Thesis IPZ Final
Thesis IPZ Final
Thesis IPZ Final
PhD Thesis
Ignacio Piquero-Zulaica
Supervisors:
Dr. Jorge Lobo-Checa
Prof. J. Enrique Ortega
E USKAL H ERRIKO U NIBERTSITATEA (EHU-UPV)
Author: Supervisors:
Ignacio Piquero-Zulaica Dr. Jorge Lobo-Checa
Prof. J. Enrique Ortega
in the
Nanophysics Lab
Centro de Física de Materiales (CFM-CSIC)
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].
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
Bibliography 169
xi
List of Figures
List of Tables
List of Abbreviations
Chapter 1
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]).
a b
c
d
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]
Chapter 1. Part I: Introduction and Motivation 5
a b
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
6 Chapter 1. Part I: Introduction and Motivation
a c
2 nm
1 nm 5 nm
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
8 Chapter 1. Part I: Introduction and Motivation
a b
10 nm
c
d
4 nm
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
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
• 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
• 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.
14 Chapter 2. Experimental Techniques and Theoretical Methods
a In-built QMB b
Thermocouple
Quartz crucible
W filament
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
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).
16 Chapter 2. Experimental Techniques and Theoretical Methods
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.
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 ).
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◦ .
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].
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
20 Chapter 2. Experimental Techniques and Theoretical Methods
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)
2.5. Experimental Techniques 21
KII
Vacuum
K┴ K
Surface
Solid k┴ k
kII
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].
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
22 Chapter 2. Experimental Techniques and Theoretical Methods
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].
2.5. Experimental Techniques 23
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].
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 >.
24 Chapter 2. Experimental Techniques and Theoretical Methods
~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
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.
2.5. Experimental Techniques 25
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
0.1 0.1
0.00
0.2 0.2
0.3 0.3 0.02
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].
26 Chapter 2. Experimental Techniques and Theoretical Methods
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]).
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,
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
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
28 Chapter 2. Experimental Techniques and Theoretical Methods
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.
2.5. Experimental Techniques 29
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
30 Chapter 2. Experimental Techniques and Theoretical Methods
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
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
-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
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 ).
32 Chapter 2. Experimental Techniques and Theoretical Methods
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].
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
2.5. Experimental Techniques 33
a Model
c
NC-AFM
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.
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.
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,
~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
36 Chapter 2. Experimental Techniques and Theoretical Methods
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.
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].
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.
38 Chapter 2. Experimental Techniques and Theoretical Methods
TABLE 2.2: Noble metal (111) vacuum level and bulk band gap edges. Values have
been taken from ref. [124]
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
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
Au(788)
c c
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]).
Chapter 3. Part II: An Introduction to Assembly and Electronic Bands of
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
Chapter 3. Part II: An Introduction to Assembly and Electronic Bands of
43
Nanoporous Networks
a b
c d
• 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
Chapter 3. Part II: An Introduction to Assembly and Electronic Bands of
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].
• 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?
Chapter 3. Part II: An Introduction to Assembly and Electronic Bands of
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].
Chapter 4
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].
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
0.0
0.0 Clean 0.0 0.25 ML 0.0 0.50 ML 0.0 0.70 ML
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
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*
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
4.2. Results and Discussion 51
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
Chapter 4. Temperature Dependence of the Partially Localized State in a 2D
52
Molecular Nanoporous Network
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)
kII (Å-1) kII (Å-1)
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.
4.2. Results and Discussion 53
a b c
-0.22
0.75 ML 136 K 136 K 0.50 ML
-0.24
-0.26
Int. (arb. units)
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
Chapter 4. Temperature Dependence of the Partially Localized State in a 2D
54
Molecular Nanoporous Network
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].
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].
Chapter 4. Temperature Dependence of the Partially Localized State in a 2D
56
Molecular Nanoporous Network
-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
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
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]
a Single Wall c e g
b d f h
Double Wall
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
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
60
Barrier Widths
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
5.2. Results and Discussion 61
-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
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)
Intensity (arb. units)
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
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
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)
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
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
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
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
5.2. Results and Discussion 65
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
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
68
Barrier Widths
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.
5.4. Supplementary Information for This Chapter 69
a b c d
e f g h
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.
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
70
Barrier Widths
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.
5.4. Supplementary Information for This Chapter 71
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).
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.
As it has been already mentioned, such halogen-bonded networks are only stable
below RT. Above this temperature, the surface catalyzed Ullmann coupling reaction
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
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
The simulations for this chapter have been performed in collaboration with Dr. Za-
karia M. Abd El-Fattah (Al-Azhar University, Egypt).
Simulation Procedure
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
kII (Å-1) kII (Å-1)
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.
Chapter 5. Precise Engineering of Quantum Dot Array Coupling Through Their
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)
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)].
5.4. Supplementary Information for This Chapter 75
SW
DW
Chapter 6
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.
a b c d
10 nm 5 nm 5 nm 2 nm
e f g h
10 nm 5 nm 5 nm 2 nm
(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)
6.2. Results and Discussion 79
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
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
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
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
-0.2
-0.2 -0.2 -0.2
E-EF (eV)
E-EF (eV)
E-EF (eV)
-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 )
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
-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
6.2. Results and Discussion 81
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.
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
82
Dot Arrays
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
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
Tunneling Voltage (V)
(V)
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
6.2. Results and Discussion 83
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]).
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).
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.
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
86
Dot Arrays
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
g h i ARPES
Cu(111)
Intensity (arb. units)
Intensity (arb. units)
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)
Tunneling Voltage (V)
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.
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
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. Supplementary Information for This Chapter 89
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).
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
90
Dot Arrays
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
Tunneling Voltage (V)
Tunneling Voltage (V)
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. Supplementary Information for This Chapter 91
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
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.
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
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. Supplementary Information for This Chapter 93
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].
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
Chapter 6. Tunable Energy and Mass Renormalization from Homothetic Quantum
94
Dot Arrays
of the projected bulk bands. For d∗ < d0 , on the other hand, the surface state shifts
to higher energies.
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
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).
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.
7.2. Results and Discussion 97
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)
STM/STS
STM/STS
PC
H
Cu
1 nm 1 nm 1 nm 1 nm M
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
Chapter 7. Effective Determination of Surface Potential Landscapes from
98
Metal-Organic Nanoporous Overlayers
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
7.2. Results and Discussion 99
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
1 nm 1 nm 1 nm min.
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
7.2. Results and Discussion 101
-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
Chapter 7. Effective Determination of Surface Potential Landscapes from
102
Metal-Organic Nanoporous Overlayers
(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
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*│(%)
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.
(0.016 V)
1 nm 1 nm
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.
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].
Chapter 7. Effective Determination of Surface Potential Landscapes from
106
Metal-Organic Nanoporous Overlayers
250°C
-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.
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. Supplementary Information for This Chapter 107
• 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
-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
Chapter 7. Effective Determination of Surface Potential Landscapes from
108
Metal-Organic Nanoporous Overlayers
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
7.4. Supplementary Information for This Chapter 109
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
1 nm
m*=0.58 m0
n=1 n=2 n=3 n=4
• 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
Intensity (arb. units)
-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
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
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)
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
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
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)
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].
-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
-0.23
-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 )
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
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
116
Quantum Dots
a 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
8.2. Results and Discussion 117
empty 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.
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
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
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
-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)
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
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
120
Quantum Dots
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.
8.4. Supplementary Information for This Chapter 121
a Xe Core Levels b
-5
-5
-6
-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).
Chapter 8. Configuring Electronic States in an Atomically Precise Array of
122
Quantum Dots
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.
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].
8.4. Supplementary Information for This Chapter 123
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
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
Zigzag Covalent Polymers
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]).
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
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
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
(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
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
132
One-Dimensional Organic Semiconductors
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
10.2. Results and Discussion 133
a b
Condensed chain
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
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
134
One-Dimensional Organic Semiconductors
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
10.2. Results and Discussion 135
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
-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).
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
136
One-Dimensional Organic Semiconductors
(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
10.2. Results and Discussion 137
a c
Intensity (arb. units)
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
1.5
2.0
2.5
b
Line position at (a) inset
-2.1 V
aaa
bbb
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
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
138
One-Dimensional Organic Semiconductors
a b
-2.5
-2.0
-1.5
-1.0
1.0
1.5
2.0
2.5
N = 10
bbb
aaa
N=7
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
10.2. Results and Discussion 139
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
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 (°)
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. Supplementary Information for This Chapter 141
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].
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.
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
142
One-Dimensional Organic Semiconductors
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.
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
10.4. Supplementary Information for This Chapter 143
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)].
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
144
One-Dimensional Organic Semiconductors
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
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.
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.
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
146
One-Dimensional Organic Semiconductors
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
-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 )
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.
Chapter 10. Electronic Structure Tunability by Periodic meta-Ligand Spacing in
148
One-Dimensional Organic Semiconductors
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.
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.
10.4. Supplementary Information for This Chapter 149
N=4 DFT
2.4
2.4
STS 3.8
3.8
Bandgap from DFT (eV)
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
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.
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
-0.5 -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:
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 – –
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
a 2π/L 2π/a* b
kx integrated Sample Bias (V)
0.0
0.0
-2.5
-2.0
-1.5
-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.
156 Chapter 11. Conclusions and Outlook
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:
a b c
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).
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
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.
a b
4
E-EF (eV)
2
c +2.3 V
1
G X
Appendix A
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
-0.5 -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)
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 – –
a 2π/L 2π/a* b
kx integrated Sample Bias (V)
0.0
0.0
-2.5
-2.0
-1.5
-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)
Appendix B
List of Publications
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
[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
[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-
wandte Chemie International Edition 46. 5 (2007), pp. 710–713.
[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”.
In: Physical Review B: Condensed Matter and Materials Physics 88. 24 (2013),
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
[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
[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.