Reductive coupling desorption of methanol on reduced SrTiO3(110)

Article abstract of DOI:10.1039/a807150k

The adsorption and reaction of methanol with SrTiO₃(110) has been studied using temperature-programmed desorption and X-ray photoelectron spectroscopy. The main organic products are methane and ethane, though the exact ratio of these depends on sample history and defect levels in the surface and bulk. It is likely that the methanol dissociates to form adsorbed methyl groups associated with reduced Ti³⁺ and Ti²⁺ centres which were formed by Ar+ bombardment. We tentatively associate ethane formation with the latter sites and dialkyl formation, while methane production may be associated with the former sites and monoalkyl formation.

Full text of DOI:10.1039/a807150k

View Article Online / Journal Homepage / Table of Contents for this issue
Reductive coupling desorption of methanol on reduced SrTiO (110)
3
Tatsuya Yoshikawaab and Michael Bowkera
a Catalysis Research Centre, Department of Chemistry, University of Reading, W hiteknights, P.O.
Box 224, Reading, UK RG6 6AD
b Materials & Functions Research L aboratory, Nippon Shokubai Co. L td., 5-8 Nishi Otabi-cho,
Suita, Osaka 564, Japan
Received 14th September 1998, Accepted 22nd December 1998
The adsorption and reaction of methanol with SrTiO (110) has been studied using temperature-programmed
3
desorption and X-ray photoelectron spectroscopy. The main organic products are methane and ethane, though
the exact ratio of these depends on sample history and defect levels in the surface and bulk. It is likely that the
methanol dissociates to form adsorbed methyl groups associated with reduced Ti3` and Ti2` centres which
were formed by Ar` bombardment. We tentatively associate ethane formation with the latter sites and dialkyl
formation, while methane production may be associated with the former sites and monoalkyl formation.
The structure of SrTiO and the (110) surface is shown in
Fig. 1. The bulk is of the perovskite type and the (110) plane
1
Introduction
3
The oxidative addition/reductive elimination process at a
metal ion site is a crucial feature in some homogeneous cata-
lysts system as has been demonstrated for WilkinsonÏs Rh
catalyst1 or for the Co catalyst in the oxo process.2 This step
can be described as follows:
consists of Ti, O and Sr. Although the coordination around Ti
is identical, there are two types of O. It is widely accepted that
Ti in the fully oxidised state is in the ]4 oxidation state (d0)
and in O is in the [2 oxidation state. Because of this, the fully
oxidised surface could work as an acidÈbase centre for
adsorbates at ambient conditions. In this work, we examine
Scheme 1
In general terms, as a result of the oxidative addition, both the
coordination number and the oxidation number of the metal
centre increase by two units. The reductive elimination is the
opposite process of the oxidative addition.
Whereas the importance of the above chemistry in homoge-
neous catalysis is established, the process has rarely been dis-
cussed in the area of heterogeneous catalysis for gasÈsolid
systems. This may reÑect either the difficulty in the heter-
ogeneous case of characterising the adsorbateÈsolid inter-
action, or a di†erence in the chemistry involved in the
catalytic cycles between the two systems. Gates3 states that
““reactivity of metal clusters is not a good basis for prediction
of the reactivity of ligands adsorbed on a surface. Nonetheless,
the concept of isolated reactive sites on surfaces as molecular
analogues is extremely valuable and borne out by a mass of
experimental results.ÏÏ He stresses the importance of local
structures for catalytic activity and surface reactivity, rather
than the bulk properties, such as electronic structure.
The surface science approach, which examines the
adsorbateÈsolid interaction under UHV conditions, could
provide an insight into the chemistry shown above because of
its ““cleanlinessÏÏ. In this sense, transition metal oxides would
be suitable for the investigation, since they are used as cata-
lytic materials and transition metal ions can change their oxi-
dation state and coordination number in such circumstances.
Pierce and Barteau4 found cyclotrimerisation products on
reduced TiO (001) and ascribed this reactivity to the presence
2
of Ti2`. In the present study, a SrTiO (110) single crystal was
3
chosen as the model surface and examined by XPS (X-ray
Fig. 1 (a) Bulk structure of SrTiO which is a perovskite lattice and
3
photoelectron spectroscopy) and TPD (temperature-pro-
grammed desorption), and probed using a model reactant,
methanol.
(b) the (110) surface structure, showing the two types of oxygen which
may exist on the perfect termination. The numbers in (b) correspond
to those shown in (a).
Phys. Chem. Chem. Phys., 1999, 1, 913È920
913

View Article Online
the adsorptionÈdesorption properties of methanol on the
reduced surface prepared by in situ Ar` sputtering.
stantially; however, an almost completely oxidised surface was
easily obtained by oxygen treatment at 820 K.
2.3 Experimental procedure
2
Experimental
After the pre-treatment of the sample, methanol was dosed to
the main chamber at a constant pressure in the range
5 ] 10~9È4 ] 10~6 mbar for varying lengths of time. TPD or
XPS spectra were measured with the background pressure
below 3 ] 10~9 mbar.
2.1 Apparatus
All the experiments were carried in a conventional UHV
chamber with a base pressure in the 10~10 mbar range. The
main background masses in the chamber were 2 (hydrogen),
15,16 (methane) and 28 (carbon monoxide). SrTiO (110)
(Pikem) was mounted without any pre-treatment. The back
3
2.4 XPS (X-ray photoelectron spectroscopy)
side and a small part of the front side was covered with gold
foil. It was either heated by the radiation of a W Ðlament from
the back side or cooled by liquid nitrogen constant Ñow into
the copper block mounted above it and connected to the gold
foil. The temperature was monitored with a chromelÈalumel
thermocouple, which was sandwiched between the gold foil
and the sample surface at the front side. A glass vial Ðlled with
methanol (99.8]%, BDH) was connected to the dosing line of
the main chamber, and puriÐed via freezeÈthaw cycles.
Al-Ka (1486.6 eV) radiation was used for the photoemission
measurements. The photoelectrons were collected by a con-
ventional CHA analyser (100 mm/, VSW Ltd) with a pass
energy of 20 eV. During measurement, the sample was earthed
through the thermocouple line.
In general, XPS has been shown to be useful for the analysis
of chemical states due to the chemical shift of the core level
binding energy for di†erent valence states. The states chosen
for the analysis were Ti 2p , 2p , O 1s, Sr 3p , 3p and
3@2
1@2
3@2
1@2
C 1s, respectively.
2.2 Sample treatment
The binding energy reference for the peak position was
either Sr 3p or 3p at 279.5 and 269.1 eV, respectively, on
The sample was oxidised by heating in 1 ] 10~6 mbar O
1@2
3@2
2
the assumption that the position in the binding energy scale
would not change with the experimental conditions, since the
spectral shapes were hardly a†ected throughout the treat-
ments shown in Table 1. By doing this, the charging e†ect on
the measured kinetic energy of photoelectrons is corrected.
When the Au 4f (from the sample holder) was used as a refer-
ence, the peak positions were not consistent for the series of
the spectra, which is owing to charging of the oxide, while the
Au was truly earthed.
(99.0%, ARGO) at 820 K for 10 min and subsequently cooled
in oxygen to 353 K. Ar (99.999%, ARGO) ion sputtering (at
500 K, 4.0 kVÈ1 lA in the scanning mode) was used for the
reduction of the sample, and since the ion gun was di†erentia-
lly pumped the pressure in the main chamber was 5 ] 10~7
mbar. Annealing at 820 K tended to reoxidise the surface sub-
2.5 Angle dependence of XPS
XPS spectra were taken at a glancing emission angle in the
present paper unless otherwise noted. This is to get precise
information on the composition of the surface layer(s). The
angle was adjusted by turning the manipulator connected to
the sample support. The angle between the X-ray gun and the
analyser was about 75¡.
Fig. 2(A) demonstrates that the intensity both of C 1s and
Sr 3p signals changed with variation of the electron take-o†
angle. As the C was on the surface, the change of the C 1s is
mainly the result of the e†ect of the change of X-ray Ñux
intensity. On the other hand, the change of the Sr signal
would be a†ected by variation of both the inelastic mean free
path of the photoelectrons and X-ray Ñux.
The angle dependence of the Sr/C atomic ratio normalised
by cos h is shown in Fig. 2(B) and this supports the above
consideration. The area of Sr 3p (normalised to C integral)
3@2
changed by more than 6 times between glancing and normal
angle. Consequently, as expected, the spectra taken at a glanc-
ing angle are much more surface sensitive than the normal
angle analysis.
2.6 TPD (Temperature-programmed desorption)
The sample was radiatively heated from the back and was
positioned 2 cm from the quadrupole mass analyser (VG). At
the beginning of the heating, typically in the 300È350 K range,
there was an induction period of non-linear heating which was
followed by a constant heating rate, typically D5 K s~1., over
the range 350È650 K.
Fig. 2 (A) Angle dependence of the Sr 2p and C 1s spectra of
SrTiO (110) covered by the contaminant carbon species, measured at
(a) 0¡, (b) 20¡, (c) 60¡, (d) 70¡ and (e) 80¡ from the surface normal,
3
Results
3
showing the increase of the relative intensity of Sr/C at high angles.
3.1 Sample pre-treatment
(B) Dependence of the C 1s/Sr 3p area ratio upon the angle emis-
3@2
After the oxidation treatment, the XPS spectrum of Ti 2p
sion with respect to the surface normal, calculated from Fig. 2(A) (C
1s cross section 1, Sr 4.4).
showed a simple peak characteristic of the Ti4` ion [Fig.
914
Phys. Chem. Chem. Phys., 1999, 1, 913È920

View Article Online
quantitative peak analysis in Table 2, it was apparent that the
topmost layer(s) were richer in oxygen than the sublayers (No.
1, 2). This might suggest that the top layer on a nearly fully
oxidised surface is, at least partially, covered by oxygen.
As a result of Ar` ion sputtering, the Ti 2p spectrum
changed its shape as shown in Fig. 3(b). Broadening of the
spectrum indicates the creation of new valence states, owing
to some conversion of Ti4` into lower oxidation state(s). The
peak area was estimated to be decreased by about 20%. The
O 1s [Fig. 3(c)] spectrum has become broader with asym-
metric tailing to a higher binding energy region, and could not
be Ðtted with a single Gaussian. The intensity of O signal has
decreased by about 30%, indicating that oxygen species were
preferentially removed by sputtering. The Sr 3p
shape (FWHM) was almost unchanged by the treatment as
shown in Fig. 3(a) and Table 1.
spectral
3@2
3.2 TPD of methanol on highly reduced SrTiO (110)
3
Thermal desorption after methanol dosing was carried out on
the reduced SrTiO (110) crystal. After the reduction by Ar`
3
sputtering, methanol was dosed at room temperature with a
base pressure in the machine of around 5 ] 10~10 mbar. Fig.
4 shows a typical TPD example after dosing 4 ] 10~6 mbar
for 1800 s. The main features were the desorption of ethane,
methane and some hydrogen around 400 K, and carbon
monoxide/hydrogen around 700 K. A broad feature was
observed for methanol over a wide range of temperature,
possibly owing to background desorption from the sample
support assembly. Ethane was identiÐed by its cracking
pattern which has major masses in the order 28, 27, 30, 26 and
29. In the Ðgure only mass 27 and 28 are shown. The proÐles
of CO and hydrogen at high temperatures have a similar
shape, suggesting coincident desorption.
3.3 XPS
The analysis of the reaction pathway was attempted by XPS.
Fig. 5 shows the XPS spectra at several points in the TPD
Fig. 3 XPS spectra of SrTiO for oxidised and sputtered surfaces: (a)
3
Sr 3p, (b) Ti 2p, (c) O 1s.
3(b)]. The peak features for each element are summarised in
Table 1. The values of FWHM (full-width half-maximum)
were 2.1 and 3.0 eV for Ti 2p and Ti 2p , respectively, for
FAT (Ðxed analyser transmission) 44 mode and each peak
could be Ðtted with a single Gaussian. From the results of
Fig. 4 TPD spectrum following methanol adsorption on reduced
3@2
1@2
SrTiO (110) at room temperature. The numbers refer to the relative
3
molecular mass of the species desorbed.
Table 1 Binding energya and FWHM (eV) of each component (FAT 44)
Sr
Ti(IV)
O
Element
chemical state
3p /eV
3p /eV
2p /eV
2p /eV
1s/eV
3@2
1@2
3@2
1@2
Oxidisedb
Oxidisedd
Sputterede
269.1(3.2)c
269.1(3.2)
269.1(3.3)
279.5(3.2)
279.5(3.2)
279.5(3.3)
458.8(2.1)
458.6(2.1)
458.4f
464.4(3.0)
464.5(3.0)
464.2f
530.7(2.2)
530.5(2.2)
530.2f
a Peak position. b Treated in oxygen at 820 K, analysed at ]20¡. c The value of FWHM is shown in parentheses. d Treated in oxygen at 820 K,
analysed at ]80¡ (glancing angle). e Sputtered, analysed at ]80¡ (glancing angle). f Results by Gaussian function Ðtting.
Phys. Chem. Chem. Phys., 1999, 1, 913È920
915

View Article Online
Table 2 Quantitative XPS peak analysisa of the methanol dosed surface
A.O.N.
[(4) ] 2]
(5)
O/Ti
(1)
Sr/Ti
(2)
O/Sr
(3)
(1)È(2)
(4)
(Curve-Ðtting)
(6)
(6)È(5)
(7)
Conditions
1b
2
3-Ae
3-Be
3.05
3.18
2.70
3.23
1.05
1.09
1.25
1.33
2.90
2.92
2.07
2.43
2.00
2.09
1.45
1.89
4.00
4.18
2.90
3.79
4
4
3.05
3.22
0
Ox.c
[0.18
Ox.,c gl.d
Sp.,f gl.d
Sp.,f gl.d
0.15
[0.57
a See details in the Section 4.4 (A.O.N. \ average oxidation number). b Analysed at ]20¡ sensitivity factors were decided based on this data,
cross section, Sr: 4.4, Ti: 7.9, O: 3.4. c Treated in oxygen at 820 K. d Analysed at ]80¡ (glancing angle). e Symbol A: pre-treated by Ar`
bombardment, B: after methanol dose. f Sputtered.
experiment, i.e. after pre-treatment, after methanol dosing and
after TPD.
After methanol dosing, a C 1s peak centred around 286.5
eV has appeared with FWHM of 2.2 eV, and a high binding
energy shoulder has appeared in the O 1s, owing to the
adsorbed species. The Ti 2p region has changed, the peak
becoming narrower with the adsorption of methanol. This
means that some of the reduced Ti has been oxidised
(oxidative addition). The Sr 3p peak shape was almost unaf-
fected by this adsorption.
After TPD, the C 1s intensity decreased to the background
level. Almost all the C species were desorbed by heating. The
shoulder of the O 1s, which had been formed by methanol
adsorption, disappeared and the oxygen species are therefore
either desorbed or incorporated as lattice oxygen. The Ti 2p
spectrum has become broader again, suggesting that the re-
reduction of Ti species has occurred. This e†ect is not merely
due to heating. The oxidation of Ti proceeds by the di†usion
of oxygen from the bulk to the surface. Therefore, the
broadening of the Ti 2p was accompanied by desorption of
the C-containing species. Since no signiÐcant C peak was
found after TPD, we can safely say that reductive-elimination
has occurred.
3.4 Curve-Ðtting of Ti 2p
The deconvolution of the Ti 2p spectrum of the reduced
sample is complex. First of all, the titanium ion can take
several oxidation states, which are supposed to be separated
by a few eV. Each oxidation state has two peaks, which are
5È6 eV apart in the spectrum due to spinÈorbit interaction in
the photo-emission process. Therefore the position and the
shape (FWHM) corresponding to each species has to be
decided to deconvolute the Ti 2p spectrum. Secondly, owing
to the large step in background signal on each side of the peak
envelope, the correct background subtraction is essential for
this analysis. Nevertheless, we attempt here to quantify the
nature of the reduced surface. The identiÐcation of the peak
positions of lower valence states of Ti was done by several
other workers5h9 and is summarised in Table 3. It is possible
to assign the oxidation states to Ti2`, Ti3` and Ti4` for the
reduced SrTiO (110).
3
Table 3 Ti 2p peak positions (eV)a for lower oxidation states
3@2
Ti2`
Ti3`
Ref.
[3.7
[3.5
[3.9
È
[1.8
[1.7
È
5
6
7
8
[1.5
[3.8
È
9
[3.9
[1.8
This work
Fig. 5 XPS spectrum after methanol TPD: (a) Sr 3p, C 1s, (b) Ti 2p,
(c) O 1s.
a Referenced to Ti4` peak position.
916
Phys. Chem. Chem. Phys., 1999, 1, 913È920

View Article Online
Table 4 Comparison of the deconvolution reults for Ti 2p spectra taken at di†erent FAT modes
FAT 22
FAT 44
Relative
Relative
Resolution
2p /eVa
2p /eV
ratio (%)b
2p /eVa
2p /eV
ratio (%)b
3@2
1@2
3@2
1@2
Ti2`
Ti3`
Ti4`
454.4(2.2)
456.5(2.1)
458.4(1.9)
460.1
462.0
464.2
27.6
24.8
47.6
454.5(2.6)
456.6(2.4)
458.4(2.1)
460.0
462.2
464.2
28.6
25.5
45.9
a The value of FWHM is shown in parentheses. b Calculated Ðgures based on the peak area.
Background subtraction was carried out according to the
procedure of Procter and Hercules.10 After background sub-
traction, the spectrum was found to Ðt well with the above
assumption, i.e. Ti 2p consisted of three paired species as
shown in Fig. 6. The extreme low and high binding energy
edge shape is informative, because it may reÑect the signal of a
single species. In the present case, it was concluded that one
species exists, since the edges were found to Ðt with a single
Gaussian function for di†erent resolution measurement (FAT
44 and 22). The results are summarised in Table 4. The peak
envelope could not be Ðtted by two contributing states (i.e.
Ti3` and Ti4`). The lowest binding energy peak was assigned
to Ti2`, because it is located 3.9 eV lower than that of Ti4`
(Table 3). The possibility of the existence of Ti1` or Ti0
species was excluded by these considerations.
These deconvolution procedures were applied to the spectra
after methanol dosing and the results are shown in Fig. 6 and
Tables 5 and 6. It is clearly demonstrated in Table 5 that, after
the adsorption, the Ti4` proportion has increased with a
decrease in the amount of Ti2` after methanol adsorption,
whereas the ratio of Ti3` species remained constant. The
result therefore indicates that methanol adsorption is accom-
panied by the oxidation of Ti2` species to Ti4` species.
3.5 Estimation of the analysis depth of XPS
The escape depth, j, which is the distance from the surface at
which the signal intensity decreases to (1/e) of that from the
surface layer, can be expressed in eqn. (1).
Fig. 6 Deconvoluted Ti 2p spectra of Fig. 5(b): (a) sputtered surface,
P
(b) methanol dosed surface.
S P e(~X@j) dx
(1)
where, S \ XPS peak area, j \ escape depth and
X \ distance from the surface. This equation can be applied
to the results of angle variation measurements by replacing j
with jcos h, since the Sr/C data showed the cos h dependence
in Fig. 2(B). The analysis depth at the glancing angle (80¡) was
calculated from the escape depth at a normal angle. The
results are shown in Fig. 7 for three di†erent escape depths,
namely 20, 30 and 40 A. The analysis depth at the glancing
angle would range between 3.4È6.8 A for these escape depths.
As the distance between the (110) planes is 2.8 A in the ideal
crystal, the analysis depth would be ca. 2È3 layers from the
Table 5 Ti 2p deconvolution results for methanol adsorption
Ti2`
Ti3`
Ti4`
No.a
(%)
(%)
(%)
A.O.N.b
*xc
C/Tid
3-A
3-B
34.2
26.0
26.6
26.0
39.3
48.0
3.05
3.22
È
0.17
È
0.75
a No. corresponds to that of Table 2. b Average oxidation number,
calculated by the eqn. (3). c The change of A.O.N. after methanol dose.
d Atomic ratio of C to Ti based on C 1s and Ti 2p area (cross section
C: 1.0, Ti: 7.9).
Table 6 Ti 2p deconvolution results for various amount of methanol adsorption
No.a
Ti2` (%)
Ti3` (%)
Ti4` (%)
A.O.N.b
*xc
C/Tid
MeOH dosee/L
4-Af
4-Bf
32.4
29.9
26.5
25.8
41.2
44.3
3.09
3.14
È
0.05
È
0.49
È
6
5-A
5-B
41.2
34.9
25.4
27.4
33.3
37.7
2.92
3.03
È
0.11
È
0.78
È
6
6-A
6-B
39.5
33.3
27.7
26.3
32.8
40.4
2.93
3.07
È
0.14
È
0.77
È
50
7-A
7-B
36.4
26.7
27.1
27.8
36.4
45.6
3.00
3.19
È
0.19
È
0.88
È
500
a,b,c,d See captions of Table 5. e The dosed amount of methanol. f Symbol A: Ar` sputtered, B; methanol dosed.
Phys. Chem. Chem. Phys., 1999, 1, 913È920
917

View Article Online
Scheme 2
dered anatase. They proposed dialkyl species as the ethane
precursor formed at temperatures greater than 723 K as
shown in species (I) below. By analogy with the
organometallic compounds, in the present case a dialkyl
species or dinuclear species I(b), would be a possible precur-
sor, and this is formed at room temperature in these experi-
ments (lattice oxygen is not shown in species I).
Fig. 7 Simulated photoelectron decay curve based on eqn. (1), for
three escape depths: (a) 20, (b) 30, (c) 40 A; (a@), (b@) and (c@): simulation
for angle variation (h \ 80¡) for the three escape depths, showing the
increase of the Ðrst layer contribution to the total at glancing exit.
The binding energy (286.5 eV) of the C species is lower than
that of molecular methanol (287.1 eV, dosed and measured at
170 K), and higher than that of carbidic carbon at 285 eV. On
the other hand, the binding energy of the oxygen ““shoulderÏÏ
formed by room temperature adsorption was D2 eV lower
than that of methanol adsorbed at 170 K. This indicates that
the oxygen species are more basic than methanol oxygen
(assuming that at 170 K methanol is physisorbed). Therefore,
it would have an anionic character similar to the hydroxide or
methoxide ion. Thus, the C species is either a methyl or
methoxy species. These considerations support the above pre-
cursors (I). [Methyl and hydroxy groups can be replaced by
methoxy and hydrogen in species (I).] The fact that the
desorption spectra of ethane showed Ðrst-order features,
namely an asymmetric shape, would also lend some support
to this proposal.
surface. Therefore, by using eqn. (1) the contribution of the
Ðrst layer to the XPS peak was estimated as 25È45% on the
assumption that Ðrst layer depth is 2 A. It should be noted
that the actual surface would have a roughness, by Ar` bom-
bardment, that would a†ect the decay features of photoelec-
trons to some extent.
4
Discussion
The signiÐcant features of the reaction of methanol with the
reduced SrTiO (110) surface can be summarised as follows:
3
1. Methanol is adsorbed on the reduced SrTiO surface at
3
room temperature.
2. Ethane was formed during TPD (CwC bond formation).
3. A part of the reduced Ti was reoxidised by the adsorp-
tion process (oxidative-addition).
4. The desorption process resulted in the reduction of Ti
(reductive-elimination).
4.3 QuantiÐcation of the amount of desorbed ethane
By using eqn. (2) the amount of ethane desorbed was esti-
mated.
h \ (v/RT )A
(2)
where h \ surface coverage, v \ pumping capacity, R \ gas
constant, T \ temperature and A \ area of the TPD spec-
trum
4.1 Adsorption of methanol on Ti2‘
From the deconvolution results of Ti 2p XPS spectra, new
insights for the methanol adsorption process at room tem-
perature can be derived. A two-electron transfer from Ti2` to
methanol moieties has occurred during the adsorption
process. In general, Ti2` species are strong reducing agents,
which are known to be able reduce even water. The presence
of Ti2` is the result of preferential removal of oxygen by Ar`
sputtering and this also would decrease the coordination
number of Ti. Therefore the Ti2` ion, having both a low coor-
dination number and high reduction potential, would have a
strong affinity for the oxygen in the methanol molecule and
appears to reduce it during the adsorption, as shown by the
alkane products in TPD. This is discussed further in the next
section.
In the case of Fig. 4, the amount of desorbed ethane was
estimated as 2.5 ] 1013 molecules. On the other hand, the
number of Ti cations on the fully oxidised SrTiO (110) surface
3
is estimated at 1.6 ] 1014 atoms based on the assumption that
the surface has no cation defects. The number of Ti cations on
the sputtered surface was estimated to be 20% less than that
of the oxidised surface (Section 3.1). The ratio of ethane mol-
ecules desorbed to surface Ti atoms is then 0.20 on the
assumption that the top layer has an equivalent Ti concentra-
tion to the analysed region. The amount of methanol
desorbed as ethane therefore corresponds to 40% of the
number of surface Ti cations.
4.4 Average oxidation number (A.O.N.) of Ti
The A.O.N. can be deÐned in the following way. From the
results of the deconvolution of Ti 2p spectrum, the relative
ratio of each species was calculated based on the peak area in
Table 5. The sensitivity factor for each state was assumed as
equal.
4.2 CwC bond formation
CwC bond formation is
a
well-known reaction in
organometallic compound decomposition. It often proceeds
by an intramolecular reductive-elimination process as shown
in Scheme 2.
Average oxidation number of Tix`
Ethane is the coupling product of methanol-derived C
species. To the best of our knowledge, this is the Ðrst obser-
vation of ethane as a main TPD product from methanol on
single crystal metal oxides. Carrizosa et al.11 observed ethane
as one of the TPD products of methanol adsorption on pow-
x \ ; n ] C
(3)
i
i
i
where n is the oxidation number of species i (2È4) andC is the
i
i
relative concentration of species i.
918
Phys. Chem. Chem. Phys., 1999, 1, 913È920

View Article Online
On the other hand, there is another way to elucidate the
A.O.N. In the composition of the reduced strontium titanate,
as Ðtting results have shown, the Sr 3p spectral shapes were
almost unchanged throughout the treatment. Therefore we
can assume the oxidation number of strontium would be con-
stant, ]2. From the requirement of electrical neutrality, it is
also assumed that the same amount of oxygen played a part
in the compensation of the plus (]) charge of Sr. Thus, the
relative concentration of each component can be calculated,
based on the area of each component. Therefore, the average
oxidation state of Ti was also deÐned as follows. A.O.N. of
Tix`
was applied to the surface with adsorbed methanol. The
results are summarised in Tables 2 and 7. The di†erences of
the two deÐnitions range between 0.5 and 1.0 and this large
di†erence can not be explained by either the shielding e†ect of
the adsorbed species (hence the increase in the O/Ti ratio), or
by the di†erence of the escape depth for each element
(assuming that all the adsorption is accompanied by oxidation
of surface Ti). Therefore, it is suggested that there are other
type(s) of adsorption, which do not accompany the oxidation
of surface Ti. The adsorption would be either acidÈbase or
molecular type, probably on both Ti3` and Ti4` sites as
shown below [(II), acidÈbase type]. The possibility of adsorp-
tion on a Sr2` site cannot be excluded, but the fact that the
Sr/Ti ratio has increased about by 10% after the adsorption of
methanol may suggest the preferable adsorption on Ti sites.
x \ (C [ C ) ] n /C
(4)
where C is the concentration of species i, n is the absolute
O
Sr Ti
0
i
0
value of the oxidation number of oxygen.
Although the oxygen species in a reduced sample consists of
more than one species as was suggested by the O 1s shape, n
was assumed to be 2. Cross sections for each element were
0
decided based on the spectrum area for the nearly fully oxi-
dised sample measured at a 20¡ emission angle. The values
tabulated by ScoÐeld12 were modiÐed to Ðt the stoichiometry.
The results are summarised in Tables 2 and 7. The two deÐni-
tions for the A.O.N. of Ti were in good agreement, indicating
the average oxidation state of Ti ion on the reduced surface to
be around ]3.0.
4.7 The reaction pathway for desorption of ethane, methane,
CO and hydrogen
Another example of the TPD which was obtained in this work
is shown in Fig. 8. This is a typical example which was
observed at the beginning of the series of experiments. The
results of Fig. 4 were towards the end of the series of experi-
ments (four months later). In Fig. 8 hydrogen and methane
desorption were clearly observed around 400 K. Therefore it
4.5 Analysis of the scheme with the quantiÐed data
By combining the data discussed so far, the process for ethane
desorption can be proposed (semi)quantitatively. The amount
of ethane is 20% (40% methanol) of Ti atoms in the Ðrst layer
(Section 4.3). The Ðrst layer contribution to the total signal
would be in the range 25È45% (Section 3.5). Two views can be
drawn for the extremes of our estimate with an assumption
that Tix` species in the subsurface region do not change their
oxidation state by methanol adsorption. For the 45% case,
since about 8% of Ti2` is converted into Ti4` as a whole, the
amount of Ti2` reacted in the Ðrst layer is 18% by the above
assumption. In this case, 16% of Tix` ions in the Ðrst layer
still remains as Ti2`. The number of reacted Ti2` and ethane
evolved are then in good agreement. This would support the
precursor species to be that illustrated in species I(a) above.
On the other hand, for the 25% case, since reacted Ti2`
would be about 32%, almost all the Ti2` species on the
surface would have been converted to Ti4` species. In this
case, the precursor species would be I(b). For both cases,
ethane molecules will be desorbed as the decomposition
product from isolated metal centre(s).
4.6 The application of A.O.N. deÐnitions for methanol dosed
surfaces
Fig. 8 TPD spectrum following the adsorption of methanol on
reduced SrTiO (110) from early in the experimental period. Methane
3
Although the two deÐnitions of A.O.N. are in agreement for
the clean sputtered surface, the results di†ered greatly when it
is more prevalent and appears at a low temperature than ethane. The
numbers refer to the relative molecular mass of the species desorbed.
Table 7 Quantitative XPS peak analysis of the methanol dosed surfacea
A.O.N.
[(4) ] 2]
(5)
A.O.N.
(curve-Ðtting)
(6)
O/Ti
(1)
Sr/Ti
(2)
O/Sr
(3)
(1)È(2)
(4)
(6)È(5)
(7)
4-A
4-B
2.76
3.20
1.25
1.38
2.21
2.32
1.51
1.82
3.02
3.64
3.09
3.14
0.07
[0.50
5-A
5-B
2.70
3.17
1.19
1.19
2.27
2.66
1.51
1.98
3.02
3.96
2.92
3.03
[0.10
[0.93
6-A
6-B
2.54
3.23
1.16
1.27
2.19
2.54
1.38
1.96
2.76
3.92
2.93
3.07
0.17
[0.85
7-A
7-B
2.60
3.32
1.14
1.26
2.28
2.63
1.46
2.06
2.92
4.12
3.00
3.19
0.08
[0.93
a See captions of Tables 2 and 6.
Phys. Chem. Chem. Phys., 1999, 1, 913È920
919

View Article Online
can be deduced that methane is the product of the reaction
This suggests that the reactivity of the surface relates to the
between a CH moiety and hydrogen according to reaction
(III). It may be that the onset of hydrogen mobility dictates
degree of the reduction (No. 4, 5) and the reduced SrTiO (110)
was almost saturated with less than 50 L of methanol dosed.
3
3
the formation of methane.
The change of A.O.N. at low coverage (4-B) is smaller than
expected. This may imply that the adsorption at the early
stage was mainly the type shown in reaction (II) in Section 4.6.
CH (a) ] H(a) ] CH
(III)
(IV)
3
4
H(a) ] H(a) ] H
2
When the methane desorption was distinct, little CO and
hydrogen desorption at high temperature could be observed.
High temperature CO and hydrogen desorption would be the
sequential decomposition products of the CH moiety perhaps
because of a shortage of available hydrogen on the surface (or
5 Conclusion
The adsorption of methanol on SrTiO (110) has been studied
3
using TPD and XPS. On the highly reduced surface Ti2` was
3
present and ethane was a major product from methanol coup-
ling. As a result of the deconvolution of the Ti 2p XPS
spectra, it is suggested that two kinds of adsorption sites are
involved in methanol adsorption. Ti2` is ascribed to be the
oxidative adsorption (addition) site. On the other hand, Ti3`
or Ti4` is suggested to be active for acidÈbase adsorption of
methanol. As the reduction of Ti was observed after the
desorption, the whole process for ethane production can be
written as oxidative-addition and reductive-elimination. This
type of process is well-known in the Ðeld of organometallic
chemistry; however, this is the Ðrst clear example of this type
of reaction and shows the change in the oxidation number at
the metal centre involved. Oxygen removed from methanol is
incorporated into the surface as lattice oxygen, Ðlling
vacancies in the bulk structure.
in the subsurface) in reaction (III). The precursor species could
be methoxide on Ti or methyl on lattice O and be identical
with CH (a) in the above reaction scheme.
3
CH O ] CO ] H
(V)
3
2
Alternatively, it can be written as
CH (a) ] O(a) ] CO ] H
(VI)
3
2
The path for the production of ethane is likely to parallel that
of methane. The di†erence between them is that hydrogen is
required for methane production, but not for ethane (Figs. 4,
6). The reason for the change with time during the experimen-
tal period is most likely to be owing to slight changes in the
bulk reduction state of the sample, which in turn would a†ect
its semiconducting properties and its surface reactivity. Upon
removal from the system the sample was a blue colour, indica-
tive of bulk reduction, whereas initially it was pale opaque
yellow.
Acknowledgements
The authors are grateful for the Ðnancial support of Nippon
Shokubai and the University of Reading during the course of
this work.
4.8 Small dose of methanol
To examine the surface reactivity, the dosed amount was
varied. The results of XPS analyses are shown in Tables 6 and
7 and Fig. 9. Even for a few Langmuirs dose of methanol, its
adsorption was conÐrmed by changes in the XPS spectrum.
References
1
J. Young, J. Osborn, F. Jardine and G. Wilkinson, J. Chem. Soc.,
Chem. Commun., 1965, 131.
2
3
4
O. Roelen, German Pat. 849548, 1938.
B. C. Gates, Catalytic Chemistry, Wiley, New York, 1992.
See, for example, K. Pierce and M. Barteau, J. Phys. Chem., 1994,
98, 3882.
5
6
C. Sayers and N. Armstrong, Surf. Sci., 1978, 77, 301.
A. Carley, P. Chalker, J. Riviere and M. Roberts, J. Chem. Soc.,
Faraday T rans. 1, 1987, 83, 351.
7
8
9
L. Ramquist, K. Hamrin, G. Johansson, A. Fahlman and C. Nor-
dling, J. Phys. Chem. Solids, 1969, 30, 1835.
N. Beatham, A. Orchard and G. Thornton, J. Phys. Chem. Solids,
1981, 42, 1051.
H. Franzen, M. Umana, J. McCreary and R. Thorn, J. Solid State
Chem., 1976, 18, 363.
10 A. Procter and D. Hercules, Appl. Spectrosc., 1984, 38, 505.
11 I. Carrizosa, G. Munuera and S. Castanar, J. Catal., 1977, 49,
265.
12 J. H. ScoÐeld, J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129.
Fig. 9 Relationship between the increment (*x) of the average oxi-
dation number of Ti and that of C/Ti per unit change.
Paper 8/07150K
920
Phys. Chem. Chem. Phys., 1999, 1, 913È920