Combined far infrared RAIRS and XPS studies of TiCl4 adsorption and reaction on Mg films

Article abstract of DOI:10.1016/j.susc.2005.04.038

In recent years there has been an increase in interest in the study of model Ziegler-Natta catalysts used for the polymerisation of ethene and propene. Particular attention has focused on catalysts consisting of TiCl₄ on activated MgCl₂ accompanied by a co-catalyst, usually triethylaluminium (AlEt₃). As part of a wider project on the characterisation of model Ziegler-Natta catalysts we have investigated the interaction of TiCl₄ with metallic Mg films grown on a Au surface using X-ray photoelectron spectroscopy (XPS) and far infrared reflection absorption infrared spectroscopy. Somewhat surprisingly, the infrared spectra show little variation as a function of exposure to TiCl₄. A very broad asymmetric vibrational band grows in with maximum intensity at 382 cm ^-1. Three prominent low frequency shoulders are observed at approximately 360, 320, and 260 cm^-1. For monolayer coverages of Mg the main band at 382 cm^-1 is narrower, less asymmetric and accompanied by a prominent shoulder at 398 cm^-1, which increases with increasing exposure to TiCl₄. TiCl₄ exposure in the presence of 5 × 10^-8 Torr of ethyl benzoate results in a change in line shape with low frequency broadening and a small shift in the frequency of the band. These spectra are discussed in the light of the possible constituent species making up the surface layer.

Full text of DOI:10.1016/j.susc.2005.04.038

Surface Science 587 (2005) 78–87
www.elsevier.com/locate/susc
Combined far infrared RAIRS and XPS studies
of TiCl4 adsorption and reaction on Mg films
a
b
M.J. Pilling , A. Amieiro Fonseca b,1, M.J. Cousins b,2, K.C. Waugh ,
c
M. Surman , P. Gardner
a,
*
a
School of Chemical Engineering and Analytical Science, Faraday Building, The University of Manchester, Manchester M60 1QD, UK
b
School of Chemistry, The University of Manchester, Manchester M60 1QD, UK
c
CCLRC Daresbury Laboratory, Daresbury, Warrington WA4 4AD, UK
Available online 23 May 2005
Abstract
In recent years there has been an increase in interest in the study of model Ziegler-Natta catalysts used for the poly-
merisation of ethene and propene. Particular attention has focused on catalysts consisting of TiCl4 on activated MgCl2
accompanied by a co-catalyst, usually triethylaluminium (AlEt3). As part of a wider project on the characterisation of
model Ziegler-Natta catalysts we have investigated the interaction of TiCl4 with metallic Mg films grown on a Au sur-
face using X-ray photoelectron spectroscopy (XPS) and far infrared reflection absorption infrared spectroscopy. Some-
what surprisingly, the infrared spectra show little variation as a function of exposure to TiCl4. A very broad asymmetric
vibrational band grows in with maximum intensity at 382 cmÀ1. Three prominent low frequency shoulders are observed
at approximately 360, 320, and 260 cmÀ1. For monolayer coverages of Mg the main band at 382 cmÀ1 is narrower, less
asymmetric and accompanied by a prominent shoulder at 398 cmÀ1, which increases with increasing exposure to TiCl4.
TiCl4 exposure in the presence of 5 · 10À8 Torr of ethyl benzoate results in a change in line shape with low frequency
broadening and a small shift in the frequency of the band. These spectra are discussed in the light of the possible con-
stituent species making up the surface layer.
ꢀ 2005 Elsevier B.V. All rights reserved.
Keywords: Infrared spectroscopy; RAIRS; XPS; Adsorption; Thin films; Ziegler-Natta catalysts; Magnesium chloride; Titanium
chloride
*
Corresponding author. Tel.: +44 161 306 4463; fax: +44 161 2004430.
E-mail address: peter.gardner@manchester.ac.uk (P. Gardner).
1
Present address: Johnson Matthey Plc, Technology Centre, Blounts Court, Sonning Common, Reading RG4 9NH, UK.
Present address: Johnson Matthey Catalysts, P.O. Box 1, Belasis Avenue, Billingham TS23 1LB, UK.
2
0039-6028/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.04.038

M.J. Pilling et al. / Surface Science 587 (2005) 78–87
79
1. Introduction
reaction mechanism. In this study we utilise far
IR reflection absorption infrared spectroscopy
(RAIRS) to investigate a model Ziegler-Natta cat-
alyst system, and to determine the feasibility of the
technique in providing further molecular level
information. We have chosen to use a similar
model catalyst to that used by Somorjai [3] i.e. thin
films of magnesium deposited on an inert gold foil
substrate subsequently exposed to TiCl4. As an
extension to this study we have also chosen to
investigate the role of Lewis bases in the catalytic
process by incorporating ethyl benzoate into our
model catalyst.
Polypropylene is one of the most industrially
important polyolefins and its efficient production
requires a catalyst with both high activity and
good stereospecificity. The most successful catalyst
to date for this system has been the MgCl2 sup-
ported Ziegler-Natta catalyst. This consists of a
pre-catalyst composed of titanium tetrachloride
chemisorbed on defective magnesium chloride,
which is reduced and alkylated by an alkyl alumin-
ium co-catalyst, typically AlEt3. Lewis bases are
added to act as internal or external electron donors
[1] and enable good control of the stereoselectivity
of the active catalyst. This ensures that highly iso-
tactic polypropylene is produced.
2. Experimental
Despite the importance of these catalysts, the
catalytic process remains poorly understood on a
molecular level, and the reaction mechanism still
needs elucidating. Characterisation of the active
centres in the ÔrealÕ catalyst has been hindered by
the inherent complexity of the catalyst, low con-
centration of active sites and its sensitivity to water
and oxygen in the surrounding ambient. In order
to circumvent these problems, recently there have
been a number of surface science studies directed
at investigating the structure and properties of
the active centres using model catalysts under
highly controlled conditions.
Somorjai et al. [2,3] have investigated the sur-
face properties of model Ziegler-Natta catalysts
fabricated by exposing thin films of magnesium
on an inert gold foil substrate at 300 K to TiCl4
and AlEt3. Through this pioneering work valuable
information was provided on the electronic prop-
erties and composition of these model catalysts
including the Ti oxidation state distribution,
adsorption site distribution, and important corre-
lations between surface structure and poly-
propylene tacticity. However a molecular level
understanding of the catalytic process remains elu-
sive and furthermore there are currently conflicting
theories on the role of the Lewis bases in the sys-
tem [4,5].
The far infrared RAIRS measurements were
performed on the dedicated IR beamline 13.3 at
the SRS, Daresbury, the basic elements of which
have been described in detail elsewhere [6]. The re-
cently modified experimental arrangement consists
of a UHV chamber coupled with an independently
pumped fast entry lock/preparation chamber. The
main analysis chamber is equipped with an ion
gun for sample cleaning, an X-ray source and
hemispherical analyser for XPS, an RGA and an
electron flood gun for electron irradiation of the
sample. The chamber was also equipped with
two leak valves for gas admittance and a shuttered
evaporator to allow deposition of thin films of Mg
or MgCl2. The evaporator consists of a ceramic
tube charged with the deposition material, which
is heated to the desired temperature via a
resistively heated filament surrounding the
assembly.
The sample used in the experiments consists of
a 4 cm2 polycrystalline gold foil attached to a sam-
ple plate with aluminium strips. The sample could
be heated by electron bombardment to tempera-
tures greater than 1000 K and rapidly cooled to
below 100 K. Argon ion sputtering was used to
clean the sample and surface cleanness was
checked by XPS prior to each experiment. The
sample was not annealed following sputtering.
Deposition of magnesium occurred at normal
incidence to the gold surface at 300 K. The magne-
sium flux was controlled by adjustment of the
Surface sensitive vibrational spectroscopic tech-
niques could potentially provide further informa-
tion on the nature of the catalytically active sites
and the molecular components involved in the

80
M.J. Pilling et al. / Surface Science 587 (2005) 78–87
temperature of the cell and varying the time of
exposure. A shutter directly in front of the source
aperture allowed accurate and reproducible dosing
of magnesium.
3. Results
3.1. Exposure of Mg/Au films to TiCl4
The thickness of the magnesium layer on the
gold foil was calculated by monitoring the attenu-
ation of the Au4f7/2 photoelectron peak. Assuming
an inelastic mean free path of the Au4f7/2 electron
The far infrared RAIRS spectra obtained fol-
lowing the exposure of a 1 ML thick film of mag-
nesium on gold at 300 K to increasing amounts of
titanium tetrachloride are shown in Fig. 1. Even at
low exposure the bands are intense. The spectrum
obtained following 0.025 L TiCl4 shows a broad
˚
in magnesium to be 22 A [7], and the thickness of
˚
1 monolayer (ML) of magnesium to be 2.6 A [2] a
4 min dose at 370 K was calculated to be equi-
valent to a 3 ML film. XPS spectra of the grown
films revealed that the magnesium layers were
reproducible and there was little or no evidence
of oxygen contamination. However, once the
source had been calibrated, XPS spectra of the
grown films were not recorded prior to the RAIRS
experiment to minimise contamination of the
films.
absorption band centred at 382 cmÀ1
. With
increasing exposure the band increases in intensity
and two smaller broad features develop at 0.325 L
centred around 290 cmÀ1 and 464 cmÀ1. At higher
exposure the high frequency feature shifts from
101.2
The magnesium chloride source was calibrated
similarly to magnesium via the attenuation of the
Au4f7/2 photoelectron peak. For this calculation
we have assumed that the mean free path of the
Au4f7/2 photoelectron in magnesium chloride is
101.0
(i)
(ii)
100.8
(iii)
˚
similar to that in magnesium i.e. 22 A. Given that
MgCl2 grows layer by layer on polycrystalline gold
100.6
464 cm-1
˚
[8] and a monolayer is 5.9 A thick [9], we have cal-
(iv)
culated that evaporation for a duration of 100 s
corresponds to approximately 2 ML of magnesium
chloride.
100.4
(v)
100.2
TiCl4 and ethyl benzoate were introduced into
the chamber from the vapour phase above the
liquid, following purification by several freeze–
pump–thaw cycles. The leak valve used for dosing
TiCl4 was fitted with a doser facing the sample to
enhance the partial pressure at the surface while
minimising contamination of the UHV chamber.
Pressures quoted in this paper are those obtained
from the nude ion gauge positioned on the oppo-
site side of the chamber from the leak valves.
Therefore, doses quoted here have not been cor-
rected for the sample pressure enhancement due
to the presence of the doser or the ionisation cross
sections of the gases.
(vi)
100.0
450 cm-1
99.8
290 cm-1
99.6
99.4
382 cm-1
398 cm-1
99.2
200
300
400
Wavenumber / cm-1
500
600
The fabricated model catalysts were investi-
gated by far infrared RAIRS, and spectra were ob-
tained from the co-addition of 256 interferrograms
at 8 cmÀ1 resolution.
Fig. 1. RAIRS spectra of 1 ML Mg/Au at 300 K exposed to (i)
0.025 L, (ii) 0.075 L, (iii) 0.325 L, (iv) 5.8 L, (v) 30.8 L, and (vi)
125 L of TiCl4.

M.J. Pilling et al. / Surface Science 587 (2005) 78–87
81
464 cmÀ1 to 450 cmÀ1, and the main band at
382 cmÀ1 also starts to become asymmetric as a
weak shoulder grows in at 398 cmÀ1. With increas-
ing exposure up to 125 L the 382 cmÀ1 main band
and the shoulders at 290 cmÀ1 and 464 cmÀ1 satu-
rate but the high frequency feature at 398 cmÀ1
continues to grow relative to the main band.
XPS measurements performed on the sample
after RAIRS revealed the presence of magnesium,
titanium and chlorine on the surface. Peaks were
observed at 458.3 eV, 464.0 eV, corresponding to
the Ti2p3/2, and Ti2p1/2, levels (Fig. 2(a)) and at
1178.9 eV (K.E) due to the MgKLL Auger, Fig.
2(b). This peak can unambiguously be assigned
to the MgKLL of MgCl2 [10]. In addition, the ab-
sence of a peak at 1185.5 eV indicates that within
the detection limits, all the magnesium has been
converted to magnesium chloride through reaction
with TiCl4 [3]. Peaks at 271.6 eV and 200.6 eV (not
shown), corresponding to the Cl2s and Cl2p,
respectively, were also observed.
(>10 ML) was grown on the clean gold substrate.
Fig. 3 shows the RAIRS spectra obtained follow-
ing exposure of the thick magnesium film to
increasing amounts of titanium tetrachloride.
Upon exposure of the multilayer film to 50 L TiCl4
an intense band with a broad line shape appears at
382 cmÀ1. Examination of the band reveals that it
is actually composed of a series of bands and is ob-
served as several weak shoulders on the main band
centred at 273, 324, 360 and 466 cmÀ1. With
increasing exposure the band at 382 cmÀ1 intensi-
fies and the shoulders appear slightly more prom-
inent. The intensities of the bands saturate at
exposures greater than 200 L, but there is a fre-
quency shift of the bands at 382 cmÀ1 and
466 cmÀ1 to 378 cmÀ1 and 435 cmÀ1, respectively.
(Unfortunately due to S/N problems, low exposure
data on the thick Mg films was not obtained dur-
ing the beamtime.)
Assignment of the absorption bands observed
in the RAIRS data is problematic due to the com-
plexity of the spectra. The XPS spectra revealed
the presence of titanium and magnesium chloride
To investigate the effect of increasing the thick-
ness of the magnesium film, a thick multilayer
130000
Ti2p3/2
50000
1178.9 eV
45000
125000
458.3 eV
Mg KLL
40000
Ti2p1/2
120000
464.0 eV
35000
30000
25000
20000
15000
10000
5000
0
115000
110000
105000
100000
95000
90000
453
458
463
468
473
1160
1165
1170
1175
1180
1185
1190
a
Kinetic energy / eV
b
Binding energy / eV
Fig. 2. XPS spectra of 1 ML of Mg exposed to 125 L TiCl4 showing (a) the Ti2p and (b) MgKLL Auger regions.

82
M.J. Pilling et al. / Surface Science 587 (2005) 78–87
100.5
100
99.5
99
irradiation of the sample with 1 KeV electrons in
a 5 · 10À7 Torr background pressure of TiCl4.
Detailed AES and angular resolved XPS charac-
terisation of these films revealed them to consist
of 3–4 ML of TiCl2 with 1 ML of TiCl4 chemi-
sorbed on the TiCl2 surface. To facilitate interpre-
tation of our RAIRS spectra we have attempted to
reproduce this system to enable us to obtain a
vibrational spectrum for TiCl2 and TiCl4.
Fig. 4 shows the RAIRS spectra obtained for the
gold foil irradiated with 1 KeV electrons in a 5 ·
10À7 Torr background pressure of TiCl4. Following
50 L exposure of TiCl4 two weak bands appear in
the spectrum at 398 cmÀ1 and 470 cmÀ1. Increased
(i)
98.5
98
(ii)
(iii)
101.8
101.6
97.5
97
(i)
(iv)
(v)
101.4
(vi)
(ii)
101.2
378 cm-1
(iii)
96.5
101.0
200
300
400
Wavenumber / cm-1
500
600
(iv)
100.8
Fig. 3. RAIRS spectra of thick multilayer Mg/Au at 300 K
exposed to (i) 50 L, (ii) 100 L, (iii) 150 L, (iv) 200 L, (v) 250 L,
and (vi) 300 L TiCl4.
(v)
100.6
on the sample surface, suggesting that in agree-
ment with SomorjaiÕs work on Mg films [2], TiCl4
is reacting with Mg to form magnesium chloride
and reduced titanium chloride. A good starting
point for the analysis would be to determine if
the bands we observe in the RAIRS spectra are
consistent with the presence of magnesium chlo-
ride and/or reduced titanium chloride. Unfortu-
nately there is little vibrational data on MgClx
and TiClx in the research literature.
100.4
470 cm-1
100.2
100.0
99.8
398 cm-1
99.6
200
300
400
500
600
Wavenumber/ cm-1
3.2. Electron irradiation induced titanium chloride
growth on gold at 300 K
Fig. 4. RAIRS spectra of electron irradiation induced titanium
chloride growth on Au at 300 K. Exposures of (i) 50 L, (ii)
100 L, (iii) 200 L, (iv) 300 L, and (v) 400 L with 1 KeV electron
energy.
Magni [11] reported that ultrathin films of tita-
nium chloride could be grown on gold at 300 K by

M.J. Pilling et al. / Surface Science 587 (2005) 78–87
83
exposure causes no change in the band at 470 cmÀ1
,
but the 398 cmÀ1 band continues to grow in inten-
sity until it finally saturates at 400 L. Since the rel-
ative intensities of the bands at 398 cmÀ1 and
470 cmÀ1 change as a function of exposure it is rea-
sonable to assign them to two different types of spe-
cies. AES data from Magni [11] reveals that for
similar conditions to those employed in our system,
the atomic ratio of Cl:Ti is close to 4 at the start of
growth but decreases as the films grow thicker. This
would be in agreement with the major constituent in
the film being TiCl4 at the start of the growth pro-
cess, but with the proportion of TiCl2 increasing
as growth proceeds. Given that the RAIRS band
at 470 cmÀ1 saturates at low exposure and the
398 cmÀ1 band continues to grow, this would
suggest that the two bands in the RAIRS spectra
could potentially be due to TiCl4 and TiCl2,
respectively.
100.3
100.1
99.9
99.7
99.5
99.3
99.1
98.9
98.7
In an attempt to determine if the 470 cmÀ1 band
is due to TiCl4 we have exposed the gold foil at
100 K to increasing amounts of TiCl4 to form a
physisorbed multilayer. The RAIRS spectrum ob-
tained (Fig. 5) gives rise to a single absorption band
520 cm-1
due to the m3 mode of molecular TiCl4 at 520 cmÀ1
.
The observation of a band at 470 cmÀ1 in Fig. 3
would be consistent with the m3 mode of TiCl4,
shifted down from the multilayer frequency due
to its interaction with the surface. Hastie [12]
reported that TiCl2 in a nitrogen matrix has an
infrared active absorption mode at 434 cmÀ1. It is
therefore possible that the band observed in Fig.
3 at 398 cmÀ1 is due to TiCl2 which is shifted down
in frequency from the matrix isolated value. On the
basis of this information, and in conjunction with
the electron spectroscopy findings of Somorjai for
this system, we assign the band at 398 cmÀ1 to
TiCl2.
200
300
400
500
600
Wavenumber / cm-1
Fig. 5. RAIRS spectrum following exposure of gold at 100 K
to 8 L TiCl4.
a flux of MgCl2 normal to the surface while contin-
uously collecting RAIRS spectra. Fig. 6 shows the
RAIRS spectrum obtained for films of MgCl2 of
increasing thickness deposited on the gold foil
at 300 K. For a film of nominally 2 ML MgCl2
a band is observed at 389 cmÀ1 with what
appears to be a weak shoulder at approximately
408 cmÀ1. There is also a broad band of low inten-
sity at this same coverage, at approximately
260 cmÀ1 and a weak shoulder at 245 cmÀ1. With
increasing film thickness the band at 260 cmÀ1
and its shoulder intensify. However the band at
260 cmÀ1 gradually shifts to 282 cmÀ1 while the
shoulder at 245 cmÀ1 appears to stay at the same
frequency. The band at 389 cmÀ1 and its shoulder
at 408 cmÀ1 also intensify with increasing thick-
ness of the film but the relative intensity of these
3.3. Multilayer films of magnesium chloride
To further aid interpretation of the RAIRS
spectra obtained for the TiCl4/Mg/Au system
we have also used RAIRS to obtain a vibra-
tional spectrum for multilayers of MgCl2 depos-
ited on the gold substrate. Magnesium chloride
was deposited by heating the source to evapora-
tion temperature, and opening the shutter to allow

84
M.J. Pilling et al. / Surface Science 587 (2005) 78–87
102.0
101.0
preliminary investigation we look at the effect of
co-dosing ethyl benzoate and titanium tetrachlo-
ride on a 2.5 ML thick magnesium film on the gold
foil at room temperature. The sample was exposed
to a dynamic pressure of 5 · 10À8 Torr of ethyl
benzoate followed by a variable dynamic pressure
of TiCl4, and RAIRS spectra were consecutively
acquired with a collection duration of 100 s each
(Fig. 7). The total exposure of the sample was
nominally 20 L of ethyl benzoate and 15 L TiCl4.
The first spectrum is recorded in a background
pressure of ethyl benzoate only, and no absorption
bands are observed. The spectrum obtained in the
background of ethyl benzoate and a constant
pressure of 5 · 10À8 Torr TiCl4 shows a strong
101.0
99.0
98.0
97.0
96.0
(i)
(ii)
(iii)
(iv)
282 cm-1
100.4
(v)
(v)
100.2
407 cm-1
(vi)
100.0
389 cm-1
(i)
200
300
400
500
600
Wavenumber / cm-1
99.8
Fig. 6. RAIRS spectra of multilayers of MgCl2 grown on gold
at 300 K. Coverages of MgCl2 are (i) 2, (ii) 4, (iii) 6, (iv) 8, (v)
10, and (vi) 12 ML.
99.6
(ii)
(iii)
two features remains the same. The positions of
these bands are consistent with transmission far
infrared spectroscopy performed on a-MgCl2 and
b-MgCl2 [13] where absorption bands have been
99.4
(iv)
observed at 407 cmÀ1
276 cmÀ1 and 248 cmÀ1
,
372 cmÀ1
,
329 cmÀ1
,
99.2
99.0
.
3.4. Influence of electron donors on the model
catalyst
200
300
400
500
600
Wavenumber / cm-1
Our study of the model catalyst was extended to
investigate the influence of electron donors and
therefore form a better approximation to the real
industrial catalyst. Ethyl benzoate was chosen as
the electron donor because of its general use in
the catalyst for polypropylene production. For this
Fig. 7. RAIRS spectra of 2.5 ML of Mg on Au, at 300 K
exposed to a background pressure of 5 · 10À8 Torr of ethyl
benzoate and a dynamic pressure of (i) 0 Torr, (ii) 5 · 10À8 Torr
and (iii) 1.5 · 10À7 Torr of TiCl4. Spectrum (iv) is as for (iii) but
after both leak valves have been closed and gases have been
pumped away. For comparison, spectrum (v) is of 2.5 ML of
Mg on Au, at 300 K exposed to 5 L of TiCl4 only.

M.J. Pilling et al. / Surface Science 587 (2005) 78–87
85
absorption band at 387 cmÀ1 and contains a very
broad shoulder centred around 290 cmÀ1. With in-
creased pressure of TiCl4 (1.5 · 10À7 Torr) there is
a slight increase in intensity of the absorption band
at 387 cmÀ1 and the shoulder at 290 cmÀ1. How-
ever there is no increase in absorption band inten-
sity below 260 cmÀ1. The spectrum obtained after
stopping admittance of TiCl4 and ethyl benzoate
into the chamber is identical to the previous spec-
trum indicating that the film is stable without the
dynamic pressure present.
Included in Fig. 7 is a RAIRS spectrum of
2.5 ML film of magnesium on gold exposed to
approximately 5 L of TiCl4 only. Calculation of
the exposure of the sample with respect to TiCl4
for the spectrum recorded for a dynamic pressure
of 5 · 10À8 Torr titanium tetrachloride in the
background of ethyl benzoate reveals it had been
exposed to approximately 5 L. Comparison of
these two spectra allows us to investigate the effect
of the ethyl benzoate on the vibrational spectrum.
The spectrum obtained in the presence of ethyl
benzoate and TiCl4 is considerably broader with
significantly more intensity below 350 cmÀ1. Also
noticeable is that the most intense band in the
spectra obtained with dynamic pressures of TiCl4
and ethyl benzoate occurs at 387 cmÀ1 compared
to 382 cmÀ1 for the spectra obtained for Mg/Au
exposed to TiCl4 only. It is apparent that, in our
model catalyst, ethyl benzoate is influencing the
chemistry occurring on the surface, and this is
borne out by the XPS spectrum (Fig. 8) recorded
under the same conditions. The photoelectron
peaks in the XPS spectrum at 286.6 and 532 eV
are assigned to the C1s and O1s demonstrating
that the ethyl benzoate has reacted at our model
catalyst surface. The XPS spectrum is very similar
to that recorded by Mori [14] where MgCl2 was
ground in a ball grinder before reaction with
TiCl4.
12000
O1s
4. Discussion
C1s
MgKLL
10000
Assignment of the bands in the RAIRS spectra
obtained following exposure of magnesium films
to TiCl4 is problematic. The spectra consist of
broad features and there is considerable overlap
between the expected vibrational frequencies of
MgCl2 and TiClx. Interpretation of the infrared
absorption bands observed for exposing the mono-
layer magnesium films to TiCl4 is aided by com-
parison to the RAIRS spectra of TiClx (Fig. 4)
and MgCl2 (Fig. 6) on gold. The band observed
for low exposure of TiCl4 at 382 cmÀ1 is in a sim-
ilar position to the most intense band in the MgCl2
spectrum at 389 cmÀ1 suggesting that it could
potentially be due to the presence of magnesium
chloride on the surface. At larger exposure of
Ti2p
Cl2p
8000
Cl2s
Au4f
6000
4000
2000
0
TiCl4 a second band is observed at 290 cmÀ1
,
and this would also be consistent with MgCl2 for-
mation since it occurred in the multilayer MgCl2
films at 282 cmÀ1. The absorption band at
464 cmÀ1 can be assigned by comparison to the
spectrum we obtained for electron induced tita-
nium chloride growth on gold (Fig. 4). On this sys-
tem a weak band was observed at 474 cmÀ1 which
75
175
275
375
475
575
Binding Energy /eV
Fig. 8. XPS spectrum of 2.5 ML Mg/Au exposed to a dynamic
pressure of 5 · 10À8 Torr ethyl benzoate and 15 L TiCl4.

86
M.J. Pilling et al. / Surface Science 587 (2005) 78–87
saturated at low exposures of TiCl4. We assigned
this band to TiCl4 interacting strongly with the
surface. Since the band at 464 cmÀ1 is in close
proximity to this, it is likely that it is due to
TiCl4—either within or at the surface of the film.
It is difficult to determine the origin of the shift
with increased exposure of TiCl4 in the 464 cmÀ1
band to 450 cmÀ1. Somorjai [3] used XPS to dem-
onstrate that exposure of thin magnesium films on
gold to TiCl4 resulted in titanium oxidation states
of 4, 3, 2 and 0. Since no magnesium chloride
bands occur in this region of the infrared spectrum
we could speculate that the shift in the band could
potentially be due to the formation of increasing
amounts of TiCl3 although we currently have no
evidence for this.
and the RAIRS bands at 382 cmÀ1 and
398 cmÀ1, it is unlikely that either of these are
due to TiCl4 or TiCl3. However we cannot rule
out the possibility of these two bands being due
to TiCl2 in different molecular environments.
The spectra obtained from exposing a thick film
of magnesium to increasing amounts of TiCl4
showed bands in similar positions to those in the
1 ML Mg/Au system. Bands were observed at
273 cmÀ1, 382 cmÀ1 which is possibly due to
MgCl2 and 466 cmÀ1 which we assign to TiCl4.
We cannot be certain of the phenomenon respon-
sible for the shifting of the TiCl4 band from
466 cmÀ1 to 435 cmÀ1. However it is possible that
the shift is due to the same effect observed for the
monolayer Mg/Au system where the band was ob-
There is some uncertainty into the origin of the
band at 398 cmÀ1 observed following high expo-
sure of monolayer Mg/Au to TiCl4. Multilayer
MgCl2 has an intense band at 389 cmÀ1 with a
weaker shoulder at 408 cmÀ1, and they both grow
at the same rate as the MgCl2 film thickness in-
creases. The 408 cmÀ1 feature shifted down in fre-
quency could potentially be the origin of the band
we observe at 398 cmÀ1. The other possibility
comes from the spectrum obtained for the electron
induced growth of titanium chloride (Fig. 4),
which featured a strong band at 398 cmÀ1 which
we assigned to TiCl2. Given that this absorption
frequency exactly matches that observed in the
spectrum the band could potentially be TiCl2. Fur-
thermore evidence for this is provided by the band
at 398 cmÀ1 growing in at a different rate to the
bands we assigned to MgCl2. Different rates of
band growth are expected to occur if they are
due to different types of species.
served to shift from 464 cmÀ1 to 450 cmÀ1
.
In the monolayer Mg/Au system exposed to
TiCl4 we observed a band at 398 cmÀ1 and as-
signed it to TiCl2. For the thick films of magne-
sium on gold exposed to TiCl4 it is difficult to
determine if there is a band at this position due
to the strong absorption band from MgCl2 at
382 cmÀ1. Interestingly a band was observed at
324 cmÀ1 for exposure of the multilayer magne-
sium film to TiCl4, but this band was not observed
in the multilayer spectrum of MgCl2 on gold or for
the monolayer magnesium film. However this
band has been observed for far infrared transmis-
sion spectra of solid MgCl2 at 329 cmÀ1 [13]. It is
therefore possible that the MgCl2 band only oc-
curs for considerably thick films.
The preliminary RAIRS investigation of the
influence of the ethyl benzoate on the model cata-
lyst produced some interesting results. Perhaps the
most striking observation was the increased
absorption intensity at 250–350 cmÀ1. TiClx vibra-
tional modes generally occur above 350 cmÀ1 so it
is unlikely that a titanium-chlorine species is
responsible for the increased intensity in this
region of the spectrum. However a magnesium-
chlorine related species would be expected to have
vibrational modes in the 250–350 cmÀ1 region. In
addition to the increased absorption intensity, a
shift in vibrational frequency to 387 cmÀ1 from
382 cmÀ1 (which was assigned to MgCl2). If the
ethyl benzoate is causing a shift in the vibrational
frequency of magnesium chloride then this would
Another possibility is that the absorption bands
at 382 cmÀ1 and 398 cmÀ1 are due to different
forms of TiClx. Somorjai revealed that exposure
of thin magnesium films on polycrystalline gold
produced TiClx with x = 0, 2, 3 and 4. Studies of
the infrared active modes of TiClx clusters in an
argon matrix [12] revealed that TiCl4, TiCl3, and
TiCl2 have absorption bands at 503 cmÀ1
,
497.1 cmÀ1, and 458.2 cmÀ1, respectively, while
TiCl2 in a nitrogen matrix has a band at
434.5 cmÀ1. Therefore given the difference between
the infrared absorption frequency in the clusters

M.J. Pilling et al. / Surface Science 587 (2005) 78–87
87
be interesting. There has recently been confusion
Acknowledgements
over the role of the ethyl benzoate and whether
it coordinates with only magnesium chloride, only
on a vacancy on the titanium compound, or both.
The XPS spectrum obtained for the ethyl benzoate
modified model catalyst provided evidence of car-
bon and oxygen on the surface. However due to
the low resolution of the spectrum it is impossible
to determine if there has been any chemical shifts
in the titanium or magnesium peaks. Further
work is required to attempt to understand the
influence of the ethyl benzoate on the model
catalyst.
We would like to acknowledge financial support
from EPSRC; Dr. Alan Brisdon, Ian Crossley and
Jackie Greenhall for their assistance with prepara-
tion of the TiCl4 samples and Terry Lee for his
assistance at Daresbury.
References
[1] P.C. Barbe, G. Cecchin, L. Noristi, Adv. Polym. Sci. 81
(1987) 1.
[2] S.H. Kim, E. Magni, G.A. Somorjai, J. Phys. Chem. B 104
(2000) 5519.
[3] E. Magni, G.A. Somorjai, J. Phys. Chem. B 102 (1998)
8788.
5. Conclusions
[4] K. Soga, T. Shiono, Y. Doi, Makromol. Chem. 189 (1988)
1531.
We have utilised far infrared RAIRS to investi-
gate model catalysts fabricated from the reaction
of titanium chloride with thin films of magnesium
on gold. In this study we have demonstrated the
difficulties in using RAIRS to investigate complex
catalytic systems. However through an analysis of
the complex spectra obtained we have managed to
provisionally assign the spectral features to the
presence of MgCl2, TiCl4, TiCl2 and possibly TiCl3
in agreement with SomorjaiÕs studies [3]. The influ-
ence of ethyl benzoate on the model catalyst was
also investigated with RAIRS. Significant differ-
ences in the vibrational spectra were observed
due to the presence of ethyl benzoate, suggesting
that far infrared RAIRS is a highly sensitive tech-
nique for investigating the effect of electron donors
on these model catalysts.
[5] M.C. Sacchi, I. Tritto, P. Locatelli, Prog. Polym. Sci. 16
(1991) 331.
[6] D.A. Slater, P. Hollins, M.A. Chesters, J. Pritchard, D.H.
Martin, M. Surman, D.A. Shaw, I.H. Munro, Rev. Sci.
Instrum. 63 (1992) 1547.
[7] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal.
17 (1991) 911.
[8] E. Magni, G.A. Somorjai, Appl. Surf. Sci. 89 (1995) 187.
[9] D.H. Fairbrother, J.G. Roberts, G.A. Somorjai, Surf. Sci.
399 (1998) 109.
[10] E. Magni, T.I. Koranyi, G.A. Somorjai, Langmuir 16
(2000) 8113.
[11] E. Magni, G.A. Somorjai, J. Phys. Chem. 100 (1996)
14786.
[12] J.W. Hastie, R.H. Hauge, J.L. Margrave, High Temp. Sci.
3 (1971) 257.
[13] M.A. Einarsrud, H. Justnes, E. Rytter, H.A. Øye, Polyhe-
dron 6 (5) (1987) 975.
[14] H. Mori, K. Hasebe, M. Terano, J. Mol. Cat. A 140 (1999)
165.