Structure and Reactivity of Diesel Soot Particles
from Advanced Motor Technologies
vorgelegt von
Diplom-Physiker
Manfred Erwin Schuster
Mistelbach
Von der Fakultät II-Mathematik und Naturwissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften
Dr. rer. nat.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. Thomas Möller
Berichter/Gutachter: Prof. Dr. Robert Schlögl
Berichter/Gutachter: Prof. Dr. Mario Dähne
Berichter/Gutachter: Prof. Dr. Cécile Hébert
Tag der wissenschaftlichen Aussprache: 18.11.2010
Berlin 2011
D 83
per aspera ad astra
Zusammenfassung
Dieselruße und andere Feinstaubpartikel können im Lungengewebe gesundheits-
schädlich wirken. Rußpartikel im Lungenbereich sind ein möglicher Grund für
Lungenkrebs. Teilchengröße und Reaktivität sind wichtige Faktoren für die Ge-
sundheitsschädlichkeit dieser Partikel. Je kleiner die Teilchen sind, desto weiter
können sie ins Lungengewebe eindringen. Die Oberflächenfunktionalisierung dieser
Teilchen entscheidet dann wie stark die Reaktivität im Lungenbereich ist. In der
vorgelegten Arbeit, wurde eine systematische Studie durchgeführt, um sowohl die
Partikel als Ganzes, also auch die Oberflächenbeschaffenheit von zwei Standards,
Euro IV und Euro VI, zu charakterisieren. Die Bindungszustände wurden mittels
EELS untersucht, während die strukturelle Ordnung mittels HRTEM und SEM
untersucht wurden. NEXAFS und XPS Untersuchungen gaben Aufschluss über
die Oberflächenfunktionalisierung der Proben. Die Proben wurden weiters einer
Testreaktion unterzogen, in der eine Behandlung im Auspuffsystem simuliert wurde.
Danach wurden die Proben mit den unbehandelten verglichen und eine Struktur-
Reaktivitätskorrelation hergestellt. Die Ergebnisse der Euro IV und Euro VI Proben
wurden mit Referenzproben verglichen.
Die grundlegenden Erkenntnisse dieser Studie sind:
1. Die Mikrostruktur der beiden Dieselruße ist durch eine Zwiebel-Struktur cha-
rakterisiert, wobei sich auf der Oberfläche der Euro IV Probe Oberfläche molekulare
Kohlenstoffe befinden.
2. Die thermische Behandlung entfernt die molekularen Kohlenstoffe von der Euro
IV Oberfläche und führt des weiteren zu einer höheren strukturellen Ordnung der
Probe.
3. Die Euro VI Probe weist im Vergleich zu Euro IV eine signifikant höhere und
heterogenere Sauerstofffunktionalisierung auf.
4. Die Reaktivitätsmessungen der beiden Proben mittels TPO zeigt, ein unter-
schiedliches Verbrennungsverhalten mit einer höheren Verbrennungstemperatur für
die Euro VI Probe. Sobald der Verbrennungsprozess eingesetzt hat, verbrennt die
Euro VI Probe stärker als die Euro IV Probe.
Die höhere Reaktivität der Euro VI Probe erklärt sich durch die höhere Funktiona-
lisierung mit Sauerstoff und durch die Umwandlung dieser Sauerstoffe während der
Testreaktion von weniger stabilen zu thermisch stabileren Sauerstoffgruppen. Diese
Studie hat auch gezeigt, dass eine Oxidations- und Hitzebehandlung mit höherer
Temperatur notwendig ist, um eine Restrukturierung der Proben zu verhindern und
damit komplette Verbrennung zu ermöglichen.
Abstract
Diesel engine soot and other fine dust particles have been shown to be hazardous
if they enter the human body, penetrate the lungs, and irritate lung tissue. These
particles have been linked to the possible causes of lung cancer. Particle size and
reactivity are important factors in this process. The smaller the particles, the
further they can enter the lung system. More crucially, reactivity determines the
environmental accessibility of such soot particles. This reactivity is defined by the
microstructure of the system and the functionalization of carbon atoms, which is
mainly determined by carbon-oxygen and carbon-hydrogen bonds.
A systematic study of both bulk and surface characterization is missing in lit-
erature. The aim of this work is to determine the microstructure and reactivity
of two soot models, namely two Emission Standards, Euro IV and Euro VI, by
combining these surface and bulk sensitive techniques. As reference samples, soot,
generated by spark discharge of graphite electrodes, served as model for atmospheric
particles, Highly Ordered Pyrolytic Graphite (HOPG), Nanotubes (NC 3100) and
Flammruss were used. HRTEM, SEM and EELS were used in this study for the
bulk characterization of the microstructure and electronic structure of carbonaceous
samples. Surface sensitive methods such as NEXAFS spectroscopy and XPS were
applied to probe the surface species which affect the reactivity and toxicity of the
particles.
The Euro IV and Euro VI samples were subsequently subject to TPO, in which
oxidation and heat experiment simulates a possible treatment in the exhaust system.
The result of this treatment was compared to results for untreated samples in
order to visualize the changes, to elucidate the structure-reactivity correlation, and
establish a reaction scheme for the different soot samples. The main results obtained
in this work are the following:
1. The microstructure of the Euro IV and VI samples is determined by an onion-
like arrangement of the graphene layers, building up a "core-shell structure", with
additional molecular-carbon units present on the surface of the Euro IV sample.
2. The microstructure of the Euro IV sample is altered by the oxidation treatment,
resulting in a loss of this "disordered" carbon units on the surface, approaching a
similar microstructure as the Euro VI sample, which did not alter its structural
state as a result of the oxidation treatment.
3. The oxygen-surface functionalization differs significantly in amount and het-
erogeneity between the two samples. Additionally, the modifications of the oxygen-
carbon bonds due to the oxidation treatment are also quite different.
4. The reactivity of the carbons shows a different starting point for combustion
of the two samples and and a different combustion rate, which can be explained by
the results obtained by the data gathered during this project.
Two factors, much higher functionalization of the Euro VI sample and the con-
version of less stable oxygen-functional groups to highly stable oxygen-functional
groups during the oxidation treatment, explain the higher reactivity of this sample
as well as its higher accessibility to the environment and subsequently a higher
reactivity in the human body. Furthermore, oxidation and heat treatment have to
occur in a much faster and higher temperature regime to ensure that soot particles
undergo complete combustion before a restructuring of the soot samples can occur,
thus hindering further combustion.
8
Table of Contents
1 Introduction 1
1.1 TheProblemofSoot.......................... 1
2 Structure of soot 5
2.1 Sootformation ............................. 5
2.2 Carbon.................................. 7
2.2.1 sphybridization......................... 8
2.2.2 sp2hybridization........................ 8
2.2.3 sp3hybridization........................ 9
2.2.4 Chemistry of Carbon . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Functional Groups on Carbon Surfaces . . . . . . . . . . . . . . . . 11
2.3.1 Generation of Surface Oxygen Functional Groups . . . . . . 13
2.3.2 Structure of Carbon-Oxygen Functional Groups . . . . . . . 15
2.4 Materialsstudied............................ 15
2.4.1 Emission Standards . . . . . . . . . . . . . . . . . . . . . . . 17
3 Characterization techniques 19
3.1 Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.1 Transmission Electron Microscopy . . . . . . . . . . . . . . . 20
3.1.2 Electron Energy Loss Spectroscopy . . . . . . . . . . . . . . 24
3.1.3 Scanning Electron Microscopy . . . . . . . . . . . . . . . . . 32
3.2 NEXAFS ................................ 33
3.2.1 Normalization and background corrections . . . . . . . . . . 37
3.3 XPS ................................... 38
3.3.1 Initial state effects . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.2 Final state effects . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.3 Inelastic background . . . . . . . . . . . . . . . . . . . . . . 41
3.3.4 Shakeup/off........................... 41
4 HRTEM on Soot 43
4.1 Structural analysis of the as-received samples . . . . . . . . . . . . 43
4.1.1 Particlesize........................... 43
4.1.2 HRTEM............................. 45
4.1.3 Curvature measurements . . . . . . . . . . . . . . . . . . . . 47
i
Table of Contents
4.2 Effect of oxidation treatment . . . . . . . . . . . . . . . . . . . . . . 48
4.2.1 Particlesize........................... 48
4.2.2 HRTEM............................. 51
4.2.3 Curvature measurements . . . . . . . . . . . . . . . . . . . . 52
5 EELS 61
5.1 LowLossregion............................. 61
5.2 C-Kedgeregion............................. 66
6 NEXAFS 71
6.1 CarbonNEXAFS............................ 71
6.1.1 C K NEXAFS vs C K EELS . . . . . . . . . . . . . . . . . . 79
6.2 OxygenNEXAFS............................ 80
7 XPS 89
7.1 C1sXPS................................. 90
7.2 O1sXPS................................. 96
7.3 Depthprofile .............................. 100
8 Discussion 109
9 Conclusion 119
10 Acknowledgement 121
ii
List of Figures
1.1
Anual death rates for different cancer sites in the USA between
1930-1989................................. 2
2.1 Growing steps for soot and fullerenes via aromers. . . . . . . . . . . 6
2.2 Schematic mechanism of the formation of soot particles . . . . . . . 7
2.3 Molecules relevant for the combustion process . . . . . . . . . . . . 7
2.4
Families of polycrystalline carbon materials relevant for surface chem-
istry.................................... 9
2.5 Variations of Carbon: Graphite, Diamond, C60 and Nanotube. . . . 10
2.6
Schematic representation of the structure of chemical active oxygen
functionalgroups ............................ 16
3.1
Signals generated by the interaction of a high energy electron with a
thinspecimen............................... 20
3.2 Schematic path of an electron in a TEM. . . . . . . . . . . . . . . . 21
3.3 The two basic operation modes of the TEM imaging system. . . . . 22
3.4 Mechanism of mass-thickness contrast in a BF image. . . . . . . . . 24
3.5 Electron energy-loss spectrum for carbon. . . . . . . . . . . . . . . . 25
3.6 Energy-level diagram of a solid. . . . . . . . . . . . . . . . . . . . . 27
3.7 Scatteringdiagram............................ 29
3.8 Post-column attached Gatan Tridiem Filter. . . . . . . . . . . . . . 30
3.9
Variation of the C1s fine structure of graphite due to variations of
the collection semi-angle (β0) and due to the specimen orientation. . 32
3.10
Final state dependence upon the direction of the momentum transfer
Q in the case of a) a small scattering angle and b) a large scattering
angle. .................................. 33
3.11 NEXAFS spectra of polymers. . . . . . . . . . . . . . . . . . . . . . 36
3.12 XPS emission process. . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.13
XPS spectrum obtained from a Euro VI soot sample using
Eexc
= 1200 eV.
40
4.1 SEM micrograph representative for the investigated soot samples. . 44
4.2 TEM overview for the determination of the soot particle size. . . . . 44
4.3 Radii distribution for the as-received Euro IV sample. . . . . . . . . 45
4.4 Radii distribution for the as-received Euro VI sample. . . . . . . . . 46
iii
List of Figures
4.5
Definition of "graphene length" and "length" for the curvature mea-
surements................................. 48
4.6 HRTEM to calculate curvature values. . . . . . . . . . . . . . . . . 49
4.7 Radii distribution for the oxidized Euro IV sample. . . . . . . . . . 50
4.8 Radii distribution for the oxidized Euro VI sample. . . . . . . . . . 51
4.9
Overview and magnified HRTEM images of as-received (a) and oxi-
dized(b)GfGSoot............................ 53
4.10
Overview and magnified HRTEM images of as-received (a) and oxi-
dized(b)EuroVISoot. ........................ 54
4.11
Overview and magnified HRTEM images of as-received (a) and oxi-
dized(b)EuroIVSoot. ........................ 55
4.12
Overview and magnified HRTEM images of as-received (a) and oxi-
dized(b)Graphite............................ 56
4.13 Curvature distribution of the as-received and oxidized GfG sample. 57
4.14
Curvature distribution of the as-received and oxidized Euro IV sample.
58
4.15
Curvature distribution of the as-received and oxidized Euro VI sample.
59
5.1
Low loss region of the Euro IV, VI and GfG samples compared to
HOPG.................................. 62
5.2
Zero and low loss region of the Euro IV, VI and GfG samples compared
to HOPG after removal of multipe scattering . . . . . . . . . . . . . 63
5.3
Low loss region of the as-received(20) / oxidized(500) Euro IV sample
with the difference spectrum inserted . . . . . . . . . . . . . . . . . 64
5.4
Low loss region of the as-received(20) / oxidized(500) Euro VI sample
with the difference spectrum inserted . . . . . . . . . . . . . . . . . 65
5.5
Low loss region of the as-received(20) / oxidized(500) GfG sample
with the difference spectrum inserted . . . . . . . . . . . . . . . . . 65
5.6
C K-edge spectra of the as-received and oxidized Euro IV, VI and GfG
samples compared to HOPG measured under magic angle conditions 68
5.7
Spectra of HOPG with 3 Gaussian curves used for the fitting of the
hybridizationratio ........................... 68
6.1
C K NEXAFS of HOPG, NC 3100, Flammruss 101, GfG, Euro IV
andEuroVI............................... 74
6.2
C K NEXAFS of the as received, oxidized Euro IV sample and the
difference spectra of the untreated and oxidized Euro IV sample . . 77
6.3
C K NEXAFS of the as received, oxidized Euro VI sample and the
difference spectra of the untreated and oxidized Euro VI sample . . 77
6.4 Difference spectra of the untreated Euro IV and Euro VI sample . . 78
6.5 Difference spectra of the oxidized Euro IV and Euro VI sample . . . 79
6.6
Carbon NEXAFS and EELS spectra of the samples under investigation.
81
iv
List of Figures
6.7
O K NEXAFS of HOPG, Flammruss 101, NC 3100, GfG, Euro IV
andEuroVI............................... 82
6.8
O K NEXAFS for the as received and oxidized Euro IV and VI samples.
83
6.9
O K NEXAFS spectra of the as received/oxidized Euro IV sample
and the difference spectrum. . . . . . . . . . . . . . . . . . . . . . . 84
6.10
O K NEXAFS spectra of the as received/oxidized Euro VI sample
and the difference spectrum. . . . . . . . . . . . . . . . . . . . . . . 85
6.11 Difference spectra of the Euro IV and VI samples. . . . . . . . . . . 86
7.1 C1s XPS of HOPG, Flammruss 101, NC 3100, Euro IV and Euro VI 92
7.2
C1s XPS of the as received(20
◦
C) and oxidized(500
◦
C) Euro IV sample
94
7.3
C1s XPS of as received(20
◦
C) and oxidized(500
◦
C) and Euro VI sample
94
7.4 Difference spectra of the as received vs oxidized Euro VI sample . . 95
7.5 Difference spectra of the as received vs oxidized Euro IV sample . . 95
7.6 O(1s) XPS of Flammruss 101, NC 3100, GfG, Euro IV and Euro VI 98
7.7 Universal Curve of electron mean free path vs electron energy. . . . 102
7.8
Evolution of the absolute amount of the oxygen species for the as
received (a) and oxidized (b) Euro IV sample with increasing photon
energy................................... 105
7.9
Evolution of the absolute amount of the oxygen species for the as
received (a) and oxidized (b) Euro VI sample with increasing photon
energy................................... 106
7.10
Evolution of the absolute amount of the oxygen species for the as
received and oxidized Euro IV (a) and Euro VI (b) sample with
increasing photon energy. . . . . . . . . . . . . . . . . . . . . . . . . 107
8.1 Fitting of Raman spectrum. . . . . . . . . . . . . . . . . . . . . . . 111
8.2
Mass conversion vs temperature for GfG, Euro IV/VI soot and Graphite.
113
8.3 Reaction sketch for the Euro VI sample . . . . . . . . . . . . . . . . 116
8.4 Reaction sketch for the Euro IV sample . . . . . . . . . . . . . . . . 117
v
List of Tables
2.1 Carbonisomers............................. 11
2.2
EU Emission Standards for HD Diesel Engines, g/kWh (smoke in
m−1
)
18
4.1
Radii distribution of the as-received and oxidized the primary Euro
IVandVIparticles............................ 50
4.2
Curvature and length values for the as-received and oxidized GfG,
Euro IV and VI particles . . . . . . . . . . . . . . . . . . . . . . . . 52
5.1
Peak position of the major features in the samples under investigation;
before and after removal of plural scattering contributions . . . . . 66
5.2 Calculated sp2hybridization from the acquired EELS spectra . . . 69
6.1 NEXAFS C K assignment according to Lehmann et al. . . . . . . . 72
6.2 NEXAFS O K assignment. . . . . . . . . . . . . . . . . . . . . . . . 73
6.3 NEXAFS C K assignment according to Braun et al. . . . . . . . . . 73
7.1
Assigment for the functional groups used in the fits for the investigated
samplesinthiswork .......................... 91
7.2 FWHMtable .............................. 93
7.3 XPSO1sassignment .......................... 97
7.4
Oxygen and carbon concentration in Euro IV, Euro VI, NC 3100 and
Flammruss101 ............................. 99
7.5
Absolute oxygen abundance for the individual contributions in the
Euro IV and Euro VI samples . . . . . . . . . . . . . . . . . . . . . 100
7.6
Absolute oxygen abundance calculated from the depth profile mea-
surements. Values on the right side reflect more bulk sensitive, while
the left side values are determined from a more surface sensitive
measurement............................... 101
7.7
Absolute oxygen abundance for the individual contributions in the
Euro IV and Euro VI samples with increasing Eexc ......... 104
vii
1 Introduction
1.1 The Problem of Soot
The formation and emission of soot by combustion processes have concerned scientists
and engineers for a long time [1]. Already in 1972 it was discussed in the scientific
community that "...from the point of view of smoke emission from the stack however
it is vitally important that all the soot particles should be burned before the
temperature of the flame drops too low and quenches combustion..." [1]. Such
aerosol particles are of central importance for atmospheric physics and chemistry,
climate and public health. In urban areas, soot particles emitted by diesel engines
account for a major fraction of air pollutants. Its not only that the soot emission
from combustion applications is a sign of poor combustion conditions and efficiency
losses, but also that those emissions are associated with carcinogenic polycylcic
aromatic hydrocarbons (PAHs)[2]. It has been shown that the incidence of lung
cancers increased strongly between 1970 and 1990 (Figure 1.1) which is partly
attributed due to diesel soot particle interaction with the lymphatic system[3].
It has been studied extensively how the bio-organism / human health is affected
by exposure to soot samples in terms of carcinogenicity [4, 5]. Ultrafine soot particles
are able to penetrate the lung walls inducing inflammation deep into the alveolar
region of the lungs [6
–
8], where they may slow down the clearance mechanisms
and provide absorption sites for toxic pollutants. Ultrafine particles, smaller than
100 nm, have been demonstrated to have enhanced toxicity per unit mass inversely
proportional to the particle size [9].
Recently it has been demonstrated that the environmental accessibility not only
depends on the particle size but also on the structure of the soot particles. Different
combustion conditions have a strong impact on the structure of the soot particles.
The accessibility to the environment has been the topic of several studies, in which the
1
1 Introduction
Figure 1.1:
Anual death rates for different cancer sites in the USA between
1930-1989 [3].
reactivity and absorption properties of various substances towards soot particles, and
their behavior in the atmosphere is monitored [10
–
12]. Additionally, studies relating
to the structural properties of environmentally relevant carbon materials have arisen
in the last few years [13
–
20]. However no systematic study of the surface and bulk
properties of soot particulates with respect to their reactivity is reported in literature.
Usually research focuses either on the bulk structural characterization provided
by high resolution transmission electron microscopy (HRTEM) [20, 21], electron
energy loss spectroscopy (EELS) [21, 22], Raman [23] or on the determination of
the surface species by means of X-ray photoemission spectroscopy (XPS) or near
edge X-ray absorption fine structure (NEXAFS) spectroscopy [13, 15, 24–26].
In order to evaluate the potential environmental accessibility of soot samples,
the aim of this work is to reveal the structural features relevant for the reactivity
of soot. The approach to achieve this goal consists of a combination of structural
information obtained from bulk and surface characterization techniques applied
towards the detailed characterization of as received soot samples as well as soot
samples subject to oxidation treatments. The structural modifications that the
soot samples undergo upon oxidation give further insight into the reactivity and
environmental accessibility of the samples.
2
1.1 The Problem of Soot
Such an approach allows us to draw a complete picture of the soot samples and
to propose a hypothetical structure for the soot samples. Furthermore a possible
post treatment is proposed that can alter the environmental accessibility of the soot
samples.
Chapter 2 presents a short introduction on the formation of soot and a general
review of carbon materials.
Chapter 3 describes the techniques applied in this work, such as TEM, EELS,
NEXAFS and XPS.
Chapter 4 describes the structural investigations on bulk properties of the soot
samples by means of SEM and HRTEM.
Chapter 5 illustrates the electronic structure studies carried outby means of EELS
for the samples under investigation.
Chapter 6 presents the surface investigations by means of NEXAFS and correlates
them with EELS measurements of the bulk of the investigated samples.
Chapter 7 presents the investigation of the surface functional groups by means of
XPS and reveal modifications in the composition of the functional groups.
Chapter 8 correlates the results obtained from the various techniques with each
other.
Chapter 9 reports the main results from this work.
3
2 Structure of soot
2.1 Soot formation
Soot is formed during high temperature pyrolysis or combustion of hydrocarbons
and consists of carbon with small amounts of other elements such as hydrogen
and oxygen. Soot particles show a clear dependency in their reactivity from their
microstructure. It is important to study the soot formation due to the fact that the
microstructure depends on its synthesis conditions.
According to literature [2, 27], the first step in the production of soot takes place
by formation of aromatic species from aliphatic hydrocarbons in the fuel. This
process involves the conversion of primary molecular species which include free
radicals and ions into relatively larger particles containing many thousands of atoms
and therefore much greater
C/H
ratios, through condensation or addition reactions.
Those species are referred as polyaromatic hydrocarbons (PAHs) [2, 27]. PAHs
are two-dimensional carbon-rich molecules. Naphthalene is considered to be the
smallest, while larger PAHs can have molecular formulas such as
C320H50
. These
PAHs are in the most common case built up of hexagonal benzene rings [28]. Soot is
only formed - from a thermodynamic approach - when m becomes larger than 2y in
CmHn+yO2−→ 2yCO +n
2+(m−2y)Cs
with
Cs
being solid carbon. This means that the C/O ratio in this case exceeds
unity. Experiments show that limits of soot formation are usually put on a level
with the onset of luminosity, which does not occur at C/O
≈
1but near to a region
of C/O
≈
0
.
5[2]. Thus, it must be concluded that soot formation is kinetically
controlled.
Following the formation of primary PAHs the system can undergo different reaction
pathways depending on the conditions such as temperature and oxygen content in
the flame. Lower temperature and oxygen enriched atmosphere favor soot formation
5
2 Structure of soot
Figure 2.1: Growing steps for soot and fullerenes via aromers [28].
while high temperature is required for fullerene formation [28]. The existence of 5
membered rings being bordered by benzene rings is the necessary step to the later
formation of fullerenes.
Figure 2.1 shows paths to fullerenes and soot via aromers. While the formation
of fullerenes from aromers occurs mainly through unimolecular reactions at high
temperatures, soot particles are formed in bimolecular addition reactions of small
unsaturated hydrocarbons (e.g.,
C2H2
) at lower temperatures. Small fullerenes
could possibly be formed in very small quantities by the oxidative degradation of
larger fullerenes [28].
While Figure 2.1 shows the nucleation steps, Figure 2.2 presents a schematic
representation of the growth steps for soot formation, including later steps of
coagulation, surface growth and particle aggregation.
Continuing PAH growth leads to soot particles with a diameter of roughly one nm.
In this process the first step is formation of particle-like structures by conglomeration
of molecules [30]. This particle inception takes place at molecular masses between
500 and 2000 a.m.u. The growing process may occur as a combination of surface
growth, due to pyrolysis processes that produce smaller HCs, in particular acetylene
[29, 31], and coagulation. Alternatively the process is assumed to take place by
dehydrogenation to atomic carbon or perhaps to
C2
radicals, which then condense
to solid carbon. Alternatively a large hydrocarbon molecule is formed initially and
then hydrogenated to give soot [32].
6
2.2 Carbon
Figure 2.2: Schematic mechanism of the formation of soot particles [29].
Figure 2.3: Molecules relevant for the combustion process [33].
During the formation process a variety of products can be formed which differ in
their carbon-, oxygen- and hydrogen-content. Therefore soot is characterised by a
distribution of the molar masses of the particles formed and by a different surface
chemistry of the soot particles. Aromatic domains are connected by aliphatic chains
and coagulate with each other to form the soot particles. A list of molecules relevant
in combustion is given in Figure 2.3. Those species can be considered to coexist
as carbon unit termination on the surface of soot particles and their probability
depends on the combustion condition in the flame.
2.2 Carbon
Carbon atoms possess many possible configurations of their electronic states. The
resultant configurations of atomic orbitals interact with one another to form what
are known as hybrid bonding orbitals. These hybridized orbitals are useful in
7
2 Structure of soot
explaining the molecular bonding orbitals which give rise to the structure and shape
of carbonaceous materials. Carbon is the 6th element in the periodic table, sitting in
column IV. Each carbon atom has six electrons that occupy
1s22s2and 2p2
atomic
orbitals. The core electrons are the 2 strongly bound electrons filling up the
1s2
orbitals. The 4 electrons in the
2s2
2p2
orbitals are more weakly bound and called
valence electrons.
The valence electrons in the crystalline phase fill up the
2s2
2px 2py 2pz
orbitals
which form covalent bonds in carbon materials. A mixing of the 2s and 2p atomic
orbitals is called hybridization, specifically, the mixing of a single 2s electron with
n=1,2,3 2p electrons is referred to as
spn
hybridization [34]. This is possible due to
the fact that the energy difference between the upper 2p energy levels and the lower
2s level is small compared to the binding energy of the chemical bonds, making it
possible for the electron wave functions of the valence electrons to mix with each
other. This mixing changes the occupation of 2s and 2p orbitals to enhance the
binding energy of the carbon atoms with their neighboring carbon atoms. Therefore
three different hybridizations are possible in carbon, sp, sp2and sp3.
2.2.1 sp hybridization
sp hybridization is characterized by a linear combination of the 2s orbital and one
of the 2p orbitals. This leads to an elongation of the wavefunction in the direction
of the participating 2p orbital. The stronger overlap of the wavefunction gives rise
to an higher binding energy. Acetylene (
HC ≡CH
) is a typical example for sp
hybridization, with a carbon-carbon triple bound.
2.2.2 sp2hybridization
The 2s and two 2p orbitals form the basis of
sp2
hybridization. An example of this
configuration is polyacetylene (
(CH−
−CH−)n
), with the carbon atoms arranged in a
zig-zag chain with a C-C angle of 120
◦
. The 3
sp2
orbitals form sigma bonds directed
outwards in a trigonal planar geometry, with the remaining
π
orbital extending
above and below the plane available for πbonding via 2p-2p overlap.
8
2.2 Carbon
Figure 2.4:
Families of polycrystalline carbon materials relevant for surface chem-
istry [35].
2.2.3 sp3hybridization
Methane is an example for
sp3
hybridization characterized by tetrahedral
σ
-bonding
to its four nearest hydrogen neighbors. Elongated wavefunctions are realized here by
an overlap of the 2s and the 3 2p orbitals. Generally spoken in a
spn
hybridization,
n+1 electrons belong to the carbon atom occupied in the hybridized
σ
orbital and
4-(n+1) electrons are in
π
orbital.
sp3
hybridization is realized with the four valence
electrons occupying 2
s1
and 2
p3
states as
σ
bonding states. To excite the 2
s2
2p2
solid ground state to 2
s1
and 2
p3
an energy of the difference between the 2s and 2p
( 4 eV) levels is required. Nevertheless the covalent bonding energy for
σ
-orbitals
( 3-4 eV per bond) is larger than the 2s-2p energy separation.
2.2.4 Chemistry of Carbon
Carbon chemistry is quite complex, exhibiting a wide range of nanostructured and
macroscopic varieties. These porous and non porous forms exceed the historical
definition of dual allotropes of graphite and diamond [35, 36]. Figure 2.4 shows a
scheme for the various nanostructured carbon allotropes.
9
2 Structure of soot
Figure 2.5: Variations of Carbon: Graphite, Diamond, C60 and Nanotube.
Carbon occurs in sp,
sp2
and
sp3
atomic hybridization. Its surface chemistry
is strongly connected to chemical bonding within the bulk. The presence of
sp2
carbon configuration leads to planar graphene layers, assuming that the structure
contains only the energetically most stable six-membered rings. This can lead to
well-ordered graphene layers (graphite) or turbostratically (normal rotated graphene
layers) disordered stacks of graphene layers. If the synthesis conditions allow for
the presence of less stable, non six-membered rings, strained three-dimensional
variations (Figure 2.5) of
sp2
hybridized carbon such as carbon onions, fullerenes
and nanotubes will be formed. This deviations from planarity, later referred to
as "curvature", increases the reactivity of those carbon allotropes significantly. A
structural deviation, such as a change in curvature of the graphene layers will result
in a vast variation of the functional group properties [35]. In the case of bent
sp2
carbons, all surface atoms are potential binding sides for heteroatoms. In contrast,
to that planar
sp2
bonded carbon offers only anchoring points on the zigzag and
armchair faces of the graphene layers. Due to the "semi-metallic" / aromatic chemical
bonding of
sp2
hybridized carbon no chemical functionalization will occur on the
basal planes at the surface of graphitic materials. Therefore surface chemistry on
sp2
carbon involving covalent bond formation can only occur on defect sides [37, 38]
or on prismatic faces.
The crystal structure of graphite is hexagonal with a stacking of honeycomb
carbon hexagon planes in a sequence of ABAB, fulfilling a P
63
/mmc symmetry.
Values for the lattice constants were first fixed by Kelly [40] with
a0
= 0.2462 nm
10
2.3 Functional Groups on Carbon Surfaces
Table 2.1: Carbon isomers [39]
Dimension 0-D 1-D 2-D 3-D
isomer C60 nanotube graphite diamond
fullerene carbyne fiber amorphous
hybridization sp2sp2(sp) sp2sp3
density 1.72 1.2-2.0 2.26 3.515
[g/cm3]2.68-3.13 22-3
Bond Length 1.40(C=C) 1.44(C=C) 1.42(C=C) 1.54(C-C)
[Å] 1.46(C-C) 1.44(C=C)
electronic semiconductor metal or semimetal insulator
properties Eg= 1.9eV semiconductor Eg= 5.47eV
and
c0
= 0.6707 nm at room temperature. This leads to an in-plane bond length of
0.1421 nm and an inter-plane spacing of 0.3354 nm. Alternative to ABAB stacking,
ABCABC stacking sequence is also observed resulting in a rhombohedral unit cell.
This forms is always present together with the hexagonal symmetry. Turbostratic
graphite is defined by a random stacking of graphene layers ideally formed of a
two-dimensional structure with non-interacting layer planes. In this case the lattice
cell constant of graphite with
c0
= 0.6707 nm is not longer valid. The value for
the interlayer separation of 0.3354 nm in ideal graphite expands to > 0.344 nm for
turbostratic carbon.
If individual carbon layers were arranged in a completely random manner with
respect to rotation about the axis perpendicular to the planes, while remaining
parallel to each other (turbostratic carbon), the reciprocal lattice points for this
two-dimensional structure would be circles, with indices (h,k), except for the (00l)
reflections which would remain discrete spots [34].
2.3 Functional Groups on Carbon Surfaces
As already shown in Figure 2.3 a large variety of molecules are involved in the
combustion process. As these reactant molecules are rearranged to become product
molecules, different soot samples under different combustion conditions result in
a wide distribution of final products. Therefore it is important in this work to
determine the composition of surface functional groups present on the samples
subject of interest.
11
2 Structure of soot
The following section will briefly summarize the functional groups present on
carbonaceous materials taking into account those hetero-atoms such as H, O and N.
Oxygen surface groups have the strongest effect on the physico-chemical properties
such as: wettability, catalytic, electrical and chemical reactivity of carbon blacks
[32]. These carbon-oxygen surface compounds are the source of the property by
which a carbonaceous material becomes useful or effective in certain respects. The
amount and the nature of the surface oxide being formed, depends on the nature
of the carbon surface, its formation history, its surface area and the treatment
temperature.
Beside oxygen, hydrogen is also present in most of the carbons. This is expected
as carbons are pyrolyzed residues or organic compounds. The carbon-hydrogen
surface compounds are more stable than carbon-oxygen surface compounds. The
hydrogen is adsorbed as atoms already at room temperature although the rate
of chemisorption is measurable around 200
◦
C. It has been shown by Bansal et al.
[41] that the chemisorption of hydrogen takes place at carbon atoms situated at
the edges and corners of the graphitic crystallites and that the chemisorption of
hydrogen interstitially between the basal planes does not occur significantly. Bansal
et al. [41] showed moreover that the chemisorption of hydrogen involves different
sites connected to varying activation energies. The hydrogen in carbon is present as
chemisorbed water and also as part of hydroxyl, phenol and hydroquinone groups.
Additionally it is bonded to carbon atoms in the form of C-H bonds.
Usually untreated carbon samples (e.g. carbon black) do not contain significant
amounts of nitrogen moieties. However by the use of gas-phase reaction, substantial
amounts of carbon-nitrogen surface moieties can be introduced. The carbon-nitrogen
surface species are stable to thermal desorption. It is necessary to apply heat
treatment at around 900–1200
◦
C to desorb nitrogen, which is observed mostly as
free nitrogen with smaller amounts of HCN, cyanogens and ammonia [24].
From the point of view of reactivity, functional groups containing oxygen atoms
are the most studied and most important ones. Therefore, the generation process of
oxygen surface functional groups will be presented subsequently in greater depth.
12
2.3 Functional Groups on Carbon Surfaces
2.3.1 Generation of Surface Oxygen Functional Groups
The generation of oxygen functional groups on carbon can be considered intermediate
states of the oxidation process from carbon to CO/CO2[42].
The carbon forms are metastable against exposure to oxygen-containing gases and
oxidising agents such as nitric acid, sulphuric acid, hydrogen peroxide etc. These
reagents and gases including dioxygen, ozone, nitric oxide,
CO2
and steam can be
used to form oxygen functional groups. The oxidation reactions occur in two steps
[43] as shown below:
The first step is the reductive activation of the reagent by producing an oxygen
di-anion.
[Cx] + R−O→[Cx]2+ +O2−+R
This can occur by chemisorption and reductive dissociation at the basal planes
of
sp2
carbon [44] or by direct activation of a radical centre formed by dangling
bonds of
sp2
/
sp3
carbon atom at the surface. These centers are very reactive and
therefore they will be only present after aggressive activation, as shown in [45], of
carbon in extremely inert environments or during in-situ gasification and during
formation of solid carbon from atomic carbon sources.
As shown in [46], the direct activation reaction occurs like:
−C∗+O2→ −C−O+O∗(2.1)
−C−C+O∗→CO +−C∗(2.2)
−C−C−O+O∗→ −CO2+C∗(2.3)
The carbon is modified by oxygen in an adsorption-reduction sequence as shown in
[47]. Dioxygen as reagent for the adsorption-reduction process occurs like:
Cbasal +O2→C−O2ads (2.4)
C−O2ads + 2e−→C−O22−
ads (2.5)
C−O22−
ads + 2e−→Cbasal + 2O2−(2.6)
The electrons that are used in that process come from the conduction band of the
graphite valence electrons. Therefore diamond, or prismatic faces of
sp2
carbon
cannot be used as a substrate for oxygen activation.
13
2 Structure of soot
The second reaction is the diffusion of activated oxygen dianion to a site of
covalent bond formation, which must be a prismatic edge/defect site in
sp2
carbons.
When the initial C-O bond is formed, several possibilities exist [48] for further
reactions. These are controlled by the boundary conditions and reaction kinetics.
•If the temperature is too high the C-O complex will desorb as CO.
•
At too high temperature and under excess of activated oxygen a
CO2
group
will form and desorb as carbon dioxide.
•
At intermediate temperature a carbon oxygen group with low disturbance of
the graphene layer will form.
•At very low temperature a complex carbon oxygen group with strong distur-
bance of the graphene layer will form.
Reaction temperatures and abundance of activated oxygen are important param-
eters determining the chemical nature of resulting functional groups as well as the
specific ease with which a given carbon-oxygen functionality is formed. Hence it
is observed that structurally inhomogeneous carbon surfaces reacting with oxygen
at temperatures below that required for gasification result in a wide variety of
oxygen-containing moieties. The result of treatment above these temperatures is
the titration of reactive surface sites with oxygen species and subsequent ’cleaning’
by gasification.
The surface chemistry of carbon is dominated by the consequences of oxygen
bonding to free defect sites. A large number of
sp2
defect sites and of most
sp3
defect sides are saturated by hydrogen atoms [49] which desorb as alkanes at
temperatures above 1400 K. Most of the carbon samples that were prepared by
combustion techniques or that have been exposed after synthesis to ambient air
carry a sufficient number of carbon-oxygen functions. A significant interaction with
polar molecules occurs rendering them hydrophilic and allowing for aggregation to
complex superstructures.
The extent of hydrophilic interaction will depend on the abundance and chemical
constitution of the carbon-oxygen functions, which is determined by the local defect
density of the carbon.
14
2.4 Materials studied
2.3.2 Structure of Carbon-Oxygen Functional Groups
As already discussed before the surface chemistry of carbon is dominated by the
consequences of oxygen bonding to free defect sites [42]. The majority of
sp2
and
sp3
defective sides are saturated with hydrogen atoms [49] and therefore do not play
a significant role in the surface chemistry of carbon. Those passivated hydrogenated
sides [41] are together with the basal graphene surface contribute significantly to
the hydrophobicity of carbon.
Most carbon samples prepared by combustion techniques, such as soot, or exposed
to atmosphere after synthesis, such as functionalized carbon nanotubes, contain a
sufficient number of carbon-oxygen functionalities rendering the surface hydrophilic.
The polar nature of the oxygen functionality increases the hydrophilic nature of the
surface sufficiently to interact with water and polar surface species [50]. During
thermal treatment such functionalities are involved in aggregation to more complex
superstructures, as as observed for soot [51].
Structures of oxygen functional groups are displayed in Figure 2.6. Carboxylic
and acid anhydride functional groups react acidic in water. Those structures contain
the carbon dioxide molecular fragment already as excellent and thermodynamically
highly preferred leaving groups and therefore it is not surprising that these most
reactive groups are also the least stable functions [42].
Depending on the abundance and density of neighboring groups desorption of
carbon dioxide from these functionalities occurs from 300 K up to approximately
723 K. Phenolic and carbonyl groups (quinoid structures if resonance-stabilized with
the aromatic n-electrons) are only weakly acidic against water and require stronger
bases than water for their safe identification. The thermal stability and thus their
abundance is much higher than those of the strongly acidic functions. Thermal
desorption leads to CO and requires temperatures up to 1273 K for complete removal.
Lactones and other ether functions react moderately acidic, in particular if they are
adjacent to other oxygen functional groups.
2.4 Materials studied
In this study the focus of the work is to compare the microstructure and functional
properties of two emission standard samples, namely Euro IV and Euro VI. Euro
15
2 Structure of soot
Figure 2.6:
Schematic representation of the structure of chemical active oxygen
functional groups
IV soot samples were collected under World Harmonized Transient Cycle conditions
(WHTC) from the undiluted raw exhaust of a heavy duty test engine that fulfills
Euro 4 exhaust limits. Samples of Euro VI soot were taken under European Transient
Cycle conditions (ETC) from the undiluted raw exhaust of a heavy duty test engine
that is designed to fulfill Euro 6 exhaust limits. More detailed information on
those samples will be provided subsequently. The samples were compared with a
reference sample from TU Munich, called GfG soot. GfG soot is genereated by a
spark discharge aerosol generator (GfG 1000, Palas, Germany) and in the context
of this project it is intended to be an upper reactivity limit of the samples under
investigation.
The second standard used in this study is a HOPG (Highly oriented pyrolytic
graphite) sample from Mateck/Germany(HOPG ZYB). Additionally, during this
study the Euro samples are compared to other carbon samples such as Flammruss
101 and Nanocyl 3100 nanotubes, to compare the influence of different structural
order and functional group content. All soot samples were deposited on Bekipor
Fecralloy metal fibre filter material (Bekaert, Belgium) with a fibre diameter of
≈10 µm and a filter diameter of 47 mm.
16
2.4 Materials studied
2.4.1 Emission Standards
European emission standards define the acceptable limits for exhaust emissions of
new vehicles sold in the European Union member states. The emission standards are
defined in a series of European Union directives staging the progressive introduction
of increasingly stringent standards.
Currently, emissions of Nitrogen oxides (NOx), Total hydrocarbon (THC), Non-
methane hydrocarbons (NMHC), Carbon monoxide (CO) and particulate matter
(PM) are regulated for most vehicle types, including cars, lorries, trains, tractors
and similar machinery, barges, but excluding seagoing ships and aeroplanes. For
each vehicle type, different standards apply. Compliance is determined by running
the engine at a standardized test cycle. European emission regulations for new
heavy-duty diesel engines are commonly referred to as Euro I to VI.
The emission standards apply to all motor vehicles with a "technically permissible
maximum laden mass" over 3.500 kg, equipped with compression ignition engines or
positive ignition natural gas (NG) or LPG engines. The regulations were originally
introduced in 1992 with the Euro I standards, followed up by Euro II regulations in
1996. Euro III standards were introduced starting from 2000, as well as Euro IV/V
standards, valid from 2005/2008. Euro VI emission standards were introduced in
2009, and will become effective in 2013/2014.
TableTable 2.2 contains a summary of the emission standards and their imple-
mentation dates. Dates in the tables refer to new type approvals; the dates for all
type approvals are in most cases one year later. Information about the test cycles
used for those emission standards can be found elsewhere [52].
17
2 Structure of soot
Table 2.2:
EU Emission Standards for HD Diesel Engines, g/kWh (smoke in
m−1
)
Tier Date & category CO HC NOx PM Smoke
Euro I 1992 < 85 kW 4.5 1.1 8.0 0.612
1992 > 85 kW 4.5 1.1 8.0 0.36
Euro II 1996.10 4.0 1.1 7.0 0.25
1998.10 4.0 1.1 7.0 0.15
Euro III 1999.10 EVVs only 1.5 0.25 2.0 0.02 0.15
2000.10 2.1 0.66 5.0 0.10 0.8
Euro IV 2005.10 1.5 0.46 3.5 0.02 0.5
Euro V 2008.10 1.5 0.46 2.0 0.02 0.5
Euro VI 2013.01 1.5 0.13 0.4 0.01
18
3 Characterization techniques
In this chapter the most important characterization techniques used in this work
are briefly presented. In the first section transmission electron microscopy (TEM)
is discussed and the following sections deal with electron energy loss spectroscopy
(EELS), near edge x-ray absorption fine structure (NEXAFS) and x-ray photoemis-
sion spectroscopy (XPS).
3.1 Electron Microscopy
Electron microscopy characterization techniques as they are used in a SEM (Scanning
Electron Microscopy) and TEM (Transmission Electron Microscopy) are powerful
tools for the local investigation of the morphology, size distribution, electronic
structure and nanostructure of carbonaceous materials.
Figure 3.1 shows the different signals that are generated by the interaction of a
fast electron with a thin sample. Signals detected in a SEM and TEM are marked
accordingly. When electrons interact with a target material, they can scatter
elastically or inelastically. The elastically scattered electron can be forwarded
through the sample or backscattered. Inelastically scattered electrons transfer some
of their kinetic energy depending on the type of interaction, to the emission of a
photon (X-ray emission), an electron (Auger-electron), or simply, the generation
of heat. In contrast to light microscopes the obtainable resolution using TEM is
not limited by the wavelength of visible light (400-700 nm), as the probing entities
are no longer photons but electrons. The resolution is determined by the Rayleigh
formula [53]
rR=0.61λ
µsinβ (3.1)
with
µ
being the refracting index of the viewing medium and
β
being the semiangle
of collection. The resolution can therefore be increased in a TEM by varying the
19
3 Characterization techniques
Figure 3.1:
Signals generated by the interaction of a high energy electron with a
thin specimen [54].
wavelength (0.00251 nm for 200kV and 0.00197 nm for 300 kV) which is achieved
by changing the acceleration voltage of the incident electrons.
3.1.1 Transmission Electron Microscopy
A TEM provides highly coherent electrons of sufficient energy and the ability to
investigate samples in terms of morphology, structure and orientation, combined
with the advantage of high spatial resolution. Figure 3.2 shows an illustration of the
general setup of a TEM used for the study of the morphology and microstructure
of various samples. The electron source is usually a field emission gun (FEG) or
LaB6 crystal which works as a cathode from which the electrons are extracted to an
anode. Depending on the working mode of the TEM, different extraction voltages
are used (e.g. 3.9 keV for TEM, 4.5 keV for STEM). In this case the first anode
is used to pull out the electrons and the second one is used to accelerate them to
the desired energy (80, 200, 300 keV). Electrons get emitted by the electron source
20
3.1 Electron Microscopy
Figure 3.2:
Schematic path of an electron in a TEM equipped with X-ray and
Electron Energy Loss Spectroscopy detector.
and are accelerated and pre-focused by the electron gun. By the use of condenser
lenses a parallel beam is formed that interacts with the specimen. After passing the
illumination system the electron beam interacts with the sample positioned within
the objective lens and subsequently goes through the intermediate lenses and the
projection lens. Below this setup the fluorescent screen is situated to render the
image visible.
The advantage of a FEG lies in the highly coherent electron beam which is
necessary for high resolution TEM (HRTEM). The instrument resolution can further
be enhanced by a
Cs
-corrector for the reduction of spherical abberations of the
objective lens until atomic resolution is achieved.
Figure 3.3 shows the two operation modes of the microscope: The diffraction
mode generates a diffraction pattern of the probed specimen on the viewing screen,
21
3 Characterization techniques
Figure 3.3:
The two basic operation modes of the TEM imaging system consist of
(A) diffraction mode: projecting the diffraction plane onto the viewing screen and
(B) image mode: projecting the image onto the screen. In each case the intermediate
lens selects either the BFP (A) or the image plane (B) of the objective lens as its
object. NB: simplified diagram that shows only 3 lenses [54].
22
3.1 Electron Microscopy
whereas the imaging mode projects an image of the specimen. Varying the strength
of the excitation of the intermediate lenses shifts between these two modes.
In diffraction, the back focal plane of the objective lens is used as the objective
plane of the first intermediate lens. In imaging mode the object of the intermediate
lens is the image plane of the objective lens. Digital image recording can be realized
by coupling a CCD camera to the imaging filter (e.g. Gatan Tridiem Filter) which
is used to display the variations in intensity of the transmitted electron beam. TEM
investigations are limited to very thin samples (< 200 nm) due to the fact that
electrons interact strongly with matter and the electron beam has to be transmitted
through the sample.
Different types of contrast are observed for a transmission electron microscope.
"Contrast" means the appearance of features in an image. Contrast in bright field
(BF) and dark-field (DF) TEM images is usually "diffraction contrast", or variations
in intensity of diffraction across the sample. Diffraction contrast and the appearance
of features in BF and DF images depend sensitively on how the Laue condition
(∆
k
=
g−s
) is satisfied. The direction of
k
is adjustable by tilts and
g
is a
reciprocal lattice vector of the crystal. The alternative "mass thickness contrast"
(Figure 3.4) is generally weaker and overshadowed by the stronger effects of electron
diffraction, except in cases where there are large differences in atomic number or
when diffraction is weak.
"Phase contrast" and "Z-contrast"("high-angle annular dark-field imaging" (HAADF))
offer better spatial resolution than conventional HRTEM imaging methods, but
require more sophisticated instruments, operator skill, and usually more interpreta-
tion. HRTEM phase contrast imaging uses coherent elastically-scattered electrons
while HAADF ("Z-contrast") imaging is formed from incoherent elastically-scattered
electrons. For incoherent scattering the intensities, I, from individual atoms, rather
than the wavefunction amplitudes,
ψ
are summed. Phase differences and inter-
ferences that are central issues for HRTEM imaging are irrelevant for HAADF
imaging. Each atom can be considered an independent scatterer because there is
no constructive or destructive interference between the phases of wavefunctions
emanating from the different atoms[55]. The incoherent images of the HAADF
method are interpreted more directly in terms of atom types and positions.
23
3 Characterization techniques
Figure 3.4: Mechanism of mass-thickness contrast in a BF image[54].
3.1.2 Electron Energy Loss Spectroscopy
Electron Energy Loss Spectroscopy (EELS) is a widespread and established technique
for the characterization of the electronic structure and chemical bonding in materials
[56, 57]. EELS can be applied with several techniques, but always involves the
bombardment of a sample with a monoenergetic beam of electrons. The technique
is frequently used in association with Transmission Electron Microscopy (TEM) or
Scanning Transmission Electron Microscopy (STEM). The electrons impinging on
the sample can lose energy by a variety of mechanisms. These losses can provide
information regarding the composition of the sample in TEM or STEM. Plasmon
losses are a frequent cause of energy loss. Plasmons are collective excitations of the
electron gas in the material and are typically several electron Volts in magnitude.
Phonon losses can also occur, which are much smaller, meaning that the energy
spread of the monoenergetic beam must be particularly narrow to detect such
losses. Phonons are quantized lattice vibrations within the solid. When the primary
electrons traverse the sample, most of the electrons scatter elastically. However,
a small amount of electrons scatter inelastically and a certain amount of energy
characteristic of the elements present in the sample is lost. This energy can be
24
3.1 Electron Microscopy
Figure 3.5:
Electron energy-loss spectrum for carbon showing the low-loss region
and the core-loss region including the ELNES (energy-loss near edge structure)
region. The two parts were acquired independently but from the same region. The
intensity scale in the core-loss region is multiplied by a factor of 20.
used to excite electrons in the sample to higher energy levels, induce plasmon
oscillations, phonons or optical transitions. Hence, the energy of the electrons
excited by the sample are related to the elements in the sample. Furthermore, the
amount of energy lost is influenced by the chemical environment of excited atom in
the sample. Therefore EELS can detect if e.g. a carbon atom is bonded in
sp2
or
sp3hybridization.
The Electron Energy-Loss Spectrum
The beam of electrons that have traversed the sample is directed into an EELS
detector which separates the electrons according to their kinetic energy and generates
an electron energy-loss spectrum, plotting the scattered intensity (electron counts) as
a function of the loss of kinetic energy of the incident fast electrons. The energy-loss
spectrum of a carbon sample as shown in Figure 3.5, consists of two parts, the
low loss and core loss region. The spectrum is shown as-acquired, i.e. without any
post-processing.
The low loss part is dominated by the zero-loss peak and the plasmon peak. The
zero-loss peak shows the electrons that are transmitted through the sample without
any measureable energy loss, including electrons that were scattered elastically in
25
3 Characterization techniques
the forward direction and those which have excited phonon modes, for which the
energy loss is less than the experimental energy resolution (0.8 eV in this case).
Beside the zero-loss the plasmon peak is often the strongest feature in the low-loss
region consisting of the area between 0 and 50 eV energy loss. The plasmon peak
is a result of the electrons that have interacted with the weakly bound outer-shell
electrons of the atoms in the specimen, such as plasmon oscillations. Plasmons
oscillations are longitudinal wave oscillations of weakly bond electrons that are
induced by the interaction of the ”
electron cloud
”of the sample and the penetrating
electron. The valence electrons in a solid can be thought of as a set of coupled
oscillators; excitation of these oscillators is similar to plasmons in metals.
The second part of the spectrum, the core-loss region, is related to primary
electrons interacting with tightly bond core electrons in the sample. The sharp
peaks are a result of excitations of core electrons to the continuum or unoccupied
states. The ionization losses are characteristic of the atom involved and so the
signal is a direct source of elemental information, just like the characteristic X-rays.
The ELNES (energy-loss near edge structure) region of the spectrum provides
information about the chemical environment and local bonding geometry in the
sample. If the energy loss spectrum is recorded from a sufficiently thin region of
the specimen, each spectral feature corresponds to a different excitation process.
In thicker samples, the spectral shape is more complicated as multiple scattering
events occur and hence the total energy loss is the sum of the individual losses.
Within some limitations, signal due to plural scattering can be removed from the
spectrum by deconvolution techniques [56]. In the case of plasmon scattering, the
result is a series of peaks at multiples of the plasmon energy.
The probability of exciting the electrons in the sample depends on the binding
energy of the electrons. It decreases with increasing energy-loss due to the decrease
of the cross-section of the scattering event. This leads to a decrease in the signal-to-
noise ratio at higher energy losses and therefore longer acquisition times are needed
to measure core-loss spectra at higher energy.
Excitations from core (inner shell) atoms give rise to absorption edges in Energy
Loss Spectroscopy comparable to X-ray absorption edges as displayed in Figure 3.6.
As these edges are present at energies
E
50
eV
for most elements, far beyond the
interaction energy of valence electrons in the solid state, the target electrons can be
26
3.1 Electron Microscopy
Figure 3.6: Energy-level diagram of a solid [56].
treated as bound to a single nuclei. Valence or conduction electrons interact strongly
with one another and their influence is reflected in the low loss region E ≈25 eV.
Figure 3.6 shows a diagram for the excitation process for inner shell atoms as
they are present in solids. K- and L-shell core levels and valence band (shadded)
are displayed.
EF
defines the Fermi level and and
Evac
the vacuum level. Primary
processes of inner- and outer-shell excitations (left side) and secondary processes
such as photon and electron emission (right side) are displayed [56].
The intensity, I, of a spectrum as a function of energy,
E
, and angle, Θ,is given
by Fermi´s golden rule [56].
I(E, Θ) ∝4γ2
a2
0q4Zψi(q·r)ψ∗
fd3r
2
ρ(E)(3.2)
γ
is the relativistic factor,
a0
is the Bohr radius and
q
is the scattering vector.
The expression in brackets refers to the transition matrix and
ρ
is the density of
electron states (DOS). In the transition matrix,
ψi
is the initial core state and
ψf
is
the unoccupied final state.
The shape of the spectrum is determined by the transition matrix, which varies
slightly with the energy. The overlap between
ψi
and
ψf
decreases with increasing
energy-loss resulting in smaller matrix elements. The dipole selection rule ∆
L
=
±
1
determines the possible transitions between
ψi
and
ψf
in the dipole approximation.
The density of states term given by
ρ
varies mainly within the first few eV of
the ionization edge. Therefore the shape of the spectrum in this region, known
27
3 Characterization techniques
as ELNES region, is modified by changes in the DOS. As the core-level states are
highly localized at the site of the excited atom, significant differences in the ELNES
region can be observed due to changes in the chemical environment of the probed
atom.
Dipole Matrix Elements
Due to the high localization of the core level states, the local DOS of the excited
atom is reflected in the ELNES spectra. Core level transitions are restricted by
dipole selection rules and therefore the acquired ELNES spectra corresponds to
a site and angular momentum projected DOS [56]. For an anisotropic solid the
ELNES depends on the crystallographic orientation of the probed sample relative
to the momentum transferred from the incident to target electron. Fine structures
in EELS arise when an inner-shell electron in a state
|ii
undergoes a transition
to an unoccupied state
|fi
. The double differential crossection, differential in
energy loss E and solid angle Ω, for the excitation process has, within the first Born
approximation and the independent electron model, the form [58]
d2σif (Θ)
dΩdE =4γ2
a2
0q4hf|ei~q~r |ii
2δ(Ef−Ei−E)(3.3)
where
~q
=
~
k0−~
k
is the momentum transfer/scattering vector and
a0
is the Bohr
radius. The fast incident electron enters the sample with energy
E0
and wave vector
~
k
and leaves it with the wave vector
~
k0
and energy
E0−E
, with
E
being the energy
loss. The momentum transfer
~q
is related to the scattering angle Ωthough the
conservation of energy and momentum(Figure 3.7)
q2=k2(Θ2+ Θ2
E)(3.4)
where Θ
E
=
E/
2
E0
and
E0
=
¯h2k2/
2
m0
.
ei~q~r
can be developed in a Taylor
series
ei~q~r
= 1 +
~q~r −i~q~r2
+
...
In our study we use conditions fulfilling the dipole
approximation (
|~q~r|
1) and therefore only contributions up to the first order of
ei~q~r
have to be considered. Assuming that
|ii
and
|fi
are orthogonal to each other
(i.e. we are in a single particle picture) the differential cross section becomes:
d2σif (Θ)
dΩdE =4γ2
a2
0q4|hf|~q~r |ii|2δ(Ef−Ei−E)(3.5)
28
3.1 Electron Microscopy
Figure 3.7:
Scattering diagram showing the incident electron momentum
~
k
and
the momentum transfer
~p
for a scattering angle Θ. The minimum(parallel) and
transverse momentum transfer are indicated too [59].
Considering the radial and angular part to be separable, the two different cross-
sections for 1
s→π∗
and 1
s→σ∗
transitions can be calculated separately. They
have to be calculated separately as they correspond to different energy transitions.
Those equations will not be presented and discussed here but can be found elsewhere
[60–64].
EELS Detector
EEL spectra were acquired with a Gatan Tridiem Imaging Filter with an energy
resolution of 0.8 eV (FWHM of the zero loss signal). In Figure 3.8 a schematic
drawing of the used Imaging Filter is shown.
The images, diffraction patterns or spectra can be observed in real-time via a
TV-rate camera or can be acquired by a CCD camera directly into a computer.
Electrons are selected by a variable entrance aperture of 5.0 mm, 2.0 mm or 1 mm in
diameter. After passing the entrance aperture the electrons travel down through the
spectrometer and are deflected through 90
◦
on the basis of the velocity-dependent
Lorentz force ~
F:
29
3 Characterization techniques
Figure 3.8:
Post-column attached Gatan Tridiem Filter. Electrons are led
through an tunable entrance aperture and separated according to their energy in
the magnetic prism.
~
F=q(~v ×~
B)(3.6)
Electrons with greater energy loss are deflected more strongly than those suffering
zero loss. A spectrum is thus formed in the dispersion plane which consists of a
distribution of electron counts versus energy loss. Besides bending the electron
beam and creating energy dispersion, the magnetic prism also has a focusing action
and creates an energy-dispersed, focused image of the TEMs projector lens crossover.
The first two quadrupoles before the slit increase the dispersion of the spectrometer
onto the slit and the quadrupoles after the slit have two functions. They can either
compensate for the energy dispersion in the image plane of the magnet and project
a magnified image of the specimen onto the CCD, or they can project an image
of the spectrum at the slit onto the CCD camera. In the first mode the system
produces images containing electrons of a specific energy selected by the slit (energy
filtered images) and in the second mode it operates like a PEELS (parallel EELS
detector system).
30
3.1 Electron Microscopy
Anisotropy-magic angle
Carbon and a lot of other interesting materials exhibit an anisotropic electronic
structure due to crystalline anisotropy or induced anisotropy as a result of strain,
electrical and magnetic perturbations [65]. At the nanoscale the proximity of the
surface may also induce additional anisotropy. Attempts to study such electronic
anisotropy with EELS are going back to studies of Raether [66] and coworkers in the
early 1970s. Further work was done by Zeppenfeld [67, 68] by studying extensively
low-energy losses in graphite. Around a decade later when parallel recording became
available it was possible to study the much weaker region of energy loss near edge
structures (ELNES) by means of EELS [59, 60, 69–72].
Due to a considerable finite collection angle, determined by the width of the SEA
(spectrometer entrance aperture) in the diffraction plane, not only directions of
q
corresponding to the center of the SEA contribute to the spectra, but also others
[63]. A consequence of this is that at high collection angles the spectrum does not
represent the electronic structure of the selected anisotropy axis any more. In the
case of graphite with the c-axis parallel to the incident beam this would mean that
instead of the π∗peak, the σ∗peak becomes dominant.
Figure 3.9 shows the variations of the fine structure in the C1s absorption spectra
of graphite as a function of the collection semi-angle (
β0
) and as a function of the
specimen orientation, defined by an angle
γ
between the c-axis of graphite and
the incident beam direction respectively [65]. It also presents the kinetic relation
between the incident electron wave-vector
k0
, the scattering electron wave-vector
kf
, and the resulting momentum transfer vector
q
, which is in relation to the
scattering angle Θ. Also shown is the decomposition of the momentum transfer
vector along the directions parallel (
qII
) and perpendicular (
q⊥
) to the direction of
the incident electron beam. Neglecting this anisotropy effect can cause errors in the
determination of the
sp2
/
sp3
ratio in carbon materials [73, 74]. Theoretical work
to determine the value for the magic angle has been done by a number of groups
[60–62].
The relativistic magic angle was found to be 1.46 ΘE[75] for 200 keV electrons.
As shown above, the spectrometer collection angle is a crucial parameter to probe
anisotropy in the local final DOS with EELS. Therefore, varying the collection
angle, as shown in Figure 3.10, probes, for the case of small angles, only states
31
3 Characterization techniques
Figure 3.9:
Variation of the C1s fine structure of graphite due to variations of
the collection semi-angle (β0) and due to the specimen orientation [65].
that are oriented parallel to the incident beam, while large collection angles cause
additionally excitations into orbitals perpendicular to the incident beam. For the
interpretation of fine structures in EELS measurements it is therefore necessary to
not only to fulfill the dipole approximation but additionally to fix the spectrometer
collection angle.
3.1.3 Scanning Electron Microscopy
In the Scanning Electron Microscope (SEM) the surface of a sample is imaged by
scanning it with a beam of electrons in a raster scan. The images are formed by
secondary electrons and/or elastically backscattered electrons. The probing depth
of the electrons in a SEM can be varied by changing their acceleration voltage.
This is used to get information about the surface morphology. Using only the back
scattered electron (BSE) for forming the image indicates the elemental composition
of a sample, as the intensity of the BSE signal is strongly related to the atomic
number (Z) of the specimen. Standard SEM instruments are equipped with an EDX
detector, which gives information about the elemental composition of the sample.
32
3.2 NEXAFS
Figure 3.10:
Final state dependence upon the direction of the momentum transfer
Q in the case of a) a small scattering angle and b) a large scattering angle.
3.2 NEXAFS
X-ray absorption spectroscopy (XAS) has advanced rapidly in the last decades,
particularly with synchrotron radiation based techniques [76]. The high brightness
of the synchrotron X-ray source has allowed tremendous improvements in spectral
resolution and the development of X-ray microscopies which apply X-ray absorption
spectroscopy to heterogeneous materials at high spatial resolution.
Near edge X-ray absorption fine structure (NEXAFS) spectroscopy [76] involves
excitation of core-level electrons to unoccupied or partly occupied molecular orbitals.
The resulting variety of phenomena, such as; absorption of energy, fluorescence or
emission of photons are used for the assessment of the bonding environment for a
specific element [77]. NEXAFS has been successfully used for the environmental
study of coal [78, 79], plant fossils [80] and soot [13–15].
As an x-ray passes through matter, it is absorbed to an extent which depends
on the nature of the substance, the thickness of the sample, and the density of
the sample. The absorbed photons cause excitation of the inner shell electrons
of the atoms in the substance. These excited inner shell (core) electrons can be
promoted to unoccupied energy levels to form a short lived excited state or they can
be removed completely to form an ionized state. NEXAFS selects a specific atomic
33
3 Characterization techniques
species through one of its edges and probes its bonding to intra-molecular and, to
less amount, extra-molecular neighbors, such as surface atoms. NEXAFS is capable
of detecting specific molecular bonds, e.g. C-C, C=C,
C≡C
, and C-H bonds in
hydrocarbons, and also to determine intra-molecular bond lengths and derive exact
orientation of molecules and functional groups present on the probed surface or in
the probed solid [76]. Traditionally, X-ray absorption spectroscopy was described in
terms of absorption edges which relate to the onsets of inner-shell ionization.
There is an absorption edge associated with each inner shell energy level of
an atom, such that all elements have an X-ray absorption edge in the soft X-ray
energy range (100–1200 eV). The amount of a particular element can be determined
quantitatively from the difference in the x-ray absorption just above and just below
its absorption edge.
While this traditional aspect of X-ray absorption provides the basis for elemental
analysis, the modern technique as executed at synchrotron facilities is much more
powerful. Compared to laboratory X-ray sources such as X-ray tubes, synchrotron
radiation is by many orders of magnitude brighter and more easily tuneable. Syn-
chrotron radiation is produced by electrons accelerated to relativistic speed that are
forced to radiate when they travel through strong magnetic fields. It can identify
quantitatively the chemical structure of the element from the fine scale details of
the absorption spectrum that occur at each edge, the so-called near-edge x-ray
absorption fine structure or NEXAFS [76]. NEXAFS methods can be applied to
nearly all elements and posses the advantage to interrogate the composition of minor
or trace elements in soot samples.
These features, which can be as much as 10 times stronger than the absorption
edge jump, correspond to electronic excited states in which an inner-shell electron
has been excited to unfilled molecular orbitals or conduction bands. As the x-
ray energy is increased throughout an absorption edge, first there is structure
associated with excitation to the lowest unoccupied molecular orbital, which is a
π∗
orbital for unsaturated carbonaceous molecules (double or triple bonds), followed
by structures associated with higher energy unoccupied molecular orbitals, typically
of
σ∗
character associated with saturated (single) chemical bonds, and then direct
inner-shell ionization.
The unoccupied electronic structure and thus the inner-shell excited states are
determined by the geometric and electronic (bonding) structure of the sample.
34
3.2 NEXAFS
NEXAFS spectra differ significantly even for rather similar molecular structures, as
shown in Figure 3.11 for some common polymers [81].
This means that the NEXAFS spectrum of each polymer can be used as a
fingerprint. In many cases, enough is known about how chemical structure and X-
ray absorption spectral features are related to allow one to identify unknown species
from measured NEXAFS spectra. Individual spectral features, particularly the low
energy
π∗
features, are often sufficient for qualitative identification in reasonably well
characterized systems, and they can serve as useful energies for selective chemical
contrast in X-ray microscopy. There is a characteristic NEXAFS structure at the
absorption edge of each element in a sample. Thus the combined study of C 1s, N 1s
and O 1s NEXAFS is a very powerful tool in polymer analysis. Finally, comparisons
with NEXAFS spectra of pure standards provides a means to derive quantitative
composition maps of the components of a complex material from a series of X-ray
microscopy images.
If the energy of an incident photon is high enough to match the binding energy
of an electron, the photon absorption probability suddenly increases in the so called
X-ray absorption edge, which is at about 290 eV for carbon, so that the electron
will be completely removed from the bonding orbital in an ionization event. At
an energy slightly below the absorption edge, the energy is not sufficient to ionize
the electron but it may be strong enough to promote it to a more weakly bound
orbital. The probability for this process depends on the occupation of this orbital
[82]. Chemical binding energies are in the range of 2-10 eV, the occupancy of orbitals
within
≈
2-10 eV of the absorption edge is in turn modified by the chemical binding
state of the atom [76, 83]. Those pre-edge resonances (features between 284 and 290
eV) are known as near-edge X-ray absorption fine structures (NEXAFS) or X-ray
absorption near edge structures (XANES). In carbon their width is
≈
0.1-0.2 eV,
which is determined by a combination of a relatively long core-hole lifetime and
a shorter lifetime of the excited state that is coupled to it. Beside the pre-edge
resonances from coupling to bound states, the inner-shell electrons can also be
promoted to classically unbound states, such as the 1
s→σ∗
transition, that have
broad energy widths of several electron volts due to their short lifetime.
NEXAFS measurements have been performed at the synchrotron radiation facility
BESSY II (Berliner Elektronenspeicherringgesellschaft für Synchrotronstrahlung),
the 3rd generation synchrotron of the HZB (Helmholtz-Zentrum für Materialien
35
3.2 NEXAFS
und Energie) in Berlin using monochromatic radiation of the ISISS (Innovative
Station for In Situ Spectroscopy) beamline as a tunable X-ray source [84]. C K-edge
and O K-edge spectra have been obtained in the Auger electron yield (AEY) mode
by using the electron spectrometer as a detector. The kinetic energy window of
the spectrometer has been set to 260-320 eV for C K-edge spectra and 520-560 eV
for O K-edge spectra, respectively, and a pass energy of 50 eV has been applied.
Simultaneously with the AEY spectra, a positively biased wire was used to obtain
absorption spectra in the total electron yield mode (TEY). The monochromator
was scanned in a continuous mode through the energy range of the C K and O K
absorption edges, i.e. the monochromator was not swept in a step-wise manner for
data acquisition but was driving with a constant speed, in this case 250 meV/sec,
without stopping for data retrieval. Data have been taken continuously and the
actual energy position at every data point was read back from the monochromator
control. Raw spectra contained approximately 3000 data points. Typically, data
reduction of a factor 4 was performed during analysis by averaging adjacent data
points during data manipulation to increase the signal to noise ratio of the spectra.
Absolute energy calibration has been done via the
π∗
resonance of TEY spectra of
gas phase
O2
at 530.9 eV [85, 86] and the second order gas phase
O2
at 265.45 eV.
The spectral resolution was about 0.1 and 0.15 eV at the C K- and O K-edge,
respectively. C K-edge raw spectra have been intensity normalized with a carbon
free Ag-foil.
3.2.1 Normalization and background corrections
The measured NEXAFS signal depends directly on the incident X-ray intensity,
and the spectra need to be corrected for variations of this intensity over time and
as a function of the photon energy [76]. Intensity variations may occur due to
instabilities of the electron beam in the storage ring. Modulations in the X-ray
intensity with photon energy are due to energy-dependent reflectivity changes of the
X-ray optics in the beam line which focus (mirrors) and diffract (gratings) the X-ray
beam emitted by the storage ring. For monolayer concentrations of molecules on
a surface, the measured signal typically has a large background contribution from
the supporting material. This has to be considered when extracting the NEXAFS
signature of the chemisorbed molecule. Several approaches are used historically
37
3 Characterization techniques
to normalize the signal from the sample. Depending on the type of background
corrections used, different sources of background problems in NEXAFS will be
eliminated. Which correction type is appropriate depends on the sample, energy
range and on the detection conditions.
It has to be pointed out that the raw signal acquired during a NEXAFS experiment
is the result of a combination of factors, of which the signal from the species of
interest may not always be the most significant contribution.
In our work we use "division by a reference monitor". This reference provides a
measure of the fluctuations in the X-ray intensity emitted by the storage ring. For
the C K NEXAFS the spectra were corrected by division of the signal from a clean
surface, which was in our case the signal of a clean Ag foil measured in the region
of the C K edge area. This Ag foil is used to cancel out the carbon signal which is
due to carbon contamination on the mirrors.
3.3 XPS
X-ray photoelectron spectroscopy (XPS) is an surface analysis technique widely used
for the determination of the elemental composition, nature of chemical bonds and
electronic structure of the surface region of a sample. It is based upon the principle of
the photoelectric effect, explained by Einstein in 1905 [87], in which monochromatic
electron radiation is used to excite bound electrons above the vacuum level. Since
for solids, photoelectrons can escape only from a depth in the order of nanometers
(5-10 nm), XPS is a surface sensitive technique. Depending on the photon energy,
which is usually in the range of X-ray radiation (100-1500 eV), core electrons from
the atoms can be photoemitted. A synchrotron source, such as the synchrotron
radiation facility BESSY II [88], offers the possibility to tune the photon energy in
a wide range between 5 and 5000 eV. Another advantage of a tuneable x-ray source
is the possibility of depth profiling measurements as the inelastic mean free path of
the emitted electrons depends on their kinetic energy.
When a sample is irradiated by a beam of soft X-rays (200-2000 eV), atomic core
electrons are emitted from the matter as consequence of the photoelectric effect.
The kinetic energy (KE) of those emitted electrons is given by
38
3.3 XPS
KE =hν −BE −φs(3.7)
with
hν
being the photon energy, BE the binding energy of the atomic orbital
from which the electron is emitted and φsis the spectrometer work function [89].
By measuring the kinetic energies of the emitted photoelectrons, it is possible to
determine their binding energy (BE), intended as the energy needed to remove an
electron from the ground-state to the vacuum level.
3.3.1 Initial state effects
When the energies of core levels are investigated in detail, small shifts in the binding
energies of the electrons can be found. Those core level shifts can be induced due
to the initial state (Figure 3.12) structure of the particle under investigation [90].
The electron configuration in and around the atom is affected by the chemical
bonding, oxidation state and electro-negativity of neighboring atoms. This results
in the appearance of a shift or shoulders of the main peaks. Thus, the observed BE
chemical shift gives information about the electronic configuration of the atom under
investigation. Conceptually, the XP spectrum of a model compound can be used as
finger-print for the recognition of species on the surface. The limitation for resolving
those slight chemical shifts is given by the peak width, which is restricted by the
instrumental resolution, but determined by the lifetime of the positive core-hole,
due to the Heisenberg uncertainty relation, created as a result of the photoemission
process.
3.3.2 Final state effects
Final state effects (Figure 3.12) reflect the relaxation energy of the system as
response to the energy difference between the excited electron system of a particle
after the loss of a photoelectron and relaxation of the electron system. Differences
depending on the material can be observed: in metals the hole state created by the
emitted photoelectron is completely shielded by conduction electrons and cores of
neighboring atoms; for "isolating" materials the number of conduction electrons and
neighboring atoms is much lower than for a metal and therefore the screening of
the positive hole will be much less.
39
3.3 XPS
3.3.3 Inelastic background
Figure 3.13 shows a XPS spectrum taken over an wide range (0-1100 eV) obtained
from the Euro VI soot sample using
Eexc
= 1200 eV. Binding energy is plotted versus
the intensity of photoemitted electrons. Figure 3.13 shows that the peaks are sitting
on a background of secondary electrons that originate from higher kinetic-energy
electrons. Therefore, the background increases stepwise after each major ionization
peak. This feature is a result of inelastic electron energy loss that occurs as electrons
from deep core levels, with a depth over the inelastic mean free path of the electron,
loose their kinetic energy. Usually the background signal is subtracted by curve-
fitting, but this can introduce some errors in quantitative intensity determination
and peak position shifts [91].
3.3.4 Shake up/off
These events occur when the outgoing photoelectron excites a valence electron to a
previously unoccupied state, or electron-hole formation (shake-up). An excitation
of the valence electron above the vacuum level can also occur and is called shake-off.
For those transitions it is necessary, that the photoelectron release some of its kinetic
energy. Hence, new features in the XPS spectrum must lie on the high binding
energy side of a direct photoemission transition. The shape of this feature is not
always reflected by a sharp peak, as photoelectrons tend to fall into the energy
region of inelastic secondary electrons and therefore often show no discrete structure.
In the case of our soot samples a shake-up feature can be observed which is used for
the "alignment" of the spectra.
XPS is the most suitable and most applied analytical techniques for the charac-
terization of the chemical configuration of the C-heteroatom bonds on the surface
of carbon material. As a consequence, a large volume of literature [92
–
94] has been
dedicated to the analysis of C1s and O1s XPS spectra, aimed at the assessment of
the chemical nature of the surface species in carbon-based materials.
X-ray photoelectron spectroscopy (XPS) has been performed at the synchrotron
radiation facility BESSY II using monochromatic radiation of the ISISS beamline
as a tuneable X-ray source [84]. Samples were introduced into the reaction cell,
≈
2 mm distant from an aperture to the differentially pumped stages of the lens
41
3 Characterization techniques
system of the hemispherical analyzer (Phoibos 150, SPECS). Details of the set-up
are described elsewhere [95].
The excitation energy used for C1s/O1s core level spectra were 585/830 eV,
respectively, resulting in a high surface sensitivity with an inelastic mean free path
(IMFP) of the photoelectrons of about 0.86 nm. The spectra have been normalized
to the impinging photon flux determined by a cleaned Ag foil and corrected for
the fraction of higher order as well as the electron current in the storage ring.
Quantitative XPS data analysis was performed by using theoretical cross sections
[96]. The O1s envelopes were fitted using mixed Gaussian-Lorentzian component
profiles after subtraction of a Shirley background [97] using Casa XPS software [98].
In the spectra in which the species abundance was close to the detection limit, a
linear background has been subtracted. The fitting was done by fixing the peak
maximum within
±
0.1eV for all spectra and applying a full width half maximum
(FWHM) of 1.4-1.6 eV. The C1s envelopes were fitted as following: The graphitic
peak was fitted in an asymmetric peak, well described by a Doniach-Sunjic model
[99]. The other peaks were fitted by using a mixed Gaussian-Lorentzian component
profiles after subtraction of a Shirley background. Atomic ratios were calculated
using atomic sensitivity factors [100].
42
4 HRTEM on Soot
In this chapter the structural analysis of the various soots and carbonaceous samples
is reported. The analysis by transmission electron microscopy will be separated into
two parts: first the analysis of the as-received samples will be shown and discussed;
in the next step the effect of the oxidation treatment will be shown by presenting the
study on the oxidized samples and their comparison with the as-received samples. In
this way the data presented is split and it should be easier to follow the modifications
that the samples undergo due to the oxidation treatment.
4.1 Structural analysis of the as-received samples
For TEM characterization the soot samples, originally collected on stainless steel
grids, were dispersed in chloroform for 30 seconds in an ultrasonic bath and a drop
was deposited on a Cu-grid coated with an amorphous carbon film. To avoid any
effect of the solvent on the soot samples parallel to the above mentioned preparation
technique TEM samples were also prepared by scratching of the soot particles from
the stainless steel grid and depositing them directly onto the TEM grids.
4.1.1 Particle size
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM)
analysis was undertaken to determine the micro- and macrostructure of the as-
received soot samples. SEM measurements (Figure 4.1) reveal the spherical shape
of the Euro IV and Euro VI soot particles. These "primary particles" have a strong
tendency to agglomerate with each other (Figure 4.2).
Particle size measurements were carried out on the acquired TEM overview micro-
graphs for the diesel soot samples (Euro VI and IV) to determine the distribution of
secondary structural parameters. About 150 particles were measured per sample.
43
4 HRTEM on Soot
Figure 4.1: SEM micrograph representative for the investigated soot samples.
Figure 4.2: TEM overview for the determination of the soot particle size.
44
4.1 Structural analysis of the as-received samples
Figure 4.3: Radii distribution for the as-received Euro IV sample.
It should be mentioned that the analysis of those structural parameters is hindered
by the fact that the secondary soot particles cluster together to build agglomerates
with sizes outside the depth of focus, as shown in the right part of Figure 4.2.
Therefore it was necessary to confine the observations to the outer parts of the
agglomerates, where it is still possible to determine the individual sizes of the
particles. The histograms of the radii distribution for as-received Euro IV (Figure 4.3)
and Euro VI (Figure 4.4) soot samples are shown subsequently.
Statistical particle size measurements give a particle size of 28
±
12 nm for the
as received Euro IV sample and values of 24
±
8nm for the as-received Euro VI
sample.
4.1.2 HRTEM
TEM and HRTEM investigations were carried out to deduce the microstructure of
the different carbonaceous samples. Using this techniques we present a comparable
structural analysis of the as-received graphite powder, Euro IV, Euro VI, and GfG
soot samples. The HRTEM results obtained here for GfG and Euro IV soot samples
are in good agreement with recent studies reported by Su et al. [19] and Müller et
al. [17, 18] on those samples.
45
4 HRTEM on Soot
Figure 4.4: Radii distribution for the as-received Euro VI sample.
HRTEM reveals that the samples contain, besides graphene BSUs, twisted ribbons
with molecular units "sitting" on them. As the investigated particles are thicker
than the depth of focus in the microscope, the two-dimensional projection may
lead to less straightforward interpretation. The projection of the three dimensional
objects may indicate that some graphitic structure is present, which is in the case
of GfG, Euro VI, and Euro IV only due to the selected defocus value. In particular
the contrast "gaps" following the rounded contour of a "graphene" layer arises from
tilting of the graphene ribbon with respect to the projection. Multiple segmentation
seems to indicate apparent small BSU. For a detailed description see [101].
High-resolution micrographs of the untreated GfG soot (Figure 4.9a) show very
fine agglomerates. The graphene segments are strongly bent and connected to long
chainlike agglomerates, which is in good agreement with previous studies [17, 18].
Micrographs of the untreated Euro VI sample reveal a more pronounced long-range
order and higher structural order compared to the as-received GfG soot sample. On
the surface of the untreated Euro VI soot (Figure 4.10a) small fullerenoid particles
with an onion-like structure can be observed. These particles are similar to those
found on the Euro IV (Figure 4.11a) sample as discussed subsequently. No distinct
morphological differences were found between Euro IV and Euro VI soot. There is a
similar arrangement of graphene layers explained by a similar soot formation process
46
4.1 Structural analysis of the as-received samples
characterized by oxygen-lean high-temperature conditions. On the surface of the
as-received Euro IV sample, small fullerenoid-like particles with a diameter of 1-3 nm
were found at the surface. Similar structures were recently reported for untreated
Euro IV soot [17
–
19]. HRTEM of the graphite powder sample (Figure 4.12a) shows
the presence of stacks of graphene layers with some molecular-type carbon on the
surface of the untreated sample. This sample exhibits over the predominant volume
well-defined long-range ordering, being in contrast to the GfG, Euro IV, and Euro
VI samples.
4.1.3 Curvature measurements
The different degree of bending of the graphene layers is an important indicator
for its defect content, as deviations from planarity are linked with vacancies that
affect the local reactivity. This was already theoretically [102] calculated and
experimentally observed [18, 103]. To quantify the lattice fringes in the HRTEM
images, the complex phase contrast image has to be converted into discriminable
fringes that can be analyzed by image analysis algorithms. The images are processed
in the following way: Fourier Transformation and Filtering of the diffraction features
corresponding for carbon in the HRTEM micrographs, skeletonization of the line
segments in the resulting binary picture, post-processing and quantitative analysis
of the observed fringes. In this way it is possible to create a binary picture, in
which the lattice fringes are separated as discrete line segments [104, 105]. Such
micrographs are shown in Figure 4.6 with the graphene segments extracted and
highlighted for clarity. This was executed by use of the DIGITAL MICROGRAPH
software package (Version 1.80.70 for GMS 1.8.0) from GATAN. The measuring
procedure used in this work was already used and proven to be useful in a previous
study [103]. It is important to point out that only the fraction of ribon elements
projected parallel to the electron beam and not all weakly interconnected "graphene
units" as assumed from observing the HRTEM images are evaluated. Curvature is
defined as the length ratio between an uninterrupted contrast line of a ribbon and
the shortest distance between its terminal points as illustrated in Figure 4.5.
The advantage of this fringe analysis is its sensitivity to the level of molecular
organization (nanostructure) of the graphitic sample.
47
4 HRTEM on Soot
Figure 4.5:
Definition of "graphene length" and "length" for the curvature
measurements.
The results of the statistical measurements of the curvature analysis for the as
received GfG, Euro VI and Euro IV soot samples are discussed subsequently and
summarized in Table 4.2, Figure 4.13(a), Figure 4.15(a) and Figure 4.14(a). A clear
trend can be seen for the different samples. The untreated GfG sample is by far
the sample which is bent strongest with a value of 0
.
607
±
0
.
117. The Euro VI and
IV samples show higher values for the curvature index indicating weaker bending
as compared to the untreated GfG soot. The value of 0
.
65
±
0
.
107 for the Euro
VI is slightly lower compared to 0
.
67
±
0
.
076 for the Euro IV sample. This would
indicate an higher C-C bond distance which can be then correlated with a possible
higher amount of heteroatoms/oxygen atoms in the sample. In Table 4.2 the length
of the graphene segments for the different samples is also displayed.
4.2 Effect of oxidation treatment
4.2.1 Particle size
The quantitative morphological characterization of Euro IV and Euro VI with the
calculated values before and after oxidation treatment are summarized in Table 4.1.
The histograms of the radii distribution for the oxidized Euro IV (Figure 4.7) and
oxidized Euro VI (Figure 4.8) sample are shown subsequently. It is shown that
the values of 24
±
8 nm for the as received Euro VI sample and 28
±
12 nm
for the as-received Euro IV sample are affected in different ways by the oxidation
treatment. The oxidation treatment has no observable effect on the particle size
distribution of the Euro VI sample, as the value is after oxidation treatment 24
±
6
nm. For the Euro IV sample the statistical measurements reveal that the particle
size distribution after oxidation treatment is in the range of 22
±
6 nm. This shows
that the Euro IV particle size distribution becomes more narrow, indicating that
48
4.2 Effect of oxidation treatment
(a) Starting HRTEM micrograph
(b) Cutout from HRTEM micrograph used for
determining the graphene segments
(c) Final picture with the measured graphene
segements marked in red
Figure 4.6:
HRTEM to calculate curvature values, showing the original micro-
graph, the cutout from which the graphene segments are taken and the colorized
graphene layers used for the measurement.
49
4 HRTEM on Soot
Figure 4.7: Radii distribution for the oxidized Euro IV sample.
Table 4.1:
Radii distribution of the as-received and oxidized the primary Euro
IV and VI particles.
Radius [nm]
sample mean value standard distribution
Euro IV untreated 14 6
Euro IV oxidized 11 3
Euro VI untreated 12 4
Euro VI oxidized 12 3
the bigger particles shrink and that very small particles are completely removed by
to the oxidation treatment. Accordingly, the oxidation occurs in a transport-limited
way and progresses with a reaction front from the outer surface to the center of
the tertiary structure (agglomerate). The secondary structure is oxidized sphere
by sphere and no extensive concerted oxidation throughout the bulk of the sample
occurs, which is shown by nearly no changes in the particle size for the Euro VI
sample as presented in Table 4.1. This precludes the abundance of highly reactive,
poorly-ordered or molecular carbon as a binder place between the supra molecular
twisted graphene ribbons. A reactive binder place would have led to porosity, and
thus to simultaneous shrinkage of all remaining secondary structural units, which
could be present within the Euro IV samples.
50
4.2 Effect of oxidation treatment
Figure 4.8: Radii distribution for the oxidized Euro VI sample.
4.2.2 HRTEM
Observations on the GfG soot micrographs of the sample oxidized up to 773 K
(Figure 4.9b) show less chain-like orientation compared with the as-received sample.
The particles seem "more tightly" connected than the untreated one, suggesting
an increase of structural order for GfG soot upon oxidation. Micrographs of the
untreated and oxidized Euro VI sample reveal a more pronounced long-range order
and higher structural order compared to untreated GfG soot.
The as-received Euro IV (Figure 4.11a) sample, with poorly ordered surface carbon,
is modified due to the oxidation treatment (up to 773 K) by a preferential removal
of the poorly ordered carbon structures, as seen by comparing the microcgraphs
shown in Figure 4.11.
The preferential oxidation of the highly disordered fraction should lead to an
apparent increase in overall ordering if its volume fraction was significant in the
deep parts of the secondary structure. This suggests that the heat treatment and
oxidation result in the removal of these minor molecular layers, modifying the overall
order and graphitization of the Euro IV soot slightly. HRTEM of the graphite
powder sample (Figure 4.12b) shows only minor morphological differences before
and after oxidation. The sample is composed of stacks of graphene layers with
some molecular-type carbon on the surface of the untreated sample. This sample
51
4 HRTEM on Soot
Table 4.2:
Curvature and length values for the as-received and oxidized GfG,
Euro IV and VI particles
Curvature fibre length [nm]
sample mean value standard distribution mean value standard distribution
GfG untreated 0.61 0.12 1.30 0.98
GfG oxidized 0.63 0.12 1.34 0.72
Euro IV untreated 0.67 0.08 0.89 0.59
Euro IV oxidized 0.71 0.08 1.40 0.84
Euro VI untreated 0.65 0.11 2.01 1.19
Euro VI oxidized 0.66 0.11 1.06 0.68
exhibits over the predominant volume well-defined long-range ordering. This being
in contrast to the GfG, Euro IV, and Euro VI samples. Due to its graphitic structure,
the sample is expected to be much less reactive than the other samples.
4.2.3 Curvature measurements
The results of the statistical measurements of the curvature analysis for the as-
received and oxidized GfG, Euro VI and Euro IV soot are listed in Table 4.2. The
histograms of the curvature distribution for the oxidized GfG (Figure 4.13b), Euro
IV (Figure 4.14b) and Euro VI (Figure 4.15b) samples are displayed and discussed
here.
It is observed that after the heat and oxidation treatment, the curvature of the GfG
sample is reduced allowing one to conclude that the graphitic order increased and
reactivity decreased. The material is expected to self-inhibit its oxidation.
During the oxidation treatment the curvature for Euro VI and IV soot weakly
decreases within in the range of its standard deviations. A slightly higher bending
for untreated and oxidized Euro VI soot compared to Euro IV soot suggests that
Euro VI soot can be characterized by higher defect density and thus by increased
concentration of reactive functional groups [18].
52
4.2 Effect of oxidation treatment
Figure 4.9:
Overview and magnified HRTEM images of as-received (a) and oxi-
dized (b) GfG Soot.
53
4 HRTEM on Soot
Figure 4.10:
Overview and magnified HRTEM images of as-received (a) and
oxidized (b) Euro VI Soot.
54
4.2 Effect of oxidation treatment
Figure 4.11:
Overview and magnified HRTEM images of as-received (a) and
oxidized (b) Euro IV Soot.
55
4 HRTEM on Soot
Figure 4.12:
Overview and magnified HRTEM images of as-received (a) and
oxidized (b) Graphite.
56
4.2 Effect of oxidation treatment
(a) as-received GfG
(b) oxidized GfG
Figure 4.13:
Curvature distribution of the as-received and oxidized GfG sample.
57
4 HRTEM on Soot
(a) as-received Euro IV
(b) oxidized Euro IV
Figure 4.14:
Curvature distribution of the as-received and oxidized Euro IV
sample.
58
4.2 Effect of oxidation treatment
(a) as-received Euro VI
(b) oxidized Euro VI
Figure 4.15:
Curvature distribution of the as-received and oxidized Euro VI
sample.
59
5 EELS
In this part the study on the electronic structure of the carbonaceous samples by
means of EELS is presented and discussed. The discussion of the data will be
separated into two subsections: First the electronic effects vibrations visible in
the low loss region of the EEL spectra, such as plasmon features and electronic
transitions, are compared and discussed. This is followed by an investigation of the
carbon K edge with a focus on molecular hybridization. Therefore, the ratio between
sp2
and
sp3
hybridized carbon atoms is determined. Finally the modifications in the
electronic structure of the different soot samples are discussed.
5.1 Low Loss region
In this part, measurements of the low loss region (0-100 eV energy loss) of Euro
IV, Euro VI and GfG soot are presented and compared with a graphene low loss
spectra. The two distinctive features present in Figure 5.1 are the peaks in the area
between 4 and 5 eV and between 20 and 25 eV energy loss. The first one corresponds
to electronic transitions between the
π
and
π∗
states [56, 106] and the latter one
reflects plasma resonance frequencies (plasmon peaks) which are determined by
the density of valence electrons. Due to the different valence electron density of
graphite and diamond the corresponding plamon peaks can be found at different
energies: 27e eV for graphite and 33 eV energy loss for diamond. Amorphous carbon
exhibits a peak at an energy loss around 23 eV or less [107]. For graphite and glassy
carbon a strong maximum energy position around 4.8 and 4.5 eV [106] is found. It
is known [108] that glassy carbon is a precursor material to graphite in the sense
that the two materials are structurally related and that the former can be converted
into the latter by pyrolysis at sufficiently high temperatures [106]. A shift of the
π→π∗
peak to lower binding energies has already been observed and is discussed
in literature [109, 110] as a result of the curvature of the shells. The coupling of
61
5 EELS
Figure 5.1:
Low loss region of the Euro IV, VI and GfG samples compared to
HOPG
the electrons on spherical shells/ bent graphenes is different from that in the planar
case. This observation is in agreement with theoretical work done by Yannouleas et
al. [111].
The energy values reported in literature are directly measured from the spectra
still containing the zero loss, and therefore also include plural/multiple scattering
contributions. Those contribution shift the energy position of the features slightly to
higher binding energies. As shown in Table 5.1, where the zero and low loss region
of the as-received and oxidized Euro IV, Euro VI and GfG samples are presented
together with the HOPG reference, it is clearly visible that the removal of plural
scattering contributions shifts the energy positions of the features to lower energies.
The Fourier-log method in the DIGITAL MICROGRAPH software package (Version
1.80.70 for GMS 1.8.0) from GATAN was used to remove the zero loss signal of the
spectra and to remove plural scattering contributions, bearing in mind that the
condition of a sample with homogeneous thickness is not completely fulfilled. As
62
5.1 Low Loss region
Figure 5.2:
Zero and low loss region of the Euro IV, VI and GfG samples compared
to HOPG after removal of multipe scattering
already shown in the previous chapter, the particle size of the different samples
is in the same range. We can assume to make the same systematic error in all
cases and therefore the removal of plural scattering contributions remains a fairly
accurate measurement. Depending on the thickness of the sample the difference
varies. In Figure 5.1 and Figure 5.2 the respective spectra to the energy values given
in Table 5.1 are shown.
The
π→π∗
feature is only present as a small bump in the as received GfG
sample. This indicates that the material contains only a minor amount of
sp2
(or sp)
hybridized carbon and that the predominant bonding is of the
sp3
or diamond-like
form. After the oxidation treatment a peak at 3.7 eV energy loss can be found for the
GfG sample, indicating that a significant amount of
sp2
(or
sp
) hybridized carbon
was formed during the oxidation treatment. The evolution of the energy position
of the
π→π∗
and plasmon peak features for GfG are shown in Table 5.1. The
difference spectra of oxidized minus as-received GfG are plotted in Figure 5.5. It can
be seen that some intensities at the higher binding energy side in the plasmon peak
region are lost which can be explained as follows. The broad plasmon peak of the
as received GfG sample is a combination of graphite, defective
sp2
carbons and also
amorphous carbon. Due to the formation of
sp2
(or sp) hybridized carbon during the
63
5 EELS
Figure 5.3:
Low loss region of the as-received(20) / oxidized(500) Euro IV sample
with the difference spectrum inserted
oxidation treatment, the plasmon contribution from defective
sp2
carbons decreases
and the overall peak looses signal at the high binding energy side - therefore the
peak shifts to lower binding energies. Parallel to that it can be seen that a sharper
π→π∗features is formed in the oxidized sample.
The difference spectrum of the Euro VI sample is plotted in Figure 5.4. The
maxima position, as shown in Table 5.1, of the
π→π∗
feature and the plasmon
peak stay constant before and after the oxidation treatment, which indicate no
distinctive changes in the electronic configuration. It should be mentioned that a
higher intensity for the
π→π∗
feature can be found for the untreated sample and
that the treated sample has a slightly narrower plasmon peak feature. This can
be explained by some loss of
sp2
(or
sp
) hybridized carbon during the oxidation
treatment which reduces the intensity in the
π→π∗
feature. Parallel to that the
narrowing of the plasmon peak can be explained with an loss of amorphous carbon
species.
The low loss spectra of Euro IV as received/oxidized is plotted together with the
related difference spectra in Figure 5.3. It is visible quite clearly that the spectral
shape matches quite well and that also also seen in Table 5.1 the peak position of
the plasmon peak are unaffected by the oxidation treatment. From this results no
64
5.1 Low Loss region
Figure 5.4:
Low loss region of the as-received(20) / oxidized(500) Euro VI sample
with the difference spectrum inserted
Figure 5.5:
Low loss region of the as-received(20) / oxidized(500) GfG sample
with the difference spectrum inserted
65
5 EELS
Table 5.1:
Peak position of the major features in the samples under investigation;
before and after removal of plural scattering contributions
peak position peak position
(plural scattering removed)
sample π→π∗
/
eV
plasmon
peak / eV
π→π∗
/
eV
plasmon
peak / eV
Graphite
6.3 6.7
±
0.1
26.7 ±0.2 4.6 24.8
GfG untreated 23.3 ±0.4 21.7
GfG oxidized 5.6 ±0.3 23.1 ±0.3 3.7 20.9
Euro IV untreated 5.6 ±0.1 23.8 ±0.2 3.9 21.7
Euro IV oxidized 5.7 ±0.2 23.7 ±0.2 4.2 21.7
Euro VI untreated 5.5 ±0.2 23.5 ±0.1 3.8 21.4
Euro VI oxidized 5.5 ±0.1 23.4 ±0.2 4 21.4
significant changes in the electronic configuration of the Euro IV sample can be
observed.
5.2 C-K edge region
To analyse the electronic structure in the bulk of the investigated carbonaceous
samples C K-edge EELS measurements were carried out. The bonding properties of
the carbon atoms are revealed by this analysis and conclusions about the carbon
hybridization can be made. To make such a study relying on anisotropic samples as
the ones under investigation it is necessary to use magic angle conditions.
The Euro IV, VI samples and GfG soot were compared with the reference samples
HOPG, Flammruss 101 and NC 3100 nanotubes. The spectra of the analyzed
samples were then fitted with three Gaussian curves so as to determine the area
under the curves for quantifying the ratio of
sp2
to
sp3
bond in the soot samples, as
shown in Figure 5.7. HOPG is fitted and used as the internal standard for 100 %
sp2
hybridized carbon atoms. EELSMODEL [112] software was used for the fitting of
the acquired core loss EEL spectra.
Figure 5.6 displays the EEL spectra of the investigated samples acquired under
magic angle conditions. The features visible in the carbon K-edge represent different
components of unoccupied electronic states. The
π∗
feature at an energy loss of
66
5.2 C-K edge region
285 eV represents transitions of C 1s electrons to unoccupied
π∗
whereas the
σ∗
feature at 291 eV reflects that of transition to unoccupied
σ∗
state. The
π∗
feature
is typical for
sp2
-hybridized carbon. The percentage of
sp2
bonded carbon of these
samples calculated according to 5.1 is displayed in Table 5.2.
The results for untreated GfG soot indicate a good correlation between HRTEM
and EELS data. The poor parallel alignment of the graphene units visible in
Figure 4.9 fits quite well to the low abundance of
sp2
bonded carbon in the sample
as obtained by EELS. GfG soot undergoes a change in its microstructure during the
oxidation process, which leads to a substantial increase of the degree of graphitization.
The abundance of
sp2
bonded carbon increases from an initial value at room
temperate of 56% to a final value of 79% in the treated sample. Euro VI shows
nearly no changes in hybridization before (89%) and after heat treatment (88%).
The higher initial amount of
sp2
bonded carbon in the Euro VI sample compared
to that in the Euro IV sample is seen in the slightly more pronounced
π∗
presence
at 285 eV. In contrast to that the Euro IV sample shows an increase in the degree
of graphitization upon oxidation treatment from 83 % to 90 %
sp2
bond. This is in
agreement with HRTEM investigation, explained as an increase in graphitization of
Euro IV by the removal of minor adventitious molecular layers and by slight changes
in the curvature of the sample. The flattening of the graphene layers during the
heat treatment also fits with the increase of the fraction of
sp2
hybridized carbon
atoms in the sample.
sp2% = harea(π∗)
area(π∗+σ∗)isample
harea(π∗)
area(π∗+σ∗)i100%sp2reference
(5.1)
67
5 EELS
Figure 5.6:
C K-edge spectra of the as-received and oxidized Euro IV, VI and
GfG samples compared to HOPG measured under magic angle conditions
Figure 5.7:
Spectra of HOPG with 3 Gaussian curves used for the fitting of the
hybridization ratio
68
5.2 C-K edge region
Table 5.2: Calculated sp2hybridization from the acquired EELS spectra
sample sp2%
GfG as received 56 ±4
GfG oxidized 79 ±3
Euro IV as received 83 ±2
Euro IV oxidized 90 ±3
Euro VI as received 89 ±2
Euro VI oxidized 88 ±2
HOPG 100
69
6 NEXAFS
One of the challenges in the application of synchrotron-based NEXAFS spectroscopy
is the complexity of molecular components present in the sample. In the following
tables Table 6.1, Table 6.2 and Table 6.3 peak assignments for the O K and C K
region will be presented that are the result of a literature study on work done on
polymers, soils and organic matter.
6.1 Carbon NEXAFS
Figure 6.1 shows C(1s) NEXAFS spectra of HOPG, GfG, NC, FL, Euro IV and
Euro VI. The C (1s) for graphite-like materials is composed of the
π∗
feature at
around 285.6 eV and the
σ∗
feature at around 292 eV [125
–
128]. The area assigned
to
σ∗
states contains not only states of
σ∗
character, but contains also contributions
from extensions of the continuum of
π∗
states that started in the area 282-287 eV
are present. The first resonance peak at 285.6 eV is originated mainly from the
excitation to the
π∗
antibonding orbital at
sp2
(C=C bond) site from C 1s level. The
features between 286 and 290 eV are assigned in literature to different components.
In the literature [129], [130] the band at 286-287 eV is assigned to the C 1s (C-R)
→
1
π∗
C=C transition that is characteristic for functionalized aromatic groups. In
the work done by Lenardi et al. [130] on CNTs the resonance peaks located at 286.8,
287.8 and 288.8 eV were assigned to originate from C 1s
→π∗
(C=O), C 1s
→σ∗
(C-H) and C 1s →π∗(O=C-OH species).
Possible contamination of C-H species on the surface that would give rise to
σ∗
(C-H) transitions are reported [131], [132] to be at about 288 eV. Small peaks in the
area between 287-290 eV with a more distinct peak at 288.4 eV have been attributed
to "interlayer states" caused by poor alignment between the graphene layers [133]
of disrupted but pure graphite. It is also mentioned in literature [119] that the
feature at 288.4 eV is caused by oxidation of the sample that generated carboxylic
71
6 NEXAFS
Table 6.1: NEXAFS C K assignment according to [83]
C Forms Bond Example Transition
Peak En-
ergy (eV)
Aromatic C
and quinone
C
C=O
Quinone-type
C Protonated
and alkylated
aromatic C
Heteroatom
substituted
aromatic C
1s→π∗
283-284.5
[78, 79]
Aromatic C
and double-
bonded alkyl
C
C=C
Protonated
and alkylated
aromatic C
1s→π∗
284.9-285.5
[78, 79,
113]
Carbonyl substi-
tuted aryl C
w Alkene C
Aromatic
C with side
chain and
N-substituted
aromatic C
C-OH
C=O
R−
(
C
=
O
)
−R
’
C=N C-N
Carbonyl C in
aromatic ring
Aromatic C at-
tached to amide
group Phenol
C Carbonyl C
Pyrimidine C
1s→π∗
286-287.4
[78, 79,
114]
Aliphatic C C-H
Aliphatic C of
CH3
,
CH2
and
CH nature,
1s−3p/σ∗
287-288.5
[78, 79,
115, 116]
Carboxylic C
R-COOH
COO C=O
(
NH2
-C-
O
R−
(
NH2
)
−R0
Carboxylic C,
Carboxyamide
C, Carbonyl C
1s→π∗
288-288.7
[78, 79, 114,
116, 117]
O-alkyl C C-OH
Polysaccharides,
alcohols
and other
hydroxylated-
and ether-linked
C
1
s→
π∗
/1s3p/
σ∗
289.2-289.5
[78, 79,
118]
72
6.1 Carbon NEXAFS
Table 6.2: NEXAFS O K assignment according to [83]
O Forms Bond Example Transition Peak Energy (eV)
Ketones, aldehydes C=O, HC=O Vanillin 2s→π∗530.6-531.3 [119]
Carboxylic acids COOH, CONH Alanine, cysteine, 2s→π∗532-532.7 [119–122]
carbox-amides hydroxamic acid
Alcohols C-OH, C-N-O Ethanol 1s→σ∗534.1 [122, 123]
Ethers C-O-C, O-C-N Diethylether 1s→σ∗535.4-535.6 [122, 123]
Water H2O1s→σ∗535, 537-542 [124]
Table 6.3: NEXAFS C K assignment as done by[22]
eV Transition Functional Group
283.7 1s→π∗Quinone
1s→π∗Protonated / alkylated
284.9−285.5Aromatic and PNA
1s→π∗Carbonyl substituted
285.8−286.4Aromatic, phenolic
1s→π∗Aromatic C-OH
Ketone-C=O
287.1−287.4Aliphatic
1s→π∗Aromatic carbonyl
287.7−288.3C=O
287.6−288.2 1s−3p/σ∗CH3,CH2,CH
288.2−288.6 1s→π∗COOH
289.3−289.5 1s−3p/σ∗C−OH,alcohol
73
6 NEXAFS
Figure 6.1:
C K NEXAFS of HOPG, NC 3100, Flammruss 101, GfG, Euro IV
and Euro VI
74
6.1 Carbon NEXAFS
acid attached to the graphite planes given the long absorption associated with this
moiety in other materials, if the modifications are small enough. Structural features
in the area 284-285 eV were observed by Entani et al. [134] and were explained
in terms of edge-derived electronic states. An explanation for this feature in the
samples of GfG, Euro IV, VI and FL may indicate that a broad surface state is
present, which is centered at the Fermi level and is partially occupied. In literature
[134] it is considered that these states (E and E*) originate from the edge states
appearing at the zigzag edge of graphite. Such a feature present in our spectra
Figure 6.1 will be assigned to molecular carbon present on the surface of the soot
samples. The upper curve in Figure 6.1 shows HOPG (pure
sp2
hybridization).
The 1s
→π∗
resonance can be used as a finger print for
sp2
coordinated carbon
atoms. The onset occurs at around 285 eV with a peak maximum at 285.6 eV. The
1s
→σ∗
transitions onset above 291 eV excitation energy is in good agreement
with literature [128, 135]. The HOPG sample for this measurement was cleaved but
not heated up. This was done with the aim to identify contributions of C-H and
C-O bonds in the NEXAFS spectrum of HOPG. Such features are not present in
typical HOPG spectra found in literature [128, 135] as those samples are usually
heated up to remove surface contaminations. The HOPG spectrum in Figure 6.1
shows clearly such contributions in the energy range between 286-288 eV. which
is useful for the interpretation of similar features in the soot samples. Beside this
"contamination sensitive" measurements the spectral resolution in the C K NEXAFS
spectra is quite high, as indicated by the high resolution of the
π∗
and
σ∗
features
while the signal reaches nearly zero intensity before the
σ∗
feature. The NC 3100
sample reveals a relatively sharp onset of the
π∗
feature. The NC 3100 sample
shows the
π∗
peak maximum shifted slightly to the higher energy compared to the
HOPG sample but still at lower energy position than the soot samples. The weaker
C-C bonds resulting from the curvature of the graphene sheets and by the larger
interlayer spacing can explain the shift observed for the nanotube sample. The
lower energy position and sharper resolved
π∗
peak compared to the soot samples
is explained by the higher structural order present in the nanotube than in the
soot samples. Our results are in agreement with results obtained by Chen et al.
[136] who performed some experiments on MWNTs (15-20 nm in diameter) and
revealed a shift and a relative loss of
π∗
intensities for nanotubes compared to HOPG.
75
6 NEXAFS
For the lamp black (FL 101) sample a less pronounced
π
state is found with
respect to HOPG. The resonance shows an increased width which could indicate a
splitting of the
π
states. Our results for lamp black fit with previous XAS results
in literature [128]. A clear difference in the structure between the untreated (GfG
20) and the heated GfG sample (GfG 500) compared to the two emission standard
samples is visible. In particular the resonances in the region 285-290 eV are clearly
different all along the investigated samples. The as received GfG sample has a
much higher intensity in the region
≈
287.5 eV relative to the intensity at 285 eV
compared to the emission standard samples. After the oxidation treatment the GfG
sample shows a sharp feature at
≈
287.5 eV. The intensity in the region between
288-291 eV remains significantly high for the GfG samples which is in clear contrast
to the other samples under investigation. Additionally, the onset of the
π∗
peak in
the GfG samples starts already at around 284 eV which we assign to the presence of
"disordered/molecular carbon". While the GfG sample presents a strong signal at
≈
287.5 eV, the other samples are characterized by a broad feature ranging from 285.5
to 286.5 eV. The nature of this feature is not well established. Strong resonances
at 286 and 288 eV have been reported for
C60
[137]. The highly conjugated double
bound in benzene gives rise to a
π∗
resonance at 288 eV [138]. Therefore the
fullerene-like structure of GfG could be an explanation of the features visible in
the C K NEXAFS. The C K NEXAFS data presented here indicate that the GfG
sample consists of higher contributions of heteroatoms than the Euro samples and
that the structural order is variing strongly from the one for the Euro samples. This
is also supported by the HRTEM and EELS data discussed in previous chapters.
The difference method [94, 139] has been used for better comparison of the spectral
shapes. The Euro IV sample after oxidation treatment does not show substantial
structural modifications as shown by NEXAFS in Figure 6.2. The onset of the
tail of the
π∗
resonance is shifted to slightly lower energies for the Euro IV 20
sample compared to Euro IV 500 as shown in the difference spectra Figure 6.2. The
additional features assigned to oxygen functional groups and
σ∗
transitions do not
undergo modifications during the oxidation treatment. Therefore we can conclude
that the changes in the features in the difference spectra of the Euro IV sample
can be attributed to a slight increase of
sp2
hybridized carbon ( 285 eV) while the
oxygen content did not change significantly.
76
6.1 Carbon NEXAFS
Figure 6.2:
C K NEXAFS of the as received, oxidized Euro IV sample and the
difference spectra of the untreated and oxidized Euro IV sample
Figure 6.3:
C K NEXAFS of the as received, oxidized Euro VI sample and the
difference spectra of the untreated and oxidized Euro VI sample
77
6 NEXAFS
Figure 6.4: Difference spectra of the untreated Euro IV and Euro VI sample
The Euro VI 20 sample has 2 peaks at around 287.5 and 288 eV, which should
be related to carbon-oxygen bonds. Those signals are less significant visible in the
Euro VI sample oxidized to 500
◦
C as shown in Figure 6.3. The Euro VI samples
show negative features between 284-286 eV in the difference spectra presented in
Figure 6.3 parallel to positive features at 289 to 290 eV. The main
π∗
peak is not
moved in position as shown in the difference spectrum, only the relative intensities
in the region 284.5 to 286 eV are stronger than those in the region of 287 to 288
eV for the oxidized sample. If we assume that the oxidation treatment did not
change the
sp2
/
sp3
bonding configuration than this would mean that the oxidation
treatment results in a loss of features in the region 287 to 288 eV, assigned to oxygen
functional groups.
The difference spectrum of the untreated Euro IV and Euro VI sample, plotted
in Figure 6.4, shows a negative intensity in the area around 284 to 285.5 eV and
further negative intensities onwards from 286.5 eV. The
π∗
of the Euro VI sample is
at lower binding energy than the Euro IV, which indicates a higher hybridization
of
sp2
bonded carbon in the as received Euro VI sample. This agrees with EELS
data of these samples, already shown and discussed in Chapter 5. Additionally, the
negative features in the difference spectrum indicate that the as received Euro VI
sample contains a higher amount oxygen functional groups. After the oxidation
78
6.1 Carbon NEXAFS
Figure 6.5: Difference spectra of the oxidized Euro IV and Euro VI sample
treatment the differences become smaller as shown in Figure 6.5. This can be
explained by the fact that the amount of carbon atoms in
sp2
hybridization after the
oxidation treatment are approaching similar values for the Euro IV and VI sample.
This fits again with EELS data that showed that the Euro IV sample increases
in structural order during the oxidation process and approaches the order of Euro
VI soot. Considering the results shown in Figure 6.2, Figure 6.3, Figure 6.4 and
Figure 6.5 it is valid to assume that the oxidation treatment increases the structural
order of the Euro IV sample while the Euro VI sample seems to be less altered. A
more detailed and accurate analysis of the oxygen functional groups will be given
subsequently by the analysis of O K NEXAFS and O 1s XPS.
6.1.1 C K NEXAFS vs C K EELS
NEXAFS and EELS are both techniques that use the same physical principle.
Therefore it is valid to consider that in a homogeneous sample the spectral shape of
the C K NEXAFS and EELS spectrum should fit with each other. In Figure 6.6 the
corresponding NEXAFS and background substracted EELS spectra are aligned with
each other. To facilitate better comparison the intensities are normalized to the
π∗
feature. The NEXAFS spectra are in black while the EELS spectra are colored.
It is clearly visible that the NEXAFS spectra allow to individuate more spectral
79
6 NEXAFS
features than the EELS spectra. The reason for this and the narrower line-widths,
is the significantly higher energy resolution obtainable by synchrotron radiation
source compared to routine TEM EELS measurements.
EEL spectroscopy reveals higher intensities in the high energy region, especially
the
σ∗
feature is better visible by means of EELS than NEXAFS. The HOPG
spectrum obtained by EELS differs to the NEXAFS one as the "contaminations"
on the surface have less influence on the final spectral shape for the bulk technique
EELS than for the surface sensitive technique NEXAFS.
The comparison shows that the two techniques do not only differ significantly in
resolution but also in the relative intensities of the
π∗
and
σ∗
features, indicating
differences between the bulk and the surface.
6.2 Oxygen NEXAFS
In Figure 6.7 the O K NEXAFS spectra of the investigated samples are shown.
The intensities have been normalized through the ring current to compensate for
the decay in the photon flux with time. Additionally the intensities for the GfG
samples are divided by a factor of five to fit the same intensity range as the other
soot samples. This indicates the high oxygen amount of the sample which is also
shown by the significant features related to oxygen in the C K NEXAFS around
287 eV presented in Figure 6.1.
Hardly any analysis of the structural features present in the O K NEXAFS of
carbon samples can be found in literature. Most of the published work (examples
are summarized in Table 6.2) was carried out on model compounds or follows a
phenomenological approach: a peak at 533 eV peak was already ascribed in literature
[140] to the O1s
→π∗
excitation of carboxylic groups (COOH), while the feature
at 533.6 eV is reported in literature [93] to be due to a carbonyl group attached to
an alkyl chain. According to the experimental XPS measurements and theoretical
work carried out on model compounds by Clark and coworkers [141], the peak at
534.5 eV is supposed to be originated from O1s
→π∗
excitation of C-OH oxygen in
COOH group. Spectrum intensity of around 540 eV becomes larger with increase of
number of oxygen atoms in a molecule: This region is known to be composed from
σ∗resonance and excitation to continuum [76].
80
6.2 Oxygen NEXAFS
Figure 6.6:
Carbon NEXAFS and EELS spectra of the samples under
investigation.
81
6 NEXAFS
In the spectra of the GfG samples a clear peak at around 530 eV is visible. In
literature [142] it has been shown that the O1s spectra of all iron oxides can be
assigned to 530.1 eV. It is stated by Schedel-Niedrig and coworkers [142] that the
O 1s binding energy is independent of the iron oxide phase. Due to the fact that
Fe2p
XPS measurements revealed a presence of iron in our measurements, we can
assign this first peak to
FexOy
of the supported Fe-mesh. This
FexOy
contribution
makes the analysis more complicate since it is difficult to discriminate which part
of the spectra is due to the oxygen functional group on the surface of the carbon
samples and which part is due to the iron oxide of the support. Right afterwards at
around 531 eV another peak is visible which is to our knowledge not assigned in
literature. Through all the samples the GfG samples reveal the broadest peak in
the area between 537 and 542 eV. After the heating and oxidation process the only
observable differences in the GfG sample is a decrease in the intensity of the signal
in the region 530-532 eV. The Euro IV sample shows a less intense signal at around
531 eV compared to the GfG sample and the
FexOy
peak is completely missing.
The feature at 531 eV gets broader during the oxidation process up to 500
◦
C for the
Euro IV sample. The intensity in the area 533-535 eV is slightly more intense in
the Euro IV samples which can indicate a higher amount of heteroatoms compared
to the GfG sample.
Figure 6.7:
O K NEXAFS of HOPG, Flammruss 101, NC 3100, GfG, Euro IV
and Euro VI
82
6.2 Oxygen NEXAFS
Figure 6.8:
O K NEXAFS for the as received and oxidized Euro IV and VI
samples. The spectra are normalized to the edge jump at 550-560 eV bearing in
mind the different amount of oxygen present on the various samples
The edge jump at 550-560 eV gives an indication about the total amount of oxygen
in the sample. This feature is usually used for the normalization of the spectra
that have to be compared. XPS measurements and depth profile measurements as
shown and discussed in Chapter 7 show that this assumption is not valid in our
case. As the information depth obtained by O K NEXAFS is comparable to that
of XPS measurements with the kinetic photon energy of 1030 eV for oxygen and
785 eV for carbon we can consider the amount of oxygen determined by O 1s XPS
measurements to be comparable to that measured by O K NEXAFS. Therefore it is
valid to normalize the edge jump in the Euro spectra by a factor that represents
the different amount of oxygen present within the samples. A spectrum normalized
by this values for the Euro samples is shown in Figure 6.8.
With this now normalized spectra of Euro IV and Euro VI sample we can generate
difference spectra to reveal in a clear way the modifications that the samples undergo
due to the heat and oxidation treatment.
Figure 6.9 displays the O K NEXAFS for the Euro IV sample before and after
oxidation treatment. The difference in intensities between the as received and
oxidized sample is shown by the "red filled" curve, labeled as "Euro IV 20 - Euro
IV 500". The difference spectra in Figure 6.9 reveals that no significant changes in
83
6 NEXAFS
Figure 6.9:
O K NEXAFS spectra of the as received/oxidized Euro IV sample
and the difference spectrum which reveals modifications due to oxidation/heat
treatment
the oxygen amount and distribution of species can be observed as an result of the
oxidation and heat treatment. The two more significant features in the difference
spectrum are those at 529.7 and 530.6 eV. The feature below 530 eV is a result
of the sharpening of the peak at 530.6 eV after the oxidation treatment, which is
probably due to the fact that the peak at 530.6 eV for the as received sample is not
only due to oxygen-carbon bonds but also contains signal contributions from the
support. Such an
FexOy
signal would be sitting at 529.7 eV and would explain this
difference in the spectrum before and after the treatment. The positive feature at
530.6 eV can be attributed to changes in the C=O double bond amount.
The changes in the O K NEXAFS for the Euro VI sample are shown in Figure 6.10.
Already at first glance it is obvious that the Euro VI sample undergoes significant
changes due to the oxidation treatment, which differs significantly from the Euro
IV sample which has no striking changes at all.
The spectra shown in Figure 6.10 show that the Euro VI sample releases a
significant amount of its oxygen due to the oxidation and heat treatment. A loss
of features sitting at around 530 eV, 531.5 eV, 534 eV, 537 eV and 539 eV can be
observed. The difference at 530 eV could be due to
FexOy
contributions that are
not present with the same amount in both samples but the peak could be also due
84
6.2 Oxygen NEXAFS
Figure 6.10:
O K NEXAFS spectra of the as received/oxidized Euro VI sample
and the difference spectrum which reveals modifications due to oxidation/heat
treatment
to pyrone/quinone oxygen groups. The positive signal at 531.5 eV is assigned to a
loss of C=O species. The features at higher energy are associated to a splitting of
π
states (534 eV) that occur for conjugated C=O double bonds[143] while the latter
features are assigned to σtransitions[143].
The significant differences in the oxygen content and distribution between the
Euro IV and Euro VI sample can be additionally shown by a direct comparison
between the as received/oxidized Euro IV and Euro VI samples. Those difference
spectra are displayed in Figure 6.11.
The difference spectra in the case of the fresh samples are displayed without filling
while the differences between Euro IV and VI after oxidation treatment are shown
with a grey filling. The intense negative features reveal the significant higher oxygen
content of the as received Euro VI sample in comparison to the Euro IV sample. It
is also possible to determine features that are in a much stronger amount present in
the Euro VI sample such as the ones at an photon energy of 531-531.5, 534, 536.5
and 539 eV. The second peak at 531.2 eV is assigned to C=O in a graphitic structure
while features in the region of 533.5 eV are attributed to carbonyl groups attached
to an alkyl chain. We assign the peak at 535 eV to an increase of C-OH bonds.
85
6 NEXAFS
Figure 6.11:
Difference spectra of the Euro IV and VI samples. The plots for
the as received and oxidized ones are overlaid with each other.
After the oxidation treatment the difference spectrum of Euro IV and VI shows
still a significant negative intensities above 534 eV. By looking at the difference
spectra for the Euro IV sample (Figure 6.9), Euro VI sample (Figure 6.10) and
EuroIV minus Euro VI (Figure 6.11) it is clear that the oxygen amount of the Euro
VI sample decreases significantly while the amount and distribution for the Euro IV
sample does not undergo remarkable changes. Nevertheless it has to be mentioned
that the Euro VI sample looses a significant amount of species at 530.5 eV during
the oxidation process resulting in an final amount that is even less then the amount
of this species for the Euro IV sample. This lets us conclude that this species gets
more easily removed from the Euro VI sample than from the Euro IV sample. The
differences in the species at around 531.2 eV gets significantly reduced during the
oxidation treatment leaving a slightly higher amount of this species on the Euro VI
sample than on the Euro IV sample. A strong decrease of the relative amount of
species between 536 and 540 eV can be observed for the oxidized Euro VI sample
compared to Euro IV. This difference spectra lets us individuate another species
sitting at 540 eV which is usually assigned to σtransitions.
Those features in the difference spectra gives information about five species present
in the Euro VI sample. As NEXAFS does not suffer from any charging or chemical
shift these energy values can be used for the subsequent fitting of the O1s XPS.
86
6.2 Oxygen NEXAFS
The conclusions drawn from these difference spectra are that the Euro IV sample
seems to contain more species at 530.5 eV but less species at higher energies than the
Euro VI sample. It is obvious that the O K NEXAFS spectrum changes significantly
for the Euro VI sample before and after the oxidation and heat treatment. Those
spectra (Figure 6.10) reveal that the oxygen amount on the Euro VI sample gets
significantly removed during the oxidation treatment, which is in contrast to the
Euro IV sample (Figure 6.9) which does not seem to loose an remarkable amount of
oxygen during the oxidation treatment. This indicates that the Euro VI sample is
much more accessable to modifications than the Euro IV sample which seems to
present some "inertness" to the heat and oxidation treatment.
87
7 XPS
Carbon and Oxygen XPS measurements were executed to determine and assign
the different oxygen species present in the sample. To monitor the changes of the
oxygen surface species due to the oxygen and heat treatment temperature programed
In-situ XPS measurements were executed. Furthermore depth profile measurements
with varying photon energies were used to probe the surface and subsurface oxygen
composition.
X-ray photoelectron spectroscopy (XPS) has been performed at the synchrotron
radiation facility BESSY II using the ISISS beamline as a tuneable X-ray source
[84]. Samples were transferred into the reaction cell, 2
mm
away from an aperture
to the differentially pumped stages of the lens system of the hemispherical analyzer
(Phoibos 150, SPECS). Details of the set-up are described elsewhere [95].
The excitation energy used for C1s/O1s core level spectra were 585/830 eV,
respectively, resulting in a high surface sensitivity with an inelastic mean free path
(IMFP) of the photoelectrons of about 0.86 nm. The spectra have been normalized
to the impinging photon flux that has been determined by a clean Ag foil and
corrected for the fraction of higher order as well as the electron current in the
storage ring. Quantitative XPS data analysis was performed by using theoretical
cross sections [96]. The O1s spectra were fitted using mixed Gaussian-Lorentzian
functions after subtraction of a Shirley background [97] using Casa XPS software
[98]. In the spectra in which the species abundance was close to the detection
limit, a linear background has been subtracted. The fitting was done by fixing the
peak position within
±
0.1 eV for all spectra and constraining the full width half
maximum (FWHM) of 1.4 - 1.6 eV. The C1s spectra were fitted as following: The
graphitic peak was fitted in an asymmetric peak, well described by a Doniach-Sunjic
model [99]. The other peaks were fitted by using a mixed Gaussian-Lorentzian
component profiles after subtraction of a Shirley background.
89
7 XPS
7.1 C1s XPS
The analysis of the XP C1s spectrum has been extensively studied earlier in the
fields of polymer and graphite fibers science [92
–
94]. It was observed that the main
peak in the C1s spectrum of carbon material presents a long tail on the high binding
energy side. A lot of effort has been made in the scientific community [144] in
order to elucidate the nature of the asymmetry. It was shown that the high binding
energy tail has an intrinsic nature and must be included in the fitting procedure as
asymmetric peak after the removal of the extrinsic background.
The asymmetry in the high BE side of the C1s peak of defect-free graphite is
generally accepted to be associated with the excitation of conduction electrons which
occurs in metal or semimetals as consequence of the relaxation of the core-hole
generated by the primary photoemission process [145, 146]. The peak line-shape
is well described by the Doniach-Sunjic function [99] but the degree of asymmetry
depends on the structure defectiveness. Disruption of the HOPG structure by ion
bombardment produces a decrease of the intensity and the degree of asymmetry of
the main C1s peak [147]. This fact is a consequence of the disruption of the extended
electron delocalization in graphite which introduces a gap at the Fermi edge, and
thus reducing the occurrence of those phenomena responsible for the tail in the high
BE side of the C1s peak. Indeed, the presence of defects in the graphitic structure
leads to more localized electronic states which give arise to another component
in the C1s at higher BE than the main "graphitic" peak. For
sp3
bonded carbon
materials such as polymers, the asymmetry was correlated to the excitation of
atomic oscillations associated with the bonds within the polymer at the expenses
of the kinetic energy of the emitted photoelectrons. It is, thus, assumed that the
observed asymmetric lineshape in the XP spectra is a sequence of symmetric peaks
shifted to higher BE than the main peak. Since these peaks overlap with the primary
peak, all must be included in the peak model. In reality, nano-carbon materials
are heterogeneous and composed by a mixture of chemically bonded or physically
interacting
sp2
and
sp3
carbons, thus, rather than using a synthetic line-shape, the
XP spectrum of a model compound can be used as line-shape for that component
in the broad spectrum of complex carbon system. It is commonly used in literature
to fit the experimental C1s data with several contributions. For nano-structured
graphitic-like carbon material, the BE of the main "graphitic" peak may vary within
90
7.1 C1s XPS
Table 7.1:
Assigment for the functional groups used in the fits for the investigated
samples in this work
functional group Binding Energy /eV
sp3285.1
C-O 286.3
C=O 287.3
C-OOH 288.8
π→π∗290.7 294.3
few tenths of electron volts, usually the value reported for HOPG vary within 284.6
eV [25, 128] and 284.4 eV. The peaks above 285 eV are assigned to defective
sp2
graphitic structure and the
sp3
carbon species. At a further higher BE than the
sp3
carbon, all the carbon hetero-bonds are found.
The assignments of the structures in C 1s spectra are based on the following
assumptions [35]: For defect-poor graphite the binding energy was found at around
284.6 eV [128]. The
sp2
and
sp3
C-C and C-H bonds give rise to a C1s signal
at 285 eV. Carbon heterobonds will lead to a shift of the C1s signal to higher
binding energy. The more electronegative the hetero-element relative to carbon,
the larger the chemical shift to higher binding energy. Additional increments are
due to carbon-heteroatom double bonds and by 1,1 substitution of the C-C or C-H
bonds. In literature there is a large variety of chemical shift values for carbon-
oxygen for several carbon materials. Studies dealing with the characterization of the
oxygenated species on the surface of carbon fibers, show a variation of O1s binding
energies as already predicted with theoretical calculations and semi-empirical model
calculations 30 years ago by Clark et al. [93].
Table 7.1 presents the energy positions that were assigned for the specific functional
groups in this work. It has to be mentioned that features above 296 eV are different
for the samples investigated. This is due to the different structure of the samples
which is reflected also in different satellite structures in this region.
Plasmon satellite features occur from 5.6 to 7.2 eV for
sp2
carbons and from 11.3 to
12.5 eV in
sp3
carbon at higher BE than the main line [148] which hinder the chemical
shift identification [92, 149, 150] of highly oxidized C-O functional groups. Features
at 286.1 (C-OH), 287.6 (C=O) and 289.1 (C=O(OR)) are assigned according to
literature [35]. It has been shown[150] that the influence of the defect structure
91
7 XPS
Figure 7.1:
C1s XPS of HOPG, Flammruss 101, NC 3100, Euro IV and Euro VI
on the absolute position of the main line which is used as "internal" standard for
binding energy scale is often not considered. To avoid any misinterpretation of our
data the spectra are therefore not aligned to the main line but to the shake up at
291 eV. Furthermore the fitting of the main line has to be executed by use of a
asymmetric model, which is best fit by a Doniach-Sunjic model [99]. The asymmetry
is mainly due to the coupling of the core hole state created during the photoemission
to the semimetallic valence band states of graphite and by the convolution of the
primary photoemission with the phonon and plasmon loss spectra [145, 151].
Figure 7.1 shows the C1s spectra for the samples investigated and compared with
C1s of HOPG taken as reference. The C1s of HOPG represents the lineshape of
almost pure, defect-free graphite measured as a standard with a line width that
is mainly determined by the spectral resolution[152]. The graphs are aligned to
the shake up at 291 eV. The maxima for all samples with respect to the HOPG
92
7.1 C1s XPS
Table 7.2: FWHM table
Sample FWHM/eV
HOPG 0.48
FL 101 0.85
NC 3100 0.57
EURO VI 500 0.9
EURO VI 20 1
EURO IV 500 0.95
EURO IV 20 0.97
are shifted to higher binding energy (as shown by the inset in Figure 7.1), having
a higher FWHM and exhibiting a high intensity region between 285 and 291eV
attributed to oxygen functional group [24, 153].
Table 7.2 presents the energy position and full with at half maximum (FWHM)
for the main peak in the C1s spectra of the investigated samples. These values
indicate that the NC 3100 and HOPG sample have a comparable amount of carbons
present with
sp2
hybridization. The value of 0.57 eV compared to 0.48 eV for HOPG
indicates the high graphitization of this sample. The much higher values for the
soot samples, between 0.85 and 1 eV, indicate a higher amount of defects in the
structure, as a higher amount of defects leads to a broader distribution of binding
energies which are then reflected in a broader line width.
Interesting to observe is that the FWHM of the main peak in the Euro IV sample
decreases after the oxidation treatment, which would indicate that the structural
order increases. Figure 7.2 shows an attempt to fit the C1s spectrum of the as
received and oxidized Euro IV sample. It can be observed that the feature at 285.1
eV decreases in intensity after the oxidation treatment, while the features at higher
binding energy do not seem to be altered by the treatment. This decrease of the
species attributed to
sp3
bonded carbons and the decrease in the FWHM indicate a
increase of structural order for the Euro IV sample due to the oxidation treatment.
The main differences for the Euro VI are a decrease of the intensity at 285,1 and
286,3 eV as seen in Figure 7.3, which would correspond to a loss of
sp3
bonded
carbons and C-O.
Difference spectra of the as received vs oxidized Euro VI sample (Figure 7.4)
indicate an loss of
sp2
hybridized carbon (284.4 eV) atoms and also a strong loss
93
7 XPS
Figure 7.2:
C1s XPS of the as received(20
◦
C) and oxidized(500
◦
C) Euro IV
sample
Figure 7.3:
C1s XPS of as received(20
◦
C) and oxidized(500
◦
C) and Euro VI
sample
94
7.1 C1s XPS
Figure 7.4: Difference spectra of the as received vs oxidized Euro VI sample
Figure 7.5: Difference spectra of the as received vs oxidized Euro IV sample
95
7 XPS
of oxygen functional groups indicated by the positive signal between 285 and 287
eV. This indicates that the surface of the Euro VI sample undergoes modifications
and a significant amount of oxygen is lost with a parallel loss of carbon due to the
oxidation. The difference spectra of the as received and oxidized Euro IV sample
reveal a different response to the oxidation treatment. From the difference spectra
in Figure 7.5 it is clearly visible that the oxidation treatment increases the amount
of
sp2
hybridized carbon with a parallel loss of
sp3
hybridized carbon atoms. This is
also supported by the slight shift to lower binding energies of the oxidized Euro IV
sample with respect to the as received one. This can be explained by a restructuring
of the carbon atoms of the surface to increase the structural graphitic order of the
Euro IV sample.
7.2 O1s XPS
The assignment of the components in the O1s spectrum is quite controversial in
literature. As reported by Henschke and coworkers [44] even under vacuum condition
the presence of a film of molecular water interacting with the O species gives arise to
a large O1s signal. The description of the O1s XP spectrum consider the following
BE regions [43, 92, 154
–
161] which are summarized in Table 7.3: at the lowest BE
side a component at 530.7 eV is usually attributed to a highly conjugated form
of carbonyl oxygen such as quinine or pyridone groups; the components around
531.1-531.8 eV are assigned to a carbon-oxygen double bond; the region at around
532.4-533.6 eV is assigned to the carbon-oxygen single bond in hydroxyl group and
in ether-like configuration. In particular, most of the fitting procedure reported
in literature includes two components: the first component at around 532.6 eV
which is the lowest thermal stable species while the second component at around
533.6 eV is desorbed at higher temperature. However from literature analysis the
precise assignment of these components to a C-O single bond chemical configuration
is very hard since the position of water is quite debated in literature [157
–
160].
Exhaustive studies of adsorption of water carried on polycrystalline graphite have
lead to the identification of very broad peak centered at 533 eV [159]. Furthermore,
some authors assign the signal above 535 eV to adsorbed water and/or oxygen.
96
7.2 O1s XPS
Table 7.3: XPS O1s assignment according to [162]
O functional groups Binding Energy (eV) Material
Quinones 530.9 CNT [154]
531.1 AC [158]
530.7±0.1CNF [163]
C=O 531 CF [92]
531.7, 532 CNT [155]
531.4 ACF [156]
530.6 AC [157]
532.3 AC [158]
531.6 CF [160]
531.1-531.8 CF [161]
530.3 Graphite [43]
531.4±0.1CNF [163]
OH 534 CF [92]
534.2 AC [158]
533.6 CF [160]
532.6 CF [161]
533 Graphite [43]
533.7±0.1CNF [163]
C-O 532.5 CF [92]
533.3-533.8 CNT [155]
532.3 AC [157]
533.3 AC [158]
532.4 CF [160]
533.5 CF [161]
531.6 Graphite [43]
532.4±0.1CNF [163]
H2O533.4 CNT [154]
533.5 ACF [156]
536.3 AC [157]
535.9 AC [158]
533 Graphite [159]
535-536 CF [160]
535.2 CF [161]
535 ±0.1CNF [163]
97
7 XPS
Figure 7.6:
O(1s) XPS of Flammruss 101, NC 3100, GfG, Euro IV and Euro VI
98
7.2 O1s XPS
Table 7.4:
Oxygen and carbon concentration in Euro IV, Euro VI, NC 3100 and
Flammruss 101
sample % of oxygen % of carbon
Euro IV as received 6.3 93.7
Euro IV oxidized 6.1 93.9
Euro VI as received 12.1 87.9
Euro VI oxidized 9.2 90.8
NC 3100 2.1 97.9
FL 101 6 94
The assignment of the species in the samples under investigation is illustrated in
Figure 7.6 and detailed values for the evolution of the single contributions can be
found in Table 7.5. A decrease of the species at around 533.7 and 531.5 eV can be
observed for the Euro VI sample after the oxidation treatment, as shown in Figure 7.6,
while the features for the Euro IV sample do not follow this trend.Table 7.5 clearly
shows the relative increase of carbonyl groups situated at around 530.6 eV from an
initial percentage of 5 to 17 %of the total oxygen content.
The atomic concentration of oxygen and carbon for the investigated samples are
displayed in Table 7.4. It is shown that the heating and oxidation process leads
to a reduction of the total amount of oxygen bond in the soot samples. The NC
3100 sample has a very low oxygen content which fits with the C K NEXAFS
spectrum that shows a narrow C K edge peak and a nearly oxygen function free area
between 286-290 eV. The amount of oxygen in the FL 101 sample lies between the
values for the nanotube and the soot samples. This also agrees with C K NEXAFS
measurements. It is displayed there that the FL 101 and NC 3100 sample have a
quite similar shape as the HOPG sample.
For the GfG sample charging problems occurred during the measurements with
synchrotron synchrotron. Nevertheless the GfG samples were investigated in an
ESCA machine and data for O1s and C1s XPS are available, but due to different
calibration of the energy window scale, the different resolution and the different
depth information in the two machines a direct comparison is not possible. Those
charging problems occur often for highly functionalized carbons. Therefore this
charging can be seen as an indicator for a high amount of heteroatoms on the surface
of the GfG samples. This is supported by the relatively low amount of
sp2
bonded
99
7 XPS
Table 7.5:
Absolute oxygen abundance for the individual contributions in the
Euro IV and Euro VI samples
Binding Energy functional Euro IV Euro IV Euro VI Euro VI
/eV group as received oxidized as received oxidized
529.8 FexOy0.043 0.024 0.149 0.192
530.6 quinone 0.044 0.319 0.22 0.835
531.4 C=O 1.56 1.456 4.072 2.458
532.5 C-O 1.623 1.374 3.052 2.131
533.7 C-OH 2.599 2.556 4.044 2.995
> 534 absorbates 0.381 0.367 0.646 0.563
carbon in the GfG samples as revealed by EELS and the poor structure visible by
HRTEM.
7.3 Depth profile
In this part depth profile XPS measurements on the Euro IV and VI samples are
presented. The measurements on the O1s and C1s core levels were executed with the
aim to get information about changes in the distribution of the oxygen functional
groups in the samples as a function of the depth. Deeper layers of the samples
are probed by using high photon energies since the corresponding photoelectrons
have higher kinetic energy and consequently higher escape depths. Greater sample
depths are in our way obtained with a higher energy the X-ray source, by giving
the core electrons greater kinetic energy. The drawback of this approach is, that
correspondingly the photoemission cross-section decreases with increasing photon
energy, which may possibly lead to sensitivity problems.
Depth profiling is used to determine the elemental composition of heterogeneous
samples [164]. By use of this method chemically distinguishable layers can be
exposed to the photons and can be measured[82]. In this work, photon-energy-
dependent depth-profiling is used to determine the concentration profile of the
subsurface oxygen species within the first few nanometers below the surface. The
escape depth of electrons depends on their kinetic energy. The relation of inelastic
mean free path versus kinetic energy is plotted in Figure 7.7 which is called the
universal curve of electron mean free path. Those values were theoretically predicted
100
7.3 Depth profile
Table 7.6:
Absolute oxygen abundance calculated from the depth profile mea-
surements. Values on the right side reflect more bulk sensitive, while the left side
values are determined from a more surface sensitive measurement.
surface →bulk
EexcO1s/C1s/eV 630 / 385 830 / 585 1030 / 785
Euro IV as received
absolute oxygen abundance 12.7 6.3 5.5
Euro IV oxidized
absolute oxygen abundance 15.4 6.1 5.2
Euro VI as received
absolute oxygen abundance 21.7 12.2 11.2
Euro VI oxidized
absolute oxygen abundance 19.4 9.2 8
by Penn [165] and experimentally determined by Rhodin and Gadzuk [166] and
Somorjai [167].
The mean free path curve Figure 7.7 has a broad (note the log-log scale) minimum
around a kinetic energy of about 70 eV, which gives around 5 Å. This means that
most of electrons with this kinetic energy, which has left the solid without suffering
an inelastic scattering event must originate from the first few layers. In this work
we tune the photon energy for probing the C1s/O1s region from 385/630 eV over
585/830 to 785/1030 eV. This is equivalent to a kinetic energy of the photoelectrons
of 100, 300, 500 eV which is following to the values given from Figure 7.7 equivalent
to a mean free path of 5 Å, 7 Åand 9 Å. It has to be mentioned that those values
were not obtained for "soot-like" carbon but rather graphitic carbon. Due to the
quite different electron density we have to assume that the real mean free path for
our system is one order of magnitude bigger.
The evolution of the absolute oxygen abundance of the as received/oxidized Euro
IV/VI samples depending on the photon energy are shown in Table 7.6. It can
be seen that the chemical composition of the surface and the near-surface region
varies significantly for the Euro IV and Euro VI samples. The as received Euro VI
sample has an nearly twice as high oxygen content on the surface compared to the
as received Euro IV sample. By using higher photon energy to get also contribution
from deeper layers it can be clearly seen that the oxygen content for the as received
Euro VI sample decreases significantly. A change of the photon energy from 630/385
101
7 XPS
Figure 7.7: Universal Curve of electron mean free path vs electron energy[168]
eV to 1030/785 eV changes the oxygen amount from 21.7 to 11.2 %. In comparison
to that the corresponding measurements for the as received Euro IV sample show
a oxygen amount of 12.7 decreasing to 5.5 %. This indicates that the amount of
oxygen functional groups present on the as received Euro VI is significantly higher
than on the Euro IV sample. This higher amount of functionalization will be related
to an higher reactivity and therefore higher accessibility to the environment. This
high oxygen concentration also fits with TPO results showing an higher reactivity
for the Euro VI sample compared to the Euro IV soot.
Depth profile measurements after the oxidation treatment for the Euro IV and
VI soot samples are also shown in Table 7.6. The oxidized Euro IV sample shows
beside a higher oxygen content on the surface a similar behavior as the as received
Euro IV sample. This indicates that the oxidation/heat treatment does not have
big effect on the oxygen/carbon composition in the Euro IV soot. This indicates an
lower reactivity of this sample.
For the oxidized Euro VI sample a significant changes compared to the as received
Euro VI sample can be observed. The oxygen content of the oxidized sample drops
from an initial value of 19.4 % over 9.2 to 8 % with increasing photon energy. In
comparison to the as received Euro VI sample this represents an loss of
≈
30 % of
the subsurface oxygen. This means that the high subsurface oxygen amount in the
102
7.3 Depth profile
as received sample gets removed during the oxidation treatment, which leads to the
conclusion that the Euro VI sample is highly reactive.
The distribution of the single oxygen species for the depth profile measurements
are summarized and discussed in Table 7.7 and graphically visualized in Figure 7.8,
Figure 7.9 and Figure 7.10 for the Euro IV and Euro VI samples. As already shown
in Table 7.6 the Euro IV sample has much lower oxygen content subsurface than on
the surface. The main contributions for the as received Euro IV sample are C=O,
C-O and mainly C-OH bonds. For the more bulk sensitive measurements those
species decrease in intensity but remain the most significant ones with a stronger
decrease for C=O than for the other species. The oxidized Euro IV sample shows
that within the first
≈
5 Å a significant increase of quinone species can be observed,
while the amount of C=O stays constant. Additionally C-O and C-OH contributions
increase also but in a less pronounced way than the quinone-like species. Beside the
formation of quinone-like species in the oxidized Euro IV sample the evolution of
the species for the more bulk sensitive measurements is consistent with the results
obtained for the as received one.
As we already know from the previous discussion and from the data shown in
Table 7.6 the amount of oxygen in the as received Euro VI sample is significantly
higher than the one present in the as received Euro IV sample. While for the Euro
VI 20 sample an amount of quinone-like species comparable to those in the Euro
IV 20 sample is found, the percentage of C=O and C-O is nearly twice as high in
the Euro VI 20 sample. This persists when measuring more subsurface sensitive.
Important to mention is that the species at
≈
530
.
6
eV
assigned to conjugated
C=O, such as quinone species, does not significantly decrease which is in contrast
to the data obtained for the Euro IV 20 sample. Those data indicate that the
as received Euro VI sample is more highly functionalized, due to the presence of
quinone-like species and parallel to that a higher degree of oxidation states is present,
due to higher contributions in the range of C-OH species, than in the Euro IV 20
sample.
After the oxidation and heat treatment the composition of species in the Euro
VI sample is altered differently than the Euro IV sample. A strong increase of the
species at 530.6 eV is observed while a significant loss of C=O and C-O functional
groups is observed. The amount of C-OH groups is reduced more significantly in the
subsurface than in the surface region. The more significant increase of quinone-like
103
7 XPS
Table 7.7:
Absolute oxygen abundance for the individual contributions in the
Euro IV and Euro VI samples with increasing Eexc
sample Photon Energy FexOyquinone C=O C-O C-OH absorbates
/eV 529.8 530.6 531.4 532.5 533.7 > 534
Euro IV 20 630 / 385 0.07 0.31 3.34 2.66 5.56 0.77
830 / 585 0.04 0.04 1.56 1.62 2.59 0.38
1030 / 785 0.02 0.05 0.55 1.67 2.55 0.66
Euro IV 500 630 / 385 0.02 0.99 3,44 3,30 6,72 0,92
830 / 585 0.02 0.32 1.46 1.37 2.56 0.37
1030 / 785 0.02 0.15 0.99 1.31 2.44 0.32
Euro VI 20 630 / 385 0.24 0.33 7.59 4.57 7.71 1.21
830 / 585 0.15 0.22 4.07 3.05 4.04 0.65
1030 / 785 0.06 0.32 3.54 2.75 3.97 0.59
Euro VI 500 630 / 385 0.35 1.10 5.56 3.78 7.37 1.22
830 / 585 0.19 0.84 2.46 2.13 2.99 0.56
1030 / 785 0.13 0.21 2.41 1.72 2.98 0.55
species in the surface and subsurface region and the loss of C-OH functional groups
especially in the subsurface region in the Euro VI sample due to the oxidation
process are the most significant and important differences to the Euro VI sample.
104
7.3 Depth profile
(a)
(b)
Figure 7.8:
Evolution of the absolute amount of the oxygen species for the as
received (a) and oxidized (b) Euro IV sample with increasing photon energy.
105
7 XPS
(a)
(b)
Figure 7.9:
Evolution of the absolute amount of the oxygen species for the as
received (a) and oxidized (b) Euro VI sample with increasing photon energy.
106
7.3 Depth profile
(a)
(b)
Figure 7.10:
Evolution of the absolute amount of the oxygen species for the as
received and oxidized Euro IV (a) and Euro VI (b) sample with increasing photon
energy.
107
8 Discussion
In this chapter the results from the previous sections are summarized and correlated
with each other. By combining the data obtained from bulk characterization such as
HRTEM, EELS and Raman with results from surface sensitive studies, as obtained
by NEXAFS and XPS, we are able to draw a picture of the samples. In addition to
this, the oxidation process is discussed with reference to the modifications that the
Euro samples undergo during the oxidation process and how this is related to the
reactivity and accessibility to the environment.
HRTEM studies as executed in this work and shown in Chapter 4 enable us to
deduce the microstructure of the Euro IV, Euro VI and GfG soot samples. The long
range order of the GfG sample differs significantly from the two emission standard
samples. The Euro soot particles are built up of stacked graphene layers (BSU),
which form graphitized spherical particles.
Additionally the Euro IV sample is partially encased with a layer of molecular
carbon while the surface of the Euro VI sample is free of molecular carbon debris.
While the Euro samples are characterized by spherical carbonaceous particles that
tend to agglomerate to bigger units, the GfG sample does not show long range
order and presents structural features which can be assumed to have a chain-like
character.
Curvature measurements have been performed to evaluate deviations from pla-
narity for the graphitic segments. Defects present in the graphitic segments such as
heteroatoms or pentagons/heptagons (non-six membered rings) as well as
sp3
hy-
bridized carbon bonds may induce bending. The degree of bending is representative
for deviations from planar
sp2
bonding in defect free graphite to
sp3
configuration
in aliphatic hydrocarbons. The GfG sample show a strong bending of the graphitic
segments while the bending for the Euro samples is significantly less pronounced.
109
8 Discussion
Among the Euro samples the Euro VI soot shows stronger bending than the Euro
IV sample.
The Euro VI sample shows a higher amount of
sp2
hybridization, as revealed by
EELS, than the Euro IV sample, which is mainly due to the fact that the Euro IV
sample contains a significant amount of molecular carbon, adding additional
sp3
hybridized carbon signal to the EELS spectrum. The GfG sample has a very low
amount of
sp2
hybridized carbon compared to the Euro samples. This is consistent
with the HRTEM and curvature analysis that reveal a chain-like structure for the
GfG sample, in which the sample is mainly built up by bonding of
sp3
hybridized
carbon atoms.
The bulk studies of the Euro VI sample show a somehow "inertness" to the heat
and oxidation process. Neither high resolution TEM micrographs could show any
structural modifications nor changes in the electronic structure could be determined
by means of EELS. Moreover The ratio of
sp2/sp3
hybridized carbons was unaffected
and the plasmon features did not show modifications. Additionally the size of the
Euro VI soot particles did not undergo measurable changes during heat treatment
and oxidation. Although the statistical curvature measurements did not show strong
changes, a significant shortening of the graphene segments could be measured.
In contrast, the Euro IV sample showed changes during the treatment. High
resolution TEM shows that molecular carbon units on the surface of the Euro
IV sample were removed during the oxidation treatment. We speculate that the
presence of these molecular carbon units is due to a temperature-gradient in the
combustion cell which favors regions with relatively low temperature. In such areas
it would be possible that small PAHs "survive" and then quench on the surface of
the Euro IV soot sample. The observations regarding removal of molecular carbon
are supported by an increase of the amount of
sp2
hybridized carbon atoms in the
Euro IV sample during the oxidation treatment. Parallel to that we could show by
curvature measurements that a flattening and elongation of the graphene fragments
is observed during the oxidation treatment. Together with EELS this shows that the
structural order increases during the oxidation treatment for the Euro IV sample.
Particle size measurement reveal a shrinkage of the particle size and a narrowing of
the size distribution, implying that very small Euro IV soot particles are completely
combusted during the oxidation treatment and that the bigger particles shrink,
110
which could be explained by the loose of molecular carbon units during the oxidation
treatment, revealed by HRTEM (Chapter 4).
The GfG sample shows already by means of bulk characterization unexpected
modifications for a so called "soot reference sample". As mentioned above the GfG
sample has a significantly different morphology than the Euro samples and undergoes
the most striking change during the oxidation process amongst the samples under
investigation. HRTEM showed a "more densely packed" morphology after the
oxidation process, which is also supported by EELS measurements that show a
enormous change in structural order. A significant amount of
sp3
hybridized carbon
is transformed into atoms bonded in
sp2
configuration. As seen in the C K EELS
measurements this change can be attributed to the loss of intensity at 287 eV which
is assigned to C-H bonds.
The results obtained in this study were correlated with Raman studies executed
in a collaboration with the TU Munich. The aim of this collaboration was to give
a scientific/physical background and explanation to a fitting procedure for carbon
Raman spectra. This fitting procedure developed by this group is presented in
Figure 8.1. Prof. Niessners groups establish a hypothesis based on the model of
Sadezky et al.[23] that it is possible to determine the structure of a sample by fitting
the Raman spectrum of this sample.
Figure 8.1:
Raman spectra with fitted curves shown by the example of the as-
received Euro VI sample [23]
111
8 Discussion
The results obtained in this work for the as-received samples are fitting with
Raman experiments [51] that show a significantly different D and G peak shape
for the Euro samples compared to the GfG sample. While the Euro samples show
sharp D and G band peaks the GfG sample shows strongly overlapping D and
G band peaks. The two Raman peaks of Euro IV and Euro VI soot are more
separated than the GfG soot, which lead to the assumption that they present a
more homogeneous structure with a lower content of molecular carbon than the GfG
soot. HRTEM and EELS experiments showed that this assumption is valid. Raman
spectroscopy showed slightly variations in the D1 FWHM upon oxidation for Euro
VI soot compared to Euro IV soot, which suggests the presence of a more disordered
and hence slightly more reactive structure for Euro VI soot. This is in agreement
with the curvature analysis and the XPS analysis of the oxygen functional groups.
For the oxidized samples EELS data are consistent with Raman data [51] that only
show minor changes in the spectral shape for the Euro IV and Euro VI samples.
The most obvious changes are visible for the GfG sample as result of the oxidation
and heat treatment. Those modifications are also observed by HRTEM and EELS
indicating that a condensation reaction occurs. Overall, it could be demonstrated
that the fitting procedure applied on the Raman spectra and the interpretation
concluded from that points in the same direction as the results obtained by HRTEM
and EELS experiments.
Reactivity towards oxygen was determined by TPO measurements of the GfG
and Euro samples. As expected from the microscopy studies, the GfG sample is
highly reactive, as shown in Figure 8.2. The low oxidation temperature indicates
that the GfG sample is not build up of graphitic segments but rather molecular
hydrocarbon chains.
In contrast to the EELS and high resolution TEM data which show no modifica-
tions for the Euro VI sample and slight modifications for the Euro IV sample, TPO
reveals that the Euro VI sample is more reactive than the Euro IV sample. The
onset sitting at lower temperature of the mass conversion for the Euro IV compared
to the Euro VI sample may be explained by the loss of the molecular carbon units
sitting on the surface of the Euro IV sample. The Euro VI sample is oxidized later
(onset at higher temperature) than the Euro IV sample but when the oxidation
begins the rate of oxidation is faster. The much higher accessibility to structural
modifications for the GfG sample compared to the Euro samples was already shown
112
Figure 8.2:
Mass conversion vs temperature for GfG, Euro IV/VI soot and
Graphite [51].
by means of HRTEM, EELS and Raman. The TPO measurements show the same
behave in a very clear way. With this studies we can clearly say that the GfG soot
is under no circumstances a valid model soot material. For further studies one has
to be aware of this problem and think about another more suitable "soot reference
sample".
XPS and NEXAFS measurements were undertaken to gain insight into the
electronic configuration and its changes during oxidation treatment. Furthermore
these surface techniques enable one to deduce the oxygen functional groups and
reveal how their presence and mobility affects environmental accessibility of the Euro
samples in different ways. C1s XPS data as presented in Chapter 7 show an increase
in the structural graphitic order for the Euro IV sample, which is accompanied by a
simultaneous loss of defective
sp2/sp3
states during the oxidation process. This is
consistent with statistical curvature measurements on HRTEM micrographs and
EELS measurements that show the same behavior as discussed above. The same
modifications are observed by C K NEXAFS analysis of the Euro IV sample which
also shows an increase in the intensity of the resonance at 285 eV after oxidation,
which can be attributed to
sp2
hybridized carbon atoms. In addition to that we
observe a decrease of the
σ∗
resonance signal, associated to a loss of
sp3
hybridized
carbon atoms. The loss of
sp3
carbon species and the decrease of the signal below
113
8 Discussion
285 eV are consistent with the burning of molecular carbon, sitting on the surface of
the soot particles during oxidation, which was observed also by means of HRTEM.
Those results let us conclude that the Euro IV sample during the oxidation preserved
a more graphitic structure while disordered surface carbon is preferentially removed.
The XPS C1s analysis of the Euro VI sample reveals a loss of defective
sp2/sp3
carbons (at 285 eV) and a loss of oxygen functional groups (>285 eV). This is also
observed in the C K NEXAFS spectra of this sample as loss of structural features
in the region between 286-288 eV. The different sharpness of the resonance in the
NEXAFS C K spectra shows stronger graphitic order for the Euro VI compared to
the Euro IV sample. This is a general results from all the characterization techniques
used in this thesis. Also, C 1s XPS and C K NEXAFS reveal a stronger feature in
the region assigned to carbon heteroatom species for the Euro VI compared to Euro
IV soot.
O1s XPS data reveal a higher concentration of surface and subsurface oxygen for
the as received Euro VI than for the Euro IV sample. Both samples show a roughly
50 %lower subsurface oxygen content compared to the surface oxygen amount
(Chapter 6).
Upon oxidation, Euro VI loses a significant amount of surface and subsurface
oxygen, whereas the Euro IV sample retains a similar amount of subsurface oxygen,
and the surface oxygen amount increases slightly.
For the Euro IV sample we observe an increase of surface C-O, C-OH species and
a strong increase of the species at 530.7 eV (quinone-like oxygen species). This is
explained by the generation of additional C=O double bonds in the neighborhood
of the originally present C=O double bonds (531.7) in a conjugated double bond
chemical configuration (quinone-like configuration) which is induced by the oxidative
thermal treatments.
By analyzing the difference spectra it was possible to obtain more specific infor-
mation from the O1s NEXAFS spectra. A clear difference between Euro IV and
Euro VI samples can be seen. The Euro VI sample contains a higher heterogeneity
of oxygen functional groups with more C=O, C-O and C-OH contributions than
the Euro IV sample but a similar amount of quinone groups. In conclusion, the
depth profile measurements support that the Euro VI sample is highly oxidized and
undergoes a significant reduction in its oxygen content due to the oxidation/heat
114
treatment, whereas the lower oxygen amount of the Euro IV sample gets slightly
enriched during the oxidation treatment.
The "porosity" of the GfG sample, which is a result of the chain-like "structure"
facilitates a fast oxidation/combustion process. Therefore such enormous changes,
as observed by HRTEM, EELS, TPO and RAMAN can be observed for the GfG
sample. For the Euro samples in contrast no such "porosity" could be observed. A
"core-shell particle concept" with the Euro particles having an "onion-like" structure
can be assumed for the Euro particles. Such a arrangement can be observed in the
high resolution micrographs in Chapter 4. The result of the study undertaken here
makes it possible to conclude that the oxidation/combustion process occurs from
outside to the inside of the soot particles. Such a combustion process is slowed
down by the structure of the particles. This "burning" shell by shell fits with the
particle size measurements that show a narrowing of the Euro IV particle size after
the oxidation treatment. This can be explained by the removal of outer "shells".
Parallel to that the carbon atoms inside the particle rearrange themselves and
increase their structural order. The Euro VI sample does not undergo changes in
the particle size distribution, which would mean in the "core-shell particle concept",
that the outer shell hinders the oxidation/combustion treatment. This would mean
that the particles are not combusted at all or if combustion starts the Euro VI
particles combust completely, due to the higher reactivity and heterogeneity of
oxygen functional groups in the Euro VI sample than in the Euro IV sample. Such
high amount of oxygen species heterogeneously in nature and therefore in reactivity,
might lead to decomposition and oxidation which would explain the complete
combustion. Important to mention is the fact that the Euro VI sample does not
only reduce the oxygen content, but it changes the distribution of functional groups.
Less stable oxygen groups are removed during the oxidation treatment or converted
into more stable ones, as shown by the depth profile study.
Based on those experimental results we can draw a reaction scheme for the Euro
samples:
Euro VI reaction pathway
For the Euro VI sample we assume a reaction sketch during the oxidation treatment
as displayed in Figure 8.3. Due to the fact that we have more bending, a higher
115
8 Discussion
Figure 8.3: Reaction sketch for the Euro VI sample
oxygen content and longer graphene fragments in the Euro VI sample we can
assume that the graphene fragments are decorated by oxygen species with lower
thermal stability (species at higher binding energy). These graphene fragments are
interconnected by an "oxygen-bridge".
During the oxidation treatment, the higher oxygen content of these species favors
the complete combustion of the particles in the outer part of the soot agglomerates.
Considering that we have a gradient of oxygen we can assume that we have for
the outside particles an oxygen rich atmosphere while inside the agglomerates a
reducing atmosphere is present. Therefore the particles inside undergo condensation
and thermal decomposition of oxygen species. Considering the high oxygen content
of this sample the probability of neighboring oxygen species is very high. Therefore
intramolecular condensation reactions may occur. This could explain the shortening
of the fibre length measured by HRTEM analysis, where an average fibre length of
2 nm gets reduced to around 1 nm.
The occurring of condensation reaction is supported by the modification of the
distribution of the oxygen species observed by O1s XPS. The nature of the oxygen
species is changed as
H2O
and
CO2
is lost while quinone groups are formed. The
more stable oxygen groups are formed while less stable ones are removed or converted
into more thermical stable ones.
Euro IV reaction pathway
For the Euro IV sample a different reaction scheme is assumed, as shown in
Figure 8.4. The graphitic segments are characterized by a lower oxygen content.
116
Figure 8.4: Reaction sketch for the Euro IV sample
This means that quite probably those graphene clusters are not interconnected by
an "oxygen-bride", as indicated by the lower curvature. The lower oxygen content
and the observation of the growing of the graphene fragment lengths suggest that
we possibly have intermolecular condensation. In the inner shell of the particles
intermolecular dehydrocyclisation [169] between different graphene domains occurs
leading to the increase of the graphene domain size.
During the oxidation we preferentially remove the molecular carbon units sitting
on the surface. However the low oxygen content suppresses the complete combustion
of the particles and favors thermal condensation of the core. The surface after the
117
8 Discussion
oxidation treatment is slightly enriched with oxygen species which would be the
precursor state of further oxidation.
These two pathways are not exclusive for either sample, however the study
undertaken reveals that there is a preferential pathway for the condensation reaction
for either Euro IV and VI.
118
9 Conclusion
The combination of HRTEM, EELS, Raman, TPO with XPS and NEXAFS as
executed in this work makes it possible to reveal all information necessary for the
understanding of the reactivity of carbonaceous samples, shown on the example of
two emission standards, Euro IV and Euro VI. This thesis points out that despite
the fact that surface and bulk characterization techniques are complementary they
are indispensable to fully explore and understand the properties of a sample subject
of investigation for reactivity studies.
On the basis of the results presented and discussed in the previous chapters we
can propose a hypothetical model for the structure of Euro IV and Euro VI. With
respect to their growing conditions it is possible to conclude a difference in reactivity
for the Euro IV and Euro VI soot sample. From the data obtained in this study for
the Euro IV and Euro VI samples we can conclude a structural model:
The Euro VI sample consists of a highly graphitic core with a highly oxidized
surface. This can be due to the fact that the temperature in the engine was higher
and therefore the carbon not exposed to oxygen gets graphitic and carbon exposed
to oxygen gets highly functionalized. O 1s XPS depth profiling proves the highly
functionalized and reactive surface of the Euro VI samples. The different results
obtained by NEXAFS and EELS for the Euro VI sample support furthermore the
idea of the heterogeneous sample with the highly inert core (therefore no changes in
sp2/sp3
content observed by EELS) and the highly reactive surface (changes during
oxidation observed by NEXAFS).
The Euro IV sample was probably oxidized at lower temperature and therefore
more molecular carbon is present. This leads to more changes in the bulk during
heating. This is supported by NEXAFS features at 290 eV which are more pro-
nounced than in the Euro VI sample, indicating more hydrocarbons in the Euro VI,
and by difference spectra. Additionally EELS revealed an increase in the relative
sp2content.
119
9 Conclusion
The GfG sample has a poorly graphic order with respect to the Euro samples
and undergoes significant structural changes during the heating treatment. This
can be explained by the poor structural order (chain-like) and the high amount of
oxygen in the sample. The high porosity given by the chain-like structure and the
high oxygen content lead during the heat treatment to a significant reordering of
the structure, most probably due to a condensation reaction, which is most likely to
occur for such a highly molecular functionalized sample.
A conclusion of this work for the GfG sample is that it cannot be used as a
reference for soot emitted by diesel engines due to significant differences in structure
and oxidation treatment. A reference sample that is much more reactive than the
other samples under investigation cannot represent a reference.
This work is relevant for health risk assessment as it shows clearly that the
newest type of emission standard "Euro VI" is more reactive and accessible to the
environment than the current standard in use, "Euro IV". It is a clear evidence
that the route of the European Commission to limit only the amount of emitted
incomplete combustion products is not the best approach. Rather than be limiting
to the reduction of the emission, a sustainable approach should be addressed to
reduce the reactivity of the emitted particles toward the environment.
The onion-like structure of the Euro samples hinders the combustion process. For
the Euro IV sample the conditions show already some effect as the particle size
shrinks due to the combustion process. For the Euro VI sample in contrast the
conditions are too mild to obtain total combustion. Contrary to the expectations
the treatment does not decrease the reactivity of the Euro VI particles but rather
makes them more reactive as it leads to an enrichment of the most stable basic
oxygen functional groups, which are the most reactive species. Therefore the thermal
treatment has to be optimized to ensure a complete combustion of the soot particles.
A faster heating to higher temperatures has to be applied to prevent a reordering
of the inner "shells" of the particles, which would lead to a certain "inertness" to
combustion. If the heating occurs fast enough to prevent the formation of highly
stable oxygen functional groups and to prevent the restructuring of the inner shells
of the particles further combustion is possible.
120
10 Acknowledgement
I want to thank all the people who supported me during my stay in Berlin. First of
all I want to express my special gratitude to Prof. Dr. Robert Schlögl, who gave
me the opportunity to do this work in the department of Inorganic Chemistry at
the Fritz Haber Institute.
The assessment of this thesis by the Technische Universität Berlin is gratefully
acknowledged: in particular I want to thank Prof. Dr. Mario Dähne and the
committee members of my doctoral board. I want to thank my supervisor, Dr. Dang
Sheng Su for his advice and assistance during the development of the scientific work.
My thank goes also to all the members of my group, "Microstructure". Especially
I want to thank Gisela Weinberg for introducing me to the SEM and for being
always helpful and Norbert Pfänder for discussion. I want to thank also Hermann
Sauer who was sharing his EELS experience with me and encouraged me to be more
critical with my data.
I thank the members of the Surface Analysis group, especially Dr. Axel-Knop-
Gericke for giving me the possibility to measure at BESSY and Dr. Raoul Blume for
his help in measuring and assisting me in the analysis of the XPS data. Furthermore
I want to thank Dr. Michael Hävecker for his help in measuring, teaching and
discussing the analysis of the NEXAFS data and for his critical comments.
I am thankful to Thomas Patric Cotter for correcting the English content and
Dr. Marc-Georg Willinger for discussion. My thank goes also to Stefanie Kühl for
helping me to refresh my LaTEX knowledge and showing me the "small" tricks.
My special thanks goes also to Dr. Rosita Arrigo who was always supporting and
giving a lot of contribution in helping me to overcome the gap between a physicists
and chemists view.
I want to thank Prof. Cécile Hébert for encouraging me to do my PhD at the
FHI and for introducing me to microscopy during my early studies.
Finally I want to thank my family who supported me along the way.
121
Bibliography
[1] U. Colombo, M. W. Thring, Combustion Science and Technology
5(1972) 189.
[2] B. S. Haynes, H. G.Wagner, Progress in Energy and Combustion Science
7(1981) 229.
[3] W. Stöber, U. R. Abel, International Archives of Occupational and
Environmental Health 68 (1996) S3.
[4] H. G. Neumann, Chemosphere 42 (2001) 473.
[5] J. Kagawa, Toxicology 181-182 (2002) 349.
[6] R. S. Hamilton, T. A. Mansfield, Atmospheric Environment 25A 25
A(1991) 715.
[7] A. Seaton, D. Godden, W. MacNee, K. Donaldson, The Lancet
345 (1995) 176.
[8]
R. M. Harrison,J. Yin, The Science of The Total Environment
249
(2000) 85.
[9] K. Donaldson, X. Y. Li, W. MacNee, Journal of Aerosol Science
29 (1998) 553.
[10] D. Stadler, M. J. Rossi, Phys. Chem. Chem. Phys. 2(2000) 5420.
[11] A. M. Nienow, J. T. Roberts, Annual Review of Physical Chemistry
57 (2006) 105.
[12] V. H. Grassian, The Journal of Physical Chemistry A 106 (2002) 860.
123
Bibliography
[13] A. Braun, N. Shah, F. E. Huggins, K. E. Kelly, A. Sarofim, C. Jacobsen, S.
Wirick, H. Francis, J. Ilavsky, G. E. Thomas, G. P. Huffman, Carbon
43 (2005) 2588.
[14] A. Braun, B. S. Mun, F. E. Huggins G. P. Huffman, Environmental Science
& Technology 41 (2007) 173.
[15] A. Braun, Journal of Environmental Monitoring 7(2005) 1059.
[16] J.-O. Müller, D. S. Su, R. E. Jentoft, U. Wild, R. Schlögl, Environmental
Science & Technology 40 (2006) 1231.
[17] J.-O. Müller, D. S. Su, R. E. Jentoft, J. Kröhnert, F. C. Jentoft,R. Schlögl,
Catalysis Today 102-103 (2005) 259.
[18] J.-O. Müller, D. S. Su, U. Wild, R. Schlögl, Physical Chemistry Chemical
Physics 9(2007) 4018.
[19] D. S. Su, R. E. Jentoft, J.-O. Müller, D. Rothe, E. Jacob, C. D. Simpson, Z.
Tomovic, K. Müllen, A. Messerer, U. Pöschl, R. Niessner, R. Schlögl,
Catalysis Today 90 (2004) 127.
[20]
D. S. Su, J.-O. Müller, R. E. Jentoft, D. Rothe, E. Jacob, R. Schlögl, Topics
in Catalysis 30-31 (2004) 241.
[21] G. A. Botton, G. Burnell, C. J. Humphreys, T. Yadav, J. C. Withers,
Journal of Physics and Chemistry of Solids 58 (1997) 1091.
[22] A. Braun, A. Kubatova, S. Wirick, S. B. Mun, Journal of Electron
Spectroscopy and Related Phenomena 170 (2009) 42.
[23] A. Sadezky, H. Muckenhuber, H. Grothe, R. Niessner, U. Pöschl, Carbon
43 (2005) 1731.
[24] U. Wild, N. Pfänder, R. Schlögl, Fresenius’ Journal of Analytical Chemistry
357 (1997) 420.
[25] H. Werner, D. Herein, J. Blöcker, B. Henschke, U. Tegtmeyer, T.
Schedel-Niedrig, M. Keil, A. M. Bradshaw, R. Schlögl, Chemical Physics
Letters 194 (1992) 62.
124
Bibliography
[26] Di Stasio, Stefano and Braun, Artur, Energy & Fuels 20 (2006) 187.
[27] B. S. Haynes, H. G. Wagner, Zeitschrift für Physikalische Chemie
133 (1982) 201.
[28] K.-H. Homann, Angewandte Chemie International Edition 37 (1998) 2434.
[29] B. A. A. L. van Setten, M. Makkee, J. A. Moulijn, Catalysis Reviews:
Science and Engineering 43 (2001) 489.
[30] H. Bockhorn (ed.), Soot Formation in Combustion. Springer Verlag, Berlin,
Heidelberg, 1994.
[31] I. M. Kennedy, Progress in Energy and Combustion Science 23 (1997) 95.
[32] J.-B. Donnet, R. C. Bansal, M.-J. Wang, Carbon black. Science and
Technoloy, 1993.
[33] J. Warnatz, U. Mass, R. W. Dibble, Combustion - Physical and Chemical
Fundamentals, Modeling and Simulation, Experiments, Pullutant Formation.
Springer Verlag, 2001.
[34]
M. S. Dresselhaus, G . Dresselhaus, K. Sugihara, I. L. Spain, H. A. Goldberg,
Graphite fibers and filaments, Springer series in Materials Science, Vol. 5.
Springer-Verlag, 1988.
[35] R. Schlögl, Handbook of Heterogeneous Catalysis, p. 357. Wiley-VCH, New
York, 1997.
[36] R. Janoschek, Chemie in unserer Zeit 22 (1988) 128.
[37] H. Kroto, Science 242 (1988) 1139.
[38] V. Elings, F. Wudl, J. Vac. Sci. Technol. A 6(1988) 412.
[39] R. Saito, G. Dresselhaus, M. S. Dresselhaus, Physical Properties of Carbon
Nanotubes. Imperial College Press, 2007.
[40] B. T. Kelly, Physics of Graphite. Applied Science, 1981.
[41] R. C. Bansal, F. J. Vastola, P. L. Walker Jr., Carbon 9(1971) 185.
125
Bibliography
[42]
R. Schlögl, Surface composition and structure of active carbons, in Handbook
of Porous Solids, ed. Prof. Dr. Ferdi Schüth, Prof. Dr. Kenneth S. W. Sing,
Prof. Dr. Jens Weitkamp, pp. 1863–1900. 2008.
[43] R. Schlögl, G. Loose, M. Wesemann, Solid State Ionics 43 (1990) 183.
[44]
B. Henschke, H. Schubert, J. Blöcker, F. Atamny, R. Schlögl, Thermochimica
Acta 234 (1994) 53.
[45] H. X. You, N. M. D. Brown, K. F. Al-Assadi, B. J. Meenan, Journal of
Materials Science Letters 12 (1993) 201.
[46] F. Atamny, J. Blöcker, A. Dübotzky, H. Kurt, O. Timpe, G. Loose, W.
Mahdi, R. Schlögl, Molecular Physics: An International Journal at the
Interface Between Chemistry and Physics 76 (1992) 851.
[47] J. A. Moulijn, F. Kapteijn, Carbon 33 (1995) 1155.
[48] J. Yang, G. Mestl, D. Herein, R. Schlögl, J. Find, Carbon 38 (2000) 715.
[49]
U. Kürpick, G. Meister, A. Goldmann, Applied Physics A: Materials Science
& Processing 55 (1992) 529.
[50] H.-P. Boehm, Angewandte Chemie 78 (1966) 617.
[51] M. Knauer, M. E. Schuster, D. S. Su, R. Schlögl, R. Niessner, N. P. Ivleva,
The Journal of Physical Chemistry A 113 (2009) 13871.
[52] Http://ec.europa.eu/environment/air/legis.htm.
[53] O. Scherzer, Journal of Applied Physics 20 (1949) 20.
[54]
D. B. Williams, C. B. Carter, Transmission Electron Microscopy - A textbook
for Materials Science. Springer, 1996.
[55] B. Fultz and J. M. Howe, Transmission Electron Microscopy and
Diffractometry of Materials. Springer, 2008.
[56]
R. F. Egerton, Electron Energy Loss Spectroscopy in the Electron Microscope.
Plenum Press, New York, 1986.
126
Bibliography
[57] V. J. Keast, A. J. Scott, R. Brydson, D. B. Williams, J. Bruley, Journal of
Microscopy 203 (2001) 135.
[58] H. Bethe, Annalen der Physik 397 (1930) 325.
[59] R. D. Leapman, P. L. Fejes, J. Silcox, Phys. Rev. B 28 (1983) 2361.
[60] C. Souche, B. Jouffrey, G. Hug, M. Nelhiebel, Micron 29 (1998) 419.
[61] N. K. Menon, J. Yuan, Ultramicroscopy 74 (1998) 83.
[62] H. Daniels, A. Brown, A. Scott, T. Nichells, B. Rand, R. Brydson,
Ultramicroscopy 96 (2003) 523.
[63] B. Jouffrey, P. Schattschneider, C. Hébert, Ultramicroscopy 102 (2004) 61.
[64] P. Schattschneider, Fundamentals of Inelastic Electron Scattering. Springer
Verlag, 1986.
[65] J. Yuan, Z. Wang, X. Fu, L. Xie, Y. Sun, S. Gao, J. Jiang, X. Hu, C. Xu,
Micron 39 (2008) 658.
[66] H. Raether, Springer Tracts Vol. 88 (1980).
[67] K. Zeppenfeld, Zeitschrift für Physik A Hadrons and Nuclei 243 (1971) 229.
[68] K. Zeppenfeld, Zeitschrift für Physik A Hadrons and Nuclei 211 (1968) 391.
[69] C. Hébert-Souche, P. H. Louf, P. Blaha, M. Nelhiebel, J. Luitz, P.
Schattschneider, K. Schwarz, B. Jouffrey, Ultramicroscopy 83 (2000) 9.
[70] K. Lie, R. Høier, R. Brydson, Phys. Rev. B 61 (2000) 1786.
[71] D. S. Su, C. Hébert, M. Willinger, R. Schlögl, Micron 34 (2003) 227.
[72]
O. Stéphan, P. M. Ajayan, C. Colliex, F. Cyrot-Lackmann, É. Sandré, Phys.
Rev. B 53 (1996) 13824.
[73] S. D. Berger, D. R. McKenzie, P. J. Martin, Philosophical Magazine Letters
57 (1988) 285.
127
Bibliography
[74] Y. X. Wang, H. Chen, R. W. Hoffman, J. C. Angus, Journal of Materials
Research 5(1990) 2378.
[75] Hébert, C. and Schattschneider, P. and Franco, H. and Jouffrey, B.,
Ultramicroscopy 106 (2006) 1139.
[76] J. Stöhr, NEXAFS Spectroscopy. Springer Verlag, 1996.
[77] A. P. Hitchcock, J. J. Dynes, G. Johansson, J. Wang, G. Botton, Micron
39 (2008) 311.
[78] G. D. Cody, R. E. Botto, H. Ade, S. Wirick, International Journal of Coal
Geology 32 (1996) 69.
[79] G. D. Cody, H. Ade, S. Wirick, G. D. Mitchell, A. Davis, Organic
Geochemistry 28 (1998) 441.
[80] C. K. Boyce, G. D. Cody, M. Feser, C. Jacobsen, A. H. Knoll, S. Wirick,
Geology 30 (2002) 1039.
[81] H. Ade, A. P. Smith, H. Zhang, G. R. Zhuang, J. Kirz, E. Rightor, A.
Hitchcock, Journal of Electron Spectroscopy and Related Phenomena
84 (1997) 53.
[82] M. P. Seah, W. A. Dench, Surface and Interface Analysis 1(1979) 2.
[83] J. Lehmann, D. Solomon, J. Brandes, H. Fleckenstein, C. Jacobson, J.
Thieme, Synchrotron-based near-edge x-ray spectroscopy of natural organic
matter in soils and sediments, in Biophysico-Chemical Processes Involving
Natural Nonliving Organic Matter in Environmental Systems, ed. Dr. Nicola
Senesi, Dr. Baoshan Xing, Pan Ming Huang, pp. 729–781. 2009.
[84] M. Hävecker, F. Senf, W. Eberhardt, R. Schlögl„ to be submitted to Nucl.
Instr. Meth. A.
[85] Y. Ma, C. T. Chen, G. Meigs, K. Randall, F. Sette, Phys. Rev. A
44 (1991) 1844.
[86] A. P. Hitchcock, C. E. Brion, Journal of Electron Spectroscopy and Related
Phenomena 18 (1980) 1.
128
Bibliography
[87] A. Einstein, Annalen der Physik 7(1905) 132.
[88] Http://www.helmholtz-berlin.de/.
[89] J. Moulder, W. Stickle, P. Sobol, Handbook of X Ray Photoelectron
Spectroscopy (P/N 624755). Perkin-Elmer, Physical Electronics Division,
1992.
[90] K. J. Børve, T. D. Thomas„ Journal of Electron Spectroscopy and Related
Phenomena 107 (2000) 155.
[91] S. Tougaard, P. Sigmund, Phys. Rev. B 25 (1982) 4452.
[92] E. Desimoni, G. I. Casella, A. Morone, A. M. Salvi, Surface and Interface
Analysis 15 (1990) 627.
[93]
D. T. Clark, H. R. Thomas, Journal of Polymer Science: Polymer Chemistry
Edition 14 (1976) 1701.
[94]
A. Proctor, P. M. A. Sherwood, Journal of Electron Spectroscopy and Related
Phenomena 27 (1982) 39.
[95] E. M. Vass, M. Hävecker, S. Zafeiratos, D. Teschner, A. Knop-Gericke, R.
Schlögl, Journal of Physics: Condensed Matter 20 (2008) 184016.
[96] J. J. Yeh, I. Lindau, Atomic Data and Nuclear Data Tables 32 (1985) 1.
[97] D. A. Shirley, Phys. Rev. B 5(1972) 4709.
[98] XPS casa software, http://www.casaxps.com/.
[99] G. K. Wertheim, P. H. Citrin, Photoemission in Solids I Vol I (1979) 197.
[100] D. Briggs, M.P. Seah, Practical Surface Analysis by Auger and X-ray
Photoelectron Spectroscop. 1990.
[101]
G. M. Jenkins, K. Kawamura, L. L. Ban, Proceedings of the Royal Society of
London. Series A, Mathematical and Physical Sciences 327 (1972) 501.
[102] J. M. Carlsson, M. Scheffler, Phys. Rev. Lett. 96 (2006) 046806.
129
Bibliography
[103] J.-O. Müller, Investigations on Environmental Carbons. Ph.D. Thesis, TU
BERLIN, 2005.
[104] H.-S. Shim, R. H. Hurt, N. Y. C. Yang, Carbon 38 (2000) 29.
[105] A. Goel, P. Hebgen, J. B. Vander Sande, J. B. Howard, Carbon
40 (2002) 177.
[106]
D. R. McKenzie, R. C. McPhedran, N. Savvides, L. C. Botten, Philosophical
Magazine Part B 48 (1983) 341.
[107]
C. Weissmantel, K. Bewilogua, C. Schürer, K. Breuer, H. Zscheile, Thin Solid
Films 61 (1979) L1.
[108]
G. M. Jenkins, K. Kawamura, Polymeric carbons–carbon fibre, glass and char.
Cambridge University Press, 1976.
[109] L. Joly-Pottuz, N. Matsumoto, H. Kinoshita, B. Vacher, M. Belin, G.
Montagnac, J. M. Martin, N. Ohmae, Tribology International 41 (2008) 69.
[110] T. Cabioc’h, J. C. Girard, M. Jaouen, M. F. Denanot, G. Hug, Europhysics
Letters 38 (1997) 471.
[111]
C. Yannouleas, E. N. Bogachek, U. Landman, Phys. Rev. B
53
(1996) 10225.
[112] J. Verbeeck, http://www.eelsmodel.ua.ac.be/.
[113] A. P. Hitchcock, P. Fischer, A. Gedanken, M. B. Robin, The Journal of
Physical Chemistry 91 (1987) 531.
[114] A. P. Hitchcock, H. D. H. Stöver, L. M. Croll, R. F. Childs, Aust. J. Chem.
58 (2005) 423.
[115] A. P. Hitchcock, D. C. Newbury, I. Ishii, J. Stohr, J. A. Horsley, R. D.
Redwing, A. L. Johnson, F. Sette, The Journal of Chemical Physics
85 (1986) 4849.
[116] A. P. Hitchcock, I. Ishii, Journal of Electron Spectroscopy and Related
Phenomena 42 (1987) 11.
130
Bibliography
[117] I. Ishii, A. P. Hitchcock, The Journal of Chemical Physics 87 (1987) 830.
[118] I. Ishii, A. P. Hitchcook, Journal of Electron Spectroscopy and Related
Phenomena 46 (1988) 55.
[119] S. G. Urquhart, H. Ade, The Journal of Physical Chemistry B
106 (2002) 8531.
[120] S. C. B. Myneni, Reviews in Mineralogy and Geochemistry 49 (2002) 485.
[121] Y. Zubavichus, O. Fuchs, L. Weinhardt, C. Heske, E. Umbach, J. D.
Denlinger, M. Grunze, Radiation Research 161 (2004) 346.
[122] D. C. Edwards, S. C. B. Myneni, The Journal of Physical Chemistry A
110 (2006) 11809.
[123]
G. D. Cody, H. Ade, M. O’D Alexander, T. Araki, Tohru A. Butterworth, H.
Fleckenstein, G. Flynn, M. K. Gilles, C. Jacobsen, A. L. D. Kilcoyne, K.
Messenger, S. A. Sandford, T. Tyliszczak, A. J. Westphal, S. Wirick, H.
Yabuta, Meteoritics & Planetary Science 43 (2008) 353.
[124] C. D. Cappa, J. D. Smith, K. R. Wilson, B. M. Messer, M. K. Gilles, R. C.
Cohen, R. J. Saykally, The Journal of Physical Chemistry B
109
(2005) 7046.
[125] J. W. Chiou, C. L. Yueh, J. C. Jan, H. M. Tsai, W. F. Pong, I.-H. Hong, R.
Klauser, M.-H. Tsai, Y. K. Chang, Y. Y. Chen, C. T. Wu, K. H. Chen, S. L.
Wei, C. Y. Wen, L. C. Chen, T. J. Chuang, Applied Physics Letters
81 (2002) 4189.
[126] S. Waidmann, M. Knupfer, J. Fink, B. Kleinsorge, J. Robertson, Journal of
Applied Physics 89 (2001) 3783.
[127]
J. A. Brandes, G. D. Cody, D. Rumble, P. Haberstroh, S. Wirick, Y. Gelinas,
Carbon 46 (2008) 1424.
[128] F. Atamny, J. Bloecker, B. Henschke, R. Schloegl, T. Schedel-Niedrig, M.
Keil, A. M. Bradshaw, The Journal of Physical Chemistry 96 (1992) 4522.
[129] S. G. Urquhart, A. P. Hitchcock, R. D. Priester, E. G. Rightor, Journal of
Polymer Science Part B: Polymer Physics 33 (1995) 1603.
131
Bibliography
[130]
C. Lenardi, P. Piseri, V. Briois, C. E. Bottani, A. Li Bassi, P. Milani, Journal
of Applied Physics 85 (1999) 7159.
[131]
P. R. Bressler, M. Lubbe, D. R. T. Zahn, W. Braun, J. Vac. Sci. Technol. A
15 (1997) 2085.
[132] J. Kikuma, K. Yoneyama, M. Nomura, T. Konishi, T. Hashimoto, R.
Mitsumoto, Y. Ohuchi, K. Seki, Journal of Electron Spectroscopy and Related
Phenomena 88-91 (1998) 919.
[133] D. A. Fischer, R. M. Wentzcovitch, R. G. Carr, A. Continenza, A. J.
Freeman, Phys. Rev. B 44 (1991) 1427.
[134]
S. Entani, S. Ikeda, M. Kiguchi, K. Saiki, G. Yoshikawa, I. Nakai, H. Kondoh,
T. Ohta, Applied Physics Letters 88 (2006) 153126.
[135] P. E. Batson, Phys. Rev. B 48 (1993) 2608.
[136] P. Chen, X. Wu, X. Sun, J. Lin, W. Ji, K. L. Tan, Phys. Rev. Lett.
82 (1999) 2548.
[137] L. J. Terminello, D. K. Shuh, F. J. Himpsel, D. A. Lapiano-Smith, J. Stöhr,
D. S. Bethune, G. Meijer, Chemical Physics Letters 182 (1991) 491.
[138] J. L. Solomon, R.L. Madix, J. Stöhr, Surface Science 255 (1991) 12.
[139] A. Proctor, P. M. A. Sherwood, Analytical Chemistry 54 (1982) 13.
[140] O. Plashkevych, V. Carravetta, O. Vahtras, H. Ågren, Chemical Physics
232 (1998) 49.
[141] D. T. Clark, J. Peeling, L. Colling, Biochimica et Biophysica Acta (BBA) -
Protein Structure 453 (1976) 533.
[142] T. Schedel-Niedrig, W. Weiss, R. Schlögl, Phys. Rev. B 52 (1995) 17449.
[143] V. C. Felicíssimo, A. Cesar, Y. Luo, F. Gel’mukhanov, H. Ågren, The
Journal of Physical Chemistry A 109 (2005) 7385.
[144] A. M. Salvi, J. E. Castle, Journal of Electron Spectroscopy and Related
Phenomena 94 (1998) 73.
132
Bibliography
[145] P. M. Th. M. van Attekum, G. K. Wertheim, G. K., Phys. Rev. Lett.
43 (1979) 1896.
[146] Speranza, G. and Minati, L., Surface Science 600 (2006) 4438.
[147] D. Q. Yang, E. Sacher, Surface Science 504 (2002) 125.
[148] F. R. McFeely, S. P. Kowalczyk, L. Ley, R. G. Cavell, R. A. Pollak, D. A.
Shirley, Phys. Rev. B 9(1974) 5268.
[149] E. Desimoni, G. I. Casella, A. M. Salvi, Carbon 30 (1992) 521.
[150] R. Schlögl, H.P. Boehm, Carbon 21 (1983) 345.
[151] Y. Baer, Journal of Electron Spectroscopy and Related Phenomena
24 (1981) 95.
[152]
K. C. Prince, I. Ulrych, M. Peloi, B. Ressel, V. Cháb, C. Crotti, C. Comicioli,
Phys. Rev. B 62 (2000) 6866.
[153]
P. M. A. Sherwood, Journal of Electron Spectroscopy and Related Phenomena
81 (1996) 319.
[154] N. Keller, N. I. Maksimova, V. V. Roddatis, M. Schur, G. Mestl, Y. V.
Butenko, V. L. Kuznetsov, R. Schlögl, Angewandte Chemie International
Edition 41 (2002) 1885.
[155]
T. I. T. Okpalugo, P. Papakonstantinou, H. Murphy, J. McLaughlin, N. M. D.
Brown, Carbon 43 (2005) 153.
[156] J. A. Maciá-Agulló, D. Cazorla-Amorós, A. Linares-Solano, U. Wild, D. S.
Su, R. Schlögl, Catalysis Today 102-103 (2005) 248.
[157] M. Walczyk, A. Swiatkowski, M. Pakula, S. Biniak, Journal of Applied
Electrochemistry 35 (2005) 123.
[158]
J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freitas, J. J. M. Órfão, Carbon
37 (1999) 1379.
[159] B. Marchon, J. Carrazza, H. Heinemann, G. A. Somorjai, Carbon
26 (1988) 507.
133
Bibliography
[160] E. Desimoni, G. I. Casella, T. R. I. Cataldi, A. M. Salvi, T. Rotunno, E. D.
Croce, Surface and Interface Analysis 18 (1992) 623.
[161] C. Kozlowski, P. M. A. Sherwood, J. Chem. Soc., Faraday Trans. 1
81 (1985) 2745.
[162] R. Arrigo, Nitrogen Functionalization of CNFs and Application in
Heterogeneous Catalysis. Ph.D. Thesis, TU BERLIN, 2009.
[163]
R. Arrigo, M. Hävecker, S. Wrabetz, R. Blume, M. Lerch, J. McGregor, E. P.
J. Parrott, J. A. Zeitler, L. F. Gladden, A. Knop-Gericke, R. Schlögl, D. S.
Su, Journal of the American Chemical Society 132 (2010) 9616.
[164] H. Bluhm, M. Hävecker, A. Knop-Gericke, E. Kleimenov, R. Schlögl, D.
Teschner, V. I. Bukhtiyarov, D. F. Ogletree, M. Salmeron, The Journal of
Physical Chemistry B 108 (2004) 14340.
[165] D. R. Penn, Phys. Rev. B 13 (1976) 5248.
[166] T.N. Rhodin, J.W. Gadzuk, The Nature of the Surface Chemical Bond.
North-Holland, 1979.
[167] Somorjai, Gabor A., Chemistry in two dimensions : surfaces. Cornell
University Press, Ithaca :, 1981.
[168] A. Zangwill, Physics at Surfaces. Cambridge University Press, 1988.
[169] Dong, G. L. and Hüttinger, K. J., Carbon 40 (2002) 2515.
134
