu(111)
hn kx
ky
F IGURE 2.3: The ARPES laboratory setup. (a) The ARPES chamber is divided into
the preparation chamber (in blue) and the analysis chamber (in green). The prepa-
ration chamber is equipped with all the necessary tools for sample cleaning. An
exchangeable molecular evaporator (in orange) is used in order to perform the
molecular deposition. A vertical manipulator connects the preparation chamber
with the ARPES analysis chamber. The ARPES analysis chamber is equipped with
an ARPES setup. A discharge lamp (in yellow) is used in order to generate the in-
cident beam (b), then the monochromator (in purple and (c)) is used to monochro-
matize the light at the desired energy (for instance He Iα ; 21.22 eV) getting rid of
satellites. The light source impinges on the samples (e) and the emitted photo-
electrons are analyzed in a hemispherical analyzer (d). (f) A channelplate detector
provides a defined angular range in the vertical direction (kx ) while the manipula-
tor polar angle (θ) around its vertical axis (z) is used to measure the dispersion in
the horizontal axis (ky ).
cryogenic manipulator with a controlled temperature PID. Regarding the incoming

photon energy of the light, this is properly defined by means of a monochromator.
This prevents the presence of satellites and refocuses the light, which is essential
for high angular resolution. In addition, the electron energy analyzer is one of the
core components on ARPES systems. Our ARPES system uses a hemispherical en-
ergy analyzer which is made by two concentric hemispheres with radius R1 and R2 .
A constant voltage V is applied between the two hemispheres, and only electrons
with enegy window Ep = e∆V /(R1 /R2 − R2 /R1 ) can go through the hemispherical
analyzer and be counted by a CCD detector. The angular mode of the analyzer al-
lows us to measure multiple energy distribution curves (EDCs) simultaneously on
a channelplate window of ±7◦ (a 2D snapshot of energy versus momentum), which
at the He Iα excitation is large enough to map the complete relevant k range in one
direction without any sample rotation. The angles in the orthogonal direction re-
quired for obtaining Fermi surface mappings (FSMs) are reached by a subsequent
change of the manipulator rotation around the z axis (θ polar angle) as indicated in
Figure 2.3 (e, f). In our usual operation conditions, our ARPES setup features en-
ergy/angle resolution of 40 meV/0.1◦ .
2.5 Experimental Techniques

2.5.1 Angle-Resolved Photoemission Spectroscopy
Photoemission spectroscopy (PES) defines all techniques based on the application

of the photoelectric effect as a tool for probing the collective behavior of electrons
in a solid, that is, to study the electronic structure of solids [85]. PES makes it pos-
sible the understanding of fundamental physics laws which dominate the behavior
of bound electrons, the chemistry of various materials and even has practical ap-
plications in several areas such as material science and surface chemistry. Ultravio-
let photoemission spectroscopy (UPS), X-ray photoemission spectroscopy (XPS) and
angle-resolved photoemission spectroscopy are some examples of this family.
Among the previous examples, ARPES represents a powerful and direct method
to study the valence band electronic structure of solids because it directly probes the
momentum-dependent electronic structure of solids, supplying detailed informa-
tion on the band dispersion and Fermi surface topology. These results are achieved
by measuring, at the same time, both the kinetic energy and angular distribution of
the electrons photoemitted from a sample [85–89].
2.5.2 Principles of ARPES
The PES technique is based on the photoelectric effect, initially discovered by Hertz
in 1887 [90], and later explained by Einstein [91]. The photoelectric effect describes
the mechanism by which a metal emits electrons (thus called photoelectrons) when
light shines upon it. The geometry of a PES experiment is illustrated in Figure 2.4 (a):
2.5. Experimental Techniques 19
a b
F IGURE 2.4: Geometry and energetics of ARPES. (a) In an ARPES experiment the
emission direction of the photoelectron is specified by the polar (θ) and azimuthal
(ϕ) angles. (b) The electron energy distribution produced by the incoming photons,
and measured discarding the background signal as a function of the kinetic energy
Ekin of the photoelectrons (right), can be referred to the density of states inside the
solid (left) measured as a function of the binding energy EB . Panels have been
adapted from ref. [89].
a beam of monochromatic photons with energy hν (supplied either by a gas dis-

charge lamp or by a synchrotron beamline) is incident on the sample (typically single
crystal), and those electrons which have absorbed incident photons can be excited
from an initial state with energy Ei to a final state with energy Ef [85]. If the energy
of the photon is high enough, an electron gains sufficient energy to escape from the
solid, is ejected into the vacuum with a certain kinetic energy Ekin and a certain di-
rection defined by the emission angles θ and ϕ. The term photoelectron is used for
such ejected electrons.
By collecting the photoelectrons with an electron energy analyzer characterized
by a finite acceptance angle, one measures their kinetic energy Ekin for a given emis-
sion direction. Knowing the work function of the material (φ), the kinetic energy
(Ekin ) of the photoelectron, and the energy of the incident radiation (hν), we can ob-
tain the binding energy (EB ) of the electrons inside the solid from the conservation
of energy (Equation 2.1).
Ekin = hν− | EB | −φ (2.1)
The EB is the minimum energy necessary for extracting a bound electron from a
solid, the electron work function is a measure of the potential barrier at the surface
that prevents the valence electrons from escaping into the vacuum, and is typically
4-5 eV in metals [88, 89, 92].
The energetics of the photoemission process is shown in Figure 2.4 (b). The en-
ergy level diagram of the electrons inside the solid is shown on the left side while
the PES spectrum measured by the analyzer without the background is shown on
the right side [89]. Core levels give rise to sharp peaks in the photoemission spec-
trum. However, in this work we are more interested in the valence electrons which
are close to the Fermi level since they are the ones responsible for various physical
properties of solids (DC conductivity, magnetism, superconductivity, etc). There-
fore, while core electrons correspond to the strongly bound electrons which are re-
sponsible for the formation of the inner orbitals of a certain material, valence elec-
trons are responsible for the formation of the energy bands. These shared electrons
are weakly bound to the nuclei and their modification can result in drastic changes
in the material properties.
While in UPS experiments it is only possible to measure the kinetic energy of the
electrons integrated over momentum space, in ARPES it is also possible to get infor-
mation about the internal momentum k of the electrons, and hence to obtain the dis-
persion of the electronic valence bands in the reciprocal lattice inside the solid. This
is done by measuring the angle at which the electrons are emitted from the sample
(θ and ϕ). The geometry of an ARPES experiment is shown in Figure 2.4 (a): the an-
alyzer collects the photoelectrons and measures both their kinetic energy Ekin and
their emission direction, which finally provides the K in vacuum. From these values
one can obtain information respectively on the binding energy and the momentum
√
P 2mEkin
k of the electrons inside the solid. The magnitude of K is given by K = ~ = ~
and the magnitudes of its components Kx , Ky and Kz are determined in terms of the
polar (θ) and azimuthal (ϕ) angles as follows:
1p
Kx = 2mEkin sinθ cosϕ (2.2)
~
1p
Ky = 2mEkin sinθ sinϕ (2.3)
~
1p
Kz = 2mEkin cosθ (2.4)
~
From the previous equations, the momentum K in vacuum is then completely deter-
mined and given by K = Kx + Ky + Kz .
Regarding the momentum k of an electron inside the solid, the in-plane compo-
nent of K, the one lying in the plane of the surface of the sample, is Kk = Kx + Ky
while the component of K perpendicular to the surface is K⊥ = Kz . Given that the
momentum component of the emitted photoelectron parallel to the sample surface
is conserved (Figure 2.5), because of translational symmetry along the surface, then
the component of the electron momentum parallel to the surface is also conserved at
the interface, i.e. going from inside the solid to the vacuum,
kk = Kk (2.5)
KII
Vacuum
K┴ K
Surface
Solid k┴ k
kII
F IGURE 2.5: Schematic representation of the momentum k of the electron inside

the solid and the momentum K in vacuum. The component of the momentum
parallel to the surface is conserved because of the conservation of the translational
symmetry going from inside to outside the solid, which means that the vectors kk
and Kk have the same magnitude. On the contrary, the perpendicular component
of the momentum is not conserved when the electron escapes from the solid.
therefore, the parallel component of the momentum inside the solid (kk ) is fully
determined by measuring the emission angle of the photoelectron and its kinetic en-
ergy in vacuum.
1p
q
kk = K2x + K2y = 2mEkin sinθ (2.6)
~
Unfortunately, the perpendicular component of the momentum is not conserved
across the sample surface because of the sudden change of potential along the z axis
(k⊥ 6= K⊥ ). However, if a nearly free-electron description for the final bulk states is
assumed, the perpendicular component k⊥ can be determined as a function of Ekin ,
θ and the inner potential V0 , which corresponds to the energy of the bottom of the
valence band referenced to the vacuum level EV :
1p
k⊥ = 2m(Ekin cos2 θ + V0 ) (2.7)
~
Fortunately, such uncertainty in k⊥ is less relevant for low-dimensional systems
since their band dispersion is negligible along the z axis and thus, is exclusively
determined by kk . Note that in real ARPES experiments, one requires atomically-
flat surfaces for an ideal conservation of surface parallel-momentum [88, 89].
2.5.3 Theoretical Description of the Photoemission Process
The most common semi-classical approach for explaining the photoemission pro-
cess, particularly when PES is used as a tool to map the electronic band structure
of solids, is the three-step model. This model decomposes the photoemission event
into three independent steps: optical excitation between the initial and final bulk
a b
F IGURE 2.6: Pictorial representation (a) of three-step and (b) one-step model de-
scriptions of the photoemission process. Panels have been adapted from ref. [89].
Bloch eigenstates, travel of the excited electron to the surface and escape of the pho-
toelectron into vacuum after transmission through the surface potential barrier [Fig-
ure 2.6 (a)]. A more exact and computationally complex description is provided by
the one step model, which is solely based in quantum mechanics. In the following
we will discuss both models starting by the three step one.
Three-Step Model
The base hypothesis of this model is that the photoemission intensity can be decom-
posed as a product of three independent contributions which do not interfere with
each other. The three independent steps are the following:
• Step 1: Optical excitation of the electron from the initial bulk eigenstate | ψi >
into the final excited bulk eigenstate | ψf >. It contains all the information
about the intrinsic electronic structure of the material.
• Step 2: Propagation of the excited electron to the surface. Generally the propa-
gation of the excited electron to the surface is described in terms of an electron
mean free path λ, which describes the probability of the excited electron to ar-
rive at the surface of the sample without undergoing any inelastic scattering
and so without suffering any change in its kinetic energy or momentum. The
curve for an electron mean free path has a universal trend, with a minimum at
about 5 to 20 Å at the kinetic energies region of 20 to 100 eV, which are typical
energies for ARPES experiments (Figure 2.7). This rather small electron escape
depth implies that many PES experiments are sensitive only to the topmost
surface layers, which is the reason why photoemission is considered a surface
sensitive technique [93].
F IGURE 2.7: Universal curve. The curve for an electron mean free path for different
materials show that its values closely follow the same trend, with a minimum at
about 5 to 20 Å at the kinetic energies region of 20 to 100 eV, which is the typical
energies for ARPES experiments [89].
• Step 3: Escape of the photoelectron into vacuum after transmission through

the surface potential.
However, this model is just an approximation. From a quantum-mechanical point of

view, photoemission should not be described in terms of several independent events
but rather as a one-step process.
One-Step Model
The one-step model [Figure 2.6 (b)] can be explained as an optical transition between
initial and final states consisting of many-body wave functions that obey appropri-
ate boundary conditions at the surface of the solid. The photoemission process is
described as a transition from an initial state ψi (initial electron wavefunction) to a
final state ψf (final electron wavefunction). Such a transition occurs with a probabil-
ity which is dictated by the Fermi’s golden rule:
1
Pi,f = | Mi,f |2 δ(Ef − Ei − hν) (2.8)
~
where δ(Ef − Ei − hν) is the energy conservation law and dictates that a transition
between ψi and ψf can only occur when hν = Ef − Ei , i.e. when the photon en-
ergy matches the energy difference between the energies of initial and final states.
Mi,f corresponds to the interaction matrix element and contains all the informa-
tion regarding the electronic interactions in a system and is directly connected to
the Hamiltonian which describes the many-body energetic balance of the material,
Mi,f =< ψf | H | ψi >.
2.5.4 Electronic Structure of Noble Metal (111) Surfaces
The electronic structure (valence band) of noble metal surfaces is composed of d-

bands, which are occuppied and highly-dispersive sp-bands that cross the Fermi
level. In addition to these, surface state electrons, known as Shockley states, are
also present on the surface in the form of a two-dimensional electron gas. 2DEGs
have played a pivotal role in fundamental condensed-matter physics for decades.
While in semiconductors 2DEGs derive from the confinement of bulk states into a
thin layer, electronic surface states on metals have a different origin. Understanding
surface states requires understanding what happens to the Hamiltonian of the sys-
tem when we approach a surface, i.e. when the bulk material is abruptly interrupted.
The existence of a surface implies a symmetry breaking (caused by the termination
of the inifinite crystal by the surface) along the direction orthogonal to the surface,
and this can introduce new evanescent solutions to the Schrödinger equation in the
projected bulk band gaps residing only at the surface. We can describe a bulk state
[Figure 2.8 (a)] as an oscillating wave which exponentially decays into the vacuum
as it reaches the surface. A surface state, on the other hand, resides at the material
surface, decaying both into the vacuum and into the bulk [Figure 2.8 (b)]. Therefore,
for a surface state to exist, it must lie in an energy gap between the bulk bands, oth-
erwise it is named a surface resonance.
Prototypical examples are found at the Brillouin-zone center of the noble metal
(111) surfaces [94]. These Shockley states appear in a gap of projected bulk bands
along the Γ-L line. Electrons of the surface state behave like free electrons parallel to
the surface and form a 2DEG. Thus the dispersion relation can be described approx-
imately as
~2 2
E(kk ) = E0 + k (2.9)
2m∗ k
where E0 is the fundamental energy of the surface-state, m∗ is the effective mass of a
surface-state electron, and kk is the wave vector parallel to the surface [95]. As it will
be observed along this thesis, such Shockley states, which originate from pz orbitals
a Surface b Surface
Bulk state
Surface state
Bulk Vacuum Bulk Vacuum
F IGURE 2.8: Surface and bulk electronic states. (a) A bulk state can be described
as an oscillating wave which exponetially decays into the vacuum as it escapes the
surface. (b) A surface state instead resides mainly at the material surface, so it will
decay both into the vacuum and into the bulk.
within the sp inverted band gap around the L point, play an important role in the
study of lateral quantum confinement effects at metal surfaces [18, 96–98].
Figure 2.9 summarizes the ARPES band dispersion of Schockley states for (111)-
terminated coinage metals. Figure 2.9 (a-c) shows the E vs kk dispersion relation of
the Cu, Au and Ag Shockley states as a grayscale plot. The surface states can clearly
be detected as parabolic curves open at the Fermi energy, positioned inside the gap
of the projected band structures of the bulk. Each one of them is characterized by
a particular band bottom energy, effective mass and Fermi wave-vector (kF ), which
are summarized in Table 2.1. The effective mass depends on the strength of the
potential variations in the surface plane. Note that the particular character of each
surface state is mainly affected by the lower edge of the projected bulk bands [99].
The Shockley surface state on Au(111) [Figure 2.9 (b)] served as an early model
system exhibiting the characteristic spin momentum locking of a Rashba-split 2DEG.
LaShell et al. [101] explained this splitting as due to a spin-orbit coupling, that breaks
the spin degeneracy in the system. The dispersion in this case is described by two
parabolic subbands of equal effective mass and energy at the band bottom, which
is characteristic for a Rashba momentum splitting of a free-electron-like band near
the Brillouin-zone center. However, it was not until the impoved resolution of elec-
tron spectroscopic techniques incorporating lasers that the spin-orbit splitting of the
a Cu(111) b Au(111) c Ag(111)

0.0 0.0 -0.02
Binding energy (eV)
0.1 0.1
0.00
0.2 0.2
0.3 0.3 0.02
0.4 0.4 0.04

0.5 0.5 0.06
0.6 0.6
0.08
0.7 0.7
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
angle off normal (°) angle off normal (°) angle off normal (°)
d e f
4 4 4
Tilt angle (°)
0 0 0
-4 -4 -4
-4 0 4 -4 0 4 -4 0 4
angle (°) angle (°) angle (°)
F IGURE 2.9: Dispersion and Fermi surface maps (FSM) of Noble metal (111) Shock-
ley states. (a, b, c) Dispersion of the Cu(111), Au(111) and Ag(111) surface states.
The spin-orbit splitting of Au(111) is clearly visible in (b), while the one of Cu(111)
could recently be achieved by using a focused laser excitation of 6 eV [inset in (a)].
(d, e, f) FSMs of the investigated Cu, Au and Ag noble metal surface states centered
at Γ. The two individual Fermi surfaces in (e) form two closed concentric circles
connected to the inner and outer branches of the spin-orbit split parabolas. Panels
have been adapted from refs. [94, 100].
Shockley surface state on Cu(111) could be observed as well [94]. For doing so, a fo-
cused (∼ 3 µm) 6 eV continuous-wave laser was used as the excitation source [insets
in Figure 2.9 (a, d)]. Figure 2.9 (d, e, f) shows the Fermi Surface Maps (FSM) of the
investigated Cu, Au and Ag noble metal surface states centered at Γ. The circular
shape of the FSM is an indication of the isotropic dispersion of these surfaces. The
radius of the ring depends on both the band bottom and the effective mass of the
surface state and gives information on the number of electrons the 2DEG contains
2
kF
(electron density n = 2π ). In all cases the ring is centered at Γ, where the center of the
gap is. Note that photoemission with the discharge lamp generally integrates over
a large sample area of about 1 mm2 , with all its steps and defects and other imper-
fections. Hence, preparing a homogeneous atomically-flat sample in the illuminated
area and avoiding the aging effect due to contamination are key for obtaining well-
resolved datasets.
TABLE 2.1: Noble metal (111) parabolic Shockley state dispersion fit results with
TS = 30 K (values have been taken from ref. [100]).
E0 [meV] m∗ /me kF [Å−1 ]

Ag 63 ± 1 0.397 0.080
Au 487 ± 1 0.255 0.167/0.192
Cu 435 ± 1 0.412 0.215
2.5.5 Low Energy Electron Diffraction
LEED is the principal technique for determining surface structures in reciprocal

space by means of the diffraction of an electron plane wave at the sample sur-
face. When the diffraction pattern is recorded, the analysis of the spot positions
yields information on the size, symmetry and rotational alignment of the adsorbate
unit cell with respect to the substrate unit cell. In a LEED experiment, monochro-
matic electrons are accelerated from a filament and focused with a lens system [Fig-
ure 2.10 (a)]. After reflection from the sample surface, only the elastically-scattered
electrons contribute to the diffraction pattern: the lower energy (secondary) elec-
trons are removed by energy-filtering grids placed in front of the fluorescent screen
that is employed to display the pattern. The fluorescent screen then detects only the
elastically backscattered electrons. The LEED experiment uses a beam of electrons
in the range 10-200 eV and beam size of around 0.5 mm2 . These electrons can be
diffracted by lattices of atomic dimensions. This is so because the de Broglie wave-
length of the low energy electrons (λ = hp ) is comparable to the interatomic distances
in the crystal (a few Å). Since low energy electrons are strongly back-scattered by the
target electronic cloud, their elastic penetration is low ∼ 10 − 50 Å, making LEED a
surface sensitive technique. The sample itself must be a single crystal with a well-
ordered surface structure in order to generate a back-scattered electron diffraction
pattern. To explain the diffraction pattern it is usual to work in reciprocal space (~gi )
instead of in the real space (~ai ). The reciprocal surface lattice is defined as follows,
2π(~aj × ~n)
~gi = , i, j = 1, 2 (2.10)
| ~ai × ~aj |
The Laue conditions, which express the conditions required for diffraction to oc-
cur, in a two dimensional system are given by,
(~ki − ~kf ) · ~ai = 2πk, i = 1, 2 (2.11)
On the other hand, energy conservation requires that,
| ~kf |=| ~ki | (2.12)
Both conditions can be represented in the Ewald construction [Figure 2.10 (b)].
Here, rods represent the reciprocal lattice of the two-dimensional surface. The initial
momentum vector (~ki ) is drawn ending at the origin of the reciprocal lattice. There-
fore, a circle of radius (| ~ki |) will contain all the possible final momentum vectors
(~kf ), according to energy conservation q (Equation 2.12). The radius of the circle de-
∗
pends on the energy of electrons as 2m ~E2 kin . The intersenction between the circle
and the rods gives the ~kf vectors fulfilling the Laue conditions and showing diffrac-
tion maxima. Hence, the diffraction pattern observed on the screen is nothing but
a projection of the reciprocal lattice, which reflects the symmetry, the size, and the
shape of the unit cell in real space.
a b Incident
Beam (10) Diffracted
Grids Beam
Electron Real space

Gun kf φ
ki kf
2θ
Sample
Δk
2θ
ki φ
Fluorescent
Screen
+V
(00)
F IGURE 2.10: LEED schematics. (a) The LEED experiment uses a beam of electrons
of a well-defined low energy incident normally onto the sample. The sample itself
must be a single crystal with a well-ordered surface structure in order to generate a
back-scattered electron diffraction pattern. Only the elastically-scattered electrons
contribute do the diffraction pattern: the lower energy (secondary) electrons are
removed by energy-filtering grids placed in front of the fluorescent screen that is
employed to display the pattern. (b) The diffraction pattern is an image of the
reciprocal space. The Ewald sphere is showing the Laue condition for the existence
of a diffracted beam, ~ki − ~kf = ∆k
~ = ~g
In this thesis the LEED is used to gain information about the symmetry and su-
perlattice constants of the on-surface self-assembled nanoporous networks and poly-
meric arrays with respect to the crystal surface. The LEED is also used for checking
the cleanliness of the (111) noble metal surfaces after sputtering and annealing, as
well as the quality and homogeneity of the terraces on vicinal surfaces [102].
In the following we will discuss different examples yielding particular LEED
patterns. For a crystalline surface, for instance Cu(111), six sharp spots are observed
at low energies when the surface is atomically-flat hosting ∼ 100 nm size or larger
terraces. The six-fold symmetry of the LEED pattern is directly related to the (111)
termination symmetry of the crystal [Figure 2.11 (a)].
Regular step arrays on a vicinal surface cause the spots to split in two or more
satellite spots. The split spots are due to the superperiodicity from the step array.
This happens because the vertical rods in the Ewald sphere are repeated due to the
extra periodicity introduced by the step array. The splitting is oriented perpendic-
ular to the steps (it gives the step flow direction), thus the step direction can be
determined by LEED [102]. A typical LEED pattern of a vicinal Cu(111) surface is
2π
displayed in Figure 2.11 (b). In addition, the size of the spot splitting is d , so that
the terrace width (d) can also be determined by LEED analysis. Clearly separated
spots indicate a well-defined terrace width distribution (TWD). In addition a low
background and sharp spots indicate good crystal quality free of defects.
For two-dimensional molecular nanoporous networks, superlattice spots pro-
vide information of the commensurability, periodicity, chirality and single or multi-
domain scenarios present. For instance, Figure 2.11 (c) shows a flower-shape metal-
organic nanoporous network on Cu(111) [104]. This particular nanoporous network
forms a single-domain and commensurate 8 × 8 superstructure on the surface. This
means that its periodicity (∼ 2 nm) is eight times the interatomic distance of Cu(111).
The high quality of the LEED pattern confirms that these molecules form a long-
range ordered metal-organic coordination network (MOCN) extending up to the
mm range.
a aaa
aaa
F IGURE 2.11: LEED patterns compared to STM images. (a) LEED pattern of the
six-fold Cu(111) surface and corresponding STM image highlighting the existance
of large terraces separated by monoatomic high steps. Ripples observed in the
STM image correspond to the standing waves of the 2DEG after scattering at step
edges with oscillations of 15 Å. LEED parameters: 80 eV; STM image adapted from
ref. [19]. (b) LEED pattern obtained at a highly stepped region of a curved Cu(111)
crystal. The splitting of the spots is directly related with the average terrace size. 30
x 30 nm2 STM image of Cu(335) vicinal surface with a terrace size of 8.1 Å. LEED
parameters: 152 eV; STM image adapted from ref. [103]. (c) LEED pattern corre-
sponding to a 8 x 8 molecular DCA nanoporous superlattice on Cu(111), demon-
strating its commensurate and single domain character of the overlayer extending
in the long-range. The STM image (20 × 20 nm2 ) corroborates the high degree
ordering of such metal-organic assembly (Courtesy of Leyre Hernández). LEED
parameters: 50 eV and STM parameters: I=300 pA/V=-1 V
2.5.6 Scanning Tunneling Microscopy and Spectroscopy
In STM a sharp metallic tip (normally W or PtIr that is ideally terminated by one sin-
gle atom) is scanned over the surface and the tunneling current between the tip and
the slighlty biased sample is measured [Figure 2.12 (a)]. The most important prereq-
uisite of the scanned surface or material is that it has to be conductive. Theoretical
analysis of electron tunneling through the vacuum barrier showed that the tunnel-
ing current IT depends on the local density of states (LDOS) of the sample [105]. For
low bias voltages Vs the tunneling current It can be written in terms of LDOS of the
sample (Ds ), the tip-sample distance d, and the work functions of the sample (Φs )
and the tip (Φt ):
r !
2m Φs + Φt
It ∝ Ds (EF ) exp −2d (2.13)
~2 2
The sign of the bias voltage Vs determines whether the STM probes filled (Vs < 0
V, electrons tunnel into the tip) or empty (Vs > 0 V, electrons tunnel into the sample)
states. Hence, due to the exponential dependence of the tunneling current on the
distance d, it is possible to measure topographic contrast of the surface. There are
two main operating modes that depend on the feedback control (height) between the
tip and the sample: constant current or constant height mode. In constant current
mode [Figure 2.12 (b)], the z position of the tip is adjusted in a feedback loop during
the scan so that the tunneling current is kept at a desired value. The change of tip
height is then recorded as a function of the lateral coordinate. In constant height
mode the feedback loop is disabled, so the z is fixed while the tunneling current It is
recorded.
a b
x, y scanning unit
z piezo
motion of
tip the tip at
feedback scanning direction constant
current
tunnel current
F IGURE 2.12: Schematic drawing of scanning tunneling microscopy (STM). (a)

Piezoelectric elements position a metallic tip with fine resolution in three dimen-
sions over the sample surface. (b) A feedback loop adjusts the tip-sample separa-
tion z to keep the tunneling current constant (adapted from ref. [106]).
2.5.7 Scanning Tunneling Spectroscopy
The signal detected by STM is the convolution of surface electronic and topographic
landscapes. In order to measure the surface electronic states, it is necessary to de-
convolute the electronic information from the topographic structure. This is accom-
plished through scanning tunneling spectroscopy. STS enables the local, energy-
resolved investigation of a sample surface density of states (DOS) by measuring the
differential conductance (dI/dV ), which is approximately proportional to the DOS.
STS enables the local characterization of physical and chemical properties of con-
ducting samples. Typically, a small AC modulation (Vrms < 100 mV at 1 kHz) is
superimposed on the DC bias voltage while a lock-in amplifier records the first har-
monic of the signal that is in phase with the modulation. Unlike ARPES, STM/STS
can probe both occupied and unoccupied states depending on the selected bias volt-
age polarity [99]. In addition, it has the capability to investigate the electronic struc-
ture at the atomic scale, while simultaneously providing topographic information.
Therefore, STS is perfectly suited to study adsorbates positioned on substrates and
confinement effects of surface state electrons, providing a complementary approach
to k-vector resolving photoemission spectroscopy techniques.
A comparison between STS and ARPES was done on Cu(111) by Crommie et
al. [108]. They found a sharp increase of the differenctial conductande dI/dV signal
at -0.45 V and explained this increase as electrons tunneling from the occupied sur-
face state of Cu(111) into empty states of the tip. Therefore, such a sharp increase of
the dI/dV signal is recognized as the onset of the two-dimensional surface-state band
of the noble metal (111) surfaces [Figure 2.13 (a)]. Note that, apart from temperature
induced variations on the onset energy of the surface state E0 [109], this is usually
a 0.0 Cu(111) 5 K b 0.0
-0.1
-0.2 -0.2
E-EF (eV)
-0.3
Energy (eV)
-0.4 -0.4
-0.5 Cu(111)
-0.6 -0.6 G 150 K
-0.2 -0.1 0.0 0.1 0.2
kII (Å-1)
-0.8
dI/dV (arb. units)
F IGURE 2.13: STS vs ARPES Cu(111) surface state measurements. (a) Experimental
dI/dV spectrum taken from the bare Cu(111) sample highlights the step-like char-
acter of the surface state [107]. (b) ARPES band structure of Cu(111) Shockley state
shows its parabolic dispersion with a well-defined band bottom energy, effective
mass and Fermi wave-vector (kF ).
lower in STS measurements than ARPES [Figure 2.13 (b)] by 5-10 meV because the
electric field induced by the presence of an STM tip affects the surface electronic
structure, via the so-called Stark effect [95].
2.5.8 Noncontact Atomic Force Microscopy
Since the first systematic achievement of atomic resolution in 1995 [110], noncon-
tact atomic force microscopy (NC-AFM) became an important tool for real-space
high-resolution imaging. The tip-sample interaction force gradient is detected via
the resonance frequency shift ∆f of an oscillating cantilever. The atomic resolu-
tion mainly arises from the short-range interactions [111]. In NC-AFM, atomic res-
olution on molecules can be achieved by employing functionalized tips, which en-
hances the chemical sensitivity of the tip. For instance, CO molecule funtionaliza-
tion can be used in order to increase the lateral resolution down to the submolecular
regime [112] [Figure 2.14 (a)]. The requirements of the functionalization agent are
the following: inert tip to prevent the molecule to be imaged from being picked up
or displaced by the tip, small-size of the tip apex for improved lateral resolution,
atomic relaxations of the tip apex due to the forces acting on it and its charge distri-
bution (tip dipole) at its apex [112, 113].
The CO tip funtionalization [Figure 2.14 (b)] can be schematically summarized
as follows: CO molecules are deposited on the surface once the prepared sample is
in cryogenic conditions (normally at 4 K). A CO molecule is identified and picked up
by approaching the tip vertically towards the molecule by applying a certain voltage
(C atom is attached to the tip while the O atom points towards the molecule being
imaged). Finally, the tip functionalization is confirmed by detecting an image con-
trast change or by checking the non-presence of the CO molecule on the surface.
For instance, the constant-height AFM image of pentacene with a CO function-
alized tip is shown in Figure 2.14 (a). At this particular height, the mixture of attrac-
tive and repulsive forces felt by the CO tip is responsible for the submolecular reso-
lution. Indeed, the maximal attractive forces are observed above the hollow sites of
the carbon rings while the repulsive force originating from Pauli repulsion above the
molecules is responsible for the bright features at the positions of atoms and bonds.
In atomically resolved AFM images of molecules, the overall interaction forces are
attractive due to the electrostatic and vdW forces, but the atomic constrast stems
from repulsive contributions of the shorter-ranged Pauli repulsion. Such atomic res-
olution due to Pauli repulsion above the atoms and bonds becomes only apparent
for the smallest accessible tip heights [Figure 2.14 (c)]. The CO tip also tends to tilt
under the influence of the interaction forces. This tilting gives rise to an apparent
sharpening of the bonds and to distortions of the imaged molecular structures. In
addition, the increased brightness at the pentancene molecular ends and differences
between lower and upper edges stem from the non-planar adsorption geometry of
pentance. Hence, NC-AFM also provides information on the molecule adsorption
a Model
c
NC-AFM
F IGURE 2.14: NC-AFM measurements on pentacene. (a) Model of pentacene and

constant-height AFM measurement of pentacene on Cu(111) acquired with a CO
functionalized tip. (b) Schematic representation of the functionalization process of
the tip with a CO molecule. (c) Submolecular resolution of the pentacene molecule
depending on the molecule-tip separation. (a) and (c) have been adapted from
ref. [114].
geometry on the surface [113, 114].

2.6 Theoretical Methods

In this section we introduce the main modelling and theoretical methods that have
been used throughout this thesis, i.e. the combined electron plane wave expansion
(EPWE) and electron boundary elements method (EBEM), density functional theory
(DFT) and the phase accumulation model (PAM). Note that these calculations have
been performed by our collaborators.
2.6.1 Electron Boundary Elements Method and the Electron Plane Wave
Expansion Method for Periodic Systems
The combined EPWE [115] and EBEM [17, 116] have been developed by Prof. Dr.
García de Abajo and represents a scalar variant of the electromagnetic PWE/BEM
extensively used for solving Maxwell’s equations and optical response for arbitrary
shapes. It is based on Green’s functions for finite geometries and electron plane wave
expansion for periodic systems. For the band structure calculations, the particle-
in-a-box model is extended to infinite 2D systems by defining an elementary cell
and using periodic boundary conditions. Within the EPWE code, solutions of the
Schrödinger equation are represented as a linear combination of plane waves and a
satisfactory convergence is achieved with a basis set consisting of ∼ 100 waves [34,
115, 117].
In this thesis the local electronic properties of nanoporous networks such as
LDOS at a single nanopore as well as band structure calculations and photoemis-
sion intensity simulations of periodic arrays and one-dimensional polymeric chains
are performed by using this method. The Schrödinger equation for electrons expe-
riencing an effective potential V (for instance when surface electrons encounter a
molecular barrier in a hexagonal molecular nanoporous network) is written as:
−~2 2mef f
(∇2 + k 2 − V (x, y))ψ = 0 (2.14)
2mef f ~2
q
2mef f E
where k = ~2
is the electron momentum, E being its energy, and m∗ is the
effective mass. In the EPWE method, both the wavefunctions and the periodic po-
tential are Fourier expanded to obtain the following Equation 2.15, which is then
solved numerically by terminating the expansion at some finite reciprocal lattice
vectors (gmax , where gmax = 10 is a good compromise for capturing all the details
such as band dispersion and energy gaps).
~2 k 2 X
( − E)C(k) + UG C(k − G) = 0 (2.15)
2mef f
G=gmax
2.6. Theoretical Methods 35
Here C and U are the Fourier coefficients of the wavefunctions and potential, respec-
tively, and G is the reciprocal lattice vector. Further details on the simulation proce-
dure of each particular nanoporous network system and one-dimensional polymers
studied in this thesis are provided in each chapter.
2.6.2 Density Funcitonal Theory
In 1998 Walter Kohn was awarded with the Nobel Prize in Chemistry for his devel-
opment of DFT, which was introduced in two seminal papers in the 60’s [118, 119].
DFT is presently the most successfull approach to compute the electronic structure
of matter. Its applicability ranges from atoms, molecules and solids to nuclei and
quantum and classical fluids. In its original formulation, DFT provides the ground
state properties of a system and the electron density plays a key role. For instance, in
the chemistry field, DFT predicts a great variety of molecular properties: molecular
structures, vibrational frequencies, atomization energies, ionization energies, elec-
tric and magnetic properties, reaction paths, etc.
Since the field of molecular electronics came to birth, propelled by the possibility
of manipulating single molecules on surfaces, it became desirable to understand the
relation between conduction channels and molecular orbitals. For such purposes ab
initio quantum chemistry DFT became of great use [120]. In this thesis DFT has been
used for two purposes: on the one hand, to elucidate how single metal atoms such
as Co can influence the electronic properties of a 2DEG on a Au(111) sample, and, on
the other hand, for characterizing the electronic band structure and morphology of
zigzag-shaped organic semiconducting chains. More details on these two particular
examples are provided in Chapter 6 and Chapter 10.
In the following the basic principles of DFT will be introduced.
The Many-Body Problem
Calculating the properties of materials requires solving the Schrödinger equation for
a system of interacting Ne electrons and Nn nuclei, i.e. diagonalizing the many body
Hamiltonian,
Ĥ = T̂n + T̂e + V̂ne + V̂ee + V̂nn (2.16)
which can be written as,
~2 X 2 X ZI e2 1X e2 X ~2 1 X ZI ZJ e 2
Ĥ = − Oi − + − O2I +
2me ~i | 2 | ~ri − ~ri | 2MI 2 ~I − R~J |
i i,I | ~
ri − R i6=j I I6=J
|R
(2.17)
where, indices i, j are for electron degrees of freedom and I, J for nuclei, Mn denotes
the mass of nucleus at position R ~ I and me denotes the mass of the electron at ~ri . The
first two terms denote the kinetic energies of nucleus (T̂n ) and electron (T̂e ) respec-
tively. The other three terms describe the Coulomb interaction energy between the
electron - nucleus (V̂ne ), electron - electron (V̂ee ) and nucleus - nucleus (V̂nn ).
In practice, solving the Schrödinger equation using the full many-body Hamil-
tonian, as shown in Equation 2.17, is not easy and thus, approximations are usually
made. In general the Born-Oppenheimer approximation or adiabatic approximation
is used. By fixing the positions of the nuclei the ionic kinetic energy term vanishes.
Further, the ion-ion interaction term can be replaced by a constant energy term and
electron-ion interactions by an external potential. This reduces the Hamiltonian in
the following form,
~2 X 2 1 X e2
Ĥ = − Oi + + Vext + EII (2.18)
2me 2 | ~ri − ~rj |
i i6=j
with Vext and EII being the external potential and the ion-ion interaction energy,
respectively. These simplifications are possible due to the large ratio between the
masses of nuclei and electrons. Also, the motion of nuclei is very slow, and their
position quasi fixed, while electrons react almost instantaneously to changes in the
external potential. Within this scenario, it is valid to neglect the kinetic energy terms
for the nuclei.
Density Functional Theory: Hohenberg-Kohn Theorems and Kohn-Sham Equa-

tions
Hohenberg and Kohn used the simplified Hamiltonian from Equation 2.18 as the
starting point for the development of DFT. This theory, which is based on two theo-
rems, states that the properties of a system of interacting particles can be described
as a functional of the ground state density. DFT reduces the complex many body
problem of 3N variables (N electrons each having 3 spatial coordinates) to simply
3 variables, where they appear in the electron density functional. According to
Hohenberg-Kohn first theorem, the ground state density no (~r) of any system of inter-
acting particles is sufficient to uniquely, except for a constant, determine the external
potential Vext (~r) acting on the particles. The second theorem describes the possibil-
ity to define a universal functional E[n], valid for any external potential Vext (~r), for
determining the energy in terms of the density no (~r). Then, the exact ground state
density no (~r) is the one which minimizes this energy functional. Therefore, know-
ing the functional E[n] is sufficient to determine the ground state energy and density.
The ground state total energy functional can be written as,
Z
EHK [n] = T [n] + Eint [n] + d3 rVext (~r)n(~r) + EII (2.19)
2.6. Theoretical Methods 37
T [n] being a functional of the kinetic energies of the interacting particles, Eint [n]
describing the interaction energy of the electrons and EII the interaction energy of
the nuclei. Using a functional of the form of Equation 2.19 to obtain the ground
state density leaves one with the following problem: in principle, the ground state
density is sufficient to determine the properties of a material, but the DOS cannot
be directly extracted from the density. For physically solving this problem, in 1965
Kohn and Sham proposed a new approach, where they assumed that the ground
state properties of a system of interacting particles can be described by the ground
state of a non-interacting system. Based on this assumption, they introduced an
auxiliary system of independent electrons with its density given by,
N
X
n(~r) = | φKS r) |2
i (~ (2.20)
i=1
In Equation 2.20, φKS r) denote the single particle wave-functions of the non inter-
i (~
acting auxiliary system, the so-called Kohn-Sham orbitals [121].
2.6.3 Phase Accumulation Model
In PAM, the surface state energies are calculated assuming constructive interference
of free-electron waves reflected back and forth at the crystal-vacuum interface [122,
123]. This is represented in the following equation:
ΦB + ΦC = 2πn (2.21)
Here ΦC and ΦB are the crystal and barrier phase changes, respectively, and n is
an integer number with n ≥ 1 for image and n = 0 for surface states. The energy
dependences of both ΦC and ΦB are given by the following expressions:
r
E − EL
ΦC = 2arcsin (2.22)
EU − EL
and r !
3.4 eV
ΦB = π −1 (2.23)
EV − E
Where EL , EU , and EV are the lower and upper edges of the crystal bulk gap
along ΓL and the vacuum energy, respectively.
The PAM nicely predicts the band bottom energy of Au(111), Cu(111) and Ag(111)
Shockley states when using the initial values summarized in Table 2.2. The surface
state enegy is found when the ΦC + ΦB cuts the n = 0 phase (see Figure 2.15). The
PAM is used in this thesis to study the effect that a metallic overlayer causes on the
surface state fundamental energy, i.e. acting on the crystal phase ΦC [123]. Further
details on this procedure are included in Chapter 6.
TABLE 2.2: Noble metal (111) vacuum level and bulk band gap edges. Values have
been taken from ref. [124]
Au(111) Cu(111) Ag(111)

EV (eV) 5.3 4.94 4.74
EU (eV) 3.5 4.25 4
EL (eV) -1.05 -0.91 -0.3
a 5
5
4
4
FB
3
3
E-EF (eV)
2
2 F B+ F C
1
1
0
0
FC
-1
-1
Au(111)
-2
-2
b 5
5
-1 0 1 2 3 4 5
44 FB
3
F B+ F C
E-EF (eV)
22
1
0
0
FC
-1
-1
Cu(111)
-2
-2
c 5
5
-1 0 1 2 3 4 5
4
4
FB
3
3
F B+ F C
E-EF (eV)
2
2
1
1
FC
0
0
-1
-1
Ag(111)
-2
-2
-1
-1 0
0 1 2
2 3
3 4
4 5
phase
F IGURE 2.15: PAM solutions for the energy position of the Shockley state for
Au(111) (a), Cu(111) (b) and Ag(111) (c) respectively. The upper and lower edges
of the crystal bulk gap along ΓL are indicated in yellow, the crystal phase (ΦC ) is
indicated in green, the vacuum phase in pink (ΦB ) and their addition (ΦC +ΦB ) in
blue. The solution for the energetic position of the surface state comes at n = 0.
39
Chapter 3
Part II: An Introduction to

Assembly and Electronic Bands of
Nanoporous Networks
The discretization of the electronic bands into distinct energy levels in nanometer-
size solid systems when quantum confinement takes place is a fundamental result
of Quantum Mechanics. In nanostructures, confinement is usually related to the re-
flection of the electronic wave function at their surfaces or interfaces by combining
different materials, giving rise to diverse optical and transport properties of semi-
conductor Quantum Dots, Wires and Wells [125].
As indicated in the previous chapter, noble metal (111) surfaces host 2DEGs,
which are ideal systems to study low-dimensional electronic properties dealing with
electron scattering at the existing nanostructures [126–131]. It is known that if elec-
trons are confined to structures with size comparable to the de Broglie wavelength,
that is, if the lateral dimensions are in the nanometer range, quantum-size effects
emerge. In this regard, STM studies have demonstrated that Shockley electrons
can be confined parallel to the surface by artificially built nanoscale assemblies [95],
for example man-made quantum corrals [18, 19], metal atom and molecular res-
onators [132], vacancy and ad-atom islands [133, 134], vicinal surfaces [80] and
molecular nanoporous networks [15, 16]. In the following paragraphs a brief ac-
count for these systems is provided since they represent the foundations for the sec-
ond part of this thesis
Taking advantage of the cryogenic temperatures in combination with the abil-
ity to position single atoms with the tip of an STM, in the 90’s quantum corrals
were assembled by repositioning Fe atoms on a Cu(111) surface [18]. The quantum
mechanical interference patterns of 2DEG as standing waves could be imaged [Fig-
ure 3.1 (a) (left)]. Indeed, such nanostructures confine electrons into discrete eigen-
states, shown in the LDOS taken at the center of the quantum corral [Figure 3.1 (a)
(right)]. These features could be qualitatively understood by considering the text-
book particle in a box system and modelling the Fe atoms as an impenetrable bound-
ary, a hard-wall barrier [135]. Even though such a perfect quantum box should show
Chapter 3. Part II: An Introduction to Assembly and Electronic Bands of
40
Nanoporous Networks
eigenstates or resonances as delta functions (practically widthless), the experimen-

tal peak widths lead instead towards considering corrals as leaky barriers [19]. This
means that in quantum corrals only a small fraction of the incident amplitude is re-
flected by their walls [135]. This inelastic absorptive channel was proposed to be
related to coupling surface electrons with bulk states [19, 136]. However, it turned
out that it could also be explained by using elastic scattering theory (finite-height
potential barriers) [137, 138].
In a similar fashion, quantum resonators built atom by atom on a metal surface
can also trap surface state electrons [139, 140]. Figure 3.1 (c) shows such atomic
arrangements (Ag adatoms forming walls on Ag(111)), where surface electrons are
confined so the movement of the electrons is only constrained perpendicular to the
barriers [140]. Interestingly, the formation of strongly anisotropic diffusion chan-
nels were observed. Thus, additionally deposited single atom adsorbates appeared
guided by the surface state confinement features along such channels [139].
A greater mass-production of highly regular low-dimensional nanostructures
on surfaces was achieved with hexagonal islands [144] or hexagonal and triangular
vacancy islands formed on Ag(111) [Figure 3.1 (b) (left)]. The higher yield demanded
little effort compared to the atomic manipulation examples, since it was obtained by
self-assembly after a very soft sputtering of the surface [98, 133, 141, 145]. These
nanostructures were phenomenologically similar to quantum corrals, but the dis-
crete energy levels could be easily tuned with the cavity size (ranging from 30 to
100 Å wide). It was found that the standing wave patterns do not correspond to in-
dividual eigenstates, but reflect superpositions of several of them that are very close
in energy due to intrinsic and thermal broadening [98, 133]. The term lossy scatter-
ing (reflection coefficient < 1 at the barrier) [141] was proposed as the most plausible
cause, which reflects the fact that electrons might be scattered out of surface states
into bulk states, greatly contributing to the energy width of the confined states [146]
[Figure 3.1 (b) (right)].
The existence of highly ordered 1D nanostructures at the macroscopic level made
possible to study such confined electronic structures with averaging techniques such
as ARPES. Vicinal surfaces are among the simplest periodic surface formations gen-
erated by introducing a small angle miscut with respect to a high symmetry plane of
the crystal. This results in terraces periodically separated by monoatomic steps [103,
130, 142, 147]. 1D electron confinement effects are observed in single terraces [97,
148] [Figure 3.1 (d) (left)], although coherent coupling of the two-dimensional sur-
face state between terraces and the formation of Bloch wave states is detected when
probing vicinal surfaces [95] [Figure 3.1 (d) (right)]. These propagating superlat-
tice electronic bands can be understood and their physical nature well-captured by
applying the 1D-Kronig Penny (KP) model [103], where steps are considered as re-
pulsive square-shaped finite potential barriers U0 ×b (where U0 corresponds to the
height of the barrier and b to its width) [149].
Not only steps or adatoms scatter the surface electrons, but also molecules. They
41
Nanoporous Networks
a Quantum Corral b Vacancy Islands
c Metal atom Resonators d Vicinal Surfaces
Au(788)
c c
e Molecular Resonators f Molecular Nanoporous Networks
c c
F IGURE 3.1: A journey into confining nanostructures. (a) Circular quantum corral,
142.6 Å in diameter, built from 48 Fe atoms on the Cu(111) surface (adapted from
ref. [18]). Experimental and theoretical voltage dependence of dI/dV, with the STM
tip located at the center of the circular corral (adapted from ref. [136]). (b) STM
image of hexagonal vacancy islands on a Ag(111) surface and dI/dV spectrum at
the center of a hexagonal vacancy island fitted with a superposition of Lorentzians
(adapted from ref. [141]). (c) Differential conductance for the Ag resonator at two
different positions of the STM tip: near the center (red) and slightly outside the
resonator (blue) (adapted from ref. [140]). (d) (Left) Topographic image of a step
resonator, profile along the green line and color scale dI/dV spectra acquired along
the same line (adapted from ref. [142]); (Right) The Shockley state in Au(788),
showing step superlattice bands and gaps (adapted from ref. [103]). (e) (Left) Re-
active Lander (RL) molecule model and dI/dV maps highlighting standing wave
patterns (adapted from ref. [143]); (Right) Self-assembled regular methionine grat-
ing on Ag(111). Tunneling spectrum taken at Ag patches in between the molecular
chains demonstrating the 1D confinement of surface state electrons (adapted from
ref. [132]). (f) (Left) Purely organic kagome network and experimental dI/dV spec-
tra demonstrating a strong position-dependent modulation of the LDOS (adapted
from ref. [16]); (Right) Topograph of a Ph6Co nanopore on Ag(111) and the experi-
mental dI/dV spectra demonstrating a strong position dependent modulation of the
nearly constant DOS of the pristine surface (adapted from ref. [17]).
42
Nanoporous Networks
create standing wave patterns of surface state electrons by scattering them with the
aromatic π system of the molecular backbone [143] [Figure 3.1 (e) (left)]. Notably,
the concepts proposed by Supramolecular Chemistry [150] could be successfully
transferred to metal surfaces producing a myriad of self-assembled molecular ar-
rangements with different symmetries and periodicities capable of interacting with
the 2DEG [1]. For instance, one-dimensional arrays of molecular resonators were
among the first examples [Figure 3.1 (e) (right)] [132]. Here, surface state resonances
could be tuned by playing with the separation between the scattering molecular bar-
riers while the 2DEG would remain practically unaffected in the direction parallel to
the barriers. Adding up more complexity into the molecular design, by the proper
selection of ligands, molecular linker dimensions, metal coordination agents and
the templating effect of the surface, nanoporous networks were built. Such robust
structures were mainly stabilized by hydrogen-bonds [50] or metal-organic coordi-
nation [9, 16, 47, 151, 152], showing long-range order and a low amount of defects.
Interestingly, by the use of the local STM/STS technique, evidence of electron lo-
calization [Figure 3.1 (f) (left)] was shown inside the two-dimensional nanocavities
defined by the molecular barriers [15–17, 129, 153, 154]. Such confined states were
shown to be sensitive to the nanocavity size and shape [Figure 3.1 (f) (right)].
However, confinement of surface electrons within the pores of such molecular
nanoporous networks turned out to be quite leaky. Owing to the finite scattering bar-
rier represented by the molecular backbones (far from hard-wall potentials), some
degree of coupling was allowed between neighboring pores. According to the con-
cepts of band theory [155], this situation gives rise to the formation of distinct elec-
tronic bands, as confirmed by Lobo-Checa et al. [15] for the 3deh-DPDI metal-organic
nanoporous network on Cu(111) [Figure 3.2]. STS measurements inside the ∼ 2 nm
wide nanocavities [yellow cross in Figure 3.2 (a)] showed electronic confinement into
different localized states [Figure 3.2 (c)], in a similar fashion to previous examples in
Figure 3.1 (f). In addition, dI/dV maps at the resonance peak energy [Figure 3.2 (b)]
confirmed the localization of the 2DEG inside each nanocavity, resembling a quan-
tum dot. However, the large broadening of this confined state already suggested
certain interpore coupling [156]. ARPES measurements detected the presence of a
shallow dispersive, cosine-shape electronic band matching the STS energy [black
dashed line in Figure 3.2 (d) (top)]. In comparison to the pristine Shockley state, the
fundamental energy of the band is shifted towards the Fermi level due to confine-
ment effects and the bandwidth gives an indication of the degree of QD intercou-
pling. A faint onset attributed to the second confined state was then detected close
to the Fermi level separated by 120 ± 30 meV from the first confined state (appear-
ing as a plateau region in STS). The magnitude of the gap between states has been
attributed to the lateral scattering strength of the potential barriers (in this case the
DPDI molecular backbone) [157].
All the aforementioned examples summarized in this introduction build the
ground for the second part of this thesis. Here, we study the electronic properties
43
Nanoporous Networks
a b
c d
F IGURE 3.2: Electronic band structure of 3deh-DPDI nanoporous network gener-

ated on Cu(111). (a) STM image of the 3deh-DPDI nanoporous network. (b) Si-
multaneously recorded dI/dV map at the confined state resonance maximum (-0.22
V). (c) STS spectra obtained at 5 K on the clean surface state (red) and inside a
pore of the 3deh-DPDI metal-organic nanoporous network (black). (d) Energy dis-
persion curves of the pristine state (red dashed line) and the confined state (black
dashed line) measured along the ΓM high symmetry direction and visualized as
the second derivative of the photoemission intensity for two different molecular
coverages. Both EDCs at 0.52 ML and 0.73 ML were acquired at RT. All panels have
been adapted from ref. [15].
of different kinds of nanoporous networks grown on (111)-terminated noble metal

surfaces. For this family of nanostructures in particular, we try to find answers to
the following fundamental questions:
• Does the new electronic band formed in a nanoporous network [Figure 3.2] stem from
the Shockley state, as it is generally assumed?
Traditionally, confinement STS spectra inside single pores and the new electronic
bands steming from nanoporous networks have been assigned to the pristine sur-
face state [15, 129]. However, direct experimental evidence to this assignment is
desirable. In Chapter 4 we take advantage of the fact that the fundamental energy
of Shockley states is related to the position of the projected bulk band gaps, which
are highly sensitive to temperature-induced variations of the lattice constant, caus-
ing the supported Shockley state to shift in energy. Therefore in a similar fashion to
Paniago et al. [109], we study the temperature dependence of the shallow dispersive
band formed by the 3deh-DPDI metal-organic network on Cu(111) [15] and observe
44
Nanoporous Networks
that it follows the trend of the pristine Cu surface state. This evidences that both
bands have the same origin. This work has been published in ref. [158].
• Is it possible to tune or engineer the degree of QD intercoupling in a nanoporous

network by altering the separation among identical QDs?
In Chapter 5 we show that it is possible to tune the confinement and intercou-

pling properties of QDs by precisely engineering the barrier width on surface self-
assembled organic nanoporous networks. The selective barrier width is obtained by
substitution of a single atom (O vs S) in the haloaromatic compounds used. As a re-
sult two similar arrays of QDs are obtained that share the same QD dimensions but
different barrier width: in one case they are separated by just one molecule while
two molecules space the other. We observe that such barrier separation tunability
strongly affects the QD intercoupling degree as evidenced by the width variations
of the confined states in STS and bandwidth in ARPES. Both complementary tech-
niques are further supported by EBEM/EPWE and DFT calculations, which give
conclusive evidence of such interpore-coupling tunability. This work has been pub-
lished in ref. [54].
• How is the band structure modified when the pore size is linearly reduced in scalable
(homothetic) nanoporous networks?
Tunability of confined state energies has already been achieved in nanoporous net-
works by varying the pore dimensions, i.e. QD size, through its organic units and
observed in STS measurements on a single nanopore level [16, 17, 159]. However,
such effects on the electronic band structure have not yet been studied with ARPES.
In Chapter 6 we form two homothetic (scalable) Co-coordinated nanoporous net-
works on Au(111) where the QD size is tuned while keeping the interpore coupling
unaffected. Apart from confirming the pore size effects on the electronic band struc-
tures by opening zone boundary gaps at different energies, we find an unexpected
downward energy shift of the fundamental energy and a lightening of the effective
mass. These counterintuitive shifts are gradual and dependent on the QD size. As
the effect of the metal-organic overlayer upon the 2DEG onset has been traditionally
overlooked, we infer that metal-organic layer-substrate interactions in the form of
adatom hybridization effects and geometrical variations are at the origin of such un-
expected effects. DFT and PAM simulations provide insight into the nature of this
phenomenom which we infer to be quite general since such effects are also observed
on a similar Cu-coordinated nanoporous network.
• Can we parametrize and model in a rational way the scattering effect of the nanoporous
network upon the surface 2DEG?
45
Nanoporous Networks
Even though many theoretical models have been able to accurately explain con-
finement effects in quantum corrals [135, 136], vicinal surfaces [103], vacancy is-
lands [133, 141] and resonators [140], we find that the understanding of the confine-
ment properties and interaction effects between a nanoporous network and a 2DEG
is still elusive. To explain the scattering effects of these novel organic nanoporous
overlayers, Electron Boundary Element Method in combination with the Electron
Plane Wave Expansion has been used, which is key to understand the observed con-
finements and interpore couplings. It serves not only for modelling local confine-
ment effects inside each QD [16, 17, 129], but also to simulate the electronic band
structures arising from interdot coupling [54, 129, 160]. In Chapter 7 we perform
EBEM/EPWE simulations on the widely studied 3deh-DPDI nanoporous network.
We conclude that, if the appropriate scattering geometry is designed and ARPES
band structure and dI/dV curves at different positions of the metal-organic network
are used as experimental input, a repulsive scattering barrier geometry (even for the
metal linkers) is required, together with a 2DEG renormalization to account for the
overlayer-surface vertical interactions.
• Since nanoporous networks can serve as ideal templates for hosting additional adsor-
bates, is it possible to tune the electronic states of the QDs by the direct repositioning
of weakly interacting atomic adsorbates (such as Xe) in the nanocavities?
Based on the sensitivity of surface states to adsorbates and knowing that regular
nanoporous structures stand out as ideal host templates [153, 161–166], in Chap-
ter 8 the 3deh-DPDI metal-organic network on Cu(111) is used to adsorb weakly ph-
ysisorbed Xe inside the nanocavities. STS measurements show that confined states
shift in energy towards the Fermi level upon filling all the pores simultaneously with
Xe. This tendency, is qualitatively corroborated by ARPES measurements. In addi-
tion, by STM tip repositioning of individual Xe atoms inside the nanocavities, it is
also possible to study with STS the local modification of the confined state energy
and QD intercoupling for different Xe occupancy configurations. These Xe-filling
configurations are accurately simulated with EBEM/EPWE, delving into the intrica-
cies of confinement and interpore coupling effects. This work has been published in
ref. [53].
In the following, Chapters 4 to 8 will have a paper-like format. In particular,

each chapter will start with a brief introduction indicating the main motivation. Af-
ter this, the main results will be described and discussed supported by a number of
figures. The chapter will end with its particular conclusions and additional supple-
mentary information whenever not previously described in Chapter 2.
47
Chapter 4
Temperature Dependence of the

Partially Localized State in a 2D
Molecular Nanoporous Network
4.1 Introduction
Quantum states have always been related to a modification of the intrinsic surface
states of the substrate material. In this chapter we study the thermally induced en-
ergy shifts of the electronic bands formed in a metal-organic nanoporous array to
confirm it. By probing regions of the surface where molecular network patches of
3deh-DPDI coexist with pristine Cu regions, we observe by ARPES, that both, the
pristine Shockley state of Cu(111) and the partially localized states from the molec-
ular network, shift by the same amount upon sample temperature variation. These
findings provide clear evidence that partially localized states on the pores of a net-
work originate from the pristine Shockley state. This work has been published in
ref. [158].
4.2 Results and Discussion

We study by ARPES the temperature dependence of the electronic structure that
arises from the interaction of the Cu-coordinated triply dehydrogenated perylene
derivative DPDI (4,9-diaminoperylene quinone-3,10-diimine) porous network gen-
erated on Cu(111). For doing so, we use a large (15 × 15 mm) Cu(111) single crystal.
DPDI [Figure 4.1 (a)] [164, 167, 168] is sublimated from a Knudsen cell (∼ 515 K)
onto the clean Cu(111) surface held at room temperature (RT) in a gradient cover-
age geometry from 0 to 1 monolayer (ML), following the procedure described in
Section 2.3 [see Figure 4.1 (c)]. 1 ML is defined as the amount of molecules that sat-
urates the surface in the closed-packed assembly [168]. After the deposition of the
well-defined coverage gradient of DPDI, the sample is annealed to 550 K [159]. This
step is critical since DPDI undergoes a triple dehydrogenation process [3deh-DPDI
Chapter 4. Temperature Dependence of the Partially Localized State in a 2D
48
in Figure 4.1 (a)] and the necessary Cu surface adatoms become available for coor-
dination to generate a p(10 × 10) commensurate and homogeneous metal-organic
network. A model of this array is shown in Figure 4.1 (b), which is formed in the
lower coverage regions (up to 0.73 ML) [168, 169]. This nanoporous arrangement
leads not only to a characteristic LEED pattern [Figure 4.1 (d)] but also to a transfor-
mation of the Cu Shockley state into a shallow dispersive band visible in the ARPES
channel plate detector close to the Fermi energy [cf. Figure 4.2 (a)] [15].
The coverage gradient allows us to study different network density regions
within the same sample preparation [see Figure 4.1 (c)]. In essence, well-formed
molecular network islands progressively colonize the surface as the molecular cov-
erage is increased, as shown by the STM images of Figure 4.1 (e, f) (0.45 ML and
0.65 ML, respectively) [167]. Above the critical coverage of ∼ 0.73 ML, the porous
network gradually collapses into a close-packed assembly [168].
Unlike other nanoporous networks, which remain mobile at the ARPES mea-
surement temperatures (150 K) and can only be stabilized by saturating its coverage
(network fully covering the surface), 3deh-DPDI metal-organic network islands re-
main stable (even as small island formations) up to 720 K [168]. As a result, ARPES
probes the signal stemming from clean Cu regions and network patches simultane-
ously. This is illustrated in Figure 4.2 (a), where the formation of a n=1 partially local-
ized state band (n=1 PLS) is monitored with increasing coverage, while the parabolic
Cu Shockley state (Cu SS) becomes fainter and eventually vanishes. For the pristine
case, the fundamental energy of the 2DEG at 150 K is at -0.4 eV and is character-
ized by an effective mass (m∗ /m0 ) of ∼ 0.4 [100]. As the 3deh-DPDI nanoporous
network islands start to populate the surface (from 0.25 ML to 0.5 ML) the Cu SS is
still visible but becomes broader and shifts slightly towards the Fermi level (∼ 40
meV). This happens because the 2DEG scatters with island edges that progressively
become closer to each other. At the same time, a second contribution starts to build
up close to -0.25 eV. This intensity corresponds to the n=1 PLS band which originates
from the coupling of the n=1 confined state of each nanopore with the neighboring
ones. When the islands are small (0.25 ML), the number of nanopores or QDs is not
sufficient to develop a coherent and well-defined electronic band. This shallow dis-
persive band with its band bottom energy at -0.25 eV, bandwidth of ∼ 90 meV and
m∗ /m0 ∼ 0.56 is ultimately best observed when the network practically covers the
surface (0.7 ML) [15, 53]. Note that at this coverage there is no sign of the pristine Cu
SS. This transition can also be nicely followed by looking at the Energy Distribution
Curves at the Γ point [Figure 4.2 (b)]. In the same way, the intensity of the pristine
Cu SS (in grey) decreases with increasing molecular coverage and the n=1 PLS band
evolves until it is fully formed (in red). The relative intensities do not exhibit a one
to one ratio. In particular, the peak area of the n=1 PLS has been reported to be ∼ 30
% with respect to the Cu SS for the optimal coverage of 0.73 ML [15]. One possible
reason is the appreciable defect concentration and the relative small average domain
4.2. Results and Discussion 49
a b
550 K
-3H2
DPDI 3deh-DPDI
c d
p(10x10)
e f
F IGURE 4.1: A variable density of nanoporous network realized through a coverage

gradient geometry. (a) Thermally induced triple-dehydrogenation of DPDI acts as
an exoligand in the formation of the metal-organic network on Cu(111) [169]. (b)
Structural model for the Cu-coordinated 3deh-DPDI nanoporous network forming
a p(10 x 10) superstructure (adapted from ref. [167]). (c) Schematic view of the
DPDI deposition gradient performed on Cu(111). (d) LEED pattern obtained at 52
eV showing the Cu(111)-p(10x10) DPDI network lattice (adapted from ref. [168]. (e)
and (f) STM images for 0.45 ML (left) (4 K, -0.8 V, 5 pA) and 0.65 ML (right) (297 K,
2 V, 6 pA) network coverage. The network structure is maintained throughout the
coverage gradient up to ∼ 0.73 ML.
50
a Pristine Cu(111) SS RD DPDI network RD 0.25ML 3deh-DPDI network

DPDI network RD 0.5ML coverage
DPDI network RD 0.7ML
0.0
0.0 Clean 0.0 0.25 ML 0.0 0.50 ML 0.0 0.70 ML
gap n=1 PLS

F (eV) -0.1
-0.1 -0.1 -0.1 -0.1
E-EF (eV)
E-EF (eV)
E-EF (eV)
E-EF (eV)
-0.2
-0.2 -0.2 -0.2 -0.2
E-E
-0.3
-0.3 -0.3 -0.3 -0.3
-0.4
-0.4 -0.4 -0.4 -0.4
 M  M M  M M  M
-0.5
-0.5 -0.5 -0.5 -0.5
-0.2
-0.2 -0.1
-0.1 0.0 0.1
0.1 0.2 -0.2 -0.1
0.2 -0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1
0.2 -0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1
0.2 -0.2 -0.1 0.0 0.1 0.2
0.2
-1 -1 -1-1) -1-1) -1 -1
kkx II(Å(Å) ) kkx II(Å
(Å ) kkx II(Å
(Å ) kk II (Å )
x (Å )
b Coverage dependent EDCs
EDC at  formation of the network
at upon c d Fermi wave-vector (kF)
Cu SS
0.0
0.0 0.25 ML DPDI network DPDI (PLS) Cu SS
Cu SS
0.5 ML DPDI network
0.7 ML DPDI network
-0.25
-0.25 0.25 ML DPDI network
0.5 ML DPDI network
Int. (arb. units)

-0.1
-0.1 0.7 ML DPDI network
E-EF (eV)
(eV) -0.30
-0.30
(eV)
-0.2
-0.2
E-EFF (eV)
n=1 PLS
E-E
-0.3
-0.3 -0.35
-0.35
E-E
Cu SS
-0.4
-0.4
-0.40
-0.40
-0.5
-0.5
0.0
0.0 0.2
0.2 0.4
0.4 0.6
0.6 -0.2 -0.1
-0.2 -0.1 0.00.0 0.10.1 0.20.2
-0.6
-0.6
DPDICoverage
Coverage (ML)
(ML) kII (Å-1)
Int. (arb. units) DPDI
Intensity (arb. units)
e
a2*
a2 M
 K
a1 a1*
Nanoporous network (Real Space) Reciprocal Space
F IGURE 4.2: Coverage dependent network island induced electronic band forma-
tion measured with ARPES. (a) Coverage gradient formation of the partially local-
ized state band. Between 0.25 ML and 0.6 ML, both Cu SS and PLS bands coexist.
(b) Energy distribution cuts at Γ highlight the intensity variations for both Cu SS
and PLS bands as the coverage of the network islands is increased until fully cov-
ering the surface. (c) Fundamental energy variations of Cu SS and PLS bands with
coverage. Scattering effects produce a slight shift of the Cu SS when the network
islands become predominant (0.5 ML). The fundamental energy of the PLS is also
better defined when the network practically covers the surface (0.7 ML). (d) Fermi
wave-vector (kF ) spectra with coverage. Slight shrinking of the wave-vector is ob-
served due to the formation of a hexagonal Surface Brillouin Zone produced by the
network. (e) Representation of the surface unit cell (real space) and the Brillouin
zone (reciprocal space) of the 3deh-DPDI network.
size that the networks exhibit compared to an atomically perfect Cu surface. In ad-
dition, lossy scattering effects with the network barrier may also be responsible for
such intensity decrease and increased background [141, 146]. Moreover, the energy
position of both Cu SS and the n=1 PLS is slightly sensitive to the coverage of the
network islands on the surface [Figure 4.2 (c)]. As the density of islands becomes
more predominant (0.5 ML), the Cu SS scatters with the network edges and presents
a minor ∼ 40 meV shift towards the Fermi level. Similarly, the fundamental energy
of the n=1 PLS is better defined when the network fully covers the surface. Fig-
ure 4.2 (d) captures the evolution of the Fermi wave-vector (kF ) with coverage. A
tendency towards smaller values can be observed. Such kF shrinking evidences a
distortion of the Fermi surface into a new hexagonal Surface Brillouin Zone. This
happens because the highly ordered hexagonal nanoporous network sets a new and
periodic scattering landscape for the surface 2DEG and new zone boundaries and
symmetry points (M and K points) appear in reciprocal space. Figure 4.2 (e) shows
the conversion from real space unit cell of 3deh-DPDI network to the first Brillouin
zone in reciprocal space. The p(10 × 10) unit cell of 3deh-DPDI network is defined
with unit-cell vectors a~1 and a~2 (|a~1 | = |a~2 | ≡ a ≈ 2.55 nm). The unit cell in reciprocal
space with the first Brillouin zone (black hexagon) and the symmetry Γ, M and K
points are defined by a~∗1 and a~∗2 . The length of the unit-cell vectors amount to |a~∗1 |
√ √
= |a~∗2 | = 4π/ 3a ≈ 0.283 Å−1 . The distances between the critical points are 2π/ 3a
≈ 0.142 Å−1 (ΓM) and 4π/3a ≈ 0.164 Å−1 (ΓK). This matches Figure 4.2 (a) where
the ARPES spectral functions along ΓM show the shallow dispersive band that fol-
lows a cosine shape with an opening of a gap at each M point. This energy gap
corresponds to the separation between the n=1 and n=2 PLS bands.
Once the formation of the network islands and its coverage dependent n=1 PLS
band have been well characterized and introduced, we proceed to study its energy
dependence with temperature. ARPES spectral functions close to the Fermi level
were systematically acquired as a function of sample temperature (from 315 K to 136
K) for two coverage regions: the first, where the molecular network completely cov-
ered the surface (∼ 0.75 ML) and the second, where it coexisted with molecule-free
areas (∼ 0.50 ML) [close to the coverage of Figure 4.1 (e, f)]. For each temperature,
a normal emission ARPES spectral function is captured in the channelplate detector
for each coverage region. This is done by sequentially moving the z-axis of the ma-
nipulator. Then the sample temperature is fractionally increased (∆T ≈ 15 K) and
the measurement is repeated.
Figure 4.3 (a) shows the ARPES spectral function corresponding to the temper-
ature extremes (315 K and 136 K) for ∼ 0.50 ML DPDI. The intensity distributions
I(E − EF , k) reflect a coexistence of the Cu SS band originating from the molecule-
free areas and the n=1 PLS band arising from the coupling between pores [15]. While
the former has an upward parabolic dispersion [85], the latter exhibits a lower bind-
ing energy at Γ and a limited bandwidth together with an apparent increase of the
effective mass [53].
As the temperature of the sample decreases from 315 K to 136 K, both electronic
bands of Figure 4.3 (a) become better defined and gain amplitude [85, 170]. As re-
ported by Paniago et al. [109], the bottom of the Cu SS band shifts towards higher
binding energy [171], but is difficult to discern the behavior of the n=1 PLS band
from the raw data. To follow its evolution, the energy distribution curves of the
ARPES spectral funcitons at Γ (k = 0) are represented in Figure 4.3 (b) for both end
52
a 0.50 ML DPDI b
0.0 0.0
-0.1 -0.1
E-EF (eV)
E-EF (eV)
-0.2 n=1 -0.2
-0.3 PLS -0.3
SS
-0.4 -0.4
-0.5 -0.5
-0.2 -0.1 0.0 0.1 0.2-0.2 -0.1 0.0 0.1 0.2 Int. (arb. Units)
kII (Å-1) kII (Å-1)
c d
0.75 ML DPDI
0.0 0.0
-0.1 -0.1
E-EF (eV)
E-EF (eV)
-0.2 n=1 -0.2
PLS
-0.3 -0.3
-0.4 -0.4
-0.5 -0.5
-0.2 -0.1 0.0 0.1 0.2-0.2 -0.1 0.0 0.1 0.2 Int. (arb. Units)
F IGURE 4.3: Temperature dependence of the PLS band and Cu SS band at the two
extreme temperatures studied. In contrast to ref. [15], raw ARPES data is presented
here to directly compare the resulting EDCs. (a) ARPES experimental spectral func-
tion acquired at 315 K and 136 K for the coverage of 0.50 ML, in which regions of the
network and molecule-free areas coexist. (b) Smoothened, normal emission (kk = 0)
EDCs for both temperatures in (a). The horizontal lines mark the energy positions
of the Cu Shockley state (SS) and the partially localized state (PLS) band minima
obtained from a two component Lorentzian fit. (c) ARPES experimental spectral
function acquired at 315 K and 136 K for the coverage of 0.75 ML featuring the sur-
face completely covered by the network. (d) Smoothened, normal emission (kk =
0) EDCs for both temperatures in (c) with horizontal lines indicating the energy
positions of the PLS band minimum.
temperatures, red being the higher temperature spectrum and blue the lower one.
We observe that the well-known Shockley state band minimum (peak closest to Ei
≈ −0.4 eV) is shifted away from the Fermi energy by ∼ 30 meV as the sample is
cooled down. The same trend is observed on the shoulder feature located in the
proximity of Ei ≈ −0.25 eV. This peak corresponds to the n=1 PLS and is in part
masked by its proximity to the Cu SS. For a better visualization of the thermal en-
ergy shift of this state, it is best to study the higher coverage regions that lack the
molecule-free areas, such that the Shockley state is quenched. This case is shown in
Figure 4.3 (c), corresponding to 0.75 ML and exhibiting only bands attributable to
the porous network. Once more, a clear energy shift for the band minimum and a
sharpening of the features are observed as the temperature is lowered. The compar-
ison between the temperature dependent EDCs, represented in Figure 4.3 (d), shows
that the n=1 PLS also shifts around 30 meV to higher binding energies as the sample
is cooled down.
a b c
-0.22
0.75 ML 136 K 136 K 0.50 ML
-0.24
-0.26
Int. (arb. units)
Int. (arb. units)

-0.28
E-EF (eV)
-0.30
-0.32
-0.34
315 K
315 K -0.36
-0.38
-0.5 -0.4 -0.3-0.2 -0.1 0.0 -0.5 -0.4 -0.3-0.2-0.1 0.0
E-EF (eV) E-EF (eV) 150 200 250 300
Temperature (K)
F IGURE 4.4: Full range temperature dependence of the PLS and Cu SS. Normal
emission EDC temperature waterfalls for 0.75 ML (a) and 0.50 ML (b) coverages.
A clear trend towards higher binding energy and a sharpening of the features is
observed as the temperature is lowered. (c) Temperature variation of the energy
position of the PLS and SS band minima as extracted from the fit for each spectral
line represented in the waterfall. The fit was carried out using one (0.75 ML) or
two (0.50 ML) Lorentzian components with a linear background convoluted with
a Fermi function. The dashed lines correspond to variation rates with a common
slope for the three components, all of which agree with the Shockley state variation
reported in ref. [109].
The nature of the energy shift observed for the n=1 PLS is still undefined at this
stage and so we must follow the temperature transition at intermediate points. In
this regard, the Shockley state of Cu(111) has been shown to linearly shift with tem-
perature, with a slope of (1.8 ± 0.1) x 10−4 eV/K [109]. This linear behavior is also
present for Ag(111) and Au(111) Shockley states, each one possessing a particular
rate. For the nanoporous network, EDC waterfall plots at normal emission (Γ) for
the studied temperature range are shown in Figure 4.4 (a, b) corresponding to full
network coverage (0.75 ML) and to coexisting regions of molecular network and
molecule-free areas (0.50 ML). We observe that the energy position of the band min-
imum of the n=1 PLS as well as the Cu SS shift in a progressive way as the temper-
ature is varied. There are no abrupt energy changes of the features, which sharpen
up as the temperature is lowered.
A quantitative analysis of the temperature dependence of these states is achieved
by Lorentzian fitting with a linear background convoluted with a Fermi function
from the spectra in Figure 4.4 (a, b) (see Section 4.4.3). Figure 4.4 (c) summarizes
the results from such analysis and shows the variation of the initial state energy
with sample temperature. As expected, the Cu SS band (yellow dots) follows the
red dashed line which accounts for a rate of (1.8 ± 0.1) x 10−4 eV/K, in agreement
with Paniago et al. [109], resulting in an increase of 32 ± 2 meV in binding energy
54
for the investigated temperature range. The other two sets of magenta points in Fig-
ure 4.4 (c) correspond to the n=1 PLS for the 0.50 ML (full circles) and 0.75 ML (open
circles) and they differ in energy by ∼ 30 meV. We attribute this offset to the cov-
erage effect in its fundamental energy already discussed in Figure 4.2. For 0.5 ML
the n=1 PLS is shifted by 33 ± 5 meV within the probed temperature range, which
agrees with the variation of the Cu SS, while a slightly smaller overall shift of 26
± 5 meV is measured for 0.75 ML. This is still consistent with the Cu SS behavior
within the experimental error. As the temperature dependence is unique for each of
the electronic states in Cu(111) [171, 172], the demonstration that the n=1 PLS band
associated with the nanoporous network exhibits an identical temperature variation
of its energy to the Cu SS, supports that both have the same electronic nature. The
energy shift is due to temperature induced variations of the lattice constant and this
affects bulk and surface differently. According to Knapp et al. [171], the magnitude
of the energy shift for the surface band, which is 50%- 75% larger than typical shifts
for bulk bands, may reflect a thermal distortion of the lattice at the surface. Further-
more, each bulk band exhibits its own temperature dependence, i.e. the Λ3 and Λ1
d-bands as well as the sp-band all have different temperature coefficients, most likely
due to their different orbital character. At the same time, all these temperature trends
are different from the Cu SS. Moreover, according to Knoesel et al. [172], the image
states of Cu(111) show an opposite rate with respect to the Shockley state, since they
are related to the bulk band projection at the top of the gap (L1 ). All this demon-
strates that the Cu SS has a unique temperature dependence, different from the rest
of electronic states. The fact that the n=1 PLS follows precisely the same trend of the
SS evidences their common origin. For some closed packed molecular assemblies,
such as CoPc on Au(111), similar trends for the Shockley derived interface state have
been observed. However, in this case, it is also argued that on organic/metal inter-
faces the Shockley state energy shift with temperature not only originates from the
sum of the changes in the Au substrate lattice constant but also slight temperature
dependent interfacial bonding distance variations have to be considered [173].
4.3 Conclusions
Upon the 3deh-DPDI nanoporous network formation, the surface state electrons are
subjected to a 2D finite periodic potential imposed by the network structure. The
network thus alters the Shockley state giving rise to the scattering phenomena ca-
pable of generating partially localized states and new band structures with differ-
ent dispersion relations [15, 129]. In this experimental study we validate the origin
and nature of such PLS observed in a 2D metal-organic nanoporous network self-
assembled on the Cu(111) surface. By studying the temperature dependence of two
different coverage regimes of the network, we corroborate that the n=1 PLS existing
4.4. Supplementary Information for This Chapter 55
in the pores of the network originates from the Cu(111) Shockley state. The ener-
gies of both states display the same temperature dependence, directly relating the
n=1 PLS to the Cu SS observed on the pristine metal surface. This experimental
demonstration supports that the surface state electrons are affected by the periodic
molecular potential of the network. Therefore, as it will be shown in the next chap-
ters, semiempirical models such as the Electron Boundary Elements Method [116] in
combination with the Electron Plane Wave Expansion [115] can be confidently used
to derive the observed electronic states and to model the surface potential landscape
created by molecular nanoporous networks [129].
4.4 Supplementary Information for This Chapter

4.4.1 ARPES Measurements
ARPES measurements were performed in UHV conditions (base pressure of 1 × 10−10

mbar) with a lab-based experimental setup equipped with a display-type hemi-
spherical electron analyzer (SPECS Phoibos 150), an energy/angle resolution of 40
meV/0.1◦ and a monochromatized Helium I (hν = 21.2 eV) source. Variable temper-
ature measurements were performed using a closed-cycle He cryogenic manipulator
with a controlled temperature range between 320 K and 30 K.
4.4.2 STM Measurements
STM measurements performed by the group of Prof. T. A. Jung from the University
of Basel (Switzerland). They were carried out at both low temperature (Omicron
Nanotechnology GmbH with Nanonis SPM control system) and room temperature
(home built STM). The bias voltages given here refer to a grounded tip. The mea-
surements were performed with Pt-Ir tips (90% Pt, 10% Ir) prepared by mechanical
cutting and followed by in-situ sputtering with Ar+ ions and controlled indentation
in the bare Cu(111) substrate. STM data were acquired in constant current mode and
were processed with the WSxM software [174].
56
4.4.3 Method Used for the ARPES Spectral Deconvolution
• For the quantitative analysis performed in Figure 4.4, we performed a Lorentzian

fitting with a linear background convoluted with a Fermi function. Figure 4.5
shows the fittings performed for Cu(111) surface state (one Lorentzian compo-
nent), 0.5 ML DPDI network (two Lorentzian components) and 0.75 ML DPDI
network (one Lorentzian component) coverages respectively. These spectra
were taken at 150 K. For 0.5 ML DPDI case, the Cu SS appears slightly shifted
to lower BEs, while the opposite trend is observed for the n=1 PLS.
Cu(111) 0.5 ML DPDI 0.75 ML DPDI

-0.6
-0.6 -0.4
-0.4 -0.2
-0.2 -0.0 -0.6
0.0 -0.6 -0.4
-0.4 -0.2
-0.2 0.0 -0.6
-0.0 -0.6 -0.4
-0.4 -0.2
-0.2 -0.0
0.0
E-EF (eV) E-EF (eV) E-EF (eV)
F IGURE 4.5: Cu SS and DPDI n=1 PLS spectra fittings. From left to right: Cu SS,
0.5 ML DPDI and 0.75 ML DPDI spectra fittings using a Lorentzian component for
each peak and a linear background convoluted with a Fermi function (not shown).
57
Chapter 5
Precise Engineering of Quantum

Dot Array Coupling Through Their
Barrier Widths
5.1 Introduction
Quantum dots are known for confining electrons within their structure. When-
ever they periodically aggregate into arrays and cooperative interactions arise, novel
quantum properties suitable for technological applications show up. Control over
the potential barrier existing between neighboring QDs is essential to alter their mu-
tual crosstalk or coupling. In this chapter we show that precise engineering of the
barrier width can be experimentally achieved by a single atom substitution (sulphur
vs oxygen) in a haloaromatic compound, which in turn tunes the degree of QD inter-
coupling. We achieved this by generating self-assembled, halogen bond stablilized,
long-range ordered nanoporous networks that confine the surface 2DEG. Indeed,
these extended QD arrays form up on bulk silver (Ag(111)) and thin silver films on
gold (3 ML Ag/Au(111)) alike, maintaining their overall structure, confinement and
interdot coupling properties. This work has been published in ref. [54]

Our concept to control the interpore barrier width while maintaining the pore size
[Figure 5.1 (a, b)] is based on the halogen bond versatility to generate artificial nanos-
tructures [12, 175, 176]. The two hexagonal nanoporous networks show single-
molecular and double-molecular separation between practically identical pores, which
extend into long-range ordered network films. This has been achieved by just a
single atom substitution (sulphur vs oxygen) in the haloaromatic precursors used,
reminiscent of a butterfly effect. We employed two molecules [Figure 5.1 (c, d)]: 3,9-
dibromodinaphtho [2,3-b:2’,3’-d]thiophene (Br-DNT) [177] and 3,9-dibromodinaphtho
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
58
Barrier Widths
a Single Wall c e g
b d f h
Double Wall
F IGURE 5.1: QD arrays generated by single-wall (SW) and double-wall (DW)

nanoporous networks that confine the surface 2DEG. (a, b) Schematic represen-
tations of the concept behind SW and DW networks. (c, d) Chemical structures
and electrostatic potential maps of Br-DNT and Br-DNF. (e, f) Large scale STM to-
pographies for the SW network generated with Br-DNT and the DW network with
Br-DNF. Insets show close-views of each network. (g, h) High-resolution atomic
force microscopy (AFM) images of the SW network and DW network. Measure-
ment parameters: tunneling current I= 5 pA, bias voltage V= 200 mV (e, f); V= 0
mV, oscillation amplitude A= 60 pm (g, h).
[2,3-b:2’,3’-d]furan (Br-DNF). Intermolecular electrostatic attraction between the pos-

itive cap (σ hole) and the electron-rich regions (negative belt of bromine atoms in Fig-
ure 5.1 (c) or oxygen atom of the furan core in Figure 5.1 (d)) are responsible for the
condensation into two distinct molecular networks [178, 179] (see Figure 5.2). Our
STM images (acquired by our collaborator Dr. Shigeki Kawai from MANA/NIMS,
Japan) [Figure 5.1 (e, f)] show extended (over 100 nm) nanoporous network forma-
tions on Ag(111) with small amount of defects. Such hexagonal nanoporous assem-
blies, are the most energetically favorable and stable below RT. Above this tempera-
ture, thermal fluctuations overcome the halogen bond strength (∼ 33 kJmol−1 ) [178]
and the nanoporous structure is destabilized while the surface catalyzed Ullmann
coupling reaction starts to dominate. This leads to the formation of covalent one-
dimensional polymers [180] (see Section 5.4.2).
Detailed structures are derived from atomic force microscopy (AFM) with a CO
functionalized tip [112, 114]. The pores of the Br-DNT network are separated by
single molecules [Figure 5.1 (g)], whereas two molecules are required in the Br-DNF
network [Figure 5.1 (h)]. Note that bright spots at the nodal sites of the latter are
CO molecules adsorbed for tip functionalization (see Section 5.4.2). While the con-
densation of Br-DNT happens solely through trigonal halogen bonding [176, 181]
[Figure 5.2 (a)], the furan group presence in Br-DNF introduces higher interaction
complexity due to the electronegativity of oxygen and it also tends to participate
in the assembly [Figure 5.2 (b)]. The O· · · Br-C bonding (oxygen is faintly observed
a b
SW DW
F IGURE 5.2: Bonding motifs that stabilize the SW and DW formations. (a) The
formation of SW network occurs via a triple halogen bond. Br atoms form a triangle
that stabilizes the whole structure. (b) The formation of DW network is induced by
the furan group presence in Br-DNF that introduces higher interaction complexity
due to the electronegativity of oxygen.
in AFM) is apparently stronger (based on the electric potentials of Figure 5.1 (c, d))
than the Br· · · Br-C homo-halogen bond, leading to a shorter bond. Both hexago-
nal arrays are commensurate with the pristine Ag(111) surface, according to density
functional theory (DFT) calculations performed by Dr. Ali Sadeghi from Shahid Be-
heshti University (Iran) (see Section 5.4.4). Indeed, the Br-DNF network interpore
distance is larger (by 14%) than that of Br-DNT, as a consequence of the molecular
pairing. Note, however, that the enclosed pore areas remain identical to both assem-
blies since the perimeter of each pore is constructed by 6 molecules with the same
dimensions in a hexagonal conformation. Essentially, we confirm structurally that
these arrays are extended model systems to investigate the 2DEG confinement and
interpore coupling, as they feature identical QDs separated by different wall widths.
For clarity, we will hereafter refer to the Br-DNT and Br-DNF networks as Single-
Wall (SW) and Double-Wall (DW) networks, respectively.
Next we consider the QD local electronic structure using STS, as this technique
measures the LDOS of the surface at a particular bias voltage [19]. Figure 5.3 (a)
shows conductance (dI/dV) spectra acquired at the center of the pores of SW and DW
networks and referenced to the clean Ag(111) substrate. The pristine Ag(111) surface
state reveals its presence by a strong variation in the conductance at -65 meV, corre-
sponding to the surface state band edge (black curve). Indeed, clear energy shifts of
the pristine surface state onset (-65 meV) are induced by the QD confinement, peak-
ing at 72 meV for the SW network and 45 meV for the DW network. These values
are unexpectedly inverted since a stronger confinement (higher peak energy) is an-
ticipated for the wider barrier (DW) [19]. It could be argued that it originates from
a difference in the molecule-substrate interaction affecting the potential amplitude,
where Br-DNT would have a larger value than Br-DNF [182]. However, not only a
larger interaction is expected for Br-DNF due to the extra oxygen electronegativity
but also the measured full width at half maximum of the confined state for the DW
60
Barrier Widths
dI/dV (arb. units)
-100 0 100 200

Sample bias (mV)
b SW V = 72 mV c DW V = 45 mV
F IGURE 5.3: Local electronic structure of the SW and DW networks. (a) Conduc-
tance (dI/dV) spectra on the pristine Ag(111) (black), on the pore centers of the SW
network (red) and DW network (green). (b, c) dI/dV maps on the SW and DW net-
works acquired on the peak maxima in (a), enlarged in the corresponding insets.
Measurement parameters: I= 10 pA, V= 72mV, modulation voltage Vac = 10 mV,
oscillation frequency fac = 513 Hz (b); I= 10 pA, V= 45 mV, Vac = 10 mV, fac = 515 Hz
(c).
network (32 meV) is narrower that that for the SW network (45 meV), suggesting
a stronger confinement [19]. The conductance maps at those particular peak ener-
gies show high conductance within the pores, exhibiting the dome shape of the n=1
confined state [Figure 5.3 (b, c)] [17, 133, 141]. Note that some pores rarely present
brighter contrast (indicated by blue arrows), which we assign to defects (see Sec-
tion 5.4.2). Moreover, the AFM images show both molecules lying flat on the surface
[Figure 5.1 (g, h)], suggesting a similar and weakly interacting regime [128].
Since these networks are analogous to coupled QDs [15], due to the leaky molec-
ular potential barriers, in principle, one should expect the formation of shallow dis-
persive bands in photoemission. Therefore, we should perform ARPES measure-
ments in both these networks. However, this technique endures the intrinsic restric-
tion of being sensitive only to the occupied electronic structure, so the confined states
from the SW and DW networks on Ag(111) become undetectable [100, 183, 184]. In
order to push both confined state peaks into the occupied region, we grew SW and
a Au(111) 1 ML Ag/Au(111) 2 ML Ag/Au(111) 3 ML Ag/Au(111)

0.0
0.0 0.0 0.0 0.0
-0.1
-0.1 -0.1 -0.1 -0.1
E-EF (eV)
-0.2
-0.2 -0.2 -0.2 -0.2
-0.3
-0.3 -0.3 -0.3 -0.3
-0.4
-0.4 -0.4 -0.4 -0.4
-0.5 -0.5
-0.5
-0.5 -0.5
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2-0.2 -0.1
-0.2 -0.1 0.0
0.0 0.1
0.1 0.2
0.2-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2 -0.2 -0.1
0.2-0.2 -0.1 0.0
0.0 0.1
0.1 0.2
0.2
kII (Å-1) kII (Å-1) kII (Å-1) kII (Å-1)

b c -0.15
-0.15 0.40
0.40
Au(111)
0.0
0.0
1ML Ag/Au(111)
-0.20
-0.20 0.38
0.38
2ML Ag/Au(111)
3ML Ag/Au(111)
-0.1
-0.1 -0.25
-0.25 0.36
0.36
E-EF (eV)
0.34
0.34
E-EF (eV)
m*/me
-0.30
-0.30
-0.2
-0.2
0.32
0.32
-0.35
-0.35
-0.3
-0.3 0.30
0.30
-0.40
-0.40
Band bottom @ 
0.28
0.28
-0.4
-0.4 -0.45
-0.45
Effective Mass 0.26
0.26
-0.50
-0.50
-0.5
-0.5
Au(111) 1 ML
1ML Ag/Au(111) 2 ML
2ML Ag/Au(111) 3 ML
3ML Ag/Au(111)
Au(111)
Ag/Au(111) Ag/Au(111) Ag/Au(111)
F IGURE 5.4: Modification of the Au(111) 2DEG with Ag film thickness. (a) ARPES
maps along ΓM measured directly on Au(111), after 1 ML, 2 ML and 3 ML of Ag
deposition (Tsample = 130 K) and subsequent annealing to 450 K. According to the
EDCs at Γ, the surface state shifts its energy towards the Fermi energy in defined
steps (b) and increases its effective mass (c). Values shown in (c) have been adapted
from ref. [157].
DW onto a 3 monolayer (ML) Ag thin film on Au(111), which shifts the Ag(111)
surface state by -100 meV while preserving its 2DEG character (Figure 5.4). The for-
mation of Ag thin films on Au(111) can be monitored by following the evolution of
the Shockley state (its fundamental energy and sharpness) with Ag coverage. This
transition from the Au(111) Shockley state up to the Shockley state corresponding to
3 ML Ag/Au(111) is shown in Figure 5.4 (a) [157]. Note that the spin-orbit splitting
of Au(111) is not observable in this channelplate image because of the low resolu-
tion conditions used in exchange of a faster acquisition. From 1 ML Ag/Au(111) to
3 ML Ag/Au(111) the surface state shifts towards the Fermi level and its effective
mass is increased. The energy shift can be observed in the EDCs at the Γ point [Fig-
ure 5.4 (b)]. Here, the peak width can also serve as a direct indication of the quality
of the film formed. In Figure 5.4 (c), the values of the fundamental binding en-
ergy and the effective mass are summarized for each complete monolayer (adapted
from ref. [157]). Note that the surface state will become pure Ag like, when 10 ML
Ag/Au(111) is reached (E= -65 meV, m∗ = 0.4 m0 ). This means that for the 3 ML
Ag/Au(111) case, even though the 2DEG is quite similar, there is still 100 meV dif-
ference between both fundamental energies.
These networks cannot be formed for 1 ML and 2 ML Ag/Au(111) films (see Sec-
tion 5.4.1). However, when using the 3 ML Ag film the ARPES data in Figure 5.5
exhibit distinct network bands that are characteristic of coupled QD arrays [15, 53].
If we compare them with the pristine parabolic surface state, they shift their minima
(band bottom at Γ) to higher energy due to confinement in the pores and deviate
62
Barrier Widths
3ML Ag/Au(111) SW DW
Raw Data a b c
0.0
E-EF (eV)
-0.1
-0.2
G G M G M G G M G M G
-0.3
2nd derivative d e f
0.0
E-EF (eV)
-0.1
-0.2
G G M G M G G M G M G
-0.3
-0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
kII (Å-1) kII (Å-1) kII (Å-1)
F IGURE 5.5: 2DEG modification induced by the SW and DW network potential

barriers along ΓM. (a-c) ARPES map along ΓM, obtained on the 3ML Ag/Au(111)
as well as on the SW and DW networks. The EDCs close to Γ were fitted using a
Lorentzian component and a linear background convoluted with a Fermi function
(black dahsed lines). (d-f) Second derivative maps of the above raw data for an
improved visualization of the second SBZ. The white dashed lines correspond to
the EBEM/EPWE calculated electronic bands generated by altering the 2DEG with
the molecular surface potentials.
rapidly from the initial quasi free-electron like parabolic dispersion (as evidenced
by the weak side replicas away from kk '±0.1 Å−1 ). Such replicas are generated by
the scattering potential at the molecular barriers and always appear with weaker
intensity than the main band from the first Surface Brillouin Zone [157]. Note that,
these subtle effects can only be observed whenever a single domain, long-range or-
dered system exists on the surface reaching the micron-size regime. The network
band periodicities (evidenced by its cosine-like shape in the ΓM high symmetry di-
rection) relate to the real space network periodicity and match our DFT calculations
and STM data (cf. Table 5.1). Contrary to STS, we find that the band minimum shift
(taking as reference the onset of the 3 ML Ag/Au(111) surface state) is larger for the
DW than the SW network (by 10 meV) and exhibits a narrower bandwidth (almost
half). This confirms a lower coupling between adjacent QDs for the DW case due to
the doubling of the barrier.
Moreover, the characteristic cosine-shape band observed when following ΓM
TABLE 5.1: Extracted ARPES experimental parameters from the electronic bands
3 ML Ag film SW/Ag film DW/Ag film

Band bottom -160 meV -120 meV -110 meV
Band width — 92 meV 51 meV
M point — 0.120 Å−1 0.104 Å−1
Interpore distance — 3.02 nm 3.49 nm
(DFT) — (3.03 nm) (3.45 nm)
m∗ /m0 0.38 0.47 0.59
SW DW
a b
0.0
E-EF (eV)
-0.1
-0.2
-0.3 M K G K M K M K G K MK
0.1 0.1
c d
0.0
0.0 0.0
E-EF (eV)
-0.1
-0.1 -0.1
-0.2
-0.2 -0.2
-0.3
-0.3
M K G K M
-0.3
K M K G K MK
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2

F IGURE 5.6: 2DEG modification induced by the SW and DW network potential
barriers along ΓK. (a, b) ARPES map along ΓK, obtained in the SW and DW net-
works. The EDCs close to Γ were fitted using a Lorentzian component and a linear
background convoluted with a Fermi function (black dashed lines). (c, d) Second
derivative maps of the above raw data for an improved visualization of the second
Surface Brillouin Zone. Contrary to the cosine-shape band structure of the ΓMΓ
high symmetry direction, the bands look even flatter in the ΓKM direction.
symmetry direction of these QD systems, appreciably changes its dispersion when

following the ΓK high symmetry direction (30◦ rotated from ΓM). This is highlighted
in Figure 5.6 where SW and DW bands become flatter beyond the K points. This is a
clear indication that the isotropic paraboloid of the pristine 2DEG becomes distorted
due to the newly formed hexagonal shape SBZ, which is induced by the honeycomb
lattice of the SW and DW nanoporous networks (see Figure 5.7).
To emphasize the SBZ induced distortions on the 2DEG, isoenergetic cuts in
second derivative (kx vs ky ) are compared for the three cases: 3 ML Ag/Au(111),
SW and DW respectively [Figure 5.7]. The selected energies correspond to the band
bottom region of the first confined states (-110 meV), the lower edge of the M point
gap for DW (-65 meV) and the lower edge of the M point gap for SW (-30 meV).
For the last case, no cut for DW is shown, since this energy falls at the gap between
the first and second confined states. In all cases, the pristine 2DEG shows a circu-
lar shape with different radius typical of an isotropic electronic state. However, at
-110 meV [Figure 5.7 (a, d, g)], the band bottoms of SW and DW are nicely observed
replicating even at the second SBZ (as dim purple circles). At -65 meV [Figure 5.7 (b,
e, h)], while SW exhibits some circular shape (with a smaller wave-vector than the
pristine), DW shows a clear hexagonal distortion because it has reached the zone
boundary (M point). We observe the same isoenergetic morphology for the SW at
higher energies (-30 meV). Such observations provide evidence of the modulation of
the 2DEG upon the formation of a honeycomb nanoporous network.
At this point STS and ARPES results seem to be in contradiction. To shed light
into these discrepancies, we performed model calculations with the Electron Bound-
ary Elements Method in combination with the Electron Plane Wave Explansion [115,
64
Barrier Widths
(E = -110 meV) (E = -65 meV) (E = -30 meV)

0.2
0.2 a 0.2 b 0.2 c
3 ML Ag/Au(111)
0.1
0.1 0.1 0.1
ky (Å-1)
0.0
0.0 0.0 0.0
-0.1
-0.1 -0.1 -0.1
-0.2
-0.2 -0.2 -0.2
0.3
0.3 0.3 0.3
d-0.2 -0.1 0.0 0.1 0.2
e-0.2 -0.1 0.0 0.1 0.2 -0.2
f -0.1 0.0 0.1 0.2
0.2
0.2 0.2 0.2
0.1
0.1 0.1 0.1
SW
ky (Å-1)
0.0
0.0 0.0 0.0
-0.1
-0.1 -0.1 -0.1
-0.2
-0.2 -0.2 -0.2
0.2
0.2 g-0.2 -0.1 0.0 0.1 0.2
0.2 h-0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1
-0.2 -0.1 0.0
0.0 0.1
0.1 0.2
0.2
kII (Å-1)
0.1
0.1 0.1
K
ky (Å-1)
DW
0.0
0.0 0.0 G M
-0.1
-0.1 -0.1
-0.2
-0.2 -0.2
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2
kx (Å-1) kxI (Å-1)

F IGURE 5.7: SW and DW induced SBZ distortions of the 2DEG observed in kx vs ky
isoenergetic cuts. (a-c) Cuts corresponding to the isotropic Shockley state of 3 ML
Ag/Au(111) at different energies. (d-f) Identical isoenergetic cuts corresponding to
the SW induced electronic band. A clear hexagonal distortion is observed as one
moves from its fundamental energy (-110 meV) up to the top of the band or M point
at -30 meV. (g, h) Isoenergetic cuts corresponding to the bottom and top of the DW
induced confined state band. The hexagonal distortion becomes apparent when the
zone boundary is reached at the top of the band. All isoenergetic cuts are shown
in second derivative to emphasize the weak replicating bands away from the first
SBZ.
116, 129]. The simulations were done in collaboration with Dr. Zakaria M. Abd El-
Fattah (Al-Azhar University, Egypt). We used simplified structures (straight beads)
for the molecules (inset in Figure 5.8), which agree well with our DFT calculated
iso-potential surfaces (see Figure 5.16). Note that this geometry became valid af-
ter corroborating that a more complicated shape of the barrier yields practically the
same results. In EBEM/EPWE the Schrödinger equation is solved for independent
electrons (2DEG) of effective mass m∗ within periodic domains containing two dif-
ferent potential areas: zero for Ag sites (pores and substrate) and constant non-zero
potential (Vef f ) for the molecular positions (walls) that scatter the electrons. The
m∗ and Vef f parameters are determined by an iterative fitting using both STS and
ARPES data (more details on the simulation process are included in Section 5.4.3).
The agreement of these calculations to the experimental data (both STS and ARPES)
turns out to be excellent, confirming that both substrates (bulk and thin film) are
equivalent for our study. The only requirement is a 100 meV shift to account for the
LDOS (arb. units)
-100 0 100 200

E-EF (meV)
F IGURE 5.8: EBEM/EPWE simulations of the local electronic structure. Calculated
dI/dV spectra, obtained after fitting the experimental data with EBEM/EPWE. The
ARPES fit was shifted by 100 meV for direct comparison with the dI/dV spectra
in Figure 5.3 (a). Insets show both molecular geometries used in EBEM/EPWE
calculations that closely match the STM topographies [Figure 5.1 (e, f)] and DFT
calculated electric field profiles (Section 5.4.4).
different 2DEG onset between Ag(111) and 3 ML Ag/Au(111). We obtain a com-

mon repulsive scattering amplitude of Vef f = 140 meV at the molecular sites for
both SW and DW, but different effective masses: m∗Ag = 0.38 m0 , m∗SW = 0.49 m0 and
m∗DW = 0.54 m0 . Such effective mass increase suggests a change in the electron wave-
function overlap with the crystal substrate concomitant to an enhancement of the
pore confinement that leads to a reduction of the QD coupling when going from SW
to DW (the effect of the effective mass increase is discussed in Section 5.4.3). The
reproduced band structures [white dashed lines in Figure 5.5 (d-f)] match excep-
tionally well with our ARPES data, validating the expected pore coupling difference
and periodic long-range order of these arrays. EBEM/EPWE simulated bands also
mimic very nicely the spectral intensity and band replicas beyond the first BZ (see
Figure 5.15 in Section 5.4.3). Moreover, the calculated local electronic structure (Fig-
ure 5.8) also agrees with the STS data of Figure 5.3 (a), not only in the peak values
but also in shape. Therefore, the EBEM/EPWE simulations bring conclusive con-
sistency when comparing the experimental STS on Ag(111) and ARPES on 3 ML
Ag/Au(111), corroborating that these techniques are complementary to each other.
The inverted order of the STS energy peaks observed in Figure 5.3 (a) can now
be explained. Coupled QDs give rise to bonding and anti-bonding continuum states
when set in arrays [160]. The fundamental energy is established by the bonding
state and the overall bandwidth (proportional to the QD intercoupling) is limited
by the anti-bonding ones [conceptually explained in Figure 5.9 (b) for two coupled
QDs]. The reduced bandwidth of the DW network compared to the SW (by∼ 45%,
66
Barrier Widths
Bonding
Antibonding
b c
Antibonding
Bonding
F IGURE 5.9: Localization of bonding and antibonding states for two coupled QDs.
(a) Simplified model supporting a bonding (red) and antibonding (green) state. (b)
EBEM calculated LDOS as a function of electron energy and position along the axis
through the centers of the QDs. Ψn and Ψ∗n label the bonding and antibonding
states, respectively. (c, d) Two-dimensional LDOS maps at the energies of the an-
tibonding (c) and bonding (d) states. The QDs have a square shape. Note that the
bonding state shows considerable intensity between QDs (at the molecular sites) in
comparison to the antibonding. Figure has been adapted from ref. [160]
cf. Table 5.1) confirms the lower interpore coupling imposed by the wider barri-
ers. However, this does not explain yet the larger STS peak shifts at the pore center
with respect to the ARPES fundamental energies. The underlying reason is that the
STS technique reveals an enhanced sensitivity to probe the anti-bonding state [53,
160]. According to Seufert et al. [160] the wavefunction shape for the bonding state
is more spread out than the anti-bonding one [Figure 5.9 (c, d)]. Thus, as shown in
Figure 5.9 (a) the antibonding state (in green) peaks more abruptly at the pore center
than the bonding state (in red), yielding a higher conductance (for a particular tip
height). Consequently, the peak lineshapes are generally asymmetric with maxima
displaced towards the top of the band (Figures 5.3 and 5.8), which in ARPES matches
the M point energy (after shifting 100 meV).
Another way of explaining this particular STS lineshape can be done from the
band structure perspective: at the bottom of the band, close to the Γ point, electrons
have the largest possible wavelenght of the state (λ= 2π
k ), which means that they are
sensitive to a large fraction of the overall potential of the nanoporous network. Thus,
they spread out over the network and limit their nanopore contribution at the cen-
ter when probed by the STS tip. Contrarily, electrons close to the M point, exhibit
the shortest wavelength of the state, which is of the order of the network interpore
periodicity. This means that they are more prone to be trapped within the QD walls
as they are highly sensitive to the periodic potential landscape (molecular network)
[as already observed in Figure 5.3 (b, c)]. As a consequence, the overall shape of the
5.3. Conclusions 67
LDOS at the pore center reflects a wide and asymmetric shape, gathering more in-
tensity for the higher energy states (cf. Figure 5.8). In fact, the width of the STS peak
is a fingerprint for the degree of coupling between neighbouring QDs. Nevertheless,
the STS is still sensitive enough to the fundamental energy (bonding state) as onsets
are sometimes observed in the spectra. In particular, these can be deduced from Fig-
ure 5.8 (at -20 meV for the SW and -10 meV for the DW), matching the ARPES energy
minima (Γ point) after shifting 100 meV. Therefore, the bandwidth and correspond-
ing QD interaction could be estimated from STS peak width whenever sharp cutoffs
show up in the spectra.
The value of Vef f = 140 meV from our model calculations yields an overall bar-
rier of Vef f ·d = 0.7 eV·Å per molecule, which is small compared to other networks
already reported [16, 129]. Such weak potential barrier can originate from the weak
interaction between the haloaromatic compound and the substrate or by the absence
of metallic coordination in our arrays. Moreover, the use of solely the substrate’s
2DEG m∗ does not provide a good agreement with the SW and DW electronic bands
and we recurrently use an increased value (m∗SW = 0.49 m0 and m∗DW = 0.54 m0 ) close
to the experimental one in order to match the shallow dispersions (cf. Table 5.1).
This suggests that, besides the lateral scattering at the molecule network, there is a
subtle change in the electron wavefunction overlap with the crystal substrate. Note,
however, that we expect this vertical overlap to be practically identical for both net-
works, given that Vef f is the same. Therefore, the additional increase of m∗ when
going from SW to DW barriers, with the associated flattening of the bands, suggests
a correlation with QD intercoupling. In essence, the m∗ increase and band flattening
(reduction in band width) could be considered like fingerprints for increased elec-
tron localization and reduced interdot coupling. Finally, we point out that charge
transfer effects (from sulphur and oxygen) are ruled out from being detrimental in
the observed confinement and coupling effects. This is because the EBEM/EPWE
model considers only homogeneous periodic repulsive scattering potentials and can
precisely simulate the experimental observations.
5.3 Conclusions
In this chapter we have shown that precise engineering of QD array coupling is
possible by modifying just the barrier width (without affecting the QD’s size). These
halogen-bond stabilized organic nanoporous networks are generated on bulk (Ag(111))
and Ag films (3 ML Ag/Au(111)) alike by substitution of a single atom (sulphur vs
oxygen) in the precursor molecule, reminiscent of a butterfly effect. The extended
and periodic nature of these arrays provides access to their distinct band structures,
which are directly compared with their local density of states and merged through
calculations. Such complementary experimental and theoretical synergy provides
complete fundamental insight into the nature of QD intercoupling processes. Even
68
Barrier Widths
though substrate contributions cannot be discarded, our findings clearly suggest

that the reduction of the QD coupling (from SW to DW) is associated with a flatten-
ing of the band dispersion and increase of the effective mass.

5.4.1 ARPES Sample Preparation
A clean Au(111) surface was in-situ prepared by repeated cycles of standard sputter-
ing and annealing (Ar+ sputtering at energies of 1.2 keV, followed by annealing to
800 K). To form Ag films of controlled monolayer on Au(111) a gradient evaporation
of Ag with the substrate at 150 K was performed. Afterwards, to improve the Ag
thin film quality, the sample was annealed to ∼ 450 K [185].
Br-DNT and Br-DNF molecules were deposited from a Knudsen cell heated at
440 K on the substrate at ∼ 130 K in order to make sure that the networks were
formed. Note that these halogen-bonded assemblies are only stable below RT.
The presented Ag thin film engineering turns out of utmost importance in bring-
ing the SW and DW confined state peaks from the unoccupied region down to the
occupied region [compare Figures 5.3 (a) and 5.5 (a)]. Both SW and DW only formed
on 3 ML Ag/Au(111), even though the film quality for 1 ML and 2 ML was excellent.
It is not discarded that the herringbone reconstruction induced surface corrugation,
atomic scale defects, modified diffusion rates as compared to the flat bulk Ag(111)
are affecting the formation of such long-range ordered assemblies. Hence, we spec-
ulate that beyond 3 ML, both networks should form indistinctly of the number of
Ag layers.
Even though Au(111) and Ag(111) have an almost identical lattice constant (∼
2.89 Å), SW and DW networks do not form on Au(111). This is probably due to the
change of diffusion and surface adsorption energies induced by the presence of the
herringbone reconstruction. The herringbone induced corrugation dominates over
the formation of the SW network as nicely illustrated in Figure 5.10 (a-h). As the
deposition coverage is increased, with the sample temperature kept at 150 K [Fig-
ure 5.10 (a-d)], chain-like structures, distorted hexagonal pores and closed-packed
assemblies can be identified, always guided by the herringbone pattern. Each one
of these structures can be better observed in the close-up images [Figure 5.10 (e-h)].
Such differences between Ag(111) and Au(111) illustrates the intricate balance be-
tween molecule-molecule and molecule-substrate interactions required for the sta-
bilization of long-range ordered systems.
a b c d
e f g h
F IGURE 5.10: Molecular arrangements of Br-DNT deposited on Au(111) at 130 K.

(a-d) A series of STM topographic images, with increasing molecular coverage and
(e-h) corresponding close views. The surface corrugation prevents the long-range
formation of 2D hexagonal molecular networks, facilitating chain structures follow-
ing the herringbone reconstruction. We stress that if an ideally regular 2D porous
network as the one observed in (f) would fully extend over the whole surface, the
obtained band structure would be different from the SW and DW cases shown for
Ag(111) because the pores have different symmetries (three-fold vs six-fold) and
exhibits different interpore dimensions. Measurement parameters: Vtip =-200 mV
for all images, I= 10 pA in (a) and (b); I = 2 pA in (c) and (d); I = 5 pA in (e); I = 20
pA in (f); I = 2 pA in (g); I = 20 pA in (h).
5.4.2 STM/AFM Measurements
All experiments were performed by Dr. Shigeki Kawai from MANA/NIMS (Japan)
with an Omicron STM/AFM with a qPlus configuration [186], operating at 4.8 K in
UHV. A clean Ag(111) surface was in-situ prepared by repeated cycles of standard
sputtering and annealing. The W tip of a tuning fork sensor was ex-situ sharpened
by focused ion beam milling technique and was then in-situ covered with Ag atoms
by contacting to the sample surface.
3,9-dibromodinaphtho [2,3-b:2’,3’-d]thiophene (Br-DNT) [177] and 3,9-dibromo-
dinaphtho [2,3-b:2’,3’-d]furan (Br-DNF) were deposited on Ag(111) surfaces at 150
K from a crucible of Knudsen cell. The resonance frequency of the self-oscillating
qPlus sensor was detected by a digital lock-in amplifier (Nanonis: OC4 and Zurich
Instruments: HF2LI and PLL). In STM mode, the tip was biased while the sample
was electronically grounded. The topographic images were taken in a constant cur-
rent mode. In AFM mode, the tip apex was terminated by a CO molecule [114] and
all images were taken at a constant height mode.
70
Barrier Widths
Detection of CO Molecules and Furan Group in DW Network
The appearance of bright spots at the coordination sites in DW network [Figure 5.1 (h)]
may initially induce to consider the formation of a metal-organic coordination with
surface Ag adatoms. However, upon a closer look at Figure 5.1 (h), one can inmeadi-
ately notice that not all coordination sites appear bright (5 out of 6 corners). Hence,
this leads towards considering the presence of an adsorbate. This is indeed the cur-
rent scneario since CO molecules were intentionally dosed for tip funtionalization
purposes. In order to further corroborate this, a series of AFM images are obtained
at the same position while altering the tip-sample separation [Figure 5.11 (a-c)]. Due
to accidental CO manipulation the initially vacant region [highlighted with a red ar-
row in Figure 5.11 (a)] becomes occupied by a CO molecule in the subsequent scan
repetitions [Figure 5.11 (b, c)].
Further understanding of the intricate bonding mechanism for the formation of
the DW network is given by AFM. However, certain elements become more com-
plicated to be detected than others, and this depends on their height with respect
to the surface or their electronic configuration [187–189]. Hence, as illustrated in
a b c
d e f
F IGURE 5.11: Adsorption of CO molecule on the nodal site and AFM imaging of
the oxygen atom of the furan group. (a-c) AFM image series of the DW network
when CO is intentionally dosed into the system for the preparation of the CO func-
tionalized tip. Vertical black arrows indicate the slow scan directions. The tip and
sample separation in (a) is smaller than those in (b) and (c) by 80 pm. In (a), four
CO molecules are found at the nodal sites. By accidental manipulation, the node
indicated by the red arrow becomes filled in (b). The horizontal white arrow marks
the discontinuous frequency shift detected. Measurement parameters: V= 0 mV,
oscillation amplitude A= 60 pm. (d-f) Close view AFM image of the furan moiety
in Br-DNF, taken in constant height mode at different tip-sample distances. (d) is
taken closer than (e) by 80 pm. The O-C bond (f) appears in (d) while it vanishes in
(e). The naphthalene moieties are flat on the surface and no steric stress is induced
by the adsorption at the furan moiety.
Figure 5.11 (d-f), when the tip comes closer to the sample (by 80 pm), the initially
unobservable oxygen atom [Figure 5.11 (e)] becomes visible in Figure 5.11 (d), ex-
hibiting an excellent matching with the expected chemical structure of the Br-DNF
molecule sketched in Figure 5.11 (f).
Defect Concentration in SW and DW Networks
STM measurements can provide a good idea about the quality of the formed nanoporous
network. As evidenced in Figure 5.1 (e, f), such QD arrays are very stable at low tem-
peratures (below RT) and due to the self-healing character of the halogen bond, such
structures self-correct until an almost perfect nanoporous organic layer is formed.
However, some electronic irregularities were detected when looking at the confined
state features at certain pores in Figure 5.3 (b) (blue arrows) apparently missing in
Figure 5.3 (c). Looking at their corresponding STM images [Figure 5.12 (a, b)], we
can identify such imperfections as Br-DNT single molecules or possible Br adatoms
trapped inside some nanocavities (yellow arrows). However, the amount of such
trapped entities is so scarce that they are irrelevant to the ARPES signal.
a b
F IGURE 5.12: Network imperfections after array formation. STM topographic im-
ages of SW (a) and DW (b) networks simultaneously recorded with the dI/dV maps
shown in Figure 5.3 (b, c). The yellow arrows mark defects that we assign to pos-
sible Br adatoms or excessive Br-DNT at the pore and which modulates its local
electronic structure. In (a), only three defective pores out of a total of 132 (2.3%) can
be observed. Such amount of imperfections is irrelevant to the ARPES signal.
From 2D Halogen-Bonded Networks to 1D Covalent Polymers
As it has been already mentioned, such halogen-bonded networks are only stable
below RT. Above this temperature, the surface catalyzed Ullmann coupling reaction
72
Barrier Widths
takes place. Molecules are dehalogenated and a C-C covalent bond is formed, trig-
gering the formation of one-dimensional covalent conjugated polymers on the sur-
face. This effect is illustrated in Figure 5.13 where the SW halogen-bonded network
(a) evolves into one-dimensional polymer chains (b) after annealing the sample at
600 K [180, 190].
a b
F IGURE 5.13: Halogen bond stabilized 2D nanoporous network transformation to

1D covalent polymers upon annealing to ∼ 600 K. (a) 50 x 50 nm2 STM image
showing the SW network configuration. Black spots on the Ag surface correspond
to CO molecules. STM parameters: V = 200 mV; I = 50 pA, T= 6.5 K. (b) 100 x
100 nm2 STM image highlighting the one-dimensional covalent polymer formation.
STM parameters: V = 200 mV; I = 50 pA, T= 77 K.
5.4.3 EBEM/EPWE Simulations
The simulations for this chapter have been performed in collaboration with Dr. Za-
karia M. Abd El-Fattah (Al-Azhar University, Egypt).
Simulation Procedure
In the following, the EBEM/EPWE simulation procedure is explained step by step:

the first step is to parametrize the geometrical structure of the networks that will
represent the potential barriers that scatter the 2DEG. The complexity of the geome-
tries can be simplified since the molecular electrostatic clouds, that are responsible
for the potential barriers, are generally smooth and relatively featureless. We choose
the width and length of the straight beads to match the molecular width (d ∼ 5 Å)
and the network dimensions. Note that this step of considering the scattering bar-
rier geometry (length and width) as accurate as possible to the STM observations of
the backbone dimensions is of key importance for a successful matching and under-
standing of confinement and coupling effects in such nanoporous systems. After this
parametrization, the Vef f (scattering potential at the molecules) and m∗ parameters
are iterated to fit the experimental data. With Vef f the scattering of the surface state
with the molecular barriers is defined and both the band bottom (Γ point in ARPES
and onset of bonding state in STS) and opening of the gaps are quantitatively simu-
lated. The effective mass parameter is inversely proportional to the curvature of the
band so by varying its value, the top of the band or antibonding state is accurately
placed in energy. Note that changes in m∗ do not alter significantly the energetic
position of the band bottom of the first confined state, but it affects the lower edge of
the M point gap or anti-bonding state. This is the reason why we are forced to recur-
rently use a larger effective mass than for the pristine case in order to fit the ARPES
experimental bands [Figure 5.14 (a, b)]. From Figure 5.14 it is evident that when the
pristine surface state m∗ is used (black curves), the band dispersion strongly devi-
ates from the experimental bands (in yellow). We achieve the best ARPES and STS
agreement for the red curve case. Finally, once the iterative fitting is satisfactory for
both STS and ARPES matching, the electronic band intensity can also be obtained
from EBEM/EPWE (Figure 5.15). The simulations show a very nice agreement with
the experimental raw data and the second derivative with replicating bands appear-
ing as weak intensity bands at the second SBZ, similar to Figure 5.5 (e, f).
a SW b DW
0.05
E-EF (eV)
0.00
-0.05
-0.10
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4
F IGURE 5.14: Influence of the effective mass m∗ upon the calculated band structure
for the same geometry and molecular potential Vef f using EBEM/EPWE. (a) Cal-
culated electronic bands along ΓM for the SW geometry using the effective masses
of 0.38 me (clean substrate case, black), 0.49 me (optimal agreement for both ARPES
and STS, red), and 0.56 me (best fit for only the ARPES data but deviating slightly
from the STS, blue). (b) Calculated electronic bands along ΓM for the DW geometry
using the effective masses of 0.38 me (clean substrate case, black), 0.54 me (optimal
agreement for both ARPES and STS, red), and 0.65 me (best fit only for the ARPES
data but deviating slightly from the STS, blue). The results of these calculations can
be compared with the experimental band structure represented as yellow transpar-
ent traces. It is observed that the pristine m∗ = 0.38 me deviates substantially from
the experimental bands and larger effective masses are required to fit all the exper-
imental data. The differences found between the red and blue calculations in (a)
and (b) falls practically within the experimental error.
74
Barrier Widths
a 3 ML Ag/Au(111) SW DW
0.00
0.00 G 0.00 0.00
E-EF (eV)
E-EF (eV)
E-EF (eV)
E-EF (eV)
-0.05
-0.05 -0.05 -0.05
-0.10
-0.10 -0.10 -0.10
-0.15
-0.15 -0.15 -0.15
G M G M G G M G M G
-0.20
-0.20 -0.20 -0.20
b -0.2 -0.1 0.0
G
0.1 0.2 -0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2
0.00
0.00 -1 0.00 -1 0.00 -1
E-EF (eV)
E-EF (eV)
E-EF (eV)
E-EF (eV)
k|| (Å ) k|| (Å ) k|| (Å )

-0.05
-0.05 -0.05 -0.05
-0.10
-0.10 -0.10 -0.10
-0.15
-0.15 -0.15 -0.15
G M G M G G M G M G
-0.20
-0.20 -0.20 -0.20
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1 0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1 0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1 0.1 0.2
0.2
kII (Å-1-1) kII (Å-1-1) kII (Å-1-1)
k|| (Å ) k|| (Å ) k|| (Å )
F IGURE 5.15: Electronic band structure simulated by EBEM/EPWE. (a) Simulated

ARPES EDCs along ΓM high symmetry direction for the pristine 3 ML Ag/Au(111),
SW and DW cases. Such bands exquisitely agree with the experimental ARPES
data shown in Figure 5.5 (a-c). The replicating bands away from the prominent
spectral function close to kk = 0 can also be nicely seen when the second derivative
is taken in (b). However, in the simulations the gap between n=1 and n=2 localized
states is better defined than in the experiments (Figure 5.5). The possible reason
is the ubiquitous inhomogeneities present when these networks are formed that
broadens our experimental spectra, increases the background and washes out the
gaps along with the inherent broadening of the bands due to lossy scattering effects
and the proximity to the Fermi energy.
5.4.4 Ab-Initio Calculations
DFT calculations have been performed by Dr. Ali Sadeghi from Shahid Beheshti
University (Iran). The calculations are carried out within the local density approx-
imation of DFT as implemented in the BigDFT code [191]. A wavelet basis set is
used to expand the wavefunction of the valence electrons while the core electrons
are removed using norm-conserving HGH pseudopotentials [192]. Calculating the
electrostatic potential (and the electric field) for a surface system is uniquely precise
by the Poisson solver of this DFT code that allows to apply periodic boundary con-
ditions along two in-plane directions while keeping free boundary conditions in the
out of plane direction [193].
In the SW molecular network, the structure is stabilized through purely halogen
bonding in a trimer configuration [Figure 5.16 (a)] [194]. The halogen atom, when
covalently bonded to a molecule, induces an area of positive electrostatic potential
on the outermost portion of the atom along the bond axis (σ-hole) [176, 178, 194]
[Figure 5.1 (c)], thus stabilizing the network by the attraction between positive and
negative electrostatic poles on the Br atoms [Figure 5.16 (a)]. The structure trans-
forms in the DW network as halogen bonding is combined with Br · · · O bonds
generating the double molecular barrier between nanocavities [Figure 5.16 (b)].
SW
DW
F IGURE 5.16: Models for SW and DW networks on Ag(111). Corresponding mod-

els predicted by DFT calculations for the arrangement of the SW (a) and DW (b)
halogen based networks on Ag(111). Br atoms are shown in brown, O in red, S
in yellow and Ag by large gray spheres. The hexagonal lattice of the DW net-
work matches a (12 x 12) surface unit cell. This results in an inter-pore distance
of 12a = 3.45 nm (a is the Ag-Ag distance) and the molecular layer lattice vector
aligns along the [11̄0] direction.
Conversely, the hexagonal lattice of the SW net-
10 −1
work exhibits a surface unit cell concomitant with atomic positions of
1 11
the Ag substrate and presenting a√small angle of 4.7◦ with the [11̄0] direction. The
inter-pore distance in this case is 111a = 3.03 nm. The 4 Å difference between the
inter-pore distances in the two cases is very close to the width of a molecular unit.
Therefore, the size of the nanocavities by SW and DW networks are practically
identical. The color maps in (a) and (b) illustrate the distribution of the normal
component of electric field E, on a plane 3 Å from the molecular layers (i.e. at the
position of the metal surface). The plots share the same color scale, where red corre-
sponds to E < −0.015 V/Å, blue to E > −0.015 V/Å, and green to zero. The local
electric field, acting as a scatterer for the surface electrons, becomes significant only
underneath the molecular units. The effective geometry and width of the barriers
is estimated from these plots. The width per molecular unit turns out to be d ≈ 5
Å, coinciding with the one used in the EBEM/EPWE models.
77
Chapter 6
Tunable Energy and Mass

Renormalization from Homothetic
Quantum Dot Arrays
6.1 Introduction
In this chapter we show that the fundamental energy of confined states within nano-
porous networks can exhibit downward shifts from EF accompanied by a lowering
of the effective masses in the presence of small gap openings at zone boundaries.
We observe these effects by ARPES in two homothetic (scalable), cobalt coordinated
metal-organic arrays obtained by self-assembly on Au(111). Specifically, we find that
the effective mass lowering and downward energy shift are gradual, i.e. dependent
on the network dimensions. Local scanning tunneling microscopy and spectroscopy
measurements agree with these findings as evidenced by the gradual downward
shift of the density of states onset and the presence of confined states at the nanocav-
ities. EBEM/EPWE simulations, DFT calculations and PAM provide insight into the
nature of this phenomenon, which we infer is related to overlayer-substrate interac-
tions in the form of adatom hybridization effects and geometrical variations of the
metal-organic assembly.

The studied cobalt coordinated networks are practically homothetic (scalable) and
generated from two cyano-polyphenyl precursors deposited on Au(111). Specifi-
cally, we used para-hexaphenyl-dicarbonitrile (Ph6) and para-terphenyl-dicarbonitrile
(Ph3) molecules that coordinate with Co atoms in a 3:2 stoichiometry. These tectons
are thermally and sequentially evaporated (molecules first, then Co) followed by a
mild annealing at 400 K and result in two scalable, periodic, long-range, single do-
main and practically defect-free QD arrays, as shown in Figure 6.1 (a, e) and named
hereafter Ph6Co (top row) and Ph3Co (bottom row). In agreement with previous
work [17], these networks present sixfold symmetry with unit cell vectors of 5.78 nm
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
78
Dot Arrays
a b c d
10 nm 5 nm 5 nm 2 nm
e f g h
10 nm 5 nm 5 nm 2 nm
F IGURE 6.1: STM characterization and EBEM/EPWE scattering regions of Ph6Co

and Ph3Co QD arrays. (a-c) STM images of Ph6Co network on Au(111). STM pa-
rameters: V = -1 V; I = 10 pA, Tsample = 77 K. (d) 2D potential geometry used
for Ph6Co EBEM/EPWE modelization. (e-g) STM images of Ph3Co network on
Au(111). STM parameters: V = -2V; I = 20 pA and Tsample = 77 K for (e); V = -1
V; I = 10 pA and Tsample = 4.5 K (f, g). (h) 2D potential geometry used for Ph3Co
modelization. According to (b, f) both networks grow 30◦ rotated from the [110] di-
rection of the surface. The herringbone periodicity is unchanged for both networks
(b, f), demonstrating the weak chemical interaction with the Au surface [10, 195].
(for Ph6Co) and 3.53 nm (for Ph3Co) along the [112] direction and, respectively en-
close pore areas of 24 nm2 and 8 nm2 [Figure 6.1 (c, g)]. Note that the chemical inter-
action of both networks with the substrate must be rather weak based on the fact that
the herringbone reconstruction is not lifted or visually modified [Figure 6.1 (b, f)] [10,
195]. We experimentally probe these networks at two labs, first mesoscopically by
means of ARPES (see Section 6.5.1) and then at the nanoscale by means of a STM
/ STS setup at 5 K (see Section 6.5.2). We shed light on these experimental data by
means of EBEM/EPWE simulations, DFT calculations and PAM, all of which are de-
tailed in the Supplementary Information of this chapter (Section 6.5). This work has
been done in collaboration with the group of Prof. M. Stöhr (University of Gronin-
gen, Netherlands) that performed the LT-STM/STS, Zakaria M. Abd El-Fattah (Al-
Azhar University, Egypt) that contributed with the PAM and EBEM/EPWE simula-
tions and the group of Prof. A. Arnau (University of the Basque Country UPV-EHU,
Spain) that performed the DFT calculations.
Ph6Co and Ph3Co networks are formed on Au(111), where the surface state on-
set can be comfortably accessed by ARPES and STS [Figures 6.2 (a), 6.4 (j)]. Never-
theless, obtaining the corresponding QD array band structure by means of ARPES is
exceedingly challenging as the networks must be periodic, extended, almost defect-
free and fully spread over the probed surface [15, 54]. To achieve such demanding
conditions we evaporate the molecules and Co adatoms by creating two orthogo-
nal shallow gradient depositions on a large (15 mm in diameter) circular Au(111)
substrate, ensuring the existence of an area hosting optimal coverage and the exact
3:2 stoichiometry [158]. The sample is then mildly annealed to 400 K in order to
improve the quality of the networks formed (see Section 6.5.1 for more details on
the preparation process). Figure 6.2 (b, c, e, f) shows the second derivative of the
ARPES spectral density from Ph6Co and Ph3Co along the ΓM and ΓK high symme-
try directions. The dispersion intensities and zone boundary gaps vary depending
on the SBZ direction followed and the size of the QDs [Figure 6.2 (d)]. We unam-
biguously observe a gradual downshift of the fundamental energy (Γ point) towards
higher binding energies as the pore size is reduced, which can be quantified in the
normal emission EDCs (Fig. 6.2 (g) and Table 6.1). Note that this goes in the op-
posite direction to the energy shift expected from conventional lateral confinement
scenarios [18, 97, 98]. Simultaneously to this downshift, we observe a reduction in
the effective mass (Figure 6.2 (i) and Table 6.1), leading to an effective 2DEG Fermi
wave-vector (kF ) pinning [Figure 6.2 (h)]. The partial confinement of the substrate’s
2DEG is inferred from the presence of small gaps (observed as slight intensity vari-
ations) at the symmetry points, which denotes weak scattering from the network
molecular barriers [157]. Note that the absence of spin-orbit splitting in our data
when the networks are present on the surface, in no way outrules this effect, as it
could be masked by intrinsic broadening of the ARPES lineshape [157, 173, 196].
To find out the surface potential landscape that the networks generate to the
2DEG, we perform EBEM/EPWE simulations. Firstly, the geometry of both systems
[Ph6Co and Ph3Co in Figure 6.1 (d, h)] is generated following STM images [see Fig-
ure 6.1 (c, g)], assigning two repulsive scattering potential sites: Vmol = 250 meV
for molecules (in green) and VCo = 50 meV for the Co regions (in purple). For
more details on the scattering potential landscape assignment see Section 6.5.4 and
Chapter 7. Such potentials perfectly reproduce the experimental band dispersion
and energy gaps (∼ 25 meV for Ph6Co and ∼ 30 meV for Ph3Co at M), reflecting
the weak scattering strength (second Fourier component) of the networks [cf. su-
perimposed red bands in Figure 6.2 (b, c, e, f)]. However, repulsive scattering is
known to shift upwards the 2DEG fundamental energy (at Γ), opposite to what is
observed here. Thus, the ARPES dispersions can only be matched by EBEM/EPWE
TABLE 6.1: ARPES experimental binding energies at Γ and effective masses

(columns EB Γ
and m∗ /m0 ) for the substrate and the two networks. Matching the
network band structures by EBEM/EPWE simulations demands downshifting the
SS references accompanied by a reduction of the effective masses with respect to
Ref,Γ
the Au(111) Shockley state (columns EEBEM and m∗,Ref
EBEM /m0 ).
Γ (eV)
EB m∗ /m0 Ref,Γ
EEBEM (eV) m∗,Ref
EBEM /m0
Au(111) 0.45 0.255 0.45 0.26
Ph6Co 0.49 0.24 0.52 0.24
Ph3Co 0.55 0.22 0.59 0.21
80
Dot Arrays
a Au(111) b Ph6Co
P6 kr 2Der GM c Ph3Co
P3 kr 2Der GM
0.0
0.0 0.0 0.0
E-EF (eV) -0.1

-0.1 -0.1 -0.1
-0.2
-0.2 -0.2 -0.2
E-EF (eV)
E-EF (eV)
E-EF (eV)
-0.3 -0.3 -0.3 -0.3
-0.4 -0.4 -0.4 -0.4
-0.5
-0.5 -0.5 -0.5
-0.6
-0.6
G
-0.6
M G M G M G M
-0.6
G M G M G
P6 kr 2Der GK P3 kr 2Der GK
-0.2
-0.2
-0.1
-0.1
0.00.0
0.1
0.1
0.2 0.2 -0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2
e f
-1 -1 -1
k (A )
II kII (A ) kII (A )
kII (Å-1) 0.00.0 0.0
d -0.1
-0.1 -0.1
E-EF (eV)
-0.2 -0.2 -0.2
E-EF (eV)
E-EF (eV)
K -0.3
-0.3 -0.3
M
K -0.4
-0.4 -0.4
G M G M -0.5
-0.5 -0.5
-0.6 -0.6
K M K G K M K
-0.6
M K G K M
-0.2
-0.2
-0.1 -0.1
0.00.0
k (A )
II
-1 0.1 0.1 -0.2
0.2 -0.2 0.2 -0.1
-0.1 0.0
0.0
k (A )
II
-1
0.1
0.1 0.2
0.2
Intensity (arb. units) kII (Å-1) kII (Å-1)
g h i
0.0
0.0 Fermi wave-vector (kF)
Au(111) 0.27
0.27
Ph3Co Au(111)
units)
Ph6Co Ph3Co 0.26

0.26
(arb.Units)
Ph6Co
-0.2 0.25
0.25
(eV)
E-EF (eV)
m*/m0
0.24
0.24
Int.(Arb.
0.23
0.23
E-E F
-0.4
-0.4 0.22
0.22
Int.
0.21
0.21
0.20
0.20
-0.6
-0.6 -0.2
Au(111) Ph6Co
Au(111) Ph6Co Ph3Co
Ph3Co
-0.2 -0.1
-0.1 0.0 0.1 0.2
EDCs at G kkIIII(Å -1
-1
(A ))
Int. (Arb. Units)
F IGURE 6.2: Ph3Co and Ph6Co electronic band structure studied with ARPES. (a)
Electronic structure of the Au(111) Shockley state, highlighting its well-known spin
orbit splitting. ARPES spectral density obtained along ΓM (b, c) and ΓK (e, f)
high-symmetry directions for both Ph6Co and Ph3Co nanoporous networks. The
EBEM/EPWE simulated band structure appears superimposed in red, showing
nice agreement with the intensity modulations, gaps and replicating bands. (d) Re-
ciprocal space directions followed along the new SBZ originated by the nanoporous
networks. (g) EDCs at normal emission (Γ point ) for pristine Au(111) (green),
Ph6Co (blue) and Ph3Co (red). A gradual downshift of the fundamental energy as
the pore size is reduced (∆EP h6Co = 40 meV and ∆EP h3Co = 100 meV with respect
to the Au SS) is observed. (h) Fermi wave-vector (kF ) features are plotted for both
Ph3Co (red) and Ph6Co (blue) networks and compared to the pristine Au surface
state (green). (i) The effective mass reduction tendency as the pore size decreases
(from Ph6Co to Ph3Co). ARPES measurements are shown in second derivative
(TSample = 150 K).
when adopting higher binding energy references and smaller effective masses than
the pristine Au(111) SS (Table 6.1). In other words, the original 2DEG of the Au(111)
surface fails when used as scattering reference for both these networks. Figure 6.3 (a,
b, d, e) summarizes the EBEM/EPWE simulated band structures, which match the
experimental findings of the gradual downward energy shift from Ph6Co to Ph3Co
in Figure 6.3 (c) and the effective Fermi wave-vector (kF ) pinning with respect to the
pristine Au(111) in Figure 6.3 (f).
Before exploring these 2DEG reference shifts in more detail by using local spec-
troscopic techniques (STM/STS), we should verify that these networks confine the
a Ph6Co
P6 kr 2Der GM BEM b Ph3Co
P3 kr 2Der GM BEM c
0.0
0.0
0.0
0.0 0.0
Au(111) ARPES
Ph3Co EBEM
-0.1
-0.1 -0.1 Ph6Co EBEM
-0.2
-0.2
(eV)
E-EFF (eV)
-0.2
(eV)
-0.2 -0.2
E-EF (eV)
E-EF (eV)
-0.3
-0.3 -0.3
F
E-E
E-E
-0.4
-0.4
-0.4
-0.4 -0.4
-0.5
-0.5 -0.5
-0.6
-0.6
-0.6
-0.6
M G M G M G M
-0.6
G M G M G EDCs at G
P3 kr 2Der GK BEM
P6 kr 2Der GK BEM
-0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2 Intensity (arb. units)
Int. (Arb. Units)
d 0.0
0.0 -1
kII (A ) 0.0 e -1
kII (A ) f
-0.1
-0.1 -0.1
Fermi wave-vector (kF)
Au(111) ARPES
-0.2
-0.2 -0.2
units)
F (eV)
Ph3Co EBEM
(arb.Units)
(eV)
E-EF (eV)
Ph6Co EBEM
-0.3
-0.3 -0.3
F
E-EE-E
Int. (Arb.
-0.4
-0.4 -0.4
-0.5
Int.
-0.5 -0.5
-0.6
-0.6
K M K G K M K
-0.6
M K G K M
-0.2 -0.2 -0.1
-0.1 0.0
0.0-1 0.1 0.1 0.2 -0.2
0.2 -0.2 -0.1
-0.1 0.0
0.0 -1
0.1
0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1 0.2
0.2
kII (A ) kII (A ) -1-1)
kII (Å-1) kII (Å-1) kkII II(A
(Å )
F IGURE 6.3: Ph6Co and Ph3Co EBEM/EPWE simulated bands in second deriva-
tive. The ΓM high-symmetry direction is shown in (a, b) while ΓK in (d, e). The
gradual downward shift is corroborated by the EDCs at Γ (c) and kF pinning in
(f). For the latter two cases (c, f) the pristine Au(111) ARPES data is shown as a
reference.
Au SS. We do this for Ph6Co network expecting the same behavior for Ph3Co [17,
166]. Figure 6.4 (a) (middle) shows two position dependent experimental conduc-
tance (dI/dV ) spectra for Ph6Co, one located at the pore center (C) and the other
halfway (H) (see inset). The lineshapes together with the differential conductance
maps in Figure 6.4 (b - e) exhibit clear confinement eigenstates [15–17, 53, 54, 166].
Note that for ascending energies a transition from the n=1 confined state to the
higher order confined states (n = 2, 3 and 4) is observed. This electron localization
mirrors the one observed for the same network on Ag(111) [17, 166]. To compare
them directly we must adapt the reported dI/dV spectra of ref. [17] by normalizing
∗,P h6Co
the energy axis by the ratio of their corresponding effective masses (mAg /m∗,P
Au
h6Co
=
0.41/0.24) and shifting the onset to the Au reference (−485 meV at 5 K). The agree-
ment (lineshape and peak energies) between the two datasets is satisfactory [cf.
middle and top of Figure 6.4 (a)], proving that these 2DEGs are similarly confined
by Ph6Co on both Ag(111) and Au(111). The expected confined state energy vari-
ation with the reduction in pore size (from Ph6Co to Ph3Co) is summarized in
Section 6.5.3, also agreeing with previous observations of Ph6Co and Ph4Co on
Ag(111) [17].
Once confinement is demonstrated, we can address the 2DEG reference shift
upon network formation using local techniques. Prior to this, we recall from the
previous chapter that fundamental energies of QD arrays (LDOS onset) are inher-
ently ill-defined by STS. The reason is that the dI/dV lineshapes of partially local-
ized states are broad (by the interdot coupling) and quite asymmetric (the maxima
are displaced towards higher energy). Figure 6.4 (j), shows dI/dV blow-ups at the
pore center of the two, Ph3Co and Ph6Co networks and of the Au SS as reference.
82
Dot Arrays
a dI/dV EBEM LDOS

b f
Ph6Co/Ag
Ph6Co
Ph6Co Ag(111) C
Ag(111) Center
Center
Ph6Co Ag(111) Halfway
Ag(111) Center
Center
n=1
Ph6Co/Ag
Ph6Co H
Halfway
dI/dV (arb. units)
Ph6Co Au(111)
Ag(111) Center
Halfway
Ph6Co
Au(111) Halfway
Ph6Co Ag(111)
Ph6Co/Au
Ph6Co Au(111)
Center
C
Halfway
Center
Au(111) Center
Halfway
n=1
n=2 n=4 EBEM LDOS
Ph6Co LDOS
Ph6Co/Au
EBEM
Ph6Co Center
Au(111) Halfway
H
Center
Au(111) Halfway
Halfway
EBEM LDOS Center
Halfway
EBEM LDOS Center
u.)
EBEM LDOS
LDOS Halfway
Halfway
u.)
EBEM
-0.39 V -0.42 V
(a.
(a.
(a. u.)
u.)
(a.
c g
dI/dV
dI/dV
dI/dV
dI/dV
Ph6Co Ag(111) Center

Ph6Co
Ph6Co Ag(111)
Ag(111) Center
Halfway n=2
Ph6Co
Ph6Co Ag(111)
Au(111) Halfway
Center
Ph6Co
Ph6Co Au(111)
Au(111) Center
Halfway
Ph6Co Au(111) Halfway
EBEM LDOS C
EBEM LDOS Center
EBEM
EBEM LDOS
LDOS Center
EBEM LDOS H
Halfway
EBEM LDOS Halfway
-0.25 V -0.29V
u.)
(a.u.)
d h
dI/dV(a.
dI/dV
-0.6
-0.6 -0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4
-0.6
-0.6 -0.4
-0.4 -0.2
Tunneling
Tunneling Voltage
-0.2 Voltage0.0
(V) (V)
0.0 0.2
0.2 0.4
0.4
-0.6
-0.6 -0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 n=3,4
j Tunneling Voltage
Tunneling
ARPES Tunneling
Voltage (V)
Tunneling Voltage
Voltage (V)
(V)
(V)
dI/dV (arb. units)
-0.1 V -0.14V
e i
-0.6 STS
-0.4 -0.2 0.0 0.2 0.4
-0.6 -0.4 -0.2 0.0 0.2 0.4
Tunneling Voltage (V) n=4
Tunneling Voltage (V)
Ph3Co/Au(111)
Ph6Co/Au(111)
Au(111) SS 0.05 V -0.04V
-0.70
-0.7 -0.6
-0.60 -0.5
-0.50 -0.4
-0.40 -0.3
-0.30
Tunneling Voltage
(V)
F IGURE 6.4: Ph6Co confinement effects and downward shift observed by

STM/STS. (a) dI/dV spectra at the pore center (C, black) and halfway (H, red)
for three Ph6Co datasets (see inset): Experimental conductance of this network
on Au(111) (middle), corresponding EBEM/EPWE conductance simulations using
the ARPES parameters (bottom), and experimental spectra of Ph6Co on Ag(111)
adapted from ref. [17] and normalized to Au(111) (top). Each spectrum is made up
of the characteristic confined states that alternate depending on the wavefunction
spatial distribution, i.e. n = 1 and n = 4 peaks at the pore center spectrum and
n = 2 and n = 3 in the halfway one. (b - e) Experimental dI/dV maps reproducing
standing wave patterns of the different energy levels n showing reasonable agree-
ment with the EBEM/EPWE simulated ones at similar energies (f - i). (j) Zoom-in
onto the experimental dI/dV onset for the pristine Au(111) SS (green), Ph6Co net-
work (blue) and Ph3Co (red) probed at the center of the pores. A gradual down-
shift of the fundamental energy is observed as the pore is reduced, agreeing with
the ARPES data of Figure 6.2 (g). The ARPES energetic position is indicated by the
vertical lines (temperature corrected by 30 meV).
For Ph3Co (in red) the LDOS onset is clearly away from the Au SS, whereas for
Ph6Co (in blue) is much closer to it. In the latter case, we could simulate and match
the experimental conductance spectra and maps [Figure 6.4 (a, f-i)] using the same
scattering parameters and effective mass reduction as for fitting the Ph6Co ARPES
bands [Figure 6.2 (b, e)]. In these LDOS maps we observe slight discrepancies for
the higher energy maps [Figure 6.4 (h, i) vs Figure 6.4 (d, e) ], which we ascribe to
weak potential variations introduced by the herringbone reconstruction. Indeed the

Ph6Co unit cell (5.78 nm) is large enough to host both fcc and hcp regions within
a single pore, which was not accounted for in the EBEM/EPWE simulations. Such
slight variations become evident as higher energies are probed.
In essence, the STS shifts qualitatively agree with the ARPES results, as observed
in Figure 6.4 (j) (vertical lines), supporting a change of the 2DEG reference upon the
metal-organic nanoporous network presence on the surface. We believe that subtle
downward energy shifts, such as the ones present in Ph3Co and Ph6Co, may also ex-
ist in other published metal-organic networks [17, 166, 182], but become practically
undetectable unless STS measurements are complemented with photoemission ex-
periments that can unambiguously define the reference substrate energy for the QD
array (see Section 6.3).
Many factors might be responsible for these counterintuitive, downward energy
shifts set by the 2DEG reference modification. Likely, it can be attributed to gen-
eral overlayer-substrate interactions in the form of charge transfer (doping effects),
hybridization effects with metal adsorbates or geometrical variations (overlayer-
substrate distance) that may renormalize the surface state band. As the shift is grad-
ual, being larger for Ph3Co than Ph6Co, and the networks are homothetic, it could
be induced by charge transfer from the Co adatoms (in numbers ranging from 0.015
monolayers (ML) to 0.005 ML) to the Au SS, similar to the downshift induced by
alkali metals [197]. However, the fact that m∗ decreases such as to keep the Fermi
wave-vector (kF ) practically pinned suggests unaltered occupancy of the 2DEG (the
2
kF
electron density n is directly related with kF by n = 2π ) [126, 198].
To examine the possible Co/Au hybridization, we have performed DFT calcu-
lations of Co atom arrays onto a non-reconstructed Au(111) surface. Figure 6.5 (a)
shows the calculated band structure from two selected supercells: the 2×2 (0.25 ML
of Co) on the left and the 3×3 (0.11 ML of Co) on the right. These superstructures
introduce an evident difference in the folding of the Au bands (in black), but more
importantly, a clear downshift of the pristine Au SS (in red). We find that the magni-
tude of such downshift is directly related to the amount of isolated Co adatoms on
the surface (Table 6.2). The actual experimental Co coverage is much lower by about
an order of magnitude, the main consequence being that the corresponding shift is
of the order of 50 meV, comparable to the experimental observations. We assign
this effect to a coupling (hybridization) between the Co 3d-bands [indicated in blue
in Figure 6.5 (a)] with the Au(111) SS [140, 198]. The effect of a Co network (with-
out molecules) on the Au SS cannot be tested with ARPES, since at low coverages
Co atoms aggregate forming clusters at herringbone elbows [199, 200], inducing, in
fact, the opposite energy shift of the Au SS [Figure 6.6 (a)]. Figure 6.6 (b, c) show a
similar effect with increasing Ph3 and Ph6 coverages, respectively.
In addition to the aforementioned effects, that is, pure charge transfer and
84
Dot Arrays
a 2.0
0.25 ML Co 0.11 ML Co b
2.0
2.0 Ag(111)PAM
Cu(111)PAM
Au(111)PAM
1.5
1.5 C/Ag(111)
1.0
d*
1.0
1.0
E-EF (eV)
E-EF (eV)
d0
0.5
0.5
0.0
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.0
-2.0 1.5
1.5 2.0
2.0 2.5
2.5 3.0
3.0 3.5
3.5 4.0
4.0
M G M d* (Å)
F IGURE 6.5: DFT and PAM simulations on hybridization and overlayer height ef-
fects on the 2DEG reference. (a) Visualization of the Au(111) surface state (red curve
with parabolic dispersion centered at Γ) downward energy shift at two different Co
coverages. The left panel corresponds to 0.25 ML and the right panel to 0.11 ML,
as obtained using a 2x2 and a 3x3 surface unit cell, respectively. The different su-
percells introduce an evident difference in the folding of the Au bands. The blue
curves near the Fermi level correspond to Co 3d-bands. (b) Effect of the overlayer
distance on the 2DEG reference for the three (111) noble metal surfaces studied
using PAM. The full colored dots (red, orange and gray) indicate the pristine SS
bottom energy and the arrows indicate the maximum downshifts that depend on
the projected bulk gap position. On the other hand, the rightmost green curve is ob-
tained from the model developed for studying the influence of an organic overlayer
on the Ag(111) SS, starting from a neutral overlayer distance of ∼ 3.2 Å (adapted
from ref. [201]).
hybridization effects, other mechanisms could also be considered. In particular, ge-

ometrical variations of the overlayers (vertical displacements) are also known to af-
fect the SS reference [122, 123]. Specifically, molecular assemblies generally tend to
increase their separation to the surface when they form metal-organic assemblies
(adatoms embedded into the structures) [167, 202, 203]. Accordingly, two differ-
ent models (one based on the PAM for metallic overlayers [122, 123] and the other
TABLE 6.2: Energy shift of the Au(111) SS with Co adatom concentration and array
size obtained from our DFT calculations. The Co adatoms relax ∼ 2.5 Å above the
unreconstructed Au(111) surface. The calculations show an increasing downward
shift of the SS with Co content. Note that the experimental amounts of Co used are
significantly lower, corresponding to 0.015 ML for Ph3Co and 0.005 ML for Ph6Co
(very diluted).
Co concentration Array size ∆ESS (eV )

0.25 ML 2×2 -0.94
0.17 ML 3×2 -0.7
0.11 ML 3×3 -0.49
0.08 ML 3×4 -0.54
a b c
-0.41 -0.43 -0.40
-0.42 -0.41
-0.44
E-EF (eV)
-0.43 -0.42
-0.45
-0.44
-0.43
-0.46
-0.45
-0.44
-0.46 -0.47
0.00 0.04 0.08 0.12 0.14 0.15 0.20 0.25 0.30 0.00 0.10 0.20 0.30
Co Coverage (ML) Ph3 Coverage (ML) Ph6 Coverage(ML)
F IGURE 6.6: Gradual band bottom energy shift of Au(111) SS upon Co, Ph3 and
Ph6 gradient depositions. (a) (E vs Coverage) plot shows a tendency of the Au SS
to shift towards the Fermi level as the Co deposition is increased. For the diluted
amount of Co required for both Ph3Co and Ph6Co network formations, the surface
state remains unaltered within the experimental error. (b, c) (E vs Coverage) plots
for Ph3 and Ph6 gradient depositions respectively. For both cases, at low coverages
the surface state remains energetically unchanged and gradually shifts towards the
Fermi level, which is more clearly observed for Ph6.
for molecular overlayers [201]), show that significant energy downshifts are also
possible whenever the metal/molecule overlayer-surface distance increases [Fig-
ure 6.5 (b)]. For instance, for the Ag(111) case (in grey), when an Ag overlayer is
positioned at a distance (d∗ ∼ 2.3 Å, inset), the Ag SS remains at its original energy
(grey dot). As the overlayer is lifted up from the surface (d∗ > 2.3 Å), the Ag SS shifts
its reference to lower energies until it meets with the projected bulk band position
(horizontal dashed grey line). The same mechanism holds for Au(111) and Cu(111)
(see Section 6.5.5 for more details). Besides, if an organic overlayer, such as graphene
on Ag(111) is considered, a similar behavior is predicted (green curve) [201]. As the
organic layer is lifted up from the surface (beyond ∼ 3.2 Å), the Ag SS shifts down-
wards in energy. Similarly, we speculate that subtle vertical increments in both of
our networks could be another plausible reason behind the observed counterintu-
itive Au surface state renormalization.
86
Dot Arrays
6.3 Corroboration of 2DEG Renormalization in Other MOCNs

In an attempt to extend the previously observed 2DEG renormalization effects, we
grow a similar metal-organic nanoporous network to Ph3Co and Ph6Co using sin-
gle Cu atoms as coordination. For such purposes, in collaboration with the group
of Prof. N. Lin (The Hong Kong University of Science and Technology, China), we
chose TPyB (1,3,5-tri(4-pyridyl)-benzene) since we judged it could form a long-range
ordered metal-organic film on Cu(111) [154, 182] [Figure 6.7 (a)]. After molecules
are deposited on the surface following a wedge evaporation, the sample is post-
annealed to 420 K to improve the formation of the nanoporous network. When the
film is saturating the surface, a single domain and long-range ordered phase is ob-
served in STM and LEED, which corresponds to a nanoporous network with 2.65
√ √
nm periodicity and a 6 3 × 6 3 R30◦ superstructure [Figure 6.7 (b, c)].
This long-range ordered nanoporous film provides exceptional QD bands when
probed by ARPES: i) the Cu(111) Shockley state [Figure 6.7 (d)] shifts significantly to
higher binding energies, ii) clear replicas appear away from the main intensity when
following ΓM and ΓK high symmetry directions of the hexagonal SBZ that originate
from the nanoporous overlayer [Figure 6.7 (e, f)]. Indeed, EDCs at Γ show a ∼ 70
meV shift of the band bottom (fundamental energy) to higher binding energies [Fig-
ure 6.7 (g)]. In addition, the Fermi wave-vector (kF ) seems practically unaffected,
evidencing that such a downward shift is not caused by charge transfer effects [Fig-
ure 6.7 (h)]. Finally, photoemission observations are complemented by LT-STS (5 K)
measurements performed at the pore center of the network [Figure 6.7 (i)]. The on-
set of the Cu SS shifts ∼ 70 mV to higher binding energies, in total agreement with
ARPES results (dashed vertical lines). Note that this observation contrasts with the
electronic properties reported for this nanoporous network in ref. [154], where an op-
posite shift of the fundamental energy was claimed. The reason for such discrepancy
is that the fundamental energy of partially localized states in STS was positioned at
the top of the asymmetric peaks instead of choosing its lowest energy onset (see
Chapter 5).
With this observations we unambiguously find that the surface state renormal-
izes by shifting its energy to higher BEs upon the formation of certain metal-organic
coordination networks on surfaces. We have identified such effect in single metal
atom coordination networks, in particular: Ph3Co and Ph6Co on Au(111), Ph4Co
and Ph6Co on Ag(111) [17], Ph5Cu on Cu(111) [166], Cu-TPyB and Cu-ExtTPyB [154].
Note that the most evident energy shifts are detected for the smallest pores, i.e. the
larger amount of single coordination metal atoms.
6.3. Corroboration of 2DEG Renormalization in Other MOCNs 87
a b c
[11-2]
d e f
0.0
0.0 0.0 0.0
-0.2
-0.2 -0.2 -0.2
E-EF (eV)
-0.4
-0.4 -0.4 -0.4
-0.6
-0.6 -0.6 -0.6
G M G M G M M K G K M K
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 0.3
0.3 0.4
0.4 -0.2 -0.1
-0.2 -0.1 0.0
0.0 0.1
0.1 0.2
0.2 0.3
0.3 0.4
0.4
kx (Å-1) kx (Å-1) ky (Å-1)
g h i ARPES
Cu(111)
0.0
0.0 TPyB+Cu Cu(111) SS
STS
TPyB network (GM
TPyB network (GK
-0.2
E-EF (eV)
-0.2
-0.4
-0.4
-0.6
-0.6 TPyB+Cu
Cu(111)
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2
Intensity (arb. units) -0.6 -0.5
-0.5 -0.4 -0.3 -0.2 -0.1 0.0
0.0
kII (Å-1)
F IGURE 6.7: Cu-TPyB network on Cu(111): growth and electronic properties. (a)
Schematic drawing of the Cu-TPyB network where Cu coordination atoms are
shown in purple. (b) STM image shows the long-range ordered formation, in agree-
ment with√ ref. [154].
√ STM parameters: V = 1 V; I = 100 pA. (c) LEED pattern show-
ing the 6 3 × 6 3 R30◦ superstructure of the network (LEED energy = 19 eV). (d)
Electronic ARPES band structure of the pristine Cu(111) and the network modu-
lated 2DEG along ΓM (e) and ΓK (f) high-symmetry directions . (g) EDCs at the Γ
point showing a shift of the fundamental energy of 70 meV to lower energies for
the network. (h) Fermi wave-vector (kF ) measurements for the pristine Cu SS and
network. (i) Experimental dI/dV onset for the pristine Cu SS (blue) and partially lo-
calized state at the pore center of the network (red). The vertical dotted lines mark
the position of the ARPES band bottom.
88
Dot Arrays
6.4 Conclusions
In summary, ARPES and STS results evidence a gradual energy and mass renormal-
ization of the Au(111) SS upon formation of two homothetic Co-coordinated metal-
organic networks (Ph3Co and Ph6Co). EBEM/EPWE simulations could match the
experimental data only after the 2DEG reference is shifted to higher binding ener-
gies. Notably this downshift is gradual with decreasing pore size and occurs in spite
of the demonstrated confining attributes of the nanocavities. Our EBEM/EPWE sim-
ulations can satisfactorily match our experimental data by means of weakly repul-
sive potentials for molecules and Co atoms. Overlayer-substrate vertical interactions
must be responsible for these effects on the Au SS reference. Our DFT calculations
and PAM simulations help us to disentangle possible mechanisms responsible for
these downshifts, where Au SS-Co adatom hybridization and Co/molecule vertical
displacements appear as the most plausible causes. Whatever the reason may be, we
hereby demonstrate that the effect of the 2DEG renormalization is a rather general
effect for certain single metal atom coordinated nanoporoous networks.

6.5.1 Sample Preparation for ARPES
A circular 15 mm Au(111) sample was in situ prepared using standard sputtering

and annealing cycles in the preparation chamber (base pressue ∼ 2 × 10−10 mbar).
Ph3 and Ph6 molecules were evaporated from a Knudsen cell-type home-built cru-
cible evaporator at 510 K and 670 K respectively. The pressure in the preparation
chamber was kept below 10−9 mbar with the sample at 150 K. The molecular depo-
sition was calibrated then by measuring the onset of the Au SS in ARPES. The Au SS
tends to broaden and shift towards the Fermi level as the molecular coverage is in-
creased and it can even quench for close to 1 ML coverages. This trend is illustrated
in Figure 6.6. Since the formation of Ph3Co and Ph6Co requires a particular 3:2
stoichiometry (molecule/Co), the Co deposition was performed after the molecular
vertical gradient deposition using a commercial metal evaporator with a Co rod. For
the latter case, the gradient was performed in the horizontal axis (orthogonal to the
molecular one). Then the sample was slightly annealed to 400 K in order to enhance
the surface mobility of Co and molecules and promote the coordination. Once the
right coverage was calibrated, we measured the ARPES bands looking for the key
features of a partially confined 2DEG, e.g. shift of the band bottom (generally with
molecular overlayers towards the Fermi level), opening of gaps at zone boundaries
(depends on the lateral scattering strength), changes in the dispersion and replicat-
ing bands due to the superperiodicity of the newly formed SBZ.
6.5.2 STM/STS Measurements
The experiments were carried out within the group of Prof. Meike Stöhr (Uni-
versity of Groningen, Netherlands) in a two-chamber UHV system (base pressure
of 4 × 10−11 mbar) housing a commercial LT-STM instrument (Scienta Omicron
GmbH). The Au(111) substrate was cleaned by repeated cycles of Argon ion sputter-
ing followed by annealing at 800 K. The Ph3 (Ph6) molecules were heated to 445 K
(550 K) inside a commercial molecule evaporator (OmniVac) and deposited onto the
Au(111) substrate held at room temperature. The Co atoms were deposited with an
electron beam evaporator (Oxford Applied Research Ltd). STM measurements were
performed at both 77 K and 4.5 K with a mechanically cut Pt/Ir wire in constant
current mode. All bias voltages are given with respect to a grounded tip. The STM
images were processed with the WSxM software [174]. STS measurements were per-
formed at 4.5 K by using a lock-in amplifier (typical modulation parameters used:
amplitude of 10 mV (rms) and frequency of 677 Hz).
90
Dot Arrays
6.5.3 Confined State Tunability with Pore Size
In addition to the gradual renormalization process observed for Ph6Co and Ph3Co
networks on Au(111), confined state tunability with pore size should also be ex-
pected, as already reported in [17]. Figure 6.8 summarizes such confined state tun-
ability, which is best illustrated in the EBEM/EPWE simulated LDOS curves and
maps. While in Ph6Co the experimental and EBEM/EPWE LDOS curves clearly
show the appeareance of confined state peaks, Ph3Co shows flatter spectra, mak-
ing the experimental observation of such localized states more difficult. This hap-
pens because gaps between states are more separated in energy for the smaller pore
(Ph3Co) than for Ph6Co, thus, DOS features become more planar. As the pore size
in enlarged (Ph6Co), confined states pile up in a smaller energy range and become
more visible [17, 133, 166].
n=1
dI/dV (arb. units)
n=2 n=4
n=1 n=4
n=2
Ph3Co C (STS)
Ph3Co H (STS)
Ph6Co C (STS)
Ph6Co H (STS)
Ph3Co C (EBEM)
Ph3Co H (EBEM)
Ph6Co C (EBEM)
Ph6Co H (EBEM)
-0.6
-0.6 -0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 0.6
0.6
F IGURE 6.8: Ph6Co and Ph3Co confined state energy tunability is illustrated by
comparing STS spectra and EBEM/EPWE simulated LDOS. Confined states are
better defined and observed for the largest pore size, Ph6Co. For the Ph3Co case,
the DOS adopts a much flatter configuration due to the wider energy separation be-
tween energy gaps. However, EBEM/EPWE simulated spatial LDOS maps clearly
show that confined states are sensitive to the pore size, spreading more in energy
for the Ph3Co case than the Ph6Co case.
6.5.4 EBEM/EPWE Simulations
For the band structure and LDOS simulations it is very important to follow the ge-
ometry of the scattering barriers. The periodicity of the networks (3.53 nm for Ph3Co
and 5.78 nm for Ph6Co) and size of molecules (5 Å width) are used in order to build
the geometry in EBEM/EPWE (Figure 6.1). Then a repulsive scattering potential is
assigned to the molecular region (Vmolecule = 250 meV). Note that in ref. [17], a higher
repulsive potential was used (V = 500 meV) with a narrower molecular width (2.5 Å)
but still yielding the same V ·d effective potential. In addition, we keep the scattering
potential at the Co sites as slightly repulsive (VCo = 50 meV), contrarily to previous
published works [16, 17, 129]. The reason for this is that the assigment of an attrac-
tive character to the Co region has never been clarified in previous works, hence, we
challenge such an assumption. Indeed, we systematically change the scattering po-
tential landscape of Ph6Co (Figure 6.9) in four different scenarios: Ph6 molecules and
Co region with the same repulsive scattering character and amplitude [Figure 6.9 (a,
e)], Co regions with a less repulsive character than the Ph6 molecules [Figure 6.9 (b,
f)], repulsive scattering character for Ph6 and slightly attractive for Co as in ref. [17]
[Figure 6.9 (c, g)] and the same attractive scattering potential amplitudes for both
Ph6 molecules and Co regions [Figure 6.9 (d, h)].
0.0
0.0 0.0 0.0 0.0
a b c d
-0.1
-0.1 -0.1 -0.1 -0.1
E-EF (eV)
-0.2
-0.2 -0.2 -0.2 -0.2
-0.3
-0.3 -0.3 -0.3 -0.3
-0.4
-0.4 -0.4 -0.4 -0.4
-0.5
-0.5 -0.5 -0.5 -0.5
G M K G G M K G G M K G G M K G
0.00 0.04 0.08 0.12 0.16 0.00 0.04 0.08 0.12 0.16 0.00 0.04 0.08 0.12 0.16 0.00 0.04 0.08 0.12 0.16
0 0.04 0.08 0.12 0.16 0
Vmol=+250mV; Vcobalt=+250mV 0.04 0.08
Vmol=+250mV; 0.12 0.16aaa 0
Vcobalt=+50mV 0.04 0.08 0.12 0.16 0
Vmol=+250mV; Vcobalt=-50mV 0.04 0.08 0.12 0.16
Vmol=-250mV; Vcobalt=-250mV
kII (Å-1) kII (Å-1) kII (Å-1) kII (Å-1)
Vmol = +250 meV Vmol = +250 meV Ph6Co Vmol = +250 meV Vmol = -250 meV
Vcobalt = +250 meV Vcobalt = +50 meV /Au(111) Vcobalt = -50 meV Vcobalt =- 250 meV
e f g h
dI/dV (arb. units)
-0.6 -0.4 -0.2 0.0 0.2 -0.6 -0.4 -0.2 0.0 0.2 -0.6 -0.4 -0.2Vcobalt=-50mV
0.0 0.2-0.6 -0.4 -0.2 0.0 0.2
-0.6 -0.4 -0.2
Vmol=+250mV; 0.0
Vcobalt=+250mV 0.2 -0.6 -0.4 -0.2
Vmol=+250mV; 0.0
Vcobalt=+50mV 0.2 -0.6 -0.4 -0.2
Vmol=+250mV; 0.0 0.2 -0.6 -0.4 -0.2
Vmol=-250mV; 0.0
Vcobalt=-250mV 0.2
Tunneling Voltage (V) Tunneling Voltage (V) Tunneling Voltage (V) Tunneling Voltage (V)
F IGURE 6.9: EBEM/EPWE band structure and LDOS simulations on Ph6Co using
four repulsive/attractive potential combinations. (a-d) Band structure simulations
for different scattering potentials (Vmol and Vcobalt ). (e-h) LDOS simulations at
different positions [pore center (red), molecule (purple) and cobalt (green)] using
the same scattering potentials as in (a-d). The experimental spectrum at the pore
center (in grey) is to be compared with the red LDOS.
92
Dot Arrays
The electronic band structure for the first three cases are identical [Figure 6.9 (a,
b, c)] and no effect is observed by changing the scattering character of the Co region
from the expected repulsive character to the attractive one. LDOS simulations in Fig-
ure 6.9 (e, f, g) also nicely match the experimental findings at the pore center (com-
pare red spectrum with experimental gray one). In addition, spectral features at the
Ph6 molecular backbone (in green) and in the Co region (in purple) are practically
identical, coinciding with the experimental observation reported in ref. [17]. Finally,
we consider a totally attractive scattering barrier situation [Figure 6.9 (d, h)]. For this
particular case, we immediately observe that the K point gap in the electronic band
structure is closed (dashed red circle), resembling a graphene-like behavior. This can
be explained with an inversion of the scattering barrier in the molecular honeycomb
lattice. While for the previous three examples, surface electrons are scattered by
molecules and Co atoms, which act as barriers, now the situation is reversed, elec-
trons are delocalized trough the molecule and Co regions (in green and purple) and
are scattered by the Au surface (red region). This situation resembles an artificial
graphene structure, a situation that has been already experimentally studied [204]
and recently used in order to model graphene-like structures by EPWE [34]. Such
a scattering barrier inversion also produces clear effects on the LDOS at the pore
center, where the red spectrum no longer matches the experimental observations
[Figure 6.9 (h)].
All the aforementioned observations guide us towards considering both molecu-
lar and metal coordination regions as repulsive scattering potentials, finding no ev-
idence for assigning an attractive character to the cobalt region. These observations
are further discussed and supported in Chapter 7, concluding that both molecules
and coordination metal atoms represent repulsive scattering potentials (barriers) for
surface electrons.
6.5.5 The Phase Accumulation Model
The phase accumulation model provides a good estimate of the surface state energies
by setting the constrain on the total phase accumulated by electrons reflected at the
crystal (ΦC )-vacuum (ΦB ) interface to produce constructive interference. That is:
ΦC + ΦB = 2πn (6.1)
where n is an integer and n = 0 is a solution for surface states [205]. The empirical
energy-dependence forms of ΦC and ΦB are [205, 206]:
r
E − EL
ΦC = 2 arcsin (6.2)
EU − EL
and r !
3.4 eV
ΦB = π −1 (6.3)
EV − E
where EL , EU and EV , are the lower and upper edges of the crystal bulk gap along
ΓL direction and the vacuum energy, respectively, and their reported values for the
three noble metals are given in Table 6.3.
TABLE 6.3: The parameters used in PAM calculations. Values have been taken from
ref. [124].
EL (eV) EU (eV) EV (eV) d0 (Å)

Ag(111) -0.30 4.00 4.70 2.361
Cu(111) -0.91 4.25 4.94 2.087
Au(111) -1.05 3.50 5.30 2.361
The vertical displacement of the outermost surface layer (d∗ = d0 ± ∆d), when
considered [see inset in Figure 6.5 (b)], contributes a new phase (Φ∗ ) which is ac-
quired by electrons traveling the corresponding optical path [207, 208]. In this case,
the form of the accumulated phase can be written as:
ΦC + ΦB + Φ∗ = 2πn (6.4)
where Φ∗ = 2k⊥ d∗ considering the outermost layer as a quantum well, and k⊥ stands
for the electron momentum perpendicular to the surface [207, 209]. Assuming k⊥
equals to the fundamental π/d0 wave-vector, for the neutral interplanar distance
(d0 ), Φ∗ = 2π, implying that the solution for the surface state is found at n = 1. The
value of d0 for the three noble metals is given in Table 6.3. Solving Equation 6.4 for
different d∗ we obtain the surface state energy as a function of the vertical displace-
ment of the metallic overlayer, as shown in Figure 6.5 (b) for the three noble metals.
Indeed, for d∗ > d0 the surface state shifts downwards until it meets the lower edge
94
Dot Arrays
of the projected bulk bands. For d∗ < d0 , on the other hand, the surface state shifts
to higher energies.
6.5.6 DFT Calculations
Density functional theory calculations have been performed by Prof. A. Arnau and
I. Gallardo (CFM, Spain) using the VASP code [210–212]. The interaction of the va-
lence electrons with the ion cores was described with the projector augmented wave
method (PAW) and the Perdew-Burke-Ernzerhof [213] exchange-correlation func-
tional was used. An energy cutoff of 300 eV in the plane wave expansions and
different K-point samplings, depending on the size of the surface unit cells, were
employed and checked in order to achieve good convergence. A four layer Au(111)
slab with hydrogen atoms passivating one of the two vacuum-metal interfaces was
used to model the Au(111) surface state [214], while a Co atom was placed on the
other interface at 2.5 Å vertical distance at a fcc hollow site. At this distance the
Co-Au(111) interaction is weaker than for the optimal adsorption distance but it is
the way we mimic a less reactive Co atom in the metal-organic network. Table 6.2
summarizes the effects that an array of isolated Co adatoms deposited on Au(111)
at different concentrations have on the reference energy of the Au SS. A more pro-
nounced downward shift of the Au SS is observed with increasing Co content.
95
Chapter 7
Effective Determination of Surface

Potential Landscapes from
Metal-Organic Nanoporous
Overlayers
7.1 Introduction
Determining the scattering potential landscape for two-dimensional metallic super-
lattices is key to understand fundamental quantum electron phenomena. Theoretical
and semiempirical methods have been extensively used to simulate confinement ef-
fects on superlattices with a single scatterer in the form of vicinal surfaces [103] and
dislocation networks [215] or single structures such as quantum corrals and vacancy
islands [135, 136, 141]. However, the complexity of the problem increases when
the building blocks (or scatterers) are heterogeneous, as in metal-organic networks,
since additional potentials may come into play.
The modelization of electron scattering by 2D arrays (organic and metal-organic
nanoporous networks) is often performed using EBEM/EPWE, which accounts for
the local confinement response (local density of states) [16, 17, 129] and the electronic
band structures arising from Boch-wave states generated from interdot coupling [15,
54, 129, 160]. However, it is a semiempirical method, so some assumptions need
to be made regarding the potential barrier strength, sign (repulsive/attractive) and
geometries, which often lead to arbitrary or unphysical conditions. In particular,
published works use thinner molecular geometries than the actual molecular back-
bones [16, 17], altered effective pore sizes [17, 160], attractive scattering potential
regions at the metal sites in metal-organic systems [16, 17, 129] or enlarged effective
masses (m∗ ) than the 2DEG reference [54, 129]. The evident question is whether such
ambiguous assumptions are necessary when simulating the electron confinement by
2D nanoporous networks.
Here, we answer this question by studying in depth the interaction between
the 3deh-DPDI metal-organic nanoporous network [15, 53, 129, 158, 164, 167, 168,
216] and the two-dimensional electron gas of Cu(111). Based on a combination of
Chapter 7. Effective Determination of Surface Potential Landscapes from
96
Metal-Organic Nanoporous Overlayers
STM/STS, ARPES and Kelvin probe force microscopy (KPFM) measurements, to-
gether with EBEM/EPWE simulations, we demonstrate that both local confinement
effects and interpore coupling induced electronic bands can be satisfactorily repro-
duced. This is achieved by starting from a realistic scattering geometry and then
assigning physically sound repulsive scattering potentials to its barriers (both for
molecules and adatoms). However, we require a 2DEG renormalization that afffects
the effective mass (m∗ ) and energy reference (ERef ), that agrees with our observa-
tions of previous chapters. Our explaination for these changes is assigned to the
alteration of the vacuum region upon network presence, which is known to partly
define the Shockley state [122, 201]. This experimental-theory synergy enables us to
capture the intricacies of the scattering potential landscape, and to reveal systematic
modeling procedures. Note that KPFM measurements have been performed by Dr.
Shigeki Kawai from MANA/NIMS (Japan), STM/STS by Dr. Sylwia Nowakowska
from Basel University (Switzerland) and EBEM/EPWE simulations in close collabo-
ration with Dr. Zakaria M. Abd El-Fattah from Al-Azhar University (Egypt).

The organometallic network that we study is the same as the one presented in Chap-
ter 4. It is formed by DPDI (4,9-diaminoperylene quinone-3,10-diimine), which un-
dergoes a triple dehydrogenation process when deposited on Cu(111) and heated to
250◦ C. These organic tectons coordinate with Cu adatoms, and form a long-range
ordered, commensurate metal-organic nanoporous network [15, 53, 158, 164, 167,
168, 216] [see Figure 7.1 (a)]. As indicated previously, the unit cell is composed of
3 molecules and 6 Cu adatoms and bears a periodicity of 25.5 Å. The model shown
in Figure 7.1 (b) highlights the three-fold symmetry of 3deh-DPDI network. The
three-fold symmetry of this network arises from the different registry of the Cu co-
ordination atoms with the surface: One type of Cu trimer (i) surrounds an on top
site, while the other Cu trimer (ii) surrounds a hollow site [167, 216].
a b
i ii Cu
N
C
H
ii i
2 nm i ii
F IGURE 7.1: 3deh-DPDI metal-organic network on Cu(111). (a) STM image of the
long-range ordered metal-organic network. STM parameters: I= 10 pA ; V= -1V. (b)
3deh-DPDI network model exhibiting the three-fold symmetry due to variations in
the registry of the metal linkers.
a b c
(-0.186 eV) (0.016 eV) (0.385 eV)
EBEM Model 1
Int. (arb. units)
EBEM Model 1 PC
H
Cu
M
1 nm 1 nm 1 nm 1 nm
d e f -0.4 -0.2 0.0 0.2 0.4
(0.385 eV)
Int. (arb. units)

(-0.186 eV) (0.016 eV)
STM/STS
STM/STS
PC
H
Cu
1 nm 1 nm 1 nm 1 nm M
g h i -0.4 -0.2 0.0 0.2 0.4

(-0.206 eV) (Fermi) PC
Int. (arb. units)

(0.49 eV)
EBEM Model 2
EBEM Model 2
H
Cu
M
1 nm 1 nm 1 nm 1 nm
-0.4
-0.4
-0.2
-0.2
0.0
0.0
0.2
0.2
0.4
0.4
Voltage (V)
F IGURE 7.2: EBEM/EPWE simulated and experimental LDOS for the 3deh-DPDI
metal-organic network. The top (a-c) and bottom (g-i) rows correspond to two dif-
ferent EBEM/EPWE simulations that we compare to the experimental case shown
at the center row (d-f). This STM/STS row consists of a topographic image (d), three
dI/dV maps at constant height at different energies [-0.186 eV (n=1 PLS), 0.016 eV
(n=2 PLS) and 0.385 eV (n=4 PLS)] (e) and four dI/dV spectra acquired at relevant
positions of the unit cell (see inset) (f). All experimental data can be vertically com-
pared to two EBEM/EPWE models: the top reproduces the one in ref. [129] and the
bottom is a new one proposed in this work. The scattering geometries shown in (a)
and (g) consist of three parts: Cu substrate area (shown in red), molecules (shown
as rectangles in purple or light blue) and metal coordination regions (as hexagons
in green or violet/yellow). The corresponding potential values are indicated in
Table 7.1. Experimental parameters: dI/dV maps and dI/dV spectra obtained from
a grid spectroscopy measurement (35 × 30 points) with initial tip conditions 400
mV/70 pA and lock-in frequency 513 Hz; zero-to-peak amplitude: 8 mV. A binning
of 30 meV is applied to each dI/dV map.
Figure 7.2 shows this network and compares the experimental STM/STS data
(center row) with two EBEM/EPWE simulations obtained using the scattering ge-
ometry shown in panels (a) and (g) and barrier potentials specified in Table 7.1. The
top row in Figure 7.2 assumes a scattering geometry proposed in ref. [129], which
is formed by hexagons at the Cu adatom coordination sites (in green) connected
by molecular potentials of 5.5 Å wide (in purple). In this model the metal adatom
regions dominate spatially over the molecules and weakly resembles the STM to-
pography of Figure 7.2 (d) and network model in Figure 7.1 (b). Furthermore, the
potential at the Cu coordination region presents a negative sign, denoting attrac-
tive scattering character to this region, similar to other simulated coordinated metal-
organic networks [16, 17]. Using the 2DEG reference of ERef = -440 meV (at 5 K) and
m∗ = 0.44 m0 (slightly larger than the pristine substrate), we find in the simulations
a very good match for the n=1 partially localized state (n=1 PLS) dI/dV map, but
this agreement is lost for the higher orders [vertically compare Figure 7.2 (b) and
(e)]. Note, for instance, that the LDOS central intensity is reversed for the highest
dI/dV map shown. Similarly, when comparing the dI/dV spectra at selected unit cell
98
TABLE 7.1: EBEM/EPWE potential parameters used in the models discussed

throughout this chapter. Vmol refers to space occupied by the molecule and Vmet
to the metal coordination centers, which are respectively shown as purple or light
blue rectangles, and as green or violet/yellow hexagons in Fig. 7.2 (a) and (g). The
large red areas corresponding to the Cu substrate are fixed to Vsubs = 0 eV. The
2DEG scattering reference (ERef at 5 K and m∗ ) used for the EBEM/EPWE simula-
tions is indicated in the last two columns. The first model is taken from ref. [129],
the second one is generated from a realistic geometry of the network, whereas the
third one is identical to the second model, but differentiates the two types of metal
coordination centers discussed in Fig. 7.3.
Vmol (meV) Vmet (meV) ERef (meV) m∗ /m0

Model 1 ref. [129] 1500 -100 -440 0.44
Model 2 390 390 -440 0.49
Model 3 390 440 (Type i) -440 0.49
340 (Type ii)
positions [Figure 7.2 (c, f)], we find that only the pore center spectrum (PC, in red)
acceptably matches the experimental curves, but the rest deviate strongly when the
bias exceeds -0.12 V. Firstly, at the halfway (H, in blue) spectrum, the dominant peak
at 0.32 V is experimentally absent. Secondly, for the metal center (Cu, in green), two
prominent peaks are observed at -0.09 V and 0.18 V and show a drop of the intensity
at the right side of the simulation, which are absent in the experimental data. In
particular, the strong localization at the Cu atoms for the 0.18 V state, with LDOS as
large as the fundamental n=1 PLS, is farthest from experiment, indicating unplysical
potential assignment for this region.
The pronounced discrepancies between the experimental data and the EBEM/EPWE
model generated with the parameters of ref. [129] demands the reconsideration of
the scattering potential and geometries in these simulations. Based on previous high
resolution AFM and STM work [167, 216] and the network model of Figure 7.1 (b),
we start from a realistic geometry where the repulsive nature of molecules domi-
nate, instead of the metal coordination regions [cf. Figure 7.2 (g)]. This leads to a
molecular geometry of 9 Å in length and 8 Å in width and much smaller metal co-
ordination regions that realistically fill the space left between molecules. Instead of
allowing this geometry to change, we provide flexibility to the 2DEG reference [54].
We find that using ERef = -440 meV (for 5 K) and m∗ = 0.49 m0 , which notably de-
viate from the pristine Cu(111) surface state [100, 108], together with homogeneous
repulsive barriers (Vmol = Vmet = 390 meV), we can finely capture most details of the
experimental LDOS and dI/dV spectra. Note that these are not just arbitrary values
(scattering parameters indicated as Model 2 in Table 7.1), but the result of simultane-
ously simulating the ARPES and STS experimental data [54]. These parameters turn
out to be crucial for pushing to lower energies the higher order confined states (for
instance lowering n=4 PLS to ∼ 0.49 V) [Figure 7.2 (h)]. Interestingly, the dI/dV spec-
trum on top of the metal adatom region (green curve) shows a relevant attenuation
of the spectral features above -0.1 V that agrees very well now with the experimen-
tal curves and corroborates the repulsive character of this region of the network [cf.
Figure 7.2 (f, i)].
We further backup the choice of repulsive scattering parameters for both molecules
and metal centers by performing Local Contact Potential Difference (LCPD) mea-
surements using the Kelvin probe force microscopy method [217, 218]. The KPFM
constrast is known to correspond qualitatively to the z-component of the electro-
static field Ez at a constant-height plane above the molecule. The simple interpreta-
tion is that a more positive LCPD corresponds to more negative charge below the tip.
Even though the tip radius is expected to be much larger than the local variations
of the surface potential, in Figure 7.3 (a), we observe that the molecular scattering
backbone (in red) is similar in size to the molecular dimensions [216]. In addition,
n2 (16meV average 30meV)
a 0.12 b c (0.016 V)
0.1
0.08
0.06
Volts (V)
0.04
0.02
0.0
-0.02
-0.04
1 nm -0.06 1 nm
-0.08
Vmol=0.39eV, V_Cu1=0.366eV;V_Cu2=0.415eV Vmol=0.39eV, V_Cu1=0.29eV;V_Cu2=0.49eV
Vmol=0.39eV, V_Cu1=0.39eV;V_Cu2=0.39eV Input3 Input2
d Fermi e Fermi f Fermi

max.
1 nm 1 nm 1 nm min.
ΔVmet = 0 meV ΔVmet = 50 meV ΔVmet = 200 meV
F IGURE 7.3: Corroboration of three-fold symmetry at metal coordination positions,

influence upon the confined states and EBEM/EPWE implementation. (a) Kelvin
probe force microscopy image of several pores showing local contact potential dif-
ference variations at the sub-nanometer scale confirming the existence of surface
potential variations upon the formation of the molecular network. (b) Noncontact
AFM measurements (adapted from ref. [216]) with sub-molecular resolution differ-
entiating the two types of metal coordination centers (denoted i and ii). (c) dI/dV
constant height map for the n=2 PLS (slightly above EF ) indicating the different
metal adatom regions as two colored circles (purple and green). The sensitivity to
the three-fold symmetry potential is demonstrated by the triangular shape (instead
of hexagonal) observed inside the pore. The LDOS asymmetry can be simulated in
EBEM/EPWE as a perturbation of the scattering potential. In particular, (d) shows
the unperturbed case (hexagonal case) when ∆Vmet = 0 eV (Vmet = 0.39 eV), whereas
(e) and (f) introduces potential variations of ∆Vmet = 50 meV and ∆Vmet = 200 meV,
respectively. The green metal coordination position has a higher potential value
than the violet one and, correspondingly, the distortion becomes stronger the larger
the difference between the two sites is.
100
the smaller Cu coordination regions, which appear less intense than the molecules
most probably due to their different height [167], do not drastically change its color
when compared to the pore regions, ruling out any tendency towards a sign rever-
sal of the scattering potential and therefore maintaining a repulsive character to the
surface electrons. Note however, that the LDOS simulation shown in Figure 7.2 (h)
displays a six-fold symmetry that contradicts previous experimental findings report-
ing the existence of two structurally different adatom coordination sites within the
network [167, 216]. This difference is explained as a variation of the registry with
the substrate, where the three metal adatoms binding the molecules of the network
arrange either surrounding a three-fold hollow site or a Cu atom of the substrate [see
Figure 7.1 (b)]. This subtle variation shows up in the AFM images acquired with a
CO functionalized tip [see Figure 7.3 (b)] and in the dI/dV maps whenever the inten-
sity shifts away from the pore center, as is the case of n=2 PLS [cf. Figure 7.3 (c)]. It is
worthy to note that, this is the first experimental observation of the triangular shape
of n=2 PLS, although they were predicted for hexagonal lattices with three-fold sym-
metry [163]. EBEM/EPWE can simulate this behavior by inducing a slight potential
variation (∆Vmet ) between Cu regions while keeping the molecular potential at Vmol
= 0.39 eV (Model 3 in Table 7.1). In this way, we observe a clear transition of the n=2
PLS from hexagonal to triangular, as shown in the bottom row of Figure 7.3 (d-f).
We find that the green regions correspond to a stronger potential than the purple
regions. Likely, the two types relate to a subtle energy imbalance stemming from
its vertical packing. Following the herringbone reconstruction potential difference
between fcc and hcp sites of 25meV [219], we tentatively assign the green sites to
larger potentials (fcc) and the purple to lower ones (hcp).
The synergy between our local spectroscopies and EBEM/EPWE simulation has
produced two relevant results: first, the metal coordination sites exhibit repulsive
scattering potential character compatible with the presence of three-fold symmetry
in the higher energy states and, second, a mass renormalization of the 2DEG occurs
upon the presence of the network on the Cu(111) surface. These findings should be
compatible with the ARPES data shown in Figure 7.4. This situation corresponds to
the one in which 3deh-DPDI network completly saturates the surface and no trace of
the Cu SS is observed (see Figure 7.7). It is straightforward to confirm that the 2DEG
renormalization is correct since both the simulated band structure and the isoener-
getic cuts fit exceptionally well to the experimental data when using the scattering
parameters indicated in Model 2 of Table 7.1. In Figure 7.4 (a), the second deriva-
tive of the experimental data (raw data in Figure 7.8) exhibits the expected shallow
dispersive bands of organic QD arrays [15, 54, 129]. The lower energy band corre-
sponding to n=1 PLS has a ∼ 80 meV bandwidth and shifts ∼ 150 meV towards
EF with respect to the pristine Cu surface state and increases m∗ to ∼ 0.58 m0 [15,
53, 54]. This effective mass is higher than the 2DEG reference as a result of the con-
finement induced by the nanoporous network [54], which is certainly substantial
DPDI Soleil 2Der GM reduced x DPDI Soleil 2Der GK reduced x

a 0.1
0.1 0.1
max.
0.0
0.0 0.0
E-EF (eV)
-0.1
-0.1 -0.1
-0.2
-0.2 -0.2
-0.3
-0.3 -0.3
-0.4
-0.4 -0.4
-0.5     -0.5        min.
-0.5
-0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4 -0.4
-0.4
0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4
kx (Å-1) ky (Å-1)
b 0.4
-0.24 eV -0.13 eV -0.07 eV Fermi
0.2
Experiment
ky (Å-1)
0.0
-0.2 max.
-0.4
c 0.4
-0.24 eV -0.14 eV -0.07 eV Fermi
0.2
ky (Å-1)
min.
EBEM
0.0
-0.2
-0.4
-0.2 0.0 0.2 -0.2 0.0 0.2 -0.2 0.0 0.2 -0.2 0.0 0.2
kx (Å-1)
F IGURE 7.4: Experimental and simulated electronic band structures and isoener-
getic maps of the 3deh-DPDI network saturating the Cu(111) surface. (a) Energy
dispersion maps for the two high symmetry directions ΓM and ΓK. The color-
plot represents the second derivative of the intensity to enhance the weak de-
tails with respect to energy and parallel momentum. The high quality of the net-
work allows to observe faint replica bands in adjoining Brillouin zones. Simulated
EBEM/EPWE bands using the scattering potentials of Model 2 (see Table 7.1) are
superimposed as red lines onto the experimental data and match perfectly both
high symmetry directions. Experimental (b) and simulated (c) isoenergetic cuts (kx
vs ky ) obtained at the band-bottom (-0.24 eV), lower edge of the M point (-0.13 eV),
inside the gap (-0.07 eV) and at the Fermi level. As a reference, the metal-organic
network induced hexagonal SBZ is superimposed onto the second derivative data.
Experimental ARPES parameters: hν = 21.22 eV, Ts = 10 K for (a) and 150 K for (b).
judging from the prominent energy gap (120 ± 30 meV) detected at the zone bound-
aries (M point) separating the n=1 and n=2 PLS bands [157] (see Figure 7.9).
However, the three-fold symmetry observed in the LDOS for the n=2 PLS is
absent in the fermi surface map of Figure 7.4 (b). Instead, all isoenergetic cuts (kx
vs ky ) exhibit six-fold symmetry (hexagonal shape) that diverges from the circular
and isotropic pristine Cu surface state. It could be argued that due to the averaging
character of ARPES, the illuminated (probed) area contains, with equal probability,
network patches with (i) and (ii) metal coordination regions at equivalent pore sites
102
(60◦ relative rotations). Nonetheless, the primary reason behind is the conservation
of time reversal symmetry, which requires E(k) = E(−k). Hence, isoenergetic cuts
should appear as six-fold even if a single three-fold symmetric 3deh-DPDI network
domain was present on the surface. This is in agreement with other three-fold sym-
metric surface structures such as Ag/Cu(111) superlattices, where the band structure
also appears as six-fold symmetric [215]. Indeed using the scattering parameters of
both Model 2 and Model 3 in Table 7.1, the electronic band structure agrees with
the band structure dispersion, gap size and isoenergetic shapes experimentally ob-
served. Thus, the nice agreement between theory and experiment supports and val-
idates the repulsive scattering character assigned for molecules and metals, as well
as the imposed renormalization of the 2DEG reference.
Our results prove that an appropriate EBEM/EPWE simulation of the elec-
tron confinement by QD arrays should be achieved by having at hand STM/STS
and ARPES experimental data and using realistic geometries of the overlayer before
a Overlayer Nanoporous Network

G
0
0.0
Cu111
New Reference
-100
-0.1 CS_bottom
E-EF (eV)
CS_bottom_r
CS_bottom_rr
-200
-0.2 CS_up
CS_up1
CS_up2
-300
-0.3
CS_up3
-400
-0.4 G M G M G M G M
Pristine SS Renormalized SS Renormalized + Lateral Scattering
b 200
-200
Gr/hBN
Gr/Cu(111)
Gr/Ag(111)
Gr/Au(111)
hBN/Cu(111)
150
-150 Xe/Cu(111)
│EB-EB0 │ (meV)
Ar/Cu(111)
Xe/Au(111)
Xe/Ag(111)
Ar/Ag(111)
Kr/Cu(111)
100
-100
Kr/Ag(111)
3deh-DPDInetw./Cu(111) ARPES
3deh-DPDInetw./Cu(111) EBEM
SW netw.3MLAg/Au(111) ARPES/EBEM
Rare Gases DW netw.3MLAg/Au(111) ARPES
50
-50
DW netw.3MLAg/Au(111) EBEM
Ph3Conetw./Au(111) EBEM
Ph6Conetw./Au(111) EBEM
Nanoporous Networks
00
00 20
20 40
40 60
60 80
80
│(m*-m0*)/m0*│(%)
F IGURE 7.5: Reported renormalization of 2DEGs by physisorbed overlayers. (a)

Schematic band renormalization of the Cu Shockley state resulting in an in-
crease of its reference energy and effective mass upon the presence of the stud-
ied nanoporous network. (b) Renormalization chart (∆EB vs ∆m∗ ) reported for
families of physisorbed systems onto substrates exhibiting 2DEGs. Nanoporous
networks (yellow area) [53, 54] are compared with rare gas overlayers (orange
area) [220, 221] and with graphene [222, 223] and h-BN [224] (blue area). Note
that m∗ renormalization becomes relevant only for nanoporous networks.
defining the strength of the scattering potential barriers. Moreover, the 2DEG refer-
ence requires certain flexibility, especially in m∗ , in agreement to previous work [54,
129]. The 2DEG reference changes upon the growth of the organic overlayer, as
schematically shown in Figure 7.5 (a). The resulting (leaky) confinement depends on
the detailed potential landscape exerted by the nanoporous network on the 2DEG,
defining the interpore coupling and producing new electronic bands separated by
gaps at zone boundaries [54, 160].
The 2DEG renormalization here proposed is not uncommon and has been ob-
served in numerous physisorbed overlayers, such as graphene (Gr), boron-nitride
(hBN) and rare gases physisorbed on noble metals. Figure 7.5 (b) compiles se-
lected results from these related systems where the surface state shifts towards the
Fermi level while varying its m∗ . It is interesting that even if the energy shifts can
be of similar magnitude, the effective masses do considerably increase in the case
of nanoporous systems. This is likely caused by the QD intercoupling extending
through the surface. Note that absolute values have been used to also include Ph3Co
and Ph6Co network results (Chapter 6).
Once the 2DEG reference modification has been justified, the scattering poten-
tial values call for attention. When comparing the molecular potential Vmol obtained
in this work (see Table 7.2) we find that its value (390 meV) is larger than other re-
ported cases with ARPES experimental data access [54]. The 3deh-DPDI generates
potential barriers capable of strongly confining the surface electrons and opening
significant energy gaps in the band structure. Nevertheless, the magnitude of the
value is reasonable when compared to other networks, suggesting that limitations to
the potential magnitude should exist when used to simulate the molecular scattering
max ∼ 0.5 eV.
potentials. As an educated guess, this limit could be of the order of Vmol
Note however that ultimately the effective potential that 2DEG electrons encounter
depends on the separation between the nanopores, i.e. the molecular widths. The
value here proposed would also be restricted to planar and single layer molecular
systems.
Furthermore, we demonstrate that the scattering character of the adatom metal
centers is repulsive for the 3deh-DPDI network. This follows a similar pattern to
TABLE 7.2: Scattering potential values for molecules (Vmol ) and metal centers (Vmet )
for three different types of networks. The selected networks correspond to the
one from this study (top row), the homothetic Ph6Co and Ph3Co metal-organic
networks on Au(111) (middle, Chapter 6) and the two halogen bonded SW and
DW networks on Ag(111) (bottom row, Chapter 5). The values are taken from
EBEM/EPWE simulations based on STS and ARPES datasets.
Vmol (meV) Vmet (meV)

3deh-DPDI network 390 340 / 440
Ph3Co / Ph6Co 250 50
SW / DW (ref. [54]) 140 –
104
molecular backbones and to other atomic-like surface 2DEG confining entities in

the form of step-edge adatoms, quantum-corral barriers or dislocation networks [18,
103, 108, 135, 225]. We infer that this effect is quite general and has in the past
been incorrectly considered for the case of organometallic nanoporous networks [16,
17, 129]. Indeed, the repulsive potential values proposed for Ph3Co and Ph6Co in
Table 7.2 and discussed previously in Chapter 6 are more physicaly sound and in
agreement with the present observations.
It is interesting to discuss further the implications of the observed LDOS devi-
ation from the hexagonal shape induced by scattering potential imbalance at the
metal sites. The local three-fold appearance of the higher order n=2 PLS comes not
solely as variations of the metal center registry with the substrate (as shown here in
Figure 7.3), but can also be generated from small relative displacements of the build-
ing blocks defining the pores. An example of the latter has been already reported for
chiral networks stabilized by non-metallic intermolecular bonds [153, 163]. Accord-
ingly, it is the ultimate geometry of the tectons that dictates the LDOS shape if the
scattering potential barriers have similar magnitudes.
Finally, it is plausible that the confined states at the pores could play a role in
guest filling of QD arrays through minimization of the electronic contributions [139,
163, 166]. A way of confirming this hypothesis would require the observation of cer-
tain symmetries imposed on the external species adsorbed into the QDs. In particu-
lar, the 3deh-DPDI nanopores can host up to 12 Xe atoms per pore that are clustered
into four atom bunches [164] [Figure 7.6 (left)]. Since these rare gas atoms maintain
exactly the same relative bunch orientation to all neighboring pores in an overall
three-fold symmetry [164], we infer that the n=2 PLS (closest to the EF ) could par-
ticipate in guiding the observed Xe pore condensation [Figure 7.6 (right)]. Alterna-
tively, these Xe atoms could be just marking three equivalent metal sites exhibiting
the lowest surface potential of the whole network [226].
n2 (16meV average 30meV)
(0.016 V)
1 nm 1 nm
F IGURE 7.6: Confined state influence upon the Xe adsorption. Up to 12 Xe atoms

adsorbs into the nanopores of 3deh-DPDI, as observed from the STM image ac-
quired with a Xe functionalized tip (left) (6 × 6 nm2 , I = 50 pA, V = 10 mV). Since
the packing occurs as three tetramers that follow a three-fold symmetry, it is most
likely guided by the triangular shape of the n=2 PLS, which is closest to the Fermi
level (right). The left image has been adapted from ref. [164].
7.3. Conclusions 105
7.3 Conclusions
In conclusion, we have shown that it is possible to obtain the scattering potential
landscape exerted by a nanoporous metal-organic overlayer onto a 2DEG and de-
termine the relevant confinement details and interaction effects. This is achieved by
combining semi-empirical EBEM/EPWE simulations with local and averaging elec-
tronic experimental techniques (STM/STS, AFM, KPFM and ARPES). We showed
that the scattering potential must be parametrized as realistically as possible to the
network geometry while providing flexibility to the 2DEG, which requires a slight
energy and/or mass renormalization due to the interactions between overlayer and
substrate. Our simulations unambiguously define that both the molecules and the
metal adatoms forming the network exhibit repulsive character to the surface elec-
trons. Following other works, we provide a tentative upper limit to the repulsive
scattering magnitude of the molecules for related systems. We also find that slight
perturbations in the scattering potential at the metal sites are responsible for the de-
formation of the confined states, which show up as three-fold. Our work confirms
that the confined 2DEG is sensitive to existing subtle interactions of the overlayer
with the substrate and corroborates the surface state renormalization, which pro-
vides consistency to the results obtained by these semi-empirical simulations.

7.4.1 ARPES Measurements
ARPES measurements for Figures 7.4 (b), 7.7 and 7.8 (b) were performed with our
lab-based experimental setup at 150 K as described in Chapter 2. ARPES measure-
ments of Figures 7.4 (a) and 7.8 (a) were performed at 10 K with hν = 21 eV at
Cassiopee beamline of the Soleil Synchrotron in Paris.
Sample Preparation
To avoid contributions from the Shockley surface state of bare Cu(111) to the ARPES
signal, it was crucial to achieve a homogeneous Cu-coordinated 3deh-DPDI net-
work completly covering the surface. Therefore, DPDI was sublimated onto Cu(111)
held at RT in a wedge geometry (producing a coverage gradient) in the proxim-
ity of the optimal coverage (∼ 0.75 ML) and then annealed (250◦ C) until a sharp
and intense signal emerging from the n=1 PLS was visible in the ARPES chan-
nelplate detector [15] (Figure 7.7). The annealing step is crucial for conversion of
the DPDI molecules into 3deh-DPDI molecules that will create the Cu-coordinated
network [158, 168, 169].
106
250°C
Cu(111) DPDI/Cu(111) 3deh-DPDI/Cu(111)

0.75 ML 0.75 ML
0.0
E-EF (eV)
-0.2
-0.4
-0.6
G G M G M
-0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2
kx (Å-1) kx (Å-1) kx (Å-1)
F IGURE 7.7: Electronic structure evolution from the pristine state, to the confined
case from the 3deh-DPDI network. From left to right, ARPES (E vs kx ) band struc-
ture corresponding to the pristine Cu(111) surface state, 0.75 ML DPDI on Cu(111)
and 0.75 ML 3deh-DPDI molecular network on Cu(111). The band structure clearly
evolves into a shallow dispersive band for the n=1 partially localized state (n=1
PLS) and an energy gap is created between n=1 and n=2 PLS.
STM/STS measurements have been performed by Dr. Sylwia Nowakowska from

Basel University (Switzerland) and close to liquid He temperatures (5 K). The bias
voltages provided refer to a grounded tip. All dI/dV spectra were recorded with
open-feedback loop. The dI/dV data presented in Figures 7.2 and 7.3 were extracted
from the grid spectroscopy measurements, in which an area of 5.45 × 4.3 nm2 was
mapped by acquisition of dI/dV spectra above each point with the resolution of 35
points × 30 points. The initial tip conditions amounted to 400 mV/70 pA (lock-in
frequency: 513 Hz; zero-to-peak amplitude: 8 mV). The value of the initial voltage
was chosen such that no contribution from quantum dot states or network backbone
is present [227]. Under this conditions, normalization could be performed by setting
the same dI/dV value at the setpoint energy for all other spectra. In this way artifacts
originating from local surface potential variations are minimized [17, 53, 227].
7.4.3 LCPD/AFM Measurements
KPFM measurements have been performed by Dr. Shigeki Kawai from MANA/NIMS
(Japan). All experiments were performed with Omicron STM/AFM with a qPlus
configuration, operating at 4.8 K in UHV. The W tip of a tuning fork sensor was ex-
situ sharpened by focused ion beam milling technique and was then in-situ covered
with Cu atoms by contacting to the sample surface.
7.4.4 Side Considerations to This Chapter
• In Figure 7.4 the second derivative of the raw data has been used in order to
enhance details such as the replicating bands away from the first SBZ and the
opening of gaps at zone boundaries [15]. In the following Figure 7.8, the raw
data is included for completeness.
DPDI Soleil RD GM reduced x DPDI Soleil RD GK reduced x
a 0.1 0.1 max.

0.0 0.0
E-EF (eV)
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
            min.
-0.5 -0.5
-0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4 -0.4
0.4-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4
kx (Å-1) ky (Å-1)
b 0.4 -0.24eV -0.13eV -0.07eV Fermi
0.2
Experiment
ky (Å-1)
0.0
max.
-0.2
-0.4
0.4
c -0.24eV -0.14eV -0.07eV Fermi
0.2
ky (Å-1)
min.
0.0 EBEM
-0.2
-0.4
-0.2 0.0 0.2 -0.2 0.0 0.2 -0.2 0.0 0.2 -0.2 0.0 0.2
kx (Å-1)
F IGURE 7.8: Experimental and simulated electronic band structures and isoener-
getic maps in raw data when the 3deh-DPDI metal-organic network saturates the
Cu(111) surface. (a) Energy dispersion maps for the two high symmetry directions
ΓM and ΓK. The high quality of the network allows to observe faint replica bands
in adjoining Brillouin zones. Simulated EBEM/EPWE bands using the scattering
potentials of model 2 (see Table 7.1) are superimposed as red lines onto the exper-
imental data and match perfectly both high symmetry directions. Experimental
(b) and simulated (c) isoenergetic cuts (kx vs ky ) obtained at the band-bottom (-
0.24 eV), lower edge of the M point (-0.13 eV), inside the gap (-0.07 eV) and at the
Fermi level. As a reference, the metal-organic network induced hexagonal SBZ is
superimposed. Experimental ARPES parameters: hν = 21.22 eV, Ts = 10 K for (a)
and 150 K for (b).
• The energy dispersion maps shown in Figures 7.4 (a) and 7.8 (a) correspond
to measurements performed at 10 K and hν = 21 eV at the Soleil Synchrotron
in Paris, while the isoenergetic cuts shown in Figures 7.4 (b) and 7.8 (b) have
108
been performed at our laboratory setup in San Sebastian at 150 K and Helium
I (hν = 21.22 eV). This is done because the quality of the former Energy dis-
persion maps (E vs k) at 10 K is higher while the statistical points gathered
for the isoenergetic cuts (kx vs ky ) in the latter case is larger. The only differ-
ence between both measurments is the temperature variation induced energy
shifts of the n=1 PLS band bottom which amounts to ∼ 30 meV as shown in
Figure 7.9 (a).
• The effective mass of the n=1 shallow dispersive PLS band generated by the
3deh-DPDI nanoporous network has been calculated using the same proce-
dure as in refs. [53, 54]. We consider a small momentum region of ±0.05 Å−1
around the Γ point since this region is not distorted by the gap openings at
zone boundaries [Figure 7.9 (b)].
• Figure 7.9 (c) shows that the gap opening at the zone boundary (M point) be-
tween n=1 and n=2 PLS bands amounts to 120 ± 30 meV, agreeing with previ-
ous observations [15] and evidencing a strong scattering potential at the molec-
ular backbone.
a Cu(111) 150 K b c n=1 PLS n=2 PLS
0.0
0.0 DPDI 150 K
DPDI 10 K
0.0
-0.1
-0.1
Int. (arb. units)
E-EF (eV)
E-EF (eV)
-0.2
-0.2 -0.2
-0.3
-0.3
-0.4
-0.4
-0.4
M G M
-0.6 120±30 meV
-0.5
-0.5
-0.2 -0.1 0.0 0.1 0.2
G point -0.6 -0.4 -0.2 0.0
-0.6
-0.6 kx (Å-1)
E-EF (eV)
Int. (arb. units)
F IGURE 7.9: Characterization of the 3deh-DPDI band structure. (a) EDCs taken at
the Γ point for the pristine Cu surface state (black) and 3deh-DPDI network (red)
at 150 K. The EDC for 3deh-DPDI network (blue) is taken at 10 K. Both features are
identical except for the well-known energy shift (∼ 30 meV) to higher binding en-
ergies as the temperature is decreased [158]. (b) Effective mass extraction from the
n=1 shallow dispersive PLS band. Following the procedure explained in ref. [53]
an effective mass of m∗ = 0.57 m0 is obtained. (c) M point fitting of the peaks us-
ing a Lorentzian-Gaussian distribution convoluted with a Fermi edge yields a band
gap of 120 ± 30 meV, a value that agrees well with EBEM/EPWE simulations and
previous observations [15].
• For the correct EBEM/EPWE simulation of the 3deh-DPDI induced band struc-
ture and LDOS features, an increase of the effective mass was required (m∗ =0.49 m0 ).
This requirement is not only essential for matching the dispersion of the n=1
shallow disperive PLS band of Figure 7.4 (a), but also for bringing higher con-
fined states such as n=4 PLS to lower energies. This energetic shift with in-
creasing m∗ can be conceptually understood by simulating a hard-wall po-
tential hexagonal single pore as in Figure 7.10. First of all, a large potential
amplitude is chosen [Vmol = 20 eV, Figure 7.10 (c)], which mimics an infinite
potential wall. Then the confined states are simulated for two different ef-
fective masses: 0.44 m0 and 0.58 m0 . The first value resembles the one for
pristine Cu(111) [108], while the second one corresponds to the experimen-
tal effective mass observed for 3deh-DPDI network. It immediately becomes
clear that higher confined states are more sensitive to effective mass changes
as their energy shifts are larger [Figure 7.10 (a, b)]. Note that by changing m∗
in EBEM/EPWE, confined states can be shifted in energy but the spatial fea-
tures (LDOS) appear unchanged [Figure 7.10 (c)]. This mass renormalization
becomes necessary for matching the experimental findings in 3deh-DPDI net-
work as well as in other networks [54].
a n=4
b
ΔE (m*0.44-m*0.58) (eV)
-0.5
Int. (arb. Units)
-0.4
-0.3
n=1 n=2 n=3 -0.2
-0.1
0.0
n=1 n=2 n=3 n=4
-0.4 0.0 0.4 0.8 1.2 1.6 2.0

Energy (V)
c (-0.028eV) (0.594eV) (1.417eV) (1.698eV)
m*=0.44 m0
n=1 n=2 n=3 n=4

(-0.129eV) (0.353eV) (0.985eV) (1.206eV)
1 nm
m*=0.58 m0
n=1 n=2 n=3 n=4
F IGURE 7.10: EBEM/EPWE simulations of the confined state energy variations

with m∗ inside a practically infinite potential hexagonal pore. (a) Discrete confined
state levels from n=1 to n=4 are shown for two relevant effective masses: 0.44 m0
and 0.58 m0 . While n=1 and n=4 confined states can be traced at the center of the
pore (red), n=2 and n=3 confined states shift away from it and peak at halfway
position (blue). (b) Upon increasing the effective mass, the confined states shift
towards lower energies, the higher ones being more sensitive (larger energy shifts)
to this changes. (c) EBEM/EPWE simulated LDOS are identical for both effective
masses, but peak at lower energies for the larger m∗ value.
110
• Starting from a realistic scattering geometry and allowing the 2DEG to renor-
malize in energy and/or mass has a critical effect in higher order confined
states. For instance by looking at Figure 7.11, we clearly observe that the highly
localized confined state predicted by Model 1 (blue spectrum) at 1 V is not
present in the experimental STS curve (grey). In order to simultaneously push
the energy of this confined state down and quench its intensity, the geometry
of the scattering landscape has to be changed [Figure 7.2 (g)], adopting a real-
istic geometry, and an increased effective mass of m∗ = 0.49 m0 must be used
(Model 2, red spectrum).
PC Model 1
PC Model 2
PC STS
-0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 0.6
0.6 0.8
0.8 1.0
1.0
Voltage (V)
F IGURE 7.11: EBEM/EPWE simulated LDOS curve at the pore center (PC), using
scattering parameters from Model 1 (blue) and Model 2 (red). The experimental STS
curve at the pore center is shown as grey color. While the matching for both Models
is reasonable below 0.4 V, Model 1 clearly deviates at 1 V since a very localized
state appears which is absent in the experiment. This state has been attenuated and
shifted to lower energies in Model 2 by using a reasonable scattering geometry and
by increasing m∗ , yielding a match with the experiment.
111
Chapter 8
Configuring Electronic States in an

Atomically Precise Array of
Quantum Dots
8.1 Introduction
2D arrays of electronically coupled QDs fabricated on metallic surfaces by means of
molecular self-assembly, assure ultimate precision of each confining unit and long-
range order. These molecular nanoporous templates serve as ideal candidates for
hosting adsorbates in an ordered fashion inside each nanocavity. For instance, single
Fe atoms, Bi clusters, Xe, C60 and ZnOEP molecules have been orderly trapped [161,
164–166, 228]. The periodically extended and robust 3deh-DPDI nanoporous net-
work grown on Cu(111) represents an ideal template for such purposes. An inter-
esting effect of using weakly physisorbed adsorbates such as Xe is that the energy
of the quantum states embedded in the dots can be delicately altered with Xe occu-
pancy. This is a process that can be controlled with atomic precision and engineered
at will by using atomic manipulation. We followed this procedure to study the de-
sired occupancy configurations, getting access to single pore and interpore coupling
effects by LT-STM/STS (experiments performed by Dr. Sylwia Nowakowska from
Basel University) and simulated with EBEM/EPWE (by Dr. Zakaria M. Abd El-
Fattah from Al-Azhar University). In addition, by means of ARPES, we observed a
fundamental energy shift of the n=1 PLS band to lower binding energy upon adsorp-
tion of Xe in the nanocavities, in agreement with STS. We followed such ensemble
electronic band changes by simultaneously detecting the appearance of the Xe core
levels in the spectra, from which the adsorption position could be monitored. This
work has been published in ref. [53].
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
112
Quantum Dots

The array of coupled QDs employed in this work is the 3deh-DPDI network gener-
ated on Cu(111) [Figure 8.1 (a)], extensively described in Chapters 4 and 7 [15, 167–
169]. The low intensity of the n=2 PLS close to EF and the absence of higher or-
der bound states [166] simplify the analysis of the targeted perturbation of the PLS
by adsorbates and its interaction with surrounding QDs. To modify the n=1 PLS in
these pores we chose Xe atoms, first for their well-characterized effects upon the sur-
face state electrons of Cu(111), which is dominated by Pauli repulsion [107, 196], and
second, for its preferential adsorption in the pores of the Cu-coordinated 3deh-DPDI
network with maximal occupancy of 12 atoms [inset in Figure 8.1 (b)] [164]. In this
way we can discretely modify the electronic state of QDs by STM repositioning. As
already reported by Nowakowska et al. [164], the 12 Xe occupancy inside each pore
consists of three tetramers, with Xe atoms adsorbed in on-top sites of the Cu(111)
√ √
atomic lattice in a ( 3 x 3)R30◦ overlayer structure, in agreement with previous
studies on Cu(111) [229, 230].
In our studies we characterize the 2D array of QDs by STM/STS providing site
specific, local information on the effect of Xe adsorption on the n=1 PLS and by com-
plementary ARPES giving access to the coherent part of the interaction between the
QDs.
In the following we characterize two extreme QD occupancy cases: the vacant
network and the fully filled network with 12 Xe atoms adsorbed in each QD. The
local STM/STS results are presented in Figure 8.1. The spatially resolved dI/dV line
scan (white solid line) in Figure 8.1 (a) illustrates the spectral and spatial distribution
of the n = 1 PLS within the vacant network, whose maximum is observed at -201 ± 6
mV [15]. Note that the ∼ 90 mV FWHM and its asymmetric peak shape are charac-
teristic signatures of coupled QDs [54, 160]. Upon adsorption of 12 Xe atoms in each
pore of the network (see Section 8.4.2 for experimental details) [164, 226], the n=1
PLS appears modified, shifting to lower BE, whose maximum peaks at -136 ± 7 mV,
amounting to a ∼ 60 mV shift [Figure 8.1 (b)]. A FWHM of ∼ 110 mV is measured
which is comparable to the ∼ 90 mV for the empty case observed in Figure 8.1(a).
a b Xe
3deh-DPDI network Xe/3deh-DPDI network
FWHM 0 FWHM 0
max.
~90 mV ~110 mV
-136 -136
-147 -147
-201 -201
-247 -247
min.
dI/dV 0 2.55 5.10 dI/dV 0 2.55 5.10
(arb. units) x (nm) (arb. units) x (nm)
Tunneling Tunneling
Voltage (mV) Voltage (mV)
F IGURE 8.1: QD electronic structure alteration through Xe occupation. Two differ-

ent network configurations have been measured by STS for investigating coopera-
tive interactions between the QD states: Empty 3deh-DPDI network (a) and filled
network (b) with 12 Xe atoms adsorbed in each QD are shown in their correspond-
ing upper insets. In each case, a series of dI/dV spectra was acquired along the
white solid line (indicated in the STM images) crossing three QDs. The dI/dV spec-
tra taken at the red and blue dots on the insets are plotted on the left-hand side of
the dI/dV traces (size of STM images 7.8 x 4.3 nm2 . STM parameters: V = -200 mV;
I = 700 pA (a) and V = -10 mV; I = 50 pA (b).
In the same way, such electronic effects on the n= 1 PLS when the network pores
are fully filled with Xe atoms can be studied with ARPES (Figure 8.2). The quasi free-
electron like Cu(111) Shockley state [purple line in Figure 8.2 (a)] is modulated into
a shallow dispersive band consistent with the coupling between the QDs [15], when
the network saturates the surface. Upon Xe filling of the pores (see Section 8.4.1
for preparation details), no pronounced changes are observed, apart from a certain
broadening and loss of intensity in the n=1 PLS band [Figure 8.2 (b)]. However, the
second derivative treatment of the data [Figure 8.2 (c, d)] exhibits a slight upward
shift of the fundamental energy of the band (Γ point) and a smearing of the features
due to the Xe adsorbates [196].
Some quantitative values can be extracted from the ARPES spectra (Figure 8.3
and Table 8.1): First, the dispersion of the n=1 PLS in close proximity to the Γ point
(kx = ± 0.06 Å−1 ) for the empty and Xe filled cases. We consider only this region
around the Γ point because it is far from the dispersion reversal of the zone bound-
ary gaps (M point) [Figure 8.3 (a, b)]. A slight shift of the fundamental energy (∼ 30
meV) is found when extracting and comparing EDCs at the Γ point [Figure 8.3 (c)].
We interpret this energy shift as the result of Pauli repulsion between the electronic
states of Xe and the n=1 PLS, similar to the previously reported shift of the Cu(111)
Shockley state caused by the adsorption of Xe (Table 8.1) [196]. Second, the effective
114
Quantum Dots
a Xe
b
3deh-DPDI network Xe/3deh-DPDI network
No Xe; t =-2 min Xe on pores; t = 13 min
0.0
0.0 0.0
-0.1
-0.1 -0.1
E-EF (eV)
-0.2
-0.2 -0.2
-0.3
-0.3 -0.3
-0.4
-0.4 -0.4
M G M M G M
-0.5
-0.5 -0.5
c -0.2 -0.1 0.0 0.1 0.2 d -0.2 -0.1 0.0 0.1 0.2
0.0
0.0 kx 0.0 kx
-0.1
-0.1 -0.1
E-EF (eV)
-0.2
-0.2 -0.2
-0.3
-0.3 -0.3
-0.4
-0.4 -0.4
M G M M G M
-0.5
-0.5 -0.5
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2
kx (Å-1) kx (Å-1)
F IGURE 8.2: Influence of Xe adsorption on the partially localized states hosted at

the pores of the network. (a) ARPES experimental spectral function acquired for
the empty network. Compared to the pristine Cu(111) Shockely state (in purple),
the characteristic cosine shape band develops when the n = 1 PLS couples among
neighboring pores of the network (measured along ΓM high symmetry direction).
(b) ARPES experimental spectral function after Xe adsorption in the pores. A slight
shift towards higher energies is observed. (c, d) Second derivative treatment of
the bands in (a, b) to better visualize the replicating bands and gaps at the zone
boundaries (M point).
mass can also be extracted from fitting the EDCs [Figure 8.3 (d)]. A slight flatten-
ing of the band is dectected, evidenced by a notable increase in the effective mass
(m∗ /m0 = 0.57 for the empty and m∗ /m0 = 0.66 for the filled case). This effect may
suggest a slight loss of interpore-coupling and consequently, an increased localiza-
tion of the n=1 PLS [54]. Note that such strong vertical interaction effects between Xe
and n=1 PLS are not present for the free-electron-like Cu(111) Shockley state, where
no significant change of the effective mass is observed with an adsorbed Xe layer (cf.
TABLE 8.1: Band structure parameters extracted from ARPES measurements for
the pristine substrate and the network with and without Xe. a) The values refer to
E − EF at normal emission (band bottom); b) The effective mass and the bandwidth
of the n=1 PLS along ΓM (details of how m∗ is extracted from the ARPES spectral
functions can be found in Section 8.4); c) DPDInet denotes the Cu-coordinated 3deh-
DPDI network formed on Cu(111); d) Values from ref. [196].
Sample E − EF at Γa) (meV) m∗ /m0 b) Bandwidth (meV)

Cu(111) -434 ± 2d) 0.43 ± 0.01d) —
Xe/Cu(111) -291 ± 2d) 0.44 ± 0.02d) —
DPDInet c) /Cu(111) -270 ± 10 0.57 ± 0.02 90 ± 10
Xe/DPDInet c) /Cu(111) -240 ± 10 0.66 ± 0.02 70 ± 10
a 3deh-DPDI network b Xe/3deh-DPDI network c

0.1
0.1 0.1 0.10
DPDI
Xe/DPDI
0.0
0.0 0.0 0.00
-0.1
-0.1 -0.1 -0.10
E-EF (eV)
-0.2
-0.2 -0.2 -0.20
30 meV
-0.3
-0.3 -0.3 -0.30
-0.4
-0.4 -0.4 -0.40
EDCs at G
-0.5
-0.5 -0.5 -0.50
-0.2 -0.1 0.0 0.1 0.2 -0.2 -0.1 0.0 0.1 0.2 Intensity (arb. Units)
-0.2 -0.1 0.0
kx
0.1 0.2 -0.2 -0.1 0.0
kx
0.1 0.2
kx (Å-1 ) kx (Å-1 )
d -0.21
e
-0.21 Fermi
DPDI
Xe/DPDI

-0.22
-0.22 m*/m0 = 0.66
E-EF (eV)
-0.23
-0.23
-0.24
-0.24
m*/m0 = 0.57
-0.25
-0.25 DPDI exp
Xe/DPDI exp
M G M
fit DPDI
-0.26
-0.26
fit Xe/DPDI
-60 -0.04
-0.06 -40 -0.02
-20 0.00
0
-3 0.02
20 0.04
40 0.06
60 -0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2
x10
kx (Å-1 ) kx (Å-1 )
F IGURE 8.3: Effective mass renormalization after Xe adsorption. ARPES spectral

function of the surface state region (a) without Xe and (b) after Xe adsorption fully
covering the pores (12 Xe atoms per pore) of the Cu-coordinated 3deh-DPDI net-
work on Cu(111). The data were fit using a single convoluted Lorentzian-Gaussian
component on a linear background and multiplied by a Fermi-Dirac distribution.
(c) Normal emission EDCs (kx = 0) from (a, b) where a ∼ 30 meV shift to higher en-
ergies is found. (d) From the band bottom dispersion fitting close to the Γ point, an
upward energy shift of the minimum energy and a reduction of the band dispersion
are observed when Xe is adsorbed in the pores. Using a parbolic fit to each we ex-
tract an apparent increase of the effective mass from (a) m∗empty /m0 = 0.55 ± 0.02 to
(b) m∗Xe /m0 = 0.66 ± 0.02. (e) Fermi wave-vector (kF ) measurements don’t reveal
apparent differences in the number of electrons present in between both 2DEGs
cases.
Table 8.1) [196]. Finally, in Figure 8.3 (e) we monitor the Fermi wave-vector (kF ) and
no change is detected, corroborating that the number of electrons in the 2DEG is not
affected by the Xe presence [220].
To investigate the influence of the surrounding QDs upon the electronic state
of an isolated unit, we generated different configurations by repositioning single Xe
atoms with the STM tip [231–233]. Two model arrangements were studied in detail:
a filled QD surrounded by empty ones [Figure 8.4 (a)] and an empty QD surrounded
by filled ones [Figure 8.4 (b)].
In the first model case [Figure 8.4 (a)], the spatially resolved dI/dV line scan of
the QDs for the filled pore is significantly different from the one observed for the in-
dividual pore in the Xe filled network [cf. Figure 8.1 (b)]: the Xe induced component
at lower BE dominates in intensity [Figure 8.4 (a), white arrow] exhibiting a sharp
QD state peak with a FWHM of ∼ 50 mV [Figure 8.4 (a), blue dI/dV spectrum]. We
attribute this sharpening to the additional out-of-plane confinement of the state via
the Pauli repulsion imposed by Xe and the loss of coupling with the empty neigh-
boring pores. The surrounding empty QDs still feature the same spatially resolved
dI/dV trace as the vacant array [cf. Figure 8.1 (a)]. Closer inspection reveals that the
116
Quantum Dots
a b
FWHM 0 FWHM 0 max.

~50 mV ~110 mV
-136 -136
-147 -147
-201 -201
-247 -247
FWHM FWHM
~90 mV ~150 mV min.
dI/dV 0 2.55 5.10 dI/dV 0 2.55 5.10
(arb. units) x (nm) (arb. units) x (nm)
Tunneling Tunneling
Voltage (mV) Voltage (mV)
F IGURE 8.4: QD electronic structure alteration through Xe single occupation or

depletion. Using STM atom manipulation two different network configurations
are engineered for investigating local interactions at each particular QD: Filled QD
with 12 Xe atoms surrounded by empty QDs (a) and empty QD surrounded by
filled QDs with 12 Xe atoms adsorbed in each QD (b). In each case, a series of
dI/dV spectra were acquired along the white line (indicated in the top STM images)
crossing three QD units. The dI/dV spectra taken at the red and blue colored dots
superimposed on the STM images are plotted on the left-hand side of the dI/dV
traces. Size of STM images 7.8 x 4.3 nm2 . STM parameters: V = -145 mV; I = 100
pA (a) and V = -145 mV; I = 100 pA (b).
single filled pore exhibits a QD state peak at slightly higher BE (∼ 10 meV) than
the filled network [Figure 8.1 (b)]. We tentatively attribute this to the fact that when
electronic states couple, the n=1 PLS peak broadens and the peak shape becomes
asymmetric.
In the second model case, we consider a central empty QD surrounded by filled
ones [Figure 8.4 (b)]. The spatially resolved dI/dV trace of the QD state of the empty
pore does not exhibit the Xe induced localization of the previous case. On the con-
trary, a bound state-like feature is observed, whose peak shifts to higher BEs (-247 ±
7 mV) than the reference case of the vacant network [cf. Figure 8.1 (a)] [234, 235]. The
dI/dV signal of the surrounding QDs is the same as for the filled network reference
[Figure 8.1 (b)]. In addition, the single empty pore peak looks broadened, showing a
FWHM of ∼ 150 mV. At this point, the origin and shape of this confined state cannot
be explained.
To explain all the different cases, we turn to simulating the electronic proper-
ties for all the aforementioned Xe-filling configurations using EBEM/EPWE. These
simulations have been performed in close collaboration with Dr. Zakaria M. Abd
El-Fattah (Al-Azhar University, Egypt). We first simulate the electronic structure
of the 3deh-DPDI metal-organic network on Cu(111). As we are mainly interested
on the n=1 PLS and in the pore center spectra, following Model 2 in Chapter 7
we parametrize the scattering potential landscape of 3deh-DPDI and assign both
empty pores
LDOS Empty Pores

LDOS Xe Filled Pores
Empty
dI/dV (arb. units)
filled pores
Xe-filled
-0.35 -0.30
-0.35 -0.25 -0.20
-0.30 -0.25 -0.20 -0.15
-0.15 -0.10
-0.10 -0.05
-0.05 0.00
0.00
Energy (eV)
F IGURE 8.5: LDOS at the pore center for all empty and all Xe-filled pore configu-
rations simulated with EBEM/EPWE. The scattering potential landscape for the all
empty case in shown in red while the Xe-filled case is shown in blue. The scatter-
ing potential amplitudes used and Cu SS renormalization values are indicated in
Table 8.2. Both simulated LDOS at the pore center match the experimental obser-
vations of Figure 8.1.
molecules and Cu adatoms in coordination a common repulsive scattering ampli-

tude (Vnetwork ). The Cu SS renormalization is also taken into consideration (see
Table 8.2). The LDOS at the center of an empty pore (red spectrum in Figure 8.5)
matches the peak shape and energy of the n=1 PLS. The vertical red dashed line
corresponds to the peak energy measured with STS in Figure 8.1 (a). To simulate
the Pauli repulsion effect produced by Xe adsorbates on the n=1 PLS, we slightly
increase the potential inside the pore by 80 meV (Table 8.2) so that the resulting n=1
PLS (blue spectrum) matches the experimental shift (∼ 60 mV) measured in Fig-
ure 8.1 (b) (dashed blue line). A small increase in coupling is detected since the
molecular barrier now represents a smaller effective scattering potential (Vnetwork =
310 meV) than the empty network. This is in agreement with the slight increase in
the peak width detected in Figure 8.1 (b).
Once we have verified that EBEM/EPWE can accurately simulate the Xe induced
electronic effects, we proceed to simulate the two Xe filling configurations studied
in Figure 8.4. First, we simulate the scenario where all the pores are empty (Vpore
TABLE 8.2: EBEM/EPWE reference and scattering parameters used for different
Xe-filling configurations
Configuration ERef,Γ
EBEM (eV) m∗ /m0 Vpore (meV) Vnetwork (meV)
All Pores Empty -0.45 0.49 0 390
All Pores Xe-Filled -0.45 0.49 80 390
All Empty/ One Xe-Filled -0.45 0.49 0/80 390
All Xe-Filled/ One Empty -0.45 0.49 80/0 390
118
Quantum Dots
= 0 eV) and only one single pore is filled with Xe (Vpore = 80 meV) [Figure 8.6 (a)
and Table 8.2]. The LDOS for the empty pore is taken at a pore that is completly
surrounded by empty pores (black dot in the scattering landscape inset). The LDOS
peak (red spectrum) matches the expected shape and energy for the characteristic
n=1 PLS. This means that a coherent electronic band exists from the coupling be-
tween empty pores, irrespective of the presence of a single pore filled with Xe. Such
condition supports the experimental observation that ARPES will still detect the n=
1 PLS band structure upon preseence of a small amount of defects. The LDOS at
the Xe-filled pore shows a localized state found at higher energies, nicely matching
the experimental peak position (dashed blue line). In addition, a broad weak shoul-
der is detected at ∼ −0.25 eV. This feature corresponds to the lower energy part
of the states (bonding side) of the n=1 PLS band. As mentioned already in Chap-
ter 5, bonding states are more spread out due to their large electron wavelength than
the antibonding ones and therefore can slightly leak into the isolated Xe-filled pore.
However, there is no sign of the antibonding states, since they are much more local-
ized inside each pore [54, 160].
Next, we invert the scenario, and in Figure 8.6 (b) simulate the case where all the
pores are filled with Xe and only one single pore is left empty. This configuration
corresponds to the one experimentally studied in Figure 8.4 (b). For the Xe-filled
pore, a position is chosen so that it is only surrounded by identical pores (yellow
dot in the scattering landscape inset). The blue LDOS matches the expected n=1
PLS both in shape and energy position (cf. dashed blue line for STS). Notably, the
empty pore (in red) also agrees with the experimental findings and represents an
interesting case. The overall peak shape agrees with the inverted asymmetry and
width detected experimentally in Figure 8.4 (b). The lineshape can be rationalized
as a combination of the lack of coupling of the empty pore with the neighboring Xe-
filled ones and a localized state that appears at high BEs, whose fundamental energy
is located around -0.23 eV. This localized state cannot couple with neighboring ones
since their fundamental energy is at lower energies (blue spectrum). Contrarily, the
bonding states of the Xe n=1 PLS band can leak into the isolated empty pore, induc-
ing a prominent shoulder at -0.17 eV and increasing considerably the overall peak
width of the red spectrum. Note that we practically do not observe any sign of the
antibonding states in the red spectrum due to their more localized nature.
8.3. Conclusions 119
a LDOS Empty Pore all empty one filled

LDOS Xe Filled Pore
dI/dV (arb. units)
b -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05

LDOS Empty Pore
0.00
LDOS Xe Filled Pore
all filled one empty
dI/dV (arb. units)
-0.35 -0.30
-0.35 -0.30 -0.25
-0.25 -0.20
-0.20 -0.15
-0.15 -0.10
-0.10 -0.05
-0.05 0.00
0.00
Energy (eV)
F IGURE 8.6: EBEM/EPWE simulated LDOS for isolated Xe-Filling Configurations.

(a) Situation where a single Xe-filled pore is surrounded by empty pores. The LDOS
at the Xe-filled pore matches the localized state detected experimentally and the
features at the empty pore correspond to the expected n=1 PLS. The weak shoul-
der at lower energies in the blue spectrum evidences the widely spread bonding
states leaking inside the pore. (b) Inverted scenario where a single empty pore
is surrounded by Xe-filled pores. Xe-filled pores bear the characteristic n=1 PLS
shifted in energy due to Pauli repulsion effect. The empty pore shows and inverted
asymmetric peak with two contributions: a localized state at high BEs (-0.23 eV)
and a prominent shoulder at -0.17 eV corresponding to the leaky bonding states of
neighboring (Xe-filled) pores.
8.3 Conclusions
In summary, we demonstrate that it is possible to engineer the electronic structure of
the n=1 partially localized state (n=1 PLS) in an array of QDs by globally adsorbing
Xe guests inside the nanocavities or locally repositioning Xe atoms. Due to Pauli
repulsion effects, the n=1 PLS shifts to lower binding energies (∼ 60 meV) when the
pores are fully filled with Xe atoms. This tendency is corroborated by ARPES mea-
surements, where the fundamental energy of the n=1 PLS band also shifts towards
the Fermi energy. By atomic manipulation with STM, we can design and engineer
different Xe filling configurations at the local or first neighbor scale: an empty pore
surrounded by filled ones and vice versa. Such configurations yielded highly local-
ized states accompanied by features stemming from the surrounding states of the
n=1 PLS electronic band. In contrast, the QDs surrounding the isolated one remain
practically unaffected in their LDOS lineshape. These results, have been corrobo-
rated by EBEM/EPWE modelling by using the same scattering potential landscape
120
Quantum Dots
as in model 2 of Chapter 7 (repulsive scattering potentials) along with the required

Cu SS renormalization. The additional Xe induced effect has been simulated by
introducing a shallow 80 meV potential inside the pores. Such delicate electronic
effects observed with Xe adsorbates demonstrates the consistency obtained in pre-
vious chapters of the manuscript that delves into the intricacies of confinement and
interpore coupling effects.

8.4.1 Sample Preparation and ARPES Acquisition Details
To avoid contributions from the Shockley state of bare Cu(111) to the ARPES sig-
nal, DPDI was sublimated onto Cu(111) held at room temperature (RT) in wedge
geometry (coverage gradient) in proximity of the optimal coverage. Afterwards the
sample was annealed to ∼ 520 K until a sharp and intense signal emerging from
partial localization of the QD state, as in refs. [15, 158], was visible in the ARPES
channel plate detector [Figure 8.2 (a)]. The Xe dosing experiment was started imme-
diately after the sample temperature dropped below 60 K, but was higher than 25
K to keep the adsorbate mobility [226, 236]. Xe core level (9-5 eV BE) and surface
state (close to Fermi energy) regions at normal emission were acquired alternatively
as a function of time while keeping Xe pressure in the chamber constant (5 × 10−10
mbar). The evolution of the 5p3/2 and 5p1/2 core levels and the n=1 PLS band (at
Γ) as function of the Xe exposure time is shown in Figure 8.7 (total exposure time
amounted to ≈ 30 min). At 0 Langmuir (L), without any Xe exposure, the n=1 PLS
looks as in Figure 8.2 (a) while no Xe core level features are visible [black spectrum
in Figure 8.7 (b) and dashed black line in Figure 8.7 (a)]. At 0.4 L the single Xe core
level features start to be predominant, where a doublet at 7.46/6.23 eV is observable
in Figure 8.7 (b) (blue line). This corresponds to the situation where the pores are
filled with 12 Xe atoms each. At this point, the n=1 PLS already responds by shifting
its fundamental energy ∼ 30 meV while the intensity begins to drop as seen for the
blue spectrum in Figure 8.7 (d) and dashed blue line in Figure 8.7 (c). Beyond this
dose (> 0.4 L up to 1.0 L) a less bound Xe component evolves (doublet at 7.29/6.08
eV). We attribute this new peak to the deposition of Xe atoms on top of the molecular
backbone and coordination sites. For 1.0 L and beyond, the n=1 PLS appears already
quenched [green spectrum in Figure 8.7 (d) and green dashed line in Figure 8.7 (c)]
coinciding with the Xe multilayer formation.
a Xe Core Levels b
-5
-5
-6
Int. (arb. units)

-6
E-EF (eV)
-7
-7
-8
-8
0 L Xe
0.4 L Xe
-9
-9 1.0 L Xe
0.0
0.0 0.5
0.5 1.0
1.0
Xe exposure / L -9
-9 -8
-8 -7
-7 -6
-6 -5
-5
E-EF (eV)
c d
G point (bottom of PLS)
0 L Xe
0.4 L Xe
0.0
0.0 1.0 L Xe
Int. (arb. units)
E-EF (eV)
-0.2
-0.2
-0.4
-0.4
0.0
0.0 0.5
0.5 1.0
1.0
-0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1
Xe exposure / L -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1
E-EF (eV)
F IGURE 8.7: Changes in the electronic structure of the Cu-coordinated 3deh-DPDI
network on Cu(111) upon Xe exposure monitored by ARPES. (a, b) The 5p3/2 and
5p1/2 Xe core levels and (c, d) the surface state region alternatively recorded during
Xe exposure expressed in Langmuirs. In (a) the photoemission intensity is repre-
sented as a grayscale plot, whereas in (b) selected EDCs are shown as a waterfall.
Initially, a single component (doublet at 7.46/6.23 eV), which we attribute to the
adsorption within the network pores, increases in intensity up to ∼ 0.4 L (blue line
in (a); blue spectrum in (b)). From that moment, a less bound component (doublet
at 7.29/6.08 eV) develops, which we attribute to adsorption at the network back-
bone, until completion of the full layer (green line in (a); green spectrum in (b)).
Both doublet energies are indicated as vertical dotted lines. The evolution of the
n=1 PLS at Γ is shown as a grayscale plot in (c) and as selected EDCs in (d). We
observe a strong attenuation of the state and a decrease of its BE upon increase of
Xe exposure. The ∼ 30 meV shift of the n=1 PLS is determined from the black and
blue spectra which were acquired at the same Xe exposure as the black and blue
spectra in (a, b).
122
Quantum Dots
Sample Preparation for STM/STS Measurements
All STM/STS measurements have been performed by Dr. Sylwia Nowakowska from
Basel University (Switzerland). The Cu-coordinated 3deh-DPDI network on Cu(111)
was prepared according to the preocedure described in ref. [169]. Xe of purity 99.99%
was dosed onto the sample placed in the STM (Omicron Nanotechnology GmbH
with Nanonis SPM control system) operated at 4.2 K, with the cryoshields open and
the leak valve being in line-of-sight with the sample. The sample was exposed to 120
L (Langmuir) of Xe (1.3 × 10−7 mbar for 1200 s) resulting in the increase of the sam-
ple temperature to 9 K. The STM measurements performed after cooling the sample
back to 4.2 K revealed different numbers of Xe atoms adsorbed in the pores as well
as the domain boundaries and step edges. Filling of all pores with 12 Xe atoms was
performed by subsequent annealing to 45 K followed by cooling to 4.2 K for the STM
measurements.
Repositioning of Single Xe Atoms
All the condensates discussed here were obtained by controllably removing of Xe

atoms from the pores of the network by means of STM repositioning. Noticeably,
new configurations that are not occurring spontaneously upon Xe exposure can be
created in this way [Figure 8.4 (a, b)] [53, 164].
STM/STS Measurement Details and Data Analysis
In the STM the bias voltage is applied to the tip. The bias voltages given in this chap-
ter refer to a grounded tip. STM measurements were performed in constant current
mode with Pt-Ir tips (90% Pt, 10% Ir), prepared by mechanical cutting followed by
sputtering and controlled indentation in the bare Cu(111) substrate. All STM images
were acquired with a Xe functionalized tip, which allows for obtaining atomic reso-
lution of Xe condensates [53, 164].
To avoid modification of the condensates via interaction with the tip, the sample
bias was selected within a range of -10 mV to -80 mV, whereas the tunneling current
was set within the range of 5-50 pA. The STM data were processed with the WSxM
software [174]. For better comparability of the data the color histograms of the STM
images were adjusted. Low-pass filtering was used for noise reduction.
All dI/dV spectra were recorded with open-feedback loop and with Xe function-
alized tip. Control spectra were acquired with a metallic tip and no difference was
observed in accordance with [237].
The dI/dV data presented in Figures 8.1 and 8.4 were extracted from the grid spec-
troscopy measurements, in which an area of 7.8 x 4.3 nm2 was mapped by acquisi-
tion of dI/dV spectra above each point with the resolution of 50 points x 30 points for
the empty network and 60 points x 30 points for the other three cases [16, 17]. The
initial tip conditions amounted to 400 mV/70 pA (lock-in frequency: 513 Hz; zero-
to-peak amplitude: 8 mV). As described in detail in ref. [227] the value of the initial
voltage was chosen such that no PLS or network backbone related contribution was
present. Owing to that, normalization could be performed by setting the same dI/dV
value at the setpoint energy of all dI/dV spectra. In this way artifacts originating
from local work function (surface potential) variations are minimized [227].
125
Chapter 9
Part III: Introduction to On-Surface

Synthesized One-Dimensional
Zigzag Covalent Polymers
The discovery of graphene’s exceptional electronic properties [238, 239] triggered the
research of graphene-based electronic devices. However, the absence of a bandgap
has precluded the use of this two-dimensional material in electronic applications.
A band gap can be introduced into graphene through quantum confinement in one
of its lateral dimensions, forming graphene nanoribbons [240]. Since the electronic
properties of GNRs are sensitively dependent on their width and edge structure, it
is an essential requirement to control their morphology down to the atomic level,
a degree of precision that currently cannot be obtained through top-down meth-
ods [241]. Recently established bottom-up techniques based on on-surface synthe-
sis in UHV [57, 242], have facilitated the manufacture of atomically well-defined
GNRs of various widths and edge structures [28, 29, 243]. Notably, numerous types
of armchair edge nanoribbons [Figure 9.1 (a)], as well as ribbons with chiral and
zigzag edges have been produced, the latter ones hosting low energy edge-localized
states [28]. The high degree of control obtained so far has not only facilitated fab-
rication of atomically precise GNRs of a single type, but also heterojunctions, such
as metal-semiconductor junctions [244], type-I (straddling gap), type-II (staggered
gap) junctions [23, 188] and topological GNR superlattices [25, 30] [Figure 9.1 (b)].
Recently, even multifunctional nanoporous graphene has been synthesized from
GNR fusing, representing a highly versatile semiconductor for simultaneous siev-
ing and electrical sensing of molecular species [24] [Figure 9.1 (c)]. They all consti-
tute primary building block examples for nanoelectronic applications such as high-
performance field-effect transistors and ultra-low power devices such as tunneling
field-effect transistors [32, 245, 246].
GNRs are conjugated polymers highly related to other types of organic chains
that are extensively used in industry as light emitting materials, photocatalysts, solar
cells and biosensors due to their large and tunable bandgaps [247–250]. Control over
their electronic properties is accomplished through topological functionalization of
Chapter 9. Part III: Introduction to On-Surface Synthesized One-Dimensional
126
a b c
F IGURE 9.1: On-surface synthesis of GNRs and nanoporous graphene. (a) STM
image of armchair type GNRs with varying widths obtained after laterally fusing
poly-(para-phenylene) chains (adapted from ref. [29]). (b) A bond-resolved STM
image of 7/9-AGNR superlattice shows the bond-resolved structure of the hetero-
junction interface (adapted from ref. [25]). (c) STM image (18 × 18 nm2 ) of the
nanoporous graphene overlayer synthesized on Au(111) (adapted from ref. [24]).
these π-conjugated oligophenylene chains, i.e. modification of their conductive path-

ways. In particular, changes of conjugation (from linear to cross-conjugation) by
precise transitions from para- to meta-ligand substitutions [251, 252] weaken the elec-
tronic communication between the repeating units of the polymer [253, 254]. Such
modifications have also been described as quantum interference electron pathways
[255–257], and bear predicted effects such as scarcely dispersive bands [251], wider
electronic bandgap [258], distinct optical properties [250], electronic switching ca-
pabilities [259] and low conductance properties [65, 255–257, 260]. However, pe-
riodic meta-junctioned zigzag chains may also show enhanced charge mobility as
compared to their poly-(para-phenylene) counterparts, reaching values comparable
to those of amorphous silicon [261].
Despite this wealth of industry attractive properties of cross-conjugated poly-
mers, key fundamental information, such as the predicted electronic structure awaits
experimental validation. Such deficiency of fundamental knowledge limits the con-
fidence in the existing predictions, according to which topology is expected to affect
the electronic properties of the polymer. Several obstacles are responsible for the lack
of the aforementioned experimental confirmation: i) the need of generating atomi-
cally identical chains exhibiting repeated para- to meta-ligand substituted units, ii)
the synthesis of well-aligned chains, to be probed by non-local, averaging spectro-
scopies, iii) the minimization of lateral interactions, prone to affect their intrinsic
band structure, and iv) the right choice of a support that sufficiently decouples the
electronic signal from the investigated oligophenylene chains.
To overcome such obstacles, solutions can be found within the context of Sur-
face Science. Particularly, the first prerequisite for obtaining perfectly reproducible
cross-conjugated zigzag polymers can be accomplished by bottom-up on-surface
synthesis. Surface-assisted C-C coupling processes such as the Ullmann coupling
Chapter 9. Part III: Introduction to On-Surface Synthesized One-Dimensional
127
reaction, have recently been applied to generate GNRs with different edge termina-
tions and widths [25, 27, 30, 62, 243, 262], and other types of oligophenylene chains
[26, 77, 263–265]. Secondly, the chain alignment for non-local characterization can be
achieved by the use of nanotemplated substrates, such as vicinal surfaces [81, 82, 103,
147]. These special surfaces have been successfully used for the macroscopic align-
ment of carbon-based chains, a fundamental requirement for ARPES experiments
[26, 27, 77–79]. With respect to the minimization of lateral interchain coupling, this
is an inherent feature of the Ullmann-type surface reactions [263] since the halogens
are cleaved during the synthesis positioning themselves between neighboring chains
[26, 77, 78, 265]. These adatoms are reported to laterally decouple adjoining chains,
without affecting the polymer’s band structure, except for a minimal rigid energy
shift similar to doping effects [26, 78]. Finally, the substrate plays a fundamental
role as a catalyst of the Ullmann reaction, making its choice crucial for a success-
ful oligomer coupling. Good candidates that present excellent yields, control and
reproducibility are the closed packed surfaces of coinage metals, which are exten-
sively used for Ullmann-type surface reactions.
In the following Chapter 10, we show how we have overcome all the afore-
mentioned obstacles and have generated an extended film of atomically precise
oligophenylene zigzag chains on a vicinal Ag(111) surface, as evidenced by STM and
LEED. The electronic band structure of such films has been unravelled by means of
ARPES and complemented by STS measurements on Ag(111). In this way, we de-
termine the experimental energy gap and visualize the spatial distribution of the
frontier orbitals. Such wealth of experimental information is clarified and expanded
by a comprehensive set of DFT calculations and EPWE simulations. This work has
been recently accepted for publication [266].
As previous chapters, the next one will have a paper-like format, starting with a
brief introduction indicating the main motivation. After this, the main results will
be described and discussed supported by several figures and it will end with the
conclusions and some additional supplementary information.
129
Chapter 10
Electronic Structure Tunability by

Periodic meta-Ligand Spacing in
One-Dimensional Organic
Semiconductors
10.1 Introduction
In the previous part of this thesis, the electronic structure studied from nanoporous
networks came as a consequence of the modification of the substrate’s 2DEG, which
was independent of the existing molecular states. In this chapter, the molecular
structures are dominant and we investigate possible confinement effects on 1D or-
ganic nanostructures generated with atomic precision by on-surface synthesis pro-
cesses. We use this ‘dry’ chemistry to introduce topological variations in a conju-
gated poly-(para-phenylene) chain in the form of meta-junctions. As evidenced by
STM and LEED, we produce a long-range ordered, monolayer thin zigzag chain
film on a vicinal silver crystal. These cross-conjugated nanostructures are expected
to display altered electronic properties, which are now unravelled by two highly
complementary experimental techniques (ARPES and STS) and theoretical calcula-
tions (DFT and EPWE). We find that meta-junctions dominate the weakly disper-
sive band structure, while the bandgap is tunable by altering the linear segment’s
length. These periodic topology effects induce significant loss of electronic coupling
between neighboring linear segments leading to partial electron confinement in the
form of weakly coupled quantum dots. Such periodic quantum interference effects
determine the overall semiconducting character and functionality of the chains. De-
signing such molecular organic semiconductors with distinct frontier orbitals is key
for the development of devices with desirable properties.
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
130
One-Dimensional Organic Semiconductors

In collaboration with Prof. Gottfried from Philipps-Universität Marburg (Germany),
we have produced a monolayer film of cross-conjugated zigzag chains from the sur-
face polymerization of the 4,4"-dibromo-meta terphenyl (DMTP) molecular aromatic
precursors via C-C coupling [see Figure 10.1 (a) and Section 10.4.4]. The template of
choice is a vicinal Ag(111) crystal surface with linear, monoatomic steps running par-
allel to the [112̄] direction [82] that corresponds to the so-called fully-kinked (100%
kinked) configuration of the step-edge [Figure 10.1 (c, d)]. We used this particular
substrate since it provides a higher flexibility to reconstruct and therefore accom-
modate the produced zigzag structures more efficiently (cf. Section 10.4.4). Indeed,
we can already disclose that we achieved an excellent film featuring a high yield
of well-ordered and aligned zigzag chains. This alignment is critical for the ARPES
measurements performed afterwards.
The formed zigzag chains appear practically planar on the surface [Figure 10.1 (b)]
and are covalently bonded displaying the characteristic phenyl-phenyl distance of
a ∼ 4.3 Å along the straight segments and a superperiodicity of L ∼ 2.24 nm be-
tween equivalent elbows [263, 267]. The unit cell of the chain features two straight
subunits made up of two phenyl rings (in para-positions) linked to two edge rings
acting as meta-junctions [green dots in Figure 10.1 (a)]. Note that in the STM image
these chains are separated by spherical features that are attributed to Br atoms split
off from the precursor molecules at the initial step of the on-surface reaction [29, 74,
175, 268–273]. The LEED pattern reveals that the organic chains are aligned parallel
to the steps and show long-range order as they are commensurate with the underly-
ing substrate [Figures 10.1 (b) and 10.9 in Section 10.4.4]. Particularly, the main silver
diffraction spots (red and green circles) are sided by a set of spots aligned along the
average step direction yielding a (9, 5; 0, 4) superstructure. Note that the best align-
ment is observed on a vicinal plane ∼ 3.6◦ off from the (111) crystal position [see
Figure 10.1 (c) and Section 10.4.4].
Our STM and LEED structural results contain the required ingredients (atomic
precision of the structure, defined alignment, long-range order and minimization of
lateral interactions by Br adatom presence) to expect the existence of a defined and
coherent electronic band structure from these chains. Figure 10.2 (a-c) shows the
second derivative (to enhance the details) of the ARPES spectral weight obtained
from such a film saturating the surface (raw data is shown in Section 10.4.5). The
resulting electronic structure in the direction parallel to the average step direction
and the main axis of the zigzag chains (E vs ky with kx = 0.1 Å−1 ), exhibits weakly
dispersive bands between −1.8 eV and −3.5 eV, separated by a ∼ 0.6 eV gap [Fig-
ure 10.2 (a)]. None of these ARPES features are observable on the pristine substrate
(cf. Figure 10.8). A closer inspection reveals that each one of them consists of a pair of
2π
anti-phase oscillatory bands [Figure 10.11]. The spectral intensity peaks around a∗ ,
where a∗ represents the projected phenyl-phenyl distance along the average chain
a b
[11-2]
2n
-4n Br
470 K L
Ag(111)
a* 0
DMTP_x135_49eV
Br a [11-2]
100
n
200
ky
300
kx
400
49eV
c d 100 200 300 400 500
(645)
(111) A-step
3.6°off
(423)
α
α
fully-kinked
step
Curved vicinal
B-step
Ag(111) crystal
F IGURE 10.1: Structural arrangement of the zigzag chain film grown over a vici-
nal Ag(111) surface. (a) Schematic representation of the DMTP precursor and the
resulting zigzag covalent chain, showing its characteristic lengths: phenyl-phenyl
distance (a) and its projection along the chain’s average direction (a∗ ), and poly-
mer superperiodicity (L). (b) High resolution STM image after chain synthesis on a
vicinal plane ∼ 3.6◦ off from the (111) crystal position. The zigzag chains are sepa-
rated by Br atoms and preferentially follow the step parallel direction ([112̄]). (STM
parameters V = −394 mV, I = 234 pA, Tsample = 100 K). Inset shows the LEED
pattern after the chain formation that exhibits single-domain, well-aligned arrange-
ment. The superstructure spots are in registry with the circled main spots (in red
the (0,0) and in green the substrate’s first order diffractions), implying commensu-
rability to the terrace atoms (LEED parameters: Ekin = 49 eV, Tsample = 300 K).
(c) Schematic drawing of the curved Ag(111) substrate. (d) Atomic ball model
highlighting the three step types present on close-packed vicinal surfaces. In our
case, the step termination corresponds to the one ending in the green spheres (100%
kinked) that runs along the [112̄] direction.
direction [Figure 10.2 (a)], assuring its molecular origin. This assertion is based on
the fact that the real space molecular orbital configuration relates to the k-space po-
sitions and intensities of the obtained electronic bands [274]. The faint replicas with
2π
L periodicity (vertical dashed blue lines), stem from the zigzag chain superperiod-
icity L [Figure 10.1 (a)], in agreement with the STM dataset.
The 1D nature of these zigzag chains is demonstrated by the lack of disper-
sion perpendicular to the average chain axis. Figure 10.2 (b) shows a representa-
tive cut (E vs kx ) across the center of the 6th Brillouin zone [green arrow at ky =
1.39 Å−1 in Figure 10.2 (a)], where discrete flat bands are observed. This confirms
132
a (kx = 0.1 Å-1)

b (ky = 1.39 Å-1)
c (E = -1.98 eV)
2π/L 2π/a*
0.0 0.0
-0.5 -0.5 1.5
1.5
-1.0 -1.0
E-EF (eV)
1.0
ky (Å-1)
-1.5 -1.5
-2.0 -2.0
0.5
0.5
-2.5 -2.5
-3.0
-3.0 -3.0 0.0
-3.5
-3.5 -3.5
0.0 0.5
0.5 1.0 1.0 1.5
1.5 0.0 0.5 0.0 0.5
ky (Å-1) kx (Å-1) kx (Å-1)
F IGURE 10.2: ARPES electronic band structure of the zigzag chain film grown
over a vicinal Ag(111) surface. (a) ARPES experimental band structure of meta-
junctioned cross-conjugated zigzag chains parallel to the average direction of
chains and steps (E vs ky with kx = 0.1 Å−1 ). (b) Experimental band structure
perpendicular to the chain average axis (E vs kx , with ky = 1.39 Å−1 [green ar-
row in (a)]. (c) Isoenergetic cut (kx vs ky ) at the top of the valence molecular band
[E = −1.98 eV, marked by red arrow in (b)]. The position on the crystal is the same
one as in Figure 10.1 (b). The second derivative of the intensity is shown in a linear
color scale (highest being black). ARPES parameters: hν = 21.2 eV, Tsample = 150 K.
that they stem from different molecular orbitals of the zigzag polymer [275]. The
non-dispersive character at the top of the valence band (red arrow at −1.98 eV) can
also be traced from the isoenergetic cut (kx vs ky ) shown in Figure 10.2 (c), where
1D polymer bands replicate at each Brillouin zone center, gaining intensity for the
larger ky values. These photoemission intensity modulations have been simulated
with the EPWE method in collaboration with Dr. Zakaria M. Abd El-Fattah from
Al-Azhar University (Egypt), which confirms that these features are neither affected
by the templating Ag surface nor by the presence of Br atoms intercalated between
the chains (see Figure 10.11 in Section 10.4.5).
Indeed, we experimentally find that the presence of Br embedded in between the
zigzag chains only causes a rigid shift of the molecular band structure of about 200
± 50 meV to higher energy, according to Figure 10.3 and in agreement with previous
work [26, 78]. After the Ullmann coupling reaction takes place, zigzag chains con-
dense into islands. The cleaved Br atoms accumulate between the chains (Figure 10.1
and Figure 10.3). These observations have been already reported for zigzag chains
grown on Cu(111), hyperbenzene and honeycombene closed rings on Ag(111) and
other covalent nanostructures such as PPP chains on Au(111) [78, 269–271]. The Br
atoms remain on the Ag(111) up to ∼ 600 K and can likely stabilize the long-range
ordered zigzag structures observed in the present work. According to Merino-Díez
and co-workers [78] the desorption of Br atoms intercalated in between the chains
produces a lowering of the workfunction and a shift of the electronic states to higher
a b
Condensed chain

Isolated chain
-2.5 -2.0 -1.5-2.5

-1.0 -2.0
-2.5 -0.5 -1.5
-2.0 0.0 -1.0
-1.5 -1.0
0.5 1.0
-0.5 1.5
1.0 0.0 2.0
1.5 0.5 2.5
2.0 1.0
2.5 1.5 2.0 2.5
Sample bias (V)
F IGURE 10.3: Effect of Br adatoms on the electronic structure of isolated and con-
densed chains grown on Ag(111). (a) STM image of a Br stabilized zigzag chain
island (condensed chains) and an isolated, Br-free zigzag chain obtained via tip
manipulation (Imaging parameters: 50 mV, 100 pA; frame: 15 × 15 nm2 ). (b) Cor-
responding constant-height dI/dV spectra for the VB and CB at selected positions.
The presence of Br rigidly shifts the VB and CB onsets by 200 ± 50 mV to higher
energies (STS parameters: Bias voltage modulation of 10 mVrms at 341 Hz. Close-
feedback parameters: -350 mV, 150 pA and 1200 mV, 100 pA for the negative and
positive resonances regions, respectively).
binding energies without affecting their effective mass. We obtain similar results
when we compare dI/dV spectra of tip manipulated isolated chains with island con-
densed ones [Figure 10.3 (a)]. In particular, the observed rigid shift of the frontier
orbitals [cf. Figure 10.3 (b)] implies that the N =4 zigzag chain energy gap is un-
changed. This evidences the limited effect of the halogen atoms on the electronic
properties of the polymers.
Our ARPES results suggest that the zigzag chains are largely decoupled from
the metallic substrate since the observed molecular bands do not show signs of hy-
bridization with the substrate in that energy window [77]. Besides, the chains are
semiconducting in nature with a bandgap certainly larger than 2 eV, since no other
bands closer to the Fermi energy are observed in the occupied region. The band
structure strongly contrasts with that of the poly-(para-phenylene) chains, which ex-
hibits a single, highly dispersive molecular band across the entire Brillouin zone [26,
77, 83, 264] [Figure 10.4 (b, e)]. Instead, it closely resembles the one predicted
for poly-(meta-phenylene) (called PMP hereafter) chains [251], implying that the
presence of meta-junctions strongly modifies the electronic structure of a polymeric
chain [65] (cf. Figure 10.13 in Section 10.4.7).
The weak interaction observed between the zigzag chain film and the substrate
is a favorable playground for a systematic theoretical analysis. As a first approx-
imation, we consider the polymers as free-standing and planar. On this basis, we
use DFT calculations (performed by Dr. Aran Garcia-Lekue from DIPC) to corrobo-
rate the weakly dispersive band structure observed experimentally. The calculated
134
electronic structure shown in Figure 10.4 (a) exhibits convincing qualitative agree-
ment with the experimental data. In particular, the dispersive character of the first
four valence bands (VBs) of the zigzag chain (between −1 eV and −2.5 eV) is con-
sistent with that in Figure 10.2 (a). The energy mismatch can be attributed to the
absence of a substrate in the calculations, as well as to the well-known limitation
of DFT to accurately predict HOMO-LUMO gaps. Note that the calculated bands
span from the Γ point to the Brillouin zone boundary ( Lπ ), which in the experiment
2π
appears replicated 12 times until a∗ [see Figure 10.4 (d)]. In this case a kx integrated
band structure is presented in order to capture the replicating bands. For compar-
ison, the calculations are extended to straight PPP chains [Figure 10.4 (b)] which
strongly differ in the electronic structure by exhibiting a highly dispersive single VB
in this energy window. This is in agreement with the experimental band structure
for PPP chains shown in Figure 10.4 (e), where a highly dispersive electronic band
2π
is observed peaking at a . This periodicity corresponds to the phenyl-phenyl dis-
tance along the polymer (see Section 10.4.6). Moreover, the zigzag chain exhibits a
greater bandgap than its straight counterpart, confirming its enhanced semiconduc-
tive character (cf. Figure 10.13).
The experimental value of the frontier orbital bandgap of the zigzag chains can
be obtained by low-temperature (4 K) STS. For such measurements (performed in
collaboration with the group of Prof. Dimas García de Oteyza from DIPC) we de-
posit a submonolayer coverage of DMTP molecules on Ag(111). In this way, small
zigzag island patches are formed on the surface while still allowing access to the
bare substrate for tip calibration and treatment [Figure 10.5]. Figure 10.5 (a) shows
the dI/dV spectra at the center (red) of a straight arm of a zigzag chain (cf. inset of
figure) and the Ag substrate (grey). The VB onset is detected at -2.1 V [coinciding
with the ARPES value in Figure 10.2 (b)] while the conduction band (CB) edge is at
1.6 V resulting in an overall bandgap of ∼ 3.7 eV. Therefore this value is larger than
the 3.2 eV reported for PPP chains grown on Au(111) [29].
DFT calculations can also shed light onto the effect that the periodically spaced
meta-junctions have on the overall electronic structure by comparing the spatially
resolved molecular orbitals at the Γ point with the π molecular orbitals of ben-
zene [155] [Figure 10.4 (c)]. In the PPP case [Figure 10.4 (b)], VB and CB are con-
structed by the overlap of Φ3 and Φ∗3 benzene molecular orbitals, respectively. These
orbitals present a large electronic weight on the carbon atoms linking the phenyl
rings (para-positions), giving rise to highly dispersive valence and conduction bands.
Likewise, the less dispersive character of the VB-1 and CB+1 bands can be attributed
to the orbital set that exhibits a nodal plane through the para carbon atoms (Φ2 and
Φ∗2 orbitals). Contrarily, for the zigzag chains [Figure 10.4 (a)] the VB and CB are a
combination of two degenerate orbitals [254]. In particular, the VB is made up of
Φ3 (straight sections) and Φ2 (elbows) orbitals, which is mirrored in the CB by Φ∗3
a b c
zigzag PPP
3 3
Φ*1
CB+1
CB+1
2 2 Φ*2
1 1 Φ*3
CB
E-EF (eV)
E-EF (eV)
CB
0 0 Φ3
VB
VB
-1 -1 Φ2
Φ1
-2 VB-1 -2
VB-1
Γ π/L Γ π/a
d 2π/L (kx integrated) e (kx = 0 Å-1)

kx integrated 2π/a* 2π/a
0.0
0.0 0.0
-0.5
-0.5 -0.5
-1.0
-1.0 -1.0
-1.5
-1.5 -1.5
E-EF (eV)
-2.0
-2.0 -2.0
-2.5
-2.5 -2.5
-3.0
-3.0 -3.0
-3.5
-3.5 -3.5
0.0
0.0 0.5
0.5 1.0
1.0 1.5
1.5 0.0
0.0 0.5
0.5 1.0
1.0 1.5
1.5
ky (Å-1) ky (Å-1)
F IGURE 10.4: Comparison of molecular orbital shape and band structure between
zigzag chains and straight PPP chains, as obtained from DFT calculations and
ARPES. The right plots in (a) and (b) show the calculated electronic band structure,
where (a) corresponds to the zigzag polymer and (b) to PPP. The highly dispersive
character of the PPP bands contrasts with the practically flat bands of the zigzag
chains, accompanied by a notable difference in the frontier orbital bandgap. Left
panels in (a) and (b) show the spatially resolved molecular orbitals at Γ for each
band. In a simplistic view, they can be constructed by overlapping different ben-
zene molecular orbitals, which are schematically shown in (c). (d) and (e) show the
second derivative of the ARPES spectral intensity (E vs ky ) along the zigzag and
PPP chain axis. These plots follow the periodicity of the chain unit cells ( 2π
L and
π
a ) marked with vertical dashed blue lines. For the PPP polymer film the substrate
used is a Ag(544) vicinal crystal (see Section 10.4.6).
136
(straight sections) and Φ∗2 (elbows). This orbital mixing, along with the reduced or-
bital amplitude at the meta-positions and expected phase shifts induced by momen-
tum steering at the elbows, results in a diminished orbital interaction (overlap) that
leads to a severe weakening of the electron coupling between adjacent straight seg-
ments. Indeed, the flat band character is also exhibited by the VB-1 and CB+1, even
though they mostly arise from a single type of benzene molecular orbital coupling
(Φ3 and Φ∗3 , respectively). This strong electronic effect governed by the meta-junction
is generally referred to as cross-conjugation [251, 252] or destructive quantum inter-
ference [255–257, 260].
The reduced electronic coupling between neighboring linear segments causes
electron localization, an effect that can be adequately addressed with STS. Figure 10.5 (b)
presents a color plot representing stacked dI/dV point spectra measured along a sin-
gle straight segment [black dashed line in the inset of Figure 10.5 (a)]. Aside from
clearly visualizing an overall bandgap of 3.7 eV (vertical dashed white lines), we
observe confinement in the CB within such segments where the spatial modulations
of the LDOS are consistent with the first two stationary states of a particle in a box.
In particular, their amplitudes die away at the edges of the linear segments (elbow
positions of the zigzag chains) but the lower state at 1.9 V features an antinode at the
segment’s center [the peak in the red spectrum of Fig. 10.5 (a)], whereas the second
state at 2.5 V oppositely exhibits a node at that position. Such electron confinement
effects have also been observed in related structures, as in the case of finite size PPP
chains featuring a single elbow (in meta-junction) [65] or in closed-cycle geometries
of honeycombenes [271]. In essence, we can conclude that the meta-junctions act as
scattering barriers for the polymer electrons regardless of the overall geometry, i.e. as
closed structures [271] or as edged (non-linear) chains [65].
Once we have verified that each straight segment of our zigzag chains acts as a
confining unit, reminiscent of a 1D array of weakly interacting QDs [54], it should
be possible to tune their electronic properties by modulating the straight segment’s
length. Varying the 1D QD length should affect the energy levels as well as the
corresponding frontier orbital bandgap. To do so, we co-evaporated on Ag(111)
linear precursors (DBTP molecules [26]) together with the previously used ones to
generate the zigzag chains, as shown in Figure 10.6 (a). The Ullmann coupling reac-
tion is likewise activated by post-annealing to 470 K, resulting in linear segments of
phenyl length configurations of N = 4 + 3n (N being the total phenyl number and n
the amount of DBTP precursors embedded in the straight segment). Figure 10.6 (b)
shows color plots of the dI/dV linescan for N = 7 (bottom) and N = 10 (top) QDs,
which evidence the expected squared wavefunction intensity variations in the CBs
for the same energy range as Figure 10.5 (b). Most importantly, by comparison to
the dashed white lines corresponding to the N = 4 segment, we observe that the
bandgap shrinks as the size of the segment increases (dashed blue lines). A quanti-
tative analysis of the experimentally determined bandgap is shown in Figure 10.6 (c),
1
revealing a N behavior in agreement to previous work for similar chains [65]. Such
a c
7Å
+1.9 V
-2.5 -2.0 -1.5 -1.0 -2.5 -0.5 -2.0 0.0 -1.5 0.5 -1.0 1.0 -0.5 1.5 0.0 2.0 0.5 2.5 1.0 1.5 2.0 2.5
-2.5
-2.0
-1.5
-1.0
-2.5 -2.0 -1.5 -1.0 1.0 1.5 2.0 2.5

1.0
1.5
2.0
2.5
b
Line position at (a) inset
-2.1 V
aaa
bbb
Sample Bias (V)
F IGURE 10.5: STS experimental determination of the zigzag chain’s frontier orbitals
on Ag(111). (a) Constant-height dI/dV spectra acquired at the center of a zigzag
straight arm (red point in STM image inset) and substrate (grey) (STM imaging
parameters: 50 mV, 100 pA; frame: 3.6 × 2.5 nm2 ). (b) Constant-height dI/dV lines-
can spectra along the zigzag arm (indicated by a dotted line in the STM topogra-
phy inset) for the same bias range of panel (a). The onsets of the VB and CB are
clearly defined, yielding a bandgap of ∼ 3.7 eV (STS bias voltage modulation for
(a) and (b): 10 mVrms at 341 Hz. Close-feedback parameters: -350 mV, 150 pA and
1200 mV, 100 pA for the negative and positive resonances regions, respectively).
(c) From top to bottom: ball and stick model of the zigzag chain. High-resolution
dI/dV maps acquired at constant-height with a CO functionalized STM tip at 1.9 V
and -2.1 V, i.e. close to the CB and VB onsets (frames: 3.0 × 1.5 nm2 ; bias voltage
modulation 10 mVrms at 341 Hz). Underneath each map the corresponding DFT
gas phase molecular frontier orbitals are shown for comparison.
behavior matches our DFT calculations for planar, free-standing, periodic zigzag
chains of different straight segment length, which confirm not only that the bandgap
of the cross-conjugated zigzag chains is larger than the one of its linear PPP coun-
1
terpart, but it is also tunable with a N relation (cf. Figures 10.13 and 10.14 in Sec-
tion 10.4.7).
Finally, we should discuss an additional property of this system that may affect
the frontier orbital bandgap: the relative twisting of the phenyl rings (non-planar
chain morphology). Figure 10.5 (c) shows high-resolution constant height dI/dV
maps close to the valence and conduction band onsets, namely at -2.1 eV and 1.9
eV, respectively. Both maps replicate well the molecular orbital simulations of Fig-
ure 10.4 (a), which are calculated for planar structures and are depicted below for
direct comparison. While the slight discrepancies in nodal positions are attributed to
138
a b
-2.5
-2.0
-1.5
-1.0
1.0
1.5
2.0
2.5
N = 10
bbb
aaa
N=7
Line position at (a)

9Å
N = 10
-2.5
-2.0
-1.5
-1.0
1.0
1.5
2.0
2.5
c
3.8
Bandgap (eV)
3.7
3.6
bbb
aaa
3.5
3.4
3.3 N=7
N=4 N = 7 N = 10
0.25 0.20 0.15 0.10 -2.5 -2.0 -1.5 -1.0 1.0 1.5 2.0 2.5
1/N Sample Bias (V)
F IGURE 10.6: Tuning the electronic confined states through the linear segment’s
length (QD size). (a) Schematic representation of the co-evaporated molecular pre-
cursors (DMTP and DBTP) that generate longer straight segments between the
meta-coordinated phenyls. The STM image shows two longer linear segments of
7 and 10 phenyl rings between elbows (Imaging parameters: 50 mV, 100 pA; frame:
4.5 × 2.3 nm2 ). (b) Constant height dI/dV line profiles close to the VB and CB onsets
along the two segments following the dashed lines in (a). The intensity modulation
of the CB confirms the confinement nature of the meta-junction termination. Fur-
thermore, the frontier orbital bandgap (vertical dashed blue lines) is reduced com-
pared to the N =4 case, which is indicated by the two vertical dashed white lines.
(STS parameters: bias voltage modulation of 10 mVrms at 341 Hz; close-feedback
at -350 mV, 150 pA and 1200 mV, 120 pA for the negative and positive resonances
regions, respectively). (c) Experimental bandgap extracted for the STS line profiles
revealing a linear N1 behavior (see Figure 10.14).
the CO probe functionalization [276], the intensity variations are in turn ascribed to
the twisting of the phenyl rings [29]. In order to reveal whether such morphological
effects are present, we perform constant height bond resolution imaging with a CO
functionalized tip [277]. This is done on condensed zigzag chains (surrounded by
Br atoms) that are compared with an isolated chain obtained after tip manipulation
[Figure 10.7]. Both types of 1D structures show some very weak intensity variations
at the phenyl rings, which suggest that there are slight twists in the configurations.
This would agree with the observations for PPP chains where a phenyl twisting of
20 ± 5◦ was observed with respect to the planar configuration [26]. Indeed, the sur-
rounding morphological conditions are determinant, since the twisting is different
when the zigzag chains are condensed into islands (asymmetric arms) or are isolated
(symmetric twisting of consecutive arms). Such subtle variations could be caused by

the hindrance of Br or by a different matching to the underlying substrate.
a b H
H H
H
H
H
H H
H H H
H H H H
R H H
H
H
H R
H
H H H
H H H
Condensed chain Condensed chain H
c d H
H H
H H
H H
H H
H
H H H H
H H
H H
R H H R
H H
H H H H
H H
Isolated chain Isolated chain
e f g
Two Phenyl Twist Two Phenyl Twist Single Phenyl Twist

Opposite Orientation Same Orientation
h 0.15
2 phenyl opposite orient.
i 2.56
2 phenyl same orient.
single phenyl
0.10
2.52
Energy gap (eV)
E-EFlat zigzag (eV)
0.05
2.48
0.00
2.44
-0.05
2.40
2 phenyl opoosite orientation
2 phenyl same orientation
-0.10 single phenyl
2.36
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Rotation angle (°) Rotation angle (°)
F IGURE 10.7: Influence of phenyl-phenyl twisting on the stability of the zigzag

chains and their electronic bandgap. Bond resolution STM images (current sig-
nal, post-processed with a Laplace filter) acquired at constant height with a CO-
functionalized STM tip (frames: 3.0 × 1.5 nm2 , Vtip = 2 mV) of zigzag chains con-
densed within an island (a) and an isolated chain obtained by tip manipulation (c).
These images suggest a slight phenyl twisting, which for the condensed chain is
asymmetric with respect to the central meta-junction of both straight segments (left
ascending and right descending), as shown in (b), whereas it is symmetric with
respect to the central meta-junciton for both straight arms for the isolated chain,
as shown in (d). Influence of phenyl-phenyl twisting on the stability of the zigzag
chains and their electronic bandgap. Three different phenyl twisting configurations
have been considered: two phenyl twist (opposite orientation) (e), two-phenyl twist
(same orientation) (f) and single phenyl twist (g). (h) Energy of the three twisting
configurations with respect to the fully planar configuration (zero energy). (i) Evo-
lution of the electronic bandgap with respect to the phenyl rotation angle, for the
three different configurations.
140
To determine the effect of the twisting on the frontier orbital bandgap, we have
performed DFT calculations on the N =4 zigzag chain and we consider three phenyl
twisting configurations: a single phenyl twisting, twisting of two phenyls with the
same orientation and twisting of two phenyls with opposite orientations (see Fig-
ure 10.7 (e-g)). As shown in Figure 10.7 (h), in all three cases the energy minimum is
found for twisting angles of approximately 20◦ to 35◦ with respect to the planar con-
figuration. However, the calculations show that the electronic bandgap is scarcely in-
creased compared to the planar configuration (by about 7%) [Figure 10.7 (i)]. Given
that the underlying substrate is known to decrease the phenyl twisting by flattening
the chains, we expect this effect to be smaller than ∼ 200 meV, validating the use of
free-standing, planar configuration geometries in the DFT calculations that support
our findings.
10.3 Conclusions
We have been able to synthesize and macroscopically align a saturated film of cross-
conjugated oligophenylene zigzag chains on a vicinal Ag(111) surface. We find
that these atomically precise chains remain sufficiently decoupled from each other
and from the substrate to probe their elusive band structure with ARPES, revealing
weakly dispersing one-dimensional electronic bands along the chain direction. DFT
band structure calculations and EPWE photoemission intensity simulations satisfac-
torily reproduce our experimental findings. By means of STS, we find that the zigzag
chain has a larger frontier orbital bandgap than its straight counterpart (PPP) and ob-
serve electronic confinement in each straight segment of the zigzag chains, reminis-
cent of 1D arrays of weakly interacting QDs. Such states can be tuned by changing
the length of the straight segments, affecting the frontier orbital bandgap, which fol-
1
lows a N dependency. Indeed, our molecular orbital simulations confirm that the pe-
riodically spaced meta-junctions at the elbows of the zigzag chain are the main struc-
tural feature responsible for the reduction of electronic coupling between adjacent
linear segments. These findings corroborate the important effects that the conduc-
tive path topology of a molecular wire has on its frontier orbitals, which are respon-
sible for defining its chemical, optical and electronic properties. Recent advances
in transfer techniques ensure that on-surface synthesized and well-aligned organic
nanostructures can be collectively transferred onto insulating substrates maintain-
ing their relative arrangement [24, 32, 33], which opens the path to further study
the transport and optical properties of these cross-conjugated oligophenylene zigzag
chains.

10.4.1 VT-STM and LT-STM/STS Measurements
The STM measurements on the curved Ag(111) crystal were carried out at ∼ 100 K
using a variable temperature Omicron STM with a Nanonis SPM control system.
The bias voltages given in Figure 10.1 (b) and Figure 10.9 (c, f) refer to a grounded
tip. STM data were acquired in constant current mode and were processed with the
WSxM software [174].
LT-STM/STS measurements were performed by the group of Prof. Dimas García
de Oteyza from DIPC with a commercial Scienta-Omicron low temperature system,
operating at 4.3 K. For the measurement, the bias voltage was applied to the tip while
the sample was electronically grounded. The STM tip was prepared ex-situ by clip-
ping a Pt/Ir wire (0.25 mm) and sharpened in-situ by repeatedly indenting the tip a
few nanometers (1 to 4 nm) into the Ag surface while applying bias voltages from 2
V to 4 V between tip and sample. In order to perform bond-resolved STM imaging,
the tip apex was terminated with a CO molecule, directly picked up from the sur-
face, by positioning the sharp metal tip on top of it and applying a 500 ms bias pulse
at -2 V. The imaging was performed by measuring at constant height while applying
a bias voltage to the tip within the range of 2.0 mV to 3.5 mV. For spectroscopic point
spectra and conductance maps, the dI/dV signals were measured by a digital lock-in
amplifier (Nanonis). STM images were analyzed by using the WSxM software [174].
10.4.2 Ab-Initio DFT Calculations
Ab-initio calculations were carried out by Dr. Aran Garcia-Lekue from the DIPC us-
ing DFT, as implemented in the SIESTA code [278]. The optB88-vdW functional [279],
which accounts for non-local corrections, was adopted for the exchange and correla-
tion potential. For each organic nanostructure, we considered a supercell consisting
of a chain infinite along the x axis, with vacuum gaps of 15 Å in y and z directions
in order to avoid interactions between chains in adjacent cells. A Monkhorst-Pack
k-point grid with 20x1x1 k-points was used for the Brillouin zone sampling and the
mesh cut-off for real space integrations was set to 300 Ry. We employed a double- ξ
plus polarization (DZP) basis set, and a mesh-cutoff of 300 Ry for the real-space in-
tegrations. All structures were fully relaxed until residual forces were less than 0.01
eV/ Å.
142
10.4.3 The Electron Plane Wave Expansion Method (EPWE)
In collaboration with Dr. Zakaria M. Abd El-Fattah from Al-Azhar University (Egypt),
the EPWE method is employed to simulate ARPES data. Within EPWE approach, the
photoemission intensity for a given binding energy and photoelectron wave vector
is obtained from Fermi’s golden rule applied to the in-plane wave function (an initial
state) and a normalized plane wave (a final state) for the parallel component of the
photoelectron wave function, as detailed in ref. [34]. In this semi-empirical method,
zigzag chains are considered free-standing and planar, which implies that the simu-
lated bands in Figure 10.11 are substrate independent and free of Br interactions.
10.4.4 Sample Preparation and Zigzag Alignment on a Ag(111) Curved

Vicinal Crystal
A silver crystal surface curved around the (645) direction was used as tunable vic-
inal substrate for chain formation and alignment [82]. This curved sample exhibits
(111) terraces of variable size (position dependent on its curvature) separated by
monoatomic steps oriented along the [112̄] direction. The steps are of fully-kinked
type, in which out-protruding atoms have no side neighbors [Figure 10.8 (a, b)].
Notably, periodic roughening of such step-edges bears negligible energy cost, and
hence can readily accommodate to the zigzag structure of the chains [82]. The sam-
ple was cleaned by repeated cycles of Ar+ sputtering at energies of 1.0 keV, followed
by annealing at 700 K. This produced clean and well-ordered surface step arrays as
verified by the splitting of the LEED spots along the surface curvature. The sub-
strate’s band structure has the d-bands’ onset below -3 eV, which provides a large
energy window for the zigzag chain bands undisrupted observation [Figure 10.8 (c,
d)].
The meta-junctioned haloaromatic compound, DMTP, was sublimated from a
ML
Knudsen cell at 360 K at a low flux (1 hour ) while the sample was held at 470 K [263,
280, 281]. Covalently bonded zigzag chains appeared separated by Br adatoms, sug-
gesting some influence from the latter in steering chain growth and alignment. The
sample was heated during deposition in order to bypass the organometallic phase
and directly form covalent structures. The step flexibility promoted the chain for-
mation while keeping other irregular structures or hyperbenzene macrocycles to a
minimum [272]. The structures remained densely packed up to high temperatures
(∼ 600 K), beyond which halogen desorption takes place, accompanied by chain
misalignment [243, 282].
The saturated zigzag film that presents best order was observed at the vicinal
plane ∼ 3.6◦ from the (111) region, corresponding to an average terrace size of 3.8
nm [Figure 10.8 (b)]. The data shown in Figure 10.2 corresponds to this position
of the substrate. Here, zigzag chains grow into a long-range ordered polymer film
parallel to the steps [Figure 10.9 (c)] and show a (9, 5; 0, 4) matrix superstructure in
a Kinked-step b
(111) (645)
B-step 3.6°off
(423)
α
α
(𝟏𝟏𝟐)
A-step
Curved vicinal
Ag(111) crystal
c kx0.1
d
00 kx = 0.1 Å-1
-1-1 1.5
1.5
M
sp-band
-2-2 ky (Å-1)
E-EF (eV)
1.0
1.0
-3
-3
-4
-4
d-bands
0.5
0.5
-5-5
-6-6 0.0
0.0
G ss
-7
-7
0.0
0.0 0.5
0.5 1.0
1.0 1.5
1.5 0.0
0.0 0.6
0.6
ky (Å-1) kx (Å-1)
F IGURE 10.8: Atomic and macroscopic geometry of the 100% kinked curved
Ag(111) crystal and its electronic structure ∼ 3.6◦ off the (111) plane. (a) Schematic
drawing highlighting the three step types present on close-packed vicinal surfaces.
In our case, the step termination corresponds to the one ending in the green spheres
(100% kinked) that runs along the [112̄] direction. (b) Schematic drawing of the
curved Ag(111) substrate. The crystal was polished so that the (111) region is away
from the center of the crystal by 9.27◦ . Zigzag chain films showed best alignment
at the position with a local vicinal plane of ∼ 3.6◦ off from the (111) region, where
kinked steps are 3.8 nm separated in average from each other. (c) The ARPES band
structure (E vs ky ) at kx = 0.1 Å−1 corresponds to the direction parallel to the [112̄].
The characteristic silver d-bands exist below -3 eV and the sp-band shows a large
dispersive character as it raises and crosses the Fermi level (highlighted by a side
dashed purple line). (d) Fermi Surface Map (kx vs ky at E = 0), exhibiting the char-
acteristic Shockley state close to the Γ point and the dispersive dominant sp-bulk
bands closer to the M point.
LEED [Figure 10.9 (b)]. Contrarily, in the (111) region, zigzag chains condense into
multidomains as nicely observed in the STM image [Figure 10.9 (f)] and LEED pat-
tern [Figure 10.9 (e)].
144
a b c
Vicinal Ag(111) Zigzag chains
cAg465_135 DMTP_x135_49eV
0 0
[11-2] [11-2]
3.6° off from (111)
100 100
200 200
300 300
400 400
45.4eV 49eV
100 200 300 400 500 100 200 300 400 500
d cAg465_136 e DMTP_x136_49eV f
0 0
[11-2] [11-2]
100 100
(111) region
200 200
300 300
400 400
45.4eV 49eV
100 200 300 400 500 100 200 300 400 500
F IGURE 10.9: Uniaxial zigzag chain ordering vs multi-domain structures deter-

mined by LEED and STM. Experimental LEED patterns of the silver substrate be-
fore (a) and after (b) the formation of the zigzag chain film at the ∼ 3.6◦ position off
from the (111). This pattern agrees with a (9, 5; 0, 4) matrix, which corresponds to an
elbow-elbow distance of 9 atomic spacings of the Ag substrate. (c) Corresponding
STM image showing the preferential uniaxial alignment of the zigzag chains paral-
lel to the step direction (STM parameters: size = 50 x 50 nm2 , V = -393.7 mV; I = 234
pA). Panels (d) and (e) show respectively the LEED patterns of the pristine silver
substrate before and after forming zigzag chains at the (111) region. At this posi-
tion, a hexagonal pattern around the (0,0) spot is observed which evidences that the
chains primarily follow three main directions. This is accordingly observed in the
corresponding STM image in (f), where 120◦ rotated patches co-exist as highlighted
in green, yellow and purple. (STM parameters: size = 35 x 35 nm2 , V = -425.2 mV;
I = 254 pA).
10.4.5 ARPES Band Structure and Photoemission Intensity Simulations

Using EPWE
The ARPES band structure for the zigzag chains presented in Figure 10.2 corre-
sponds to the second derivative treatement of the photoemission intensity. Such
data treatment is used to enhance the weak photoemission features. For complete-
ness, the raw data is included in Figure 10.10.
a (kx = 0.1 Å-1) b (ky = 1.39 Å-1) c (E = -1.98 eV)

kx=0.1 bin5
0.0 0.0
-0.5
-0.5 -0.5 1.5
-1.0
-1.0 -1.0
E-EF (eV)
1.0
ky (Å-1)
-1.5
-1.5 -1.5
-2.0
-2.0 -2.0
0.5
-2.5
-2.5 -2.5
-3.0
-3.0 -3.0 0.0
-3.5
-3.5 kx=0.1 bin5 -3.5
0.0 0.0 0.5 1.0 1.50.0 0.0 0.5 0.0 0.5
-0.5
-0.5 -0.5 1.5
1.5
-1.0
-1.0 -1.0
E-EF (eV)
1.0
1.0
ky (Å-1)
-1.5
-1.5 -1.5
-2.0
-2.0 -2.0
0.5
0.5
-2.5
-2.5 -2.5
-3.0
-3.0 -3.0 0.0
0.0
-3.5
-3.5 -3.5
0.0 0.5 1.0 1.5 0.0 0.5 0.0 0.5
0.5
ky (Å-1) kx (Å-1) kx (Å-1)
F IGURE 10.10: Raw data comparison to the 2nd derivative treatment for the elec-
tronic band structure of zigzag chains at the vicinal plane ∼ 3.6◦ off from the
(111) region. The ARPES spectral intensity raw data is shown in the top row
and the 2nd derivative in the bottom row. (a) ARPES band maps (E vs ky for
kx = 0.1 Å−1 ). (b) Band structure perpendicular to the chains average axis (E
vs kx , with ky = 1.39 Å−1 ). (c) Isoenergetic cut (kx vs ky ) at the top of the VB
(E = −1.98 eV). The photoemission intensity presented in (a) has been obtained by
following the direction highlighted by the black line. The band structure perpen-
dicular to the main zigzag chain axis shown in (b) is shown as a horizontal green
line.
146
To capture the photoemission intensity modulation observed in ARPES, we per-

form photoemission intensity simulations using the EPWE method [34, 117]. These
simulations (excluding the substrate and Br atoms presence) match the experimen-
tal data shown in Figure 10.11 and support the argument of the scarce effects that
the underlying Ag substrate and Br atoms have on the electronic properties of the
zigzag chains.
1/A
0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6
a ARPES b EPWE -1.8
1.8
1.8 1.8
1.6
-1.6
1.6
1.6
ky (Å-1)
ky (Å
1.4
-1.4
1.4
1.4
1/A )
-1
1.2
-1.2
1.2
1.2
1.0
-1.0
1.0
1.0
-0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 0.6
0.6 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
kx (Å-1) kx (Å-1)
= 0 Å-1
kxkx=0 = 0.15 Å-1
kxkx=0.15 = 0.35 Å-1
kxkx=0.35
c d e
-2.0
-2.0 -2.0 -2.0
E-EF (eV)
ARPES
-2.5
-2.5 -2.5 -2.5
-3.0
-3.0 -3.0 -3.0
kx0 kx=0.15 kx=0.35
-1.5
-1.5
1.0 1.2 1.4 1.6
-1.5
1.8
-1.5
f g1.0 1.2 1.4 1.6 1.8

h 1.0 1.2 1.4 1.6 1.8
-2.0 -2.0
E-EF (eV)
-2.0 -2.0
EPWE
-2.5
-2.5 -2.5 -2.5
-3.0
-3.0 -3.0 -3.0
-3.5
-3.5 -3.5 -3.5
1.0 1.2 1.4 1.6 1.8 1.0 1.2 1.4 1.6 1.8 1.0 1.2 1.4 1.6 1.8
1.0 1.2 1.4 1.6 1.8 1.0 1.2 1.4 1.6 1.8 1.0 1.2 1.4 1.6 1.8
ky (Å-1) -1
ky (Å ) -1
ky (Å )
F IGURE 10.11: Experimental ARPES data comparison to EPWE photoemission in-

tensity simulations. The top row (a, b) shows the 2nd derivative of the ARPES and
EPWE isoenergetic cuts (kx vs ky ) performed at the top of the VB (E=-1.98 eV in
ARPES). The middle row (c-e) corresponds to the 2nd derivative of the experimen-
tal ARPES spectral intensities (E vs ky ) values (indicated as dashed lines in (a)).
The bottom row (f-h) shows the equivalent photoemission intensity simulations
performed with the EPWE code. Their match is exceptional, supporting the idea
of weak electronic interactions between the zigzag chains with the substrate and
surrounding Br atoms.
10.4.6 ARPES Band Structure Comparison of Zigzag Chains vs PPP Chains.
The occupied experimental band structure of the zigzag chains is compared with
the straight PPP chains in Figure 10.12. The PPP polymer film has likewise been
grown using a vicinal Ag substrate with (544) orientation and featuring an aver-
age terrace width of 2.4 nm terminated by straight steps rotated 30◦ from the fully
kinked curved Ag(111) [see Figure 10.8 (a)]. Indeed, the step direction of Ag(544)
follows the [11̄0] direction and the terraces are slightly smaller than the one used for
the zigzag, but can still host in average 3-4 parallel PPP chains. The difference in the
electronic nature between both chain types is evident from the dataset and closely
follows the DFT calculations reported in Figure 10.4. For the zigzag chains a kx inte-
2π
grated band structure is presented in order to capture the replicating bands every L
[Figure 10.12 (b)] where L corresponds to the superperiodicity of the zigzag polymer
[Figure 10.12 (a)]. In contrast, PPP chains show a highly dispersive electronic band
2π
in the same energy window [Figure 10.12 (d)], peaking at a where a corresponds to
the phenyl-phenyl distance along the polymer [Figure 10.12 (c)].
a b kx integrated RD
2π/L kx integrated 2π/a*
N=4 zigzag 0.0 0.0
L = 2.24 nm -0.5 -0.5
a*
-1.0
(kx integrated)
-1.0
E-EF (eV)
-1.5 -1.5
kx
-2.0 -2.0
ky -2.5 -2.5
-3.0
-3.0 -3.0
-3.5 -3.5
c d 0.0 0.0 0.5 1.0 1.50.0 0.0 0.5 1.0 2π/a
1.5
PPP
-0.5 -0.5
a
-1.0
-1.0 -1.0
(kx = 0 Å-1)
E-EF (eV)
-1.5 -1.5
kx -2.0
-2.0 -2.0
-2.5 -2.5
ky -3.0
-3.0 -3.0
-3.5
-3.5 -3.5
0.0
0.0 0.5 1.0
0.5 1.5 0.0 0.5 1.0 1.5
1.5
ky (Å-1) ky (Å-1)
F IGURE 10.12: N =4 zigzag chain ARPES band structure comparison with that of
PPP. Schematic representations of the N =4 zigzag chains (a) and the poly-(para-
phenylene) chains (c). (b) and (d) show their respective raw (left) and second
derivative (right) of the ARPES spectral intensity (E vs ky ) along the chain axis.
These plots follow the periodicity of the chain unit cells ( 2π π
L and a ) marked with
vertical dashed blue lines. For the PPP polymer film the substrate used is a Ag(544)
vicinal crystal.
148
10.4.7 Band Structure Variations with Straight Segment’s Length of the

Zigzag Chains from DFT Calculations
To further investigate the evolution of the frontier orbtial bandgap and electronic
structure with the straight segment’s length, we perform additional DFT calcula-
tions. As indicated before and depicted in Figure 10.13, zigzag chains with N =4 bear
a closer resemblance to poly-(meta-phenylene) rather than to the straight, highly dis-
persive PPP counterpart. Moreover, a gradual reduction of the electronic bandgap
of the frontier orbitals is already observed (PMP > zigzag N =4 > zigzag N =7 > PPP)
as the straight segments are enlarged. Such an evolution is clearly shown in Fig-
ure 10.14 for a varying number of phenyl rings from 4 up to 11 in a planar, free-
standing configuration. In agreement with previous work [65], we find that the band
1
gap shrinks linearly as N as the size of the straight segment is increased. This also
agrees with our experimental results of Figure 10.6 (c), which are represented on
the right axis of the graph. This evidences that while the electronic band dispersion
is governed by the periodically spaced meta-junctions, the bandgap varies with the
size of the straight segments.
PMP N=4 Zigzag N=7 Zigzag PPP
1
E-EF (eV)
-1
-2
G π/L G π/L G π/L G π/a
F IGURE 10.13: Band structure comparison between PMP, zigzag and PPP chains, as
obtained by DFT calculations. The highly dispersive band structure of PPP chains
(on the right) clearly show a different behavior to the other three cases. These plots
show a progressive reduction of the electronic bandgap as the meta-junctions (el-
bows) are separated further. We attribute this to the cross-conjugated nature of the
polymers hosting periodically spaced meta-junctions.
N=4 DFT
2.4
2.4
STS 3.8
3.8
Bandgap from DFT (eV)
Bandgap from STS (V)

2.3
2.3 3.7
3.7
N=7
2.2
2.2 3.6
3.6
2.1
2.1 N=8 N=10 3.5
3.5
N=9
N=11
2.0
2.0 3.4
3.4
1.9
1.9 3.3
3.3
0.25
0.25 0.20
0.20 0.15
0.15 0.10
0.10
1/N
F IGURE 10.14: Evolution of the zigzag chain frontier orbital bandgap with increas-
ing number of phenyl rings at the straight segments. Both DFT (theory) and STS
(experimental) show a clear N1 dependency of the bandgap with increasing number
of phenyl rings inside the straight segments.
151
Chapter 11
Conclusions and Outlook
In the following, the ending conclusions from each chapter are collected and merged
into a general and global conclusion section for each separate part. As this work does
not end at this point, an outlook is included to provide hints of the interesting inves-
tigation paths to be performed for these research lines in the future.
11.1 Conclusions to Part II: Nanoporous Networks

In the second part of this thesis we have studied the electronic confinement and QD
intercoupling induced by organic and metal-organic nanoporous networks. This has
been possible by the experimental characterization of six different self-assembled
molecular nanoporous networks grown on Cu(111), Ag(111) and Au(111) substrates
as well as on thin Ag films. We have accomplished the generation of long-range or-
dered nanoporous networks in UHV conditions that fully saturate the surface. This
is achieved through performing a careful gradient coverage evaporation process
and selecting the proper molecule-molecule interactions in the form of self-healing
bonding mechanisms (halogen bonding and metalorganic coordination) as well as
picking out the right molecule-substrate interactions and commensuration. These
represent essential prerequisites to access their experimental 2DEG electronic band
structures with ARPES, as summarized in Figure 11.1. In addition, the LDOS yield-
ing confinement inside each single nanopore have been probed with STM/STS mea-
surements through collaborations with different international groups, which greatly
complement the ensemble electronic properties accessed by photoemission.
Simulations by the semi-empirical electron boundary element method in combi-
nation with the electron plane wave expansion has been extensively used through-
out this thesis. This has allowed us to model and understand the scattering poten-
tial landscape of each nanoporous network [Table 11.1] and simulate the LDOS and
electronic band structures [Figure 11.1]. In addition, density functional theory and
the phase accumulation model have been used to capture subtle overlayer-substrate
interactions such as hybridization effects and geometrical variations. All these the-
oretical methods have been obtained through collaborations. Since EBEM/EPWE
is a 2D modelling system, vertical interactions can not be directly captured but are
152 Chapter 11. Conclusions and Outlook
a DPDI Soleil 2Der GM reduced x d P6 kr 2Der GM
0.1
0.1 0.0 0.0
0.0
0.0 -0.1 -0.1
E-EF (eV)
-0.1
E-EF (eV)
-0.1 -0.2 -0.2
E-EF (eV)
-0.2
-0.2 -0.3 -0.3
-0.3
-0.3 -0.4 -0.4
-0.4
-0.4 -0.5 -0.5
     5 nm
2 nm -0.5
-0.5 -0.6 -0.6
      
-0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 -0.2
-0.2
-0.1
-0.1
0.0
0.0 -1
0.1
0.1
0.2
0.2
b kx (Å-1) e
kII (AGM
P3 kr 2Der )
0.0 0.0
-0.1 -0.1
0.0
E-EF (eV)
-0.2 -0.2
E-EF (eV)
E-EF (eV)
-0.1 -0.3 -0.3
-0.2 -0.4 -0.4
     -0.5 -0.5
5 nm
-0.3 -0.6  -0.6
   
-0.2 -0.1 0.0 0.1 0.2 -0.2
-0.2 -0.1-0.1 0.0
0.0 0.1
0.1 0.2
0.2
c
-1
kx (Å-1) f kII (A )
0.0
0.0
0.0
E-EF (eV)
-0.2
-0.2
E-EF (eV)
-0.1
-0.4
-0.4
-0.2
     -0.6
-0.6
    
-0.3
-0.2 -0.1 0.0 0.1 0.2
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 0.3
0.4 0.4
0.6
kx (Å-1) kx (Å-1)
F IGURE 11.1: ARPES electronic band structure of the studied nanoporous networks
in a nutshell. (a) 3deh-DPDI metal-organic network on Cu(111). (b, c) Single Wall
(SW) and Double Wall (DW) halogen bond stabilized nanoporous networks grown
on 3 ML Ag/Au(111) for ARPES and Ag(111) for STS. (d, e) Two homothetic (scal-
able) Ph6Co and Ph3Co metal-organic nanoporous networks grown on Au(111). (f)
Cu-TPyB metal-organic network grown on Cu(111).
accounted for as renormalization of the 2DEG. Our complete experimental and the-
oretical synergy provides an accurate understanding of the global electronic prop-
erties of the studied nanoporous networks and allows us to predict the electronic
properties of the broad family of single layer organic nanoporous materials grown
on metallic substrates.
The main conclusions from these organometallic nanoporous networks are the
following:
1. Partially localized states in nanoporous networks originate from the Shock-

ley state.
In Chapter 4 we validate the origin and nature of the partially localized states
that are generated in nanoporous networks. By studying the temperature
dependence of the band structure generated by the 3deh-DPDI nanoporous
network, we corroborate that the n=1 PLS existing in the pores of the net-
work originates from the pristine Cu(111) Shockley state [Figure 11.1 (a)]. The
energies of both states, i.e. n=1 PLS and Cu SS, display the same tempera-
ture dependence, evidencing their common origin and nature. This experi-
mental demonstration supports that the surface state electrons are affected by
the periodic scattering potential landscape exerted by molecular barriers of a
nanoporous network, partially localizing them inside each nanocavity.
11.1. Conclusions to Part II: Nanoporous Networks 153
TABLE 11.1: ARPES fundamental binding energies and extracted effective masses
(first two columns). EBEM/EPWE required 2DEG renormalized energy and masses
(third and fourth columns) and scattering potential values (last two columns) to
match the experimental data.
Γ
EB m∗ /m0 Ref,Γ
EEBEM m∗,Ref
EBEM /m0 Vmol Vmet
(eV) (eV) (meV) (meV)
3deh-DPDI/Cu(111) 0.25 0.57 0.40 0.49 390 340/440
SW/ 3ML Ag/Au(111) 0.12 0.47 0.16 0.49 140 –
DW/ 3ML Ag/Au(111) 0.11 0.59 0.16 0.54 140 –
Ph6Co/Au(111) 0.49 0.24 0.52 0.24 250 50
Ph3Co/Au(111) 0.55 0.22 0.59 0.21 250 50
Cu-TPyB/Cu(111) 0.47 0.41 – – – –
Au(111) 0.45 0.255 0.45 0.26 – –
3ML Ag/Au(111) 0.16 0.38 0.16 0.38 – –
Cu(111) 0.40 0.42 0.40 0.42 – –
2. The scattering potential landscape generated by organic and metal-organic

nanoporous networks represents a repulsive scattering potential for surface
electrons.
The 3deh-DPDI metal-organic nanoporous network is used in Chapter 7 to
show that it is possible to unambiguously obtain the scattering potential land-
scape exerted by a molecular nanoporous overlayer onto a 2DEG and deter-
mine the relevant confinement details and interaction effects. We show that the
scattering potential must be parametrized as realistically as possible to the net-
work geometry while the flexibility is provided to the 2DEG reference, which
requires a slight energy and/or mass renormalization due to the interactions
between overlayer and substrate [Table 11.1]. Our simulations unambiguously
define that both the molecules and the metal adatoms forming the network
exhibit repulsive character to the surface electrons. We also find that slight
perturbations in the scattering potential at the metal coordination sites are re-
sponsible for the deformation of the confined states spatial shape. We note
here that the higher order states are more sensitive than the lower ones to de-
viations from the correct values used for the scattering potentials.
3. Electronic states in nanoporous networks can be individually and globally

configured with physisorbed Xe adsorbates.
We have demonstrated in Chapter 8 that it is possible to engineer the electronic
structure of the n=1 PLS by selectively adsorbing Xe guests atoms inside the
nanocavities. When globaly adsorbing Xe atoms into all nanocavities, a Pauli
repulsion induced ∼ 60 meV shift of the n=1 PLS to lower binding energies is
detected in STS and corroborated by ARPES band structure measurements. In
addition, we can engineer different interpore coupling configurations by local

reposition or removal of 12 Xe atoms with the STM tip. Such configurations
yielded distinct localized states, while the electronic features of the surround-
ing identical QDs remained unaffected. Indeed, residues from the bonding
part of the surrounding PLS can still be detected in the single pore LDOS, al-
lowing us to delve into the intricacies of interpore coupling effects.
4. The quantum dot intercoupling degree can be precisely engineered in nano-

porous networks.
In Chapter 5, we precisely engineer the QD intercoupling by altering QDs bar-
rier width (from single molecule barrier to a double one) without affecting the
QD’s size [Figure 11.1 (d, e)]. Halogen-bond stabilized organic nanoporous
networks are generated on bulk Ag(111) and 3 ML Ag/Au(111) thin films alike
by substitution of a single atom (sulphur vs oxygen) in the precursor molecule.
Our findings show that the reduction of the QD intercoupling is associated
with a flattening of the band dispersion and an increase of the effective mass
that is mirrored by a narrowing of the asymmetric peak widths detected in STS
[Figure 11.1 (d, e) and Table 11.1].
5. The 2DEG can undergo an energy and mass renormalization upon the pres-
ence of nanoporous networks.
Perhaps one of the most surprising result is found in Chapter 6, where we
study the electronic properties of two homothetic (similar) Co-coordinated
metal-organic nanoporous networks [Figure 11.1 (d, e)]. Our results evidence a
gradual and unusual energy shift of the Au(111) SS to higher binding energies
and a lightening of the effective mass concomitant with a charge preservation
upon network formation. Notably this downshift is gradual with decreasing
pore size and occurs in spite of the demonstrated confining attributes of the
nanocavities [Table 11.1].
Overlayer-substrate interactions must be responsible for this effect on the Au
SS reference. DFT calculations and phase accumulation model simulations
help us to disentangle possible mechanisms responsible for these downshifts,
where Au SS-Co adatom hybridization and Co/molecule vertical displace-
ments appear as the most plausible causes. Similar renormalization effects are
also observed on a Cu-coordinated metal-organic nanoporous network [Fig-
ure 11.1 (f)], suggesting that this traditionally overlooked effect could be rather
general on MOCN with single coordination atoms.
11.2. Conclusions to Part III: Zigzag Chains 155
11.2 Conclusions to Part III: Zigzag Chains

In the third part of this thesis, we have used the on-surface Ullmann coupling re-
action to synthesize and macroscopically align a saturated film of cross-conjugated
zigzag chains with periodically spaced meta-junctions on a vicinal Ag(111) surface.
ARPES measurements reveal weakly dispersing 1D electronic bands along the
chain direction [Figure 11.2 (a)]. This band structure constrasts with the highly dis-
persive one of poly-(para-phenylene) chains and resembles more the one of poly-
(meta-phenylene), demonstrating that the electronic band dispersion is governed by
the periodically spaced meta-junctions.
We find that zigzag chains present a larger frontier orbital bandgap than the
straight PPP counterparts (3.7 eV vs 3.2 eV in ref. [29]) and show electron confine-
ment in each straight segment, reminiscent of 1D arrays of weakly interacting QDs
[Figure 11.2 (b)]. Such molecular states can be tuned by changing the length of these
1
straight segments, affecting the frontier orbital bandgap, which follows a N depen-
dency (where N is the number of phenyl rings per straight section).
These findings corroborate the important effects that the conductive path topol-
ogy of a molecular wire has on its frontier orbitals, which are responsible for defining
its chemical, optical and electronic properties.
a 2π/L 2π/a* b
kx integrated Sample Bias (V)
0.0
0.0
-2.5
-2.0
-1.5
-1.0
-2.5 -2.0 -1.5 -1.0 1.0 1.5 2.0 2.5

1.0
1.5
2.0
2.5
-0.5
-0.5 kx
N=4 straight segment
ky
(kx integrated)
-1.0
-1.0
E-EF (eV)
-1.5
-1.5 aaa
bbb
-2.0
-2.0
-2.5
-2.5
-3.0
-3.0
VB CB
-3.5
-3.5
0.0 0.5 1.0 1.5
ky (Å-1)
F IGURE 11.2: Zigzag chain electronic structure probed by (a) ARPES and (b) STS.
Shallow dispersive bands evidence the loss of electronic coupling between neigh-
boring straight segments, which are concomitant with electron localization effects,
reminiscent of QD 1D arrays.
11.3 Outlook
Metal-organic frameworks (MOFs) are an important class of materials that present
intriguing opportunities in the fields of sensing, gas storage, catalysis and opto-
electronics [283–286]. MOCNs grown on metallic surfaces are also known to show
magnetic properties [4, 5], catalytic effects [6, 7], oxidation states [8], exotic tesella-
tion patterns [9–12] and even bear the prospect of exhibiting topological band struc-
tures [13, 14]. As an outlook for the near future, the following complementing re-
search lines to this thesis are already initiated:
• Magnetic Nanoclusters on MOCNs. Highly ordered and robust MOCNs could

be used as extended templates for controlling the ordering and size of mag-
netic atom nanoclusters and as a consequence, to tune their magnetic proper-
ties. The aim would be to induce magnetic atoms to self-assemble within the
pores into ordered nanoclusters [287]. We have recently observed that Fe tri-
angular islands grown on Cu(111) translate into well-ordered nanoclusters of
controllable size when deposited on top of the Cu-DCA MOCN [Figure 11.3 (a-
c)]. The next step is to study the size of these nanoclusters as well as their re-
spective magnetic signal and compare them to the case where there is no MOF
presence.
a b c
6.0nm 6.0nm 6.0nm
F IGURE 11.3: Fe nanocluster size and assembly control when grown on Cu-DCA
MOCN. (a) Fe double atomic height triangular shape islands grown on Cu(111).
(b) Long-range ordered Cu-DCA MOCN grown on Cu(111). (c) Fe nanocluster
formation and stabilization when grown on top of Cu-DCA MOCN. This project
is done in collaboration with Leyre Hernández and Dr. Fernando Bartolomé from
ICMA (CSIC-University of Zaragoza).
• Experimental Validation of Organic Topological Insulators (OTIs). Recently,

2D MOCNs have been proposed as a flexible material platform for realizing
exotic quantum phases including topological and anomalous quantum Hall
insulators [14, 288–292]. For example, it is possible to realize honeycomb and
kagome lattices that are expected to give rise to Dirac cones and flat bands.
The DCA-Cu MOCN is one of these examples [Figure 11.4 (a)]. If a strong
spin-orbit coupling (SOC) is introduced by, for instance, using heavy elements
11.3. Outlook 157
a b
c
0.0
0.0
E-EF (eV)
ss
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2 sp-band
Cu(111)
d -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.0
0.0
E-EF (eV)
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
Cu-DCA
e 0.2
0.2 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.1
0.1
kx (Å-1)
0.0
0.0
-0.1
-0.1
-0.2
-0.2
E=-0.45 eV
Cu(111)
f 0.2 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.1
kx (Å-1)
0.0
-0.1
-0.2 E=-0.45 eV
Cu-DCA
-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

ky (Å-1)
F IGURE 11.4: Metal-organic frameworks as organic topological insulators. (a)

Schematic atomic structure of Cu-DCA film. The inset shows the DCA molecule.
The red dashed , blue dashed, and black lines outline the honeycomb lattice by the
Cu atoms, the Kagome lattice by the DCA molecules, and the unit cell, respectively
(adapted from ref. [14]). (b) 10 × 10 nm2 STM image of the Cu-DCA film grown
on Cu(111). Its single domain, long-range order and low defect concentrations are
ideal for ARPES band structure measurements. STM parameters: I=300 pA; V=-1
V. (c,d) ARPES raw band structure measurements (E vs ky ) of the pristine Cu(111)
surface and Cu-DCA film along the ΓK high symmetry direction of both substrate
and network. (e,f) Isoenergetic cuts (kx vs ky ) performed at E=-0.45 eV for the
pristine Cu(111) and Cu-DCA films respectively. The sp-band appears replicated
following the DCA superperiodicity.
as Bi atoms in coordination, this should result in the opening of topologically

non-trivial band gaps and the realization of OTIs with topological edge states.
Experimentally, direct synthesis of 2D MOCNs has been essentially confined
to metal substrates [9]. For instance, we observe that the DCA-Cu MOCN
forms a single domain, long-range ordered film on Cu(111) [Figure 11.4 (b)].
ARPES band structure measurments on this particular MOCN (E vs ky ) clearly
show how the sp-band of the Cu(111) surface [Fig. 11.4 (c)] appears replicated
following the superperiodicity of the MOCN [vertical dahsed black lines in
Fig. 11.4 (d)]. In addition, some intensity is detected at -0.45 eV, reminiscent of
non-dispersive bands. However, isonergetic cuts (kx vs ky ) performed at this
energy only highlight the replicating sp-bands without any additional feature
[cf. Figure 11.4 (e, f)]. This preliminary results suggest that the interaction with
the metallic substrate most probably masks the intrinsic electronic properties
of the MOCN [293, 294]. Therefore, depositing a 2D-OTIs on a 2D insulating
substrate such as hBN or a decoupling layer such as graphene may protect the
TI properties from quenching.
Regarding the on-surface synthesis and high-alignment of oligophenylene chains

and GNRs on vicinal surfaces, the following research lines are proposed:
• Image Potential States (IPS) in GNRs and Oligophenylene Chains. Exotic

super atom states have been detected inside graphene nanopores and bay-
shape GNR edges [24, 277]. These states are neither related to atomic orbitals
nor to the 2DEG confinement from the metallic substrate [295]. Instead, they
originate from the free electron–like image potential states (IPS) that are con-
fined at the vacuum side along the GNR edge. They can be regarded as the 2D
analog of the super atom states that develop when a graphene sheet bends into
a fullerene [296, 297]. In a similar fashion to the multibay-shape GNRs giving
rise to rather flat super atom bands [24], zigzag polymers host an ideal ge-
ometry with periodic elbows for confining such IPS and coupling them in the
long-range. Preliminary DFT calculations on the N =4 zigzag chain band struc-
ture [Figure 11.5 (a)] and molecular orbital simulations [Figure 11.5 (b)] at the
expected IPS bands [purple and green arrows in Figure 11.5 (a)] evidence con-
finement of the electron wavefunction. Such localized states observed at the
elbow regions may well correlate with the high intensities we observe in the
high-resolution dI/dV map at 2.3 V [Figure 11.5 (c)], in agreement with previ-
ous observations on similar systems [24, 277]. Such IPS could be further tuned
and their coupling altered by changing the separation between meta-junctions
either by synthesizing a pure poly-(meta-phenylene) chain or by enlarging the
straight segments lengths as well as by functionalizing the DMTP precursors
with nitrogen atoms [26].
11.3. Outlook 159
a b
4
E-EF (eV)
2
c +2.3 V
1
G X
F IGURE 11.5: Image potential states in cross-conjugated oligophenylene zigzag

chains. (a) DFT calculated electronic band structure for free-standing zigzag poly-
mers. The possible IPS state coupling induced shallow dispersive bands are high-
lighted in red. (b) The spatially resolved molecular orbitals at Γ for both bands
[purple and green arrows in (a)] show the electron wavefunction localized in the
concave region of each elbow of the zigzag chain. Interestingly, its spatial position
spreads outside the carbon backbone. (c) High-resolution constant height dI/dV
map at 2.3 V shows high intensity collimation at the elbow sites of the zigzag poly-
mer, in resemblance to the predicted IPS.
• Generation of 1D and 2D ordered structures by the right choice of substrate

and precursors. Recent advances in transfer techniques of on-surface syn-
thesized organic nanostructures onto insulating substrates [24, 32] opens the
path towards further studying transport and optical properties of the highly-
aligned cross-conjugated oligophenyelene zigzag chains. Most importantly,
since the alignment of these polymers or GNRs into films and their direction-
ality may be a necessary prerequisite to conserve during transfer and device
fabrication, the use of vicinal surfaces not only would be mandatory for pho-
toemission experiments but also could serve as an alignment tool for GNR
and polymer based device fabrication, provided that the nanotemplate is re-
peatedly conserved [33].
The development in chemical synthesis has lead to the synthesis of graphene
nanoribbons with different edge topologies. By tuning the edge termination
from armchair type to zigzag, edge states have been observed [28]. In addi-
tion, recently GNRs with non-trivial topology have been grown on metallic
substrates, giving rise to exotic topological states [25, 30]. By using the right
vicinal surface, a desirable macroscopic growth and alignment of such GNRs

would allow for the experimental observation of their bulk and edge band
structure in ARPES, greatly complementing the already available STS infor-
mation.
The current understading on molecular self-assembly for building nanoporous
networks and on-surface synthesis methods for builiding covalent 1D and 2D
nanostructures sets the grounds for designing suitable molecular precursors
for the formation of long-range ordered COFs. The main current issue is to
reduce the defect concentration appearing as tetragonal, pentagonal and hep-
tagonal pores when targeting the formation of a honeycomb COF from three-
fold halo-aromatic precursors [63, 75, 298]. Perhaps using four-fold symmetric
precursors may give rise to the formation of square-shaped COFs with a lower
defect concentration [69, 299].
161
Appendix A
Resumen de las Conclusiones

Principales
A.1 Estructura Electrónica de Redes Orgánicas Nanoporosas

Generadas en Superficies de Metales Nobles
En la segunda parte de la tesis se ha estudiado las propiedades de confinamiento
electrónico y las bandas electrónicas generadas a partir del acoplamiento entre pun-
tos cuánticos (QDs) existentes en las redes nanoporosas orgánicas y metal-orgánicas
generadas sobre superficies metálicas. Esto ha sido posible gracias a la caracteri-
zación experimental de hasta seis redes nanoporosas autoensambladas sobre sub-
stratos metálicos tales como Cu(111), Ag(111) y Au(111) además de heteroestruc-
turas metálicas compuestas por 3 capas atómicas de Ag en Au(111), todo ello en
condiciones controladas de ultra alto vacío (UHV).
Las redes nanoporosas con orden a largo alcance, aquellas que recubren por com-
pleto toda la superficie, han sido generadas mediante la selección de interacciones
intermoleculares adecuadas a modo de enlaces autocorrectores, tales como enlaces
halógenos y metal-orgánicos, junto con una correcta interacción molécula-sustrato y
registro con el sustrato. Dichos sistemas se han preparado utilizando gradientes de
recubrimiento. Este modo de preparación tan complejo representa un requisito sine
qua non para observar bandas electrónicas moduladas en medidas de fotoemisión
resuelta en ángulo (ARPES). Además, mediante el uso de técnicas espectroscópicas
complementarias como la microscopía de efecto túnel (STM/STS), se han estudiado
las propiedades de confinamiento electrónico locales, es decir, en cada nanoporo o
QD, complementando de manera satisfactoria las bandas electrónicas detectadas en
fotoemisión.
El modelo semi empírico denominado método de elementos de frontera del elec-
trón (EBEM) en combinación con la expansión de ondas planas de electrones (EPWE)
se ha utilizado de manera recurrente para simular y comprender el complejo sistema
de dispersión que cada red nanoporosa ejerce sobre el gas electrónico bi-dimensional
(2DEG) presente en las superficies metálicas usadas como substrato. Esto nos ha
permitido reconstruir y corroborar tanto las complejas bandas electrónicas como las
propiedades locales de confinamiento a nivel de nanoporo mediante las simulación
162 Appendix A. Resumen de las Conclusiones Principales
a DPDI Soleil 2Der GM reduced x d P6 kr 2Der GM
0.1
0.1 0.0 0.0
0.0
0.0 -0.1 -0.1
E-EF (eV)
-0.1
E-EF (eV)
-0.1 -0.2 -0.2
E-EF (eV)
-0.2
-0.2 -0.3 -0.3
-0.3
-0.3 -0.4 -0.4
-0.4
-0.4 -0.5 -0.5
     5 nm
2 nm -0.5
-0.5 -0.6 -0.6
      
-0.4
-0.4 -0.2
-0.2 0.0
0.0 0.2
0.2 0.4
0.4 -0.2
-0.2
-0.1
-0.1
0.0
0.0 -1
0.1
0.1
0.2
0.2
b kx (Å-1) e
kII (AGM
P3 kr 2Der )
0.0 0.0
-0.1 -0.1
0.0
E-EF (eV)
-0.2 -0.2
E-EF (eV)
E-EF (eV)
-0.1 -0.3 -0.3
-0.2 -0.4 -0.4
     -0.5 -0.5
5 nm
-0.3 -0.6  -0.6
   
-0.2 -0.1 0.0 0.1 0.2 -0.2
-0.2 -0.1-0.1 0.0
0.0 0.1
0.1 0.2
0.2
c
-1
kx (Å-1) f kII (A )
0.0
0.0
0.0
E-EF (eV)
-0.2
-0.2
E-EF (eV)
-0.1
-0.4
-0.4
-0.2
     -0.6
-0.6
    
-0.3
-0.2 -0.1 0.0 0.1 0.2
-0.2
-0.2 -0.1
-0.1 0.0
0.0 0.1
0.1 0.2
0.2 0.3
0.4 0.4
0.6
kx (Å-1) kx (Å-1)
F IGURE A.1: Resumen de la estructura de bandas de las diferentes redes

nanoporosas estudiadas con ARPES. (a) Red nanoporosa metal-orgánica 3deh-
DPDI en Cu(111). (b, c) Redes nanoporosas estabilizadas con enlaces halógenos
con barreras compuestas por moléculas individuales (SW) y dobles (DW) crecidas
sobre 3ML Ag/Au(111). (d, e) Redes nanoporosas metal-orgánicas similares com-
puestas por las moléculas Ph6 y Ph3, respectivamente, en coordinación con átomos
de Co y crecidas sobre Au(111). (f) Red nanoporosa metal-orgánica compuesta por
la molécula TPyB en coordinación con átomos de Cu y crecida sobre Cu(111).
de las densidades locales de estado (LDOS). Además, los métodos de cálculo de

teoría de funcional de densidad (DFT) y modelo de acumulación de fase (PAM) han
sido utilizados para abordar en detalle las sutiles interacciones verticales en forma
de hibridización y efectos geométricos presentes en la interfase entre la red orgánica
nanoporosa y el 2DEG. Esta sinergia entre técnicas experimentales y simulaciones
teóricas nos ha aportado una profunda comprensión sobre las propiedades electróni-
cas derivadas del ensamblaje de cada red nanoporosa en particular.
Los resultados obtenidos sirven para comprender las propiedades electrónicas
de las redes nanoporososas orgánicas y metal-orgánicas crecidas sobre superficies
metálicas. Los resultados principales se pueden resumir en los siguientes puntos
fundamentales:
1. Los estados parcialmente localizados en las redes nanoporosas se originan o

provienen del estado de superficie del sustrato.
Mediante el estudio de la dependencia con la temperatura de las bandas elec-
trónicas generadas por la red metal-orgánica nanoporosa 3deh-DPDI (capí-
tulo 4), se corrobora que el primer estado parcialmente localizado (denomi-
nado n=1 PLS) existente en los nanoporos, proviene del estado de Shockley
de Cu(111). La energía fundamental de ambos estados, es decir, de n=1 PLS
y Cu SS, muestra la misma tendencia con la temperatura, lo que evidencia
A.1. Estructura Electrónica de Redes Orgánicas Nanoporosas Generadas en
163
Superficies de Metales Nobles
TABLE A.1: Resumen de las propiedades electrónicas extraídas de las medidas de

ARPES, es decir, las energías de ligadura en el punto Γ y las masas efectivas de las
respectivas redes nanoporosas estudiadas. Además se incluyen los potenciales de
dispersión, energías de referencia y masas efectivas adoptadas en las simulaciones
con EBEM/EPWE.
Γ
EB m∗ /m0 Ref,Γ
EEBEM m∗,Ref
EBEM /m0 Vmol Vmet
(eV) (eV) (meV) (meV)
3deh-DPDI/Cu(111) 0.25 0.57 0.40 0.49 390 340/440
SW/ 3ML Ag/Au(111) 0.12 0.47 0.16 0.49 140 –
DW/ 3ML Ag/Au(111) 0.11 0.59 0.16 0.54 140 –
Ph6Co/Au(111) 0.49 0.24 0.52 0.24 250 50
Ph3Co/Au(111) 0.55 0.22 0.59 0.21 250 50
Cu-TPyB/Cu(111) 0.47 0.41 – – – –
Au(111) 0.45 0.255 0.45 0.26 – –
3ML Ag/Au(111) 0.16 0.38 0.16 0.38 – –
Cu(111) 0.40 0.42 0.40 0.42 – –
su naturaleza común. Esta demostración experimental respalda el hecho de

que los electrones de superficie son muy sensibles y por ende, se ven drástica-
mente afectados por los potenciales de dispersión periódicos generados por la
red nanoporosa molecular, localizando o atrapando parcialmente dichos elec-
trones en cada nanocavidad o nanoporo.
2. El mapa de potencial de dispersión producido por las redes nanoporosas

tiene un caracter repulsivo sobre el estado de superficie, acompañado de una
renormalización de éste tanto en masa como en energía.
Mediante el uso de la red nanoporosa metal-orgánica creada por 3deh-DPDI,
en el capítulo 7 se demuestra que es posible obtener de forma fehaciente el
mapa de potencial de dispersión que representa una red nanoporosa para el
2DEG sobre el cual se ha generado. De esta manera se pueden simular de
manera detallada y rigurosa los efectos de confinamiento electrónico e interac-
ciones más relevantes. Demostramos que el mapa de potencial de dispersión
se tiene que generar o parametrizar siguiendo detalladamente la geometría de
la red nanoporosa, permitiendo un grado de flexibilidad en la referencia del
2DEG. Esto implica que se require de una renormalización en masa y energía
causada por interacciones entre el gas electrónico y la red orgánica. Nuestras
simulaciones asignan, sin ninguna duda, un carácter repulsivo tanto a molécu-
las como a los metales de coordinación, sirviendo ambos como barreras repul-
sivas para los electrones de superficie. Además, los estados confinados en la
red 3deh-DPDI, aparecen deformados, adoptando una disposición triangular
en algunos casos, en vez de la estructura hexagonal esperada. Estas distor-
siones son debidas a pequeñas perturbaciones en el potencial de dispersión de
los metales de coordinación, que debido a su simetría triangular, provocan un

cambio espacial en los estados electrónicos.
3. Los estados electrónicos en redes nanoporosas se pueden configurar tanto

individual como globalmente mediante la fisisorción de gases nobles como
el Xe.
En el capítulo 8 se demuestra que es posible manipular de forma controlada
la estructura electrónica del n=1 PLS en la red nanoporosa 3deh-DPDI medi-
ante la absorción global y simultanea de 12 átomos de Xe en cada nanoporo
de la red. Además, mediante manipulación individual de átomos de Xe con
la punta del STM, es posible obtener diferentes configuraciones de ocupación
para alterar los estados localizados de manera individual. En el primero de
los casos, mediante medidas de espectroscopía de efecto túnel, observamos
que los átomos de Xe generan una interacción denominada repulsión de Pauli,
mediante la cual el n=1 PLS presenta un cambio de 60 meV hacia energía de
ligadura menores. Estas observaciones locales se corroboran mediante medi-
das de ARPES, evidenciando un cambio similar en la energía fundamental de
la banda electrónica correspondiente a n=1 PLS. En el segundo caso, se pro-
cede a alterar las configuraciones a primeros vecinos: un poro vacío rodeado
de poros llenos de Xe y viceversa. Estas configuraciones artificialmente con-
struidas muestran que en el poro manipulado residen estados localizados a
energías más negativas y de mayor anchura de pico para el primer caso y esta-
dos muy localizados a menores energías para el segundo de los casos. Dichos
cambios, no afectan al estado electrónico de los poros vecinos pero sí que se ob-
servan la cola de los mismos en las posiciones manipuladas. Estos resultados
nos aclaran algunos conceptos dentro de la complejidad de las interacciones
entre poros existentes en este tipo de redes nanoporosas.
4. Las redes nanoporosas permiten controlar el acoplamiento electrónico entre

QDs con un alto grado de precisión.
En el capítulo 5 mostramos que somos capaces de alterar de forma controlada
el grado de interacción entre puntos cuánticos (QDs) modificando el grado de
comunicación o acoplamiento entre QDs perfectamente ordenados en una red.
Esto se ha conseguido mediante la modificación de la anchura de la barrera
entre QDs, pasando de una sola molécula en un caso, a dos moléculas en el
segundo caso, sin afectar el tamaño del QD [Figuras A.1 (b, c)]. Ambas re-
des nanoporosas se estabilizan mediante enlaces halógenos y se generan tanto
en Ag(111) como en 3 ML Ag/Au(111). La diferencia entre ambos sistemas
nanoporosos reside, únicamente, en el uso de un mismo precursor salvo por el
canje de un atomo de azufre por uno de oxígeno.
La combinación de técnicas experimentales de ARPES y STM/STS junto con
A.1. Estructura Electrónica de Redes Orgánicas Nanoporosas Generadas en
165
Superficies de Metales Nobles
simulaciones realizadas con EBEM/EPWE, aportan una visión completa desde

el punto de vista fundamental sobre la naturaleza de los procesos de acoplamiento
entre QDs. Nuestros hallazgos demuestran de forma fehaciente que la reduc-
ción en el acoplamiento entre QDs está diréctamente asociada con una pla-
narización en la dispersión de la banda electrónica generada y un incremento
de la masa efectiva del electrón. Estos efectos son respaldados por las obser-
vaciones de STS, donde el pico muestra un carácter asimétrico correspondi-
ente a los estados parcialmente localizados en QDs, los cuales sufren un es-
trechamiento debido a la disminución en el acoplamiento.
5. Los 2DEG presentes en las superficies metálicas pueden experimentar una

renormalización de su energía y masa a causa de las redes nanopororos orgáni-
cas y metal-orgánicas ensambladas encima.
Posiblemente el resultado más llamativo se presenta en el capítulo 6, donde
hemos estudiado las propiedades electrónicas de dos redes nanoporosas metal-
orgánicas similares, ambas coordinadas con átomos de cobalto [Figuras A.1 (d,
e)]. Nuestros resultados evidencian un cambio gradual de la energía funda-
mental hacia mayores energías de ligadura acompañado de un aligeramiento
en la masa efectiva del estado de superficie de Au(111) simultáneo a una preser-
vación de la carga total del 2DEG. Notablemente, estos cambios graduales
en energía y masa están directamente relacionados con una disminución del
tamaño del nanoporo entre ambos sistemas y son compatibles con las observa-
ciones de confinamiento electrónico dentro de cada nanoporo.
Las interacciones entre las redes nanoporosas y el 2DEG deberían de ser los re-
sponsables de los cambios de referencia observados. Los cálculos teóricos real-
izados con DFT y PAM nos ayudan an entender mejor los posibles mecanismos
involucrados en estos efectos electrónicos, siendo la hibridización entre los ató-
mos de Co y el 2DEG, además de posibles desplazamientos verticales de las
moléculas, las causas más plausibles. En este sentido, hemos observado simi-
lares efectos de renormalización en otra red nanoporosa metal-orgánica, pero
esta vez, utilizando átomos de coordinación de cobre [Figura A.1 (f)], lo que
sugiere que, estos efectos que tradicionalmente han pasado desapercibidos,
podrían ser comunes a redes orgánicas nanoporosas coordinadas con un sólo
átomo.
A.2 Estructura Electrónica de Cadenas Orgánicas Covalentes

Unidimensionales Sintetizadas en Superficie
En la tercera parte de la presente tesis, hemos hecho uso de la reacción de Ullmann en
un cristal vecinal de Ag(111) para sintetizar y alinear una película constituida por ca-
denas de geometría zigzag compuestas de uniones meta periódicamente espaciadas.
Las medidas de ARPES revelan bandas electrónicas unidimensiones muy poco dis-
persivas a lo largo del eje pricipal de las cadenas [Figura A.2 (a)]. Esta estructura de
bandas contrasta con la estructura de bandas altamente dispersiva ya conocida para
el caso del poly-(para-fenileno) (PPP) y se asemeja, en cambio, al poly-(meta-fenileno)
(PMP), demostrando claramente que la estructura de bandas está gobernada y con-
trolada por las uniones meta periódicamente espaciadas.
Mediante mediadas espectroscópicas locales sobre las cadenas zigzag crecidas
en Ag(111), observamos que el tamaño de energía prohibida entre los orbitales fron-
tera es de 3.7 eV, sensiblemente mayor que el ya conocido para PPP (3.2 eV [29])
[Figura A.2 (b)]. Además, detectamos indicios claros de confinamiento electrónico
en cada segmento recto comprendido entre uniones meta, con lo que el polímero en
forma zigzag se asemeja a un sistema unidimensional de QDs débilmente acopla-
dos. Estos estados moleculares se pueden modular cambiando la longitud de los
segmentos rectos, los cuales están estrictamente definidos por el número de fenilos
(N ) que la componen, alterando en consecuencia el tamaño de energía prohibida del
1
sistema con una dependencia de N.
Estos hallazgos corroboran los efectos significativos que el recorrido topológico
tiene sobre la conductividad electrónica y los orbitales frontera, los cuales son re-
sponsables de definir las propiedades químicas, ópticas y electrónicas de la propia
cadena.
a 2π/L 2π/a* b
kx integrated Sample Bias (V)
0.0
0.0
-2.5
-2.0
-1.5
-1.0
-2.5 -2.0 -1.5 -1.0 1.0 1.5 2.0 2.5

1.0
1.5
2.0
2.5
-0.5
-0.5 kx
N=4 straight segment
ky
(kx integrated)
-1.0
-1.0
E-EF (eV)
-1.5
-1.5
aaa
bbb
-2.0
-2.0
-2.5
-2.5
-3.0
-3.0
VB CB
-3.5
-3.5
0.0 0.5 1.0 1.5
ky (Å-1)
F IGURE A.2: Estructura electrónica de las cadenas zigzag obtenidas mediante

(a) ARPES y (b) STS. Las bandas poco dispersivas evidencian una pérdida del
acoplamiento electrónico con los segmentos rectos adyacentes, lo cual coincide con
efectos de confinamiento que son reminiscentes de redes de QDs unidimensionales.
167
Appendix B
List of Publications
1. L. A. Miccio, M. Setvin, M. Müller, M. Abadía, I. Piquero-Zulaica, J. Lobo-

Checa, F. Schiller, C. Rogero, M. Schmid, D. Sánchez-Portal, U. Diebold, J.E. Or-
tega. Interplay between Steps and Oxygen Vacancies on Curved TiO2(110)
NanoLetters 16, pp. 2017-2022 (2016)
2. S. Nowakowska, A. Wäckerlin, I. Piquero-Zulaica, J. Nowakowski, S. Kawai,

C. Wäckerlin, M. Matena, T. Nijs, S. Fatayer, O. Popova, A. Ahsan, S. Fatameh
Mousavi, T. Ivas, E. Meyer, M. Stöhr, J. E. Ortega, J. Björk, L. H. Gade, J. Lobo-
Checa, T. A. Jung. Configuring Electronic States in an Atomically Precise
Breadboard of Quantum Boxes Small 12, 28, pp. 3759-3763 (2016)
3. Z. M. Abd El-Fattah, P. Lutz, I. Piquero-Zulaica, J. Lobo-Checa, F. Schiller, H.

Bentmann, J. E. Ortega, F. Reinert. Formation of the BiAg2 Surface Alloy on
Lattice Mismatched Interfaces Physical Review B 94, 155447 (2016)
4. J. Brede, J. Slawinska, M. Abadia, C. Rogero, J. E. Ortega, I. Piquero-Zulaica,

J. Lobo-Checa, A. Arnau , J. Iribas Cerdá. Tuning the Graphene on Ir(111)
Adsorption Regime by Fe/Ir Surface-Alloying 2D Materials 4, 1 (2016)
5. I. Piquero-Zulaica, S. Nowakowska, J. E. Ortega, M. Stöhr, L. H. Gade, T. A.

Jung, J. Lobo-Checa. Temperature Dependence of the Partially Localized
State in a 2D Molecular Nanoporous Network Applied Surface Science 391, pp.
39-43 (2017)
6. I. Piquero-Zulaica, J. Lobo-Checa, A. Sadeghi, Z. M. Abd El-Fattah, C. Mit-

sui, T. Okamoto, R. Pawlak, T. Meier, A. Arnau, J. E. Ortega, J. Takeya, S.
Goedecker, E. Meyer, S. Kawai. Precise Engineering of Quantum Dot Array
Coupling Through Their Barrier Widths Nature Communications 8, 787 (2017)
7. M. Abadia, M. Ilyn, I. Piquero-Zulaica, P. Gargiani, C. Rogero, J. E. Ortega and

J. Brede. Polymerization of Well-Aligned Organic Nanowires on a Ferromag-
netic Rare-Earth Surface Alloy ACS nano 11, 12, 12392-12401 (2017)
8. J. E. Ortega, G. Vasseur, I. Piquero-Zulaica, S. Matencio, M. A. Valbuena, J.

Rault, F. Schiller, M. Corso, A. Mugarza and J. Lobo-Checa. Structure and
168 Appendix B. List of Publications
Electronic States on Vicinal Ag(111) Surfaces with Densely Kinked Steps

New J. Phys 20, 7, 073010 (2018)
9. I. Piquero-Zulaica, A. Garcia-Lekue, L. Colazzo, C. K. Krug, M. S. G. Mohammed,

Z. M. Abd El-Fattah, J. M. Gottfried, D. G. de Oteyza, J. E. Ortega, J. Lobo-
Checa. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
One-Dimensional Organic Semiconductors Accepted in ACS nano (2018) DOI:
10.1021/acsnano.8b06536
10. I. Piquero-Zulaica, Z. M. Abd El-Fattah, J. Li, L. Solianyk, I. Gallardo, A. Arnau,

J. E. Ortega, M. Stöhr and J. Lobo-Checa. Tunable Energy and Mass Renor-
malization from Homothetic Quantum Dot Arrays In preparation
11. I. Piquero-Zulaica, Z. M. Abd El-Fattah, S. Nowakowska, S. Kawai, O. Popova,

M. Matena, M. Enache, M. Stöhr, J. E. Ortega, L. H. Gade, T. A. Jung and J.
Lobo-Checa. Effective Determination of Surface Potential Landscapes from
Metal-Organic Nanoporous Overlayers In preparation
12. Z. M. Abd El-Fattah, M. A. Kher-Eldem, I. Piquero-Zulaica, F. J. García de

Abajo and J. E. Ortega. Graphene: Free Electron Scattering Within an In-
verted Honeycomb Lattice Submitted for publication in 2D Materials
169
Bibliography
[1] J. V. Barth, G. Costantini, and K. Kern. “Engineering atomic and molecular

nanostructures at surfaces”. In: Nature 437. 7059 (2005), pp. 671–679.
[2] L. Bartels. “Tailoring molecular layers at metal surfaces”. In: Nature Chemistry
2. 2 (2010), pp. 87–95.
[3] A. Kumar, K. Banerjee, and P. Liljeroth. “Molecular assembly on two-dimensional
materials”. In: Nanotechnology 28. 8 (2017), p. 082001.
[4] N. Abdurakhmanova et al. “Superexchange-Mediated Ferromagnetic Cou-
pling in Two-Dimensional Ni-TCNQ Networks on Metal Surfaces”. In: Phys-
ical Review Letters 110. 2 (2013), p. 027202.
[5] T. R. Umbach et al. “Ferromagnetic Coupling of Mononuclear Fe Centers in
a Self-Assembled Metal-Organic Network on Au(111)”. In: Physical Review
Letters 109. 26 (2012), p. 267207.
[6] R. Gutzler et al. “Mimicking Enzymatic Active Sites on Surfaces for Energy
Conversion Chemistry”. In: Accounts of Chemical Research 48. 7 (2015), pp. 2132–
2139.
[7] J. Čechal et al. “CO 2 Binding and Induced Structural Collapse of a Surface-
Supported Metal–Organic Network”. In: The Journal of Physical Chemistry C
120. 33 (2016), pp. 18622–18630.
[8] Y. Li et al. “Coordination and Metalation Bifunctionality of Cu with 5,10,15,20-
Tetra(4-pyridyl)porphyrin: Toward a Mixed-Valence Two-Dimensional Coor-
dination Network”. In: Journal of the American Chemical Society 134. 14 (2012),
pp. 6401–6408.
[9] L. Dong, Z. Gao, and N. Lin. “Self-assembly of metal–organic coordination
structures on surfaces”. In: Progress in Surface Science 91. 3 (2016), pp. 101–
135.
[10] J. I. Urgel et al. “Quasicrystallinity expressed in two-dimensional coordina-
tion networks”. In: Nature Chemistry 8. 7 (2016), pp. 657–662.
[11] L. Yan et al. “Self-assembly of a binodal metal–organic framework exhibiting
a demi-regular lattice”. In: Faraday Discussions 204 (2017), pp. 111–121.
[12] Yi-Qi Zhang et al. “Complex supramolecular interfacial tessellation through
convergent multi-step reaction of a dissymmetric simple organic precursor”.
In: Nature Chemistry 10. 3 (2018), pp. 296–304.
170 BIBLIOGRAPHY
[13] X. Zhang and M. Zhao. “Robust half-metallicity and topological aspects in

two-dimensional Cu-TPyB”. In: Scientific Reports 5. 14098 (2015).
[14] L. Z. Zhang et al. “Intrinsic Two-Dimensional Organic Topological Insulators
in Metal–Dicyanoanthracene Lattices”. In: Nano Letters 16. 3 (2016), pp. 2072–
2075.
[15] J. Lobo-Checa et al. “Band Formation from Coupled Quantum Dots Formed
by a Nanoporous Network on a Copper Surface”. In: Science 325. 5938 (2009),
pp. 300–303.
[16] F. Klappenberger et al. “Dichotomous Array of Chiral Quantum Corrals by a
Self-Assembled Nanoporous Kagomé Network”. In: Nano Letters 9. 10 (2009),
pp. 3509–3514.
[17] F. Klappenberger et al. “Tunable Quantum Dot Arrays Formed from Self-
Assembled Metal-Organic Networks”. In: Physical Review Letters 106. 2 (2011),
p. 026802.
[18] M. F. Crommie, C. P. Lutz, and D. M. Eigler. “Confinement of Electrons to
Quantum Corrals on a Metal Surface”. In: Science 262. 5131 (1993), pp. 218–
220.
[19] M. F. Crommie et al. “Waves on a Metal Surface and Quantum Corrals”. In:
Surface Review and Letters 02. 01 (1995), pp. 127–137.
[20] D. R. Smith. “Metamaterials and Negative Refractive Index”. In: Science 305. 5685
(2004), pp. 788–792.
[21] K. Nakada et al. “Edge state in graphene ribbons: Nanometer size effect and
edge shape dependence”. In: Physical Review B: Condensed Matter and Materials
Physics 54. 24 (1996), pp. 17954–17961.
[22] Ting Cao, Fangzhou Zhao, and Steven G. Louie. “Topological Phases in Graphene
Nanoribbons: Junction States, Spin Centers, and Quantum Spin Chains”. In:
Physical Review Letters 119. 7 (2017), p. 076401.
[23] Y.-C. Chen et al. “Molecular bandgap engineering of bottom-up synthesized
graphene nanoribbon heterojunctions”. In: Nature Nanotechnology 10. 2 (2015),
pp. 156–160.
[24] C. Moreno et al. “Bottom-up synthesis of multifunctional nanoporous graphene”.
In: Science 360. 6385 (2018), pp. 199–203.
[25] D. J. Rizzo et al. “Topological band engineering of graphene nanoribbons”.
In: Nature 560. 7717 (2018), pp. 204–208.
[26] A. Basagni et al. “Tunable Band Alignment with Unperturbed Carrier Mobil-
ity of On-Surface Synthesized Organic Semiconducting Wires”. In: ACS Nano
10. 2 (2016), pp. 2644–2651.
[27] P. Ruffieux et al. “Electronic Structure of Atomically Precise Graphene Nanorib-
bons”. In: ACS Nano 6. 8 (2012), pp. 6930–6935.
BIBLIOGRAPHY 171
[28] P. Ruffieux et al. “On-surface synthesis of graphene nanoribbons with zigzag

edge topology”. In: Nature 531. 7595 (2016), pp. 489–492.
[29] N. Merino-Díez et al. “Width-Dependent Band Gap in Armchair Graphene
Nanoribbons Reveals Fermi Level Pinning on Au(111)”. In: ACS Nano 11. 11
(2017), pp. 11661–11668.
[30] O. Gröning et al. “Engineering of robust topological quantum phases in graphene
nanoribbons”. In: Nature 560. 7717 (2018), pp. 209–213.
[31] A. Celis et al. “Graphene nanoribbons: fabrication, properties and devices”.
In: Journal of Physics D: Applied Physics 49. 14 (2016), p. 143001.
[32] J. P. Llinas et al. “Short-channel field-effect transistors with 9-atom and 13-
atom wide graphene nanoribbons”. In: Nature Communications 8. 633 (2017).
[33] M. Ohtomo et al. “Graphene nanoribbon field-effect transistors fabricated by
etchant-free transfer from Au(788)”. In: Applied Physics Letters 112. 2 (2018),
p. 021602.
[34] Z. M. Abd El-Fattah et al. “Graphene: Free electron scattering within an in-
verted honeycomb lattice”. In: arXiv:1808.06034 (2018).
[35] G. Buchs et al. “Confined electron and hole states in semiconducting carbon
nanotube sub-10 nm artificial quantum dots”. In: Carbon 132 (2018), pp. 304–
311.
[36] R. P. Feynman. “There’s plenty of room at the bottom: An invitation to enter
a new field of physics”. In: Resonance 16. 9 (2011), pp. 890–905.
[37] N. Taniguchi, ed. Nanotechnology: integrated processing systems for ultra-precision
and ultra-fine products. Oxford science publications. Oxford [England] ; New
York: Oxford University Press, 1996.
[38] V. Balzani. “Nanoscience and nanotechnology: The bottom-up construction
of molecular devices and machines”. In: Pure and Applied Chemistry 80. 8
(2008), pp. 1631–1650.
[39] B. D. Gates et al. “New Approaches to Nanofabrication: Molding, Printing,
and Other Techniques”. In: Chemical Reviews 105. 4 (2005), pp. 1171–1196.
[40] G. Binnig and H. Rohrer. “Scanning tunneling microscopy—from birth to
adolescence”. In: Reviews of Modern Physics 59. 3 (1987), pp. 615–625.
[41] D. M. Eigler et al. “Imaging Xe with a low-temperature scanning tunneling
microscope”. In: Physical Review Letters 66. 9 (1991), pp. 1189–1192.
[42] D. M. Eigler and E. K. Schweizer. “Positioning single atoms with a scanning
tunnelling microscope”. In: Nature 344. 6266 (1990), pp. 524–526.
[43] L. Bartels, G. Meyer, and K.-H. Rieder. “Controlled vertical manipulation
of single CO molecules with the scanning tunneling microscope: A route to
chemical contrast”. In: Applied Physics Letters 71. 2 (1997), pp. 213–215.
172 BIBLIOGRAPHY
[44] J.-M. Lehn. “Toward complex matter: Supramolecular chemistry and self-
organization”. In: Proceedings of the National Academy of Sciences 99. 8 (2002),
pp. 4763–4768.
[45] G. M. Whitesides. “Self-Assembly at All Scales”. In: Science 295. 5564 (2002),
pp. 2418–2421.
[46] A. Kühnle. “Self-assembly of organic molecules at metal surfaces”. In: Current
Opinion in Colloid & Interface Science 14. 2 (2009), pp. 157–168.
[47] J. V. Barth. “Molecular Architectonic on Metal Surfaces”. In: Annual Review of
Physical Chemistry 58 (2007), pp. 375–407.
[48] E. Goiri et al. “Multi-Component Organic Layers on Metal Substrates”. In:
Advanced Materials 28. 7 (2016), pp. 1340–1368.
[49] S. Conti and M. Cecchini. “Predicting molecular self-assembly at surfaces: a
statistical thermodynamics and modeling approach”. In: Physical Chemistry
Chemical Physics 18. 46 (2016), pp. 31480–31493.
[50] G. Pawin et al. “A Homomolecular Porous Network at a Cu(111) Surface”.
In: Science 313. 5789 (2006), pp. 961–962.
[51] U. Schlickum et al. “Metal-Organic Honeycomb Nanomeshes with Tunable
Cavity Size”. In: Nano Letters 7. 12 (2007), pp. 3813–3817.
[52] N. A. A. Zwaneveld et al. “Organized Formation of 2D Extended Covalent
Organic Frameworks at Surfaces”. In: Journal of the American Chemical Society
130. 21 (2008), pp. 6678–6679.
[53] S. Nowakowska et al. “Configuring Electronic States in an Atomically Precise
Array of Quantum Boxes”. In: Small 12. 28 (2016), pp. 3759–3763.
[54] I. Piquero-Zulaica et al. “Precise engineering of quantum dot array coupling
through their barrier widths”. In: Nature Communications 8.787 (2017).
[55] R. Balog et al. “Bandgap opening in graphene induced by patterned hydro-
gen adsorption”. In: Nature Materials 9. 4 (2010), pp. 315–319.
[56] M. Ezawa. “Peculiar width dependence of the electronic properties of carbon
nanoribbons”. In: Physical Review B: Condensed Matter and Materials Physics
73. 4 (2006).
[57] R. Lindner and A. Kühnle. “On-Surface Reactions”. In: ChemPhysChem 16. 8
(2015), pp. 1582–1592.
[58] A. Richter et al. “On-surface synthesis on a bulk insulator surface”. In: Journal
of Physics: Condensed Matter 30. 13 (2018), p. 133001.
[59] A. Gourdon. “On-Surface Covalent Coupling in Ultrahigh Vacuum”. In: Ange-
wandte Chemie International Edition 47. 37 (2008), pp. 6950–6953.
[60] D. F. Perepichka and F. Rosei. “Chemistry: Extending Polymer Conjugation
into the Second Dimension”. In: Science 323. 5911 (2009), pp. 216–217.
BIBLIOGRAPHY 173
[61] J. A. Lipton-Duffin et al. “Synthesis of Polyphenylene Molecular Wires by

Surface-Confined Polymerization”. In: Small 5. 5 (2009), pp. 592–597.
[62] J. Cai et al. “Atomically precise bottom-up fabrication of graphene nanorib-
bons”. In: Nature 466. 7305 (2010), pp. 470–473.
[63] M. Lackinger. “Surface-assisted Ullmann coupling”. In: Chemical Communica-
tions 53. 56 (2017), pp. 7872–7885.
[64] E. Carbonell-Sanromà et al. “Quantum Dots Embedded in Graphene Nanorib-
bons by Chemical Substitution”. In: Nano Letters 17. 1 (2017), pp. 50–56.
[65] S. Wang, W. Wang, and N. Lin. “Resolving Band-Structure Evolution and
Defect-Induced States of Single Conjugated Oligomers by Scanning Tunnel-
ing Microscopy and Tight-Binding Calculations”. In: Physical Review Letters
106. 20 (2011), p. 206803.
[66] J. Eichhorn et al. “On-surface Ullmann polymerization via intermediate organometal-
lic networks on Ag(111)”. In: Chemical Communications 50. 57 (2014), pp. 7680–
7682.
[67] S.-W. Hla et al. “Inducing All Steps of a Chemical Reaction with the Scan-
ning Tunneling Microscope Tip: Towards Single Molecule Engineering”. In:
Physical Review Letters 85. 13 (2000), pp. 2777–2780.
[68] F. Ullmann. “Ueber symmetrische Biphenylderivate”. In: Justus Liebig’s An-
nalen der Chemie 332. 1-2 (1904), pp. 38–81.
[69] L. Grill et al. “Nano-architectures by covalent assembly of molecular building
blocks”. In: Nature Nanotechnology 2. 11 (2007), pp. 687–691.
[70] R. Gutzler. “Band-structure engineering in conjugated 2D polymers”. In: Phys-
ical Chemistry Chemical Physics 18. 42 (2016), pp. 29092–29100.
[71] O. Ourdjini et al. “Substrate-mediated ordering and defect analysis of a sur-
face covalent organic framework”. In: Physical Review B: Condensed Matter and
Materials Physics 84. 12 (2011), p. 125421.
[72] Y.-Q. Zhang et al. “Homo-coupling of terminal alkynes on a noble metal sur-
face”. In: Nature Communications 3. 1286 (2012).
[73] S. Clair, M. Abel, and L. Porte. “Growth of boronic acid based two-dimensional
covalent networks on a metal surface under ultrahigh vacuum”. In: Chemical
Communications 50. 68 (2014), pp. 9627–9635.
[74] A. Rastgoo-Lahrood et al. “Reversible intercalation of iodine monolayers be-
tween on-surface synthesised covalent polyphenylene networks and Au(111)”.
In: Nanoscale 9. 15 (2017), pp. 4995–5001.
[75] S. Whitelam et al. “Common Physical Framework Explains Phase Behavior
and Dynamics of Atomic, Molecular, and Polymeric Network Formers”. In:
Physical Review X 4. 1 (2014), p. 011044.
174 BIBLIOGRAPHY
[76] S. Linden et al. “Electronic Structure of Spatially Aligned Graphene Nanorib-

bons on Au(788)”. In: Physical Review Letters 108. 21 (2012), p. 216801.
[77] G. Vasseur et al. “Quasi one-dimensional band dispersion and surface met-
allization in long-range ordered polymeric wires”. In: Nature Communications
7. 10235 (2016).
[78] N. Merino-Díez et al. “Switching from Reactant to Substrate Engineering in
the Selective Synthesis of Graphene Nanoribbons”. In: The Journal of Physical
Chemistry Letters 9. 10 (2018), pp. 2510–2517.
[79] B. V. Senkovskiy et al. “Finding the hidden valence band of N = 7 armchair
graphene nanoribbons with angle-resolved photoemission spectroscopy”. In:
2D Materials 5. 3 (2018), p. 035007.
[80] A. Mugarza and J. E. Ortega. “Electronic states at vicinal surfaces”. In: Journal
of Physics: Condensed Matter 15. 47 (2003), S3281.
[81] M. Corso et al. “Electronic states in faceted Au(111) studied with curved crys-
tal surfaces”. In: Journal of Physics: Condensed Matter 21. 35 (2009), p. 353001.
[82] J. E. Ortega et al. “Structure and electronic states of vicinal Ag(111) surfaces
with densely kinked steps”. In: New Journal of Physics 20. 7 (2018), p. 073010.
[83] G. Vasseur et al. “Π Band Dispersion along Conjugated Organic Nanowires
Synthesized on a Metal Oxide Semiconductor”. In: Journal of the American
Chemical Society 138. 17 (2016), pp. 5685–5692.
[84] K. Oura, ed. Surface science: an introduction. Advanced texts in physics. Berlin
; New York: Springer, 2003.
[85] F. Reinert and S. Hüfner. “Photoemission spectroscopy—from early days to
recent applications”. In: New Journal of Physics 7 (2005), p. 97.
[86] F. J. Himpsel. “Angle-resolved measurements of the photoemission of elec-
trons in the study of solids”. In: Advances in Physics 32. 1 (1983), pp. 1–51.
[87] K. E. Smith and S. D. Kevan. “The electronic structure of solids studied using
angle resolved photoemission spectroscopy”. In: Progress in Solid State Chem-
istry 21. 2 (1991), pp. 49–131.
[88] A. Damascelli, Z. Hussain, and Z.-X. Shen. “Angle-resolved photoemission
studies of the cuprate superconductors”. In: Reviews of Modern Physics 75. 2
(2003), pp. 473–541.
[89] A. Damascelli. “Probing the Electronic Structure of Complex Systems by ARPES”.
In: Physica Scripta T109 (2004), p. 61.
[90] H. Hertz. “Ueber einen Einfluss des ultravioletten Lichtes auf die electrische
Entladung”. In: Annalen der Physik und Chemie 267. 8 (1887), pp. 983–1000.
[91] A. Einstein. “Über einen die Erzeugung und Verwandlung des Lichtes betr-
effenden heuristischen Gesichtspunkt”. In: Annalen der Physik 322. 6 (1905),
pp. 132–148.
BIBLIOGRAPHY 175
[92] H. Ishii et al. “Energy Level Alignment and Interfacial Electronic Structures at
Organic/Metal and Organic/Organic Interfaces”. In: Advanced Materials 11. 8
(1999), pp. 605–625.
[93] Ph. Hofmann et al. “Unexpected surface sensitivity at high energies in angle-
resolved photoemission”. In: Physical Review B: Condensed Matter and Materi-
als Physics 66. 24 (2002), p. 245422.
[94] A. Tamai et al. “Spin-orbit splitting of the Shockley surface state on Cu(111)”.
In: Physical Review B: Condensed Matter and Materials Physics 87. 7 (2013), p. 075113.
[95] H. Oka et al. “Spin-polarized quantum confinement in nanostructures: Scan-
ning tunneling microscopy”. In: Reviews of Modern Physics 86. 4 (2014), pp. 1127–
1168.
[96] Y. Hasegawa and Ph. Avouris. “Direct observation of standing wave forma-
tion at surface steps using scanning tunneling spectroscopy”. In: Physical Re-
view Letters 71. 7 (1993), pp. 1071–1074.
[97] P. Avouris and I.-W. Lyo. “Observation of Quantum-Size Effects at Room
Temperature on Metal Surfaces With STM”. In: Science 264. 5161 (1994), pp. 942–
945.
[98] J. Li et al. “Electron Confinement to Nanoscale Ag Islands on Ag(111): A
Quantitative Study”. In: Physical Review Letters 80. 15 (1998), pp. 3332–3335.
[99] P. Han and P. S. Weiss. “Electronic substrate-mediated interactions”. In: Sur-
face Science Reports 67. 2 (2012), pp. 19–81.
[100] F. Reinert et al. “Direct measurements of the L -gap surface states on the (111)
face of noble metals by photoelectron spectroscopy”. In: Physical Review B:
Condensed Matter and Materials Physics 63. 11 (2001), p. 115415.
[101] S. LaShell, B. A. McDougall, and E. Jensen. “Spin Splitting of an Au(111) Sur-
face State Band Observed with Angle Resolved Photoelectron Spectroscopy”.
In: Physical Review Letters 77. 16 (1996), pp. 3419–3422.
[102] M. Ilyn et al. “Step-doubling at Vicinal Ni(111) Surfaces Investigated with a
Curved Crystal”. In: The Journal of Physical Chemistry C 121. 7 (2017), pp. 3880–
3886.
[103] A. Mugarza et al. “Modelling nanostructures with vicinal surfaces”. In: Jour-
nal of Physics: Condensed Matter 18. 13 (2006), S27.
[104] J. Zhang et al. “Probing the spatial and momentum distribution of confined
surface states in a metal coordination network”. In: Chemical Communications
50. 82 (2014), pp. 12289–12292.
[105] C. J. Chen. Introduction to Scanning Tunneling Microscopy. Oxford University
Press, 2007.
[106] M. K. Muntwiler. “Nanostructured magnetic interfaces: case studies and new
experiment control software”. PhD thesis. University of Zürich, 2004.
176 BIBLIOGRAPHY
[107] J.-Y. Park et al. “Modification of surface-state dispersion upon Xe adsorption:

A scanning tunneling microscope study”. In: Physical Review B: Condensed
Matter and Materials Physics 62. 24 (2000), R16341–R16344.
[108] M. F. Crommie, C. P. Lutz, and D. M. Eigler. “Imaging standing waves in a
two-dimensional electron gas”. In: Nature 363. 6429 (1993), pp. 524–527.
[109] R. Paniago et al. “Temperature dependence of Shockley-type surface energy
bands on Cu(111), Ag(111) and Au(111)”. In: Surface Science 336. 1-2 (1995),
pp. 113–122.
[110] F. J. Giessibl. “Atomic Resolution of the Silicon (111)-(7x7) Surface by Atomic
Force Microscopy”. In: Science 267. 5194 (1995), pp. 68–71.
[111] R. Pawlak et al. “Atomic-Scale Mechanical Properties of Orientated C 60 Molecules
Revealed by Noncontact Atomic Force Microscopy”. In: ACS Nano 5. 8 (2011),
pp. 6349–6354.
[112] P. Jelínek. “High resolution SPM imaging of organic molecules with function-
alized tips”. In: Journal of Physics: Condensed Matter 29. 34 (2017), p. 343002.
[113] S. Kawai. “Revealing mechanical and structural properties of molecules on
surface by high-resolution atomic force microscopy”. In: Polymer Journal 49
(2017), pp. 3–11.
[114] L. Gross et al. “The Chemical Structure of a Molecule Resolved by Atomic
Force Microscopy”. In: Science 325. 5944 (2009), pp. 1110–1114.
[115] A. Mugarza et al. “Electron Confinement in Surface States on a Stepped Gold
Surface Revealed by Angle-Resolved Photoemission”. In: Physical Review Let-
ters 87. 10 (2001), p. 107601.
[116] F. J. García de Abajo et al. “Lateral engineering of surface states – towards
surface-state nanoelectronics”. In: Nanoscale 2. 5 (2010), pp. 717–721.
[117] A. Mugarza et al. “Measurement of electron wave functions and confining
potentials via photoemission”. In: Physical Review B: Condensed Matter and
Materials Physics 67. 8 (2003), p. 081404.
[118] P. Hohenberg and W. Kohn. “Inhomogeneous Electron Gas”. In: Physical Re-
view 136. 3B (1964), B864–B871.
[119] W. Kohn and L. J. Sham. “Self-Consistent Equations Including Exchange and
Correlation Effects”. In: Physical Review 140. 4A (1965), A1133–A1138.
[120] J. C. Cuevas et al. “Theoretical description of the electrical conduction in
atomic and molecular junctions”. In: Nanotechnology 14. 8 (2003), R29–R38.
[121] B. Gunter Kretz. “Electronic and transport properties of 2D Dirac materials:
graphene and topological insulators”. PhD thesis. Universidad del País Vasco
UPV-EHU, 2018.
BIBLIOGRAPHY 177
[122] P. M. Echenique and J. B. Pendry. “The existence and detection of Rydberg

states at surfaces”. In: Journal of Physics C: Solid State Physics 11. 10 (1978),
pp. 2065–2075.
[123] Z. M. Abd El-Fattah et al. “Modifying the Cu(111) Shockley surface state by
Au alloying”. In: Physical Review B: Condensed Matter and Materials Physics
86. 24 (2012), p. 245418.
[124] Z. M. Abd El-Fattah. “Surface states manipulation via surface/interface de-
fects and adsorbates”. PhD thesis. Universidad del País Vasco (UPV-EHU),
2012.
[125] P. Harrison. Quantum Wells, Wires and Dots: Theoretical and Computational Physics
of Semiconductor Nanostructures. Chichester, UK: John Wiley & Sons, Ltd, 2005.
[126] F. Forster et al. “Systematic studies on surface modifications by ARUPS on
Shockley-type surface states”. In: Surface Science 600. 18 (2006), pp. 3870–3874.
[127] M. C. Cottin et al. “Anisotropic scattering of surface state electrons at a point
defect on Bi(111)”. In: Applied Physics Letters 98. 2 (2011), p. 022108.
[128] K. Müller et al. “Multimorphism in molecular monolayers: Pentacene on Cu(110)”.
In: Physical Review B: Condensed Matter and Materials Physics 79. 24 (2009),
p. 245421.
[129] N. Kepčija et al. “Quantum confinement in self-assembled two-dimensional
nanoporous honeycomb networks at close-packed metal surfaces”. In: The
Journal of Chemical Physics 142. 10 (2015), p. 101931.
[130] F. Baumberger et al. “Tailoring Confining Barriers for Surface States by Step
Decoration: CO / Vicinal Cu(111)”. In: Physical Review Letters 88. 23 (2002),
p. 237601.
[131] J. Lobo-Checa et al. “Tuning the Surface State Dimensionality of Cu Nanos-
tripes”. In: Physical Review Letters 93. 13 (2004), p. 137602.
[132] Y. Pennec et al. “Supramolecular gratings for tuneable confinement of elec-
trons on metal surfaces”. In: Nature Nanotechnology 2. 2 (2007), pp. 99–103.
[133] J. Li et al. “Tunnelling spectroscopy of surface state scattering and confine-
ment”. In: Surface Science 422. 1-3 (1999), pp. 95–106.
[134] G. Rodary et al. “Quantization of the electron wave vector in nanostruc-
tures: Counting k -states”. In: Physical Review B: Condensed Matter and Ma-
terials Physics 75. 23 (2007), p. 233412.
[135] E. J. Heller et al. “Scattering and absorption of surface electron waves in
quantum corrals”. In: Nature 369. 6480 (1994), pp. 464–466.
[136] G. A. Fiete and E. J. Heller. “Colloquium : Theory of quantum corrals and
quantum mirages”. In: Reviews of Modern Physics 75. 3 (2003), pp. 933–948.
178 BIBLIOGRAPHY
[137] H. K. Harbury and W. Porod. “Elastic scattering theory for electronic waves
in quantum corrals”. In: Physical Review B: Condensed Matter and Materials
Physics 53. 23 (1996), pp. 15455–15458.
[138] A. I. Rahachou and I. V. Zozoulenko. “Elastic scattering of surface electron
waves in quantum corrals: Importance of the shape of the adatom poten-
tial”. In: Physical Review B: Condensed Matter and Materials Physics 70. 23 (2004),
p. 233409.
[139] N. N. Negulyaev et al. “Direct Evidence for the Effect of Quantum Confine-
ment of Surface-State Electrons on Atomic Diffusion”. In: Physical Review Let-
ters 101. 22 (2008), p. 226601.
[140] J. Fernández et al. “Manipulation of the surface density of states of Ag(111)
by means of resonators: Experiment and theory”. In: Physical Review B: Con-
densed Matter and Materials Physics 94. 7 (2016), p. 075408.
[141] H. Jensen et al. “Electron dynamics in vacancy islands: Scanning tunneling
spectroscopy on Ag(111)”. In: Physical Review B: Condensed Matter and Materi-
als Physics 71. 15 (2005), p. 155417.
[142] J. E. Ortega et al. “Scattering of surface electrons by isolated steps versus
periodic step arrays”. In: Physical Review B: Condensed Matter and Materials
Physics 87. 11 (2013), p. 115425.
[143] L. Gross et al. “Scattering of Surface State Electrons at Large Organic Molecules”.
In: Physical Review Letters 93. 5 (2004), p. 056103.
[144] Ph. Avouris. “Real space imaging of electron scattering phenomena at metal
surfaces”. In: Journal of Vacuum Science & Technology B: Microelectronics and
Nanometer Structures 12. 3 (1994), p. 1447.
[145] K. Schouteden and C. Van Haesendonck. “Lateral Quantization of Two-Dimensional
Electron States by Embedded Ag Nanocrystals”. In: Physical Review Letters
108. 7 (2012), p. 076806.
[146] L. Niebergall et al. “Electron confinement in hexagonal vacancy islands: The-
ory and experiment”. In: Physical Review B: Condensed Matter and Materials
Physics 74. 19 (2006), p. 195436.
[147] J. E. Ortega et al. “One-dimensional versus two-dimensional surface states
on stepped Au ( 111 )”. In: Physical Review B: Condensed Matter and Materials
Physics 65. 16 (2002), p. 165413.
[148] L. Bürgi et al. “Confinement of Surface State Electrons in Fabry-Pérot Res-
onators”. In: Physical Review Letters 81. 24 (1998), pp. 5370–5373.
[149] J. E. Ortega et al. “Interplay between structure and electronic states in step
arrays explored with curved surfaces”. In: Physical Review B: Condensed Matter
and Materials Physics 83. 8 (2011), p. 085411.
BIBLIOGRAPHY 179
[150] J.-M. Lehn. Supramolecular Chemistry: Concepts and Perspectives. Weinheim, FRG:
Wiley-VCH Verlag GmbH & Co. KGaA, 1995.
[151] S. Stepanow et al. “Surface-Assisted Assembly of 2D Metal–Organic Net-
works That Exhibit Unusual Threefold Coordination Symmetry”. In: Ange-
[152] K. Müller, M. Enache, and M. Stöhr. “Confinement properties of 2D porous
molecular networks on metal surfaces”. In: Journal of Physics: Condensed Mat-
ter 28. 15 (2016), p. 153003.
[153] J. Wyrick et al. “Do Two-Dimensional Noble Gas Atoms Produce Molecular
Honeycombs at a Metal Surface?” In: Nano Letters 11. 7 (2011), pp. 2944–2948.
[154] S. Wang et al. “Tuning two-dimensional band structure of Cu(111) surface-
state electrons that interplay with artificial supramolecular architectures”.
p. 245430.
[155] R. Hoffmann. “A chemical and theoretical way to look at bonding on sur-
faces”. In: Reviews of Modern Physics 60. 3 (1988), pp. 601–628.
[156] P. Liljeroth et al. “Variable Orbital Coupling in a Two-Dimensional Quantum-
Dot Solid Probed on a Local Scale”. In: Physical Review Letters 97. 9 (2006),
p. 096803.
[157] D. Malterre et al. “ARPES and STS investigation of Shockley states in thin
metallic films and periodic nanostructures”. In: New Journal of Physics 9. 10
(2007), pp. 391–391.
[158] I. Piquero-Zulaica et al. “Temperature dependence of the partially localized
state in a 2D molecular nanoporous network”. In: Applied Surface Science 391
(2017), pp. 39–43.
[159] A. Shchyrba et al. “Covalent assembly of a two-dimensional molecular “sponge”
on a Cu(111) surface: confined electronic surface states in open and closed
pores”. In: Chemical Communications 50. 57 (2014), pp. 7628–7631.
[160] K. Seufert et al. “Controlled Interaction of Surface Quantum-Well Electronic
States”. In: Nano Letters 13. 12 (2013), pp. 6130–6135.
[161] M. Stöhr et al. “Lateral Manipulation for the Positioning of Molecular Guests
within the Confinements of a Highly Stable Self-Assembled Organic Surface
Network”. In: Small 3. 8 (2007), pp. 1336–1340.
[162] D. Kuhne et al. “Rotational and constitutional dynamics of caged supramolecules”.
In: Proceedings of the National Academy of Sciences 107. 50 (2010), pp. 21332–
21336.
[163] Z. Cheng et al. “Adsorbates in a Box: Titration of Substrate Electronic States”.
In: Physical Review Letters 105. 6 (2010), p. 066104.
180 BIBLIOGRAPHY
[164] S. Nowakowska et al. “Interplay of weak interactions in the atom-by-atom

condensation of xenon within quantum boxes”. In: Nature Communications
6. 6071 (2015).
[165] R. Zhang et al. “Two-Dimensional Superlattices of Bi Nanoclusters Formed
on a Au(111) Surface Using Porous Supramolecular Templates”. In: ACS Nano
9. 8 (2015), pp. 8547–8553.
[166] M. Pivetta et al. “Formation of Fe Cluster Superlattice in a Metal-Organic
Quantum-Box Network”. In: Physical Review Letters 110. 8 (2013), p. 086102.
[167] M. Matena et al. “On-surface synthesis of a two-dimensional porous coor-
dination network: Unraveling adsorbate interactions”. In: Physical Review B:
Condensed Matter and Materials Physics 90. 12 (2014), p. 125408.
[168] M. Stöhr et al. “Controlling Molecular Assembly in Two Dimensions: The
Concentration Dependence of Thermally Induced 2D Aggregation of Molecules
on a Metal Surface”. In: Angewandte Chemie International Edition 44. 45 (2005),
pp. 7394–7398.
[169] A. Shchyrba et al. “Controlling the Dimensionality of On-Surface Coordina-
tion Polymers via Endo- or Exoligation”. In: Journal of the American Chemical
Society 136. 26 (2014), pp. 9355–9363.
[170] R. Paniago et al. “High-resolution photoemission study of the surface states
near Γ on Cu(111) and Ag(111)”. In: Surface Science 331-333 (1995), pp. 1233–
1237.
[171] J. A. Knapp et al. “Temperature dependence of bulk and surface energy bands
in copper using angle-resolved photoemission”. In: Physical Review B: Con-
densed Matter and Materials Physics 19. 6 (1979), pp. 2844–2849.
[172] E. Knoesel, A. Hotzel, and M. Wolf. “Temperature dependence of surface
state lifetimes, dephasing rates and binding energies on Cu(111) studied with
time-resolved photoemission”. In: Journal of Electron Spectroscopy and Related
Phenomena 88-91 (1998), pp. 577–584.
[173] H. Yamane and N. Kosugi. “Site-Specific Organic/Metal Interaction Revealed
from Shockley-Type Interface State”. In: The Journal of Physical Chemistry C
120. 42 (2016), pp. 24307–24313.
[174] I. Horcas et al. “WSXM: A software for scanning probe microscopy and a tool
for nanotechnology”. In: Review of Scientific Instruments 78. 1 (2007), p. 013705.
[175] W. Wang et al. “Single-Molecule Resolution of an Organometallic Interme-
diate in a Surface-Supported Ullmann Coupling Reaction”. In: Journal of the
American Chemical Society 133. 34 (2011), pp. 13264–13267.
[176] S. Kawai et al. “Extended Halogen Bonding between Fully Fluorinated Aro-
matic Molecules”. In: ACS Nano 9. 3 (2015), pp. 2574–2583.
BIBLIOGRAPHY 181
[177] T. Okamoto et al. “V-Shaped Organic Semiconductors With Solution Process-

ability, High Mobility, and High Thermal Durability”. In: Advanced Materials
25. 44 (2013), pp. 6392–6397.
[178] T. Clark et al. “Halogen bonding: the σ-hole: Proceedings of “Modeling inter-
actions in biomolecules II”, Prague, September 5th–9th, 2005”. In: Journal of
Molecular Modeling 13. 2 (2007), pp. 291–296.
[179] G. R. Desiraju et al. “Definition of the halogen bond (IUPAC Recommenda-
tions 2013)”. In: Pure and Applied Chemistry 85. 8 (2013), pp. 1711–1713.
[180] S. Kawai et al. “Organometallic Bonding in an Ullmann-Type On-Surface
Chemical Reaction Studied by High-Resolution Atomic Force Microscopy”.
In: Small 12. 38 (2016), pp. 5303–5311.
[181] J. Shang et al. “Assembling molecular Sierpiński triangle fractals”. In: Nature
Chemistry 7. 5 (2015), pp. 389–393.
[182] W. Wang et al. “Cooperative Modulation of Electronic Structures of Aromatic
Molecules Coupled to Multiple Metal Contacts”. In: Physical Review Letters
110. 4 (2013), p. 046802.
[183] A. A. Kordyuk. “ARPES experiment in fermiology of quasi-2D metals (Re-
view Article)”. In: Low Temperature Physics 40. 4 (2014), pp. 286–296.
[184] D. Lu et al. “Angle-Resolved Photoemission Studies of Quantum Materials”.
In: Annual Review of Condensed Matter Physics 3 (2012), pp. 129–167.
[185] H. Cercellier et al. “Interplay between structural, chemical, and spectroscopic
properties of Ag / Au ( 111 ) epitaxial ultrathin films: A way to tune the
Rashba coupling”. In: Physical Review B: Condensed Matter and Materials Physics
73. 19 (2006), p. 195413.
[186] F. J. Giessibl. “High-speed force sensor for force microscopy and profilom-
etry utilizing a quartz tuning fork”. In: Applied Physics Letters 73. 26 (1998),
pp. 3956–3958.
[187] S. Kawai et al. “Atomically controlled substitutional boron-doping of graphene
nanoribbons”. In: Nature Communications 6. 8098 (2015).
[188] G. D. Nguyen et al. “Atomically precise graphene nanoribbon heterojunc-
tions from a single molecular precursor”. In: Nature Nanotechnology 12. 11
(2017), pp. 1077–1082.
[189] R. A. Durr et al. “Orbitally Matched Edge-Doping in Graphene Nanorib-
bons”. In: Journal of the American Chemical Society 140. 2 (2018), pp. 807–813.
[190] G. Reecht et al. “Oligothiophene Nanorings as Electron Resonators for Whis-
pering Gallery Modes”. In: Physical Review Letters 110. 5 (2013), p. 056802.
[191] L. Genovese et al. “Daubechies wavelets as a basis set for density functional
pseudopotential calculations”. In: The Journal of Chemical Physics 129. 1 (2008),
p. 014109.
182 BIBLIOGRAPHY
[192] C. Hartwigsen, S. Goedecker, and J. Hutter. “Relativistic separable dual-space

Gaussian pseudopotentials from H to Rn”. In: Physical Review B: Condensed
Matter and Materials Physics 58. 7 (1998), pp. 3641–3662.
[193] L. Genovese et al. “Efficient solution of Poisson’s equation with free bound-
ary conditions”. In: The Journal of Chemical Physics 125. 7 (2006), p. 074105.
[194] Thai Thanh Thu Bui et al. “The Nature of Halogen · · · Halogen Interactions:
A Model Derived from Experimental Charge-Density Analysis”. In: Ange-
[195] M. Ruiz-Osés et al. “Non-Covalent Interactions in Supramolecular Assem-
blies Investigated with Electron Spectroscopies”. In: ChemPhysChem 10. 6 (2009),
pp. 896–900.
[196] F. Forster, S. Hüfner, and F. Reinert. “Rare Gases on Noble-Metal Surfaces:
An Angle-Resolved Photoemission Study with High Energy Resolution † ”.
In: The Journal of Physical Chemistry B 108. 38 (2004), pp. 14692–14698.
[197] E. Bertel and N. Memmel. “Promotors, poisons and surfactants: Electronic
effects of surface doping on metals”. In: Applied Physics A Materials Science
and Processing 63. 6 (1996), pp. 523–531.
[198] C. Liu et al. “Interaction between Adatom-Induced Localized States and a
Quasi-Two-Dimensional Electron Gas”. In: Physical Review Letters 96. 3 (2006),
p. 036803.
[199] C. Didiot et al. “Interacting quantum box superlattice by self-organized Co
nanodots on Au(788)”. In: Physical Review B: Condensed Matter and Materials
Physics 76. 8 (2007), p. 081404.
[200] P. Mishra et al. “Spatially Resolved Magnetic Anisotropy of Cobalt Nanos-
tructures on the Au(111) Surface”. In: Nano Letters 17. 9 (2017), pp. 5843–5847.
[201] N. Armbrust et al. “Model potential for the description of metal/organic in-
terface states”. In: Scientific Reports 7. 46561 (2017).
[202] Q. Sun et al. “Dehalogenative Homocoupling of Terminal Alkynyl Bromides
on Au(111): Incorporation of Acetylenic Scaffolding into Surface Nanostruc-
tures”. In: ACS Nano 10. 7 (2016), pp. 7023–7030.
[203] Y.-L. Zhao et al. “Donor/Acceptor Properties of Aromatic Molecules in Com-
plex Metal–Molecule Interfaces”. In: Langmuir 33. 2 (2017), pp. 451–458.
[204] K. K. Gomes et al. “Designer Dirac fermions and topological phases in molec-
ular graphene”. In: Nature 483. 7389 (2012), pp. 306–310.
[205] N. V. Smith. “Phase analysis of image states and surface states associated
with nearly-free-electron band gaps”. In: Physical Review B: Condensed Matter
and Materials Physics 32. 6 (1985), pp. 3549–3555.
BIBLIOGRAPHY 183
[206] F. Schiller et al. “Electronic structure of Mg : From monolayers to bulk”.

p. 125106.
[207] J. E. Ortega and F. J. Himpsel. “Quantum well states as mediators of magnetic
coupling in superlattices”. In: Physical Review Letters 69. 5 (1992), pp. 844–847.
[208] Y. Hasegawa, T. Suzuki, and T. Sakurai. “Modification of electron density
in surface states: standing wave observation on Pd overlayers by STM”. In:
Surface Science 514. 1-3 (2002), pp. 84–88.
[209] M. A. Mueller et al. “Probing interfacial properties with Bloch electrons: Ag
on Cu(111)”. In: Physical Review B: Condensed Matter and Materials Physics 40. 8
(1989), pp. 5845–5848.
[210] G. Kresse and J. Hafner. “Ab initio molecular dynamics for liquid metals”. In:
Physical Review B: Condensed Matter and Materials Physics 47. 1 (1993), pp. 558–
561.
[211] G. Kresse and J. Furthmüller. “Efficient iterative schemes for ab initio total-
energy calculations using a plane-wave basis set”. In: Physical Review B: Con-
densed Matter and Materials Physics 54. 16 (1996), pp. 11169–11186.
[212] G. Kresse and J. Furthmüller. “Efficiency of ab-initio total energy calculations
for metals and semiconductors using a plane-wave basis set”. In: Computa-
tional Materials Science 6. 1 (1996), pp. 15–50.
[213] J. P. Perdew, K. Burke, and M. Ernzerhof. “Generalized Gradient Approxi-
mation Made Simple [Phys. Rev. Lett. 77, 3865 (1996)]”. In: Physical Review
Letters 78. 7 (1997), pp. 1396–1396.
[214] N. Gonzalez-Lakunza et al. “Formation of Dispersive Hybrid Bands at an
Organic-Metal Interface”. In: Physical Review Letters 100. 15 (2008), p. 156805.
[215] Z. M. Abd El-Fattah et al. “Lifshitz Transition across the Ag/Cu(111) Super-
lattice Band Gap Tuned by Interface Doping”. In: Physical Review Letters 107. 6
(2011), p. 066803.
[216] S. Kawai et al. “Van der Waals interactions and the limits of isolated atom
models at interfaces”. In: Nature Communications 7. 11559 (2016).
[217] C. Hückstädt et al. “Work function studies of rare-gas/noble metal adsorp-
tion systems using a Kelvin probe”. In: Physical Review B: Condensed Matter
[218] A. Sadeghi et al. “Multiscale approach for simulations of Kelvin probe force
microscopy with atomic resolution”. In: Physical Review B: Condensed Matter
[219] L. Bürgi, H. Brune, and K. Kern. “Imaging of Electron Potential Landscapes
on Au(111)”. In: Physical Review Letters 89. 17 (2002), p. 176801.
184 BIBLIOGRAPHY
[220] J. Ziroff et al. “Adsorption energy and geometry of physisorbed organic molecules
on Au(111) probed by surface-state photoemission”. In: Surface Science 603. 2
(2009), pp. 354–358.
[221] F. Forster et al. “Surface and interface states on adsorbate covered noble metal
surfaces”. In: Surface Science 532-535 (2003), pp. 160–165.
[222] H. González-Herrero et al. “Graphene Tunable Transparency to Tunneling
Electrons: A Direct Tool To Measure the Local Coupling”. In: ACS Nano 10. 5
(2016), pp. 5131–5144.
[223] J. Tesch et al. “Structural and electronic properties of graphene nanoflakes on
Au(111) and Ag(111)”. In: Scientific Reports 6. 23439 (2016).
[224] S. Joshi et al. “Boron Nitride on Cu(111): An Electronically Corrugated Mono-
layer”. In: Nano Letters 12. 11 (2012), pp. 5821–5828.
[225] D. Malterre et al. “Symmetry breaking and gap opening in two-dimensional
hexagonal lattices”. In: New Journal of Physics 13. 013026 (2011).
[226] H. Dil et al. “Surface Trapping of Atoms and Molecules with Dipole Rings”.
In: Science 319. 5871 (2008), pp. 1824–1826.
[227] W. Krenner et al. “Assessment of Scanning Tunneling Spectroscopy Modes
Inspecting Electron Confinement in Surface-Confined Supramolecular Net-
works”. In: Scientific Reports 3.1454 (2013).
[228] J. Teyssandier, S. De Feyter, and K. S. Mali. “Host–guest chemistry in two-
dimensional supramolecular networks”. In: Chemical Communications 52. 77
(2016), pp. 11465–11487.
[229] M. A. Chesters, M. Hussain, and J. Pritchard. “Xenon monolayer structures
on copper and silver”. In: Surface Science 35 (1973), pp. 161–171.
[230] Th. Seyller et al. “Observation of top-site adsorption for Xe on Cu(111)”. In:
Chemical Physics Letters 291. 5-6 (1998), pp. 567–572.
[231] D. M. Eigler, C. P. Lutz, and W. E. Rudge. “An atomic switch realized with
the scanning tunnelling microscope”. In: Nature 352. 6336 (1991), pp. 600–603.
[232] A. Yazdani, D. M. Eigler, and N. D. Lang. “Off-Resonance Conduction Through
Atomic Wires”. In: Science 272. 5270 (1996), pp. 1921–1924.
[233] G. Kichin et al. “Single Molecule and Single Atom Sensors for Atomic Reso-
lution Imaging of Chemically Complex Surfaces”. In: Journal of the American
Chemical Society 133. 42 (2011), pp. 16847–16851.
[234] L. Limot et al. “Surface-State Localization at Adatoms”. In: Physical Review
Letters 94. 3 (2005), p. 036805.
[235] F. E. Olsson et al. “Localization of the Cu(111) Surface State by Single Cu
Adatoms”. In: Physical Review Letters 93. 20 (2004), p. 206803.
BIBLIOGRAPHY 185
[236] J.-Y. Park et al. “Adsorption and growth of Xe adlayers on the Cu(111) sur-
face”. In: Physical Review B: Condensed Matter and Materials Physics 60. 24 (1999),
pp. 16934–16940.
[237] H. C. Manoharan, C. P. Lutz, and D. M. Eigler. “Quantum mirages formed
by coherent projection of electronic structure”. In: Nature 403. 6769 (2000),
pp. 512–515.
[238] K. S. Novoselov et al. “Electric Field Effect in Atomically Thin Carbon Films”.
In: Science 306. 5696 (2004), pp. 666–669.
[239] A. H. Castro Neto et al. “The electronic properties of graphene”. In: Reviews
of Modern Physics 81. 1 (2009), pp. 109–162.
[240] Y.-W. Son, M. L. Cohen, and S. G. Louie. “Energy Gaps in Graphene Nanorib-
bons”. In: Physical Review Letters 97. 21 (2006), p. 216803.
[241] A. Narita et al. “New advances in nanographene chemistry”. In: Chemical
Society Reviews 44. 18 (2015), pp. 6616–6643.
[242] Q. Shen, H.-Y. Gao, and H. Fuchs. “Frontiers of on-surface synthesis: From
principles to applications”. In: Nano Today 13 (2017), pp. 77–96.
[243] D. G. de Oteyza et al. “Substrate-Independent Growth of Atomically Precise
Chiral Graphene Nanoribbons”. In: ACS Nano 10. 9 (2016), pp. 9000–9008.
[244] P. H. Jacobse et al. “Electronic components embedded in a single graphene
nanoribbon”. In: Nature Communications 8. 119 (2017).
[245] P. B. Bennett et al. “Bottom-up graphene nanoribbon field-effect transistors”.
In: Applied Physics Letters 103. 25 (2013), p. 253114.
[246] V. Passi et al. “Field-Effect Transistors Based on Networks of Highly Aligned,
Chemically Synthesized N = 7 Armchair Graphene Nanoribbons”. In: ACS
Applied Materials & Interfaces 10. 12 (2018), pp. 9900–9903.
[247] S. Günes, H. Neugebauer, and N. S. Sariciftci. “Conjugated Polymer-Based
Organic Solar Cells”. In: Chemical Reviews 107. 4 (2007), pp. 1324–1338.
[248] G. Li, W.-H. Chang, and Y. Yang. “Low-bandgap conjugated polymers en-
abling solution-processable tandem solar cells”. In: Nature Reviews Materials
2. 8 (2017), p. 17043.
[249] H. Masai and J. Terao. “Stimuli-responsive functionalized insulated conju-
gated polymers”. In: Polymer Journal 49 (2017), pp. 805–814.
[250] P. Guiglion and M. A. Zwijnenburg. “Contrasting the optical properties of the
different isomers of oligophenylene”. In: Physical Chemistry Chemical Physics
17. 27 (2015), pp. 17854–17863.
[251] S. Y. Hong et al. “Origin of the Broken Conjugation in m-Phenylene Linked
Conjugated Polymers”. In: Macromolecules 34. 18 (2001), pp. 6474–6481.
186 BIBLIOGRAPHY
[252] M. H. van der Veen et al. “Molecules with Linear pi-Conjugated Pathways
between All Substituents Omniconjugation”. In: Advanced Functional Materi-
als 14. 3 (2004), pp. 215–223.
[253] C. García-Fernández et al. “Exploring the Relation Between Intramolecular
Conjugation and Band Dispersion in One-Dimensional Polymers”. In: The
Journal of Physical Chemistry C 121. 48 (2017), pp. 27118–27125.
[254] A. A. Kocherzhenko, F. C. Grozema, and L. D. A. Siebbeles. “Single molecule
charge transport: from a quantum mechanical to a classical description”. In:
Physical Chemistry Chemical Physics 13. 6 (2011), pp. 2096–2110.
[255] D. Z. Manrique et al. “A quantum circuit rule for interference effects in single-
molecule electrical junctions”. In: Nature Communications 6. 6389 (2015).
[256] C. M. Guédon et al. “Observation of quantum interference in molecular charge
transport”. In: Nature Nanotechnology 7. 5 (2012), pp. 305–309.
[257] T. Markussen, R. Stadler, and K. S. Thygesen. “The Relation between Struc-
ture and Quantum Interference in Single Molecule Junctions”. In: Nano Letters
10. 10 (2010), pp. 4260–4265.
[258] P. A. Limacher and H. P. Lüthi. “Cross-conjugation”. In: Wiley Interdisciplinary
Reviews: Computational Molecular Science 1. 4 (2011), pp. 477–486.
[259] A. L. Thompson et al. “Using Meta Conjugation To Enhance Charge Separa-
tion versus Charge Recombination in Phenylacetylene Donor-Bridge-Acceptor
Complexes”. In: Journal of the American Chemistry Society 127. 47 (2005), pp. 16348–
16349.
[260] T. Tada and K. Yoshizawa. “Molecular design of electron transport with or-
bital rule: toward conductance-decay free molecular junctions”. In: Physical
Chemistry Chemical Physics 17. 48 (2015), pp. 32099–32110.
[261] J. Terao et al. “Design principle for increasing charge mobility of π-conjugated
polymers using regularly localized molecular orbitals”. In: Nature Communi-
cations 4.1691 (2013).
[262] L. Talirz, P. Ruffieux, and R. Fasel. “On-Surface Synthesis of Atomically Pre-
cise Graphene Nanoribbons”. In: Advanced Materials 28. 29 (2016), pp. 6222–
6231.
[263] Q. Fan et al. “Surface-Assisted Organic Synthesis of Hyperbenzene Nanotroughs”.
In: Angewandte Chemie International Edition 52. 17 (2013), pp. 4668–4672.
[264] M. Abadía et al. “Polymerization of Well-Aligned Organic Nanowires on a
Ferromagnetic Rare-Earth Surface Alloy”. In: ACS Nano 11. 12 (2017), pp. 12392–
12401.
[265] L. Cai et al. “Direct Formation of C–C Double-Bonded Structural Motifs by
On-Surface Dehalogenative Homocoupling of gem -Dibromomethyl Molecules”.
In: ACS Nano 12. 8 (2018), pp. 7959–7966.
BIBLIOGRAPHY 187
[266] I. Piquero-Zulaica et al. “Electronic Structure Tunability by Periodic meta-

Ligand Spacing in One-Dimensional Organic Semiconductors”. In: ACS Nano
(2018).
[267] C. J. Judd et al. “Ullmann Coupling Reactions on Ag(111) and Ag(110); Sub-
strate Influence on the Formation of Covalently Coupled Products and Inter-
mediate Metal-Organic Structures”. In: Scientific Reports 7.14541 (2017).
[268] M. Di Giovannantonio et al. “Insight into Organometallic Intermediate and
Its Evolution to Covalent Bonding in Surface-Confined Ullmann Polymeriza-
tion”. In: ACS Nano 7. 9 (2013), pp. 8190–8198.
[269] Q. Fan et al. “Surface-Assisted Formation, Assembly, and Dynamics of Pla-
nar Organometallic Macrocycles and Zigzag Shaped Polymer Chains with
C–Cu–C Bonds”. In: ACS Nano 8. 1 (2014), pp. 709–718.
[270] M. Koch et al. “Substrate-controlled linking of molecular building blocks:
Au(111) vs. Cu(111)”. In: Surface Science 627 (2014), pp. 70–74.
[271] M. Chen et al. “On-Surface Synthesis and Characterization of Honeycombene
Oligophenylene Macrocycles”. In: ACS Nano 11. 1 (2017), pp. 134–143.
[272] Q. Fan et al. “On-Surface Pseudo-High-Dilution Synthesis of Macrocycles:
Principle and Mechanism”. In: ACS Nano 11. 5 (2017), pp. 5070–5079.
[273] M. Di Giovannantonio et al. “On-Surface Growth Dynamics of Graphene
Nanoribbons: The Role of Halogen Functionalization”. In: ACS Nano 12. 1
(2018), pp. 74–81.
[274] H. Offenbacher et al. “Orbital tomography: Molecular band maps, momen-
tum maps and the imaging of real space orbitals of adsorbed molecules”. In:
Journal of Electron Spectroscopy and Related Phenomena 204 (2015), pp. 92–101.
[275] G. Koller et al. “Intra- and Intermolecular Band Dispersion in an Organic
Crystal”. In: Science 317. 5836 (2007), pp. 351–355.
[276] L. Gross et al. “High-Resolution Molecular Orbital Imaging Using a p -Wave
STM Tip”. In: Physical Review Letters 107. 8 (2011), p. 086101.
[277] J. Hieulle et al. “On-Surface Route for Producing Planar Nanographenes with
Azulene Moieties”. In: Nano Letters 18. 1 (2018), pp. 418–423.
[278] J. M. Soler et al. “The SIESTA method for ab initio order- N materials simula-
tion”. In: Journal of Physics: Condensed Matter 14. 11 (2002), pp. 2745–2779.
[279] J. Klimeš, David R Bowler, and Angelos Michaelides. “Chemical accuracy for
the van der Waals density functional”. In: Journal of Physics: Condensed Matter
22. 2 (2010), p. 022201.
[280] Q. Fan, J. M. Gottfried, and J. Zhu. “Surface-Catalyzed C-C Covalent Cou-
pling Strategies toward the Synthesis of Low-Dimensional Carbon-Based Nanos-
tructures”. In: Accounts of Chemical Research 48. 8 (2015), pp. 2484–2494.
188 BIBLIOGRAPHY
[281] Q. Fan et al. “Confined Synthesis of Organometallic Chains and Macrocycles

by Cu-O Surface Templating”. In: ACS Nano 10. 3 (2016), pp. 3747–3754.
[282] P. J. Goddard, K. Schwaha, and R. M. Lambert. “Adsorption-desorption prop-
erties and surface structural chemistry of bromine on clean and sodium-
dosed Ag(111)”. In: Surface Science 71. 2 (1978), pp. 351–363.
[283] C. H. Hendon et al. “Grand Challenges and Future Opportunities for Metal–Organic
Frameworks”. In: ACS Central Science 3. 6 (2017), pp. 554–563.
[284] L. E. Kreno et al. “Metal–Organic Framework Materials as Chemical Sen-
sors”. In: Chemical Reviews 112. 2 (2012), pp. 1105–1125.
[285] L. Zhu et al. “Metal–Organic Frameworks for Heterogeneous Basic Cataly-
sis”. In: Chemical Reviews 117. 12 (2017), pp. 8129–8176.
[286] V. Stavila, A. A. Talin, and M. D. Allendorf. “MOF-based electronic and opto-
electronic devices”. In: Chemical Society Reviews 43. 16 (2014), pp. 5994–6010.
[287] R. Baltic et al. “Magnetic properties of single rare-earth atoms on graphene/Ir(111)”.
In: Physical Review B: Condensed Matter and Materials Physics 98. 2 (2018), p. 024412.
[288] Z. F. Wang, N. Su, and F. Liu. “Prediction of a Two-Dimensional Organic
Topological Insulator”. In: Nano Letters 13. 6 (2013), pp. 2842–2845.
[289] Z. F. Wang, Z. Liu, and F. Liu. “Organic topological insulators in organometal-
lic lattices”. In: Nature Communications 4. 1471 (2013).
[290] L. Dong et al. “Two-Dimensional π-Conjugated Covalent-Organic Frameworks
as Quantum Anomalous Hall Topological Insulators”. In: Physical Review Let-
ters 116. 9 (2016), p. 096601.
[291] Ya-ping Wang et al. “Discovery of intrinsic quantum anomalous Hall effect in
organic Mn-DCA lattice”. In: Applied Physics Letters 110. 23 (2017), p. 233107.
[292] X. Zhang et al. “Theoretical Discovery of a Superconducting Two-Dimensional
Metal–Organic Framework”. In: Nano Letters 17. 10 (2017), pp. 6166–6170.
[293] Z. Yang et al. “Two-dimensional delocalized states in organometallic bis-
acetylide networks on Ag(111)”. In: Nanoscale 10. 8 (2018), pp. 3769–3776.
[294] H. Sun et al. “Deconstruction of the Electronic Properties of a Topological In-
sulator with a Two-Dimensional Noble Metal–Organic Honeycomb–Kagome
Band Structure”. In: The Journal of Physical Chemistry C 122. 32 (2018), pp. 18659–
18668.
[295] Y.-Q. Zhang et al. “Intermolecular Hybridization Creating Nanopore Orbital
in a Supramolecular Hydrocarbon Sheet”. In: Nano Letters 16. 7 (2016), pp. 4274–
4281.
[296] M. Feng, J. Zhao, and H. Petek. “Atomlike, Hollow-Core-Bound Molecular
Orbitals of C60”. In: Science 320. 5874 (2008), pp. 359–362.
BIBLIOGRAPHY 189
[297] J. Zhao et al. “The Superatom States of Fullerenes and Their Hybridization
into the Nearly Free Electron Bands of Fullerites”. In: ACS Nano 3. 4 (2009),
pp. 853–864.
[298] M. Fritton et al. “The influence of ortho -methyl substitution in organometallic
self-assembly – a comparative study on Cu(111) vs. Ag(111)”. In: Chemical
Communications 54. 70 (2018), pp. 9745–9748.
[299] T. Lin et al. “Steering On-Surface Polymerization with Metal-Directed Tem-
plate”. In: Journal of the American Chemical Society 135. 9 (2013), pp. 3576–3582.

Thesis IPZ Final

Uploaded by

Copyright:

Available Formats

Thesis IPZ Final

Uploaded by

Document Information

Original Description:

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

Thesis IPZ Final

Uploaded by

Copyright:

Available Formats

Electronic Bands of Nanoporous Networks

and One-Dimensional Covalent Polymers

Electronic Bands of Nanoporous Networks

A thesis submitted in fulfillment of the requirements for the degree of

October 19, 2018

The studied electronic structure from nanoporous networks correspond to the

List of Abbreviations xvii

1 Part I: Introduction and Motivation 1

2 Experimental Techniques and Theoretical Methods 11

The Many-Body Problem . . . . . . . . . . . . . . . . . . . . . . 35

3 Part II: An Introduction to Assembly and Electronic Bands of Nanoporous

4 Temperature Dependence of the Partially Localized State in a 2D Molecular

5 Precise Engineering of Quantum Dot Array Coupling Through Their Bar-

6 Tunable Energy and Mass Renormalization from Homothetic Quantum Dot

6.5.4 EBEM/EPWE Simulations . . . . . . . . . . . . . . . . . . . . . . 91

7 Effective Determination of Surface Potential Landscapes from Metal-Organic

8 Configuring Electronic States in an Atomically Precise Array of Quantum

9 Part III: Introduction to On-Surface Synthesized One-Dimensional Zigzag

10 Electronic Structure Tunability by Periodic meta-Ligand Spacing in One-

10.4.5 ARPES Band Structure and Photoemission Intensity Simula-

11 Conclusions and Outlook 151

A Resumen de las Conclusiones Principales 161

B List of Publications 167

1.1 Top-Down and Bottom-Up Fabrication Methods . . . . . . . . . . . . . 2

2.1 ARPES Gradient Evaporation Method . . . . . . . . . . . . . . . . . . . 14

3.1 A Journey Into Confining Nanostructures . . . . . . . . . . . . . . . . . 41

4.1 Coverage Dependent Formation of 3deh-DPDI Molecular Network . . 49

5.1 QD Arrays Generated by Single-Wall (SW) and Double-Wall (DW)

5.2 Bonding Motifs That Stabilize the SW and DW Formations . . . . . . . 59

6.1 STM Characterization and EBEM/EPWE Scattering Regions of Ph6Co

7.1 3deh-DPDI Metal-Organic Network on Cu(111) . . . . . . . . . . . . . 96

7.9 Characterization of the 3deh-DPDI Band Structure . . . . . . . . . . . . 108

8.1 QD Electronic Structure Alteration Through Xe Occupation . . . . . . . 113

9.1 On-Surface Synthesis of GNRs and Nanoporous Graphene . . . . . . . 126

10.1 Structural Arrangement of the Zigzag Chains . . . . . . . . . . . . . . . 131

11.1 ARPES Electronic Band Structure of the Studied Nanoporous Networks152

11.2 Summary of the Zigzag Chain’s Electronic Structure Probed by ARPES

A.1 Resumen de la Estructura de Bandas de las Diferentes Redes Nanoporosas

2.1 Noble Metal (111) Shockley State Dispersion Fit Results . . . . . . . . . 26

5.1 Extracted ARPES Experimental Parameters from SW and DW Networks 62

6.1 ARPES Experimental BEs at Γ and Effective Masses of Ph3Co and

7.1 EBEM/EPWE Potential Parameters Used in the Models Discussed Through-

8.1 Band Sructure Parameters Extracted From ARPES Measurements on

11.1 Summary of ARPES and EBEM/EPWE Values Obtained for Nanoporous

A.1 Resumen de las Propiedades Electrónicas Extraídas de las Medidas de

2DEG Two-Dimensional Electron Gas

OTIs Organic Topological Insulators

To family and friends, and to you.

Part I: Introduction and Motivation

the Physics of Interfaces [1, 2].

F IGURE 1.3: Self-assembled molecular nanoporous networks on noble metal sur-

STM/STS (site specific) ARPES (ensemble information)

F IGURE 1.4: Electronic coupling of QD arrays on surfaces. (a) Conceptual

scattering strength. As a result, we engineer the surface band structure by locally

The resulting electronic structure from nanoporous networks that we study is

F IGURE 1.5: On-surface synthesized 1D and 2D covalent nanostructures. (a)

F IGURE 1.6: Ullmann coupling reaction and long-range ordering on surfaces.

semiconducting nanostructures. In the third part of this thesis, we introduce period-