Surface and interface structure of
electrochemically grafted ultra-thin organic
films on metallic and semiconducting materials
vorgelegt von
MSc Phys.
Ecatherina (Katy) Roodenko
aus Tel-Aviv
von der Fakult¨at II - Mathematik und Naturwissenschaften
der Technischen Universit¨at Berlin
zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften
Dr. rer. nat.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. E. Sedlmayr
Berichter: Prof. Dr. N. Esser
Berichter: Prof. Dr. C. Thomsen
Tag der wissenschaftlichen Aussprache: 14.12.2007
Berlin 2008
D 83
1
Parts of this work were already published in:
Gensch M., Roodenko K., Hinrichs K., Hunger R., G¨uell A. G., Merson A.,
Schade U., Shapira Y., Dittrich Th., Rappich J., Esser N. ”Molecule-solid inter-
faces studied with infrared ellipsometry: ultrathin nitrobenzene films.”, J. Vac.
Sci. and Technol. B 23 1838 (2004).
Rappich J., Merson A., Roodenko K., Dittrich Th., Gensch M., Hinrichs K.,
Shapira Y. ”Electronic properties of Si surfaces and side reactions during elec-
trochemical grafting of phenyl layers.”, J. Phys. Chem. B 110 1332 (2006).
Roodenko K., Gensch M., Heise H. M., Schade U., Esser N., Hinrichs K. ”Influ-
ences of thick film inhomogeneities on the ellipsometric parameters.”, Infrared
Phys. and Technol. 49 39 (2006).
G¨uell A. G., Roodenko K., Yang F., Hinrichs K., Gensch M., Sanz F., Rap-
pich J. ”Interface properties and passivation of p-Si(111) surfaces by electro-
chemical organic layer deposition.”, Mater. Sci. and Eng. B 134 273 (2006).
Roodenko K., Rappich J., Gensch M., Esser N., Hinrichs K., Hunger R. ”Time-
resolved Synchrotron XPS monitoring of irradiation-induced nitrobenzene reduc-
tion for chemical lithography.”, J. Phys. Chem. B 111 7541 (2007).
Roodenko K., Rappich J., Gensch M., Esser N., Hinrichs K. ”Studies of electro-
chemically grafted thin organic layers on inorganic surfaces with infrared spec-
troscopic ellipsometry.”, Appl. Phys. A 90 175 (2008).
Contents
1 Introduction 4
2 Electrochemical surface modification 7
2.1 Aryl diazonium compounds: tailoring of the surface properties . 8
2.2 Electrochemical grafting . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Electrochemical cell . . . . . . . . . . . . . . . . . . . . . 10
2.2.2 Charge transfer from electrode into electrolyte . . . . . . 13
2.2.3 Side reactions during the electrochemical grafting processes 16
2.3 Preparation of silicon surfaces . . . . . . . . . . . . . . . . . . . . 17
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 Optical modeling 20
3.1 Fundamental transitions and overtones . . . . . . . . . . . . . . . 20
3.2 Lorentz dispersion model . . . . . . . . . . . . . . . . . . . . . . 22
3.2.1 Extension of Lorentz dispersion model for amorphous solids 24
3.3 Propagation of polarized light in stratified media . . . . . . . . . 25
3.4 Application of the optical models for simulations of IR ellipso-
metric spectra: an example of hydrogen–passivated Si(111) surface 30
4 Experimental methods 33
4.1 Infrared Spectroscopic Ellipsometry (IRSE) . . . . . . . . . . . . 34
4.1.1 IRSE setup . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1.2 Measurements of the ellipsometric parameters . . . . . . . 34
4.1.3 Broadband sources of IR radiation . . . . . . . . . . . . . 36
4.1.4 Detectors of IR radiation . . . . . . . . . . . . . . . . . . 38
4.2 X-ray photoelectron spectroscopy . . . . . . . . . . . . . . . . . . 44
4.2.1 Deconvolution of XPS spectra . . . . . . . . . . . . . . . . 46
4.2.2 Evaluation of the XPS spectra . . . . . . . . . . . . . . . 46
5 Optical properties of organic thin films 50
5.1 IR properties of tetrafluorborate aryldiazonium compounds . . . 51
5.2 Nitrobenzene on Au, Si(111) and TiO2surfaces . . . . . . . . . . 54
5.2.1 IRSE characterization of nitrobenzene films . . . . . . . . 54
5.2.2 Determination of optical constants . . . . . . . . . . . . . 58
2
CONTENTS 3
5.2.3 Thickness determination and studies of the chemical com-
position of nitrobenzene films using combined XPS and
IRSE analysis . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2.4 IRSE studies of temperature–induced desorption . . . . . 64
5.3 Methoxybenzene on Au, Si(111) and TiO2surfaces . . . . . . . . 66
5.4 Electrochemical grafting on porous silicon . . . . . . . . . . . . . 68
5.4.1 IRSE characterization of PSi: comparative studies with
Si(111) and Si(001) . . . . . . . . . . . . . . . . . . . . . . 71
5.4.2 Organic modification of porous silicon . . . . . . . . . . . 73
5.4.3 The Si–C bond: discussion . . . . . . . . . . . . . . . . . 75
5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6 Passivation and oxidation of Si surfaces 78
6.1 Stability of H-passivated Si (111) surfaces . . . . . . . . . . . . . 79
6.2 Oxidation under atmospheric conditions . . . . . . . . . . . . . . 82
6.3 Determination of the optical parameters in mid–IR spectral range
for SiOxlayer forming under ambient conditions on Si(111) surface 86
6.4 SiOxinterface formation during the electrochemical grafting . . . 89
6.4.1 Stability of the organic films on oxidized surfaces to HF
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7 X–ray induced reduction of nitrobenzene 97
7.1 Overview of the X–ray irradiation induced changes on the ob-
served core level spectra . . . . . . . . . . . . . . . . . . . . . . . 99
7.2 Deconvolution of the N1s core level . . . . . . . . . . . . . . . . . 101
7.2.1 Dynamics of the integrated intensities . . . . . . . . . . . 103
7.3 Deconvolution of the C1s core level . . . . . . . . . . . . . . . . . 104
7.3.1 Dynamics of the integrated intensities . . . . . . . . . . . 106
7.4 Deconvolution of the O1s core level . . . . . . . . . . . . . . . . . 106
7.4.1 Dynamics of the integrated intensities . . . . . . . . . . . 109
7.5 Deconvolution of the Si2p core level and dynamics of the inte-
grated intensities . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
8 Concluding remarks 115
A Simulations of the IRSE spectra 118
A.1 Spectroscopic properties of thin films . . . . . . . . . . . . . . . . 118
A.2 Best-fit calculations . . . . . . . . . . . . . . . . . . . . . . . . . 119
A.3 Multiple-angle measurements routine . . . . . . . . . . . . . . . . 123
Acknowledgments 140
Chapter 1
Introduction
The aim of this work was to characterize the electrochemically deposited ultra–
thin organic films on metallic and semiconducting surfaces. The understanding
of the thin film composition, of the orientation of molecules in the organic lay-
ers and of the film/substrate interface structure is essential for optimizing the
preparation conditions. Improvement of such hybrid organic/inorganic materi-
als is important in many engineering applications, as for instance in photovoltaic
and other optoelectronic technologies.
Electrochemistry is a non-vacuum technique and it does not require elevated
temperatures for the deposition of organic molecules [1]. It is typically carried
out in liquid electrolytes, and allows a direct reaction between the radicals in
the electrolyte and the electrode surface. Electrochemistry can be used for
deposition of organic molecules in the sub-monolayer regime. The control over
the electrode potential dictates the deposition rate, interface properties and
the structure of the organic layer. For understanding of the electrochemical
processes governing the structure and composition of hybrid organic/inorganic
materials characterization of the surface and interface properties is essential.
In this work, the deposition of ultra-thin organic films was performed from
benzenediazonium salt solutions. Deposition occurs when the voltage is applied
on the electrode, which leads to the formation of radicals in the electrolyte
through the reduction of the benzenediazonium salts. The free radicals are
highly reactive and their interaction with the electrode leads to the growth of
ultra thin organic layers on its surface. There is a large variety of benzenediazo-
nium compounds with different chemical groups bounded to the benzene ring.
This work presents the characterization of the nitrobenzene (C6H5NO2), bro-
mobenzene (C6H5Br), methoxybenzene (C6H5OCH3) and 4–methoxydiphenyl
amine (C6H5–NH–C6H5–OCH3) thin films on metallic and semiconducting elec-
trodes.
Such diversity of functional groups attached to the benzene ring offers an
opportunity to tailor the surface properties. Covalent attachment of the or-
ganic molecules influences, for example, the surface electron affinity, passivates
surface gap states and changes its adsorption behavior, chemical reactivity, wet-
4
CHAPTER 1. INTRODUCTION 5
ting, radiation absorption, adhesion, and biocompatibility [2, 3, 4]. Therefore,
applications of the hybrid organic/inorganic materials are inherently diverse and
stretch from photovoltaic applications [5,6,7] to biosensor technology [8,9,10].
Although the technological relevance of organically modified materials is
greatly recognized, their physical and chemical properties, such as reactivity,
molecular structure and their optical and electronic properties are the subjects of
fundamental research. Improvement of the performance of the organic/inorganic
devices demands a good understanding of the growth mechanisms and of the
surface and interface structure. Therefore the characterization of electrochemi-
cally modified surfaces was undertaken in this work.
The characterization work presented in this thesis was performed using sur-
face sensitive techniques, such as infrared spectroscopic ellipsometry (IRSE)
and X-ray photoelectron spectroscopy (XPS). The measurements addressed the
following important questions:
Structure of the ultra-thin organic films
Side reactions that take place during the deposition
Structure of the organic/inorganic interfaces
Interface stability to the oxidation under atmospheric conditions
Further possibilities of surface engineering by X-ray irradiation of the thin
organic layers
The choice of the IRSE method as a characterization technique was due to
its high surface sensitivity to organic adsorbates, easiness of application and a
possibility of measurements under ambient atmospheric conditions. The XPS
technique provided complementary information on chemical composition and
thickness of the organic and interface layers.
Part of this work was dedicated to development of the simulations routines
based on the phenomenological optical models, which were necessary for inter-
pretation of the IRSE spectra. This allowed us to determine optical parameters
of the ultra thin organic films, their thickness and molecular orientation.
Since the uncontrolled oxidized interfaces are unwanted in device engineer-
ing, studies of interface oxidation and search for the ways to prevent it are
essential. One of the aims of this work was to characterize the interface silicon
oxide layer that forms between the grafted organic thin films and the silicon
substrates. It was of interest to study such oxidized interfaces that form under
different conditions: first, as a result of the side reactions with the surround-
ing aqueous solution, and second, as a result of the exposure to atmospheric
conditions.
The results presented in this work rely on the cross–correlated analysis which
involved mainly IRSE and XPS techniques. It allowed to perform a quantita-
tive study of the oxide formation at the silicon/film interface. Spectra delivered
by XPS technique provided information on chemical bonds in thin films and
organic/inorganic interfaces. Deconvolution of the core level spectra gave an
CHAPTER 1. INTRODUCTION 6
insight onto the sub-oxide structure of the Si/SiOx/film interfaces. Studies of
the surface roughness as a result of the oxidation taking place at the organ-
ically protected and unprotected surfaces were performed using atomic force
microscopy (AFM).
A special attention in this work was given to the possibility to modify
the chemical structure of thin films by X-ray irradiation. This work explores
the reduction of nitrobenzene on silicon surfaces (Si-C6H4NO2) to aniline (Si-
C6H4NH2) upon X–ray irradiation. This subject is of interest since it allows a
biological compatibility of the surface through the NH2bio–reactive functional
group. The components of the reduction process are proposed upon a detailed
deconvolution of the observed core levels.
This dissertation summarizes the work which addressed the above issues.
Chapter 2 introduces electrochemistry as a surface modification method. The
principles of the electrochemical cell are described along with the model for the
grafting procedure. Creation of the radicals in the electrolyte solution and their
subsequent attachment to the surface of the solid electrode are discussed. In
addition, possible pathways for side reactions that may take place upon charge
transfer in the aqueous electrolyte are presented.
Chapter 3 outlines the mechanisms of infrared absorption through the molec-
ular vibrations in the investigated material. Furthermore, the ellipsometric pa-
rameters as well as the models that were used for IRSE spectral interpretations
are discussed. The dispersion model for simulations of the IRSE spectra and the
calculations of the radiation propagation in a stratified media are presented. In
Chapter 4, experimental methods and data analysis techniques are introduced.
This includes a detailed presentation of the components of the IRSE setup, and
the discussion of the XPS data analysis.
In chapter 5, characterization of metallic and semiconducting surfaces mod-
ified with various benzene derivatives is presented. An emphasis is placed on
methoxybenzene (C6H5OCH3) and nitrobenzene (C6H5NO2) modified surfaces.
Simulations of the IRSE spectra are applied to evaluate the optical properties
of the ultra thin organic films. Comparison with the data obtained from XPS
measurements is performed for a cross referenced analysis of the thickness and
the chemical composition of the surface adsorbates.
Chapter 6 introduces a detailed characterization of the SiOxinterface be-
tween the organic films and silicon surfaces. Here, oxidation of the silicon sur-
faces during the electrochemical grafting and its prevention are presented. A
comparative analysis between the stability of the organically modified surfaces
and the unmodified hydrogen passivated silicon surfaces to oxidation in atmo-
spheric condition is performed.
Chapter 7 presents the process that converts the NO2nitro groups of elec-
trochemically grafted nitrobenzene on Si surfaces into NH2amino groups upon
X–ray irradiation. This chapter proposes a detailed mechanism for this conver-
sion and suggests several intermediate species on the reaction pathway.
The last chapter provides a survey and conclusions of this work.
Chapter 2
Electrochemical surface
modification with
ultra–thin organic films
Functionalization of surfaces with organic thin films allows to control the ma-
terial interfacial properties, which is important in development of device tech-
nology [11,3]. Methods for deposition of thin organic films on metallic or semi-
conducting surfaces can be in general subdivided into physical [12, 13] and
chemical [14,15] modifications. In case of a physical modification, physisorption
of molecules on substrates takes place with a Coulomb interaction between sur-
face and organic molecules. Chemical modification is achieved through a cova-
lent bond of the deposited molecules with the surface (chemisorption). Surface
preparation methods are diverse, stretching from ultra-high vacuum deposi-
tion [16,17,18,15] to wet–chemical preparation [19,20]. Wet–chemical prepara-
tion is of advantage for technological applications, since it can be carried out
under atmospheric conditions in suitable solutions. Non–vacuum methods for
surface preparation include Langmuir–Blodgett, spin–coating, electrochemical
grafting and many others [11,21].
This work focuses on characterization of ultra-thin organic films deposited
electrochemically from aryldiazonium salts on inorganic electrodes. The fol-
lowing sections are organized as follows: first, a motivation for the organic
modification from the aryldiazonium compounds will be given. Next, a detailed
description of electrochemical method and its application to thin film prepara-
tion is presented. The issues related to charge transfer between the electrode
and electrolyte, as well as possible side reactions connected with the grafting
process, are discussed.
7
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 8
2.1 Aryl diazonium compounds: tailoring of the
surface properties
Covalent attachment of the organic molecules to solid surfaces enables to tai-
lor the surface properties by tuning of the electron affinity and surface dipole
moment [6,22]. Molecules used in this work allow to control surface properties
by changing the functional group, X, attached at the para–position of the di-
azo compound from which the molecules are grafted on the surfaces, as shown
schematically in Fig. 2.1. Changing the functional group X of the molecules
X
N BF
2 4
+ -
functional
group
benzene
ring
diazonium
group
counterion
tetrafluorborate
Figure 2.1: Schematic drawing of the aryl diazonium tetrafluorborate
molecule with diazonium group N+
2and a variable group X to change be-
tween the electron acceptor and electron donor–like molecular properties.
In this work, the studies were performed with X=Br, NO2and OCH3.
changes their electronic properties and influences the properties of the host
surface accordingly [2, 3, 4]. In this work, surfaces modified with 4–bromo–
(X=Br), 4–nitro– (X=NO2) and 4–methoxy– (X=OCH3) benzenediazonium-
tetrafluoroborate compounds (4–BrBDT, 4–NBDT, and 4–MeBDT, respectively)
were studied.
Fig. 2.2 shows the energy band diagrams as proposed by Hunger et al [4]
based on the observation delivered by XPS measurements. Fig. 2.2 (a) shows
the schematic band diagram for a functionalized silicon surface and presents
the definitions of the related surface parameters. The work function WF is
defined as energy difference between the vacuum level, Evac and the Fermi level,
EF. The electron affinity of the surface, χ, is defined from the bottom of the
conduction band Ecb to the vacuum level Evac. The electron affinity χcan be
viewed as the modified ”intrinsic electron affinity” of the Si, χSi, by a dipole
contribution δwhich depends on the charge distribution at the interface and
within the adsorbate layer [4]:
χ=χSi +δ(2.1)
The step potential δdefined such that an increase of the electron affinity corre-
sponds to δ > 0.
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 9
Fig. 2.2 (b–e) show the effects of the modification of the silicon surfaces
with benzene derivatives carrying different functional groups X. The functional
group X influences molecular dipole moment [23]. When molecules attach to
the surface, they change the surface electron affinity χ. The electron affinity
depends on the molecular dipole moment and orientation of the molecules on
the surface. Thus, variation of the functional group X should in general enable
tailoring of the surface properties. However, processes that govern molecular
Si(111)-H
Evb
Ecb
Evac
EF
cSi cWF
Evbm
Eg
eVbb
d
eV =0.47eV
bb
d=
-0.27eV
eV =0.09eV
bb
d=
+0.36eV
eV =0.13eV
bb
eV =0.57eV
bb
Si(111)-C H NO
652
Si(111)-C H OCH
653
d=
+0.33eV
d=
-0.33eV
Si(111)-C H Br
65
a.
b. c.
e.d.
Figure 2.2: a. Energy band diagram of a functionalized silicon surface
with band bending, eVbb, and a surface dipole, δ, modifying the intrin-
sic electron affinity of silicon, χSi. b–e: Energy band diagrams of the
hydrogen-terminated p–Si(111)–H (not an ideal H–termination [4])(b);
nitrobenzene grafted onto p–Si(111) (c); methoxybenzene/p–Si(111) (d);
bromobenzene/p–Si(111) (e). After Hunger et al [4].
orientation of the electrochemically grafted molecules on the surfaces are not
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 10
well understood. As will be shown in the following sections, one of the possible
side reactions is the polymerization of the radicals in the electrolyte, which leads
to formation of unordered organic films carrying polymerized units. Besides,
oxidation of the organic/silicon interface may take place during the grafting in
aqueous solution. In this work characterization of the electrochemically grafted
films and the organic/inorganic interfaces was performed, in order to elucidate
the parameters that influence those unresolved issues of the interface and thin
film structure.
2.2 Electrochemical grafting
Electrochemical grafting is a method of covalent attachment of organic molecules
on a surface. A typical electrochemical cell consists of electrodes and ionic
conductor, or electrolyte. Application of a potential permits a charge–carrier
transfer between the surface and species in solution. Deposition of organic
monolayers proceed through a formation of radicals in electrolyte and their
attachment to electrode through the surface reactive sites. Electrochemically
grafted organic films construct the main scope of this thesis, thus a detailed
discussion of this method is presented in the following subsections.
2.2.1 Electrochemical cell
A typical electrochemical cell is shown in Fig. 2.3. This arrangement is known
as three-electrode cell, where the current is passed between the working electrode
and the counter (or auxiliary) electrode. The potential of the working electrode
is measured relatively to the reference electrode. The electrochemical grafting
V
i
Auxiliary
(counter)
electrode
Reference
electrode
Working
electrode
Power
supply
Figure 2.3: Principle of three-electrode electrochemical setup.
for instance proceeds upon application of the voltage on the working electrode.
When the working electrode is driven to negative potentials, the energy of the
electrons in the electrode is raised. When the Fermi level of the electrode is
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 11
above the lowest unoccupied molecular orbital (LUMO) of the species in the
solution, the electron transfer from the electrode to the molecules in the solution
is thermodynamically favorable (Fig. 2.4, upper panel). A flow of electrons from
EFermi
LUMO
HOMO
Electrode Solution
potential
+
-
EFermi LUMO
HOMO
Electrode Solution
e
applicationofnegative
potential
EFermi
LUMO
HOMO
Electrode Solution
potential
+
-
EFermi
LUMO
HOMO
Electrode Solution
e
applicationofpositive
potential
A +e A-
A -e A+
Figure 2.4: Representation of reduction (upper panel) and oxidation
(lower panel) processes of species A in solution. The shown molecular
orbitals of species A are the highest occupied molecular orbital (HOMO)
and the lowest vacant molecular orbital (LUMO). After ref. [24].
electrode to solution is called reduction current. By lowering the energy of the
electrons in a working electrode below the highest occupied molecular orbital
(HOMO), a thermodynamically favorable energy for electron transfer from the
species in the electrolyte to the electrode can be reached (Fig. 2.4, lower panel).
A flow of the electron from solution to electrode is called oxidation current.
In this work, the reported potential is referred to Au–electrode in the same
solution. The calculations of the charge transfer between the electrolyte and
the solid, taking into account the polarization fluctuations in solution can be
found in the textbooks – see for example [1].
The main part of this work is dedicated to the studies of surfaces organically
modified though electrochemical grafting from aryl diazonium salts. A model for
deposition of organic layers from aryl diazonium compounds on Si surfaces was
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 12
proposed recently by Allongue et al [25]. The initial stage involves formation of
the aryl radical through the reduction of the aryl diazonium ion:
≡N+
2−⊘−X+e−−−−−−→ •⊘−X+N2(2.2)
H
Si
Si
Si Si Si Si
Si Si Si
Si
HH
N2
H
Si
Si
Si Si Si Si
Si Si Si
Si
HHH
e-
XX
N+
N
+
where ⊘stand for benzene ring and •designates a radical. In the next step,
hydrogen is being removed from the Si surface leaving a dangling bond behind:
≡Si −H+•⊘−X−−−−−→ ≡ Si •+H−⊘−X(2.3)
H
H
Si
Si
Si Si Si Si
Si Si Si
Si
HH
SiSi
Si
Si
H
Si
Si Si Si
Si Si Si
Si
H
Si
X
X
The Si dangling bond can now react with aryl radicals in the electrolyte:
≡Si •+N+
2−⊘−Xe−, -N2
−−−−−→ ≡ Si −⊘−X(2.4)
e-
Si
H
Si
Si Si Si
Si Si Si
Si
H
Si
X
X
Si
Si
H
Si
Si Si Si
Si Si Si
Si
H
N+
N
+
,-N2
Organic monolayers from aryl diazonium salts can also be grafted on other
surfaces, for example, Au or TiO2[26,27,28].
This work also presents studies on surfaces modified with Grignard reagent
methylmagnesium halide (CH3MgX, where X = Cl, Br, or I). This method for
modification of silicon surfaces was explored for example by Boukherroub et
al [29] for deposition of alkyl adsorbates on H:Si(111). Electrochemical deposi-
tion leads to the reaction in which CH3MgX is split in an electrochemical process
into •CH3and MgX+, and the •CH3radical can be attached to the available
silicon dangling bond [30]. Studies of so methylated porous silicon surfaces are
presented in section 5.4.2.
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 13
2.2.2 Charge transfer from electrode into electrolyte
Electrochemical deposition can be monitored by recording a current flowing
between the electrodes in electrochemical cell as a function of time. Fig. 2.5
shows the current transients during electrochemical grafting of 4–nitrobenznene
(4–NB) onto hydrogenated Si(111) at different negative potentials [31]. Initially,
no charge is flowing between the electrode and electrolyte. However, with the
addition of the diazonium salt (at time t=0 sec in Fig. 2.5), the current increases
due to the initiation of the charge transfer process. After the molecules are
grafted on the surface, they block the further charge transfer from the electrode
into the electrolyte, thus the current levels out at longer deposition times. Such
process is called a self–limiting process. This situation is schematically depicted
in Fig. 2.6, where the blocking effect of the created film is represented by a
potential barrier. On the other hand, grafting of the molecules on a chemically
-0.8VSi(111)
-0.9V
-1.2V
1.25mM4-NBDT
in0.01MHSO
2 4
Current( )mA
0
-50
-100
-150
0 5 10
Time(sec)
Figure 2.5: Current transients during electrochemical grafting from 4–
NBDT onto hydrogenated Si(111) at different negative potentials, typical
for a self–limiting process. Time scale has been set to zero when the
diazonium salt was added to the solution. Adopted from ref. [31].
oxidized Si(111) surface shows a non-self-limited behavior, as shown in Fig. 2.7.
Also here, the addition of the molecules into electrolyte is characterized by a
steep increase of the current, as indicated by arrow in Fig. 2.7. However, in
contrast to the previous example, only a slight decay in current as a function of
time can be seen due to the growing film thickness and thick organic layers can
be deposited on chemically oxidized Si(111) surfaces. Hartig et al [2] explained
these differences on basis of the molecular orientation. Deposition of the p–
nitrobenzene parallel to the amorphous SiO2surface allows charge-transfer via
the π-bonding system, while on hydrogenated Si(111) surface, charge transfer is
limited by molecules oriented perpendicularly with the molecular plane to the
surface [2]. However, it is also possible that the properties of the silicon oxide
layer itself are responsible for such non–limiting grafting behavior. The charge
transfer from the silicon through the silicon oxide layer can be related to the
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 14
e-
e-
Si
electrolyte
film
electrolyte
Si
a. b.
Figure 2.6: Schematic drawing of the silicon/electrolyte system and of
the electron–transfer related potentials. a. Initial situation of the elec-
tron transfer under applied potential when no film was yet grown. b.
Formation of the grafted film blocks the electron transfer from the elec-
trode into electrolyte. When the film layer thickness is in the range of
20–25˚
A , tunneling is in principle possible [1]. Charge transfer can also
be assisted by an externally applied potential.
conduction through the defects in the oxide film or incorporation of protons
from electrolyte solution [32,33]. The surface of the chemically oxidized silicon
is rough [34,2,35], which also affects the charge transport from the electrode into
electrolyte. It will be shown in the following sections that the results obtained by
Current( )mA
Time(sec)
-1
-10
-100
10 100 10001
addition
ofdiazonium
0.01MHSO
2 4
-1.2V
Figure 2.7: Time dependence of the current during the electrochemical
grafting of p-nitrobenzene molecules at the chemically oxidized Si(111)
surface, typical for a non self–limiting process. The time position of
adding the diazonium salt is marked by an arrow. Adopted from ref. [2].
infrared spectroscopic ellipsometry did not indicate any preferential orientation
on neither of the substrates. However, such charge-transfer behavior might be
dictated by the very first layer deposited on various substrates. This situation
was not detected by us, due to the high reactivity of the molecules and their
polymerization during the grafting process, as will be presented further in this
work.
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 15
Studies of reversibility of grafting are frequently performed using the stan-
dard method in electrochemistry of cyclic voltammetry. Cyclic voltammetry has
become a useful technique for initial electrochemical studies of new systems and
can reveal information on complicated electrode reactions. It is based on chang-
ing of the applied voltage on the electrodes, and measurements of the resulting
charge transfer. Fig. 2.8 shows schematically the principle of cyclic voltamme-
try. As the applied potential increases, the oxidation occurs and a current due
to the electron transfer from the species that are being oxidized to the electrode
flows. When the potential is reversed at time t=t0, reduction occurs and the
current due to the electron transfer from the electrode to the species that are
being reduced flows. The detailed theory and calculations of the charge trans-
fer between the electrolyte and the solid, based on calculations of Marcus and
Gerischer, can be found in the textbooks – see for example [1]. The earlier
time(sec)
Potential(V)
i(A)
Potential(V)
i(A)
(+)
(-)
(-) (+)
t0
RedOx+e-
RedOx+e-
Figure 2.8: Left: cyclic potential sweep and the resulting signal as a
function of time. Right: the resulting signal as a function of a potential.
studies on grafting of organic molecules from benzenediazonium compounds re-
vealed that the functional group of the benzenediazonium has a strong influence
on the grafting potential of the molecules on the electrode surface. This was
shown for example in ref. [36] on metals. Deposition of thin organic films from
aryl diazonium salts on TiO2surface and monitoring using cyclic voltammetry
was performed by Merson et al [26]. The results are summarized in Fig. 2.9
for nitrobenzene (NB), methoxybenzene (MeB) and bromobenzene (BrB) de-
position. All cyclic voltammograms exhibited current increase due to reduction
process of the aryl diazonium salts and formation of the organic layer at the
electrode. The absence of the oxidation peak during the back scan revealed
an irreversible process due to the decomposition of the diazonium cation. The
potential of maximum current (Upeak) was found to vary for each type of the
grafted molecules. For NB Upeak occurred at -0.87 V, for BrB at -0.95 V and for
MeB at -1.04 V. The Upeak position was found to be correlated with the dipole
moment values of the studied molecules. In addition, a double-peak structure
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 16
-1.5 -1.0 -0.5 0.0
Potential(V)
0.0
-0.1
Current(mA)
-NO
-OCH -Br
2
3
Figure 2.9: Cyclic voltammetry of TiO2in a 5 mM solution of nitroben-
zene compound (-NO2),bromobenzene compound (-Br) and methoxyben-
zene compound (-OCH3) in 0.01 M H2SO4.The results are taken from
ref. [26]
was observed on each voltammogram. The double peak was explained by the
presence of two different sites (namely -Ti-O- and -Ti-OH) at the TiO2sur-
face from which electrons can be transfered into the benzene compound. The
presence of -Ti-OH surface groups is caused by protonation in acid electrolyte.
2.2.3 Side reactions during the electrochemical grafting
processes
In addition to the desired process of molecular grafting, side reactions can take
place in the electrochemical cell. Eq. 2.5 shows the reaction for hydrogen evolu-
tion, upon which the charge transfer from an electrode reduces protons in acidic
solution, forming atomic hydrogen.
H++e−→ •H(2.5)
Additional side reactions involve reactions of the radicals with water. For elec-
tron acceptors, such as nitrobenzene (R=NO2), possible reaction is shown in
eq. 2.6:
R−⊘•δ++δ−OHδ+
2→R−⊘−OH +•H(2.6)
While for electron donors, such as methoxybenzene (R′=OCH3), the reaction
may lead to creation of the OH radicals in the electrolyte:
R′−⊘•δ−+δ+H2Oδ−→R′−⊘−H+•OH (2.7)
The reactions described in Eqs. 2.5, 2.6, 2.7 lead to creation of •Hatoms or
•OH radicals that can participate in a creation of (and a reaction with) the
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 17
dangling bonds at Si surfaces:
•H+H−Si →H2+•Si (2.8)
•OH +H−Si →H2O+•Si (2.9)
The •OH radicals can also react with Si dangling bonds which can lead to the
back-bond oxidation at the Si surface:
Si −OH +Si −OH →Si −O−Si +H2O(2.10)
Additional side-reactions in the electrolyte involves polymerization of organic
radicals:
R−⊘•+•⊘−R→R−⊘−⊘ −R(2.11)
As can be seen from the above equations, the side reactions require electron
transfer from the electrode into electrolyte and will influence the current that
flows between the working and the counter electrode upon application of voltage.
This makes it difficult to estimate the reaction rates from the measured current
and additional characterization techniques are necessary for measurements of
the amount of grafted molecules.
2.3 Preparation of silicon surfaces
Two types of organically modified silicon substrates were used in this work:
Si(111) and porous silicon. The main break-through in preparation of ideally
hydrogen-terminated Si surfaces by chemical etching was performed by Chabal
and co-workers. They discovered that the pH of a fluoride solution has an
influence on flattening of the treated surfaces [37, 38]. The pH change can be
achieved through the buffering of HF solutions by addition of NH4F and/or
NH4OH [38]. This work prompted many studies and it was soon established
that pure 40% NH4F solution (pH=7–8) leads to the formation of hydrogen
terminated (111) surface terraces with up to some hundreds of nm width [39,
37,38].
Fig. 2.10 shows schematically the model for HF etching as proposed by
Trucks et al [40]. Initially, removal of the hydroxide groups from the silicon
surface and termination of the Si dangling bond with fluorine is illustrated
(Fig. 2.10(a)). The ionic nature of the Si–F polarizes the silicon back bond as
shown in Fig. 2.10(f). This polarization allows the insertion of HF into a Si–Si
bond. Steps (b)–(e) in Fig. 2.10 represent the etching sequence terminating
with the removal of the surface silicon atom as SiF4unit. From the other hand,
porous silicon, as it name suggests, consists of the pores that can be created
by etching of silicon atoms from atomically flat silicon surfaces as discussed
in every detail elsewhere [41, 42, 43, 44, 45]. The importance of porous silicon
(PSi) was recognized since the discovery of Canham in 1990 that nanocrystalline
porous silicon can emit visible light through photoluminescence at room tem-
perature [46]. However, while the construction of working devices based on PSi
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 18
Si
Si
Si Si
F
H-F H-F
Si
Si
Si Si
F
H
F
Si
Si Si
Si
Si Si
F
H
F
F
H
Si
Si
Si Si
H-F
HF
+H2O
+
Si
Si
Si Si
OH
Si
Si
Si Si
F
+2+
F
Si
FF
Si
Si Si
H
HH
Si
Si
Si
a.
b. c. d. e.
(d )
-(d )
+
f.
Si Si F
F
Figure 2.10: HF etching process as proposed by Trucks et al [40]. See
text for the description of the process.
was reported [47,48,41], stable and strong photoluminiscence is hard to achieve.
One of the promising approaches to overcome this problem can be the grafting
of special molecules on to the PSi surfaces [20].
The primer interest in characterization of electrochemically modified PSi in
this work was its larger surface area, and thus a higher amount of the grafted
molecules per investigated spot as compared to the organically modified Si(111)
and Au surfaces. In turn, this should enable a higher signal–to–noise ratio
for weaker absorption bands, among them the Si–C or Si–O–C bonds that are
expected to be formed in electrochemical grafting. Additionally, we expected to
observe in the obtained IRSE spectra the effects due to the orientation of the
grafted molecules on the walls (or the pores) of PSi.
In this work, porous silicon was formed by an anodic treatment in 50%
HF:ethanol (1:1) for several minutes. The IRSE spectra obtained from freshly
prepared as well as organically modified porous silicon are presented in sec-
tion 5.4.
2.4 Summary
In this chapter, electrochemical methods for organic modification of solid sur-
faces were presented. These methods do not require ultra–high vacuum systems
and can be performed under room temperature conditions. By deposition of
suitable molecules, surface properties (for example, surface band-bending [4])
can be controlled. However, electrochemical grafting is a complex process which
involves side–reactions such as surface oxidation and polymerization of reactive
radicals. In the next chapters characterization of the electrochemically modi-
CHAPTER 2. ELECTROCHEMICAL SURFACE MODIFICATION 19
fied surfaces is presented using dedicated surface sensitive analytic techniques,
such as Infrared Spectroscopic Ellipsometry (IRSE) and X-ray Photoelectron
Spectroscopy (XPS). These methods enable identification of surface adsorbates,
reveal the properties of organic/inorganic interfaces and the structure of thin
films. Based on these methods, the results of studies on influences of the prepa-
ration conditions on the structure of the organically modified surfaces are pre-
sented in this work. Stability of the modified surfaces to ambient conditions
and their chemical reactivity upon X–ray irradiation will be discussed.
Chapter 3
Optical modeling
Surface analytical methods based on Fourier-Transform Infrared (FTIR) spec-
troscopy constitute an important part of this work. FTIR spectroscopy is a
widely used method for investigations of vibrational absorption properties of
liquids, gases and solids. FTIR–based transmission and reflection techniques
play a significant role in studies of molecular ordering and orientation on var-
ious surfaces: from studies of carbon monoxide chemisorption on metals [49]
and preparation of flat hydrogen-terminated Si(111) surfaces [50] to studies of
ultra-thin organic films properties [20].
This chapter is dedicated to the discussion of the interpretation of the FTIR
spectra based on optical models. The opening section introduces a model for
absorption of the infrared radiation by excitations of the molecular vibrations.
Next, we introduce the models that were used throughout this work for the
IR data evaluation. These include the classical Lorentz dispersion model and
a model for propagation of the electromagnetic radiation in stratified media.
Finally, example demonstrating the application of these models for evaluation
of the measured spectra is given.
3.1 Fundamental transitions and overtones
In IR spectroscopy, the vibrational excitation is achieved by sample irradiation
with a broad-band radiation source. The molecule is excited to a higher vibra-
tional state directly by absorption of IR radiation. Atoms bound in a molecule
can have only quantized energies. A simple way to describe molecular vibrations
is to consider a diatomic molecule. The vibration of two nuclei in a diatomic
molecule can be reduced to the motion of a single particle with reduced mass µ,
whose displacement qfrom it’s equilibrium position is equal to the change of the
internuclear distance. With m1and m2being masses of two nuclei, the reduced
mass reads 1
µ=1
m1+1
m2. Assumption of a simple parabolic potential function
is shown by a dashed curve in Fig. 3.1. In this case, the system represents a
20
CHAPTER 3. OPTICAL MODELING 21
harmonic oscillator, with the potential approximated by:
V=1
2Kq2(3.1)
where K is the force constant. The Schr¨odinger wave equation becomes:
d2Ψ
dq2+8π2µ
h2(E−1
2Kq2)Ψ = 0 (3.2)
The eigenvalues of this equation are:
E=hν(m+1
2) = hceν(m+1
2) (3.3)
where eν= 1/λ is the wavenumber with dimension cm−1, notation commonly
used in vibrational spectroscopy, h is the Planck constant and mis the vibra-
tional quantum number with integer values 0,1,2... The vibration frequency is
then given by:
ν=1
2πsK
µ(3.4)
or
eν=1
2πcsK
µ(3.5)
The actual potential however can be approximated more precisely by adding a
cubic term (solid line in Fig. 3.1):
V=1
2Kq2−Gq3(K >> G) (3.6)
Which leads to the following eigenvalues:
Em=hcwe(m+1
2)−hcxewe(m+1
2)2+.... (3.7)
where weis the wavenumber corrected for anharmonicty, and xeweindicate the
magnitude of anharmonicity. Eq. 3.7 shows that the energy levels of anharmonic
oscillator are not equidistant, and the separation decreases slowly as νincreases.
The anharmonicity is responsible for appearence of overtones and combination
vibrations. From Eq. 3.7, one obtains:
(Em−E0)/hc =mwe−xewe(m2+m) + ... (3.8)
Thus resulting in the following relations for the fundamental and the over-
tone vibrations:
Fundamental: eν1=we−2xewe
First overtone: eν2= 2we−6xewe
CHAPTER 3. OPTICAL MODELING 22
Second overtone: eν3= 3we−12xewe
Where the values of weand xewecan be determined by observation of the
overtones in the IR and Raman spectra. The coupling between several funda-
Internucleardistance
Energy
Continuum
n=0
hn
1
2
Figure 3.1: Morse potential energy curves [51] and vibrational energy lev-
els for a diatomic molecule. Dashed line: parabolic potential, dotted line:
cubic potential, solid line: actual potential including the anharmonicity
correction. Adopted from ref. [52].
mental modes gives rise to the combination absorption bands. Both combina-
tion and overtone bands are typically much weaker than the fundamental bands.
Fig. 3.2 shows transmission IR spectrum obtained from a low density polyethy-
lene (LDPE), which demonstrates the absorption peaks due to fundamental
modes of vibration marked between 650 and 1650 cm−1, and the absorption
peaks arising due to the combinations of those bands, as marked between 1650
and 2650 cm−1[53]. Some of the combination bands include contributions from
the Raman–active modes in the spectral range between 1000 and 1800 cm−1, as
assigned in detail by Nielsen and Holland [53].
It is important to mention selection rule for detection of the molecular vi-
brations using IR techniques. The vibration is IR–active if the transition dipole
moment µis changed during the vibration:
∂µ
∂q 6= 0 (3.9)
Where qis the normal coordinate. Thus, for example, vibrations of homopolar
diatomic molecules such as N2or O2are not IR-active.
3.2 Lorentz dispersion model
In this work, the calculations of the optical properties of organic thin films
were based on a macroscopic approach, employing harmonic oscillator model in
CHAPTER 3. OPTICAL MODELING 23
650 1150 1650 2150 2650 3150
0.0
0.2
0.4
0.6
0.8
Wavenumber(cm)
-1
Transmission
726
1295
1305
1368 1350
726
1894*
2016*2144*
2345*2415*
Figure 3.2: Transmission spectrum of low density polyethylene (LDPE)
demonstrating the absorption peaks due to fundamental modes of vibra-
tion marked between 650 and 1650 cm−1, and due to the combinations
of those bands, as marked between 1650 and 2650 cm−1(starred). The
assignment is based on ref. [53].
Lorentz formulation. The dielectric function is defined in terms of the electric
field Eand the polarization Pthrough the Maxwell equations (SI units) [54]:
D=ǫ0E+P=ǫ0ǫE(3.10)
where Dis the dielectric displacement, ǫ0is the free space permittivity, ǫis the
dielectric function, Eis the electric field vector.
Molecular vibrations can be described as simple harmonic oscillators. The
equation of motion of a charged harmonic oscillator with mass mand charge Q
in response to the driving oscillating electric field Eis given by [55]:
md2r
dt2=QE−mω2
0~r −mΓdr
dt (3.11)
where ris the electron displacement, w2
0is the resonant frequency and Γ is the
damping constant. In this equation, the term QE(ω) arises from the driving
force generated by the electric field, mω2
0ris the restoring force, and the last
term is due to the damping force. This equation describes the molecular vibra-
tions [56,57] as well as the lattice absorption through optical phonons [55]. The
solutions for displacement from equilibrium rcan be expressed as:
r=r0exp(i(k·r)−wt) (3.12)
Substitution of Eq. 3.12 into Eq. 3.11 results in:
r0=QE
ω2
0,j −ω2+iωΓj
(3.13)
CHAPTER 3. OPTICAL MODELING 24
The charged harmonic oscillators produce a macroscopic polarization P:
P=NQr(3.14)
where Nis the density of the harmonic oscillators. Substitution of 3.14 and 3.13
into 3.10 leads to the following expression:
ǫ= 1 + NQ2
ω2
0,j −ω2+iωΓj
(3.15)
For ω≫ω0,ǫ→1 , while for ω≪ω0,ǫ→1 + NQ2
ω0. In IR spectral range,
the radiation field appears to be static to the electrons and thus high–energy
contributions due to electronic transitions can be considered constant in the
IR spectral range. This contribution is generally referred to as high frequency
dielectric constant and denoted as ǫ∞. When Eq. 3.15 is generalized to include
a collection of single harmonic oscillators and ǫ∞is included to the dielectric
function, one obtains:
ǫ(ω) = ǫ∞+X
j
(Fj
ω2
0,j −ω2+iωΓj
) (3.16)
where the summation is performed on the jth oscillator with an oscillator strength
F= 4πNQ/m, damping constant Γ and the resonant frequency ω0.
3.2.1 Extension of Lorentz dispersion model for amor-
phous solids
One of the objectives of this work is a quantitative analysis of IRSE spectra
obtained from thin films/SiOx/Si interfaces. Formation of SiOx/Si interfaces
under uncontrolled conditions (anodic oxidation in electrolyte, oxidation under
atmospheric conditions) is governed by growth of amorphous SiO2. Fig. 3.3
shows SiO2/Si(001) interfacial structure as proposed by Tu and Tersoff [58]
based on Monte–Carlo simulations. A change from a crystalline to amorphous
solid introduces a distribution in bond strength and thus in oscillator frequen-
cies. Such change can be described phenomenologically through the modification
of the Lorentz dispersion model [59, 60] taking into account the distribution of
the oscillator frequency around the resonance frequency ω0,j:
ǫ(ω) = ǫ∞+X
j
(1
√2πσ Z∞
−∞
exp(−(x−ω0,j)2
2σj2)Fj
x2−ω2+iωΓj
dx) (3.17)
where σjis the gaussian half width at half maximum for the jth oscillator dis-
tribution around resonance frequency ω0,j.
CHAPTER 3. OPTICAL MODELING 25
a-SiO2
Interface
region
c-Si(001)
Figure 3.3: SiO2/Si interfacial structure as proposed by Tu and Ter-
soff [58] based on Monte–Carlo simulations. The Si and O atoms are
represented by gold and red spheres, respectively. Each arrow points to
a row of oxygen atoms that form the bridges at the interface.
3.3 Propagation of polarized light in stratified
media
Teitler and Henvis [61] introduced a 4×4 matrix method for studies of reflection,
refraction, and transmission of light by stratified anisotropic media. The general
exposition of the method was introduced by Berreman [62]. A summary of the
method can be found in textbooks [63]. In this section, a brief description of the
method relevant to the simulations performed in this work will be given. In the
following, we assume the plane parallel electromagnetic wave whose variation
with position rand time tis described by exp[i(k·r−wt)] with the wave vector
kand angular frequency w. The Maxwell equation that include the partial
derivative with respect to time then reduce to:
∇×E=−iw
cB(3.18)
and
∇×H=−iw
cD(3.19)
where E,D,H,Bdenote the electromagnetic field vectors. In cartesian coordi-
nates the two Maxwell equations can be combined in the following matrix form:
0 0 0 0 −∂
∂z
∂
∂y
0 0 0 ∂
∂z 0−∂
∂x
0 0 0 −∂
∂y
∂
∂x 0
0∂
∂z −∂
∂y 0 0 0
−∂
∂z 0∂
∂x 0 0 0
∂
∂y −∂
∂x 0 0 0 0
Ex
Ey
Ez
Hx
Hy
Hz
eiwt =iw
c
Dx
Dy
Dz
Bx
By
Bz
eiwt
(3.20)
CHAPTER 3. OPTICAL MODELING 26
Or shortly,
OG =iw
cC(3.21)
where Ois a 6x6 symmetric matrix operator, which can be partitioned into four
3x3 submatrices:
O=0∇
−∇ 0(3.22)
With 0denoting the 3x3 zero matrix and ∇is as shown by the right upper
corner of the 6x6 matrix in Eq. 3.20. In the absence of nonlinear optical effects
and spatial dispersion, the relation between Gand Ccan be described as:
C=MG (3.23)
where the 6x6 matrix Mcarries the information on the anisotropic optical prop-
erties of the medium where the electromagnetic fields propagate. The optical
matrix M can be written in the following form:
M=ǫ ρ
ρ′µ(3.24)
where the sub–matrices ǫand µare the dielectric and magnetic tensors respec-
tively. ρand ρ′denote the optical rotation tensors. Substitution of Eq. 3.23
into Eq. 3.21 yields:
OG =iw
cMG (3.25)
The particular problem of the reflection and transmission of the monochro-
matic plane wave incident from the isotropic ambient medium (z<0) onto an
anisotropic planar structure (z>0) stratified along z–axis is shown schematically
in Fig. 3.4. The symmetry of the problem suggests that there is no variation
along the y direction of any field component so that ∂/∂y=0. If the wave vector
kxis the xcomponent of the incident wave, the variation of the fields in x–
direction goes as e−ikxx. Subsequently ∂/∂x=-ikx. This significantly simplifies
Eq. 3.22 and the problem reduces to the 4x4 matrix form:
∂
∂z
Ex
Hy
Ey
−Hx
=iw
c
∆11 ∆12 ∆13 ∆14
∆21 ∆22 ∆23 ∆24
∆31 ∆32 ∆33 ∆34
∆41 ∆42 ∆43 ∆44
Ex
Hy
Ey
−Hx
(3.26)
Which can be represented in the following form:
∂
∂z Ψ=−iw
c∆Ψ (3.27)
The matrix elements ∆ij in Eq. 3.26 are given for example in a textbook by Az-
zam and Bashara [63]. In the special case when M is constant and independent
of zover some continuous interval of z, Eq. 3.27 can be integrated:
Ψ(z+h) = L(h)Ψ(z) (3.28)
CHAPTER 3. OPTICAL MODELING 27
Erp
Eip
Eis Ers
ambient
z=h1
z=hj
z=d
substrate
Ets
Etp
x
z
y
j0
jd
Figure 3.4: Schematic drawing of stack of homogeneous anisotropic layers
with plane parallel interfaces.
where Lis the partial transfer matrix of the layer:
L=eiw
ch∆= [I−iw
ch∆−1
2!(wh
c)2∆2−1
3!(wh
c)3∆3+...] (3.29)
where Iis the 4x4 identity matrix. In case of a system consisting of multiple
layers the connection of electric and magnetic fields between the two surfaces
at z= 0 and z=dis obtained by a recursive application of the partial transfer
matrix in layer j:
L′=Y
j
L(3.30)
The total transfer matrix L′connect the surfaces at z= 0 and at z=dthrough
the following equation:
Ψ(d) = L′Ψ(0) (3.31)
When measurements are performed on bare substrates or on substrates with
one film on top, like in the case of this work, the equations of Fresnel and Drude
are sufficient to relate the ratio of reflection coefficients rp/rsto the optical
properties (complex indices of refraction and film thickness) of the surface. For
application of Fresnel equation, the following conditions should be met: (1) the
lateral dimensions of the film must be many times its thickness so that the
multiply reflected and transmitted partial waves can be summed to infinity and
(2) the source bandwidth, beam diameter, collimation degree and film thickness
must be all such that the multiply reflected and transmitted waves combine
coherently [63]. In addition, parallel surface planes of the substrates and the
CHAPTER 3. OPTICAL MODELING 28
film are assumed. The above conditions are fulfilled by the samples investigated
in this work, thus the Fresnel theory as applied in the simulations of this work
is shortly presented in the following.
Electromagnetic radiation which strikes a boundary between a medium of a
given refractive index n0and a second medium with refractive index n1under-
goes reflection and refraction at the boundary between the two media (Fig. 3.5).
The fraction of the intensity of incident light that is reflected from the interface
d
j0
j1
j2
n =n + k
1 1 1
i
n0
n =n + k
2 2 2
i
{
....
{
....
tA
rA
r01
r12
Figure 3.5: Schematic drawing of radiation propagation in a system of
ambient (refractive index n0), thin film and a substrate.
is given by the reflectance Rsand Rp, for perpendicularly and parallely polar-
ized radiation with respect to the plane of incidence. The reflectance is given
through:
Rs,p =rs,pr∗
s,p =|rs,p|2(3.32)
with r∗the complex conjugate of the reflection coefficient r, for s- and p- po-
larizations, respectively, where the Fresnel coefficients rsand rpare given by:
rs=nmcos ϕm−nm+1 cos ϕm+1
nmcos ϕm+nm+1 cos ϕm+1
(3.33)
rp=nm+1 cos ϕm−nmcos ϕm+1
nm+1 cos ϕm+nmcos ϕm+1
(3.34)
where nis the complex refractive index, and mand m+1 designate two adjacent
media. Within the layer, refracted radiation is reflected back and forth as
outlined in Fig. 3.5. The amplitude of the wave is reduced by each reflection and
additionally by the absorption on its passage through the layer. The coherent
superimposition of all these contributions upon reflection is described by Airys
equation:
rs,p
A=rs,p
01 +rs,p
12 exp(iδA)
1 + rs,p
01 rs,p
12 exp(iδA)(3.35)
CHAPTER 3. OPTICAL MODELING 29
with the phase δAgiven by:
δA= 4πdωn1cos ϕ1(3.36)
with ωthe optical frequency, d the film thickness and n1is the complex refractive
index of the thin film (Fig. 3.5).
The models discussed above were employed for simulations of the spectra
measured with the infrared spectroscopic ellipsometry. The principles of the
ellipsometry are summarized in Fig. 3.6. Upon reflection from the sample sur-
face, a linearly polarized radiation becomes in general elliptically polarized.
The measured ellipsometric parameter tan ψis defined as the ratio between the
reflection coefficients parallel and perpendicular polarized with respect to the
plane of incidence, while ∆ is the measure of the phase shift between them. In
other words, they can be described through the following relation:
rp
A
rs
A
= tan ψexp(i∆) (3.37)
linearlypolarized
radiation
Figure 3.6: Schematic drawing of the ellipsometric principle. Due to the
different complex reflection coefficients parallel and perpendicular to the
plane of incidence (rpand rs), the light changes its polarization state after
reflection at the sample surface. In general, the linearly polarized light
becomes elliptically polarized and the ellipsometric parameters tan ψand
∆ are defined through: rp
rs= tan ψexp(i∆)
Chapter 4 describes the setup which was used in this work for infrared spec-
troscopic ellipsometry (IRSE) measurements. There, a more detailed introduc-
tion of the ellipsometric parameters will be given. Analysis of limitations and
possibilities of this technique for studies of ultra-thin organic films is presented
in Appendix A.
CHAPTER 3. OPTICAL MODELING 30
3.4 Application of the optical models for simula-
tions of IR ellipsometric spectra: an exam-
ple of hydrogen–passivated Si(111) surface
The aim of this section is to demonstrate the application of the models described
in sections 3.2 and 3.3 for simulations of the measured ellipsometric spectra. As
an example, simulations of the IRSE spectra obtained from hydrogen passivated
Si(111) surfaces are performed. This case was chosen due to the well defined
orientation of hydrogen atoms on Si(111) surfaces. Specific molecular orienta-
tion of the molecules on solid surfaces influences the direction of the transition
dipole moment arising due to the molecular vibrations. This in turn can be
observed by inspection of the characteristic line shapes in the IRSE spectra.
Such characteristic line shapes can be well demonstrated in case of hydrogen
passivated Si(111) surfaces.
Fig. 3.7 shows schematically the transition dipole moments arising as a result
of the Si–H stretching and bending vibrations. The transition dipole moment
due to the δ(Si–H) bending vibrations is oriented parallel to the Si(111) surface.
Thus, it interacts with the s–polarized component of the incident radiation.
Consequently, δ(Si–H) absorption band would appear as a peak-up feature in
IRSE tan ψspectra (with tan ψ=|rp
rs|as defined by Eq. 3.37). The transition
dipole moment due to the ν(Si–H) stretching vibrations is oriented perpendic-
ular to the Si(111) surface, thus interacting with the p–polarized component of
the incident radiation. This gives rise to the absorption band appearing as a
peak–down feature in tan ψellipsometric spectra. Fig. 3.8 shows the measured
IRSE spectra obtained from hydrogen passivated Si(111) surface along with the
simulated results employing the models discussed in sections 3.2 and 3.3. The
spectra were measured using MCT and bolometer detectors in order to enable
investigations in the extended spectral range (see section 4.1.4 for details). The
tan ψspectrum resulted in two well-distinguished peaks: the peak with its max-
imum at 2082 cm−1originated from the stretching Si–H vibrations and the peak
Si
H
d+
d-H
Si
H
d+
d-H
Changeinatransition
dipolemoment
Si
H
d+
d-
Si
H
d+
d-
Changeinatransition
dipolemoment
a. b.
Figure 3.7: Schematic drawing of the transition dipole arising due to
(a) δ(Si–H) bending and (b) ν(Si–H) stretching vibrations.
CHAPTER 3. OPTICAL MODELING 31
d(Si-H) n(Si-H)
Wavenumber(cm)
-1
tany /tany
SiH SiO2
D- ( )D0
SiH SiO2
550 600 650 2050 2100 2150
I0.001
0.05
I
Figure 3.8: Ellipsometric spectra obtained from hydrogen passivated
Si(111) surface, referenced to the spectra of oxidized Si surface. δis
bending and νis stretching vibrations. Black circles: measured data; red
line: simulated spectra. The Lorentz oscillator parameters used in the fit
are described in Table 3.1.
at 626 cm−1was due to the Si-H bending vibrations. This example illustrates
the influence of the orientation of the transition dipole moment on the observed
line shapes in IRSE spectra and the success of the Lorentz oscillator model to
reproduce the measured results (the simulations are shown by a red curve in
Fig. 3.8). Table 3.1 summarizes the Si–H Lorentz oscillator values. It shows
that for the Si–H stretching vibration, the position of the oscillator frequency
ω0was at 2077 cm−1. However, the outcome of the simulations as shown in
red in Fig. 3.8 fits well the measured spectrum with the peak positioned at
2082 cm−1. This is the result of the optical effect for Lorentz oscillators with a
relatively high oscillator strength F. Such optical effect is frequently referred to
as Berreman effect [66].
The results of the simulations shown in Table 3.1 show that for the stretch-
ing vibrations, the xand ycomponents of the oscillator strength F were zero,
while for the bending vibrations there was no z-component (Fz=0). The issue
CHAPTER 3. OPTICAL MODELING 32
Vibrational Frequency ω0F Γ
mode (cm−1) (cm−2) (cm−1)
stretching 2077 Fz= 32000 4
ν(Si–H) Fx=Fy=0
bending 626 Fz= 0 4
δ(Si–H) Fx=Fy=520000
Table 3.1: Lorentz oscillator parameters for stretching and bending Si–H vi-
brational modes simulated for spectrum obtained from the hydrogen passivated
Si(111) surface as shown in Fig. 3.8. In the simulations the value for the high
frequency refractive index n∞= 1.1 was taken from the refs. [64,38]. The Si-H
layer thickness was taken as the length of the Si-H bond of 1.5 ˚
A [65].
of the simulations for determination of the orientation of molecules in thin films
will be addressed again later in this work in the discussion of the spectra ob-
tained from the electrochemically deposited thin organic layers on metallic and
semiconducting surfaces in chapter 5.
Chapter 4
Experimental methods
The chemical structure of the surface organic adsorbates can be readily found us-
ing different types of Fourier Transform Infrared (FTIR)–based spectroscopies.
IR spectroscopy is attractive as a means of studying surface species because
it can be easily adopted to versatile experimental environments as it can be
operated under atmospheric conditions. IR spectroscopy can operate both in
reflection and in transmission modes. While transmission mode requires at
least partially transparent sample for the IR radiation to reach the detector,
reflection modes can be utilized also for studies of highly reflecting metallic and
semiconducting surfaces. Reflectance methods include Attenuated Total Reflec-
tion (ATR), Multiple Internal Reflection (MIR), Diffuse Reflectance Infra-red
Fourier Transform (DRIFT), Infrared Spectroscopic Ellipsometry (IRSE), and
Reflection–Absorption IR Spectroscopy (RAIRS, or equally IRRAS for Infrared
Reflection Absorption Spectroscopy), to name a few [67]. Raman spectroscopy
is usually related to as a complementary technique to infrared absorption spec-
troscopy. The mechanism of spectral formation is however different as Raman
spectroscopy is based on inelastic scattering of photons following their interac-
tion with vibrating molecules of the sample [68, 55]. The molecular vibration
becomes Raman active if there is a modulation of the molecular polarizability.
However, Raman intensities are usually low, although different approaches, such
as resonant Raman [69] and surface enhanced Raman scattering (SERS) [70] can
be combined for enhancement of the Raman signal.
From the above mentioned techniques, this work focuses on implementation
of IRSE, which has an advantage of high surface sensitivity and does not require
a vacuum for its operation. As was outlined in Fig. 3.6, IRSE enables measure-
ments of the ratio (tan ψ) between the complex reflection coefficients parallel
(rp) and perpendicular (rs) to the plane of incidence, and the phase shift (∆)
between them – see section 3.3 for their definition.
As a complementary spectroscopic method for studies of ultra thin organic
films, X-ray Photoelectron Spectroscopy (XPS) [71, 72, 73, 74] was employed.
Application of the IR spectroscopy in conjunction with XPS enabled us to de-
termine the structure of ultra-thin films and organic/inorganic interfaces.
33
CHAPTER 4. EXPERIMENTAL METHODS 34
The following sections present the description of the above techniques as
employed in this work.
4.1 Infrared Spectroscopic Ellipsometry (IRSE)
Infrared spectroscopic ellipsometry is a powerful optical technique for investi-
gations of the optical properties of thin films [75]. Its application ranges from
studies of inorganic materials and films, such as SiO2[76, 77] or SiC [78], to
studies of optical properties of ultra thin organic films [75]. Observation of the
absorption bands due to molecular vibrations and phonons allows identification
of the molecular composition and structure of the investigated material. High
sensitivity of this method to ultra thin layers is especially valuable in the tech-
nologically relevant field of organic functionalization of surfaces for development
of hybrid organic/inorganic devices.
4.1.1 IRSE setup
The definition of the measured ellipsometric parameters was given earlier in sec-
tion 3.3 (see Fig. 3.6). A photograph of the infrared photometric spectrometer
that was employed in this research is shown in Fig. 4.1. The setup consists of
the FTIR spectrometer, polarizer P1, analyzer P2and a detector. Radiation
from the FTIR spectrometer is linearly polarized at the polarizer P1. After
reflection from the sample surface the radiation in general becomes elliptically
polarized. The polarization state of the reflected radiation is analyzed through
the analyzer. Placing a retarder in an optical path of the reflected beam en-
ables determination of the complete set of the Stokes parameters and thus an
access to additional information on the degree of polarization [79]. In general,
IRSE can provide versatile information regarding molecular composition and
structure of the investigated material. However, an access to this information
is not straightforward and depends on the energy detection range of the detec-
tors. Another important factor is the intensity of the radiation source which
becomes more important when poorly reflecting materials are investigated. The
next sections give a brief description of infrared sources and detectors that were
employed in this research.
4.1.2 Measurements of the ellipsometric parameters
In photometric ellipsometry employed in our studies, the ellipsometric param-
eters are determined from intensity measurements at four azimuthal angles of
the polarizer (00, 900, 450, and 1350) at a fixed analyzer position (450):
cos 2ψ=I(00)−I(900)
I(00) + I(900)(4.1)
sin 2ψcos ∆ = I(450)−I(1350)
I(450) + I(1350)(4.2)
CHAPTER 4. EXPERIMENTAL METHODS 35
FTIR
polarizer
analyzer
detector
sample M1
M2
M3
Figure 4.1: IRSE laboratory setup. Radiation from the FTIR spec-
trometer is linearly polarized at the polarizer. After reflection from the
sample surface the radiation in general becomes elliptically polarized.
The polarization state of the reflected radiation is analyzed through
the analyzer. M1, M2 and M3 are the mirrors that are used for radia-
tion direction and focusing. Radiation propagation path from FTIR to
detector is shown schematically by red line.
For isotropic samples with an abrupt interface, derivation of the complex
dielectric function bǫ=ǫ1+iǫ2from the ellipsometric parameters tan ψand ∆
can be performed using the following equation [56]:
ǫ1=n2−k2= sin2ϕ0(1 + tan2ϕ0(cos22ψ−sin22ψsin2∆)
(1 + sin 2ψcos ∆)2) (4.3)
ǫ2= 2nk =−(sin ϕ0tan ϕ0)2sin 4ψsin ∆
(1 + sin 2ψcos ∆)2(4.4)
where nand kare the real and imaginary parts of the complex refractive index,
bn=n+ik and bǫ=bn2. In practice, experimental measurements lead to so–
called pseudo–optical constants <bn >, which represent the optical properties
of a sample as whole, i.e. an effective value which includes the film properties
and the properties of the interfacial region between the isotropic bulk and the
film. For a description of layered anisotropic systems, a 4x4 matrix formalism
is typically used – see section 3.3 and refs. [63,80].
Introduction of the additional element in the optical path of the reflected
beam, a so–called retarder, results in an additional phase shift and improves
the accuracy of determination of ∆ in spectral regions where it is close to 00
and 1800[79]. Furthermore, introduction of a retarder allows determination
of the degree of phase polarization - a parameter that serves as a measure of
the correlation between the reflected waves. In case that the correlation is lost
completely polarization degree is zero, while for homogeneous sample and ideal
interfaces it is unity.
CHAPTER 4. EXPERIMENTAL METHODS 36
A parameter of the total degree of polarization of the radiation serves as an
indicator of the depolarization effects that can take place in the sample [81]:
P=p<cos 2ψ >2+<sin 2ψ >2(<cos ∆ >2+<sin ∆ >2) (4.5)
where the values in the brackets <cos 2ψ >,<sin 2ψ >,<cos ∆ >and
<sin ∆ >stand for the mean quantities measured in the experiment. It is
conventional to define a degree of phase polarization Pph as a sum of the average
terms of <cos ∆ >2and <sin ∆ >2:
Pph =p<cos ∆ >2+<sin ∆ >2(4.6)
Pph might not sum up to unity when depolarization in a sample takes place. De-
polarization in a sample might occur due to scattering, lateral inhomogeneities in
the refractive index of a sample, non-parallel or rough sample boundaries [81,82].
4.1.3 Broadband sources of IR radiation
Globars are the conventional broadband infrared radiation sources for FTIR
spectroscopic application. In Brucker FTIR spectrometers used in this work
globars present a silicon carbide rod of 5 to 10 mm width and 20 to 50 mm length,
which emit electromagnetic radiation when electrically heated in the range of
1500 – 2000 K. In order to achieve a good signal-to-noise (SNR) ratio for ultra-
thin organic films, we typically focused the radiation into a spot of about 6
mm2on the sample plane. Higher lateral resolution in the IR spectral range
could be achieved at the advanced synchrotron radiation source with a superior
SNR as compared to the globar source, when using a similar irradiated spot on
the sample plane. Advantage of the synchrotron source over the conventional
globar lies in its brilliance - it emits more photons per unit area into a unit solid
angle [83]. Fig. 4.2 shows the comparison between the brilliance of black–body
radiation at 1200K and the calculated BESSY II brilliance.
Using the above brilliance advantage of the synchrotron source, it is possible
to study sample inhomogeneities and molecular distribution on the sample sur-
face of organically modified samples. For instance, Fig. 4.3 (a) shows our results
from the mapping of a 430 µm thick inhomogeneous low-density polyethylene
(LDPE) sample performed at the IRIS infrared beamline at BESSY II, with
a 6 mm distance between the scanned spots [82]. There are clear differences
between the spectra obtained at different lateral locations, due to the sample
inhomogenieties [82]. Fig. 4.3 (b) shows an enlarged section of the above spec-
tra, where a fringe pattern is clearly visible when scanned with synchrotron
radiation (spectra shown in black). This pattern arises from back-and-forth re-
flection of radiation from the sample interfaces. However, the interferences are
absent from the spectra obtained with the globar source (shown in Fig. 4.3 (b)
as a smooth grey line), for which the irradiated spot was around 50 mm2. This is
because the phase correlation is completely eliminated when the radiation wave
fronts are transmitted through the inhomogeneous sample (where the thickness
CHAPTER 4. EXPERIMENTAL METHODS 37
Wavelength( m)m
Wavenumber(cm)
-1
1200KBlackbody
Bessy II
10
10
10
10
10
10
10
10
10
10
-2
-3
-4
-5
-6
-7
-8
-9
-10
10 100 1000 10000
2
Brilliance(W/mm/sr[2cm ,1A])
-1
-1 100 101000 1
Figure 4.2: Comparison between globar and BESSY II brilliance. The
brilliance was calculated using the far-field approach taking into account
the intrinsic beam size in the center of the bending magnet as well as the
source size due to projection and diffraction. The brilliance is quoted
per 1 Ampere ring current, resolution of 2 cm−1. Adopted from [83].
900 960 1020 1080 1140
0.4
0.5
0.6
0.7
0.8
1000 1500 2000 2500 3000
0.0
0.2
0.4
0.6
0.8
(b)
Transmittance
Wavenumber (cm-1)
CH3
Rock
combination
modes
CH2
bending
sceletal C-C
stretchig
CH2
Sym. and
antisym.
stretching
(a)
Transmittance
Wavenumber (cm-1)
Figure 4.3: Transmittance spectra obtained for three different LDPE
sample locations using the synchrotron radiation source (black). (b):
enlarged section of (a). Dark gray line: a spectrum taken with the
globar radiation source. Light gray line: a spectrum taken with the
globar source with an additional 1 mm2aperture placed in front of the
LDPE sample. The spectra taken with globar radiation are offset by
0.2 for clarity [82]. Peak assignment was done on the basis of ref. [84].
CHAPTER 4. EXPERIMENTAL METHODS 38
and structural inhomogeneities cause each wave front at different sample posi-
tions to experience different optical paths). The other globar spectrum, shown
in Fig. 4.3 (b), was taken with an additional aperture of about 1 mm2in front of
the sample, limiting the globar radiation to the same area in the sample plane
as when using the synchrotron radiation. However, the signal-to-noise ratio in
this case was insufficient to observe a spectral fringe pattern.
This example illustrates the importance of brilliant synchrotron sources in
investigations of inhomogeneous samples. In this research, it was employed to
study grafted organic thin films in order to access a high–spatial resolution
data [27].
4.1.4 Detectors of IR radiation
Detectors can be characterized by their responsivity, detectivity and noise equiv-
alent power. The responsivity of an infrared detector R(T, f) represents the out-
put signal voltage (or current) in response to input radiation from a black body
at absolute temperature T, with the operating electrical frequency f[85,86]:
R(T, f) = vS
Φs,bb
(4.7)
where vSis the rms signal voltage at the output of a detector, Φs,bb is rms radiant
flux from a black body. Detectivity D∗is defined as the rms signal–to–noise in a
1 Hz bandwidth per unit incident radiant power per square root detector area.
The units of D∗are √Hz/watt. In terms of measurement parameters, D* is
given by [85,86]:
D∗=√AD∆f
Φs,bb
vS
vN
(4.8)
where ADis the detector area in cm2, ∆fis the electrical bandwidth in Hz,
vS/vNis the rms signal–to–noise voltage ratio (the noise measured in the band-
width ∆f) and Φs,bb is rms radiant flux. The Noise Equivalent Power (NEP)
is the rms incident radiant power which gives rise to an rms signal voltage (or
current) equal to the rms noise voltage (or current). The NEP is related to D*
by:
NEP = √AD∆f
D∗(4.9)
The units of NEP are watts.
In this work, thermal and photon detectors for the detection of infrared
power were used. Photon detectors exhibit a selective wavelength dependence
of the response per unit incident radiation power. On the other hand, thermal
detectors are wavelength independent. Idealized representations of these two
types of response are schematically illustrated in Fig. 4.4. Most of the spectra
in this work were obtained using photon MCT (mercury–cadmium–telluride,
HgCdTe) detectors due to their higher detectivity as compared to the thermal
detectors in the mid–IR spectral range. However, the performance of liquid ni-
trogen cooled photon detectors fails below that of the liquid–He cooled thermal
CHAPTER 4. EXPERIMENTAL METHODS 39
Relativeoutputsignal(a.u.)
wavelength
photondetector
thermaldetector
Figure 4.4: Schematic representation of spectral response of photon
and thermal detectors for unit radiant power per unit wavelength in-
terval [85].
detectors in the spectral range above 10 µm (see Fig. 4.5). In this work, an
attempt was performed to extend the performance of the conventional bolome-
ter detector, which is used for far–infrared applications, to the range between
1000 cm−1to 500 cm−1for detection of the molecular vibrations in this spec-
tral range. The bolometer is a thermal infrared detector which employs an
electrical resistance thermometer to measure the temperature of the radiation
absorber. Typical composite bolometer consists of the radiation absorber with
a size appropriate to intercept the signal to be measured, a large absorptivity
over the frequency range of interest, and a low heat capacity. The supporting
substrate has a low heat capacity and large thermal conductivity, so that it re-
mains isothermal during bolometer operation. The thermometer is thermally
attached to the radiation absorber and/or the supporting substrate. It has low
heat capacity, low electrical noise, and an adequate temperature dependence of
its electrical resistance. The thermal link, which connects the thermally active
portions of the bolometer to the heat sink has low heat capacity and an appro-
priate thermal conductance for the required application. The heat sink has a
stable temperature appropriate for the application. The mechanical support for
the thermally active portion of the bolometer has low heat capacity, low thermal
conductance, and must be stiff enough that the mechanical resonant frequen-
cies are higher than the operating frequency of the bolometer. The photograph
of the bolometer employed in this work is shown in Fig. 4.6. In this work, a
composite bolometer type was used. A suitably blackened diamond absorber
thermally bonded to the bolometer. The absorbing layer thickness is generally
selected to minimize fringing effects. A parabolic Winston cone was gold-plated
to prevent tarnish and to improve thermal properties. Infrared cut-on filter con-
sisting of white polyethylene overlay on one face with diamond scatter particles
was used to prevent thermal heating of the bolometer and to allow operation up
to 1000 cm−1. Fig. 4.7 shows intensity recorded at the bolometer when three
CHAPTER 4. EXPERIMENTAL METHODS 41
Winston
cone
Window
Coldmodule
Bolometer
cavity
Filterwheel
Filterwheel
driver
Window
Figure 4.6: Photograph of a bolometer that was used in this work.
CHAPTER 4. EXPERIMENTAL METHODS 42
different outer vacuum windows were used. The main source of the bolometer
400 600 800 1000 1200 1400 1600 1800
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Wavenumber(cm)
-1
Intensity(a.u)
KBr
KRS-5
Polyethylene
Figure 4.7: Intensity detected by a bolometer through different outer
vacuum windows, consisting of wedged polyethylene overcoated on the
inner face with diamond layer, KRS-5 and KBr windows, as marked on
the figure. The spectra were recorded at the ellipsometer after a passage
through two non-crossed wire-grid polarizers and mirror optics.
noise is the thermal fluctuation noise. Other noise sources are the shot noise,
the 1/f noise related to the electric contacts, low frequency noise due to changes
in the temperature of the heat sink and the noise related to the first stages of
amplification [88].
Another thermal detector that was used in this work was a thermal DTGS
(deuterated triglycine sulfate) detector, whose operation is based on the pyro-
electric effect. In this effect, absorption of infrared radiation raises the tem-
perature of the absorber, which changes the surface charge [86, 54]. With the
appropriate external circuit, this change in surface charge leads to a signal volt-
age. The detector works under room temperature conditions. This detector
in principle also works in the desired spectral range between 1000 cm−1and
500 cm−1. The comparison of the bolometer performance in this work to that
of the DTGS detector is shown in Fig. 4.8 (a)). Although the extension of the
bolometer performance toward 500 cm−1– 1000 cm−1can be considered to be
successful (see results in sections 5.4.1 and 6.1) and it had a lower noise level
when compared with DTGS (Fig. 4.8 (a)) there were some disadvantages and
unresolved problems. The main disadvantage was the expensive liquid–He cool-
ing. One of the problems was an interference effect that we believe resulted from
the bolometer detection area (probably in an absorber which was designed for
far-infrared applications), or from the interferences in the Winston cone. These
interferences accompanied all of our measurements (Fig. 4.8 (b)). In order to
get rid of them, it was necessary to reference all of the results to the spectra
CHAPTER 4. EXPERIMENTAL METHODS 43
500 650 800 950 1100 1250
0.970
0.975
0.980
0.985
0.990
0.995
1.000
500 650 800 950 1100 1250
0.240
0.245
0.250
0.255
0.260
0.265
a. b.
Bolometer
DTGS
Bolometer
DTGS
Wavenumber(cm)
-1
tan /tany y
S1 S2
tan y
Figure 4.8: Comparison of performance between two thermal detectors:
DTGS and liquid He-cooled bolometer. a. The ratio between tan ψ
spectrum obtained from HF-treated Si(001) surface and tan ψspectrum
from NH4F–treated Si(001) surface. b. Non-referenced tan ψspectra
from NH4F–treated Si(001) surface. Interference pattern accompanies
bolometer measurements, arising from the bolometer detection area.
Spectra were obtained at 7.5 KHz scanning frequency with 3.5 mm
Jaquinot aperture in all cases.
obtained from bare substrates. This task required a careful alignment of the
whole setup, in order to assure that the radiation exactly retraces its passage
for both samples. Such effect is not surprising since the bolometer detection
area was designed for longer wavelength in far-IR region.
The limit of the thermal detectors performance is set by their main noise
source, which is the ”thermal noise”. This noise sets the ultimate limit to the
minimum signal detected by a perfect thermal detector. From thermodynamical
considerations, the fluctuations in arrival and emission rates of photon on a
thermal detector lead to a root mean square fluctuation in radiant power ∆Φ [85,
86]:
NEP = ∆Φ = q(16AdkBσǫT 5
d∆f) (4.10)
where ǫis the emissivity, σis the Stefan-Boltzmann constant (σ=5.67x10−12
Jcm−2K−4), kBis Boltzmann constant (kB=1.38x10−23JK−1) and Adis the re-
ceiving area. In ideal case, all incident power will be absorbed by a detector with
ǫ=1. At this limit, the minimum detectable power is NEP=∆Φ=5x10−12W, as-
suming T=300K, Ad=1 mm2and ∆f=1 Hz. For operation under cryogenic
conditions, or for operation in outer space with the cosmic background radia-
tion at 3 K, the limiting sensitivity becomes NEP(3K)=5.5x10−16W.
CHAPTER 4. EXPERIMENTAL METHODS 44
4.2 X-ray photoelectron spectroscopy
X-ray photoelectron spectroscopy (XPS) is a surface analysis technique which
allows to study core–level structure of the atoms comprising the sample sur-
face and thus to investigate the surface chemical composition. X-ray excitation
causes ionization of atoms, as schematically shown in Fig. 4.9. The kinetic
energy Ekin of an emitted photoelectron is given by:
Ekin =hν −EB−WF(4.11)
where hν is the photon energy, EBis the binding energy relative to the Fermi
level, and WFis the work function (Fig. 4.9) The typical depth sensitivity of
ValenceBand
ConductionBand
2p
2s
1s
EB
EF
Evac
WF
hn
photoelectron
Figure 4.9: Schematic presentation of a photoemission process. Evac is
a vacuum level, WFis a work function, EFis a Fermi potential and EB
is the binding energy.
XPS as a function of the electron energy is given by inelastic mean free path
(IMFP), as shown in Fig. 4.10. IMFP is the distance that electrons travel in
material before being inelastically scattered. In a range between 50 eV to 100
eV the escape depth is lower than 1 nm, thus information obtained in this
spectral range is very surface–sensitive. XPS analysis allows studies of core-
level and valence structure of surface atoms. An important information on
structure of bonded atoms can be revealed from observation of chemical shifts
of the respective core-levels. For instance, XPS has been extensively used in
order to establish interface structure and suboxide distribution of technologically
important SiO2/Si surfaces – see for example refs. [90,91]. Different oxidation
states of the Si atoms bonded to 0, 1, 2, 3 or 4 oxygen atoms give rise to
the chemical shifts, where the respective photoemission core-level peaks appear
at higher binding energy as the oxidation state of the Si atoms grows. XPS
became a popular and useful technique in investigations of structure of modified
CHAPTER 4. EXPERIMENTAL METHODS 45
100
10
1
0.1
Energy(eV)
110 100 1000
l(nm)
Figure 4.10: The mean free path of the electrons in solid as a function
of their kinetic energy. The doted data represent measurements from
different elements. Adopted from ref. [89].
surfaces with organic monolayers [92,4]. It allows to study chemical structure of
the organic monolayers, to analyze the interface between the substrate and the
layer and to estimate the layer thickness. In addition to the core-level peaks,
satellite peaks due to final state effects can be also observed in the XPS spectra.
Fig. 4.11 shows schematically possible processes upon removal of the electron
from the core level. The shake–up satellite occurs upon relaxation of the valence
electrons in response to the loss of the core electron. The energy required for
this transition is then absent from the kinetic energy of the photoelectron, thus
this two-electron process leads to the discrete structure on the low kinetic energy
side of the spectrum (or, equally, at a higher binding energy). In our work, we
dealt with the spectra from aromatic systems, which show shake-up satellites
with intensities of up to 10% of the primary peak. In aromatic systems the
satellite structure is due to π→π∗, excitations, involving the highest occupied
molecular orbitals and lowest unoccupied molecular orbital (HOMO-LUMO)
transitions [93,92]. These satellites therefore can serve as an indication of system
aromaticity. The shake-off satellites, on the other hand, result in emission of the
valence electron into the unbound continuum state. This process leaves an ion
with vacancies in both the core level and a valence level. In this case, the energy
separation from the main peak is much bigger than for the shake-up satellites,
and the satellites tend to fall within the region of the broad inelastic tail of
the XPS spectrum. Additional possibility is an emission of an Auger electron.
Following the creation of the core hole by X–rays ionization, the atom relaxes by
filling the hole via a transition from an outer level. As a result of the transition,
an excess energy becomes available as a kinetic energy for ejection of an electron
from the atom. The experiments described in this work were performed at
CHAPTER 4. EXPERIMENTAL METHODS 46
Evac
Unoccupied
valencelevels
Occupied
valencelevels
Corelevels
Ground
state
Core
ionization Shake-up Shake-off Auger
decay
Figure 4.11: Processes occurring upon X-ray irradiation.
the BESSY-II synchrotron facility, at the undulator beamline U49/2-PGM2
using the SoLiAS experimental station [94]. Synchrotron radiation source offers
superior resolution compared to conventional instruments, which allowed us to
perform a careful spectral deconvolution. It is also possible to tune the energy
for surface-specific measurements. Spectra were acquired in normal emission
using a Phoibos 150 MCD9 analyzer from SPECS GmbH. The photoelectron
spectra were referenced to the Fermi energy. In the next sections we briefly
discuss the methods for analysis of the XPS spectra.
4.2.1 Deconvolution of XPS spectra
The fitting of the core-levels was performed using Voigt profile, which is a con-
volution of the Gaussian and Lorentzian lineshapes. The Voigt profile takes into
account the experimental resolution, thermal (phonon) broadening and the core-
hole lifetime. Typically, Shirley [95] or Tougaard [96] background subtraction
methods are used to remove the background due to inelastic electron scatter-
ing. In Shirley background subtraction type, the background at any point is
assumed to arise from the scattering of electrons of higher kinetic energy. The
background is thus proportional to the integrated area above the high Ekin side
of the peak. The Tougaard background [96] is based on similar considerations
but takes into account the differential inelastic electron scattering cross-section
and scales it with the number of electrons emitted. In our case, both of the
above methods gave similar results, thus we used either of these backgrounds
for fitting the measured spectra.
4.2.2 Evaluation of the XPS spectra
Solid state photoemission can be described as a three–step process, upon which
the electron is first optically excited by the energy hνof the incident radiation;
then the electron travels through the solid and eventually the electron escapes
CHAPTER 4. EXPERIMENTAL METHODS 47
into the vacuum with the kinetic energy Ekin [97]. The detected core–level
intensity (number of the photoelectrons per second) from an orbital of the con-
stituent atoms is dependent on the atomic density n(number of the atoms per
cm3), the flux fof the X–ray photons impinging on the sample (in photons
cm−2s−1), the photoelectric cross–section σfor the particular transition (in
cm2per atom), the angular efficiency factor φfor the instrumental arrangement
(the angle between photon path and the emitted photoelectron that is detected),
the efficiency photoelectron process y, the area Afrom which the electrons are
detected, the efficiency of detection Tand the mean free path λ[98]:
I=nfσφyAT λ (4.12)
The set–up dependent constants f,φ,A,Tand the efficiency photoelectron
process ywe denote as S(S=fφyAT ).
For a layered system of interest (for example, SiO2/Si or organic/Si), the
intensity from the investigated atomic line is obtained by integrating over the
exponential escape probability:
Il=SnlσlZd
0
exp(−z/λl)dz =nlσlλl[1 −exp(−d/λl)] (4.13)
where Ilis the integrated peak intensity due to the layer-characteristic emission,
σlis the atomic photoionization cross section, λlis the escape depth from the
layer, and nlis the density of atoms in the layer.
The emission from the substrate is given by a similar expression where the
escape probability is multiplied by the attenuation factor exp(−d/λl) of the
overlayer:
Isub =Snsubσsub exp(−d/λl)Z∞
0
exp(−z/λsub)dz =nsubσsubλsub exp(−d/λl)
(4.14)
with σsub,λsub and nsub the atomic photoionization cross section, the escape
depth, and density of atoms in a substrate, respectively. The ratio of the equa-
tions above results in:
Il
Isub
=cl
csub
[exp(d/λl)−1] (4.15)
with cl=nlσlλland csub =nsubσsubλsub. Using Eq. 4.15, one can determine
the thickness of the overlayer.
In this work, we performed also studies of interface island-structure, for
which the modification of the above equations was necessary. The model is
outlined in Fig. 4.12, where a sample consists of a substrate, of island-structured
interfacial layer and of the organic overlayer. For our samples, this structure may
represent a Si substrate, partly oxidize SiOxinterfacial layer and the grafted thin
organic film, as will be discussed later in this work. For many organic materials,
literature cites quite diverse IMFP values or no data at all can be available. For
organic thin films in this work IMFP values from ref. [99] were used. However,
CHAPTER 4. EXPERIMENTAL METHODS 48
Organiclayer
Interfaciallayer
Substrate
q1q2
q2
qi=q1q2
q2
+ + ...
interfacelayercoverage:
q
di
interfacelayerthickness: d
i
i
dl
organiclayerthickness: dl
Figure 4.12: Layer model with island interfacial layer. Coverage and
thicknesses of the interfacial and organic layers are defined in the figure.
if one is interested in the ratio Ii
Isub with Iibeing the integrated intensity from
the interface layer, the contribution from the upper layer will be canceled out
under assumption that the organic layer with the same thickness dlcovers the
substrate and the interfacial layer. Keeping this in mind, we designate the
contribution from the organic layer as αl, which is a constant. The core–level
intensity Iifrom the interfacial layer can be obtained similarly to Eq. 4.13:
Ii=SθiαlniσiZdi
0
exp(−z/λi)dz =θiαlniσiλi[1 −exp(−d/λi)] (4.16)
with the interface layer coverage θiand thickness di. For the substrate, the
core–level intensity can be separated into two parts: one that comes from areas
not covered with the interface layer (Ia
sub) and the other that comes from below
the interface–layer, (Ib
sub). (Ia
sub) can be expressed as:
Ia
sub =S(1 −θi)αlnsubσsub Z∞
0
exp(−z/λsub)dz = (1 −θi)αlnsubσsubλsub
(4.17)
whereas for the (Ib
sub) the escape probability is multiplied by the attenuation
factor coming from the interfacial overlayer:
Ib
sub =Sθiαlnsubσsub exp(−d/λi)Z∞
0
exp(−z/λsub)dz =θiαlcsub exp(−di/λi)
(4.18)
where csub was defined earlier in this section. Combining the equations above,
we get:
Ii
Isub
=Ii
Ia
sub +Ib
sub
=ci
csub
θi(1 −exp(−di
λi))
θi(exp(−di
λi)−1) + 1 (4.19)
with ci=niσiλi. Eq. 4.19 can be solved for coverage θiand thickness diwhere
the value of Ii
Isub is estimated from two measurements at two different photon
energies.
CHAPTER 4. EXPERIMENTAL METHODS 49
Using the evaluated interface coverage θiand thickness dias an input, the
evaluation of the thickness of the organic overlayer was performed in this work
with the dedicated XPS Multiquant program [100]. Additional input data which
is required by the program is the integrated intensities of the measured XPS
lines. The program allows several geometry models where we used the one as
outlined in Fig. 4.12.
Chapter 5
Optical properties of
organic thin films
In this chapter the studies of electrochemically grafted thin organic layers on
TiO2, Si(111), Au and porous silicon surfaces are described. As was shown in
section 2.2.1, grafting of the thin films from the aryl diazonium salts starts from
the split of the benzenediazonium into the benzene radical carrying the specific
functional group and the N2. The released benzene radicals are then grafted
into the electrode. IR spectra of the aryl diazonium salts, which is presented in
this chapter, provided information on the structure of this initial compound. In
addition, these spectra exhibit the absorption bands that one should expect in
the spectra from the grafted ultra thin organic films.
After the grafting, identification of the molecular adsorbates and the orien-
tation of the molecules in thin films was performed based on the observation
of the respective absorption peaks in the IRSE spectra and their simulations
in the optical models. This chapter shows that no preferential orientation was
observed for the organic molecules grafted from the benzenediazonium salts on
Si(111), Au, TiO2and porous silicon substrates. However, IRSE spectra ob-
tained from methylated porous silicon revealed that methyl (CH3) groups orient
with their molecular axis perpendicular to the walls of the porous silicon.
Application of the optical models described in chapter 3 allowed to determine
the Lorentz oscillator parameters for nitrobenzene thin films. For this purpose,
analysis of several test samples with different nitrobenzene layer thicknesses
was performed. Since the thickness and the optical parameters of materials
are interdependent, cross–referenced studies with XPS and VIS–ellipsometry
were carried out for the determination of the thickness and high–frequency re-
fractive index. A good agreement was obtained between the measured spectra
from the nitrobenzene–modified Au and Si(111) substrates and the simulated
results which used the Lorentz oscillator parameters as determined from the test
samples with different nitrobenzene thicknesses.
The temperature–induced desorption of nitrobenzene molecules from Au sur-
50
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 51
face is investigated using the IRSE technique. Observation of the diminishing
of the NO2-related absorption peaks amplitudes established that the desorption
occurs around 1700C.
The oxidation of the interface between silicon and organic thin films was also
observed. This will be discussed in chapter 6.
This chapter is structured as follows: first, we show the IR spectra of the
aryl diazonium compounds prior to the electrochemical grafting. Then, a de-
tailed studies of the Au, Si(111) and TiO2surfaces modified with nitrobenzene
molecules are presented. This is followed by the discussion of the methoxyben-
zene properties grafted on various substrates. Finally, studies on methylated
porous silicon are introduced. A discussion of the molecular attachment on
silicon surfaces through the Si–C or Si–O–C bonds concludes this chapter.
5.1 IR properties of tetrafluorborate aryldiazo-
nium compounds
IR spectroscopy of the 4–methoxybenzenediazonium tetrafluorborate (4–MeBDT)
and 4–nitrobenzenediazonium tetrafluorborate(4–NBDT) salts was performed
for investigations of the absorption band structure of the compounds from which
the molecules were grafted on various surfaces. In addition, these measurements
allow to notice absorption peaks due to possible contaminations that may be
present on the surface, for example, resulting from the BF−
4anion or N2(dia-
zonium) cation.
Fig. 5.1 shows transmittance spectra obtained from the 4–NBDT and 4–
MeBDT pressed into a KBr tablet. Mixing a KBr powder with the solid of
interest and preparing a tablet under high pressures is a well-known technique
for transmission measurements in IR spectral range. KBr is transparent between
4000 cm−1to 400 cm−1, thus the vibrational absorption of the material of
interest is not effected by its presence. However, care should be taken when
using this technique due to a possibility of ion exchange between the KBr and
the material which is to be analyzed. Spectra obtained from 4–NBDT and
4–MeBDT (Fig. 5.1) showed a broad peak between 900 and 1200cm−1, related
to the BF−
4vibrations [106, 107]. Characteristic bands that are common to
4–NBDT and nitrobenzene and to 4–MeBDT and methoxybenzene molecules,
respectively, are listed in table 5.1. However, the presence of the diazonium
group shifts strongly the frequencies that would be expected for nitrobenzene
(NB) and methoxybenzene (MeB). Thus the assignment in Table 5.1 was given
only for absorption bands for which the direct comparison to the NB and MeB
literature–listed frequency values was possible. Other characteristic absorption
peaks were due to diazonium ν(N≡N) and ν(Aryl-N)diazo vibrations around
2200-2300 cm−1(marked with * in Fig. 5.1) and 1300-1350 cm−1(marked with
** in Fig. 5.1), respectively. Table 5.1 lists the measured absorption frequencies
which are in agreement with those found in literature [103,104,105].
It is interesting to note the relative shift observed in frequencies between
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 52
NO
Wavenumber(cm)
-1
Transmittance
2
N2
+BF
4-
OCH3
N2
+BF
4-
1 32 45 76 8
1 32 45 76 8
600 900 1200 1500 1800 2100 2400 2700
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
1.0
**
*
*
**
Figure 5.1: Transmittance spectra of 4–NBDT (gray) and 4–MeBDT
(black) in KBr tablet. The diazonium–related absorption bands are
marked by stars and are discussed in the text. The prominent bands
1–8 for each of the above compounds are listed in Table 5.1.
Band No. 4-NBDT Assignment 4-MeBDT Assignment
Fig. 5.1 cm−1[101] cm−1[102]
1 664 - 684 -
2 742 - 842 -
3 866 - 1292 (1292) δ(C–H)
4 1316** ν(Aryl-N)diazo 1346** ν(Aryl-N)diazo
5 1360 (1347) νss(NO2) 1442 (1442) δs(CH3)
6 1545 (1523) νas(NO2) 1494 (1497) ν(C-C)
7 1618 - 1585 (1588) ν(C-C)
8 2306* ν(N≡N) 2254* ν(N≡N)
Table 5.1: The observed frequencies for 4-NBDT and 4-MeBDT in KBr pellet
and the assignments of the major bands as labeled in Fig. 5.1. No assignment
was performed in the spectral range between 900 and 1300 cm−1due to the
overlap with the strong bands associated with the BF−
4absorption. Only the
bands for which there is a clear comparison with nitrobenzene and metoxy-
benzene spectra in refs. [101] and [102] were assigned. The frequencies in the
parentheses refer to the values determined from the literature. The starred fre-
quencies refer to the diazonium–related absorption bands which were assigned
in accordance to refs. [103,104,105]
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 53
the ν(N≡N) and ν(Aryl-N)diazo in OCH3and NO2p-substituted aryldiazo-
nium compounds. Accordingly to refs. [108,104, 105], bond order (the number
of bonds between a pair of atoms) changes upon attachment of the organic
group at the para-position of the benzenediazonium, as is schematically shown
in Fig. 5.2. In general, it was found that electron donor substituents reduce the
diazonium frequency, whereas acceptor substituents increase it (relatively to
X=H in Fig. 5.2) [108,104]. For electron acceptor sustituents (i. e. X=OCH3),
structure II in Fig. 5.2 becomes more dominant. There is a decrease of the N≡N
bond order which shifts the absorption peak frequency due to the diazonium to
lower wavenumbers [104, 105, 109]. In the following sections, investigations of
N N
+
X
N N
+
X
-
+
( )I
( )II
Figure 5.2: Structure of aryldiazonium upon substitution of different X
groups [108, 104]. (I):Ground state; (II):Increasing the contribution of
polarized structure upon substitution of electron-donating group.
electrochemically grafted organic films on various substrates will be discussed.
However, no contaminations from the diazonium or BF−
4anions from the ben-
zenediazonium salts were detected with infrared ellipsometry. The absence of
the BF−
4contaminations was also confirmed by XPS, where no indicative fluo-
rine F1s emission was found [4].
However, contaminations due to the wet-chemical etching could be some-
times observed on the organically modified surfaces. The wet-chemical etch-
ing of chemically prepared oxides on Si(111) surfaces is the final step to pre-
pare flat and well H-passivated Si(111) surfaces with terrace-like structures
and mono-atomic steps [39, 50, 110]. The final etching step in 40% NH4F pro-
duces atomically flat surfaces [39]. However, sometimes the residual NH+
4ions
were still present on the surfaces. Accordingly to the work of Yota and Bur-
rows [111], these contaminations are the traces of the ammonium silicon fluoride
((NH4)2SiF6) which is formed during the etching in the buffereded hydrofluoric
acid [111]. Since such contaminants are soluble in water, rinsing of the samples
in general removes their traces from the surface. Nonetheless, some of our IR
spectra contained the absoprtion bands due to the NH+
4contaminations, as can
be seen in (Fig. 5.3). During our IRSE spectral interpretation extra care was
thus taken in the discussion of the absorption bands in this spectral range.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 54
n(Si-H)
1950 2100 2250
-181.1
-181.0
-180.9
-180.8
-180.7
0.240
0.242
0.244
0.246
0.248
1200 1350 1500
-180
-179
-178
-177
-176
0.204
0.210
0.216
0.222
0.228
Wavenumber(cm)
-1
tan y
D()
0
d(NH )
4
+
Figure 5.3: IRSE tan ψ(black) and ∆ (gray) spectra obtained from
n-type Si(111) surface, prepared by a standard procedure [112]. On the
left-side panel, a typical absorption peak due to NH+
4contaminations is
observed, while on the right-hand side, a Si-H absorption feature reveals
that the H-termination was successful.
5.2 Nitrobenzene on Au, Si(111) and TiO2sur-
faces
In this section, the characterization work on nitrobenzene molecules electro-
chemically grafted on Si(111), Au and TiO2surfaces is introduced. The simula-
tions in optical models for interpretation of IRSE results and for determination
of the Lorentz oscillator parameters are discussed. The experiments on the sta-
bility of the nitrobenzene films to high temperature conditions are presented,
where the temperature–induced desorption was monitored by IRSE. Eventually,
we present our studies on the chemical composition of the electrochemically
grafted NB films using XPS technique.
5.2.1 IRSE characterization of nitrobenzene films
For all surfaces which were modified with nitrobenzene thin films, IRSE spectra
exhibited absorption bands due to symmetric and anti–symmetric stretching
vibrations of the NO2group, as can be seen in Fig. 5.4 (a). Fig. 5.4 (b) shows
the comparative plot between the spectra obtained from different surfaces in
the extended spectral range. This plot shows the relative amplitudes of the
observed vibrational bands on TiO2, Au and Si(111) surfaces. Fig. 5.4 (c)
shows SEM images of the TiO2surface before and after the modification with
the nitrobenzene layers and it will be discussed later in this section.
The assignment of the absorption bands that were observed in Fig. 5.4 was
based on the data provided in refs. [101,113,114] and is summarized in Table 5.2.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 55
Liquid NB NB on TiO2NB on p-Si(111) NB on Au Assignment
[101,114] this work this work this work [101,113,114]
1021 1014 - - δ(C-H)
1108 1110 - - ν(C–N)+ring
breathing
1174 1180 - - C-H bend
1347 1351 1353 1354 νss(NO2)
1523 1525 1531 1527 νas(NO2)
1606 1598 1600 1598 ν(C–C)ring
Table 5.2: Observed frequencies for nitrobezene in (cm−1) and assignments.
The NB frequencies on TiO2, Au and p-Si(111) were obtained in this work as
shown in the spectra in Fig. 5.4 (a,b). νss symmetric stretching vibration; νas
asymmetric stretching vibration; δbending vibration.
For a detailed spectral analysis and conclusions on molecular orientation,
simulations in optical models are necessary. Fig. 5.5 shows the calculated
line shapes for different possible orientations of the nitrobenzene on Si and Au
surfaces.
For metallic surfaces, the line shapes follow the surface selection rules [115].
These rules allow only absorption of incident IR radiation by vibrational modes
whose transition dipole moment components are perpendicular to the surface
plane. Fig. 5.6 shows the orientation of the transition dipole moment for the
symmetric (νss(NO2)) and antisymmetric (νas(NO2)) stretching vibrations of
the NO2group of nitrobenzene. Orientation of the molecules with the molecular
plane parallel to the Au surface would result in absence of the absorption bands
due to both the symmetric and antisymmetric vibrations of the NO2group (Fig.
5.5). If the molecular plane would be perpendicular to the Au surface (molecules
would be upright–standing), the νss(NO2) mode would give a transition dipole
moment perpendicular to the Au surface while the νas(NO2) would result in
a transition dipole moment oriented parallel to the Au surface. In this case,
only absorption bands due to νss(NO2) should appear in our spectra. However,
the results of the measurements presented in Fig. 5.4 (a) show a situation with
the line shapes corresponding to a random orientation as shown through the
simulations of the spectra in Fig. 5.5.
Another important result coming out of calculations presented in Fig. 5.5(b)
is the relative band amplitude of the absorption bands of the molecules grafted
on Au and Si substrates. For upright–standing molecules, Fig. 5.5 shows that
the absorption band due to the symmetric NO2vibrations on Au is twice the
amplitude of this band on Si substrate. This is due to the image–charge effect
of the transition dipole moment on metallic surfaces. It is also interesting to
compare the above-discussed spectra from NB/Si and NB/Au surfaces to the
NB/TiO2case. From a comparative plot shown in Fig. 5.4 (b), it is evident that
nitrobenzene films on TiO2surface have stronger absorption features, resulting
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 56
a. b.
1050 1200 1350 1500 1650 1800
I0.005
tan tan
y y
F S
/
c.
TiO /Ti
2
NB/TiO /Ti
2
( )
0
D D
-
F S tan tan
y y
F S
/
tan tan
y y
F S
/( )
0
D D
-
F S ( )
0
D D
-
F S
NB/Si(111)
NB/Au
NB/TiO2
tan tany y
F S
/
n(C-C)
ring
Wavenumber(cm)
-1
Wavenumber(cm)
-1
n(NO )
2
as
n(NO )
2
ss
1200 1300 1400 1500 1600 1700 1800
I0.0005
I0.05
I0.0005
I0.05
I0.01
I0.5
Figure 5.4: a. Referenced IRSE spectra of tanψFof the nitrobenzene
film to tanψSof various bare substrates in the range between 1200 cm−1
and 1800 cm−1. b. Referenced tan ψspectra plotted on the same scale
in the extended spectral range. tan ψspectrum for NB on Si (111) was
shifted by 0.002 for clarity. Black: NB on p-Si(111); red: NB on Au;
blue: NB on TiO2. c. SEM micrographs of a non-modified TiO2on
Ti-coated glass substrate (upper SEM panel) and surface modified with
NB molecules (lower SEM panel). The white lines serve as guide to the
eye and indicate the NB layer. The SEM images were taken at a tilt
angle of 300.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 57
n(NO)
ss
2
D D
F S
-tan /tany y
FS
1350 1500 1650 1350 1500 1650 1350 1500 1650
Wavenumber(cm)
-1
D D
F S
-
NB/Si
NB/Au
(a.u)
1300 1400 1500 1600
NB/Si
NB/Au
1e-3
tan /tany y
FS
Wavenumber(cm)
-1
n(NO)
ss
2
n(C-C)
ring
n(NO)
ss
2
n(NO)
ss
2
n(C-C)
ring
n(NO)
ss
2
n(NO)
ss
2
n(C-C)
ring
(a.u)
(a.u)
(a.u)
a.
b.
tan y
Figure 5.5: a. Calculated ellipsometric line shapes for Au (red) and
Si(black) NB–modified surfaces, referenced to bare substrates (arbi-
trary scaled). Molecular orientation of NB molecules is shown schemat-
ically on the insets, where the left column refers to upright–standing
molecules, central to the molecular plane parallel to the surface and the
right column shows the molecules randomly oriented on a substrate. b.
tan ψspectrum comparing the amplitudes for a 10 nm calculated film
with high-frequency refractive index of 1.46. The data from measured
substrates was employed in all calculations.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 58
n(NO)
as 2n(NO)
ss 2
Figure 5.6: Molecular transition dipole moment (symbolized by red ar-
rows) for symmetric (νss(NO2)) and antisymmetric (νas(NO2)) stretch-
ing vibrations of the NO2group of nitrobenzene.
from a higher thickness of the layer. This finding is supported by SEM images of
organically modified TiO2surface (Fig. 5.4 (c)), where a thickness of about 70
-100 nm of NB is observed. VIS– ellipsometry indicated the thickness of 100 nm
and a high-frequency effective refractive index of 1.3. The relatively low value of
the effective refractive index is a result of a less dense layer. Here, high thickness
of the films suggests occurrence of polymerization upon electrochemical grafting.
In general, polymerization can occur through the radical position at the aryl
ring, as shown in Fig. 5.7 (a). The other possibility proposed by Laforgue
et al [116] is the involvement of diazonium group in a polymerization or even
in an attachment to a surface (Fig. 5.7 (b)). However, we did not detect any
absorption due to a diazonium group in our IRSE spectra for none of the studied
molecules. Thus the most likely scenario is the polymerization through the
process shown in Fig. 5.7 (a).
R
R
R
R
R
N
N
R
R
N
N
a. b.
Figure 5.7: Polymerization of benzene derivatives upon electrochemical
grafting after [116]. a. Polymerization through radical end at the aryl
ring; b. Possible involvement of diazonium in formation of electrochem-
ically grafted thin films
5.2.2 Determination of optical constants
The aim of this section is the determination of the Lorentz oscillator parameters
for nitrobenzene thin films in the infrared spectral range. This was done through
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 59
cross–reference of the IRSE data from NB films with several thicknesses on Si
and Au substrates. We verified our calculations using complementary methods,
such as Vis–ellipsometry and XPS.
In order to calculate NB optical properties in the IR spectral range, films
with several thicknesses were prepared. As was already mentioned earlier (see
section 2.2.1) grafting of nitrobenzene films from aryl diazonium compounds
on H-passivated Si(111) surface is a self-limiting process. Thus for preparation
of a thick nitrobenzene film, the electrode potential was driven to -4 V. This
enabled an electron transfer through the already formed NB blocking layer by
means of tunneling. We denote this sample ”S1” for further discussion. The
other films were taken out of the electrolyte after grafting within 110 sec and
550 sec, respectively. We denote the sample prepared withing 550 sec of grafting
”S2” and the 110 sec ”S3”.
Samples S1 and S3 were measured by VIS–ellipsometry, resulting in high-
frequency refractive index of 1.46. The thickness of S1 was 125 nm while
that of S2 was 14 nm. Sample S3 was measured with soft X–rays using X-
rays standing waves which monitors a characteristic atomic X-ray fluorescence
signal [117] at the plane-grating monochromator (PGM) beamline for undula-
tor radiation [118,119] at the Physikalisch-Technische Bundesanstalt (PTB) at
BESSY II. Evaluation of carbon fluorescence emission resulted in 3 nm thickness
of the NB film in the S3 sample [120].
Using the above data, IRSE spectra from the three samples were fitted using
Lorentz–oscillator formalism in isotropic layer model. The results are shown in
Fig. 5.8. Oscillator parameters are summarized in Table 5.3.
Assignment Frequency ω0(cm−1) F (cm−2) Γ (cm−1)
sym. NO2stretch 1346 21207 19
asym. NO2stretch 1524 22748 23
Ring stretch 1598 5404 23
Table 5.3: Parameters of NB on Si(111) surface as fitted using Lorentz-Oscillator
model. The simulated results are shown in Fig. 5.8
We observed a good agreement between the Lorentz oscillator constants
of the NO2group–related absorption bands from 2–[4–(N–Dodecanoyl amino)
phenyl]–5–(4–nitrophenyl)–1,3,4–oxadiazole Langmuir–Blodgett (LB) films on
gold as obtained in Ref. [121] and the Lorentz oscillator constants for the NB
films in this work as shown in Table 5.3. The approach that was used in ref. [121]
for the determination of NB optical constants through the scaling of the opti-
cal constants from the LB films with the known thickness and a determined
high–frequency dielectric constants has some disadvantages. Namely, the chem-
ical environment of the nitrobenzene unit in the above LB film may have an
influence on the absorption band shapes in IRSE spectra, resulting in overesti-
mated or underestimated Lorentz oscillator parameters. The current approach
as described above bases uniquely on studies of NB films, where we combined
the data from different experimental methods and cross–referenced the results
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 60
1300 1400 1500 1600
Wavenumber (cm-1)
0.001
0.001
0.005
1300 1400 1500 1600
Wavenumber (cm-1)
0.2
0.2
1
I
I
I
I
I
I
d=115nm
d=14nm
d=3nm
a.
b.
c.
D()
0
tan y
n(NO)
ss
2
n(NO)
as
2
n(C-C)
ring
n(NO)
ss
2
n(NO)
as
2
n(C-C)
ring
Figure 5.8: Measured (black) and calculated (red) spectra for three dif-
ferent nitrobenzene films on Si(111) substrates with d=115 nm(a); d=14
nm (b); and d=3 nm (c) thicknesses. Left: ellipsometric parameter tan
ψ, right: the corresponding ∆ spectra. Calculations were performed us-
ing high–frequency refractive index n∞=1.46 and thickness as denoted
in the plots. Lorentz oscillator parameters are given in Table 5.3
from NB films with different thicknesses. Using the oscillator parameters listed
in Table 5.3, a good agreement was achieved between the measured and the
calculated lineshapes for NB on metallic Au surface, as shown in Fig. 5.9. The
results obtained from the simulations of NB films on Au and Si substrate point
out that the molecules have an isotropic distribution on these substrates. This
supports our earlier driven conclusion from comparison of the lineshapes as ob-
tained from TiO2, Si(111) and Au surfaces that the nitrobenzene does not have
any preferential orientation in the grafted films.
5.2.3 Thickness determination and studies of the chemical
composition of nitrobenzene films using combined
XPS and IRSE analysis
To investigate the chemical composition of the nitrobenzene films on Si(111)
and to cross-reference their thickness values, X-ray photoelectron spectroscopy
(XPS) was employed as a complementary method to IRSE. The deconvolution
of the core level spectra will be performed and discussed in detail in chapter 7.
Here, we shortly present the relevant core–level information for determination
of composition and thickness of the grafted NB films on Si(111) surface.
Fig. 5.10 (a) shows the Si2p, N1s, O1s and C1s core level spectra obtained
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 61
tan y
tan y
D( )
0D( )
0
Wavenumber(cm)
-1
1300 1400 1500 16001300 1400 1500 1600
1300 1400 1500 1600 1300 1400 1500 1600
n(NO)
ss
2
n(NO)
as
2
n(C-C)
ring
n(NO)
ss
2
n(NO)
as
2
n(C-C)
ring
d=5nm
d=2.3nm
I0.0005
I
0.1
I
0.1
I0.0005
Figure 5.9: Measured (black) and calculated (red) spectra for two dif-
ferent nitrobenzene films on Au substrates. Left: tan ψ, right: the
corresponding ∆ spectra. The calculation was performed using high-
frequency refractive index n∞=1.46 and thickness as denoted in the
plots. Lorentz oscillator parameters are given in Table 5.3
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 62
at 650 eV incident energy. The ratio of the integrated intensities should be
indicative of the relative atomic composition of the film. Integrating the core–
level spectra from Fig. 5.10, we obtain the C1s integrated intensity of 5310,
the N1s integrated intensity of 1280 and the O1s of 3870. The nitroben-
zene chemical formula C6H5NO2presumes the relative atomic composition of
6(C):1(N):2(O), while the above integrated intensities suggest the film compo-
sition of 4(C):1(N):3(O). However, it is readily observable that the N1s spec-
trum consists of two different peaks, positioned at 406 eV and 400 eV. While
the peak at 406 eV results from the nitrogen atom bonded to oxygen in NO2
groups, the peak at 400 eV results from nitrogen bonded to hydrogen in amino
(NH2) groups. Such amino-type contribution result mainly from the amino com-
pounds that serve as the precursors for the aryl diazonium salts from which the
nitrobenzene was electrochemically grafted onto the surface [122]. Ignoring this
peak and integrating only the signal from lower binding energies, the integrated
intensity of N1s was 875. This is in good agreement with the expected C:N (6:1)
ratio.
For O1s core-level spectrum, the situation is much more complicated. It will
be shown in chapter 7 that the O1s signal includes contributions from interfacial
oxygen in SiOxcentered at 531.4 eV. Ignoring this peak, the integration over
the rest of the O1s peak results however in integrated intensity of 3400, thus
the ratio of the corrected integrated intensities for C:N:O is 6:1:4. This relation
still does not correspond to the molecular formula of nitrobenzene. However,
according to Refs. [123,124,125], also the water–related O1s signal is centered at
533 eV, similarly to the signal from the oxygen in NO2groups. This explains the
high oxygen ratio relatively to the nitrogen and the carbon–related contributions
as obtained from integrated intensities of the respective core–level peaks. Due
to the spectral overlap, it was not possible to distinguish between the O1s signal
coming from the remnant water in the electrolyte and from the nitrobenzene
film.
Comparison between the XPS and IRSE spectra presented in Fig. 5.10 re-
veals a SiOx–related peak in the Si2p core level of the XPS spectra, while no
band in the IRSE spectra corresponding to the SiOxabsorption in the range
between 1000 to 1200 cm−1was observed. The reason might be that (a) the two
different pieces of the sample that were analyzed by XPS and IRSE were differ-
ently oxidized; (b) the above discussed water residuals from the electrochemical
grafting were triggered by X–ray irradiation to oxidize the silicon interface; (c)
oxygen atoms which are released upon the reduction of the nitrobenzene to ani-
line upon X–ray irradiation contributed to oxidation of the silicon surface (see
chapter 7 for the detail discussion of this reduction process) and (d) different
sensitivity of the IRSE and XPS to the amount of the surface silicon oxides. The
detailed discussion of the amount of the SiOxas calculated from the combined
XPS and IRSE measurements will be presented in section 6.3.
The measurements shown in Fig. 5.10 were used for the determination of
the nitrobenzene thickness. Simulations of the IRSE spectra are shown in red
in Fig. 5.10(b), in which the Lorentz oscillator parameters determined in the
previous section were used (see Table 5.3). The simulation resulted in 5 nm
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 63
399 402 405 408 411
0
100
200
300
400
500
600
528 530 532 534 536 538 540
0
250
500
750
1000
1250
1500
1750
2000
98 100 102 104 106
0
100
200
300
400
500
600
282 285 288 291 294 297
0
350
700
1050
1400
1750
2100
2450
2800
Wavenumber(cm )
-1
n(SiO )xnss
(NO )2
nas
(NO )2
Si2p
h =650eVn
C1s
h =650eVn
O1s
h =650eVn
N1s
h =650eVn
Bindingenergy(eV)
Intensity(a.u)
tan y / tan y
FS
D D ( )-
FS0
I0.002
I0.1
SiOx
a.
b.
1000 1100 1200 1300 1400 1500 1600
-NO2
-NH2
Figure 5.10: a. Si2p, C1s, N1s and O1s core level spectra obtained
from the nitrobenzene–terminated Si(111) sample, at hν=650 eV pho-
ton energy. Deconvolution of all of the above core level spectra is be
presented in detail in chapter 7. b. IRSE spectra obtained from the
other part of the same sample, where the measured spectra are shown
in black and the simulated in red. Simulations were performed using
n∞=1.46, thickness d=5 nm and Lorentz oscillator parameters as shown
in Table 5.3.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 64
nitrobenzene layer thickness. For estimation of nitrobenzene film thickness us-
ing the XPS spectra, Eq. 4.14 from section 4.2.2 was used. The Si2p integrated
intensity from the nitrobenzene–coated Si(111) surface was 445 at 650 eV, while
for bare Si(111), the integrated intensity was 9400. The calculations were per-
formed using Eq. 4.14, which expresses the attenuation of the Si2p core level
signal by the nitrobenzene layer. The data for inelastic mean free path is un-
fortunately currently absent from databases. Data for several organic materials
was published by Tanuma, Powell and Penn in ref. [99]. The data diverges from
1.86 nm for Guanine to 2.4 nm for polymethylmethacrylate (PMMA) at 650 eV.
We take the λlvalue to be 2.27 nm, typical for polyethylene and polysterene.
Thus, the thickness of the nitrobenzene is estimated to be around 7 nm. The
simulations of IRSE data from the same sample results in 5 nm thickness. The
discrepancy is the result of (a) ignoring the interface SiOxlayer and (b) the un-
certainty in the IMFP values. However, calculations based on the XMQ analysis
(see section 4.2.2) take into account the interfacial SiOxlayer. Such calculations
result in the NB thickness of 5 nm, in a good agreement with IRSE results. The
detailed discussion of the determination of the nitrobenzene layer thickness in a
model which takes into account the SiOxinterfacial layer thickness and coverage
will be presented in chapter 6.
5.2.4 IRSE studies of temperature–induced desorption
The temperature stability studies of nitrobenzene organic layers on Si(111) sur-
faces were reported in [126], where temperature–induced desorption was mon-
itored with mass spectrometer. The desorption was reported to start above
1900C, with the desorption of NO and of C2HNO2above 2000C.
In this work we performed IRSE temperature–dependent measurements of
NB on Au surfaces under UHV conditions. Fig. 5.11 shows the results of the
measurements, where plot (a) shows high–resolution spectra (1 cm−1), plot (b)
shows evolution of spectra obtained with resolution of 4 cm−1and plot (c) shows
the integrated intensities of absorption peaks due to NO2symmetric stretching
vibration (1), NO2asymmetric stretching vibration (2) and of the ring C–C
stretching vibration (3) as a function of the temperature.
The cryogenic temperatures were employed in order to find out whether the
cooling can influence the ordering of the molecules on the surface. However,
as can be seen in high–resolution spectra in Fig. 5.11 (a), no differences were
found between the absorption peaks in the spectra of NB/Au at -1600C and
the absorption peaks in the spectra obtained after the heating of the sample
up to 600C. The desorption starts around 1700C, with the diminishing of the
amplitudes of the ν(NO2) vibrations. However, the absorption peak due to
ν(C-C)ring stays nearly constant through the treatment. Above 2500C, only
the intensity due to the symmetric ν(NO2) vibration is still possible to evaluate,
while the other absorption peaks diminish in amplitude and overlap with the
atmospheric water absorption bands. The diminishing of the amplitudes of the
absorption bands due to ν(NO2) vibrations is in agreement with the desorption
of the NO units as reported in ref. [126]. However, the ν(C-C)ring has almost
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 65
-150 -50 50 150 250
0.01
0.02
0.03
0.04
0.05
0.06
Wavenumber(cm)
-1
[ C]
0
0.003
I
0.002
[ C]
0
Temperature( C)
0
IRPeak
intensity(a.u)
(1)
(2)
(3)
b.
c.
a.
I
tan /tany y
FS
Wavenumber(cm)
-1
tan y
0.001
I
I0.2
D()
0
iii.
iv.
ii.
i.
iii.
iv.
ii.
i.
(1) (2) (3) (1) (2) (3)
(1) (2) (3)
d.
n(NO)
ss 2n(NO)
as 2n(C-C)
ring
(1): (2): (3):
75 0
70 0
1300 1400 1500 1600 1700 RT
-150
140
280
240
200
170
130
90
50
-150
1300 1400 1500 1600
-160
280
160
140
120
100
60
-160
1300 1400 1500 1600 1700
Figure 5.11: Temperature–dependent IRSE spectra obtained from
NB/Au surface. a. Spectra obtained with the resolution of 1 cm−1; b.
Spectra with resolution of 4 cm−1; c. Integrated intensity of νss(NO2)
(1), νas(NO2) (2) and ν(C-C)ring (3) absorption peaks. d. Zoomed–in
IRSE spectra before the temperature–dependent desorption experiment
(ii,iv) and after it (i,iii), obtained at incident angles of 750(black) and
700(red). The arrows emphasize the shift of the ν(C–C)ring absorption
peak to higher frequencies upon sample heating.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 66
constant amplitude throughout the experiment.
Fig. 5.11 (d) shows IRSE spectra obtained from the surface directly after
electrochemical preparation (ii,iv) and after temperature–dependent desorption
(i,iii), at 750and 700angle of incidence. In this figure, a shift of about 20 cm−1
from 1600 cm−1toward higher frequencies is observed for ν(C–C)ring after ther-
mal treatment of the sample. Also the absorption bands due to the residual NO2
bonds can be still observed. These results indicate that the substantial changes
occur in the environment of the phenyl ring of the nitrobenzene, but that the
residual organic material stays on the surface. On the unheated surface, the
initial amplitudes of the bands due to the νNO2stretching vibrations are about
three to four times higher in comparison to the band due to the ν(C–C)ring
stretching vibrations which appears at 1600 cm−1. However, after the thermal
desorption, the bands due to the ν(NO2) and the ν(C–C)ring exhibit compa-
rable amplitudes. The shift of the band initially positioned at 1600 cm−1to
higher frequency by 20 cm−1can be due to the absorption by a ν(C=C) vibra-
tional mode in alkenes [127]. This is then in agreement with earlier work [126]
where the detection with mass spectrometry indicated the decomposition of
nitrobenzene into C2H2, C2HNO2and NO2fragments at temperatures above
2000C.
5.3 Methoxybenzene on Au, Si(111) and TiO2
surfaces
Spectra obtained after the grafting of MeB (methoxybenzene) films on TiO2/Ti,
Au and Si(111) surfaces are shown in Fig. 5.12. The bands marked in Fig 5.12
can be attributed to the vibrational modes of MeB molecules [102,128] and are
identified in the spectra for each of the three substrates. The band around
1178 cm−1arises from rocking vibrations of CH3groups [102]. The band
around 1250 cm−1is attributed to C-OCH3stretching vibrations. Bands at
1515 cm−1and 1610 cm−1probably arise due to the ring C–C stretching vibra-
tions [102, 128]. Another characteristic vibration for MeB around 1030 cm−1
is attributed to O-CH3stretching vibrations [102, 128]. This band appears in
spectra obtained from TiO2/Ti and Au surfaces, but is not distinguishable on
a spectrum obtained from the MeB/Si(111) surface. On Si surfaces, the range
between 1050 cm−1and 1250 cm−1is the range of the absorption bands due to
the stretching vibrations of silicon oxides [129] and the MeB vibrational modes
may be overlapping with those from the SiOxcontributions from the substrate.
The absorption signal around 1440 cm−1, which appears as peak-up feature in
the spectrum obtained from the Si surface is most likely due to residual NH+
4
ions on the surface, as was pointed out in section 5.1 [111,130]. This band was
present on the hydrogen passivated Si surface which was used for referencing
the spectrum and is the result of the pretreatment of Si. Additionally, the MeB
– related band intensities from Au and Si surfaces were of similar magnitudes,
while the MeB – related bands obtained from the TiO2/Ti surface were of much
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 67
MeB/TiO
MeB/Au
MeB/Si(111)
1480 1520 1560 1600 1640 1680
I
I
MeB/TiO2
Wavenumber(cm )
-1
Wavenumber(cm )
-1
tan /tany y
FS
tan /tany y
FS
I
0.005
0.001
0.003
2
I
0.002
MeB/Si(111)
MeB/Au
1050 1200 1350 1500 1650
Figure 5.12: Left: tan ψspectra obtained from MeB on TiO2, Au
and Si(111) surfaces; right: comparative figure for absorption band
intensities of MeB on the different surfaces in a spectral range between
1480 cm−1and 1680 cm−1.
stronger intensities.
Measurements performed by VIS-ellipsometry indicated a thickness of 55 nm
and a refractive index of 1.33 for MeB on TiO2/Ti, while thicknesses of 4 nm and
3 nm, and a refractive index of 1.46 were obtained for Au and Si surfaces, respec-
tively. The differences in refractive indices in thinner and thicker films could be
due to a different molecular structure and density in these films. Similar to the
nitrobenzene film on a TiO2/Ti surface in the previous section, a higher grafted
thickness on the TiO2/Ti surface could be due to a non self-limiting process that
was observed earlier for grafting of NB onto oxidized Si surface [2]. This differ-
ence in thicknesses is the main reason for different band amplitudes that were
observed for various substrates in the IR spectra in Fig. 5.12. For discussion
of the orientation of methoxybenzene on Au surface, we apply surface–selection
rules in a similar way as it was done in section 5.2 for nitrobenzene. Bands due
to stretching vibrations of C-OCH3and O-CH3were observed at 1250 cm−1
and 1030 cm−1. For a thin film on a metallic substrate this is only possible
if the corresponding transition dipole moments have a component in the direc-
tion of the surface normal. Assuming that the molecules have planar geometry
and that the transition dipole moments of C-OCH3and O-CH3are lying in the
molecular plane [102], the orientation of the molecular planes can not be parallel
to the substrate surface. However, a non-planar geometry, as reported by Tan
et al [131] for MeB on Pt surfaces could also be the reason for our observations.
Therefore a clear assertion for the molecular orientation is not possible for this
case.
A good agreement was found between the spectra of MeB on the Pt sur-
face as obtained by Tan et al [131] and the spectra of MeB on TiO2, Au and
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 68
Si(111) surfaces. Nonetheless, we observed certain differences in the band posi-
tion of some of the vibrational bands: for instance, Tan et al [131] reported the
phenyl-related stretching vibrations at 1603 cm−1, 1587 cm−1, 1500 cm−1and
1441 cm−1for MeB multilayers on Pt. For electrochemically grafted MeB on Au
and on TiO2, we observed these bands at 1608 cm−1, 1575 cm−1, 1513 cm−1
and 1463 cm−1, respectively. These differences probably arise from different
deposition methods that were employed in both cases. While Tan et al [131]
used vacuum deposition, where the MeB molecules were dosed on the surface
from the gas-phase, in this work the molecules were deposited electrochemically.
The electrochemical deposition requires creation of radicals which changes the
electronic [23] and the vibrational [132, 133] properties of molecules. Also the
multilayer growth mechanism may differ from that under vacuum condition
preparations, and may proceed through the polymerization of radicals [25]. All
these differences may result in a shift of the absorption band positions of elec-
trochemically grafted films relative to the frequencies observed in spectra of
liquid-phase MeB [102] or films deposited from the gas–phase.
Fig. 5.13 shows a comparison between IRSE spectra of the silicon surface
terminated with the MeB derivative, namely 4–methoxy–diphenyl–amine (4–
MDA) and the IRSE spectra obtained from the MeB terminated surfaces. The
methoxybenzene group is a composite of the both MeB and of the 4–MDA. Thus,
IRSE spectra obtained from surfaces modified with these molecules have similar
absorption features, as shown in Fig. 5.13. This figure presents a spectrum
obtained from Si(111) surface modified with 4–MDA along with a spectrum of
MeB on Au surface for comparison. The spectra from the MeB/Au were chosen
for this comparison due to a better signal to noise ratio than was obtained for
MeB/Si as was shown earlier in Fig. 5.12.
Absorption bands due to the C6H5OCH3are present in both 4–MeB and
4–MDA. Table 5.4 shows the assignment of the prominent absorption bands
marked in Fig. 5.13. The wealth of absorption bands between 1040 cm−1to
1100 −1as reported in Ref. [134] may overlap with SiOxvibrational frequencies
thus the assignment in this spectral range is not unequivocal. The bigger
4–MDA molecules need a greater volume than NB or MeB on a surface since it
consists of two phenyl rings. Section 6.2 will show that this property plays an
important role in a protection (passivation) of the Si substrate from oxidation.
5.4 Electrochemical grafting on porous silicon
Preparation of porous silicon (PSi) through electrochemical treatment in acidic
fluoride solutions was mentioned in section 2.3. In this section the IR spec-
troscopic properties of hydrogen passivated and of organically modified PSi is
presented. Specifically, the organic modification will include the nitrobenzene
(C6H5NO2) and methyl (CH3) termination.
Fig. 5.14 shows the SEM micrograph obtained from the PSi surface. Ac-
cording to refs. [41,42,43,44,45], the pores in the silicon surface are formed as
walls which are perpendicular to the Si surface plane. Fig. 5.15 schematically
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 69
1050 1200 1350 1500 1650
tan /tany y
FS
Wavenumber(cm)
-1
MeB/Au
MDA/Si(111)
145
2367
0.001
N
H
OCH3
OCH3
Figure 5.13: IRSE referenced tan ψspectra of MeB/Au (upper spec-
trum) and 4–MDA on Si(111) (lower spectrum). Spectrum of MeB/Au
is shown for comparison. The assignment of the marked bands is sum-
marized in Table 5.4.
Band No. 4-MeB/Au Assignment 4-MDA/Si(111) Assignment
Fig. 5.13 cm−1[102,131] cm−1[102,131,134]
1 1030 ν(O-CH3) not clear ν(O-CH3),
ν(C-N)
2 1110 δ(C−H) 1105 δ(C−H)
3 1178 ρ(CH3) - ρ(CH3)
4 1250 ν(Ph-O) 1240-1267 ν(Ph-O)
5 - - 1348 ν(C-N)
6 1515 ν(C-C) 1515 δ(N-H)
7 1610 ν(C-C) 1610 ν(C-C), δ(N-H)
Table 5.4: Observed frequencies for grafted 4–MeB and 4–MDA as shown com-
paratively in Fig. 5.13. ν–stretching, ρ–rocking, δ–bending vibrational modes
as assigned on the basis of the references above. Absorption bands between
1040 cm−1to 1100 cm−1in Fig. 5.13 may overlap with SiOxvibrational fre-
quencies thus the assignment in this spectral range is not unequivocal.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 70
shows orientation of the hydrogen and methyl (CH3) on such walls in porous
silicon and compares to the orientation on flat silicon surfaces. The support
of the model shown in Fig. 5.15 is provided in the following, through the com-
parison of the line shapes of the IRSE spectra obtained from the porous and
flat silicon surfaces. The next subsections are structured as follows: first, the
silicon
substrate
porous
silicon
Figure 5.14: SEM micrograph obtained from the porous silicon surface.
Si
H C
Si
Si
Si
a. b.
Figure 5.15: a. Schematic drawing of hydrogen (left) and CH3(right)
orientation in the pores of the silicon. b. Schematic drawing of the
molecules attached to the flat silicon surfaces. The molecular symmetry
axis is indicated by red arrows for the CH3groups.
spectra obtained from the hydrogen passivated silicon surfaces are presented.
The support to the model described in Fig. 5.15 is provided in accordance to
the observations of the spectral line shapes. Next, IRSE results obtained from
organically modified flat Si(111) surfaces will be compared to those obtained
from PSi surfaces. Orientation of nitrobenzene (C6H4NO2) and methyl (CH3)
grafted on porous silicon surfaces will be discussed.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 71
5.4.1 IRSE characterization of PSi: comparative studies
with Si(111) and Si(001)
In this section, IRSE spectra obtained from the flat Si(111) and Si(001) hydrogen
passivated surfaces are compared to those obtained from porous silicon (PSi),
as shown in Fig. 5.16. At lower frequencies, spectrum obtained from the PSi
surface (Fig. 5.16 (a)) exhibits absorption peaks due to the bending vibrational
modes as follows: at 636 cm−1due to Si–H bending vibrations [135, 136, 137],
at 669 cm−1due to Si–H wagging vibrations [135,136], and at 906 cm−1due to
SiH2scissors mode [135,136,137]. At higher frequencies, absorption peaks due
to stretching vibrations are observed: the mode at 2094 cm−1is assigned to Si–H
stretching vibrations [135,137], the mode at 2121 cm−1is due to SiH2stretching
vibrations [135,137] and the absorption at 2140 cm−1is due to SiH3stretching
vibrations [135, 137]. The absorption peaks due to the Si–Oy–Hxmolecular
vibrations at 2200 cm−1and around 2250 cm−1[138] were not observed.
Fig. 5.16 (b) exhibits spectra obtained from Si(001) surface in the spectral
range between 2000 and 2280 cm−1. The absorption peak around 2110 cm−1
is assigned to dihydride stretching vibrations and the feature at 2140 cm−1to
tryhydride stretching vibrations [64]. The absorption peak due to the monohy-
dride stretching vibrations is expected at 2080 cm−1, it is however at the noise
level of the obtained tan ψand ∆ spectra.
Fig. 5.16 (c) shows the spectra obtained from hydrogen–passivated Si(111)
surface, which exhibit two sharp absorption peaks due to silicon monohydrides
vibrational modes: at 2082 cm−1the mode is due to Si–H stretching vibrations
and at 626 cm−1is due to the Si–H bending vibrations [135]. Comparison
between the spectra in Fig. 5.16 reveals striking differences in band shapes.
While the spectra obtained from the well-defined Si(111) surface exhibit a peak–
up feature due to the absorption through the bending Si–H vibrations and a
peak–down feature due to the absorption by the stretching Si–H vibrational
mode, the spectra obtained from the PSi exhibit inverse features. Comparison
between the IRSE spectra obtained from the PSi and Si(001) reveals the peaks
positioned at the same frequencies due to the dihydride and trihydride stretching
vibrations in the spectral range between 2000–2200 cm−1, which however have
different line shapes.
In section 3.4 we showed that the peak–up features on H/Si(111) surfaces
arise due to the absorption through the molecular vibrations with transition
dipole moment parallel to the surface plane. On the other hand, peak–down
features arise due to the absorption through the molecular vibrations with tran-
sition dipole moment perpendicular to the surface plane (see Fig. 3.8 in sec-
tion 3.4). The inverted peak shape for the PSi suggests that the transition
dipole moments have inverse features in PSi as compared to the flat silicon
surfaces, with respect to the radiation plane of incidence.
These observations support the earlier presented model as shown in Fig. 5.15.
The lineshapes differences can be understood when orientation of the transition
dipole moment relatively to the surface plane are taken into account. The pores
in the silicon surface were formed as walls which are perpendicular to the Si
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 72
tan /tany y
FS
D D-
FS()
0
c. d(Si-H) n(Si-H)
I0.002
I0.1
600 750 900 175020002250
D()
0
a.
0.01
I
I3
n(Si-H)
d(Si-H)
dsc(Si-H)
Wavenumber(cm)
1850 1950 2050 2150 2250
tan y
I0.001
-1
tan y
2030 2100 2170 2240
x
2140(T)
2110(D)
2080(M)
?
H/Si(001)
PSi
I5e-5
I0.01
600 750 900 1750 2000 2250
H/Si(111)
b.
Wavenumber(cm)
D()
0
tan y
-1
Figure 5.16: a. IRSE spectra of hydrogenated porous silicon; b. H–
passivated Si(001) surface (from Ref. [121]); c. H–passivated Si(111)
surface referenced to oxidized Si(111) surface. δstands for bending and
νfor stretching vibrations. Panel (a) zooms in the absorption peaks
due to the ν(Si–Hx). In panel (b) the dihydride (D) and trihydride (T)
stretching vibrations with the corresponding Kramers–Kronig related
parts in ∆ spectrum are marked. For monohydride (M)–related band
at around 2080 cm−1no correspondence in ∆ was observed. The set–up
related spectral between 590 and 632 cm−1is shown in light gray on
panel (a).
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 73
surface plane, in agreement with other works [41,42, 43,44,45]. These perpen-
dicular walls have a high volume for hydrogen passivation. This is reflected in
the inverse band shapes of IRSE spectra of PSi as compared to the flat Si(001)
and Si(111) cases.
5.4.2 Organic modification of porous silicon
In this section, IRSE spectra obtained from the organically modified porous
silicon (PSi) are discussed. The focus is on two different types of modifica-
tions: nitrobenzene modification and methyl modification from Grignard com-
pounds [139,140,141], as was described in section 2.2.1.
Fig. 5.17 shows spectra obtained from CH3and CD3modified, as well as
from a non–modified PSi surfaces, for comparison. All spectra represent ab-
sorption bands due to Si–H bending vibrations in the range between 630 and
730 cm−1. Thus, methylation does not replace all of the Si–H bonds of PSi. The
absorption peak due to the rocking CD3vibration lies around 605 cm−1[142],
which unfortunately overlaps with the absorption band due to the IRSE setup
(Fig. 5.17, gray curve). Spectrum obtained from CH3/PSi exhibit an absorp-
tion band at 773 cm−1, due to the rocking CH3vibrations [143]. The mode at
906 cm−1also appears on all of the spectra and belongs to SiH2scissors vibra-
tional mode. The modes between 2060 and 2150 cm−1belong to the absorption
peaks due to SiHxstretching vibrations, as discussed in the previous section.
This is also the region where the absorption bands due to stretching vibrations
of CD3are expected [142]. The absorption peaks due to stretching vibrations
of CH3are present at 2902 cm−1(νss(C-H)CH3), 2931 cm−1(νas(C-H)CH2,
hydrocarbon contamination remnants) and at 2969 cm−1(νas(C-H)CH3) [143].
Unfortunately, the above mentioned spectral overlaps hindered the observation
of the absorption peaks due to the CD3molecular vibrations.
Similar to the discussion of Si–H orientation on the walls of the PSi, also
here the shape of the absorption peak due to the molecular vibrations can be
used to estimate the orientation of the CH3groups on the PSi walls. The peak
due to rocking vibrations at 773 cm−1has a peak-down feature, corresponding
to a transition dipole moment perpendicular to the surface plane. On the other
hand, the absorption bands between 2902 and 2980 cm−1appear as peak–up
features. The orientation of the transition dipole moments due to the methyl
stretching vibrations νss(C-H) and νas(C-H) thus should be considered. While
the transition dipole moment due to the symmetric stretching vibration is ori-
ented parallel to the symmetry axis of the methyl group, the transition dipole
moment due to the asymmetric mode has projected components both in the
direction of methyl symmetry axis and perpendicular to it [144]. The peak–up
feature due to the symmetric stretching mode can arise due to the transition
dipole moment oriented parallel to the surface (perpendicular to the walls of the
PSi.) The transition dipole moment due to the asymmetric stretching vibration
has components both parallel to the surface and perpendicular to it. However,
the appearance of this absorption peak as a peak–up feature presumes that the
major component of the transition dipole moment is also oriented parallel to
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 74
0.40
0.42
0.44
0.46
0.48
0.50
0.52
0.54
450 700 950 1200 1450
-150
-140
-130
-120
-110
-100
1500 1850 2200 2550 2900 -220
-200
-180
-160
-140
-120
-100
Wavenumber(cm)
-1
tan y
D()
0
CH/por.Si
CD/por.Si
por.Si
3
3
0.30
0.32
0.34
0.36
0.38
0.40
0.42
0.44
0.46
0.48
0.50
0.52
2750 2850 2950 3050
a. b.
d.
tan y
D()
0
c.
I0.001
I4
d(Si-H)
dsc(Si-H)
n(CH)
3
r(CH)
3
n(Si-H)
x
n(CH)
3
ss
n(CH)
3
as
n(CH)
(contam.)
2
as
Figure 5.17: IRSE spectra obtained from PSi surfaces modified with
CH3(red), CD3(black) and non–modified PSi (blue). a,b: tan ψ
spectra in the spectral range between 450 to 1500 cm−1and 1500 to
3100 cm−1, respectively. c,d: the corresponding ∆ spectra of the plots
a,b respectively. The gray plot in (a) shows the absorption peak due to
the IRSE setup, arising from the polarizers. ν: stretching; ρ: rocking;
δ: bending vibrational modes.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 75
the surface (perpendicular to the porous walls of the Si). Schematic drawing of
the orientation of hydrogen and of the CH3groups on PSi as suggested by the
observed lineshapes in the IRSE spectra was shown schematically in Fig. 5.15
(a). However, theoretical calculations involving pore parameters and the open-
ing angle of the C–H bond of the methyl group are necessary in order to give a
more precise understanding of the orientation of the methyl group in PSi.
On the other hand, the grafting of nitrobenzene on PSi does not result in any
preferential orientation as no peak–up features in the spectra were observed, as
shown in Fig 5.18. The assignment of the nitrobenzene vibrational modes was
given earlier in Table 5.2. The spectra in Fig. 5.18 do not show any absorption
peaks due to the bending vibrations of the remnant Si–H bonds between 630 and
700 cm−1. Instead, observation of the absorption features in regions (a),(b) and
(c) in Fig. 5.18 indicate the formation of SiOxon the porous surface. Region
(a) in Fig. 5.18 belongs to SiOxrelated absorption bands in Si–O–H vibrational
modes [135,145] and to the Si–O bending vibrations in O–Si–O units [136]. A
broad peak in region (b) between 1010 and 1265 cm−1arises due to the Si–
O–Si stretching vibrations [135, 145]. Spectral range (c) marked in Fig. 5.18
exhibits weak broad absorption peaks observed at 2130 and 2250 cm−1, which
were assigned to Si–H stretching mode in Si2O–SiH units and to Si–H in O3–SiH
units in PSi, respectively [136,138]. The main nitrobenzene related absorption
peaks are band (3) in Fig. 5.18 at 1350 cm−1(νss(NO2)), band (4) at 1523 cm−1
(νas(NO2)) and band (5) at 1600 cm−1(ν(C-C)ring). These appear as peak–
down features, similarly to the spectra obtained from Si(111), Au and TiO2
surfaces as discussed in the previous sections. Band (2) at 1112 cm−1was
observed also for NB/TiO2(see Table 5.2) and was assigned to a combination
C–N stretch and ring breathing vibration mode. The origin of the weak band
at 752 cm−1labeled (1) is not clear. As no mode at this frequency was listed as
resulting from the nitrobenzene molecular vibrations as adsorbed on a surface
or in a liquid state [114], one of the possibilities can be that this mode belongs
to the Si–C stretching vibrations. The detailed discussion of Si–C bond will be
presented in section 5.4.3.
The spectra in Fig. 5.18 suggest that in contrast to methylated PSi surface,
oxidation occurred for NB–modified PSi. The oxidized interface results from
the grafting in aqueous solution, which was not the case for the methylated
surfaces. In addition, no preferential orientation similar to the CH3–modified
PSi was observed. This could be due to the polymerization of radicals discussed
in section 5.2 which probably occurs during grafting on PSi, leading to a loss of
the structure of the grafted molecules.
5.4.3 The Si–C bond: discussion
In principle, strong signals from absorption peaks of PSi can be utilized for
searching for the Si–C bond, that should arise due to a covalent attachment of
the molecules to the surface. Yamada et al [146] performed HREELS studies on
methylated Si(111) surfaces and assigned Si–C stretching mode at 683 cm−1and
Si–C bending mode at 507 cm−1. Theoretical studies on methylated Si(111) sur-
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 76
600 900 1200150018002100
-165
-157
-149
-141
-133
-125
0.30
0.33
0.36
0.39
0.42
D()
0
tan y
Wavenumber(cm)
-1
a. b. c.
1 2 3 45
setup
feature
Figure 5.18: IRSE spectra of nitrobenzene on PSi. The absorption
feature between 590 and 632 cm−1was due to polarizers in the setup.
Regions a–c, as well as the absorption peaks 1–5 are discussed in the
text.
faces [147] resulted in assignment of Si–C stretching mode at around 640 cm−1
and of the bending mode in the range between 480–500 cm−1. Our spectra
of PSi (Fig. 5.17) show a weak peak at 509 cm−1which is however observed
for all of the surfaces: H-passivated PSi, CH3–modified and CD3–modified PSi.
Thus a clear assignment of the bending mode is not possible in our case. The
absorption peak due to the stretching Si–C vibrations is expected in the range
of the Si–H bending vibrations, thus we could not observe it either. As to ni-
trobenzene Si–C bond, the HREELS results of Ref. [148] showed a Si–C bond
at 615 cm−1for benzene at Si(100) and at 540 cm−1for benzene at Si(111) sur-
faces. These modes either overlap with the absorption band resultant from our
setup or outside the detection region. An additional possibility is that nitroben-
zene binds to Si surfaces through Si–O–C bond. All of our XPS spectra have
indicated suboxides species even when HF was added to the electrolyte, raising
a possibility that the grafting may require the formation of Si–O–C bond. How-
ever, these suboxides could result also from the rinsing of the sample in water
after grafting. FTIR studies of methanol on Si surfaces indicate the formation
of absorption peak due to Si–O–C between 1085 and 1100 cm−1[149]. Similar
studies by Michalak et al [150] on deuterated methanol showed absorption due
to Si–O–C bond of Si–O–CD3between 1078 and 1088 cm−1. In nitrobenzene
on PSi spectrum, this spectral range is overlapping with a broad peak due to
the vibrations of SiOx.
CHAPTER 5. OPTICAL PROPERTIES OF ORGANIC THIN FILMS 77
As was already mentioned, spectrum shown in Fig. 5.18 exhibits a weak band
at 752 cm−1labeled (1) whose origin is not clear. For instance, no mode at this
frequency was listed as resulting from the nitrobenzene molecular vibrations as
adsorbed on a surface or in a liquid state [114]. There is a possibility that this
mode belongs to the Si–C stretching vibrations. Gurthner et al [151] observed
an absorption peak at 762 cm−1for electrochemical grafting of methyl iodide on
PSi. While unfortunately this spectral range overlaps with strong absorption
peaks due to the Si–H for methyl terminated PSi (Fig 5.17), the absorption
band (1) present in the NB/PSi spectrum in Fig. 5.18 can be due to the Si–C
absorption. However a more clear assignment may be provided upon isotope
labeling of either C or H atoms in the phenyl ring, which will lead to the
shift of the possible absorption peak due to the Si–O–C or Si–C stretching
vibrations. Additionally, further extension of the spectral range toward even
lower wavenumbers may help in detection of Si–C bending vibration if this is
the case for the electrochemical attachment of the molecules to Si surfaces.
5.5 Summary
This chapter summarized the characterization work on electrochemically grafted
organic thin films. Through the combined XPS and IRSE results we showed
that electrochemical preparation leads to organic termination with the grafted
molecules on different surfaces. Cross–referencing the data from complementary
techniques helped us in quantitative determination of optical parameters and
thickness of the thin films. IRSE analysis enabled studies of molecular orienta-
tion of the organic molecules on various substrates. The obtained results point
to formation of films consisting of several layers of the organic molecules, which
is likely due to polymerization of radicals during the electrochemical procedure.
The films grafted from benzene derivatives showed no preferential orientation
on the host surfaces.
The situation was different when IRSE spectra of PSi were compared to
those obtained from Si(111) and Si(001) surfaces. First, the spectra showed
substantial differences in absorption peak shapes due to Si–H vibrational modes.
Second, absorption peaks due to molecular vibrations of CH3terminated porous
silicon showed similar trends in lineshapes as hydrogenated PSi. Based on these
obsevations, we estimated the orientation of the hydrogen and of the CH3on
porous silicon with the molecular symmetry axis perpendicular to the porous
walls. Such orientation was not detected in nitrobenzene grafted on PSi, which
exhibited features similar to those detected in unordered thick films on TiO2
surfaces.
The next chapter presents characterization of the silicon oxide interface
which forms below the organic layer as a result of the exposure to atmospheric
conditions and during the electrochemical grafting process.
Chapter 6
Passivation and oxidation of
Si surfaces
In the previous chapter, it was shown that electrochemical grafting leads to a
successful termination of the electrode surfaces with organic molecules. For sili-
con surfaces, the formation of the oxidized SiOxinterface was briefly mentioned.
In this chapter the studies on interfacial structure between grafted organic layer
and Si(111) surfaces are presented in more details. Generally, uncontrolled SiOx
interfaces are undesired in the engineering of the hybrid organic/silicon devices,
since they introduce a large number of surface electronic trap states [152]. The
understanding of the processes governing the formation of the SiOxinterface
and its prevention are thus important.
These studies were realized through the measurements delivered by IRSE,
XPS, VIS–ellipsometry and AFM methods. Deconvolution of the Si2p core level
spectra gave insight to the SiOxstructure of the sample. Cross–referencing
of the results from IRSE and XPS enabled quantification of the SiOxlayer
thickness. The model included the substrate/SiOx/nitrobenzene layer, where
the SiOxlayer was simulated taking into account the possibility of an “island”–
like structure (see Fig. 4.12 in the earlier section 4.2.2).
It is shown that in general, organic thin films slow down the oxidation when
the samples are exposed to ambient conditions. It was found that different
organic molecules dictate different oxidation kinetics of the silicon interface.
This chapter presents that the HF addition into the electrolyte can prevent
the formation of an oxidized interface. HF also was found to successfully remove
the oxidized SiOxinterface which formed when the samples were stored for
a certain time under atmospheric conditions. It was expected that such HF
treatment would also remove the organic layer grafted above the oxidized silicon
interface. However, the absorption peaks due to the overlying organic layer
remained unchanged as exhibited by the IRSE spectra after the HF treatment.
A model for such peculiar behavior is proposed.
This chapter is structured as follows: First, characterization of hydrogen-
78
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 79
passivated Si(111) exposed to atmospheric conditions is presented along with
the simulations of the degrading absorption peaks due to the Si–H molecular
vibrations as measured by IRSE. Further, we show that during storage under
atmospheric conditions, grafted organic monolayers slow down the formation of
the SiOxinterface. Using cross–correlated XPS, IRSE and VIS–ellipsometric
studies, quantitative analysis of the SiOxlayer forming on the Si(111) substrate
under ambient conditions in the initial stages of oxidation and after 1 year of
storage is performed.
6.1 Stability of H-passivated Si (111) surfaces
Optical characterization of the Si surfaces gives insight into the initial compo-
sition of the surface before the grafting as to its hydrogen termination, surface
oxides presence and contaminations. In section 5.4.1 we presented IRSE studies
of Si(111) hydrogenated surfaces and compared the results to the spectra ob-
tained from PSi. In this section, we perform studies on degradation of the Si–H
passivation as a result of exposure to ambient conditions.
Fig. 6.1 shows a series of spectra that were obtained from the Si(111) surface
upon storage under ambient air conditions. Degradation of the surface is ex-
pressed in disappearance of the initially strong ν(Si–H) absorption band around
2082 cm−1. Si surfaces are very unstable under ambient conditions and the
degradation typically occurs within 24 hours. The red curves in Fig. 6.1 show
the results of simulations. The fitting parameters are summarized in table 6.1.
The measured optical constants of oxidized Si substrate were taken to simulate
the substrate, thus the simulated spectra include the measurement noise. In
the simulations, the value for the high frequency refractive index n∞= 1.1 was
taken from the ref. [64,38]. The Si-H layer thickness was taken as the length of
the Si-H bond of 1.5 ˚
A [65].
The degradation was simulated varying the fraction of a surface coverage
with the Si-H bonds. This was done through the superposition of the reflection
coefficients, with rtotal =θrSiH + (1 −θ)rSi where θwas the coverage of the
Si surface with Si–H bonds, rSiH was the reflection coefficient calculated with
the Lorentz oscillator for simulation of the absorption peak due to the Si–H
stretching vibrations and rSi was calculated as rSiH but without the Lorentz
oscillator at this position. The superimposed quantity rtotal represents the radi-
ation reflected from H–terminated parts of a Si surface and the radiation which
is reflected from Si atoms which are not bonded to the H atoms due to the
degradation under air.
Initially, the surface is fully hydrogen-terminated, with the S-H coverage
fraction of 1. As oxidation proceeded, the Si-H coverage fraction gradually
became smaller and after 24 hours, it was fitted with the coverage value of
0.1. However, this parameter did not compensate for the Si-H peak broadening
which occurred during the oxidation. It was necessary to change the broadening
oscillator parameter Γ in order to simulate the spectrum broadening.
Integration of the coverage parameter into our simulations is based on the
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 80
2000 2050 2100 2150
Wavenumber (cm-1)
timeafter
preparation(hours)
24
10
8
6
4
2
0
24
10
8
6
4
2
0
0.0002
0.1
tan y
D()
0
Figure 6.1: Ellipsometric spectrum obtained from hydrogen passivated
Si(111) surface. Black circles: measured data; red line: simulated spec-
tra. On the right side a storage time in hours is shown beside each
spectrum. The measured optical constants of oxidized Si substrate were
taken to simulate the substrate, thus the simulated spectra include the
measurement noise. For the lowest presented absorption amplitude af-
ter 10 hours the tan ψSNR was 3 and SNR of ∆ was 7.
Frequency ω0time F Γ coverage
(cm−1) (hours) (cm−2) (cm−1)
2077 0 32000 2 1
2077 2 32000 3 0.9
2077 4 32000 5 0.75
2075 6 32000 10 0.6
2075 8 32000 16 0.5
2075 10 32000 16 0.4
2075 24 32000 16 0.1
Table 6.1: Fitting parameters for hydrogen degradation series as shown in
Fig.6.1
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 81
model proposed by Miura et al [153]. In this model, the water present in the air
is predominantly involved in the oxidation of surface Si-H bonds. At the next
step, native oxide grows where most of the surface Si-H bonds were already
eliminated (Fig. 6.2). The variable θ=θ1+θ2in Fig. 6.2 corresponds to the
coverage fraction used in our simulations.
H H
H
H
H
H
H
H
H
H H H
H
oxidized
Coverage q1
Coverage q2
H
H
O2H 0
2
HH H
SiO2
H
H
H
H
H
H
a)
c)
b)
Figure 6.2: A model proposed for oxidation kinetics of hydrogen-
passivated Si surfaces, after Miura et al [153]. a. Initially H-terminated
Si surface; b. Initiation of the surface oxidation; c. Native oxide for-
mation favorably proceeds on the oxidized surface simultaneously with
further degradation of the Si-H bonds.
The broadening of the Si–H absorption peak is also in agreement with the
literature. Ogawa et al [154] confirmed that the broadening does not result from
physically adsorbed contaminations from ambient atmosphere. They proposed
that the spectral broadening can be inferred to the oxygen entering subsurface of
silicon lattice sites and distorting the lattice, influencing Si-H vibrational prop-
erties. Broadening due to the interaction with the lattice was confirmed from
observation of the deuterium-terminated diamond C(111) surfaces as compared
to the hydrogen-terminated C(111) surface [155]. The broadening of the absorp-
tion line for deuterium-terminated C(111) was found to be stronger than for the
hydrogen-terminated C(111) diamond surface. For the deuterium on C(111) the
stretching frequency is situated closer to diamond surface phonons compared to
that of the C–H stretch. Thus the C–D stretching motions can easily relax via
a two-phonon process, resulting in broad features in the IR spectra. Dephasing
due to coupling to the lattice phonon modes on hydrogenated semiconducting
surfaces was confirmed by measurements and calculations performed by Wang
et al [156].
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 82
Other effects that can be responsible for the broadening of the absorption
peak and its frequency shift are the dynamic and static multipole coupling
mechanisms. The former is due to the transition multipole coupling, with the
dominating dipole coupling. The latter is due to interaction between perma-
nent multipoles, i.e. the chemical environment of the adsorbates [157]. The
close spacing of surface adsorbates affects the vibrational spectrum because the
motion of an individual oscillating dipole is influenced by the electric dipole field
generated by the motion of its neighbors. This effect forces adsorbates bound
to a surface to vibrate together instead of individually and shifts the vibra-
tional transition relatively to single Si-H vibration. The coupling is dependent
on the morphology of the overlayer, thus the surface structure (etch pit density,
roughness) determines the frequency shift and broadening [158]. This effect is
relevant in studies of surface morphology upon preparation of hydrogenated Si
surfaces, where etch-pits have a strong influence on the shape and frequency of
the Si-H absorption peak [158].
In case of hydrogen passivated Si(111), dynamic dipole coupling is influenced
by changes in morphology during the surface degradation. For instance, AFM
studies showed that within the first 10 to 22 hours under atmospheric conditions,
the surface roughness increased [159,160]. The roughness influences the dynamic
dipole coupling and leads to broadening of the absorption spectra and to the
frequency shift to lower frequencies [158]. Additional effect that might have an
influence on the dynamic dipole coupling is a removal of part of the Si-H bonds
during the oxidation, which may lower the overall dipole field [158]. A similar
red-shift and broadening of the Si-H absorption peak was observed by Lambers
and Hess [161] upon F2laser induced oxidation (wavelength 157 nm), and was
assigned to the loss of the dynamic dipole coupling.
6.2 Oxidation under atmospheric conditions
In order to investigate the stability of organic layers/Si(111) interfaces to atmo-
spheric conditions, we monitored IRSE spectra of H/Si(111), NB/Si(111), and
4–methoxydiphenylamine (4–MDA) on Si(111) for one year. Fig. 6.3 presents
the monitored spectra for each of the substrates (plots a,b,c). A compara-
tive graph between the above samples is shown by plot (d). As was discussed
in section 5.3, there are several absorption bands due to molecular vibration
of 4–MDA overlapping with the SiOxabsorption region. However, it is clear
from Fig. 6.3 (d) that the oxidation process of the 4–MDA modified Si surface
proceeds much slower than for the other two substrates. There are also some
differences in oxidation pattern for H/Si and NB/Si. After 18 hours of stor-
age, no bands due to SiOxcan be yet observed for NB/Si and H/Si samples.
However, after 50 hours, there appear major differences in the absorption re-
gion of SiOx. For H/Si sample, a distinguishable SiOxabsorption band can be
observed. A similar band also appears for NB/Si sample, but of about half of
the amplitude which is observed for H/Si sample. Similar tendency is seen also
after 6 days of oxidation: the amplitude of the SiOx–related absorption bands is
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 83
much smaller for NB/Si than for H/Si sample. After this time period, the oxide
thickness on H/Si sample reaches about 0.6 nm, based on IRSE simulations as
described in section 6.3. After 27 days, the amplitude of the SiO2related bands
on the spectrum obtained from NB/Si approaches the one that can be seen on
H/Si spectrum. After 388 days, the SiOx–related absorption bands become of
a similar amplitude at 1220 cm−1, with minor differences between 1140 and
1160 cm−1. The amplitude of the SiOx–related absorption bands of 4–MDA/Si
sample remains smaller than for H/Si and NB/Si during this observational pe-
riod. The thickness of the SiO2on H/Si sample was 2.5 nm after 388 days,
as obtained from VIS–ellipsometric measurements. All of the spectra exhibit a
shift of the SiOxrelated absorption band to higher frequencies as the oxidation
proceeds. This shift is due to changing oscillator parameters which is the result
of the changing bonding configuration from SiOxto SiO2interface. The next
section will present the simulations of the IRSE spectra in this region, showing
that the shift can be reproduced in optical models.
AFM images of the H/Si, NB/Si and 4–MDA/Si are shown in Fig. 6.4 after
19 days after 466 days of oxidation. The RMS roughness after 19 days of
oxidation was 0.8 ˚
A for H/Si, 2 ˚
A for NB/Si and 1.5 ˚
A for 4–MDA/Si. The
roughness value for H/Si surface is in a good agreement with refs. [160,159], but
somewhat underestimated in comparison to ref. [162] which reported the lowest
RMS roughness of 1.4 ˚
A upon NH4F treatment. After 466 days, the RMS
roughness of 6 ˚
A was observed for H/Si sample, 3 ˚
A for NB/Si and 3 ˚
A for
4–MDA/Si. The roughness of H/Si surface varies strongly with oxidation, and
eventually after 466 days was higher than the roughness observed for organically
protected Si(111) surfaces.
These results demonstrate that the organic thin films slow down the oxida-
tion of the interface, and that the size of the molecule plays an important role in
oxidation dynamics. Surface morphology is also effected by oxidation. However,
surfaces covered with thin organic films exhibited a lesser roughness than the
non–coated Si surface. The roughness after one year of oxidation was found
to be independent on the type of the grafted molecules. These properties can
be utilized in engineering of electronic devices for protection against interface
oxidation.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 84
I0.002
I0.3
5h
42h
4d
31d
388d
5h
42h
4d
31d
388d
5h
42h
4d
31d
388d
5h
42h
4d
31d
388d
I0.002
I0.3
5h
42h
4d
31d
388d
5h
42h
4d
31d
388d
I0.002
I0.3
I0.002
I0.3
18h
50h
27d
388d
18h
50h
27d
388d
b. d.
tan y / tan y
FS
D D ( )-
FS0
tan y / tan y
FS
D D ( )-
FS0
6d
6d
a. c.
-1
Wavenumber(cm )
H/Si(111) NB/Si(111)
4-MDA/Si(111) Comparativeplot
950 1100 1250 1400 1550
950 1100 1250 1400 1550
950 1100 1250 1400 1550
(NO)
as
2
n
(NO)
ss
2
n
n(SiO)
x
n(SiO)
x
n(SiO)
xn(SiO)
x
950 1050 1150 1250 1350
( N - H )
d
Figure 6.3: IRSE spectra monitoring oxidation of the H-passivated
Si(111) (a), NB–modified Si(111) (b) and 4–MDA modified Si(111) (c)
under atmospheric conditions. Time of storage is selectively shown for
spectra marked in red. (d). Comparative plot of the spectra obtained
from the above samples. In plot (d), 4–MDA/Si is shown by red line,
NB/Si by black and H/Si by gray. In plots (a,b,c) the colors are imple-
mented only for a contrast. All data was referenced to the spectrum of
freshly prepared H-passivated Si(111).
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 85
19days
4-MDA/Si
400nm
1.44nm
400nm
3.17Å
400nm
1.00nm
400nm
4.33nm
400nm
1.88nm
400nm
2.15nm
466days
0
2 mm
4.35
0
[nm]
2 mm
2 mm
2 mm
1.89
0
[nm]
0
2 mm
2 mm
0.33
0
[nm]
0
2 mm
2 mm
0
2.16
0
[nm]
2 mm
2 mm
0
1.45
0
[nm]
2 mm
2 mm
0
1.01
0
[nm]
4.33nm
0.32nm
1nm
4.33nm
4.33nm1.88nm
4.33nm2.15nm
1.44nm
NB/Si H/Si
2 mm
2 mm
2 mm
Figure 6.4: AFM images of H/Si, 4–MDA/Si and NB/Si after 19 days
(left column) and 466 days of oxidation (right column). The RMS
roughness after 19 days oxidation was 0.8 ˚
A for H/Si, 2 ˚
A for NB/Si
and 1.5 ˚
A for 4–MDA/Si. After 466 days, the RMS roughness was 6
˚
A for H/Si sample, 3 ˚
A for NB/Si and 3 ˚
A for 4–MDA/Si.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 86
6.3 Determination of the optical parameters in
mid–IR spectral range for SiOxlayer form-
ing under ambient conditions on Si(111) sur-
face
In this section we determine the Lorentz oscillator parameters for SiOxforming
on Si(111) in the initial stages of oxidation (below 1 month of storage) and sub-
sequently after 1 year of oxidation. For this, we used a cross–correlated analysis
of IRSE with XPS and VIS–ellipsometry. In the initial stages of oxidation, XPS
was used for determination of the thickness of the forming SiOxlayer. After
one year of oxidation, VIS–ellipsometry was used for this purpose.
Fig. 6.5 shows XPS results for the freshly prepared H–passivated Si(111) and
Fig. 6.6 (a) shows the same sample in its initial stages of oxidation, including the
assignment of the Si0to Si+4 peaks. Both figures contain IRSE results obtained
from the other piece of the same sample, synchronously.
IRSE spectra in Fig. 6.5 reveal a strong ν(Si-H) band due to Si-H stretching
vibration at around 2082 cm−1. At the same time, broad absorption band due
to ν(SiOx) can be observed between 1060 cm−1and 1170 cm−1. However, this
peak probably results mainly from the interstitial oxides. This is supported by
deconvolution of X-ray photoelectron spectra, which reveal only a minor sub–
oxide structure of the sample. Here, the predominance of the Si+1 peak points
out the formation of the Si–OH surface hydroxides. There is also a minor con-
tribution due to the silicon in the Si+4 oxidation state, which presumes a minor
occurrence of the silicon back–bond oxidation. Fig. 6.6 (a,b) shows the results of
99 102 105
0.0
2.0x105
4.0x105
Bindingenergy(eV)
1000 1100 1200 2000 2100 2200
I0.3
I0.0005
Wavenumber(cm )
-1
tan yD()
0
n(Si-H)
n(SiO )x
Si0
Si+1
Si+2
Si+4
Si+3
98 100 102 104 106
0
1x106
2x106
3x106
4x106
5x106
Intensity (a.u)
Figure 6.5: XPS (left) and IRSE (right) spectra from freshly prepared
hydrogen-passivated Si(111) surface. XPS data was obtained at 150 eV
photon energy. On the inset: zoom–in of the XPS spectrum showing
the sub–oxide deconvoluted components.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 87
the measured sample after several days of storage under atmospheric conditions.
The XP spectrum exhibit the predominance of the Si+4 component, pointing
out the silicon oxidized back–bonds. Calculations based on XP spectrum shown
in (Fig. 6.6 (a)) resulted in oxide thickness of 6 ˚
A . Using this data as input,
we performed simulations of the IRSE spectra. The results of the simulations
are presented along with the measured IRSE data in Fig. 6.6 (b). The sample
was measured subsequently after one year of oxidation with VIS–ellipsometry
and exhibited a thickness of 2.5 nm. Fig. 6.6 (c) illustrates the fitted IRSE
spectrum obtained from the sample after 1 year of oxidation. Table 6.2 sum-
marizes the parameters that were obtained from the best–fit calculations of the
above IRSE spectra using the Gaussian distribution of Lorentzians model [59]
(see section 3.2.1 for details), for the sample in initial stages of oxidation and
after 1 year storage under atmospheric conditions. IRSE spectra exhibited in
panels (b,c) in Fig. 6.6 mark the positions of the resonance frequencies ω0which
were used in the simulations. However, the related absorption peaks positions
are shifted in both the measured and the simulated spectra. This is the result of
the optical effect for Lorentz oscillators with a relatively high oscillator strength
F. This optical effect is frequently referred to as the Berreman effect [66].
Fit Frequency F (cm−2) Γ (cm−1)σ(cm−1)
ω0(cm−1)
Fig. 6.6 (b) 1053 281570 10 36
(initial stages of oxidation) 1146 89047 10 39
Fig. 6.6 (c) 1075 170000 10 36
(1 year of oxidation) 1185 83000 6 32.5
Table 6.2: Parameters used in simulations of SiOxabsorption peak as shown
in Fig. 6.6. The thickness was 6 ˚
A as obtained from XPS data for the sample
under initial stages of oxidation and 2.5 nm as obtained from VIS–ellipsometry
for the sample stored for 1 year under ambient conditions. The ǫ∞was 2.1.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 88
tan y / tan y
FS
D D ( )-
FS0
98 100 102 104 106
0.0
2.0x105
4.0x105
6.0x105
8.0x105
1.0x106
1.2x106
1.4x106
1.6x106
Intensity (a.u)
Si+4
Si+3
Si+2
Si+1
Si0
Bindingenergy(eV) Wavenumber(cm )
-1
1000 1100 1200 1300
tan y / tan y
FS
D D ( )-
FS0
Wavenumber(cm )
-1
1000 1100 1200 1300
I0.001
I0.05
I0.001
I0.05
a. b.
c.
w=1053
cm-1 w=1146
cm-1
w=1075
cm-1
w=1185
cm-1
0
0
00
Figure 6.6: a. XPS and b. IRSE spectra from initially hydrogen-
passivated Si(111) surface after storage under atmospheric conditions.
Si0to Si+4 deconvoluted components are marked on the XPS spectrum.
IRSE spectra show the measured data (black) and calculations (red)
using ǫ∞=2.1 and thickness of 6 ˚
A , as calculated from XPS. XPS data
was obtained at 150 eV photon energy. c. IRSE spectrum obtained after
one year of oxidation and fitted with thickness of 2.5 nm as obtained
from VIS–ellipsometry with ǫ∞=2.1. Positions of the Lorentz oscillator
used in the calculations (see Table 6.2) are marked in panels (b) and
(c).
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 89
6.4 SiOxinterface formation during the electro-
chemical grafting
As was presented in section 2.2.3, SiOxinterface can form during electrochemical
grafting in the aqueous surrounding. Formation of the oxide interface may be
prevented when HF is added to the electrolyte [163]. The goal of this work
was to compare the interfacial structure of two samples: the one on which the
nitrobenzene was grafted in HF–free solution (sample A) and the other prepared
in HF–containing solution (sample B).
Addition of HF into the electrolyte can prevent formation of the SiOxinter-
face during electrochemical grafting as outlined in Fig. 2.10 of section 2.3. The
process results in etch–back of the Si surface atoms from hydrogen–passivated
surface. Thus, when aryl diazonium salts are added into the HF–containing
electrolyte, the grafting process proceeds in parallel with the HF–etch of the
surface. Fig. 6.7 shows XPS and IRSE results obtained from the two pieces of
the same freshly–prepared samples A and B. Both XPS and IRSE spectra in
Fig. 6.7 clearly indicate that a higher amount of SiOxwas present in sample
A. IRSE spectra indicated also a higher nitrobenzene amplitude of the absorp-
tion peaks of about 20% for sample A. This is in agreement with our previous
observations of NB grafting on oxidized Si surfaces. Fig. 6.8 shows histograms
of Si+xdistribution in samples A and B. Fig. 6.8 (a) shows histogram of the
Si+xintegrated intensities, normalized to the integrated intensity of the bulk
Si+0 component. The ratios I(Si+x) / I(Si+0) are proportional to a number of
Si atoms in the intermediate oxide state. From this comparison, it is evident
that the total oxide content Si+all (Si+all=Pn
i=0 Si+x) is higher for sample A.
Also, Si+xoxide contents in sample A are higher than in sample B for x=1..3.
However, it is not the case for Si+1 component, where the normalized integrated
intensity is higher for sample B. In other words, there are more Si atoms in this
intermediate state in sample B than in sample A. Fig. 6.8 (b) presents the per-
centage of Si+xcomponents from the sum of integrated intensities Si+all. The
Si+1 integrated intensity in sample B is around 20% of the sum over the Si+x
normalized integrated intensities Si+all, while in sample A, it is only around 4%.
These results point out the predominance of the Si+1 oxidation state in sample
B and on the advanced oxidation stage in sample A (lower Si+1 intensity, higher
intensity of the fully oxidized Si+4 component.)
For calculation of the interfacial SiOxlayer thickness from XPS data ni-
trobenzene overlayer should be taken into account. This demands a knowledge
of the nitrobenzene IMFP values. To our knowledge, such values have not been
reported in literature so far. However, this can be avoided if calculations are
performed using data obtained at several photon energies, as discussed in sec-
tion 4.2.2. Equation 4.15 can be used to estimate the upper limit for SiOx
thickness. Using IMFP of 6.5 ˚
A and cl
csub = 0.84 for SiO2layer at 240 eV [90],
the estimated upper limit for the SiOxthickness is 9 ˚
A for sample A and 3.5 ˚
A for
sample B. The oxidized NB/Si(111) interface for sample B is most probably a
result of water rinsing after the grafting. Due to the nitrobenzene molecular
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 90
98 100 102 104
0
20
40
60
80
100
120
140
160
180
98 100 102 104
0
10
20
30
40
Sample
A
Sample
B
tan y / tan y
FS
D D ( )-
FS0
Wavenumber(cm )
-1
Bindingenergy(eV)
A
B
A
B
Intensity(a.u)
Intensity(a.u)
n(SiO )xnss
(NO )2
nas
(NO )2
950 1050 1150 1250 1350 1450 1550 1650
I0.002
I0.1
Figure 6.7: XPS (left) and IRSE (right) spectra obtained from the sam-
ples prepared without addition of HF into electrolyte (sample A) and
with HF (sample B). XPS data was obtained at 240 eV photon energy.
The histogram comparing the Si+xrelative distribution in samples A
and B is shown in Fig. 6.8.
0.0
0.3
0.6
0.9
1.2
1.5
1.8
2.1 Sample A
Sample B
Deconvolutedcomponents
0
5
10
15
20
25
30
35
40
45
Si+4 Si+3 Si+2 Si+1
Sicontents(%)
+x
I(Si)/I(Si)
+x 0
Si /Si
+4 0
Si /Si
+3 0
Si /Si
+2 0
Si /Si
+1 0
Si /Si
+all 0
a. b.
Deconvolutedcomponents
Figure 6.8: a. Histogram of Si+x(x=1..4) intensities normalized to
the bulk Si+0 intensity, for sample A (black) and sample B (gray), as
shown in XP spectra in Fig. 6.7. The right–most columns present a
Si+all/Si+0 ratio, where Si+all=Pn
i=0 Si+x. b. Distribution of Si+x
percentage from the total Si+all value.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 91
steric hindrance, not every Si atom is occupied by nitrobenzene molecule, living
a space for Si–OH formation of the top Si atoms and the further process of the
back-bond oxidation.
When no HF is added into the electrolyte, a competition between nitroben-
zene grafting and reaction of the Si surface with the H2O molecules in aqueous
solution presumes silicon oxide island formation, which can be the case for sam-
ple A. Island formation can also proceed for sample B as a result of the above
mentioned post–grafting oxidation during water rinsing. Thus, the actual calcu-
lation of the SiOxthickness is even more complicated if the SiOxisland structure
is taken into account. We employed Eq. 4.19 as described in section 4.2.2 to es-
timate the silicon oxide thickness and coverage, using the XPS spectra obtained
at 650 eV and at 240 eV. To calculate the NB layer thickness we used IRSE
simulations with the parameters from Table 5.3. The nitrobenzene thickness
values were cross-checked using XPS Multiquant program (see section 4.2.2),
where the obtained SiOxthickness and coverage were used as input parame-
ters. A good agreement for the thickness values was found between the IRSE
simulated results and the XPS Multiquant analysis (the difference between the
results obtained from simulations of IRSE spectra and XPS Multiquant analy-
sis lied within 15% of the obtained thickness values). Table 6.3 summarizes our
results.
Sample Thickness Coverage Product NB thickness
dSiOx [nm] θSiOx dSiOxθSiOx [nm] dNB [nm]
Sample A 0.85 0.9 0.77 7
(HF-free)
Sample B 1.3 0.2 0.26 5
(HF-treated)
Table 6.3: Estimated thickness and coverage parameters as obtained from XPS
data. The thickness and coverage of the model were defined in Fig. 4.12 of
section 4.2.2.
The results show that the total SiOxthickness dSiOxθSiOx is about three
times higher for sample A. In addition, the obtained coverage (θSiOx=0.9) for
sample A points out an almost fully formed SiOxinterface. These results sug-
gest that the competition between grafting of the NB radicals and reaction
with the H2O molecules favors the attachment of the OH radicals to Si atoms.
The relatively high contribution of the Si+4 component in the XP spectrum of
sample A in Fig. 6.7 suggests the silicon back–bond oxidation. SiOxcoverage
θSiOx for sample B was 0.2, with thickness dSiOx=1.3 nm. IRSE spectra did
not show any pronounced peak in the SiOxregion for this sample. For compari-
son, H–passivated Si(111) surface oxidized under atmospheric conditions (see in
Fig. 6.6) exhibited a much higher absorption intensity in the IRSE spectra, while
the estimated thickness of SiOxwas 0.6 nm. In case of sample B, the product
dSiOxθSiOx was 0.26 nm, which is about three times smaller than the thickness
value obtained for the above hydrogen–passivated Si stored under atmospheric
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 92
conditions. Probably the low coverage and the low overall thickness dSiOxθSiOx
play an important role in formation of the IRSE spectra. Another possibility
is that irradiation with X–rays triggers an additional oxidation of the sample
through reduction of the nitrobenzene where the release of oxygen can cause the
oxidation of the silicon interface (see chapter 7 for details of the X–rays induced
reduction.) However, as will be seen in chapter 7, the decomposition results in
larger time scales than the time that it takes to obtain a single XP spectrum.
Thus the measured intensities do reflect to some extent the initial state of the
sample as prepared, although the silicon–oxide composition might be somewhat
higher in the X–ray irradiated part of the sample than in the non–irradiated
one which was measured by IRSE. This is due to the photochemical oxidation
processes triggered by X–ray irradiation. An additional source for interface oxi-
dation can be the residual water which may be included in the electrochemically
grafted thin films as was discussed in section 5.2. The oxidation of the interface
can be caused by irradiation of these residual water inclusions [164].
6.4.1 Stability of the organic films on oxidized surfaces to
HF treatment
In this section, we explore the post–grafting oxidation of the interface between
the organic films and the silicon substrate in air and discuss the removal of this
layer in HF solution. Subsequent oxidation of samples A and B was monitored
with IRSE after two months of exposure to atmosphere. The spectra are shown
in Fig. 6.9 by black curves. For sample B, the absorption band due to SiOx
appears in the range between 1050 and 1250 cm−1, which was not seen directly
after the preparation (see Fig. 6.7 for comparison). For sample A, the amplitude
of this band becomes larger in comparison to the as–prepared sample (compare
with the results in Fig. 6.7). These results illustrate that oxidation proceeds
on the NB/Si interface upon exposure to atmospheric conditions on both of the
samples.
In order to investigate the stability of such interfacial SiOxlayer to HF
etch, the samples were etched in 2% HF solution for 20 sec. The results are
shown in Fig. 6.9 by red curve, for both sample A and sample B. While the
absorption bands due to SiOxdisappear from the IRSE spectra, the NB–related
absorption bands remained unchanged. This is a surprising result, since one
would expect that NB would be etched off from the sample with the below–
lying SiOxinterface.
For a better understanding of these results, electrochemical grafting on
chemically oxidized Si surfaces was performed, with a subsequent HF etch. In
Fig. 6.10 the IRSE spectra for the samples as prepared are shown in red and
the HF–treated samples are shown in black. In this figure, ”sample a” refers
to a bare chemically oxidized sample, ”sample b” was grafted during 110 sec
(thickness dNB=3 nm) and ”sample c” was grafted during 550 sec (thickness
dNB=14 nm). Chemical oxidation resulted in about 2 nm of silicon oxide layer,
as obtained from IRSE simulations on sample a. HF treatment resulted in
disappearance of the SiOx–related absorption peak, while NB–related absorp-
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 93
1000 1100 1200 1300 1400 1500 1600 1000 1100 1200 1300 1400 1500 1600
Wavenumber / cm-1
tan y
D()
0
I0.001
I0.1
Sample A SampleB
Oxidized
(2months)
HF-etched
Wavenumber(cm )
-1
(NO)
as
2
n
(NO)
ss
2
n
n(SiO)
x
(NO)
as
2
n
(NO)
ss
2
n
n(SiO)
x
Figure 6.9: IRSE spectra of sample A prepared without addition of HF
into electrolyte (left) and sample B prepared with addition of HF (right)
after 2 months of oxidation under atmospheric conditions (black) and
after a subsequent HF–treatment (red). The SiOx– related absorption
peak disappears in both cases upon HF–treatment, while NB-related
peaks remain unchanged.
tion bands remained unchanged. In addition, appearance of the peak due to
Si–H stretching vibration at 2082 cm−1points out on hydrogen–termination of
the available surface Si atoms. Similar results were obtained for bromobenzene
grafted on chemically oxidized Si surface: HF–etching resulted in disappearance
of the silicon oxide–related absorption peaks, while the bromobenzene absorp-
tion peaks remained nearly unchanged. Similar results were obtained for 1.5
years old samples of NB/Si, where the removal of oxide did not result in the
removal of the organic layer (spectra not shown). HF–etching presumes an
etch–back of the Si atoms, as was schematically shown earlier in Fig. 2.10. The
obtained results however point out that the electrochemically grafted molecules
are not etched together with the removed surface Si atom.
Allongue at al [165] performed similar experiments for bromobenzene grafted
on Si surfaces. They reported however different results: upon HF–dip of a 6–
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 94
I1e-3
I
1e-4
I0.1
I
0.1
a
b
c
a
b
c
a
b
c
a
b
c
tan y
D()
0
(NO)
2
ss
n
n(SiO)
x
as
n
(NO)
2(Si-H)
n
Wavenumber(cm )
-1
950 1150 1350 1550 1900 2050 2200
Figure 6.10: IRSE spectra of (a) chemically oxidized Si(111), (b) NB
grafted during 110 sec and (c) NB grafted during 550 sec on chemically
oxidized Si surfaces. Red: initial spectra after the preparation; black:
spectra obtained after HF etch.
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 95
month old sample, the bromobenzene layer was etched away together with the
underlying silicon oxide layer. They explained their results as follows: the
oxidation starts from the defects in the organic layer and at the step edges of
the Si surface, since the molecular grafting does not take place there [165]. The
oxide starts to grow as islands, followed by a nucleation and eventually the
complete layer formation. Removal of the complete oxide layer also removes
the overlying molecules. Our results above however point out on a different
situation: although the HF removes the oxide, the organic layer stays intact.
One of the possibilities is that attachment of the HF fluorine atom(s) to
a Si surface atom leads to polarization of SiO2back bonds, upon which the
Si atom becomes slightly positive, as shown schematically in Fig. 6.11. The
grafted molecules of interest were nitrobenzene and bromobenzene, molecules
of electron–acceptor type. The slightly positive and slightly negative ends of
the attached molecule are indicated in Fig. 6.11. The positively–charged end
of the phenyl ring may get repulsed from the Si atom, which becomes more
positive upon attachment of fluorine atom(s). The organic molecule can rebind
either to underlying less positive oxygen or silicon atoms. However, theoretical
calculations similar to those performed in ref. [166] for simulations of HF inter-
actions with oxidized Si surface are necessary in order to reveal the validity of
this simple model.
+HF
d+
d-
d-
d+
Xd-
d+
Xd-
HOF Si
Figure 6.11: Schematic drawing of a possible scenario for reattachment
of phenyl ring to slightly more negative atoms on a silicon surface. The
arrow indicate possible re–attachment sites for the phenyl ring.
6.5 Summary
In this chapter we discussed the SiOxinterfaces that form between the organic
films and the Si substrates as side reactions of electrochemical grafting and
under atmospheric conditions. XPS techniques gave an insight into the sub–
oxide structure of the Si surfaces. Using appropriate models, the thickness and
coverage of the SiOxinterface were quantified.
It was shown that in general, addition of HF into electrolyte prevents for-
mation of the oxidized silicon interface during the grafting. This was presented
CHAPTER 6. PASSIVATION AND OXIDATION OF SI SURFACES 96
by the absence of the SiOx–related absorption peaks in the IRSE spectra. How-
ever, XPS spectra included the SiOxpeaks in the spectra obtained from the
samples for which HF was added into the electrolyte. One possibility is that
the post–grafting water rinsing of the samples induces the formation of this in-
terface. However, there is also a possibility that grafting proceeds through the
Si–O–C bonds, explaining the oxide peaks in the XPS measurements.
Another finding of this work was that oxidation under ambient conditions
proceeds more slowly beneath the organic layer than on an unprotected silicon
surface. We found out that under larger molecule (4–MDA), the oxidation is
slowed down considerably, which is probably due to the steric hindrance of the
molecule. This finding can be useful in design of electronic devices for protecting
of silicon against oxidation.
It was also observed that HF etching of the organically modified surfaces
with an underlying SiOxinterface removes the interface, but molecules remain
on the surface. The possibility of the rebinding of the molecules to more negative
surface atoms was discussed. Theoretical calculations are necessary in order to
prove the validity of this simple model.
Chapter 7
X–ray induced reduction of
nitrobenzene to aniline on
Si(111) surfaces
In the previous chapters it was shown that electrochemistry leads to a successful
functionalization of inorganic electrodes with organic thin films. Here, a process
that converts the NO2nitro groups of electrochemically grafted nitrobenzene
on Si surfaces into NH2amino groups is presented. Such conversion of nitro
groups in aromatic compounds into amino species by electrochemical methods
is a well–established procedure and has been reported in refs. [167,168].
Recently, conversion of nitro groups to amino species by external irradia-
tions using hard and soft X-rays [169,170], electrons [171] and even visible laser
light [172] has received considerable attention. This peculiar type of irradia-
tion damage is usually an unwanted effect. Yet, it opens up new interesting
technological perspectives in the case of the nitro-to-amino conversion, such as
employing chemical lithography [171] by focused beams or shadow masks.
Laterally patterned surfaces in the micro- or even nanoscale regime can be
prepared where the surface functional groups, NO2or NH2, determine the lo-
cal surface reactivity. Thereby connectivity to further functional units can be
achieved. Such lateral chemical patterning by external irradiation has been
demonstrated previously [171,172].
A number of studies has dealt with the irradiation–induced nitro–to–amino
conversion, and there is a consensus in the literature that electrons cause the
NO2conversion. In the case of photon irradiation, the electrons generated in
the photoemission process (i.e., photoelectrons and secondary electrons) trig-
ger the conversion processes. As Moon et al. pointed out [170], the reaction
cross-section is too large to be interpreted as a photocleavage reaction under
non-resonant conditions. The details of the mechanism and chemistry of NO2
conversion are still under investigation. Concerning the source of the hydrogen
atoms or protons that are necessary for the conversion of NO2to NH2, different
97
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 98
opinions prevail. Electron induced C-H bond splitting [171] as well as solvent
or water remaining within or adsorbed on the organic layer [172] were quoted
as likely sources of hydrogen. Concerning the effect of (soft) X-ray irradiation,
the selective cleavage and complete desorption of the nitro group [169] instead
of (mainly) chemical conversion [173] were reported.
Other non-consistent findings relate to the oxygen species. Whereas refs. [169,
170,173] did not observe a decrease in the O1s core level for NO2reduction by
X-rays, changes in the O1s core level induced by low-energy electron beam ex-
posure were reported by Eck et al [171].
This chapter addresses those unresolved issues of the irradiation-induced
nitro-to-amino conversion. High-resolution in situ studies of thin nitrobenzene
layers irradiated with light from an undulator of a third generation synchrotron
source at BESSY II was performed. The chemical conversion was monitored
on-line by acquiring high-resolution X-ray photoelectron spectra.
The nitrogen-related emissions were analyzed by a detailed curve fitting,
where the measured data were fitted into up to six individual components.
In particular, shakeup satellites [174] were included in the data analysis and
interpretation.
In order to have an insight into the time–dependent changes in the spectra,
the deconvoluted integrated intensities as function of irradiation time twere
plotted. These data were fitted with an equation of the type A+Bexp(−t/α).
In this formula, Aand Bare constants, where Bcan be either positive or
negative, depending on whether a certain component was growing or decaying,
respectively. The exponential part includes t, which is the irradiation time,
and α, the constant of the decay/growth of a certain component. Here, we
report on the constant αfor some of the processes involved in the modification
of nitrobenzene during irradiation. The accuracy of the decay constant αis
dependent on background subtraction, the quality of the fit, and the amount of
data that was analyzed. Thus, the reported values of the decay constant αhave
an error bar of 25%.
Careful examination of the core level spectra of nitrobenzene monolayers
grafted on the Si substrate allowed us to identify not only the NH2groups that
resulted from the irradiation but a rather complex ensemble of amino-related
species. A considerable oxidation of the supporting silicon substrate by oxygen
released from the nitro groups was observed. Our measurements confirmed that
no decrease in the overall C1s core level was induced by X-ray irradiation, in
agreement with other authors [170, 173]. In the following sections, the results
of simultaneous X-ray irradiation and photoelectron monitoring are described.
The deconvoluted core level emissions and their assignments to the observed
components is presented. Finally, the chemical processes that take place upon
nitro-to-amino conversion is proposed, passing through the intermediate stages.
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 99
7.1 Overview of the X–ray irradiation induced
changes on the observed core level spectra
Fig. 7.1 gives an overview of the irradiation-induced changes of a nitrobenzene
layer grafted onto Si(111). The detailed spectra of the Si2p, C1s, N1s, and O1s
530 533 536 539 0
250
500
750
1000
1250
1500
1750
2000
399 402 405 408 411
0
100
200
300
400
500
600 283 286 289 292 295 0
350
700
1050
1400
1750
2100
2450
2800
98 100 102 104
0
100
200
300
400
500
600
700
Si2p
h =650eVn
C1s
h =650eVn
O1s
h =650eVn
N1s
h =650eVn
Si0
SiOx
-NH2
-NO2
Bindingenergy(eV) Bindingenergy(eV)
Intensity(a.u) Intensity(a.u)
Figure 7.1: Development of the Si2p, C1s, N1s, and O1s core levels upon
irradiation with hν=650 eV. The spectra series were obtained in 2 min
intervals, within 20 min total. The arrows on each graph emphasize the
irradiation-induced changes. Please note the scaling factors.
core levels are shown. They were obtained sequentially within 20 min of irradi-
ation/observation time at an excitation energy of 650 eV. The N1s line exhibits
the NO2–related emission at a binding energy of about 406 eV, which quickly
decays upon irradiation. On the other hand, the amino-related emission at
about 400 eV increased with time. In contrast to earlier studies [173] significant
changes were observed within the O1s core level: the dominating emission at 533
eV decayed in intensity, and the evolution of a distinct shoulder at 531.5 eV indi-
cates the formation of a new chemical species. The irradiation-induced changes
of the C1s emission are comparatively subtle, nonetheless significant. The main
emission at 285 eV is slightly reduced in intensity, which is compensated by an
increase in emission intensity around 288 eV. The overall C1s emission intensity,
however, remains unchanged, as the detailed core level analysis by curve fitting
will confirm. The Si2p emission shows the Si0doublet emission from elemental
bulk silicon at 99.24 eV (Si2p3/2, 0 min). In the course of the irradiation, the Si
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 100
2p emission increases in intensity and shifts continuously toward higher binding
energies. This energetic shift is presumably related to an increase in the surface
band bending (the silicon substrate is p–type doped), which is brought about
by the secondary electron irradiation [175, 176] The Si2p region contains also
emissions from silicon (sub)oxide species SiOx around 103 eV, which apparently
exist at the interface with the electrografted nitrobenzene layer. The irradiation-
induced binding energy shift toward higher energy was detected for both the
bulk Si0and the oxide Si+xcomponents, where Si+xdenotes the (intermediate)
oxidation states with Si bonded to x oxygen atoms.
To obtain a better insight into the silicon/nitrobenzene interface structure,
the Si2p core level was monitored in a separate experiment run on a pristine,
non-irradiated surface spot by excitation with hν=240 eV photon energy. These
spectra were used for the data analysis of the Si2p core level rather than those
obtained with hν=650 eV photon energy due to a higher surface sensitivity.
Figure 7.2 shows Si2p spectra obtained sequentially within 26 min, where the
changes in the Si+0 and Si+xcomponents are emphasized by the arrows. Because
98 100 102 104 106
0
30
60
90
120
150
180
210
Si0
SiO
suboxides
xSiO2
Si2p
h =240eVn
Bindingenergy(eV)
Intensity(a.u)
Figure 7.2: Development of the Si2p core level upon irradiation with
hν=240 eV. The spectra series were obtained in 2 min intervals, within
26 min total. The arrows on each graph emphasize the irradiation
induced changes.
of the higher surface sensitivity, the relative intensity of the SiOxrelated emis-
sions is now much higher. The center of gravity of the (sub)oxide species shifts
from about 102 to 103 eV, which is much higher than the band bending change
(0.3 eV) monitored by the Si0emission. This shift is interpreted as an increase
in the oxidation state (chemical shift). Interfacial silicon suboxides SiOxun-
dergo a transition to fully oxidized silicon oxide SiO2during irradiation. The
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 101
increase of the Si0and SiOxintensities indicates a reduction of the density of
the organic layer. All core level spectra in Figures 7.1 and 7.2 were thoroughly
analyzed by curve fitting. The results of these detailed studies will be presented
in the following sections, in the order N1s, C1s, O1s, and Si 2p.
7.2 Deconvolution of the N1s core level
The N1s core level was fitted using six components, labeled N1 to N6. Fig. 7.3
displays the fits of the first spectrum, describing the chemical state of the pris-
tine as–deposited surface, and after 1200 s of irradiation, representing a largely
converted surface containing mainly amino-related species. The fitting param-
eters for all six components are given in Table 7.1.
398 401 404 407
0
100
200
300
400
500
600
398 401 404 407 0
20
40
60
80
398 401 404 407 0
20
40
60
80
398 401 404 407
0
50
100
150
200
250
300
350
N1s
hn=650eV
d.
N6
N5
N4 N3
N2
N1
c.
b.a.
1200 sec
0 sec
Bindingenergy(eV) Bindingenergy(eV)
Intensity(a.u) Intensity(a.u)
Figure 7.3: N1s core level spectra obtained at a photon energy of 650 eV.
(a and b) Spectrum obtained in the first scan and (c and d) spectrum
after 1200 s exposure to X-rays. (b and d) Enlargement of panels a and
c, respectively. The components labeled N1 to N6 are shown in panel c
and are discussed in the text.
Fig. 7.4 shows the temporal evolution of the integrated intensities of these
components. The component N2 is centered at 406.0 eV and is related to the
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 102
N1s Assignment BE (eV) Gaussian
components width (eV)
N1 π-π∗shake–up [177] 407–407.3 1.1
N2 NO2406.0 1.1
N3 π-π∗shake–up [178] 404.4 1.1
N4 protonated imine N+H [179,180] 402.1–402.6 1.7
or diazonium –N2+H [181]
N5 nitroso (solid) –N=O [182] 401 1.7
or protonated amine N+•H [180]
N6 NH2neutral amine [180,183] 400–399.7 1.7
Table 7.1: Fitting parameters of binding energy (BE) and Gaussian width
FWHM for components N1 to N6 (as defined in Fig. 7.3 (c)). Lorentzian FWHM
was 0.135 eV for all N1s components [184]. Components were allowed to vary
during the fit in the range shown in the table. N4 and N5 components are also
in the range of the N=O–related signal in nitrosobenzene (401.6 eV in solid
nitrosobenzene) [182].
nitrogen in the NO2group of the grafted NB molecules. This peak diminished
substantially during exposure to the X-rays. The asymmetry of the amino-
related emission line at 400 eV necessitated the fitting into the two components
N5 and N6. These components grow during the X-ray exposure time. The com-
ponent N6 is centered around 400.0 eV, but it gradually shifts toward 399.7 eV
during the irradiation. The component N5 was fixed at 401.0 eV during the fits.
The binding energy of N5 is close to the one reported by Kumar et al. [179] for
the positively charged nitrogen species in polyaniline (400.8 eV), as well as to
that of solid nitrosobenzene, H6C5–N=O, which was reported at 401.6 eV [182].
Reduction of nitrobenzene that involves the nitrosobenzene intermediate state
was also reported in ferric oxide suspensions [185] and through electrochemical
reduction [167,168] with the aniline as the final product. The N1 component at
around 407 eV is attributed to the N1s shakeup satellite in nitrobenzene [177].
This is supported by the fact that it decays with a rate similar to that of the
nitrobenzene related component N2 at 406.0 eV. The position of this shakeup
satellite was observed at a lower BE than reported in the literature for gas
phase nitrobenzene [177]. According to Distefano et al. [177], the shakeup band
in the N1s core level has about 2 eV separation from the main N1s peak due to
the NO2group, while in our case, a separation of only about 1 eV was found.
This discrepancy is probably due to the different chemical environment of our
molecules, which are polymerized C6H4NO2units in a solid layer [186,187] in-
stead of isolated gas phase species. Both the N3 and the N4 components show a
more complex behavior: there is an initial increase in their integrated intensities
and a subsequent decrease. The integrated intensities of these components were
very low, and difficulties arose in the background subtraction as well as in the
peak fitting process itself. The N3 component is attributed to a shakeup pro-
cess in aminobenzene (aniline) [178] The N4 component may be related either
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 103
to a positively charged imine species [179, 180] or to –N+
2diazonium residu-
als [181]. It should be mentioned, however, that no diazonium residuals were
detected previously after similar grafting of aryldiazonium salts on various sub-
strates [181,188,189,187]. The decay of the N3 and N4 peaks that occurs after
about 150 s of irradiation might be due either to conversion of these species
further to aminobenzene or to desorption of the nitrogen in these species from
the surface.
7.2.1 Dynamics of the integrated intensities
An overall nitrogen loss from the system was detected during the irradiation,
as can be seen from Fig. 7.4. There, the total N1s plot stands for the sum of
0 7500 750
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 750 0 750 0 750 0 750 0 750
N6
N5
N4
N3
N2
totalN1s
N1
a.
0 750
0.0
0.2
0.4
0.6
0.8
1.0
N6
Extrapolation
b.
Extrapolation
ofN6
Scaling
(normalization)
factors
x120 x80 x30 x800 x80 x1300
Irradiationtime(sec)
Norm.integratedintensity
x720
Figure 7.4: (a) Normalized integrated intensity (peak area) of the N1s
core level components as a function of irradiation duration. (b) Ex-
trapolation of N6 component to zero intensity. Please note the scaling
factors.
the components N1 to N6. Loss of about 20% in nitrogen was observed after
around 450 s of irradiation. After this time, the total N1s intensity seems to
saturate. For comparison, Mendes et al. [173] reported a decrease of 18% after
447 min in aromatic NO2-containing self-assembled monolayers, using a labora-
tory MgKαsource (1253.6 eV). The much faster progress of nitro reduction in
our experiment is due to the orders of magnitude higher photon flux. In con-
trast to La et al [169], the 100% bond cleavage and desorption of NO2groups
upon synchrotron irradiation with hν=550 eV was not observed in this work.
Fitting of the growth and decay constants showed that the growth rate of NH2-
related peak N6 and the decay rate of NO2–related peak N2 were different. It
was found that the decay constant αof the N2 component was about 350 s,
while the growth constant of the N6 component was about 800 s. However,
the decay constant of the sum of components N6 + N5 + N4 +N3, situated at
lower energy to the NO2–related N2 component, was 480 s, which is closer to
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 104
the decay constant αof the N2 (350 s) peak than the decay constant of the N6
peak alone (800 s). This again supports the argument that the NB molecules
decompose to additional species and that the reduction to aniline is not a single-
product process. The remaining gap between the growth constant of the sum
of the components N6 + N5 + N4+ N5 (480 s) and the decay constant of the
NB–related N2 and N1 peaks (350 s) is due to the partial detachment of the
N atoms from the surface. Finally, it should be commented on the emissions
from reduced nitrogen species (N5 and N6) that were already present in the
first monitoring spectra. It is clear from Fig. 7.4 that these emissions do not
result from an insufficient time resolution but exist already on the as–received,
non–irradiated surface. The N1s spectrum was collected first and required less
than 60 s. Extrapolating the N6 emission to zero intensity (Fig. 7.4 (b)) shows
that it would require an irradiation time of about 380 s for these reduced species
to evolve. Thus, it can be concluded that the amino-related species were present
already on a pristine surface and were introduced probably by the amino com-
pounds that serve as the precursors for the aryl diazonium salts from which the
nitrobenzene was electrochemically grafted onto the surface. The interpreta-
tion as a contaminant was supported by ref [186], where amino–type emissions
also had been observed for electrochemically grafted bromobenzene layers. As a
second possibility, the electrochemical grafting process itself may have been ac-
companied by an electrochemical reduction of NO2as a side reaction, as pointed
out by Allongue et al. [190].
7.3 Deconvolution of the C1s core level
The C1s core level was fitted using six components (Fig. 7.5). A summary of
fitting parameters for all six components is shown in Table 7.2.
The main peak C1 was initially centered at 285.1 eV and shifted gradually
toward a lower binding energy at longer exposure time to X-rays. The energetic
separation of component C2 from C1 was 1.2 eV (at 0 s) and was found to
gradually diminish to about 1.0 eV. These two main peaks are caused by the
carbon atoms in the benzene ring [191, 192] where the C1 peak at the lowest
energy is due to neutral C-C or C-H bonds and the C2 peak is due to the carbon
on the benzene ring bonded to the nitrogen atom. The growing C2 peak has
roughly a similar constant α(200 s) as compared to the decay constant of the
C1 peak (170 s). The binding energy of the carbon bonded to a N atom in gas
phase aniline was reported to be 291.29 eV and 292.09 eV in gas phase nitroben-
zene [192]. Thus, the difference in the binding energy of carbon nonbonded to N
and bonded to N is about 1.3 eV in aniline and about 0.95 eV in nitrobenzene.
An analogue asymmetry in the C1s core level was also reported for aniline ad-
sorbed on Ag(110), [196] where the separation in the binding energy for carbon
nonbonded to N and bonded to N was 1.0 eV. Consequently, the increasing C2
component separated by about 1 eV from the main C1 peak is attributed to
carbon atoms bonded to nitrogen atoms in either nitrobenzene or aniline-like
species. An increase in intensity of the C2 component as a function of the irra-
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 105
283 287 291 295
0
450
900
1350
1800
2250
2700
283 287 291 295 0
150
300
450
600
750
283 287 291 295
0
450
900
1350
1800
2250
2700
283 287 291 295 0
150
300
450
600
750
a. C1s
hn=650 eV
d.
1200 sec
b.
C5 C6C4
C3
C2
C1
c.
0 sec
Bindingenergy(eV) Bindingenergy(eV)
Intensity(a.u) Intensity(a.u)
Figure 7.5: C1s core level spectra obtained at 650 eV photon energy. (a
and b) Spectrum obtained during the initial irradiation and (c and d)
spectrum after 1200 s. exposure to X–rays. (b and d) Enlargement of
panels a and c, respectively. Components C1 to C6 are shown in panel
c and are discussed in the text.
diation time thus suggests an increase in the number of the ring carbon atoms
bonded to nitrogen (i.e., through polymerization, [197, 198, 199]. The gradual
shift of the C1 component toward lower binding energies can be explained as a
gradual reduction of nitrobenzene to aminecontaining species. For instance, the
average observed energy for the benzene carbon atoms nonbonded to N atoms
was reported to be 290 eV in aniline, while in NB, this value was 291.1 eV in
the gas phase [192]; thus, the shift of the peak due to carbons at the benzene
ring toward lower energies was expected upon reduction of nitrobenzene. The
C3 component was separated from the C1 peak by 2.8 eV and is possibly due
to bonding to positively charged amine groups [179]. C4 is separated from C1
by about 4.9 eV and is attributed to a shakeup π–π∗satellite characteristic
of aniline [178, 194]. The growth constant αof the C4 component (1770 s) is
close to the growth constant of the sum of amino-related components N5 + N6
(1600 s), supporting assignment of the C4 component as related to the shakeup
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 106
C1s Assignment BE (eV) Gaussian
components width (eV)
C1 carbon in benzene ring 285.1–284.9 1.16
nonbonded to nitrogen
[180,187,191,192]
C2 carbon in benzene ring bonded 286.3–285.9 1.1–1.43
to nitrogen [180,187,192]
C3 carbon in benzene ring bonded 287.9–287.6 2.2
to NH+[179,193]
C4 π–π∗shake–up satellite 290.3–289.7 1.6
in aniline [194]
C5 π–π∗shake–up satellite 291.5–291.3 1.8
in benzene [194,195]
C6 π–π∗shake–up satellite 293.1–293.3 1.8
in benzene [194,195]
Table 7.2: Fitting parameters of binding energy (BE) and Gaussian width
FWHM for components C1 to C6 (as defined in Fig. 7.5 (c)). Lorentzian FWHM
was 0.125 eV for all C1s components [184]. Components were allowed to vary
during the fit in the range shown in the table.
process at the benzene rings bonded to the amino groups. The components C5
and C6 are related to shakeup π–π∗transitions at the benzene ring, which are
characteristic for benzene [195] and were also observed in both nitrobenzene and
aniline [194]. The occurrence of these peaks is thus indicative of the integrity
of the benzene ring during the investigation.
7.3.1 Dynamics of the integrated intensities
Fig. 7.6 shows a plot of the integrated intensities of the C1s deconvoluted emis-
sion peaks. The sum of the C1s components (total C1s in Fig. 7.6 ) shows
no significant change as a function of irradiation time. Thus, no desorption of
phenyl units or smaller carbonaceous fragments from the surface was observed.
The sum of the C5 and C6 components, related to π-π∗shakeup satellites of the
benzene ring, also shows no change in intensity: this indicates that the aromatic
rings remain intact during the irradiation. The irradiation affects a change of
the line shape of the C1s emission: a reduction of the C1 intensity and growth
of the C2, C3, and C4 peaks during the irradiation process were found.
7.4 Deconvolution of the O1s core level
The O1s core level was fitted using four components (Fig. 7.7). A summary of
the fitting parameters for all four components is shown in Table 7.3.
The main peak O2 was centered at 533.0 eV and is related to oxygen from the
nitro group of the NB molecule [187]. The O4 component is related to a shakeup
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 107
0 7000 700
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00
0 700 0 700 0 700 0 700
C1 C2 C3 C4 C5+C6 totalC1s
Scaling
(normalization)
factors
x3800 x1420 x630 x150 x300
Irradiationtime(sec)
Norm.integratedintensity
x5520
Figure 7.6: C1s normalized integrated intensity plots as a function of
irradiation duration. The sum of C5 and C6 corresponds to integrated
intensities of the shakeup satellite related components. Total C1s is the
sum of all C1s integrated intensities in fitted peaks.
O1s Assignment BE (eV) Gaussian
components width (eV)
O1 LBC (SiO2) Si–O–Si [200,201,202] 531.4–531.7 0.68–1.32
O2 C6H5NO2[187] 533.0 1.32
O3 nitrosobenzene N=O [203] 534–534.1 1.37–1.42
O4 π–π∗(C6H5NO2) [177] 535.5–536 1.2
Table 7.3: Fitting parameters of binding energy (BE) and Gaussian width
FWHM for components O1 to O4 (as defined in Fig. 7.7 (c)). Lorentzian width
was 0.2 eV for all O1s components [204]. Binding energies of the components
varied during the fit in the range shown in the table. Possible contribution
from nitroso groups N=O would be within the range of the components O2
and O3 (532.6-534.1 eV in solid nitrosobenzene [182,203]). HBC component of
interfacial SiOxwould be expected within the O2 component [200,201].
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 108
528 530 532 534 536 538
0
400
800
1200
1600
2000
528 530 532 534 536 538
0
400
800
1200
1600
2000
528 530 532 534 536 538
0
150
300
450
600
750
900
528 530 532 534 536 538
0
150
300
450
600
750
900
a. 0 sec
O1s
hn=650eV
Intensity (a.u)
O4
O3
O2
O1
c. 1200 sec
Binding energy (eV)
b.
d.
Figure 7.7: O1s core level spectra obtained at 650 eV photon energy.
(a and b) Spectrum obtained during the initial irradiation and (c and
d) spectrum after 1200 s exposure to X-rays. (b and d) Enlargement of
panels a and c, respectively. Components O1 to O4 are shown in panel
c and are discussed in the text.
process in the NB molecule [177]. The component O1 is generally related as a
lower binding energy component (LBC [200]) and is characteristic for fully Si
coordinated, Si-O-Si bridge bonded oxygen on oxidized Si surfaces [200, 205,
201, 202]. In the initial stages of irradiation, this peak was fitted with 1.58
eV separation from the main peak, but for fitting of the successive data, it
was positioned at about 1.32 eV separation from the main peak. Initially, the
Gaussian FWHM of the O1 component was 0.68 eV, but already after 150 s
of irradiation, the FWHM value was 1.25 eV and increased gradually at longer
exposure to X–ray irradiation. The component O3 shifted gradually toward
a lower binding energy as the exposure to X-ray irradiation continued. The
Gaussian FWHM was kept constant at 1.42 eV, except in the data from the
initial exposure, where the best fit was achieved for a FWHM of 1.37 eV. The
assignment of O3 is not straightforward: on one hand, O3 could be related to
the higher binding energy component (HBC [200]) of surface silicon oxide and
is attributed to oxygen in Si–O configurations [200,205,201,202]. However, the
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 109
observed O1-O3 separation was around 2.4 eV, whereas the HBC-LBC energy
separation was consistently reported to be 1.4 eV [205, 201, 202]. Thus, we do
not assign O3 to a silicon oxide or hydroxide species. A possible assignment
for the O3 component could be intermediate nitrobenzene reduction states such
as nitrosobenzene. The BE values for the O1s emission of aromatic nitroso
compounds reported in the literature vary considerably from 532.6 [182] to 534.1
eV [203]. Further support for this assignment will be given in section 7.6.
7.4.1 Dynamics of the integrated intensities
Fig 7.8 shows the integrated intensities for the O1s components. From Fig. 7.8
0 55011000 5501100 0 55011000 55011000 5501100
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
O1 O2 O3 O4 totalO1s
Irradiationtime(sec)
Norm.integratedintensity
Scaling
(normalization)
factors
x620 x3360 x520 x230 x3900
Figure 7.8: Normalized integrated intensity plot of the O1s emission
components as a function of irradiation duration. O1, O2, O3, and O4
components are as shown in Fig. 7.7 (c). Please note the scaling factors.
, it is evident that there is an overall loss of oxygen from the surface (total
O1s). This oxygen is released from the NO2groups of the NB molecules and
mostly desorbs into the vacuum during the irradiation process. Part of the
oxygen that leaves the NB molecules, however, participates in an oxidation
of the Si substrate since the O1 component that is related to LBC on the Si
substrate increases considerably. The growth constant αof the O1 and O3 peaks
is similar and equal to about 400 s. This constant also correlates well with the
decay constant of the NO2– related O2 component (450 s), implying that upon
irradiation–induced cleavage from the NO2group, a part of the released oxygen
incorporated into the interfacial SiOxlayer. The O4 component, related to the
shakeup process in the NB molecule, decreases with the same constant α(450
s) as the main NB–related O2 component, supporting the assignments of both
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 110
the O4 and the O2 components as related to the NO2group in nitrobenzene.
Remnants of the solvent (i.e., water) incorporated into the solid nitrobenzene
layer would exhibit an emission around 533 eV [123,124,125] (i.e., they would
overlap with the NO2–related emission O2). The parallel decay of O2and the
NO2shakeup structure O4 indicates, however, that the contribution of H2O to
the total O2 signal must be uninfluenced by irradiation. For comparison, the
observed decay constant αfor the NO2-related N2 component was 350 s. These
decay constants appear to be close within the error bar of α. These results
suggest that the O2 and N2 components are correlated and that both result
from the decomposition of the NO2group in the NB molecules.
7.5 Deconvolution of the Si2p core level and dy-
namics of the integrated intensities
The Si2p core level was fitted using five components (Fig. 7.9). A summary of
fitting parameters for all five components is given in Table 7.4. The silicon
Si2p BE (eV) Gaussian
components width (eV)
Si099.24–99.41 0.37–0.38
Si+1 Si0+ 0.97 0.8
Si+2 Si0+ 2.1 0.8
Si+3 Si0+ 2.7 0.8
Si+4 Si0+ 3.5 1.1
Table 7.4: Fitting parameters of binding energy (BE) and Gaussian width
FWHM for components Si+0 to Si+4 (as defined in Fig. 7.9 (c)). Lorentzian
width was 0.055 eV for all Si2p components [74]. For the Si2p core level, spin-
orbit splitting was fixed at 0.605 eV, and the branching ratio was set to 0.52.
Components were allowed to vary during the fit in the range shown in the ta-
ble. For assignments and fitting, parameters similar to those reported earlier in
refs. [90,206,207] were used.
bulk peak Si0shifted toward a higher binding energy during the irradiation, from
99.24 to 99.41 eV. The other components related to Si atoms in (intermediate)
oxidation states (Si+1 to Si+4 components) were fixed at certain relative energies
from the Si0component as shown in Table 7.4. Fig. 7.10 shows integrated
intensities as a function of irradiation time for the Si0to Si+4 components, as
well as the sum of the Si+1, Si+2, Si+3, and Si+4 components.
There is a slow growth of the Si0component, which is due to the continuous
decrease of the density of the organic layer through nitrogen and oxide desorp-
tion. Taking an electron inelastic mean free path of 0.9 nm for the Si2p photo-
electrons in organic matter [99], an equivalent effective thickness decrease of 0.3
nm over the total irradiation time can be estimated. There is a slight increase
in the amount of the Si+1 intermediate oxidation state. The Si+2 intermediate
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 111
100 102 104
0
30
60
90
120
150
180
100 102 104 0
20
40
60
80
100
100 102 104
0
30
60
90
120
150
180
100 102 104 0
20
40
60
80
100
a. Si2p
hn=240eV
0sec
Si+4
Si+3
Si+2
Si+1
Si0
c.
b.
1440 sec d.
Bindingenergy(eV) Bindingenergy(eV)
Intensity(a.u) Intensity(a.u)
Figure 7.9: Si2p core level spectra obtained at 240 eV photon energy.
(a and b) Spectrum obtained during the initial irradiation and (c and
d) spectrum after 1440 s exposure to X–rays. (b and d) Enlargement
of panels a and c, respectively. Components Si0, Si+1, Si+2, Si+3, and
Si+4 are shown in panel c and are discussed in the text.
state is an energetically unfavorable state for the oxidation of Si(111) [208], and
it decays substantially with increasing irradiation time. At the same time, the
Si+3 component is initially growing, but after about 480 s of irradiation, the
intensity of this component decreases. The Si+4 component is continuously in-
creasing as a function of the irradiation time. The growth constant of the sum
of all oxidation states Si+x(total Si+xin Fig. 7.10) is close to that of the O1s
LBC component (about 400 s). Taking all the individual changes together, the
line shape change of the Si 2p emission is to be described as a transition from
intermediate oxidation states (suboxides) to the fully oxidized state, SiO2. As
the intensity ratio between the (partially) oxidized (total Si+x) and the bulk Si0
emission does not significantly change, the thickness of the interfacial (sub)oxide
layer remains relatively constant. Its effective thickness was approximately 0.3
nm.
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 112
0 7500 750
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 750 0 750 0 750
0 750
Total Si+x
Si+0 Si+1 Si+2 Si+3 Si+4 totalSi+x
Scaling
(normalization)
factors
x150 x30 x50 x120 x370
Irradiationtime(sec)
Norm.integratedintensity
x60
Figure 7.10: Normalized integrated intensities of the silicon components
as a function of irradiation time. Data are normalized to the highest
integrated intensity in each data set. Please note the scaling factors.
7.6 Discussion
A careful fitting of the X-ray photoelectron spectra allowed us to follow in detail
the complicated process of nitrobenzene reduction upon X-ray irradiation. From
the N1s core level, we could conclude that during the irradiation not only a chem-
ical reduction takes place but also a partial desorption of nitrogen-containing
species. The appearance of the N5, N4, and N3 peaks, which are positioned at
higher binding energies than the uncharged aniline-related N6, proposes that
several chemical species with a higher local charge at the nitrogen core than in
a neutral amine–like environment have been created. The protonated or posi-
tively charged amino groups –NH+–Ph can be generated either in a molecular
aminobenzene–like or a polymerized polyaniline-like state.
Development of the nitroso groups during this process can be concluded from
cross-referencing the studies from N1s and O1s core levels. We tentatively assign
the component O3 to nitrosobenzene or hydroxylamino–like environments. The
similarity of the irradiation–induced nitrobenzene reduction to the electrochem-
ical reduction process as illustrated in Fig. 7.11 suggests that these intermediate
states might be formed. The binding energies of N1s and O1s for aromatic ni-
troso compounds reported in the literature exhibit a considerable scatter, but
a trustworthy reference experiment was reported by Batich and Donald [203],
which reported 400.3 eV for the N1s and 534.1 eV for the respective O1s emis-
sion. Thus, identifying O3 with emission from –N=O or –NH–OH environments,
the corresponding N1s signal would overlap with N6 (or N5). Identifying O2
with –NO2, and O3 with –N=O or NHOH only, it can be estimated that 10–20%
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 113
nitro
NO2N
O
NH
OH
NH2
a b c d
OH NX2
NX2
N
nitroso hydroxylamino amino
HNX2
NX2
N
ePossiblepolymerizedspecies
nitro
NO2
NO2N
O
N
O
NH
OH
NH
OH
NH2
NH2
a b c d
OH NX2
NX2
N
nitroso hydroxylamino amino
HNX2
NX2
N
HNX2
NX2
N
ePossiblepolymerizedspecies
Figure 7.11: Comparative representation of the electrochemical reduc-
tion series from nitrobenzene (a) to aminobenzene (d) with intermediate
states of nitrosobenzene (b) and phenylhydroxylamine (c) and sugges-
tions for possible polymerized species (e).
of the initially present NO2groups was converted into the intermediate nitroso–
or hydroxylamino–like environment after 1200 s irradiation.
Cross-referencing the Si2p and O1s core level data, we conclude that upon
irradiation-induced cleavage from the NO2group, a part of the released oxygen
is incorporated into the interfacial SiOxlayer. As a consequence, the initially
prevailing suboxide species Si+1, Si+2, and Si+3 are further oxidized, and Si+4
(SiO2) becomes the dominating chemical species. Fitting of the C1s core level
indicated that the benzene rings themselves do not detach from the surface
and do not decompose during irradiation. An increase in intensity of the C2
component, attributed to carbon atoms bonded to nitrogen atoms in either ni-
trobenzene or aniline-like species, suggests an increase in the number of the ring
carbon atoms bonded to nitrogen. We interpret this increase as a polymerization
process of benzene rings through nitrogen-containing bridges.
The reduction and polymerization processes are illustrated in Fig. 7.11 and
were derived in analogy to the electrochemical reduction process (Fig. 7.11 a-
d) [167, 168]. Intermediate states of nitrosobenzene (H5C6–N=O) and phenyl-
hydroxylamine (H5C6-NH-OH) are formed during the electrochemical reduction
process of nitrobenzene to aminobenzene (aniline) [167,168,185]. In the electro-
chemical series, this process is achieved via the supply of electrons and protons.
During the irradiation-induced reduction of nitrobenzene, electrons are supplied
via the secondary electron current through the nitrobenzene thin film; however,
the source of hydrogen to convert from nitrobenzene to aminobenzene is ob-
scure. In other studies, the electron–induced hydrogen–carbon bond splitting
was cited as the origin of the hydrogen for the nitro–to–amino conversion. Sim-
CHAPTER 7. X–RAY INDUCED REDUCTION OF NITROBENZENE 114
ilarly, we believe that it is hydrogen from the phenyl rings that is the source of
the amine-like coordination of the reduced nitrogen species; however, we suggest
an electron–induced polymerization process as depicted in Fig. 7.11e rather than
a three-step process like C–H bond splitting followed by –N–O bond splitting
and –H capture for the N–H bond formation
7.7 Summary
In this work, we investigated the effects of synchrotron irradiation on a ni-
trobenzene film grafted electrochemically on a Si(111) surface. In agreement
with earlier reports, we observe an irradiation-induced reduction of nitroben-
zene to an aminobenzene–like compound. This process can be employed for the
purpose of chemical lithography.
A detailed study of the chemical conversion process by acquiring time–
dependent spectra, by careful fitting of the high–resolution spectra, and by
calculation of the related decay and growth constants was presented. The pho-
ton irradiation-induced secondary electron current modifies the surface layer by
a reduction of nitrobenzene. The irradiation-induced reduction of nitrobenzene
is accompanied by a partial desorption (about 25%) of nitro groups. Oxygen
atoms that were freed from the remaining nitro groups were observed to oxi-
dize the silicon substrate and to convert the initially present suboxide into fully
oxidized SiO2.
The irradiation-induced reduction of NB is not a single-product process,
as a variety of reduced nitrogen species was observed. The main components
are neutral and positively charged protonated amine groups. Indications of
a polymerization of the benzene ring were found, and generally, the spectral
features comply with those reported in the literature for polyaniline [179].
Intermediate reduction states in a nitroso or hydroxylamino-like environment
were observed. The polymerization through attachment of the nitrogen to the
benzene rings [197, 198, 199] explains the hydrogen source for the creation of
the amine-like environment. There was no evidence for participation of water
solvent remnants in the chemical conversion process as claimed by others. The
carbon density remained constant during the irradiation-induced NB reduction.
Benzene rings did not detach or decompose during the X-ray irradiation, as
indicated by the persistence of the benzene-ring-related π-π∗shakeup satellites.
To conclude, detailed studies of irradiation induced damage of organic ni-
trobenzene layer on a Si(111) substrate were performed. We proposed a re-
duction scheme of nitrobenzene to amino-related species. Understanding of
this process will allow for further developments in chemical lithography and
micropatterning of hybrid organic/inorganic surfaces.
Chapter 8
Concluding remarks
In this work, structural characterization of organically modified metallic and
semiconducting surfaces and interfaces was carried out. The characterization
was performed using surface sensitive techniques, such as infrared spectro-
scopic ellipsometry (IRSE), atomic force microscopy (AFM), X-ray photoelec-
tron spectroscopy (XPS) and VIS-ellipsometry. Cross correlated studies em-
ploying those experimental methods allowed to determine the influences of the
electrochemical preparation conditions on thin films and interfaces properties,
and to improve them accordingly. Our investigations revealed a successful mod-
ification of surfaces with nitrobenzene, bromobenzene, methoxybenzene and 4–
methoxydiphenyl amine thin films through the electrochemical reduction of ben-
zenediazonium salts. It was shown that grafting from the Grignard compounds
on porous silicon led to termination with CH3and CD3groups.
Spectra obtained using IRSE techniques enabled a direct observation of sur-
face chemical composition at a sub–monolayer regime. For instance, it en-
abled to observe the degradation of the hydrogen passivated silicon surfaces
(section 6.1), to explore the structural properties of grafted organic thin films
(sections 5.2, 5.3 and 5.4) and to study the interfacial properties of the sam-
ples (sections 6.2 and 6.4). Combination with other techniques, such as XPS
and VIS–ellipsometry was important for independent determination of the thick-
nesses and optical properties of thin organic layers. Based on these observations,
conclusions on side reactions during the preparation, passivation behavior, sta-
bility and structure of the organic thin films were drawn.
Cross correlated analysis using IRSE and XPS enabled to quantify also the
properties of the silicon oxide interfaces. In particular, XPS studies provided
the discrimination between the silicon sub–oxides (Si0.. Si+4). Using appropri-
ate models, the SiOxcoverage and thickness in differently prepared surfaces, i.e.
with and without HF in the electrolyte were quantified. The main finding was
that an almost complete SiOxlayer forms at the silicon/organic film interface
in the electrolyte when HF was not added to the electrolyte, while addition of
HF was found to generally prevent the interface oxidation. However, when HF
was added to the electrolyte, the XPS results pointed out a low coverage of the
115
CHAPTER 8. CONCLUDING REMARKS 116
interfacial SiOx, while no absorption bands due to SiOxwere observed in the
IRSE spectra. This small discrepancy between the XPS and IRSE results was
argumented by additional possible oxidation processes that may take place dur-
ing the XPS measurements. First, XPS triggers the reduction of nitrobenzene
upon which the released oxygen can participate in the interface oxidation. Sec-
ond, the oxidation could be triggered through the residual water in the organic
layer from the grafting in aqueous solution when irradiated by X–rays.
Another finding of this work was a peculiar behavior of the organically mod-
ified SiO2/Si surfaces when subjected to HF etching. While HF removed the
SiO2layer, as supported by IRSE results, the organic layer was not removed
together with the SiO2interface on which it was grafted. This behavior was
reasoned by HF etching kinetics, upon which we proposed that the removal
of the SiO2was slower than the rebinding of the organic molecules to other
(slightly positive) sites on the silicon surface. For a detailed understanding of
this process further theoretical calculations are necessary.
Organic functionalization allows tailoring of the surface and interface prop-
erties, and thus it is becoming of high importance in research and industry.
Tuning of the surface properties, such as electron affinity, wettability, pH sen-
sitivity and so forth can be enabled by a change of the functional group of
the attached organic material. In this work it was shown that oxidation of the
underlying silicon surface can be slowed down through organic modification.
We proposed that the size of the molecule plays an important role in oxida-
tion kinetics, where the 4–methoxydiphenyl amine (C6H5–NH–C6H5–OCH3),
composed of two phenyl rings, slows down the interface oxidation more than
the nitrobenzene (C6H5NO2), composed of one benzene ring. These results are
important, for instance, in photovoltaic technology where the stable, well passi-
vated interfaces are crucial for the control of the charge recombination centers.
Biological compatibility of the surfaces can be realized through the bio–
reactive functional groups such as NH2. One of the aims of this work was to
study in detail the reduction of nitrobenzene (C6H5NO2) to aniline (C6H5NH2)
on silicon surface upon irradiation with X-rays. The components of the re-
duction process were proposed upon a detailed deconvolution of the observed
core levels. It was found that the irradiation induced the reduction of the NO2
to a NH2functional group, leaving the benzene rings intact. Additionally, we
proposed that polymerization of the benzene units takes place as a result of
the X–ray irradiation. The procedure of lithographic patterning through X–
ray irradiation can be employed in fabrication of biosensor arrays for biological
analysis and diagnostics.
While this work provided a wealth of information useful for understanding
and improvement of the electrochemical processes, several open questions still
remain unresolved. Many of the difficulties are related to the detectivity limits of
the experimental methods. For instance, our interpretation of the IRSE spectra
was based on the relatively strong infrared active vibrational bands. However,
weaker bands which were in the noise level could have supplied additional in-
formation on the film structure and bondings. One such problem was related
to the proof of the formation of covalent Si–C (or Si–O–C) bonds between the
CHAPTER 8. CONCLUDING REMARKS 117
organic films and the silicon substrates. Since the expected amplitude of such
absorption band is relatively week, additional techniques may be required for
confirmation of this covalent attachment. In this work, expansion of the IRSE
technique to lower frequencies was performed for the direct proof of the Si–C
or (Si–O–C) formation. Further improvement of the signal–to–noise ratio is
still necessary. This can be achieved, for example, by surface-enhanced infrared
absorption (SEIRA) spectroscopy, where evaporation of discontinuous metallic
films on the organically modified surfaces results in enhanced absorption bands.
Additional open issues are related to the lack of optical data on the studied
organic materials. This is the situation in the research of every new material,
and the related problems may be solved by cross correlated analysis employ-
ing several surface sensitive techniques. Such cross–correlated approach was
the focus of this work, where IRSE, XPS and VIS–ellipsometric studies were
performed for the determination of the optical constants of the thin films. How-
ever, improvement of those techniques and development of new surface–sensitive
methods are necessary for the analysis of organic/inorganic surfaces and inter-
faces.
Those open questions of this work introduce new interesting research possi-
bilities. Directions for future research are diverse. On one hand, it includes the
development of the well controlled, technologically relevant deposition methods
for engineering of new hybrid organic/inorganic devices. On the other hand, it
demands improvement of the surface sensitive techniques. In parallel, expansion
of the theoretical background should be performed for the interpretation of the
experimentally obtained results.
Appendix A
Simulations of the IRSE
spectra
In this appendix we discuss simulations of the IRSE spectra, and show the
possibilities and limitations of the technique for the determination of the high–
frequency refractive index and thickness of the measured thin films. Advantages
of the multiple-angle measurements will be demonstrated.
A.1 Spectroscopic properties of thin films
Fig. A.1 shows plots of the reflectance Rp, Rs and ellipsometric parameters
tan ψand ∆ for films of 5 nm, 50 nm and 500 nm and 5000 nm thickness. While
multiple minima and maxima occur across the whole spectral range for thicker
films (Fig. A.1), no such minima and maxima occur for the thinner films in the
MIR spectral range. The extrema are at higher photon energies, which would be
more sensitive for thickness determination of thinner films. Fig. A.1 implies that
for investigations of the film thickness, the choice of the spectral range might
be critical. In the next sections we present a careful theoretical analysis of the
parameters that influence the determination of the high-frequency refractive
index and film thickness in ellipsometric investigations.
In the next sections we are going to perform theoretical calculations in order
to demonstrate the results for ”model” case, which does not take into account
experimental limitations or frequency–dependent dispersion. The calculations
of thin film of a high-frequency refractive index of 1.46 will be assumed. We
will discuss two types of substrates: the one with a refractive index n2=3.48
and the other representing a metallic substrate with n2=3.48+31.7i. These
values were chosen with the same real part in the refractive index in order
to perform comparison between the two cases. Moreover, the value n2=3.48
represents well the properties of a Si substrate in the MIR spectral range, while
the value n2=3.48+31.7i is representative for Au substrate at 2000 cm−1[209].
The simplified dispersion–free approach will be justified in section A.2 for the
118
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 119
500 2500500 2500 500 2500 500 2500
500 2500500 2500 500 2500 500 2500
x1 x7 x1000 x100000
500 2500500 2500 500 2500 500 2500
x1 x7 x22.5 x225
x1 x1 x107 x10700
500 2500500 2500 500 2500 500 2500
x1 x1 x4500x45
Wavenumber(cm)
-1
5000
nm
500
nm
50
nm
5
nm
5000
nm
500
nm
50
nm
5
nm
tan (a.u)y
D(a.u.)
Rp(a.u) Rs(a.u)
Figure A.1: Calculated ellipsometric parameters tan ψand ∆ and re-
flectance Rp, Rs for films of 5000 nm, 500 nm and 50 nm and 5 nm of a
refractive index 1.46 on a substrate of n∞=3.48 (typical for Si), at the
angle of incidence 650in MIR spectral range. The numbers on the plots
indicate zoom factors relatively to the data shown for 5000 nm film.
purpose of the present discussion only. Otherwise, frequency–dependent values
of the optical constants of substrates and thin films were taken into account
throughout the rest of this thesis.
A.2 Best-fit calculations
For best-fit calculations, the least-squares routine is typically applied [210]. In
this work it was done by fitting of the measured ellipsometric parameters tan
ψand/or ∆ using models described in chapter 3. Typically for evaluation of
film’s thickness d and its refractive index n, the fit on ellipsometric parameter
∆ is performed, since the parameter ∆ is more sensitive for thickness and re-
fractive index of a film. This is demonstrated in Fig. A.2, where the results of
simulations of a thin nitrobenzene film on Au substrate are shown. The spectra
shown in gray were calculated using constant refractive index n=1.46 in the
spectral range of interest. For spectra shown in black dispersion model based
on the Lorentz oscillators was implemented (for oscillator parameters employed
in this simulation see section 5.2). The major differences between the spectra
simulated employing the dispersion and those for which the refractive index n
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 120
was wavelength–independent was in the absorption region of the NB molecule,
between 1300 and 1620 cm−1. Outside this spectral range, the spectra carry a
nearly identical character. Fig. A.2 shows that the background, the band am-
plitudes and the slope of the ∆ change in response to the thickness variation.
The difference in the slope of the ∆ is for instance 20 % between the 5 nm and
15 nm organic films. Both ellipsometric parameters tan ψand ∆ exhibited a
change in the absorption bands amplitude for spectra for calculations of which
the dispersion model was implemented.
When the absorption constants are known, they can be used in Lorentz
oscillator model for independent thickness evaluation. However, it is not the
case for many organic materials. Consequently, the best fit is typically applied to
the ∆ spectrum in the spectral range where contributions of vibrational features
are negligibly small. During such best–fit procedure, film thickness d and film
refractive index n are varied until the best fit is achieved. In our case, the fit of
Wavenumber(cm )
-1
1100 1250 1400 1550 1700 1850
tan y
D()
0
I0.001
I0.5
5nm
10nm
15nm
5nm
10nm
15nm
dispersion-free
5nm,10nmand15nm
dispersionincluded
dispersionfree
dispersion
included
Figure A.2: Simulated ellipsometric parameters tan ψand ∆ for ni-
trobenzene thin films with 5 nm, 10 nm and 15 nm film thicknesses
on Au–coated glass substrate at 700incident angle. Black: simulations
including the dispersion due to NB absorption bands; gray: no dis-
persion was assumed. Simulations were performed using models from
chapter 3 and optical constants from section 5.2. In these simulations
the measured gold–coated substrate optical constants were employed.
the ellipsometric parameters is based on Airy reflection coefficients rA(Eq. 3.35,
chapter 3). Eq. 3.35 however implies a complex dependency of rAcoefficients
(and thus the ellipsometric parameters tan ψand ∆) on a thickness d and a
refractive index n of a film. This dependency enables several {n,d}solutions
for a single measured curve. It is thus useful to investigate these dependencies
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 121
in detail. We perform a theoretical analysis of the n,d film solutions using
the following procedure: Firstly, a theoretical spectrum of the ellipsometric
parameter ∆ini is calculated at a certain angle of incidence. For convenience,
we take nini=1.46 of the thin film (representing a typical organic polymer film,
for example), and n2=3.48, simulating the refractive index of a Si substrate.
Next, we fit the ∆ini using the least–square trust region dogleg routine [211]
looking for minimum of Pi(∆ini(wi)−∆fit(wi))2, where ∆fit is calculated
through variation of the high-frequency refractive index nfit and thickness dfit
of a film, where fit stands for fitted values (to distinguish them from the initial
values at which ∆ini was calculated). Ideally, a single solution for nfit matching
nini and for dfit matching dini should be found.
Fig. A.3 shows the results of the calculations for 5 nm thin film at 650
incident angle in the infrared spectral range between 700 and 4000 cm−1. At
a single angle, the minimum of Pi(∆ini(wi)−∆fit(wi))2is achieved, however,
for a set of pair values nand d, which is situated along the minima of the 3–D
chart shown on left panel in Fig. A.3. The right panel of Fig. A.3 shows the
actual solution by the crossing white lines, which falls within the space of the
all other possible pair values nand dgiving the solution for ∆ini.
S D w - D w( ( ) ( ))
ini fit
2
iii
(a.u.)
d[nm]
n
d[nm]
n
S D w - D w( ( ) ( ))
ini fit
2
iii
(a.u.)
Figure A.3: The charts of Pi(∆ini(wi)−∆fit(wi))2as a function of the
film high frequency refractive index nand thickness d. Left: a 3D chart
view; Right: same as left panel, chart view from above. The set of the
{n,d}solutions for ∆ini is situated along the minima of the left chart
and along the dark shaded area in the right chart. The intersection of
the white lines shows the actual parameters of nini=1.46 and dini=5
nm at which ∆ini was calculated. Calculations were performed in the
infrared spectral range between 700 and 4000 cm−1at 650incident
angle. .
Fig. A.4 shows the plots of Pi(∆ini(wi)−∆fit(wi))2as a function of vary-
ing nand d, for simulated films with several different thicknesses. For thick-
nesses above 30 nm, solutions (shown as minima in the plots of Fig. A.4) could
be found. However, below this thickness value, the values of Pi(∆ini(wi)−
∆fit(wi))2could not converge to the given solution. The problem can be al-
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 122
1.3 1.4 1.5 1.6 1.7
0.000
0.002
0.004
0.006
0.008
0.010
0.012
27 30 33
S D w - D w( ( ) ( ))2
iii
(a.u.)
ini fit
n d[nm]
d=30nmn=1.46
1.3 1.4 1.5 1.6 1.7
0.00
0.05
0.10
0.15
0.20
0.25
44 48 52 56
1.3 1.4 1.5 1.6 1.7
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
62 66 70 74 78 82
d=50nmn=1.46
d=70nmn=1.46
Figure A.4: The values of Pi(∆ini(wi)−∆fit(wi))2as a function of
the film high frequency refractive index nand thickness d. Each figure
exhibits the actual parameters of dini at which ∆ini was calculated
(nini=1.46 was kept for all graphs). These values are marked by the
broken lines. Calculations were performed in the infrared spectral range
between 700 and 4000 cm−1at 650incident angle.
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 123
ready observed for the solution for the 30 nm film, where a ”flat” minimum is
achieved for the extended range of the {n,d}values.
A.3 Multiple-angle measurements routine
The purpose of using multiple angles in ellipsometry measurements is to increase
the amount of data available to characterize a material. Varying the angle of
incidence changes, for instance, the path length in the film and the reflectance
of the s- and p- polarized radiation, Rp and Rs. For evaluation of the multiple-
angle spectra, often best-fit calculations are used. Here we will describe and
analyze a method of fitting the ∆ spectra at multiple angles. Typically the sub-
strate optical constants are available from separate ellipsometric measurements.
This is also advisable since a certain variation in optical constants can prevail
due to varying doping and surface treatment (for semiconductor materials such
as Si) or due to different coating methods (such as Au/glass substrates).
The calculated data is optimized to fit the experimental spectra taken at
several different incidence angles, and the results are refined until the best fit is
reached. Following this procedure, improved consistency of the deduced values
can be attained.
The procedure for determination of the thickness and high-frequency refrac-
tive index of thin organic films generally follows the same steps as described
earlier for a single angle of incidence in section A.2. As before, we are looking
for {n, d}pairs resulting in a solution of Pi(∆ini(wi)−∆fit(wi))2= 0. As in
section A.2, however, multiple solutions of {n, d}pairs are available for spectra
at each incidence angle. The advantage of performing this procedure for several
angles of incidence is that in general a unique solution of single {n, d}pair for
the given film can be found. Fig. A.5 shows such solution for a 70 nm thick
film with n=1.46 on a Si substrate (left panel). The solution is given as an
intersection of {n,d}plots at different incident angles. However, simulations for
Au substrate result in an overlap of {n,d}solutions for all 4 simulated angles
of incidence in the spectral range between 2000 cm−1and 4000 cm−1. The rea-
son for this can be revealed from observation of wavelength-defendant plots in
Fig. A.5. For Si, there is a strong change in ∆ around 750in the spectral region
of interest (2000 cm−1– 4000 cm−1) as shown in Fig. A.6. Thus there is also
a strong change in the {n, d}solution occurring around this angle (Fig. A.5).
This is the Brewster angle region (ϕBrewster ≈740for nsubstr = 3.48, which was
taken to represent Si substrate). For nsubstr=3.48+i*31.7, which was taken to
represent the Au substrate, there is no steep angle dependence analogue to that
of the Si in the spectral range of interest, as can be seen in Fig. A.6. However,
for thicker film thickness it becomes possible to find the single {n,d}solution
also for a metallic substrate, as demonstrated in Fig. A.7 for a 250 nm thick
film. This example illustrates the importance of the substrate optical properties
on determination of the film {n,d}parameters. In experimental IRSE measure-
ments, however, the 750angle of incidence is problematic due to a high opening
angle of the radiation coming from the globar source.
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 124
1.3 1.4 1.5 1.6
60
70
80
90
nn
d[nm]
d[nm]
Si Au
60
65
70
75
80
85
1.3 1.4 1.5 1.6 1.7 1.8
750
700
650
600
nn d[nm]
d[nm]
600
700
S D w - D w( ( ) ( ))
ini fit
2
iii
(a.u.)
S D w - D w( ( ) ( ))
ini fit
2
iii
(a.u.)
700
750
600
a.
b.
c.
d.
{n,d}=
{1.45,70nm}
{n,d}=
{1.45,70nm}
750
Figure A.5: Plots of the {n, d}of thin film (d=70 nm, n=1.46) giving
solution for Pi(∆ini(wi)−∆fit(wi))2= 0. (a) and (b): substrate
with nsubstr=3.48 (Si); (c) and (d): nsubstr=3.48+i*31.7 (Au). In the
upper panel, a three dimensional representation for the Pi(∆ini(wi)−
∆fit(wi))2= 0 at 600, 700and 750is shown by three colored surface
plots. The solutions are represented by the intersections of all three
plots, where a single solution is available for Si and multiple solutions
for Au substrate. The actual solution {n, d}={1.46,70 nm}is marked.
Lower panel: only the data along the minimum of the Pi(∆ini(wi)−
∆fit(wi))2= 0 is shown. The solution for n,d for Si substrate is given
by an intersection of the individual curves for each angle. For gold
substrate, all of the solution curves for all angles coincide on the same
curve. Spectra were simulated in a spectral range between 2000 cm−1
and 4000 cm−1.
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 125
40 50 60 70 80
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
40 50 60 70 80
0.02
0.07
0.12
0.17
0.22
0.27
40 50 60 70 80
0.4
0.5
0.6
0.7
0.8
0.9
D(0)
Rp
Rs
40 50 60 70 80
-180
-160
-140
-120
-100
-80
40 50 60 70 80
0.92
0.93
0.94
0.95
0.96
0.97
0.98
40 50 60 70 80
0.990
0.992
0.994
0.996
0.998
1.000
D(0)
Rp
Angle of incidence (0)
Rs
Si
Au
2000cm
2740cm
4390cm
-1
-1
-1
Figure A.6: Plots of ∆,Rp,Rsas a function of angle of incidence for
different wavelength in MIR spectral range (see color code). Calcu-
lated for a film of 70 nm thickness with high-frequency refractive index
n=1.46.
165
190
215
240
265
290
315
1.4 1.5 1.6 1.7 1.8 1.9
750
650
550
n
d[nm]
Figure A.7: Plots of the {n, d}for a thin film with thickness d=250
nm and n=1.46 giving solution for Pi(∆ini(wi)−∆fit(wi))2=
0. nsubstr=3.48+i*31.7 (Au),in the spectral range 2000 cm−1to
4000 cm−1.
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 126
Using the measured ∆ spectra for evaluation of the refractive index and
thickness of thinner films on silicon substrate in MIR spectral range is not
straight-forward. For instance, fitting of the simulated 15 nm thin film on a Si
substrate leaded to coinciding curves (multiple solutions) at 3 different angles in
the spectral region of interest. Thus the measurements at a lower wavelength is
desirable. Alternatively, in IRSE the thickness can be scaled from the strength
of the absorption bands, but this will require the proof of the linearity of the
strength of the absorption bands with the thickness of a given thin film. For
our purposes, thickness and refractive of the ultra-thin films were measured
using VIS–ellipsometry, improving the sensitivity to {n, d}parameters of ultra-
thin films. As a complimentary method, XPS analysis was used to determine
thickness in ultra-thin nitrobenzene films.
Additional complication when working with measured data is the offset that
may arise due to detector non–linearity. However, the procedure of the mul-
tiple angle fit can be in general adapted to treat this offset problem. This is
because although each measurement shifts the ∆ offset, it keeps the rest of the
characteristic features of the spectrum without any change. The discussion is
even more simplified when paying attention that in the spectral range of interest
and for the thicknesses below 100 nm the ∆ spectra can be approximated by a
linear curve of the type ∆=aω+bwith abeing the slope of the curve and bis
the ∆ value at ω=0 (ωis the wavenumber). Using this property, the problem
can be now separated into two parts: the slope of the ∆ spectrum, which is
kept unchanged during the measurements and the shifting parameter b. For our
purposes, it makes sense to use the offset parameter bat the wavenumber ωin
the range of interest (mid–infrared). Thus in the following discussion we use
b=∆(ω= 2000cm−1).
The fitting procedure is in general the same as described earlier for multiple
angle fit of the whole ∆ spectrum. Previously, we were looking for {n, d}pairs
resulting in a solution of Pi(∆ini(wi)−∆fit(wi))2= 0. The change now is that
we are looking for the solutions for aand bindependently: abs(aini −afit) = 0
and abs(bini(w=2000 cm−1)−bfit(w=2000 cm−1) = 0 should keep. Fig. A.8
shows the solution for a and b parameters which can be seen along the minima
of the graphs. It should be pointed out that each of these parameters gives
independent {n,d}solutions. This fact was exploited by the fitting of the whole
∆ spectrum previously through looking for zeros of Pi(∆ini(wi)−∆fit(wi))2=
0. However, Fig. A.8 shows that the solutions can also be found by fitting on
the aand bcomponents of the linear ∆ spectra in the range of interest.
Fig. A.9 shows the result of the multiple-angle approach, for slope a and for
the parameter b. For each of these parameters, the single {n,d}solution can
be obtained independently. Thus, it is in general sufficient to use the slope of
the ∆ spectrum and to ignore the background shift which is caused through the
measurement. This approach can be also extended to more complicated shapes
of ∆ spectra, when a polynomial fit can be applied to the ∆ spectrum and then
the analysis can be performed on the individual polynomial coefficients.
To summarize, IRSE measurements are suitable for evaluation of thickness
and refractive index. This can be either done by scaling of the absorption bands
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 127
d[nm]
d[nm]
n
n
%divergence
fromaini
%divergence
frombini
Figure A.8: Maps showing the solution for the slope a (left panel) and
for b=∆(ω=2000 cm−1) (right panel), when the measured ∆ spectra
are approximated by a linear curve. The solution for {n,d}are given
along the minima of the graphs, where the divergence (shown in %) of
the afit and bfit from aini and bini, respectively, is zero. The calculated
range was 2000 cm−1to 4000 cm−1, at 700angle of incidence.
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
48
54
60
66
72
78
84
750
700
650
600
60
64
68
72
76
80
84
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
750
700
650
600
solutionforoffsetb solutionforslopea
d[nm]
d[nm]
nn
Figure A.9: Plot of the {n, d}of thin film giving the zero solution
for abs(bini(w=2000 cm−1)−bfit(w=2000 cm−1)2= 0 (left) and
abs(aini −afit) = 0 (right), at multiple angles of incidence, with
nsubstr=3.48. The intersection of these functions gives the unique
{nini,dini}solution of the thin film (dini=70 nm,nini=1.46). The cal-
culated range was 2000 cm−1to 4000 cm−1.
APPENDIX A. SIMULATIONS OF THE IRSE SPECTRA 128
due to molecular vibrations from the well-known ”calibration” sample, or using
the interference properties of the obtained spectra. When the latter attitude
is applied, optimal parameters (such as angle of incidence, appropriate spectral
range) should be chosen in accordance to the substrate and the film optical
properties. We have shown that the background shift that can occur during the
measurements due to different radiation passage through the optical setup can
be overcome.
In this work, we used multiple approaches and complementary techniques
(VIS–ellipsometry, XPS) for analysis of the thickness and refractive index of the
thin films.
Bibliography
[1] Morrison S. T. Electrochemistry at semiconductor and oxidized metal elec-
trodes. Plenum press, New York, 1980.
[2] Hartig P., Rappich J., and Dittrich Th. Appl. Phys. Lett., 80:67, 2002.
[3] Cahen D. and Hodes G. Advanced Mater., 14:789, 2002.
[4] Hunger R., Jaegermann W., Merson A., Shapira Y., Pettenkofer C., and
Rappich J. J. Phys. Chem. B., 110:15432, 2006.
[5] Guenes S., Neugebauer H., and Sariciftci N. S. Chem. Rev., 107:1324,
2007.
[6] Goh C., Scully S. R., and McGehee M. D. J. Appl. Phys., 101:114503,
2007.
[7] Goetzberger A., Luther J., and Willeke G. Sol. En. Mat. and Solar Cells,
74:1, 2002.
[8] Bartic C., Campitelli A., and Borghs S. Appl. Phys. Lett., 82:475, 2003.
[9] Song K.S., Hiraki T., Umezawa H., and Kawarada H. Appl. Phys. Lett.,
90:063901, 2007.
[10] Patolsky F., Zheng G., Hayden O., Lakadamyali M., Zhuang X., and
Lieber C. M. PNAS, 101:14017, 2004.
[11] Ling M. M. and Bao Z. Chem. Mater., 16:4824, 2004.
[12] Briggs D., Brewis D.M., and Konieczko M.B. J. Mater. Sci., 14:1344,
1979.
[13] Wells R.K., Badyal J.P.S., Drummond I.W., Robinson K.S., and Street
F.J. Polymer, 34:3611, 1993.
[14] Rye R.R. and Martinetz R.J. J Appl Polym Sci, 37:2529, 1989.
[15] Wolkow R. A. Annu. Rev. Phys. Chem., 50:413, 1999.
[16] Forrest S.R. Chem. Rev., 97:1793, 1997.
129
BIBLIOGRAPHY 130
[17] Umbach E., Sokolowski M., and Fink R. Appl. Phys. A, 63:565, 1996.
[18] Rochet F., Jolly F., Bournel F., Dufour G., Sirotti F., and Cantin J.L.
Phys. Rev. B, 58:11029, 1998.
[19] Quinn J. F., Johnston A. P. R., Such G. K., Zelikin A. N., and Caruso F.
Chem. Soc. Rev., 37:707, 2007.
[20] Buriak J. M. Chem. Rev., 102:1271, 2002.
[21] Ulman A. An Introduction to Ultrathin Organic Films: From Langmuit-
Blodgett to Self-Assembly. Academic Press, Boston, 1991.
[22] Ashkenasy G., Cahen D., Cohen R., Shanzer A., and Vilan A. Acc. Chem.
Res., 35:35, 2002.
[23] Hartig P., Dittrich Th., and Rappich J. J. Electroanal. Chem., 524-
525:120, 2002.
[24] Bard A. J. and Faulkner L. R. Electrochemical Methods: Fundamentals
and Applications. Wiley, 2nd ed., 2000.
[25] Allongue P., de Villeneuve C. H., Cherouvrier G., Cortes R., and
Bernard M. C. J. Electroanal. Chem., 550-551:161, 2003.
[26] Merson A., Dittrich Th., Zidon Y., Rappich J., and Shapira Y. Appl.
Phys. Lett., 85:1075, 2004.
[27] Gensch M., Roodenko K., Hinrichs K., Hunger R., Guell A.G., Merson
A., Schade U., Shapira Y., Dittrich T., Rappich J., and Esser N. J. Vac.
sci. and Technol. B., 23:1838, 2005.
[28] Roodenko K., Rappich J., Gensch M., Esser N., and Hinrichs K. Appl.
Phys. A.
[29] Boukherroub R., Morin S., Bensebaa F., and Wayner D. D. M. Langmuir,
15:3831, 1999.
[30] Fellah S., Boukherroub R., Ozanam F., and J.-N. Chazalviel. Langmuir,
20:6359, 2004.
[31] Rappich J., Merson A., Roodenko K., Dittrich Th., Gensch M., Hinrichs
K., and Shapira Y. J. Phys. Chem. B., 110:1332, 2006.
[32] Vlasov Y. G., Tarantov Y. A., and Bobrov P. V. Anal. Bioanal. Chem.,
376:788, 2003.
[33] Madou M. J., Frese K. W. Jr., and Morrison S. R. Phys. Stat. Sol. A,
57:705, 1980.
[34] Yamashita Y., Asano A., Mishioka Y., and Kobayashi H. Phys. Rev. B.,
59:15872, 1999.
BIBLIOGRAPHY 131
[35] Angermann H., Henrion W., Rebien M., and R¨oseler A. Appl. Surf. Sci.,
235:322, 2004.
[36] Bernard M. C., Chausse A., Cabet-Deliry E., Chehimi M. M., Pinson J.,
Podvorica F., and Vautrin-Ul C. Chem. Mater., 15:3450, 2003.
[37] Higashi G. S, Becker R. S, Chabal Y. J., and Becker A. J. Appl. Phys.
Lett., 58:1656, 1991.
[38] Jakob P. and Chabal Y. J. J. Chem. Phys., 95:2897, 1991.
[39] Allongue P., de Villeneuve C. H., Morin S. Boukherroub R., and Wayner
D. D. M. Electrochim. Acta, 45:4591, 2000.
[40] Trucks G. W., Raghavachari K., Higashi G. S., and Chabal Y. J. Phys.
Rev. Lett., 65:504, 1990.
[41] Foell H., Christophersen M., Carstensen J., and Hasse G. Mater. Sci.
Engineering R., 39:93, 2002.
[42] Smith R. L. and Collins S. D. J. Appl. Phys., 71:R1, 1992.
[43] Yahyaoui F., Dittrich Th., Burke Th., Aggour M., Lust S., Levy-Clement
C., and Rappich J. J. Electrochem. Soc., 149:E472, 2002.
[44] Lehmann V. and Roennebeck S. J. Electrochem. Soc., 146:2968, 1999.
[45] Lehmann V. and Goesele. Appl. Phys. Lett., 58:856, 1991.
[46] Canham L. T. Appl. Phys. Lett., 57:1046, 1990.
[47] Hirschman K. D., Tsybeskov L., Duttagupta S. P., and Fauchet P. M.
Nature, 384:338, 1996.
[48] Cullis A. G., Canham L. T., and Calcott P. D. J. J. Appl. Phys., 82:909,
1997.
[49] Blyholder G. J. Phys. Chem, 68:2772, 1964.
[50] Chabal Y.J. and Raghavachari K. Surf. Sci., 502:41, 2002.
[51] Morse P. M. Phys. Rev., 34:57, 1929.
[52] Nakamoto K. Infrared and Raman Spectra of Inorganic and Coordination
Compounds, 5th edition. Willey Interscience, New-York, 1997.
[53] Nielsen J. R. and Holland R. F. J. Molec. Spectrosc., 6:394, 1961.
[54] Kittel C. Introduction to Solid State Physics. John Willey and Sons,
New-York, 1996.
[55] Yu P. Y. and Cardona M. Fundamentals of Semiconductors. Physics and
Material Properties. Springer, Berlin, 1996.
BIBLIOGRAPHY 132
[56] R¨oseler A. and Korte E. H. Handbook of vibrational spectroscopy, volume 2.
Editors: Griffith P. R and Chalmers J. (John Wiley and Sons, Chichester,
2002), 2002.
[57] Hasegawa T., Nishijo J., Umemura J., and Theiss W. J. Phys. Chem. B,
105:11178, 2001.
[58] Tu Y. and Tersoff J. Phys. Rev. Lett., 84:4393, 2000.
[59] Brendel R. and Bromann D. J. Appl. Phys., 71:1, 1991.
[60] Naiman M. L., Kirk C. T., Aucoin R. J., Terry F. L., and Wyatt P. W.
J. Electrochem. Soc., 131:637, 1984.
[61] Teitler S. and Henvis B. W. J. Opt. Soc. Am, 60:830, 1970.
[62] Berreman D. W. J. Opt. Soc. Am, 62:502, 1972.
[63] Azzam R. M. A. and Bashara N. M. Ellipsometry and Polarized Light.
North Holland Publishing Company, Amsterdam, 1977.
[64] Chabal Y. J. Physica B, 170:447, 1991.
[65] Ong C. K. and Khoo G. S. J. Phys. C, 20:419, 1986.
[66] Berreman D. W. Phys. Rev., 130:2193, 1963.
[67] Tolstoy V.P., Chernyshova I. V., and Skryshevsky V. A. Handbook of
Infrared Spectroscopy of Ultrathin Films. Willey Interscience, New-York,
2003.
[68] Vickerman J. C. Surface analysis. The principle techniques. Wiley-VCH,
Weinheim, Germany, 1997.
[69] Sakamoto K., Hashizume H., Nagafusa M., Sato H., and Ushioda S. Surf.
Sci., 368:292, 1996.
[70] Campion A. and Kambhampati P. Chem. Soc. Rev., 27:241, 1998.
[71] Duwez A.-S. J. Elect. Spectr. Relat. Phenom., 134:97, 2004.
[72] Reinert F. and Huefner S. New Journ. Phys., 7:97, 2005.
[73] Himpsel F.J. Brazil. Journ. Phys., 23:31, 1993.
[74] Hunger R., Fritsche R., Jaeckel B., Jaegermann W., Webb L. J., and Lewis
N.W. Phys. Rev. B, 72:045317, 2005.
[75] Hinrichs K., Gensch M., and Esser N. Appl. Spectr., 59:272A, 2005.
[76] Brunet-Bruneau A., Fisson S., Vuye G., and Rivory J. J. Appl. Phys.,
87:7303, 2000.
BIBLIOGRAPHY 133
[77] Yang F., Roodenko K., Hinrichs K., and Rappich J. J. Micromech. Mi-
croeng., 17:S56, 2007.
[78] Tiwald T. E., Woollam J. A., Zollner S., Christiansen J., Gregory R. B.,
Wetteroth T., Wilson S. R., and Powell A. R. Phys. Rev. B, 60:11464,
1999.
[79] R¨oseler A. and Molgedey W. Infrared Physics, 24:1, 1984.
[80] Schubert M. Thin Solid Films, 313:323, 1998.
[81] R¨oseler A. J. Opt. Soc. Am., A 9:1124, 1992.
[82] Roodenko K., Gensch M., Heise H.M., Schade U., Esser N., and Hinrichs
K. Infrared Phys. and Technol., 49:39, 2006.
[83] Schade U., Bartl F., Hofmann K.P., Illing G., Korte E.H., Peatman W.B.,
and R¨oseler A. BESSY Annual Report, 1998:457, 1998.
[84] Snyder R. G. J. Chem. Phys., 47:1316, 1967.
[85] Keyes R. J. (editor). Optical and Infrared Detectors, 2nd edition. Springer-
Verlag, Berlin, 1980.
[86] Wolfe W. L. and Zissis G. J. The Infrared Iandbook. The Infrared Infor-
mation Analysis (IRIA) Center, Michigan, 1993.
[87] Rogalski A. Infrared Detectors. CRC, Singapore, 2000.
[88] Richards P. L. J. Appl. Phys., 76:1, 1994.
[89] Seah M.P. and Dench W. A. Surf. Interface Anal., 1:2, 1979.
[90] Himpsel F. J., McFeely F. R., Taleb-Ibrahimi A., and Yarmoff J. A. Phys.
Rev. B., 38:6084, 1988.
[91] Lu Z. H., Graham M. J., Jiang D. T., and Tan K. H. Appl. Phys. Lett.,
63:2941, 1993.
[92] Briggs D. and Seah M. P. Practical surface analysis: Auger and X-ray
photoelectron spectroscopy. John Wiley and Sons Ltd, 1996.
[93] Beamson G. and Briggs D. Molec. Phys., 76:919, 1992.
[94] Mayer T., Lebedev M. V., Hunger R., and Jaegermann W. Appl. Surf.
Sci., 252:31, 2005.
[95] Proctor A. and Sherwood P. M. A. Anal. Chem., 54:13, 1982.
[96] Tougaard S. Surf. Interface Anal., 25:137, 1997.
[97] Steiner P., H¨ochst H., and H¨ufner S. Z. Physik B, 30:129, 1978.
BIBLIOGRAPHY 134
[98] Wagner C. D., Davis L. E., Zeller M. V., Taylor J. A., Raymond R. H.,
and Gale L. H. Surf. Interf. Anal., 3:211, 1981.
[99] Tanuma S. and Powell C.J.and Penn D.R. Surf. Interface Anal., 21:165,
1994.
[100] Mohai M. Surf. Interface Anal., 36:828, 2004.
[101] Clarkson J. and Smith W. E. J. Mol. Struct., 655:413, 2003.
[102] Balfour W. J. Spectrochim. Acta Part A, 39:795, 1983.
[103] Nuttall R. H., Roberts E. R., and Sharp D. W. A. Spectrochim. Acta,
17:947, 1961.
[104] Tabei K. and Ito C. Bull. Chem. Soc. Jap., 41:514, 1968.
[105] Whetsel K. B., Hawkins G. F., and Johnson F. E. J. Am. Chem. Soc.,
78:3360, 1956.
[106] Cote G. L. and Thompson H. W. Proc. Royal Soc. London A: Math. Phys.
Sci., 210:217, 1951.
[107] Mayo D. W., Miller F. A., and Hannah R. W. Course notes on the
interpretation of infrared and Raman Spectra. John Wiley and Sons, Inc.,
Hoboken, New Jersey, 2004.
[108] Zollinger H. Diazo Chemistry I. VCH Verlagsgesellschaft, Weinheim,
1994.
[109] Glaser R., Horan C. J., Lewis M., and Zollinger H. J. Org. Chem., 64:902,
1999.
[110] Chabal Y. J., Higashi G. S., and Raghavachari K. J. Vac. Sci. Technol.
A., 7:2104, 1989.
[111] Yota J. and Burrows V. A. J. Appl. Phys., 69:7369, 1991.
[112] Angermann H., Henrion W., Rebien M., and R¨oseler A. Solar Energy
Mater. Solar Cells, 83:331, 2004.
[113] Bocharov S. and Teplyakow A. V. Surf. Sci., 573:403, 2004.
[114] Syomin D., Wang J., and Koel B. E. Surf. Sci., 495:L827, 2001.
[115] Hayden B. E. Vibrational Spectroscopy of Molecules on Surfaces. Plenum
Press, New York, 1987.
[116] Laforgue A., Addou T., and Belanger D. Langmuir, 21:6855, 2005.
[117] Jin H., Kinser C. R., Bertin P. A., Kramer D. E., Libera J. A., Hersam M.
C., Nguyen S. T., and Bedzyk M. J. Langmuir, 20:6252, 2004.
BIBLIOGRAPHY 135
[118] Senf F., Flechsig U., Eggenstein F., Gudat W., Klein R., Rabus H., and
Ulm G. J. Synchr. Radiation, 5:780, 1998.
[119] Beckhoff B., Fliegauf R., Ulm G., Pepponi G., Wobrauschek P., Fabry L.,
and Pahlke S. Spectrochim. Acta part B, 56:2073, 2001.
[120] Beatrix Pollakowski. Monitoring X–ray fluorescence signal from
nitrobenzene–modified Si(111) surface at the plane–grating monochro-
mator (PGM) beamline for undulator radiation at the Physikalisch-
Technische Bundesanstalt (PTB) at BESSY II. Private communication,
2007.
[121] Gensch M. Infrared ellipsometry for the investigation of interfacial layers
and thin organic films on silicon. Mensch und Buch Verlag, 2005. PhD
thesis.
[122] Roodenko K., Gensch M., Rappich J., Hinrichs K., Esser N., and Hunger
R. J. Phys. Chem. B, 111:7541, 2007.
[123] Henderson M. A. Surf. Sci. Rep., 46:5, 2002.
[124] Skinner W. M., Prestige C. A., and Smart R. St. C. Surf. Interface Anal.,
24:620, 1996.
[125] Grellner F., Klingenberg B., Borgmann D., and Wedler G. J. Electron.
Spectrosc. Relat. Phenom., 71:107, 1995.
[126] Rappich J., Hartig P., Nickel N.H., Sieber I., Schulze S., and Dittrich Th.
Microelec. Engineer., 80:62, 2005.
[127] Smith B. Infrared Spectral Interpretation. A Systematic Approach. CRC
press, New York, 1999.
[128] Eisenhardt C. G., Gemeschu A. S., Baumgrtel H., Chelli R., Cardini G.,
and Califano S. Phys. Chem. Chem. Phys., 3:5358, 2001.
[129] Hinrichs K., Gensch M., R¨oseler A., and Esser N. J. Phys.: Cond. Mat.,
16:s4335, 2004.
[130] Plumb R. C. and Hornig D. F. J. Chem. Phys., 23:947, 1955.
[131] Tan Y. P., Khatua S., Jenkins S. J., Yu J. Q., Spencer J. B., and King D.
A. Surf. Sci., 589:173, 2005.
[132] Syomin D., Kim J., Koel B. E., and Ellison G. B. J. Phys. Chem. B,
105:8387, 2001.
[133] Friderichsen A. V., Radziszewski J. G., Nimlos M. R. Winter P. R., Day-
ton D. C., David D. E., and Ellison G. B. J. Am. Chem. Soc., 123:1977,
2001.
BIBLIOGRAPHY 136
[134] Koval’chuk E.P., Stratan N.V., Reshetnyak O.V., Blazejowski J., and
Whittingham M.S. Solid State Ionics, 141:217, 2001.
[135] Ogata Y.H., Tsuboi T., Sakka T., and Naito S. J. Porous Mater., 7:63,
2000.
[136] Borghesi A., Sassella A., Pivac B., and Pavesi L. Solid State Communi-
cations, 87:1, 1993.
[137] Maruyama T. and Ohtani S. Appl. Phys. Lett., 65:1346, 1994.
[138] Miura T., Niwano M., Shoji D., and Miyamoto N. Appl. Surf. Sci.,
101:454, 1996.
[139] Yang F. PhD thesis, work in preparation.
[140] Teyssot A., Fidelis A., Fellah S., Ozanam F., and Chazalviel J. N. Elec-
trochimica Acta, 47:2565, 2002.
[141] Yamada T., Shirasaka K., Noto M., Kato H. S., and Kawai M. J. Phys.
Chem. B, 110:7357, 2006.
[142] Canaria C.A., Lees I.N., Wun A.W., Miskelly G.M., and Sailor M.J. Inorg.
Chem. Commun., 5:560, 2002.
[143] Webb L. J., Rivillon S., Michalak D. J., Chabal Y. J., and Lewis N. S. J.
Phys. Chem. B, 110:7349, 2006.
[144] Torres J. and Arkin I. T. Biophys. J., 82:1068, 2002.
[145] Young T. F., Chen C. P., Liou J. F., Yang Y. L., and Chang T.C. J.
Porous Mater., 7:339, 2000.
[146] Yamada T., Kawai M., Wawro A., Suto S., and Kasuya A. J. Chem. Phys,
21:10660, 2004.
[147] Ferguson G. A. and Raghavachari K. J. Chem. Phys, 125:154708, 2006.
[148] Taguchi Y., Fujisawa M., Takaoka T., Okada T., and Nishijima M. J.
Chem. Phys, 95:6870, 1991.
[149] Bateman J. E., Horrocks B.R., and Houlton A. J. Chem. Soc. Faraday
Trans., 93:2427, 1997.
[150] Michalak D. J., Rivillon S., Chabal Y. J., Esteve A., and Lewis N. S. J.
Phys. Chem. B, 110:20426, 2006.
[151] Gurtner Ch., Wun A. W., and Sailor M. J. Angew. Chem. Int. Ed.,
38:1966, 1999.
[152] Yablonovitch E., Allara D. L., Chang C. C., and Gmitter T.and Bright T.
B. Phys. Rev. Lett., 57:249, 1986.
BIBLIOGRAPHY 137
[153] Miura T., Niwano M., Shoji D., and Miyamoto N. J. Appl. Phys., 79:4373,
1996.
[154] Ogawa H., Ishikawa K., Inomata C., and Fujimura S. J. Appl. Phys.,
79:472, 1996.
[155] Cheng C.L., Lin J.C., and Chang H.C. J. Chem. Phys., 106:7411, 1997.
[156] Wang J. K., Tsai C. S., Lin C. E, and Lin J. C. J. Chem. Phys., 113:5041,
2000.
[157] Disselkamp R., Chang H. C., and Ewing G. Surf. Sci., 240:193, 1990.
[158] Newton T. A., Boiani J.A., and Hines M. A. Surf. Sci., 430:67, 1999.
[159] MacLaren D. A., Curson N. J., Atkinson P., and Allison W. Surf. Sci.,
490:285, 2001.
[160] Khan A. and Hess P. Appl. Phys. Lett., 85:4157, 2004.
[161] Lambers J. and Hess P. Thin Solid Films, 400:111, 2001.
[162] Hsiao G. S., Virtanen J. A., and Penner R. M. Appl. Phys. Lett., 63:1119,
1993.
[163] G¨uell A. G., Roodenko K., Yang F., Hinrichs K., Gensch M., Sanz F., and
Rappich J. Mater. Sci. and Eng. B., 134:273, 2006.
[164] Kanashima T., Maida O., Kohma N., Agata M., Yudate S., Ueno M.,
Okuyama M., and Ohashi H. Jap. J. Appl. Phys. part 1, 40:4195, 2001.
[165] Allongue P., de Villeneuve C. H., Pinson J., Ozanam F., Chazalviel J. N.,
and Wallart X. Electrochim. Acta, 43:2791, 1998.
[166] Hoshino T. and Nishioka Y. J. Chem. Phys., 111:2109, 1999.
[167] Smith W. H. and Bard A. J. J. Am. Chem. Soc., 97:5203, 1975.
[168] Oliveria M. C. F. Electrochim. Acta, 48:1829, 2003.
[169] La Y. H., Kim H. J., I. S. Maeng, Jung Y. J., and Park J. W. Langmuir,
18:2430, 2002.
[170] Moon J. M., Kim K. J., Kang T. H., Kim B., Kang H., and Park J. W.
Langmuir, 14:5673, 1998.
[171] Eck W., Stadler V., Geyer W., Zharnikov M., G¨olzh¨auser A., and Grunze
M. Adv. Mater., 12:805, 2000.
[172] Han S. W., Lee I., and Kim K. Langmuir, 18:182, 2002.
[173] Mendes P., Belloni P., Ashworth M., Hardy C., Nikitin K., Fitzmaurice
D., Critchley K., Evens S., and Preece J. Chem. Phys. Chem., 4:884, 2003.
BIBLIOGRAPHY 138
[174] Kl¨opffer W. Introduction to Polymer Spectroscopy. Springer-Verlag,
Berlin, 1984.
[175] Dittrich T., Schwartzkopff M., Hartmann E., and Rappich J. Surf. Sci.,
437:154, 1999.
[176] Dittrich T., Burke T., Koch T., and J. Rappich. J. Appl. Phys., 89:4636,
2001.
[177] Distefano G., Guerra M., Jones D., Modelli A., and Colonna F. P. Chem.
Phys., 52:389, 1980.
[178] Keane M. P., de Brito A. N., Correia N., and Svensson S. Chem. Phys.,
155:379, 1991.
[179] Kumar S. N., Bouyssoux G., and Gaillard F. Surf. Interface Anal., 15:531,
1990.
[180] Kohut-Svelko N., Reynaud S., Dedryevre R., Martinez H., Gonbeau D.,
and Francois J. Langmuir, 21:1575, 2005.
[181] Liu Y. C. and McCreery R. L. Anal. Chem., 69:2091, 1997.
[182] Kishi K., Takata H., and Kohnushi S. J. Electron Spectrosc. Relat. Phe-
nom., 61:83, 1992.
[183] Huang S. H., Fischer D. A., and Gland J. L. J. Phys. Chem., 100:10223,
1996.
[184] Ma Y., Chen C. T., Meigs G., Randall K., and Sette F. Phys. Rev. A,
44:1848, 1991.
[185] Klausen J., Tr¨ober S. P., Haderlein S. B., and Schwarzenbach R. P. Env-
iron. Sci. Technol., 29:2396, 1995.
[186] Hunger R., Jaegermann R. R., Merson A., Y. Shapira, Pettenkofer C.,
and J. Rappich. J. Phys. Chem. B, 110:15432, 2006.
[187] Laforgue A., Addou T., and Belanger D. Langmuir, 21:6855, 2005.
[188] Stewart M. P., Maya F., Kosynkin D. V., Dirk S. M., Stapleton J. J.,
McGuiness C. L., Allara D. L., and Tour J. M. J. Am. Chem. Soc.,
126:370, 2004.
[189] Adenier A., Cabet-Deliry E., Chausse A., Griveau S., Mercier F., Pinson
J., and Vautrin-Ul C. Chem. Mater., 17:491, 2005.
[190] Allongue P., Delamar M., Desbat B., Fagebaume O., Hitmi R., Pinson J.,
and Saveant J.-M. J. Am. Chem. Soc., 119:201, 1997.
[191] Riga J., Pireaux J. J., and Verbist J. J. Mol. Phys., 34:131, 1977.
BIBLIOGRAPHY 139
[192] Takahata Y. and Chong D. P. Bull. Chem. Soc. Jpn., 73:2453, 2000.
[193] Salavagione H. J., Acevedo D. F., Miras M. C., Motheo A. J., and Bar-
bero C. A. J. Polym. Sci., Part A: Polym. Chem., 42:5587, 2004.
[194] Ohta T., Fujikawa T., and Kuroda H. Chem. Phys. Lett., 32:369, 1975.
[195] Rennie E. E., Kempgens B., Koeppe H. M., Hergenhahn U., Feldhaus J.,
Itchkawitz B. S., Kilcoyne A. L. D., Kivimaeki A., Maier K., Piancastelli
M., Polcik N., Ruedel M., and Bradshaw A. A. M. J. Chem. Phys.,
113:7362, 2000.
[196] Solomon J. L., Madix R. J., and J. Stoehr. Surf. Sci., 255:12, 1991.
[197] Yamada K., Teshima K., Kobayashi N., and Hirohashi R. J.Electroanal.
Chem., 394:71, 1995.
[198] Gospodinova N. and Terlemezyan L. Prog. Polym. Sci., 23:1443, 1998.
[199] Zuman P. and Shah B. Chem. Rev., 94:1621, 1994.
[200] Prabhakaran K., Kobayashi K., and Ogino T. Surf. Sci., 290:239, 1993.
[201] Hollinger G., Morar J. F., Himpsel F. J., Hughes G., and Jordan J. L.
Surf. Sci., 168:609, 1986.
[202] Namiki A., Tanimoto K., Nakamura T., Ohtake N., and Suzaki T. Surf.
Sci., 222:530, 1989.
[203] Batich C. D. and Donald D. S. J. Am. Chem. Soc., 106:2758, 1984.
[204] Sakamoto K., Zhang H. M., and Uhrberg R. I. G. Phys. Rev. B, 70:35301–
1, 2004.
[205] Teraoka Y. and Yoshigoe A. Jpn. J. Appl. Phys., 41:4253, 2002.
[206] Hollinger G. and Himpsel F. J. Appl. Phys. Lett., 44:93, 1984.
[207] Oh J. H., Yeom H. W., Hagimoto Y., One K., Oshima M., Harita N.,
Nywa M., Toriumi A., and Kakizaki A. Phys. Rev. B, 63:205310–1, 2001.
[208] Niwano M., Katakura H., Takeda Y., Takakuwa Y., Miyamoto N., Hiraiwa
A., and Yagi K. J. Vac. Sci. Technol., A, 9:195, 1991.
[209] Palik E. D. Handbook of optical constants of solids. Academic Press, Inc.,
London, 1985.
[210] Tsankov D., Hinrichs K., Korte E. H., Dietel R., and R¨oseler A. Langmuir,
18:6559, 2002.
[211] Powell M. J. D. Numerical Methods for Nonlinear Algebraic Equations,
chapter 7. Editor: Rabiniwitz P., (Routledge), 1970.
Acknowledgments
This work would not be possible without the input and support from many of
my friends and colleagues. Especially I would like to express my gratitude to
the following people:
Prof. Dr. N. Esser for the supervision of this work and for helpful discus-
sions, and for the opportunity to work in the ISAS team.
Dr. K. Hinrichs for his guidance, support and for sharing his deep insight
into the field of IR ellipsometry, for his humor and friendship.
Dr. J. Rappich for his introduction to electrochemistry, preparation of the
samples, interesting discussions and collaborative work.
Dr. R. Hunger for XPS measurements and for enlightening conversations.
Dr. Th. Dittrich for helpful comments.
Dr. U. Schade for a collaborative work at the infrared beamline at BESSY II.
Dr. A. R¨oseler for useful discussions.
Prof. Dr. W. Richter for his useful comments in the course of this work.
Prof. Dr. A. Katzir for his support during my relocation to Germany and
for his hospitality during my visits at Tel Aviv university.
Prof. Dr. Y. Shapira for his kind hospitality in the Tel Aviv University.
Dr. A. Merson for sample preparation.
Dr. M. Ortolani for his help with the bolometer cooling.
Dr. M. Gensch for his collaborative work, encouragement and friendship.
Mr. F. Yang for a collaborative work, many useful discussions and for the
corrections of this thesis.
Ms. B. Pollakowski for X–ray fluorescence measurements and for helpful dis-
cussions.
Ms. I. Fischer for her dedicated assistance in the laboratory and for her help
with many set-ups.
Mr. C. Roland, for his help with the UHV system and with the cryostatic
equipment.
Mr. G. Hinte, Mr. R. Sorge, Mr. R. Herzlieb and Mr. D. Kutz for their
constructions in the workshop.
Mr. R. Tischendorf for his electronics and computer support.
Ms. S. Reichardt for her guidance and her support concerning the Minerva
fellowship.
All other members of ISAS and the former group of Prof. Richter for the
motivating atmosphere.
I am thankful to many of my colleagues and my friends - Dr. S. Chandola,
Dr. Y. Mikhailova, Dr. S.D. Silaghi, Dr. J. Lee, Dr. Th. Deniozou, Dr. C.
Cobet, Mr. C. Werner, Ms. M. Rakel, Ms. S. John Louis, Ms. D. M. Rosu, Mr.
D. Aulich, Ms. R. Passmann and Ms. S. Wenmackers for their support.
I am also thankful to all my friends abroad, who kept in touch despite the
distance.
My warmest thanks are to Udi Fuchs, for his comfort during hard days, for
a supply of cake and cups of hot chocolate, for discussions of the simulations
and for reminding me about all those forgotten π’s in the integrals when my
programs did not work as expected. I am also grateful to Udi’s parents, for
their care and support.
And of cause, I would like to say a huge Spasiba to my family - my parents
and my sister Helena, for their phone calls, their understanding and encourage-
ment.
This work was supported by Minerva fellowship.
