XRD and XPS characterisation of transition metal silicide thin films

Article abstract of DOI:10.1016/j.susc.2011.10.015

Binary transition metal silicides based on the systems Ti-Si, Fe-Si, Ni-Si and Cr-Si were fabricated on Si wafers by means of ion-beam co-sputter deposition and subsequent annealing. The crystalline structures of the phases formed were identified from the characteristic patterns acquired by means of X-ray diffraction (XRD) measurements. The phase formation sequences were described by means of the Pretorius' effective heat of formation (EHF) model. For the Ti-Si, Fe-Si and Ni-Si systems, single phase thin films of TiSi ₂, β-FeSi₂ and NiSi₂ were generated as the model predicts, while a mixture of CrSi + CrSi₂ phases was obtained for the Cr-Si system. The surface chemical condition of individual specimens was analysed by using X-ray photoelectron spectroscopy (XPS). The chemical shifts of transition metal 2p_3/2 peaks from their metallic to silicide states were depicted by means of the Auger parameters and the Wagner plots. The positive chemical shift of 2.0 eV for Ni 2p_3/2 peak of NiSi ₂ is mainly governed by the initial-state effects. For the other silicide specimens, the initial-state and final-state effects may oppose one another with similar impact. Consequently, smaller binding energy shifts of both negative and positive character are noted; a positive binding energy shift of 0.3 eV for the Fe 2p_3/2 level was shown for β-FeSi₂ and negative binding energy shifts of 0.1 and 0.3 eV were determined for CrSi + CrSi₂ and TiSi₂, respectively.

Full text of DOI:10.1016/j.susc.2011.10.015

Surface Science 606 (2012) 329–336
Contents lists available at SciVerse ScienceDirect
Surface Science
journal homepage: www.elsevier.com/locate/susc
XRD and XPS characterisation of transition metal silicide thin films
⁎
P.L. Tam, Y. Cao, L. Nyborg
Department of Materials and Manufacturing Technology, Chalmers University of Technology, SE-412 96, Gothenburg, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Binary transition metal silicides based on the systems Ti–Si, Fe–Si, Ni–Si and Cr–Si were fabricated on Si wa-
fers by means of ion-beam co-sputter deposition and subsequent annealing. The crystalline structures of the
phases formed were identified from the characteristic patterns acquired by means of X-ray diffraction (XRD)
measurements. The phase formation sequences were described by means of the Pretorius' effective heat of
formation (EHF) model. For the Ti–Si, Fe–Si and Ni–Si systems, single phase thin films of TiSi2, β-FeSi2 and
NiSi2 were generated as the model predicts, while a mixture of CrSi+CrSi2 phases was obtained for the
Cr–Si system. The surface chemical condition of individual specimens was analysed by using X-ray photoelec-
tron spectroscopy (XPS). The chemical shifts of transition metal 2p3/2 peaks from their metallic to silicide
states were depicted by means of the Auger parameters and the Wagner plots. The positive chemical shift
of 2.0 eV for Ni 2p3/2 peak of NiSi2 is mainly governed by the initial-state effects. For the other silicide spec-
imens, the initial-state and final-state effects may oppose one another with similar impact. Consequently,
smaller binding energy shifts of both negative and positive character are noted; a positive binding energy
shift of 0.3 eV for the Fe 2p3/2 level was shown for β-FeSi2 and negative binding energy shifts of 0.1 and
0.3 eV were determined for CrSi+CrSi2 and TiSi2, respectively.
Received 9 August 2011
Accepted 18 October 2011
Available online 25 October 2011
Keywords:
Transition metal silicide
Thin films
X-ray photoelectron spectroscopy (XPS)
X-ray diffraction (XRD)
Pretorius' effective heat of formation (EHF)
model
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
chemical-state information, surface chemical composition and electron
screening effect can be determined by means of XPS and AES measure-
Transition metal silicides have long been promising materials in mi-
croelectronic devices owing to their low electrical resistivity and excel-
lent chemical stability [1]. Furthermore, in the manufacturing industry,
metal silicides have been considered for high temperature structural ap-
plications because of their high temperature strength and oxidation re-
sistance [2,3]. From the high oxidation resistance point of view, it is
envisaged that metal silicides can also be the candidates for protective
coatings on engineering materials, e.g. stainless steels, to combat against
corrosion attack in aggressive environment [4].
Different binary systems for such approach include various combi-
nations of a transition metal and silicon to form the intended silicide
by means of selected thin film and coating technologies. One example
of such is sputter deposition. Here, either co-melted alloy targets or
use of elemental targets for co-deposition can be employed. To depict
the surface chemical characteristics of the thin films, surface analytical
techniques by means of X-ray emission spectroscopy (XES), X-ray pho-
toelectron spectroscopy (XPS) and/or Auger electron spectroscopy
(AES) are necessary means [5–52]. For example, Weijs et al. determined
the chemical bonding of silicon in various transition metal silicide thin
films including TiSi2, CrSi2, CoSi2, NiSi2, MoSi2 and NbSi2 through XES
measurements in combination with partial density-of-state (DOS) cal-
culations [29]. Meanwhile, the electronic structure (i.e. total DOS), the
ments [53]. At present, many studies have already been devoted to the
fundamental study of other 3d and 4d transition metal silicides [5–52].
In this context, it is of prime importance to understand how thin film
structure and phase formation sequence correlate to one another. For
this reason, surface characterisation on a series of transition metal sili-
cide thin films deposited on the Si wafer was carried out in this study
by means of XPS combined with XRD analysis for phase identification.
Among the variety of the silicide compositions, disilicides are the most
commonly used compounds and these are thus targeted in the present
study [1–3]. Crucial for the assessment of the silicide formation is the
possibility of unambiguous and consistent interpretation of XPS data.
Although large amount of literature data related to the silicide charac-
terisation can be found, the reported data, especially the core-level
binding energy (BE) position from the XPS measurements, vary
among different publications [5–15] (see Table 1). Besides the advance-
ment of the instrumental technology, it can be supposed that the incon-
sistence of data could be due to the surface cleanliness [12–14],
different conductivity of the substrate materials [5–14], calibration
practices [5–14] and the implementation of ion cleaning processes
just before the characterising work [7]. It is suggested that air exposure
after individual specimen preparation before thin film characterisation
should be the most significant factor that affects the characterised sur-
face. With reference to the experimental results from XRD and XPS
[5–17] on silicides, the approach in this study is to take advantage of
thin film preparation and subsequent surface analysis in an ultra-high
⁎
Corresponding author. Tel.: +46 31 772 12 57; fax: +46 31 772 13 13.
E-mail address: lars.nyborg@chalmers.se (L. Nyborg).
0039-6028/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2011.10.015

330
P.L. Tam et al. / Surface Science 606 (2012) 329–336
Table 1
Some of the reported XPS core-level binding energy positions of the transition metals and transition metal silicides.
Types of materials Film-substrate system/Bulk
materials
Fabrication techniques
Annealing
temperature in XPS
(°C)
X-ray source BE of transition metal Ref.
2p3/2 peak in XPS (eV)
Ti
Bulk material
MgKα
453.8
[54]
[5]
Ti
Ti/Ti–Si thin films on Si (111)
Ion beam deposition
AlKα
454.1
454.3
β-TiSi2 or
α-TiSi2
500 or 700
Ti
Ti-Si
Ti/SiOx thin films on p-Si (111)
Magnetron cathode deposition
MgKα
b300
454.5
454.8
[6]
Fe
Bulk material
MgKα
706.8
[54]
[7]
Fe
Fe–Si
Fe/Si multilayer on GaAs
Magnetron sputtering
AlKα
707.3
707.5
500
Fe
FeSi2
Bulk material
AlKα
707.0
707.4
[29]
Co-melting in the electron beam furnace
800
Ni
Bulk material
MgKα
852.3
[54]
[10]
Ni
NiSi2
Ni/Ni–Si on Si
Ion beam sputter deposition (IBSD)
AlKα
852.7
854.6
800
NiSi2
Cr
Ni/Co bilayer film on Si
Bulk material
Radio frequency (RF) magnetron sputtering
600
AlKα
MgKα
AlKα
854.5
574.1
[9]
[54]
[12]
Cr
Cr–Si
Cr/Cr–Si on amorphous Si
Plasma enhanced chemical vapour deposition (PECVD)
573.9
575.0
Cr–Si
Cr–Si on alumina
Magnetron sputtering
573.8
574.3
[13]
[14]
CrSi2
CrSi2 on Si
Molecular beam epitaxy (MBE)
450
MgKα
Table 2
Some of the reported XRD phases of the transition metal silicides [16,17].
Types of thin films
Film-substrate system/bulk
materials
Fabrication techniques
Annealing temperatures (°C),
time
X-ray source
in XRD
Phase
characterised
Ref.
[17]
Fe–Si
Fe–Si
Fe/Fe–Si on Si (111)
(boron-doped, n-type)
Evaporating deposition and
rapid thermal annealing (RTA)
600, 20 s
750, 10 s
CoKα
α-Fe, FeSi, Si
Fe–Si, β-FeSi, Si
Cr
Cr/Cr–Si on Si (100)
Electron beam evaporation deposition
CuKα
Cr, Si
[16]
Cr–Si
Cr–Si
Cr–Si
Cr–Si
(n-type)
300, 1 h
400, 1 h
500, 1 h
600, 1 h
Cr, Si
Cr, Cr3Si, Si
CrSi2, Si
CrSi2
vacuum (UHV) environment without any air exposure throughout the
processing and handling. In this way, the XPS binding energy positions
for the silicides formed could then be reassured. To aid the interpreta-
tion of the XPS core-level peak data, use of the Auger parameters and
the Wagner plots are also presented. The XRD analysis, on the other
hand, acquires information at larger depth than XPS, and is less crucial
to the change of surface condition. In certain extent, surface cleanliness,
then, is not the primary concern in XRD, but still, air exposure period be-
tween sample preparation and characterisation should be kept as short
as possible to minimise the surface interference effect such as the sur-
face oxide formation. Some characterised patterns of silicide thin films
are tabulated in Table 2. As no correlation study between XRD and
XPS characterisation about the silicide thin films has been found so
far, therefore, this is one important objective in the present study to
build such relationship.
level better than 10−6 Pa. This system includes a PHI 5500 instrument
for XPS measurements, a vacuum furnace for thermal annealing and a
sputter deposition chamber for thin film fabrication [55]. The binary-
component thin films were deposited on 10×10 mm2 Si wafers after
argon ion sputter-cleaning. The choice of Si substrate will give benefit
to the disilicide formation and will be discussed in detail in the
Discussion section. The deposition was done by sputtering the high pu-
rity targets composed of Si and Ti, Ni, Fe or Cr, respectively, using two
Table 3
The processing parameters in thin film fabrications.
Type of
silicides
Operating voltage of the
Kaufman-type ion gun (V)
As-deposited
composition
Annealing
temperature (°C)
and duration
On transition
metal target
On silicon target
2. Experiments and analyses
Ti–Si
Fe–Si
Ni–Si
Cr–Si
300
200
350
200
500
500
500
500
Ti0.40Si0.60
Fe0.40Si0.60
Ni0.33Si0.67
Cr0.20Si0.80
700, 1 h
600, 1 h
850, 1 h
650, 1 h
Preparation and characterisation of transition metal silicide thin
films were carried out in an interconnected UHV system at a vacuum

P.L. Tam et al. / Surface Science 606 (2012) 329–336
331
individually monitored Kaufman-type argon ion beams, generating
given composition ratios of the transition metal and silicon for the dif-
ferent metal silicide films fabricated. Table 3 summarises the sputtering
parameters used. After deposition, the composition of the specimens
was first characterised with XPS and then transferred to the vacuum
furnace for annealing to achieve long-range ordered structures. In all
the annealing processes, no excessive outgassing had occurred and
the chamber pressure was maintained at 10−6 Pa except for the first
minute, at which it was around 10−4 Pa. The annealing temperatures
employed are presented in Table 3. The annealed specimens were
then moved to the surface analytical instrument again for composition
and chemical analysis. The XPS instrument equipped with monochro-
matic AlKα source (hυ=1486.6 eV) was calibrated according to the
ISO 15472:2001 standard in which the binding energy (BE) is aligned
with the Au 4f7/2, Ag 3d5/2 and Cu 2p3/2 at 83.96 eV, 368.21 eV and
932.62 eV, respectively [56,57]. The absolute energy resolution of the
BE spectra is 0.8 eV, which is determined by measuring the full-width
at half maximum (FWHM) of the Ag 3d5/2 peak from a sputter-cleaned
silver sample. The smallest core-level shift that could be identified is
smaller than 0.1 eV.
It is noted that the assessment of binding energy shift of the core-
level spectra of an element from one state to another is one of the pri-
mary research topics in this study. Therefore, a fundamental under-
standing about the BE level of a characteristic peak is necessary. In
principle, the BE level in the XPS measurement is an integrated result
of the initial-state and final-state effects of the photoelectron emis-
sion process [18–24]. The initial-state effects that take place prior
to photoemission may include factors [25,26] such as ionicity (i.e.
charge transfer), valence orbital transformation and valence electron
configurations. Mathematically, these factors can be summarised in
Eq. (1):
The core level shift (ΔEb), depends on both the initial-state and final-
state effects of both chemical states, can be deduced from Eqs. (2) and
(3), as:
ΔEb
¼
1=2ðΔβ−ΔαÞ
ð6Þ
Through these equations, it implies that both physical effects can
be roughly estimated via an experimental approach.
The phases corresponding to surface chemical conditions deter-
mined by the XPS core-level spectra were confirmed by means of the
X-ray diffraction technique with a grazing incidence setup, i.e. GIXRD.
High resolution diffraction patterns of various silicide specimens were
obtained from 40 to 80° at a low scanning rate (i.e. 20 s per 0.02° in
step) with an incident angle set at 2.0° in a Bruker D8 Advance X-ray dif-
fraction instrument using CrKα radiation (λ=0.22897 nm). The dif-
fraction peaks were matched with the International Centre of
Diffraction Data (ICDD) powder diffraction files (PDF-2) [58] and the
possible phases formed were identified. These characterised phases
were then studied with the annealing processes and compared with
those phases predicted using a thermodynamic Pretorius' EHF model.
This model determines the phases formed by comparing the effective
heat of formation (ΔH′) of all possible silicide phases in every silicide
system by using the starting composition as the effective concentration
and applying the following equation [59,60]:
∘
ΔH′ ¼ ΔH
EFFECTIVE CONCENTRATION OF LIMITING ELEMENT
COMPOUND CONCENTRATION OF LIMITING ELEMENT
ꢀ
:
ð7Þ
According to the model, the first formed phase should fulfil the
criteria to be a congruent phase and it should have the lowest effec-
tive heat of formation (ΔH′) among all possible phases in one system.
Generally, non-congruent silicides have difficulty to nucleate and are
usually disregarded as the first formed phase, unless the temperature
difference between the liquidus curve and the peritectic/peritectoid
point is very small in the phase diagram. However, surplus of the
needed element that exists in the system can facilitate the non-
congruent phase formation in the subsequent phase development
process.
Eb;PE ¼ E0b;PE þ k⋅q þ VM
ð1Þ
where Eb,PE is the binding energy of a core-level photoelectron in an
atom, Eb0,PE is an energy reference and a physical constant, VM is the
Madelung potential and q is the valence charge of an atom and k is
the change in core potential resulting from a removal of a photoelec-
tron [20,21].
The final-state effects follow the photoelectron emission, i.e. the
system adjusts its orbital to a lower binding energy level, producing
the relaxation of the core holes that are induced during photoelectron
emission process. This relaxation energy (R) consists of an intra-
atomic contribution (Ria) – the contraction energy of the outer elec-
tron change toward the inner core hole – and an extra-atomic contri-
bution (Rea) – the relaxation energy associated with the rest of the
system – but it can assume that Rea alone is the dominant factor for
the final-state effects [20,21].
Both initial-state and final-state effects of each chemical state can
be graphically presented by means of the Wagner plot (or chemical-
state plot) [10,21–24] in terms of initial-state Auger parameter (β)
and final-state Auger parameter (α), which are defined and inter-
preted, respectively, in Eqs. (2) and (3) [21,22]:
3. Results
Fig. 1 illustrates the diffraction patterns of the metal silicides after
the individual thermal annealing processing. The patterns indicate
that all the metal silicide thin films show crystalline structures. As
can be seen in the figure, the thermal annealing processes give single
phases of disilicides (i.e. TiSi2, β-FeSi2 and NiSi2) for the Ti–Si, Fe–Si
and Ni–Si systems. Meanwhile, two phases of monosilicide and disili-
cide are developed for the Cr–Si system. For this binary-phase speci-
men, the intensity ratios of the characteristic peaks suggest that CrSi
is the dominant phase. The lattice parameters and the structural in-
formation for the individual silicide phases, in accordance with the
corresponding PDF data [58] are presented in Table 4.
Fig. 2 shows the core-level binding energy (BE) spectra of the
transition metal 2p3/2 peak from individual silicide specimens as
well as for their corresponding pure metal films. In the figure, the dif-
ferences of the BE positions for these two chemical states, or the
chemical shifts, are indicated for the different combinations of transi-
tion metal and Si studied. As can be seen in the figure, prominent BE
shifts are observed for the 2p state after all silicidation processes.
Meanwhile, the direction and magnitude of this shift is materials de-
pendent. For the earlier transition metals Ti and Cr there are negative
BE shifts of 0.3 eV and 0.1 eV, respectively. For the later transition
metals Fe and Ni, the shifts are positive with magnitude of 0.3 eV
and 2.0 eV, respectively. By contrast, the corresponding Si 2p peaks
do not show significant shift before and after these processes because
Ek;Auger ¼ β−3 E
ð2Þ
ð3Þ
⋅
b;PE
Ek;Auger ¼ α−Eb;PE
where Ek,Auger is the kinetic energy of the Auger peak.
In fact, with respect to a specific reference, the changes of initial-state
(Δβ) and final-state (Δα) effects are:
Δβ ¼ 2 ΔðV þ k qÞ
ð4Þ
ð5Þ
⋅
⋅
M
Δα ¼ 2 ΔR
⋅

332
P.L. Tam et al. / Surface Science 606 (2012) 329–336
a
b
-FeSi2 (PDF: 71-0642)
TiSi2 (PDF: 85-0879)
(sub.)
(004)
(022)
(313)
(220)
(422)
(332)
(sub.)
(311)
(402)
(521)
(331)
(312)
(004)
40
45
50
55
60
65
70
75
80
40
45
50
55
60
65
70
75
80
2
(o)
2
(o)
c
d
(220)
CrSi (PDF: 72-1326)
*CrSi2 (PDF: 35-0781)
NiSi2 (PDF: 43-0989)
(211)
(210)
(111)
(111)
*(003)
*(110)
*(211)
(sub.)
*(111)
(sub.)
40
45
50
55
60
65
70
75
80
40
45
50
55
60
65
70
75
80
2
(o)
2
(o)
Fig. 1. Phase identification of the transition metal silicide specimens: (a) Ti–Si, (b) Fe–Si, (c) Ni–Si and (d) Cr–Si.
the initial- and final-state effects compensate each other and thus
cannot be used for depicting any chemical information regarding
transition metal silicides. Similar results have already been reported
for different Ni–Si compounds [4,10]. Therefore, only BE positions of
the transition metal 2p signal (here specifically the 2p3/2 level) are
taken into account for the silicide determination. It should be noted
that the BE shifts shown in Fig. 2 for the Ti, Fe and Ni signals are the
representatives for the single phase disilicides of TiSi2, β-FeSi2 and
NiSi2 compared to that of the metallic state, while the BE shifts for
Cr signal in the present study show the integrated representation of
the mixed phases of CrSi+CrSi2, compared to the metallic state of Cr.
The observed BE positions of the metallic and silicide states can
then be defined in relation to the silicide phase presented in the dif-
fraction patterns and the reviewed literature [5–15,29,54]. Therefore,
the BE and FWHM (in parentheses) of the transition metal 2p3/2 level
of metallic state of Ti, Fe, Ni and Cr are at 454.2 eV (1.1 eV), 706.7 eV
(2.6 eV), 852.6 eV (1.8 eV) and 574.3 eV (1.4 eV), respectively. The
corresponding BE positions of TiSi2, FeSi2, NiSi2 and mixture of CrSi
and CrSi2 are at 453.9 eV (0.9 eV), 707.0 eV (1.7 eV), 854.6 eV
(1.6 eV) and 574.2 eV (1.1 eV). The changes of peak features noted
from this result also include the peak width narrowing and asymme-
try reduction owing to silicidation.
An important means of further understanding the observed core-
level BE shifts is the study of the characteristic Auger lines for the
same specimens. This forms the basis for the determination of the
Auger parameters [10,18–26,30]. The Auger peak selected for this
study is that of the L3M23M23 transition. The kinetic energy (KE) po-
sitions of this peak recorded from the silicide specimens and the
pure metals are summarised in Table 5. Further implications of the
comparison of core-level and Auger peak positions for the different
transition metal silicide systems will be treated in the Discussion sec-
tion, when considering the possible contribution of initial-state and
final-state effects behind the observed core-level positions through
Wagner plots. Considering the results in Table 5, it should now
Table 4
Crystal structure parameters of the transition metals and transition metal silicides [27,28,58,61].
Phase
Structure
Lattice parameter
(nm)
No. of atoms
per unit cell
Density
(g/cm3)
Distance between two
metallic atoms in the lattice
(nm)
Si
Ti
TiSi2
Fe
FeSi2
Ni
NiSi2
Cr
CrSi
CrSi2
Fd3m (A4)
P63/mmc (A3)
Fddd (C54)
Im3m (A2)
Cmca (C49)
Fm3m (A1)
Fm3m (C1)
Im3m (A2)
PC213 (B20)
P6222 (C40)
a=0.5431
8
2
8
2
16
4
4
2
4
3
2.33
4.51
4.13
7.88
4.95
8.91
4.80
7.20
5.39
4.98
0.235
0.290, 0.295
0.321
a=0.2951, c=0.4683
a=0.8236, b=0.4773, c=0.8523
a=0.2866
a=0.9863, b=0.7791, c=0.7833
a=0.3524
a=0.5416
a=0.2884
a=0.4484
a=0.4428, c=0.6369
0.249
0.249
0.383
0.250
0.283
0.306

P.L. Tam et al. / Surface Science 606 (2012) 329–336
333
a
b
Fe
Ti
FeSi2
TiSi2
2p3/2
2p3/2
460
458
456
454
452
450
712
710
708
706
704
702
Binding energy (eV)
Binding energy (eV)
c
d
2p3/2
Cr
2p3/2
CrSi + CrSi2
Ni
NiSi2
860
858
856
854
852
850
580
578
576
574
572
570
Binding energy (eV)
Binding energy (eV)
Fig. 2. Core level 2p3/2 spectra of transition metals from different silicide systems before (_____) and after (
) silicidation.
merely be noted that the energy shifts of the Auger lines are, except in
the case of Ti–Si, larger than the core-level shifts.
phase diagram. However, in the present case the starting point is a mix-
ture generated by means of co-sputtering and not the two components
in contact at an interface. In this respect, it is therefore assumed that the
as-deposited concentrations indicated by the XPS analyses could serve
as a good measure for the effective concentration of the limiting ele-
ment for the silicide formation in the present case. The EHFs (ΔH′) of
each of the possible phases were then calculated and the results are pre-
sented in Table 6. For Ti–Si system, it is then predicted that TiSi2 (i.e.
Ti0.33Si0.67) should be the first congruent phase formed from the com-
position of Ti0.40Si0.60 owing to its lowest ΔH′. This agrees with Pretori-
us' own prediction [59,60]. However, there is excessive Ti remaining
after the TiSi2 formation. This means that this Ti can further react with
the Si substrate and thus complete consumption of Ti should occur to
TiSi2. For the Fe–Si system, FeSi (i.e. Fe0.50Si0.50) is the congruent
phase with the lowest ΔH′ congruent phase and thus considered to be
the first formed phase of the as-deposited state of Fe0.40Si0.60 according
to Table 4. This is also in agreement with Pretorius' original concept
4. Discussion
The temperature chosen for the thermal annealing processing of the
co-sputtered binary-component silicide films was referenced to the lit-
erature [5–17,50–52,54]. Single silicide phases in the form of TiSi2, β-
FeSi2 and NiSi2 could be achieved for the Ti–Si, Fe–Si and Ni–Si systems,
while binary silicide phases in the form of CrSi+CrSi2 resulted in the
case of the Cr–Si system. Besides experimental XRD characterisation, a
theoretical approach to predict the observed phase formation is
employed by applying Pretorius' effective heat of formation (EHF)
model [59,60] as presented in the Experiments and analyses section. It
should be noted that the approach represents a deviation from Pretori-
us' original concept in which the initial concentration of the limiting el-
ement is taken as the concentration of the lowest eutectic of the binary
Table 5
Binding energy positions (eV) of transition metals in the XPS measurements for different characterised specimens.
Type of films
Eb (2p3/2
)
ΔEb (2p3/2
)
KE of Auger
L3M23M23
transition
Δ Auger
Auger parameters
ΔR
α (Final-state effects)
Δα
β (Initial-state effects)
Δβ
Ti
TiSi2
Fe
FeSi2
Ni
NiSi2
Cr
454.2
453.9
706.7
707.0
852.6
854.6
574.3
574.2
388.6
388.6
598.5
596.8
708.0
705.8
489.2
488.3
842.8
842.5
1751.2
1750.3
2718.6
2717.8
3265.8
3269.6
2212.1
2210.9
−0.3
+0.3
+2.0
−0.1
0
−0.3
−1.4
−0.2
−1.0
−0.9
−0.8
+3.8
−1.2
−0.15
−0.70
−0.10
−0.50
1305.2
1303.8
1560.6
1560.4
1063.5
1062.5
−0.7
−3.2
−0.9
CrSi+CrSi2
All Δs represent the differences between the silicides and the metal references.

334
P.L. Tam et al. / Surface Science 606 (2012) 329–336
Table 6
Composition of different transition metal silicide thin films by means of XPS quantitative analysis and the possible initial formed phase predicted by Pretorius' effective heat of for-
mation (EHF) model at as-deposited state [59,60].
As-deposited composition Possible phases Congruency Compound composition ΔH°
Limiting Effective concentration ΔH′
First phase
formed
(at.%)
[kJ (mol at.)−1] element (at.%) [kJ (mol at.)−1
]
Ti0.40Si0.60
Ti3Si
Ti5Si3
Ti5Si4
TiSi
TiSi2
Fe3Si
FeSi
FeSi2
Ni3Si
Ni5Si2
Ni2Si
Ni3Si2
NiSi
NiSi2
Cr3Si
Cr5Si3
CrSi
C
Ti0.75Si0.25
Ti0.625Si0.375
Ti0.56Si0.44
Ti0.50Si0.50
Ti0.33Si0.67
Fe0.75Si0.25
Fe0.50Si0.50
Fe0.33Si0.67
Ni0.75Si0.25
Ni0.71Si0.29
Ni0.67Si0.33
Ni0.60Si0.40
Ni0.50Si0.50
Ni0.33Si0.67
Cr0.75Si0.25
Cr0.625Si0.375
Cr0.50Si0.50
Cr0.33Si0.67
−53.0
−72.4
−81.0
−78.6
−57.0
−25.8
−39.3
−30.6
−37.2
−42.3
−46.9
−45.2
−42.4
−29.3
−34.4
−35.0
−30.2
−25.8
Ti
Ti
Ti
Ti
40
40
40
40
60
40
40
60
25
25
25
25
25
25
20
20
20
20
−28.3
−46.3
−57.9
−62.9
−51.0
−13.8
−31.4
−27.4
−12.4
−14.9
−17.5
−18.8
−21.2
−22.2
−9.2
C
C
Si
TiSi2
FeSi
Fe0.40Si0.60
Fe
Fe
Si
Ni0.25Si0.75
C
C
C
Ni
Ni
Ni
Ni
Ni
Ni
Cr
Cr
Cr
Cr
NiSi
Cr0.20Si0.80
C
C
−11.2
−12.2
−15.6
CrSi2
C
CrSi2
[59,60]. However, the high temperature treatment leads to the reaction
between FeSi and the remaining Si in the film and therefore FeSi trans-
forms to β-FeSi2 [7,16,54]. It should be noted that FeSi2 can be in
the form of two possible phases depending on the heat treatment tem-
perature, i.e. it can either be in β-form (or C49) (formed at 550 °C) [7] or
in α-form (or C54) (at or above 700 °C) [7,16]. In accordance with the
thermal processing conducted in our experiment, β-FeSi2 is charac-
agreement with Pretorius' own predictions [59,60]. However, as evi-
denced by the XRD results, the NiSi is totally transformed to NiSi2
after annealing. The explanation of this consequence is similar to that
of the Fe–Si system; the remaining Si reacts with the NiSi during the
annealing and leads to the formation of the Si-rich phase of NiSi2. For
the Cr–Si system, CrSi2 (i.e. Cr0.33Si0.67) is predicted from Table 4 as
the first phase to form. This is also in line with Pretorius' predictions
[59,60]. Still, the XRD results show that both CrSi2 and CrSi are formed.
It is, therefore, assumed that the prolonged heat treatment may lead to
terised to be the final form. For the Ni–Si system, NiSi (i.e. Ni0.50Si0.50
)
is the first congruent phase to form according to Table 4. This is in direct
a
b
600.0
599.5
599.0
598.5
598.0
597.5
597.0
596.5
596.0
595.5
391.0
Initial-state effect (Slope = -3)
390.5
Final-state effect (Slope = -1)
390.0
Fe
389.5
389.0
Ti
388.5
TiSi2
388.0
387.5
387.0
386.5
-FeSi2
454.2
454.1
454.0
453.9
707.1
707.0
706.9
706.8
706.7
706.6
Binding energy (eV)
Binding energy (eV)
c
d
491.0
490.5
490.0
489.5
489.0
488.5
488.0
487.5
487.0
486.5
709.0
708.5
708.0
707.5
707.0
706.5
706.0
705.5
705.0
704.5
Ni
Cr
CrSi + CrSi2
574.20
NiSi2
854.5
854.0
853.5
853.0
852.5
574.30
574.25
Binding energy (eV)
Binding energy (eV)
Fig. 3. Wagner plots for core level 2p3/2 line and L3M23M23 Auger peaks in different silicides studied. The arrows represent twice the core level shifts with respect to metal reference
caused by the final state effects (solid arrow) and initial state effects (dashed arrow), respectively.

P.L. Tam et al. / Surface Science 606 (2012) 329–336
335
the CrSi2 decomposition into CrSi and Si. Hence, the binary phase is
observed.
The XPS analyses of the 2p core-level spectra of the transition
Acknowledgements
The authors gratefully acknowledge the competent assistance from
Mr. U. Jelvestam and the discussion with Prof. V. Langer in the Depart-
ment of Chemical and Biological Engineering at Chalmers.
metal elements of the silicide revealed the existence of both positive
and negative BE shifts for their silicide states compared to their me-
tallic states. Fig. 3 shows the Wagner plots for the core-level 2p3/2
binding energies and L3M23M23 Auger kinetic energies measured for
the different silicides. The Auger kinetic energy is on the ordinate
and the photoelectron binding energy is on the abscissa oriented in
the negative direction. The intercepts of the straight lines with slopes
−3 and −1 on the ordinate (at Eb =0 outside the plot in Fig. 3) rep-
resent the initial-state effects, β, and the final-state effects, α, respec-
tively, as indicated in Eqs. (2) and (3). The quantity of the core-level
chemical shift with respect to the transition metal state due to the
initial-state and final-state effects is represented by half of the dash
and solid arrow for each case in Fig. 3, respectively.
References
[1] S. Murarka, Intermetallics 3 (1995) 173.
[2] A.D. Chrikin, V.O. Lavrenko, V.M. Talash, Powder Metall. Met. Ceram. 48 (2009)
330.
[3] J.J. Petrovic, A.K. Vasudevan, Mater. Sci. Eng., A 261 (1999) 1.
[4] P.L. Tam, Fabrication and corrosion properties of nickel silicide thin films, Chal-
mers University of Technology, Licentiate thesis, Gothenburg, Sweden, 2009
[5] S.M. Lee, E.T. Ada, H. Lee, J. Kulik, J.W. Rabalais, Surf. Sci. 453 (2000) 159.
[6] E. Buzaneva, T. Vdovenkova, S. Litvinenko, V. Makhnjuk, V. Strikha, V. Skry-
shevsky, P. Shevchuk, V. Nemoshkalenko, A. Senkevich, A. Shpak, J. Electron,
Spectrosc. 68 (1994) 707.
[7] B.A. Ortlowski, B.J. Kowalski, K. Fronc, R. Zuberek, S. Mickevičius, F. Mirabela, J.
Ghijsen, J. Alloys Compd. 362 (2004) 202.
The final-state effects shift the core level to higher binding energy for
all the silicides studied as compared to those corresponding transition
metals. All the reduced α values as seen in Fig. 3 and read in Table 5 indi-
cate that relaxation of the core holes is less favourable in silicides with re-
spect to the metal references. Because the relaxation energy is transferred
from the core hole to the outgoing photoelectron, the decreased relaxa-
tion energy (Δα=2ΔR) will then lead to the positive shift of the core-
level electrons towards higher binding energies in the XPS measure-
ments. However, this effect is less important for the early transition
metal like Ti and the late transition metal like Ni, as can be seen from
the listed α values in Table 5 and the illustrations in Fig. 3a and c.
The NiSi2 exhibits the highest positive core-level BE shift among the
silicides studied. It is because this silicidation process involves lattice
size expansion, Ni\Ni bond length extension and atomic concentration
reduction (c.f. Table 4). Accordingly, the Ni in the NiSi2 structure behaves
in an atomic-like character [44]. In this case, the change in the initial-
state effects, Δβ, becomes more significant compared to that in the
final-state effects (i.e. the relaxation energy difference), see Table 5 and
Fig. 3c. Contrary to that of NiSi2, the initial-state effects for the other
metal silicides cause negative shift of the core level (Fig. 3a, b, d),
which can compensate the positive shift caused by final-state effects
partly or totally. Depending on the relative magnitude of these two ef-
fects, the core-level 2p3/2 peak can shift either negatively (Ti and Cr sili-
cides) or positively (Fe silicide). In the case of Cr silicides, comparable
magnitude but reverse impact of the initial-state and final-state effects
give rise to a small shift of only −0.1 eV.
[8] N. Ohtsu, M. Oku, A. Nomura, T. Sugawara, T. Shishido, K. Wagatsuma, Appl. Surf.
Sci. 254 (2008) 3288.
[9] S.K. Saha, R.S. Howell, M.K. Hatalis, Thin Solid Films 347 (1999) 278.
[10] Y. Cao, L. Nyborg, U. Jelvestam, Surf. Interface Anal. 41 (2009) 471.
[11] I. Bertóti, M. Mohai, K. Kereszturi, A. Tóth, E. Kálmán, Solid State Sci. 11 (2009)
1788.
[12] D. Caputo, G. de Cesare, M. Ceccarelli, A. Nascetti, M. Tucci, L. Meda, M. Losurdo, G.
Bruno, J. Non-Cryst. Solids 354 (2008) 2171.
[13] C.-G. Kim, Thin Solid Films 479 (2005) 182.
[14] L. Haderbache, P. Wetzel, C. Pirri, J.C. Peruchetti, D. Bolmont, G. Gewinner, Surf.
Sci. 209 (1989) L139.
[15] T.L. Barr, Appl. Surf. Sci. 15 (1983) 1.
[16] S. Agarwal, A. Jain, C. Lal, V. Ganesan, I.P. Jain, Appl. Surf. Sci. 253 (2007) 4721.
[17] K. Radermacher, S. Mantl, Ch. Dieker, H. Lüth, C. Freiburg, Thin Solid Films 215
(1992) 76.
[18] S. Mandal, C.K. Ghosh, D. Sarkar, U.N. Maiti, K.K. Chattopadhyay, Solid State Sci. 12
(2010) 1803.
[19] M. Satta, G. Moretti, J. Electron, Spectrosc. 178–179 (2010) 123.
[20] G. Moretti, J. Electron, Spectrosc. 95 (1998) 95.
[21] G. Moretti, J. Electron, Spectrosc. 76 (1995) 365.
[22] C.D. Wagner, A. Joshi, J. Electron, Spectrosc. 47 (1988) 283.
[23] C.D. Wagner, L.H. Gale, R.H. Raymond, Anal. Chem. 51 (1979) 466.
[24] C.D. Wagner, Anal. Chem. 44 (1972) 967.
[25] A. Franciosi, J.H. Weaver, Surf. Sci. 132 (1983) 324.
[26] S. Hasa, I. Ohdomari, Phys. Rev. B 38 (1988) 7554.
[27] P.J.W. Weijs, H. van Leuken, R.A. de Groot, J.C. Fuggle, Phys. Rev. B 44 (1991) 8195.
[28] M. Mizuno, I. Tanaka, H. Adachi, Phys. Rev. B 59 (1999) 15033.
[29] V. Kinsinger, I. Dézsi, P. Steiner, G. Langouche, J. Phys. Condens. Matter 2 (1990)
4955.
[30] A. Aranitis, S. Diplas, P. Tsakiroloulos, J.F. Watts, M.J. Whiting, S.A. Morton, J.A.D.
Matthew, Acta Mater. 49 (2001) 1063.
[31] N. Takao, I. Tanaka, H. Adachi, Intermetallics 4 (1996) S113.
[32] A. Franciosi, J.H. Weaver, D.G. O'Neill, Y. Chabal, J.E. Rowe, J.M. Poate, O. Bisi, C.
Calandra, J. Vac. Sci. Technol. 21 (1982) 624.
[33] E. Walker, C.P. Lund, P. Jennings, J.C.L. Cornish, C. Klauber, G. Hefter, Appl. Surf. Sci.
222 (2004) 13.
[34] M. Ekman, V. Ozoliņš, Phys. Rev. B 57 (1998) 4419.
[35] L. Braicovich, Thin Solid Films 140 (1986) 79.
5. Conclusions
[36] J.J. Jia, T.A. Callcott, W.L. O'Biren, Q.Y. Dong, D.R. Mueller, D.L. Ederer, Z. Tan, J.I.
Budnick, Phys. Rev. B 46 (1992) 9946.
[37] M. Oku, T. Shishido, H. Matsuta, K. Wagatsuma, J Electron Spectrosc. 153 (2006) 75.
[38] K. Ogawa, M. Sasaki, A. Ohnishi, K. Kitaura, H. Fujimoto, A. Azuma, K. Takahashi,
M. Kamada, Phys. Procedia 11 (2011) 63.
[39] S. Hüfner, G.K. Werheim, N.V. Smith, M.M. Traum, Solid State Commun. 11 (1972)
323.
[40] J.C. Fuggle, Z. Zołnierek, Solid State Commun. 38 (1981) 799.
[41] O. Bisi, C. Calandra, J. Phys, C: Solid State Phys. 14 (1981) 5479.
[42] P.J. Grunthaner, F.J. Grunthaner, A. Madhukar, J.M. Mayer, J. Vac. Sci. Technol. 19 (1981)
649.
[43] P.J. Grunthaner, F.J. Gruntahner, A. Madhukar, J. Vac. Sci. Technol. 20 (1982) 680.
[44] A. Franciosi, J.H. Weaver, F.A. Schmidt, Phys. Rev. B (1982) 546.
[45] M. García-Méndez, M.H. Farías, D.H. Galván-Martínez, A. Posada-Amarillas, G.
Beamson, Surf. Sci. 532–533 (2003) 952.
[46] A. Gheorghiu, C. Sénémaud, R. Asal, E.A. Davis, J. Non-cryst, Solids 82 (1995) 293.
[47] L. Gregoratti, S. Günther, J. Kovac, M. Marsi, M. Kiskinova, Appl. Surf. Sci.
144–145 (1999) 255.
Transition metal slicides were synthesised as crystalline thin films by
means of sputter co-deposition onto Si wafers followed by annealing.
Through the structural, phase and surface chemical characterisations
using XRD and XPS analyses involving core-level spectra and Auger
spectra acquistion, the crystallinity and the bonding character of the sil-
icide compounds formed are depicted. With the aid of Pretorius' EHF
model, the formation sequence of the silicide phases was defined in ac-
cordance with the experimental observations involving the formation of
single phase TiSi2, β-FeSi2 and NiSi2 for the Ti–Si, Fe–Si and Ni–Si depo-
sitions, but the formation of CrSi+CrSi2 for the Cr–Si deposition needs
further investigation. The chemical shifts of the XPS 2p3/2 core level for
the transition metals in silicide states compared to metallic states were
analysed by means of the Auger parameters and the Wagner plots. The
analyses show that the large chemical shift of +2.0 eV for Ni 2p3/2
from Ni to NiSi2 is mainly governed by the initial-state effects, while
the final-state effects (i.e. mainly extra-atomic relaxation energy) are
less important. For the other silicide compounds, the binding energy
shifts were significantly smaller and less than 0.3 eV with the compen-
sating effects of initial-state and final-state effects, leading to either pos-
itive or negative chemical shift.
[48] M. Taguchi, F. Le Normand, J. Hommet, S. Rey, G. Schmerber, J.C. Parlebas, Eur.
Phys. J. B 18 (2000) 611.
[49] J.-H. Xu, Y.-N. Xu, Solid State Commun. 55 (1985) 891.
[50] G. Ottaviani, C. Nobili, Thin Solid Films 163 (1988) 111.
[51] A. Guldan, V. Schiller, A. Steffen, P. Balk, Thin Solid Films 100 (1983) 1.
[52] H. von Känel, N. Onda, H. Sirringhaus, E. Müller-Gubler, S. Goncalves-Conto, C.
Schwarz, Appl. Surf. Sci. 70/71 (1993) 559.
[53] G.G. Kleiman, J. Mole, Cataly. A: Chem 228 (2005) 137.

336
P.L. Tam et al. / Surface Science 606 (2012) 329–336
[54] C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Mullenberg, Handbook of X-ray
Photoelectron Spectroscopy, Perkin-Elmer Corporation, 1979.
[55] I. Olefjord, Applications of ESCA in Materials Technology (Corrosion, weld embrit-
tlement, and battery electrodes), PhD thesis, Chalmers University of Technology,
Gothenburg, Sweden, 1974
[58] Powder Diffraction Files (PDF-2), International Crystallography Diffraction Data-
base (ICDD), 2000.
[59] R. Pretorius, Thin Solid Films 290–291 (1996) 477.
[60] R. Pretorius, T.K. Marais, C.C. Theron, Mater. Sci. Rep. 10 (1993) 1.
[61] W.B. Pearson, Handbook of Lattice Spacings and Structures of Metals, Pergamon
Press Ltd, 1967.
[56] M.P. Seah, Surf. Interface Anal. 31 (2001) 721.
[57] B. Vincent Crist, Handbooks of Monochromatic XPS Spectra, XPS International Inc,
1999.