Polarized X-ray absorption spectroscopy and XPS of TiS3: S K- and Ti L-edge XANES and S and Ti 2p XPS

Article abstract of DOI:10.1016/j.susc.2005.03.048

Sulfur and Ti 2p XPS and polarized S K- and L_2,3- and Ti L _2,3-edge XANES spectra have been obtained from oriented ribbon-like crystals of TiS₃ grown by vapor transport. The S 2p XPS spectrum for (0 0 1) crystal faces has line widths of 0.53 eV and is completely accounted for by partially overlapped 2p_3/2/2p_1/2 doublets for sulfide (S^2-) and disulfide (S22-) species, which are separate and independent entities in TiS₃. Evidence of unsaturated surface states is lacking, consistent with a surface monolayer for the (0 0 1) growth face of disulfide atoms oriented with their charge-neutral sides outward. The S K-edge XANES spectra show pronounced anisotropy in the (0 0 1) plane of TiS₃ crystals, associated with the photoelectron transition channel S 1s→3pxσ*_u with the electric vector (E) parallel the a-axis (and the S-S bond of the disulfide group), and transitions to unoccupied antibonding orbitals of S-Ti bonds with E parallel to the b-axis (the direction of the well-known quasi-one-dimensional character of TiS₃). The XPS and ultrasoft and soft X-ray region XANES spectra confirm the surface and near-surface structural integrity of TiS₃.

Full text of DOI:10.1016/j.susc.2005.03.048

Surface Science 584 (2005) 133–145
www.elsevier.com/locate/susc
Polarized X-ray absorption spectroscopy and XPS of TiS3:
S K- and Ti L-edge XANES and S and Ti 2p XPS
*
M.E. Fleet , S.L. Harmer, X. Liu, H.W. Nesbitt
Department of Earth Sciences, University of Western Ontario London, Ont., Canada N6A 5B7
Received 1 December 2004; accepted for publication 23 March 2005
Available online 13 April 2005
Abstract
Sulfur and Ti 2p XPS and polarized S K- and L2,3- and Ti L2,3-edge XANES spectra have been obtained from ori-
ented ribbon-like crystals of TiS3 grown by vapor transport. The S 2p XPS spectrum for (001) crystal faces has line
widths of 0.53 eV and is completely accounted for by partially overlapped 2p3/2/2p1/2 doublets for sulfide (S2ꢀ) and
disulfide (S22ꢀ) species, which are separate and independent entities in TiS3. Evidence of unsaturated surface states is
lacking, consistent with a surface monolayer for the (001) growth face of disulfide atoms oriented with their charge-
neutral sides outward. The S K-edge XANES spectra show pronounced anisotropy in the (001) plane of TiS3 crystals,
associated with the photoelectron transition channel S 1s ! 3pxrꢁu with the electric vector (E) parallel the a-axis (and
the S–S bond of the disulfide group), and transitions to unoccupied antibonding orbitals of S–Ti bonds with E parallel
to the b-axis (the direction of the well-known quasi-one-dimensional character of TiS3). The XPS and ultrasoft and soft
X-ray region XANES spectra confirm the surface and near-surface structural integrity of TiS3.
ꢀ 2005 Elsevier B.V. All rights reserved.
Keywords: X-ray photoelectron spectroscopy; X-ray absorption near edge spectroscopy; Sulfides; Layered structures; TiS3
1. Introduction
cathodes of non-aqueous alkali metal batteries [3].
Consequently, its synthesis [4,5], chemical spec-
troscopy [6–10], electronic structure [7,11], electri-
cal and magnetic properties [12–15] and single-
crystal anisotropy [8,9] have been well studied.
The compound has a relatively low-temperature
stability, undergoing peritectic decomposition at
632 ꢁC, 8.95 atm.
Titanium trisulfide (TiS3) is a low-dimensional,
van der WaalsÕ bonded, layered transition-metal
sulfide [1,2]. It belongs to the class of alkali-
metal-insertable chalcogenides used in CuO-based
*
Corresponding author. Tel.: +1 519 661 3184; fax: +1 519
Titanium trisulfide crystallizes with the mono-
661 3198.
E-mail address: mfleet@uwo.ca (M.E. Fleet).
clinic ZrSe3-type crystal structure, in space group
0039-6028/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.03.048

134
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
˚
P21/m, with a = 4.958, b = 3.4006, c = 8.778 A,
and b = 97.32ꢁ [1,2] (Fig. 1). Two of the three sul-
fur atoms per formula unit are present as a single-
atom is in irregular eightfold coordination with
sulfur atoms. Six shorter Ti–S distances (of 2.45–
˚
2.49 A) form TiS6 trigonal prisms which are linked
bonded disulfide ion (S22ꢀ; S–S = 2.04 A), and the
˚
via shared faces into an infinite chain parallel to
the b-axis, making the structure quasi-one-dimen-
sional. Neighboring chains are offset by b/2 and
connected laterally via longer Ti–S bonds (of
third sulfur atom is formally sulfide (S2ꢀ). The Ti
˚
2.63 and 2.64 A) into sheets parallel to the crystal
face (001). Individual (001) sheets are bounded
above and below by a layer of disulfide groups,
and held together in the resulting two-dimensional
structure only by weak (van der WaalsÕ) forces.
Importantly for the present study, the bond axis
of the disulfide group is aligned parallel to the
crystallographic a-axis.
Sulfur 2p X-ray photoelectron spectroscopy
(XPS) measurements have been made on TiS3
[7,10] and, earlier, on the isostructural compound
ZrS3 [11]. Jellinek et al. [11] deduced a molecular
orbital scheme for the disulfide ligand in the
ground state of these two trisulfide compounds,
showing that the lowest unoccupied antibonding
orbital (3pxrꢁu) is projected parallel to the S–S
bond axis. The marked anisotropy of the TiS3
structure has been investigated by polarized infra-
red [8], Raman [16], and micro-Raman [9] spec-
troscopy. All of these vibrational studies reported
markedly different spectra for light polarized par-
allel and perpendicular to the direction of the
chain of TiS6 trigonal prisms (b-axis), and [8] in-
ferred a rather weak interchain interaction from
the very small frequency differences between the
Raman active and infrared active C point pho-
nons. X-ray absorption spectroscopy is also sensi-
tive to anisotropy of crystal structure because the
absorption edge features are dependent on both
the initial and final electron states, thus permitting
the anisotropy of conduction states to be probed.
Layer-structure compounds, like the transition-
metal chalcogenides, hexagonal nitrides, and
V2O5, have been particularly well studied in this
respect [17–21]. The last of these studies investi-
gated the polarized X-ray absorption of the hexa-
gonal CdI2-structure chalcogenides TiS2, TiSe2
and TiTe2, which have uniaxial physical (optical,
electric, magnetic) properties and maximum
anisotropy parallel to the c-axis. We presently
investigate the polarized X-ray absorption of
c
b
a
2.04
S(1)
Ti
S(2)
S(3)
(a)
{100} cleavage
(001)
crystal
face
c
b
(b)
a
Fig. 1. (a) Structure of TiS3 (type A) extending over about eight
unit cells: solid circles are Ti; open circles are S, S(1) is sulfide
(S2ꢀ) and S(2) and S(3) are disulfide (S22ꢀ); bond distances are in
angstrom; note that S–S bonds of disulfide groups (heavy lines)
are parallel to a-axis. (b) Sketch of ribbon-like crystal of TiS3 in
crystallographic orientation of part (a).

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
135
TiS3 at the S K- and L- and Ti L-edges. Whereas,
the layered-structure MX2 chalcogenides (M = Ti,
Zr, Hf, V, Nb, Ta, Mo, and W; X = S, Se and Te)
are isotropic perpendicular to the c-axis, TiS3 is
anisotropic both parallel and perpendicular to
the layer normal (d001), permitting a more direct
correlation between polarized absorption and ori-
entation of unoccupied antibonding orbitals. Also,
we show that a high-resolution S 2p XPS spectrum
for TiS3 is in precise agreement with the assumed
valence states of the sulfur atoms, and confirm
the absence of reconstruction surface states on
the (001) growth surface.
tube and reacted in a box furnace at 500 ꢁC over-
night and 700 ꢁC for 2 h. The product was then
crushed, re-tubed and heated at 1200 ꢁC for 1 h
and 1000 ꢁC for 60 h, and quenched in water.
2.2. XANES
Sulfur K- and L- and Ti L-edge XANES spectra
were collected at the Canadian Synchrotron Radi-
ation Facility (CSRF), Aladdin storage ring (Uni-
versity of Wisconsin at Madison, Wisconsin),
using simultaneous measurements in fluorescence
yield (FY) and total electron yield (TEY) modes,
with the latter indicated by current yield measured
at the sample holder. The storage ring was oper-
ated at 800 MeV, 60–180 mA. Sulfur K- and L-
edge spectra were measured on a double crystal
monochromator (DCM) beamline [22] and a Mark
IV Grasshopper monochromator beamline [23],
respectively. The DCM used an InSb (111) crystal
and had an energy resolution of about 0.6 eV
(FWHM) at 1840 eV; the Darwin width of the
crystal corresponds to an energy resolution of
0.9 eV for the S K edge. The double Bragg reflec-
tion geometry results in a beam at the sample that
is essentially 100% polarized for the S K edge [24].
The beam size at the sample was <1 · 3 mm2. The
grasshopper beamline used an 1800/mm holo-
graphic grating with an energy resolution of
<0.1 eV in the S L-edge region. The beam height
at the sample was ꢂ1 mm. The TEY recording
mode probes the near surface (estimated at
2. Experimental methods
2.1. Synthesis
Crystals of TiS3 were synthesized by vapor
transport in sealed evacuated silica glass tubes, of
7 mm internal diameter and about 170 mm in
length. Charges were prepared from stoichiometric
mixtures of high-purity Ti sponge and elemental
sulfur, and reacted in a horizontal furnace at
400–600 ꢁC for eight days, with the starting mixture
positioned in the furnace hot spot. The vapor-
transported products comprised millimeter-sized
crystals of TiS3 and TiS2, as well as elemental
sulfur. These phases were spatially separated at
the cold end of the tubes, facilitating an essentially
pure sampling of the TiS3 crystals. The titanium
sulfides were characterized by powder and single
grain X-ray diffraction, using Rigaku DMAX II
(Cu Ka X-radiation), Rigaku D/MAX-B system
(Co Ka) and Bruker D8 Discover (Cu Ka) diffrac-
tometer and Gandolfi camera (Cr Ka) procedures.
The type A structure for TiS3 was confirmed using
the X-ray diffraction criteria of Furuseth and Fjell-
˚
˚
<100 A depth for the S K edge and <50 A depth
for the S L edge), whereas the FY mode is more
representative of the bulk material [25,26]. As re-
cently noted [27], our S K-edge XANES spectra
of labeled sulfur compounds compare very favor-
ably with measurements on more recent generation
beamlines fitted with a Si(111) monochromator
[28]. This is because the core–hole lifetime broad-
ening is significantly greater than the resolution
of the Si(111) crystal. Also, the edge feature (or
white line) of the XANES spectra is often consid-
erably broadened by fine structure reflecting com-
plexity in the final states. Kravtsova et al. [27]
emphasized that no new features were evident in
S K-edge XANES spectra of metal monosulfides
recorded using a Si(111) monochromator.
˚
vag [2]. Crystals were elongated in the b-axis direc-
tion (Fig. 1b). The larger crystals were blade- or
ribbon-like on (001) with a perfect 100 cleavage
and up to 30 · 0.3 · 0.005 mm3 in size, but overall
crystal thickness was highly variable, grading
downwards to <1 lm for fibres and whiskers. Cad-
mium iodide-type TiS2 was prepared from a stoi-
chiometric mixture of Ti sponge and elemental
sulfur contained in a sealed evacuated silica glass

136
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
Sample mounts were prepared using double-
b
sided conducting carbon tape affixed to a clean
stainless steel disk. All operations for the grass-
hopper and spherical grating monochromator
(SGM) beamlines, including opening the silica
glass tubes and preparation of the mounts, were
conducted in a dry nitrogen-filled glove bag at-
tached to the port of the spectrometer sample
chambers [27,29]. For collection of polarized
absorption spectra, a textured (oriented) mount
was prepared by carefully laying larger blade-like
crystals of TiS3 in parallel alignment on the carbon
tape. After collection of the S L-edge spectra, the
textured mount was transferred to the DCM
beamline with minimal exposure to air. A separate
textured mount was prepared for the Ti L-edge
measurements (below). The experimental design
for XANES measurements on all three beamlines
allowed the electric vector (E) of the synchrotron
radiation to be oriented, at normal incidence [wave
vector k normal to (001)], parallel to either the
a-axis or the b-axis; i.e., corresponding to E perpen-
dicular and parallel to the elongation of the
textured crystals, respectively (Figs. 1 and 2). In
addition, the (001) plane of the crystals could be
tilted through an angle h up to 75ꢁ relative to E,
with the tilt angle set visually to within 3ꢁ. The
useful range of h for the DCM beamline was limited
by interference from the frame of the sample holder
to about 0–60ꢁ. Powder samples of TiS3 and TiS2
were prepared by lightly grinding a 1:1 mixture of
sulfide crystals and powdered quartz in an agate
mortar.
a
d001
k
E
Fig. 2. Experimental design for collection of XANES spectra
using textured sample mount: ribbon-like crystal of TiS3 is
viewed in projection down b-axis (cf., sketch in Fig. 1b); k is
wave vector and E is electric vector of synchrotron radiation; h
is angle made by normal to (001) and direct beam, and varied
from 0ꢁ to about 75ꢁ. Two series of measurements were made:
(1) crystal rotated about b-axis (as illustrated) and E parallel to
the a-axis and S–S bond axis at h = 0ꢁ, and (2) crystal rotated
about a-axis and E parallel to the b-axis and normal to S–S
bond axis at h = 0ꢁ; note that, in the monoclinic crystal system,
the interplanar spacing d001 is not parallel to c-axis, and its
length is [csin(b)].
delivered >2 · 1010 photons/s (when the stored
electron beam is about 100 mA); the beam size at
the sample was 1 · 1 mm2, and the pressure at
the sample was 10ꢀ8–10ꢀ9 Torr. Measurement
conditions included a photon energy range of
455–474 eV at a step size of 0.1 eV and one scan
per sample, with a total collection time of approx-
imately 8 min per sample.
Data reduction and analysis used the BAN and
BGAUSS data analysis programmes [30] and
FORTRAN77 codes. All XANES spectra were
normalized against incident intensity (I0), which
was measured simultaneously, by current yield.
Sulfur K-edge spectra were calibrated to the K-
edge peak of native sulfur at 2472.0 eV [31], after
removal of a linear pre-edge background. The Ti
L-edge XANES spectra were calibrated relative
to the Ti L-edge of rutile (TiO2) at 458.23 eV
[32]. Background subtraction was made with an
empirical Shirley-like function [33] fitted to the
pre-edge region. The background for these L-edge
The DCM (S K-edge) measurement conditions
included a photon energy range of 2455–2515 eV
at a step size of 0.25 eV, with count times deter-
mined by counting statistics and two scans per
sample, giving a total run time of approximately
29 min per sample. The grasshopper (S L edge)
measurement conditions included a photon energy
range of 150–190 eV at a step size of 0.07 eV and
two scans per sample, with a total collection time
of approximately 20 min per sample.
The SGM beamline used for collection of the Ti
L-edge spectra was monochromatized by
a
600 lines/mm grating over the photon region
240–670 eV, and yielded a photon resolution
<0.25 eV at the Ti L-edge (ꢂ455 eV); the beamline

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
137
spectra was unsystematic from sample to sample
and complicated by the composite nature of the
spectra. The adopted correction, albeit unconven-
tional, appeared adequate for the present qualita-
tive use of these spectra. Features in all of the
present XANES spectra are accurate to 0.1 eV.
error. The refined values for S 2p3/2 peak position
and FWHM were 161.2 eV and 0.53 eV for the sul-
fide atom (S2ꢀ) and 162.4 eV and 0.53 eV for the
disulfide group (S22ꢀ), and the disulfide/sulfide peak
intensity ratio was 2.021.
The S 2p XPS spectrum of TiS3 (Fig. 3a) is of
exceptional resolution, and confirms the chemical
state of the sulfur atoms inferred from earlier stud-
ies [7,10,11]. The XPS spectrum is seen to be com-
posite, and comprised of three narrow, symmetric
overlapped peaks representing S 2p3/2 and S 2p1/2
doublets for one sulfide (S2ꢀ) atom and one disul-
fide (S22ꢀ) group per formula unit, with the S 2p1/2
limb of the S2ꢀ doublet overlapping the S 2p3/2
limb of the S22ꢀ doublet. The fitted peak widths
for the sulfide (0.53 eV) and disulfide (0.53 eV)
components are significantly smaller than in previ-
ous studies (e.g., 0.7–0.75 and 0.8 eV, respectively,
in [10]), and result in minimal overlap of the three
resolved spectral components. One of the most
interesting aspects of the S 2p spectrum of TiS3
is that the two sulfur doublets provide a complete
fit to the spectrum. The fitted disulfide/sulfide peak
intensity ratio of 2.021 is within error of the ideal
value of 2, and confirms the stoichiometric integ-
rity of the near-surface monolayers of the sample.
It is evident that there is no measurable intervalent
transfer between the two sulfur species: the sulfide
and disulfide species are indeed separate and inde-
pendent entities in titanium trisulfide.
2.3. XPS
X-ray photoelectron spectroscopy (XPS) mea-
surements used a Kratos Axis Ultra instrument fit-
ted with a monochromated Al Ka X-ray source
(1487 eV) operated at 15 kV and 14 mA and a spot
size of 300 lm. The detector was at normal inci-
dence to the sample surface enabling analysis up
˚
to a depth of 50–100 A. A textured sample of
TiS3 crystals was prepared in a glove box filled
with dry Ar gas attached to the XPS spectrometer.
The analytical chamber was maintained at 2 ·
10ꢀ9 Torr. Survey spectra were collected using a
pass energy of 160 eV and a step size of 0.7 eV,
to check sample purity. High-resolution spectra
were collected at 10 eV pass energy using a step
size of 0.025 eV. All spectra were fitted by least-
squares refinement, after subtraction of a Shirley
background [33]. The spectrometer was standard-
ized to the Au 4f peak at 84.0 eV and calibrated
using the Cu 2p3/2 peak at 936.6 eV.
3. Results and discussion
The increased resolution of the new generation
of conventional XPS instruments has produced
spectra with much narrower peak widths than in
previous studies. The present values of 0.53 eV
are consistent with recent measurements of S 2p
peak widths in transition-metal sulfides (0.41–
0.55 eV [35–38]). As the inherent peak width
measured in the gas phase is 0.12 eV, these exper-
imental values for sulfides show that the resolution
of the Kratos Axis Ultra is in the order of only
0.3 eV. Consequently, excess intensity attributable
to surface chemical species formed as a result of
fracture is now resolved routinely using conven-
tional XPS instruments [35]. Although the Kratos
system does not resolve surface states in the detail
afforded by the highly surface sensitive techniques
of synchrotron radiation XPS (SRXPS) [35–38]
and synchrotron radiation photoemission [39],
3.1. S 2p XPS spectrum
The S 2p XPS spectrum of TiS3 obtained with
the Kratos Axis Ultra (Fig. 3a) has been fitted with
spin–orbit split doublet peaks for two sulfur spe-
cies, interpreted as sulfide (S2ꢀ) and disulfide
(S22ꢀ), using peak shapes of 50% Gaussian/50%
Lorentzian (G:L ratio of 1:1). Corresponding pairs
of S 2p3/2 and S 2p1/2 peaks were assigned the same
peak width (FWHM) and separated by 1.19 eV
[34,35], with the S 2p1/2 peak assigned half the
intensity of S 2p3/2 according to theory. Thus,
the binding energy, intensity and FWHM of the
S 2p1/2 peak were calculated entirely from the fitted
parameters of the S 2p3/2 peak. The G:L peak
shape ratio of 1:1 was optimized by trial-and-

138
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
S 2p
Ti 2p3/2
S22- 2p3/2
+
S2- 2p1/2
S2- 2p3/2
S22- 2p1/2
168
166
164
162
160
158
460
458
456
454
binding energy (eV)
binding energy (eV)
(a)
(b)
Fig. 3. XPS spectra for textured TiS3 crystals: (a) S 2p, showing S 2p3/2/S 2p1/2 doublets for one sulfide (S2ꢀ; dot-dot-dash peaks) and
two disulfide (S22ꢀ) atoms per formula unit; S 2p3/2 peaks are at 161.2 eV for S2ꢀ and 162.4 eV for S22ꢀ; note that the S 2p3/2 peak of S22ꢀ
overlaps the S 2p1/2 peak of S2ꢀ. (b) Ti 2p3/2: peak for pristine TiS3 is at 455.9 eV; weak dot-dot-dash peak probably represents TiO.
there is a good correspondence between the bind-
ing energies of the excess intensity from conven-
tional XPS and the surface states from SRXPS
for fractured surfaces of pyrite (FeS2), marcasite
(FeS2) and arsenopyrite (FeAsS) [35].
disulfide peaks [35], is observed in the Kratos S
2p XPS spectrum for TiS3 (Fig. 3a). This conclu-
sion is, of course, qualified by the limited resolu-
tion of the surface monolayer in conventional-
source XPS analysis. However, based on the com-
parison of conventional and synchrotron radiation
S 2p XPS spectra for pyrite-group sulfides in Fig. 2
of [35], we predict that additional (surface) fea-
tures will not be prominently present in the equiv-
alent SRXPS spectrum of TiS3. The apparent
absence of prominent surface states in the S 2p
XPS spectrum of TiS3 (Fig. 3a) is consistent with
the surfaces analyzed, which are predominately
(001) growth faces of the textured crystals (Fig.
1). By extension of this reasoning, the present
XPS spectrum confirms that the surface mono-
layer of the (001) face of TiS3 crystals is formed
of disulfide atoms. These atoms are essentially
charge neutral in respect to interactions in direc-
tions away from the TiS3 layer units, forming only
weak, van der WaalsÕ bonds with disulfide atoms
of adjacent layer units in the bulk structure. As
no covalent or ionic bonds are broken at the
(001) growth surface there is no need for surface
stabilization. Breaking of van der WaalsÕ bonds
should not result in extensive reconstruction of
the neighboring atoms as there is no sharing of
As their kinetic energy is about 1326 eV when
using an Al Ka source, S 2p photoelectrons tra-
verse approximately 15 monolayers before attenu-
ation. Ignoring packing density [40,41], the bulk S
2p signal contributes 90–95% and the surface sig-
nal about 5–10% of the total sulfur intensity [40–
42]. The majority of contemporary XPS studies
of metal sulfides have revealed additional intensity
resulting from surface atoms that have different
chemical environments than the fully coordinated
atoms in the bulk. In these covalently-bound com-
pounds, surface atoms exist in a lower coordina-
tion than in the bulk due to removal of nearest-
neighbor atoms as a result of fracture, cleavage
or scraping, or at a terminal growth surface pro-
ducing a high surface free energy. Stabilization of
these surfaces may be achieved by lowering the
surface free energy through redistribution of elec-
trons on surface atoms or surface reconstruction
[36,43,44].
No distinct surface signal, in the form of excess
intensity on the high binding energy flanks of the

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
139
electrons. The atoms at the surface should have the
3.3. XANES spectra
same coordination as those in the bulk. Addition-
ally, the absence of a high binding energy tail sug-
gests that no ligand-to-metal transfer mechanism
occurs in TiS3.
The polarized XANES spectra collected in the
soft X-ray region from oriented crystals of TiS3
using the total electron yield (TEY) are reported
in Figs. 4–7. In all cases, the fluorescence yield
spectra showed the same features as corresponding
TEY spectra, but at inferior peak-to-background
resolution. The S K-edge XANES (Figs. 4–6) are
characterized by a prominent absorption edge fea-
ture (white line) a, which represents transition of
the S 1s core electron to the lowest unoccupied
antibonding states on the S atom restricted by
the atomic selection rules DL = 1, DJ = 0 or 1,
in the electric dipole approximation. For most
S-bearing materials, peak a represents the photo-
electron transition S 1s ! S 3pr*. However, in
transition-metal sulfides unoccupied antibonding
states on the sulfur atoms may be hybridized with
empty metal 3d states [27,29,45,46], to the extent
that the white line of the S K edge may reflect as-
pects of the 3d electron configuration of the metal
atom(s) [47,48]; e.g., the edge feature of CdI2-type
TiS2 is split into two components a1 and a2
which apparently represent transition to S 3pr*
3.2. Ti 2p3/2 XPS spectrum
The Ti 2p3/2 spectrum collected from the TiS3
crystals (Fig. 3b) exhibits a strong peak centered
at about 456 eV and a small high binding energy
tail extending to about 458 eV. The G:L ratio
was adjusted to provide the best fit to the main
peak. The adopted value of 1:1 has a greater
Lorentzian component than previous studies. As
a result of more accurate electron analyzers, detec-
tors and monochromated X-ray sources, the
instrument response function is greatly reduced,
thereby revealing the underlying inherent Lorentz-
ian peak shape. A single symmetric peak was fitted
along the leading edge of the Ti 2p3/2 spectrum at
456.0 eV with a FWHM of 0.6 eV, and is consis-
tent with fully coordinated Ti4+ in the bulk. The
binding energy of the Ti 2p3/2 peak is in agreement
with that reported by [10] for TiS3. Convergence of
the refinement was obtained by fitting two peaks
to the high binding energy tail with FWHM con-
strained to the value of the main peak.
a
Even with increase in the Lorentzian compo-
nent of the G:L ratio, the fitted peak does not pro-
vide a good fit to the low binding energy side of the
Ti 2p3/2 spectrum. Similar excess low binding en-
ergy intensity has been observed in the metal peaks
of other sulfides, and remains incompletely under-
stood [35,37]. Possible explanations include the
presence of reduced Ti species at the surface and
use of an incorrect G:L function in peak fitting.
Reduced Ti may form at the surface as a result
of defects or from X-radiation beam damage.
The two additional peaks fitted to the high
binding energy tail of the Ti 2p3/2 spectrum have
binding energies of 456.5 and 457.4 eV. As a sub-
monolayer of O (1.95 at.%) was observed at the
TiS3 surface in the broad scan analysis and the
Ti 2p tail contributes 2.3% to the total signal, these
additional peaks (and high binding energy tail) are
most likely due to minor TiO and Ti oxysulfide
species, respectively [10].
b
S K edge
a
E // a-axis
b
c
45o
30o
15o
0o
a
2470
2480
2490
2500
energy (eV)
Fig. 4. S K-edge XANES spectra for textured crystals of TiS3
oriented with the electric vector (E) parallel to a-axis at h = 0ꢁ,
and rotated 15ꢁ, 30ꢁ and 45ꢁ about b-axis (cf., Fig. 2): inset
shows orientation of individual ribbon-like crystal at h = 0ꢁ;
spectra collected by TEY, and normalized to a common height
for the edge jump.

140
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
a
g
S K edge
c
Ti L2,3 edge
h
b
d
c
1. E // b-axis
f
a
b
2. powder
3. (1+4)/2
E // b-axis
a
E // a-axis
4. E // a-axis
powder
455
460
465
energy (eV)
470
475
2470
2480
energy (eV)
2490
2500
Fig. 7. Ti L2,3-edge XANES spectra for TiS3 crystals with E
parallel to a- and b-axes [i.e., E parallel and normal to S–S bond
axis, respectively, with k normal to (001)] and a powdered
sample of TiS3: features a, b, c, and d represent the Ti L3 edge
and f, g and h the Ti L2 edge; spectra collected by TEY, and
normalized to a common height for edge feature g.
Fig. 5. S K-edge XANES spectra for TiS3 crystals oriented with
E parallel to a- and b-axes, k normal to (001), compared with
spectra for the powdered sample and the average of the two
crystal spectra (dot-dash line): spectra collected by TEY, and
normalized to the edge jump.
resonances. XANES spectra for the series of mea-
surements starting with the electric vector (E) par-
allel to the a-axis and k normal to (001) (Fig. 4)
have an additional feature a0 present as a shoulder
on the high energy side of the edge feature.
Although our spectrometer did resolve the split-
ting of the edge feature into components a1 and
a2 for samples of CdI2-type TiS2 (Fig. 6) and the
intercalated sulfide NixTiS2 (x ! 0), the S K edge
of TiS3 is clearly marked by a singlet feature.
Perhaps more significant is the apparent ab-
sence of evidence of composite structure in the
XANES spectrum, because the K edge of TiS3 rep-
resents absorption by two quite distinct sulfur spe-
cies, S2ꢀ and S22ꢀ, and each of these sulfur species
has an irregular bonding sphere. The two disulfide
atoms [S(2) and S(3)] are both in pyramidal coor-
dination with one S and two Ti atoms, in a S(STi2)
cluster (Fig. 1), whereas the single sulfide atom is
in tetrahedral coordination with Ti and forms
a2
S K edge
TiS2
a1
b
E //-axis b
sulfur
E // a-axis
2470
2480
energy (eV)
2490
2500
Fig. 6. S K-edge XANES spectra for TiS3 crystals oriented with
E parallel to a- and b-axes [i.e., E parallel and normal to S–S
bond axis, respectively, with k normal to (001); dot-dash lines],
compared with XANES for powdered native sulfur and CdI2-
type TiS2, respectively; spectra collected by TEY, and normal-
ized to a common height for the edge jump.
˚
˚
two short (2.45 A) and two long (2.63, 2.64 A)
S–Ti bonds. The S K-edge feature of TiS3 is lo-
cated at about 2471.5 eV, compared with 2469.5–
2470 eV for most transition-metal monosulfides,
2471 eV for disulfides of the pyrite-group, and
2472 eV for native sulfur [49,50]. On this basis,
antibonding states mixed with empty t2g and eg
orbitals on Ti, respectively [47] (Fig. 6). The vari-
ous post-edge features (b, c, etc.) of the S K-edge
XANES are best attributed to multiple scattering

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
141
the S K edge of TiS3 might seem to be dominated
by absorption associated with the two disulfide
atoms. However, for a wide range of sulfur com-
pounds, the white line of the S K edge shifts to-
wards higher energy with increase in charge on
the metal and decrease in the metallic/covalent
character of the metal–S bond, being displaced as
high as 2477 eV for the second edge peak of the
ionically bound CaS [50]. Therefore, it is reason-
able to assume that the K edge feature of the sul-
fide (S2ꢀ) atom in TiS3 is displaced upwards into
near-coincidence with that of the two disulfide
atoms.
The S L-edge XANES spectra of transition-me-
tal sulfides are complicated by superposition of the
L3 and L2 edges shifted by the spin–orbit splitting
value of 1.1–1.2 eV, and the two dipole-allowed
photoelectron transition channels (2p ! 3sr*)
and (2p ! 3d) [27,29]. The spectra for TiS3 were
all similar to the S L-edge XANES of the isostruc-
tural compound ZrS3 [51], and are not presently
reported.
The Ti L-edge XANES spectra for the textured
single crystals show excellent resolution of the L3
and L2 edges (Fig. 7), corresponding to the Ti
2p3/2 and Ti 2p1/2 excitations, respectively. The
measured spin–orbit splitting is about 5.4 eV.
Although the Ti 2p electron has dipole-allowed
transitions to both s- and d-like final states, the
3d channel is much stronger than the 4s channel,
due to the large wavefunction overlap. In conse-
quence, the Ti 2p absorption is determined
very largely by the transition probability for
3d0 ! 2p53d1. Both L3 and L2 edges are marked
by two prominent features (c and d for L3 and g
and h for L2) which are attributed to splitting of
the 3d levels by the local crystal field of the Ti
atom. The weak pre-edge features (a and b for
L3 and f for L2) are probably attributable to mul-
tiplet core hole-3d electron interactions [52]. For
octahedrally-coordinated Ti4+ in oxide lattices, c
and g relate to transitions to empty 3d(t2g) orbital
levels and d and h to transitions to empty 3d(eg)
levels [53–55]. The measured crystal field splittings
for TiS3 are 1.9 eV at the L3 edge and 1.7 eV at the
L2 edge. The intensity relationships of the crystal
field-split peaks in this sulfide are reversed com-
pared with XANES spectra for Ti in four, five
and sixfold coordination in crystalline oxides and
in silicate glasses [32], which consistently show rel-
ative absorption intensities of d > c and h > g [55].
Interestingly, measurement of the Ti L-edge
XANES for CdI2-type TiS2, which has Ti in octa-
hedral coordination, resulted in a spectrum unlike
that of TiS3 but similar to those of the Ti oxides,
particularly Ba titanate which has Ti in fourfold
coordination [32]. The sample mount for the pres-
ent measurement on TiS2 was prepared in air, so it
is possible that the spectrum reflected contamina-
tion of the near surface of the sulfide with oxygen.
In support of this suggestion, unpublished Ti L-
edge study of binary Ti–S compositions with sam-
ple mounts prepared in air consistently yielded
broadly similar TEY spectra, with d and h stron-
ger than c and g, respectively, even though the cor-
responding FY spectra showed a progressive shift
of all features to lower energy with decrease in the
mean oxidation state of the Ti atoms. This obser-
vation may be evidence for the putative titanium
oxysulfide thin-film phase [10]. Resolution of the
second L3, L2-edge feature of TiS3 (d and h,
respectively; Fig. 7) was lost with light grinding
of the sample in the glove bag, perhaps reflecting
the fragility of the trichalcogenide crystal struc-
ture. However, all measurements on textured
mounts of TiS3 prepared in air resulted in a similar
Ti L2,3-edge XANES spectrum, suggesting that the
loss in resolution of features d and h might also
reflect minor incipient oxidation during grinding.
3.4. Polarization effects
Polarized absorption measurements were made
with E first parallel to the a-axis of the TiS3 crystals
and k normal to (001) (h = 0ꢁ), and then inclined to
the a-axis (and the plane of the structural layer
unit) at angles (h) of 15ꢁ, 30ꢁ and 45ꢁ by rotating
the crystal about the b-axis (Fig. 2). Measurements
were repeated with E first parallel to the b-axis and
k normal to (001) (h = 0ꢁ), and then inclined to b-
axis at angles of 15ꢁ, 30ꢁ, 45ꢁ, 60ꢁ, and 75ꢁ, by
rotating about the a-axis. The experimental design,
with the use of textured crystal mounts, allowed
most of the crystal anisotropy to be investigated.
When E is parallel to the a-axis (h = 0ꢁ), it is per-
pendicular to the TiS6 trigonal pyramidal chain

142
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
but parallel to the S–S bond axis of the disulfide
group (Figs. 1 and 2). Increase in h tilts E away
from the plane of the structural layer unit: k re-
mains normal to the b-axis but is no longer normal
to (001). Correspondingly, when E is parallel to
the b-axis (h = 0ꢁ), it is parallel to the TiS6 chain.
Increase in h tilts E away from the plane of the
structural layer unit (and crystal elongation): k re-
mains normal to the a-axis and the S–S bond axis
but is no longer normal to (001).
(e.g., Fig. 7) is readily accounted for, because tran-
sitions of core electrons to s final states are less
likely to reveal crystal anisotropy than transitions
to p and d states. Also, Ti4+ is in a fairly high
(eightfold) irregular coordination with sulfur and
less likely to show anisotropy for 3d orbital sym-
metry; see [56] for splitting of 3d states in the more
strongly bonded TiS6 trigonal prismatic cluster.
Nevertheless, the first shell sulfur atoms are asym-
metrically distributed, with four sulfide atoms on
one side of the Ti atom and four disulfide atoms
bridged by a strong r bond on the other side,
and the crystal field splitting of 5.4 eV revealed
by the Ti L-edge XANES is moderate. In contrast,
the S K-edge XANES spectra, which largely probe
the s ! p photoelectron transition channel, reveal
strong absorption anisotropy in TiS3. These obser-
vations are consistent with previous studies on lay-
ered structure compounds [17–20], where crystal
orientation effects are less marked for L edges than
K edges, and strongest for the K edge of the ligand
atom(s): e.g., in an early study, strong anisotropy
was evident at the Se K edge of the MoS2-type
diselenide WSe2 but no anisotropy was discerned
at the corresponding W L edge, nor at the Ta L
edge of the CdI2-type disulfide TaS2 [17].
The maximum absorption anisotropy for TiS3
occurs in the (001) plane of the layer unit, rather
than normal to the layer unit as in the MX2 chalco-
genides [17,18,21], wurtzite-type nitrides [20], and
V2O5 [19]. This difference in the orientation of the
anisotropy is probably related to the presence of
two strongly-bonded and mutually perpendicular
features in the layer unit of the trisulfide: the TiS6
chain (the quasi-one-dimensional structural unit)
parallel to the b-axis and the disulfide group paral-
lel to the a-axis (Fig. 1). Absorption with E parallel
to the b-axis and the TiS6 chain involves photoelec-
tron transitions to antibonding states associated
with resolved components of the four disulfide
atom–Ti bonds and two short sulfide–Ti bonds.
On the other hand, with E parallel to the a-axis,
the absorption involves final states associated with
the S–S bond [which gives rise to two interactions,
S(2)–S(3) and S(3)–S(2)] and resolved components
of the four disulfide atom–Ti bonds, with a minor
contribution associated with the resolved compo-
nents of the two long sulfide–Ti bonds.
Absorption anisotropy is pronounced for S K-
edge measurements, particularly in the (001) plane
of the crystals at h = 0ꢁ (Figs. 5 and 6), but not for
S and Ti L-edge measurements. For E parallel to
the a-axis (h = 0ꢁ), and k normal to (001), the S
K-edge feature a is more intense and is augmented
by an additional feature a0, whereas the multiple
scattering feature b is more prominent for E paral-
lel to the b-axis, and k normal to (001). Also, the
energy position of the edge feature a is character-
ized by shifts to 2471.3 eV for E parallel to the
a-axis and 2471.5 eV for E parallel to the b-axis.
The S K-edge XANES spectrum for the powdered
sample is closely reproduced by the average of the
two XANES spectra for k normal to (001) (h = 0ꢁ;
Fig. 5). Increase in the tilt angle (h) to 60ꢁ resulted
in no significant change in the XANES spectra for
the series of measurements starting with E parallel
to the b-axis (and perpendicular to S–S bond axis).
However, for the series of measurements starting
with E parallel to the a-axis, significant change oc-
curred at h = 30ꢁ and 45ꢁ, with progressive dimi-
nution of the shoulder a0 and strengthening of
the multiple scattering feature b (Fig. 4); the spec-
tra collected at 60ꢁ and 75ꢁ were unusable due to
interference from the sample holder. The S L-
and Ti L-edge XANES spectra were similar for
all orientations investigated (e.g., Fig. 7). How-
ever, the Ti L-edge spectrum for E parallel to the
a-axis shows weaker pre-peak features (a and f)
and evidence of complexity in the edge features
c, d, g, and h. Increase in tilt angle resulted in some
strengthening of the high-energy crystal field peaks
(d and h) for both series of measurements, with a
close approach to a common spectrum at the max-
imum tilt angle investigated (75ꢁ).
The minimal absorption anisotropy evident in
the S and Ti L-edge XANES spectra of TiS3

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
143
We note that the disulfide group has the appro-
priate orientation and ground state electronic
structure [7,11] to account for the anomalous
absorption with E parallel to a-axis (Fig. 5). In
particular, the abrupt decrease in strength of the
shoulder a0 with increase in tilt angle h beyond
15ꢁ (Fig. 4) is consistent with a dominant unidirec-
tional bond. The only unoccupied S 3p antibond-
that TiS3 is diamagnetic [12], whereas CdI2-type
TiS2 displays weak Curie–Weiss paramagnetism
[57] and, hence, probably has a finite, albeit low,
occupancy of Ti 3d states. The disulfide ion is well
known to impose a low-spin 3d electron configura-
tion and diamagnetic character on pyrite (FeS2)
[58]. Analogously to TiS3, the S K-edge XANES
spectrum of pyrite does not display crystal field
splitting [49].
ing orbital of the disulfide (S22ꢀ) ion (S 3pxrꢁ), is
u
oriented along the S–S bond axis. The two S
3py,z antibonding orbitals are p-bonding in charac-
ter and fully occupied within the valence band of
We recognize that the high energy shoulder (a0)
to the edge feature of XANES spectra for the ser-
ies of measurements starting with E parallel to the
a-axis (h = 0ꢁ; Fig. 4) could reflect a degree of mix-
ing of empty S 3pxrꢁu and Ti 3d orbitals, as was
suggested analogously for a similar feature in the
S K-edge XANES of 2H-MoS2 [18]. However,
we have noted above that p–d hybridization is very
likely suppressed in TiS3. Also, the relatively high
energy of a0 (2473.9 eV), and its abrupt diminution
with increase in inclination of E to the direction of
the S–S bond axis, is more consistent with a multi-
ple scattering satellite to the edge feature.
The component of the S K-edge white line pres-
ently associated with directed S–S bonds in TiS3 is
somewhat analogous to the r-shape resonance of
chemisorbed gas molecules [59], although the latter
tends to occur as a broad multiple scattering fea-
ture in the immediate post-edge region. Inner-shell
electron energy-loss spectroscopy (ISEELS), which
in the limit of small momentum transfer yields
information equivalent to XANES spectroscopy,
is generally used in preference to XANES for the
study of r-shape resonances [60]. Finally, study
of the bulk structure of strongly-bonded com-
pounds of the light elements using either XANES
or another electron energy-loss technique (electron
energy-loss near-edge structure, ELNES) routinely
reveals absorption edge fine structure attributable
to differential bonding characteristics [61,62].
the S22ꢀ ion, leaving 3pxrꢁ as the dominant disul-
u
fide orbital of p character at the bottom of the con-
duction band of the crystal. Absorption due to the
S 1s ! 3pxrꢁu transition is a maximum with E par-
allel to the a-axis of the TiS3 crystals (h ꢃ 0ꢁ), be-
cause E is then parallel to the bond axis of the
disulfide group. The absorption contributions
from the S–S bond are expected to dominate the
S K-edge XANES spectrum for this orientation.
In support of the present suggestion for domi-
nance of the S–S bond for E parallel to the a-axis
(h = 0ꢁ), note that the S K-edge XANES spectrum
bears some similarity to that of native sulfur (and
pyrite) but is quite unlike the S K-edge XANES of
TiS2, particularly in respect to the relative heights
of features a and b, and the lack of evidence of p–d
hybridization in the edge feature (Fig. 6). Con-
versely, when the crystals are oriented so that E
is parallel to the b-axis (and normal to the S–S
bonds), the S 1s ! 3pxrꢁu transition no longer
operates. The final states for the s ! p channel
are now predominantly unoccupied S s,pr* anti-
bonding orbitals associated with S–Ti bonds. In
consequence, the S K-edge spectrum for E parallel
to the b-axis (h = 0ꢁ) bears some similarity to the S
K-edge XANES of TiS2 (Fig. 6).
The absence of crystal field splitting in the edge
feature a shows that p–d hybridization, which is so
important in defining final states for TiS2 and
many other transition-metal sulfides, is suppressed
in S K-edge XANES spectra of TiS3. This may be
related to the unusual stereochemical feature of
bridging S–S bonds in the first coordination sphere
of Ti (Fig. 1a), which could have displaced the Ti
3d orbitals out of energy alignment with the empty
S 3p antibonding orbitals. It is also noteworthy
4. Conclusions
XPS and ultrasoft and soft X-ray region
XANES spectroscopy confirm the surface and
near-surface structural integrity of ribbon-
like crystals of TiS3 grown by vapor transport.
The S 2p XPS spectrum for (001) crystal faces

144
M.E. Fleet et al. / Surface Science 584 (2005) 133–145
has exceptionally narrow line widths (0.53 eV) and
is completely accounted for by two overlapped
doublets, for S2ꢀ (S 2p3/2 at 161.2 eV) and S22ꢀ (S
2p3/2 at 162.4 eV). There is no evidence of unsatu-
rated surface states in this conventional-source
XPS spectrum, showing that the disulfide atoms,
which interconnect the TiS3 layer units in the bulk
through van der WaalsÕ bonds, form the (001)
growth surface monolayer. The Ti 2p3/2 XPS spec-
trum is dominated by a singlet line at 455.9 eV, in
agreement with the 4+ oxidation state assumed for
the bulk, with additional weak intensity corre-
sponding to minor/trace TiO and Ti oxysulfide
species at the surface.
the Synchrotron Radiation Centre (SRC), Univer-
sity of Wisconsin-Madison, for their technical
assistance, and the National Science Foundation
(NSF) for support of the SRC under the grant
#DMR0084402. This work was supported by the
Natural Sciences and Engineering Research Coun-
cil of Canada.
References
˚
[1] S. Furuseth, L. Brattas, A. Kjekshus, Acta Chemica
Scandinavica A 29 (1975) 623.
˚
[2] S. Furuseth, H. Fjellvag, Acta Chemica Scandinavica 45
(1991) 694.
Polarized X-ray absorption measurements re-
vealed pronounced anisotropy in the (001) plane
of textured crystals at the S K edge but not at
the S and Ti L edges. When the electric vector
(E) of the synchrotron radiation is parallel to the
a-axis (and to the direction of the covalent S–S
bonds), and the wave vector is normal to (001),
the S K-edge XANES spectrum is dominated by
the photoelectron transition channel S 1s !
3pxrꢁu. The S 3pxrꢁu antibonding orbital is oriented
along the S–S bond axis and is the only unoccu-
pied S 3p antibonding orbital of the disulfide ion.
The XANES spectrum for this orientation is dom-
inated by the edge feature and is generally similar
to that of native sulfur. A markedly different S K-
edge spectrum, with a weaker edge feature, results
for E parallel to the b-axis, the direction of the
well-known quasi-one-dimensional character of
TiS3. The final states are now unoccupied anti-
bonding orbitals associated exclusively with re-
solved components of S–Ti bonds. XANES
spectra measured simultaneously by total electron
and fluorescence yield were essentially identical for
all three absorption edges investigated, demon-
strating that the near-surface structure is an
adequate description of the bulk structure.
[3] G.L. Holleck, J.R. Driscoll, Electrochimica Acta 22 (1977)
647.
[4] K. Oshima, M. Yokoyama, H. Hinode, M. Wakihara, M.
Taniguchi, Journal of Solid State Chemistry 65 (1986) 392.
[5] H.S.W. Chang, D.M. Schleich, Journal of Solid State
Chemistry 100 (1992) 62.
[6] K. Endo, H. Ihara, K. Watanabe, S. Gonda, Journal of
Solid State Chemistry 39 (1981) 215.
[7] K. Endo, H. Ihara, K. Watanabe, S. Gonda, Journal of
Solid State Chemistry 44 (1982) 268.
[8] S. Jandl, J. Deslandes, M. Banville, Infrared Physics 22
(1982) 327.
[9] P. Gard, F. Cruege, C. Sourisseau, O. Gorochov, Journal
of Raman Spectroscopy 17 (1986) 283.
[10] D. Gonbeau, C. Guimon, G. Pfister-Guillouzo, A. Levas-
seur, G. Meunier, R. Dormoy, Surface Science 254 (1991)
81.
[11] F. Jellinek, R.A. Pollak, M.W. Shafer, Materials Research
Bulletin 9 (1974) 845.
[12] H. Haakon, A. Kjekshus, E. Roest, A. Steffensen, Acta
Chemica Scandinavia 17 (1963) 1283.
[13] S. Kikkawa, M. Koizumi, S. Yamanaka, Y. Onuki, S.
Tanuma, Physica Status Solidi A 61 (1980) K55.
[14] P.L. Hsieh, C.M. Jackson, G. Gruener, Solid State
Communications 46 (1983) 505.
[15] E. Finkman, B. Fisher, Solid State Communications 50
(1984) 25.
[16] D.W. Galliardt, W.R. Nieveen, R.D. Kirby, Solid State
Communications 34 (1980) 37.
[17] S.M. Heald, E.A. Stern, Physical Review B 16 (1977) 5549.
[18] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, X.H. Feng,
K.H. Tan, Physics and Chemistry of Minerals 22 (1995)
123.
[19] E. Goering, O. Muller, M. Klemm, M.L. denBoer, S.
¨
Horn, Philosophical Magazine B 75 (1997) 229.
Acknowledgements
[20] K. Lawniczak-Jablonska, T. Suski, Z. Liliental-Weber,
E.M. Gullikson, J.H. Underwood, R.C.C. Perera, T.J.
We thank two anonymous reviewers for helpful
comments, Astrid Jurgensen and Kim Tan, Cana-
¨
Drummond, Applied Physics Letters 70 (1997) 2711.
ˇ
˚
[21] S. Bocharov, G. Dra¨ger, D. Heumann, A. Simunek, O.
ˇ
Sipr, Physical Review B 58 (1998) 7668.
dian Synchrotron Radiation Facility, and staff of

M.E. Fleet et al. / Surface Science 584 (2005) 133–145
145
[22] B.X. Yang, F.H. Middleton, B.G. Olsson, G.M. Bancroft,
J.M. Chen, T.K. Sham, K. Tan, D.J. Wallace, Nuclear
Instruments and Methods in Physics Research A 316
(1992) 422.
[42] S. Tanuma, C.J. Powell, D.R. Penn, Surface and Interface
Analysis 17 (1991) 911.
[43] W. Mo¨nch, Semiconductor surfaces and interfaces, Ch.
13Springer Series in Surface Sciences, Springer-Verlag,
1995.
[23] K.H. Tan, G.M. Bancroft, L.L. Coatsworth, B.W. Yates,
Canadian Journal of Physics 60 (1982) 131.
[44] A.S. Gibson, J.P. Lafemina, Structure of mineral surfaces.
Physics and chemistry of mineral surfaces, in: P.V. Brady
(Ed.), CRC Series in Chemistry and Physics of Surfaces
and Interfaces, CRC Press, 1996.
[24] Z. Hussain, E. Umbach, D.A. Shirley, J. Sto¨hr, J.
Feldhaus, Nuclear Instruments and Methods 195 (1982)
115.
[25] J. Kawai, H. Adachi, S. Hayakawa, S.Y. Zhen, K.
Kobayashi, Y. Gohshi, K. Maeda, Y. Kitajima, Spectro-
chimica Acta 49B (1994) 739.
[45] J.A. Tossell, Journal of Chemical Physics 66 (1977) 5712.
[46] A.V. Soldatov, A.N. Kravtsova, M.E. Fleet, S.L. Harmer,
Journal of Physics: Condensed Matter 16 (2004) 7545.
[47] Z.Y. Wu, G. Ouvrard, C.R. Natoli, Journal de Physique
IV France 7 (1997) C2-199.
[26] M. Kasrai, W.N. Lennard, R.W. Brunner, G.M. Bancroft,
J.A. Bardwell, K.H. Tan, Applied Surface Science 99
(1996) 303.
[48] S.P. Farrell, M.E. Fleet, I.E. Stekhin, A. Kravtsova, A.V.
Soldatov, X. Liu, American Mineralogist 87 (2002) 1321.
[49] D. Li, G.M. Bancroft, M. Kasrai, M.E. Fleet, X.H. Feng,
K.H. Tan, Canadian Mineralogist 33 (1995) 949.
[50] M.E. Fleet, X. Liu, S.L. Harmer, P.L. King, Canadian
Mineralogist, in press.
[27] A.N. Kravtsova, I.E. Stekhin, A.V. Soldatov, X. Liu,
M.E. Fleet, Physical Review B 69 (2004), 134109/1.
´
´
[28] M. Bonnin-Mosbah, N. Metrich, J. Susini, M. Salome, D.
Massare, B. Menez, Spectrochim. Acta B 57 (2002) 711.
[29] S.P. Farrell, M.E. Fleet, Physics and Chemistry of Min-
erals 28 (2001) 17.
[51] J.G. Chen, Surface Science Reports 30 (1997) 1.
[52] J.P. Crocombette, F. Jollet, Journal of Physics: Condensed
Matter 8 (1996) 5253.
[30] T. Tyliszcak, BAN Data Analysis Program, McMaster
University, 1992.
[31] J.C. Fuggle, J.E. Inglesfield, Unoccupied electronic states:
fundamentals for XANES, EELS, IPS and BISTopics in
Applied Physics, vol. 69, Springer-Verlag, 1992.
[32] G.S. Henderson, X. Liu, M.E. Fleet, Physics and Chem-
istry of Minerals 29 (2002) 32.
[53] F.M.F. de Groot, J.C. Fuggle, B.T. Thole, G.A. Sawatzky,
Physical Review B 41 (1990) 928.
[54] F.M.F. de Groot, J.C. Fuggle, B.T. Thole, G.A. Sawatzky,
Physical Review B 42 (1990) 5459.
[55] G. van der Laan, I.W. Kirkman, Journal of Physics:
Condensed Matter 4 (1992) 4189.
[33] D.A. Shirley, Physical Review B 5 (1972) 4709.
[34] H.W. Nesbitt, M. Scaini, H. Ho¨chst, G.M. Bancroft, A.G.
Schaufuss, R. Szargan, American Mineralogist 85 (2000)
850.
[56] R. Huisman, R. de Jonge, C. Haas, F. Jellinek, Journal of
Solid State Chemistry 3 (1971) 56.
[57] L.E. Conroy, K.C. Park, Inorganic Chemistry 7 (1968)
459.
[35] S.L. Harmer, H.W. Nesbitt, Surface Science 564 (2004) 38.
[36] S.L. Harmer, A.R. Pratt, H.W. Nesbitt, M.E. Fleet,
American Mineralogist 89 (2004) 1026.
[58] J.A. Tossell, D.J. Vaughan, Theoretical Geochemistry:
Applications of Quantum Mechanics in the Ear, Oxford
University Press, 1992.
[37] S.L. Harmer, A.R. Pratt, H.W. Nesbitt, M.E. Fleet,
Canadian Mineralogist, in press.
[59] J. Sto¨hr, J.L. Gland, W. Eberhardt, D. Outka, R.J. Madix,
F. Sette, R.J. Koestner, U. Doebler, Physical Review
Letters 51 (1983) 2414.
[38] S.L. Harmer, M. Tang, W. Skinner, H.W. Nesbitt, G.M.
Bancroft, in preparation.
[39] J.A. Rodriguez, J. Hrbek, Z. Chang, J. Dvorak, T. Jirsak,
A. Maiti, Physical Review B 65 (2002) 235414.
[40] M.P. Seah, W.A. Dench, Surface Interface Analysis 1
(1979) 2.
[60] F. Sette, J. Sto¨hr, A.P. Hitchcock, Chemical Physics
Letters 110 (1984) 517.
[61] L.A.J. Garvie, P.R. Buseck, Mineralogical Society of
America Reviews in Mineralogy 33 (1996) 821.
[62] M.E. Fleet, X. Liu, Physics and Chemistry of Minerals 28
(2001) 421.
[41] D. Briggs, M.P. SeahPractical Surface Analysis, vol. 1,
John Wiley & Sons, 1990.