Single-source CVD routes to titanium phosphide

Article abstract of DOI:10.1039/b203613b

Treatment of TiCl₄ with two equivalents of L (L = PhPH₂, Ph₂PH, PPh₃, CyPH₂, Cy₂PH, PCy₃) resulted in the formation of [TiCl₄(L)₂]. Reaction of TiCl₄

Full text of DOI:10.1039/b203613b

View Article Online / Journal Homepage / Table of Contents for this issue
Single-source CVD routes to titanium phosphide
Christopher S. Blackman,a Claire J. Carmalt,*a Ivan P. Parkin,a Leonardo Apostolico,b
Kieran C. Molloy,b Andrew J. P. Whitec and David J. Williamsc
a Department of Chemistry, Christopher Ingold Laboratories, University College London,
20 Gordon Street, UK WC1H 0AJ. E-mail: c.j.carmalt@ucl.ac.uk
b Department of Chemistry, University of Bath, Claverton Down, UK BA2 7AY
c Department of Chemistry, Imperial College of Science, Technology and Medicine,
South Kensington, UK SW7 2AY
Received 12th April 2002, Accepted 7th May 2002
First published as an Advance Article on the web 28th May 2002
Treatment of TiCl4 with two equivalents of L (L = PhPH2, Ph2PH, PPh3, CyPH2, Cy2PH, PCy3) resulted in the
formation of [TiCl4(L)2]. Reaction of TiCl4 with a stoichiometric amount of 1,2-bis(diphenylphosphino)methane
(dppm), 1,2-bis(diphenylphosphino)ethane (dppe) and 1,2-bis(diphenylphosphino)propane (dppp) affords
[TiCl4(dppm)], [TiCl4(dppe)] and [TiCl4(dppp)], respectively. X-Ray crystal structures of [TiCl4(Cy2PH)2] (5) and
[TiCl4(dppe)] (9) have been determined. Low pressure chemical vapour deposition (LPCVD) studies of all the
compounds revealed that [TiCl4(L)2] (L = CyPH2, Cy2PH and PCy3) and [TiCl4(dppm)] can form titanium phosphide
thin-films on glass.
Introduction
Results and discussion
Synthesis and characterisation
The formation of titanium() phosphide (TiP) thin films has
received little attention, although this material possesses a
number of useful properties. It is a metallic conductor, refrac-
tory (dec. < 1580 ЊC), hard and shows good resistance to
oxidation at elevated temperatures.1 Furthermore, thin films of
early transition metal phosphides have been used as diffusion
barriers in semiconductor devices.2 Bulk TiP may be prepared
by direct elemental combination3 or via the solid state meta-
thesis reaction of TiI4 with Na3P. 4 Alternatively, thin films of
TiP have been prepared by chemical vapour deposition (CVD)
involving the gas phase reaction of TiCl4 and PCl3 under an
atmosphere of hydrogen and argon at 850–1050 ЊC.5 This route
requires high reaction temperatures which in turn limits the
choice of substrate used in the film deposition process. More
recently, we have reported the first dual-source APCVD
(atmospheric pressure CVD) route to TiP coatings on glass
from the reaction of TiCl4 and ButPH2 at 500–600 ЊC.6 Single-
source precursors, which contain pre-formed Ti–P bonds, could
offer several potential advantages over the dual-source routes,
including lower deposition temperatures and easier handling of
the precursors. However, to the best of our knowledge, only one
single-source precursor to titanium phosphide thin films has
been reported to date. Silver-coloured titanium phosphide
films, of composition TiP1.1 were grown on glass and silicon
substrates using the precursor, [TiCl4(CyPH2)2], within the
temperature range 350–600 ЊC.7
Herein we describe the synthesis and characterisation of a
range of complexes of the type [TiCl4(L)2] (L = PhPH2, Ph2PH,
PPh3, CyPH2, Cy2PH, PCy3) and [TiCl4(LЈ)] (LЈ = dppm,
dppe, dppp). Two of the complexes, namely [TiCl4(Cy2PH)2]
and [TiCl4(dppe)], have been structurally characterised. Low
pressure chemical vapour deposition experiments on the com-
plexes are described, enabling assessments to be made of the
potential of the compounds to act as precursors to titanium
phosphide thin films. We are particularly interested in investi-
gating what effect changing the phosphine ligand has on the
resulting film composition.
Treatment of TiCl4 with two equivalents of a phosphine, L,
resulted in the formation of [TiCl4(L)2] [L = PhPH2 (1), Ph2PH
(2), PPh3 (3), CyPH2 (4), Cy2PH (5); Cy = cyclohexyl, Ph =
phenyl] in high yield (> 66%). All the complexes (1–5) have been
characterised by 1H and 31P NMR, the results of which indicate
that a monomeric complex [TiCl4(L)2] with either cis-or trans-
phosphines has been isolated in all reactions (Scheme 1). To
Scheme 1 L = PhPH2 (1), Ph2PH (2), Ph3P (3), CyPH2 (4), Cy2PH (5),
Cy3P (6/7).
investigate whether the cis or trans isomer has been formed the
X-ray crystal structure of compound 5 has been determined.
A crystal structure determination showed 5 to be the trans
complex illustrated in Fig. 1. The molecule has crystallographic
Ci symmetry and the geometry at titanium is slightly distorted
octahedral having cis angles in the range 87.79(3) to 92.21(3)Њ.
The Ti–Cl and Ti–P distances are unexceptional (Table 1).
The shortest approach to the P–H hydrogen atom is an inter-
molecular contact of 3.34 Å to Cl(1). There are no other
intermolecular contacts of note.
Analytical data were also obtained for compounds 1–5. The
carbon and hydrogen analyses for (2) and (4) were satisfactory.
However, for compounds 1, 3 and 5, although the hydrogen
analyses were satisfactory, the carbon analyses were con-
sistently low by 1–3%. We presume that this observation is due
2702
J. Chem. Soc., Dalton Trans., 2002, 2702–2709
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b203613b

View Article Online
Table 1 Selected bond lengths (Å) and angles (Њ) for 5
Table 2 Selected bond lengths (Å) and angles (Њ) for 9
Ti–Cl(1)
Ti–P
2.2892(8)
2.6372(8)
Ti–Cl(2)
2.2837(8)
Ti–Cl(1)
Ti–Cl(3)
Ti–P(1)
2.265(2)
2.265(2)
2.660(2)
Ti–Cl(2)
Ti–Cl(4)
Ti–P(2)
2.296(2)
2.251(2)
2.640(2)
Cl(2)–Ti–Cl(1)
Cl(2)–Ti–P
Cl(1)–Ti–P
89.00(3)
90.75(3)
92.21(3)
106.2(2)
Cl(1)–Ti–Cl(2A)
Cl(2A)–Ti–P
Cl(1A)–Ti–P
91.00(3)
89.25(3)
87.79(3)
Cl(4)–Ti–Cl(3)
Cl(3)–Ti–Cl(1)
Cl(3)–Ti–Cl(2)
Cl(4)–Ti–P(2)
Cl(1)–Ti–P(2)
Cl(4)–Ti–P(1)
Cl(1)–Ti–P(1)
P(2)–Ti–P(1)
107.74(7)
96.57(7)
92.78(7)
163.12(6)
79.31(6)
86.92(6)
88.36(6)
76.66(5)
Cl(4)–Ti–Cl(1)
Cl(4)–Ti–Cl(2)
Cl(1)–Ti–Cl(2)
Cl(3)–Ti–P(2)
Cl(2)–Ti–P(2)
Cl(3)–Ti–P(1)
Cl(2)–Ti–P(1)
96.53(7)
95.44(7)
161.79(8)
89.05(6)
85.28(6)
163.79(7)
78.63(6)
C(6)–P–C(12)
Fig. 1 The molecular structure of 5.
Fig. 2 The molecular structure of 9 showing the folding of the axial
to the formation of either titanium phosphide with carbon
impurities or metal carbide during the thermal decomposition
stage of the microanalysis procedure. Attempts at improving
the microanalysis by using a combustion aid were unsuccessful.
In contrast to the reactions described above, treatment of
TiCl4 with two equivalents of PCy3 resulted in the isolation of
a cream crystalline material. Analytical and spectroscopic data
for these crystals are consistent with the formation of a mixture
of two complexes, namely [TiCl4(PCy3)2] (6) and [TiCl4(PCy3)]
(7). Evidence for the formation of 6 and 7 was apparent from
31P NMR where two signals were observed at ϩ20.01 and
ϩ80.06 ppm. This result is not surprising given the bulky nature
of the PCy3 ligand.
The related reactions between TiCl4 and a stoichiometric
amount of a chelating phosphine, LЈ, in toluene, resulted in the
isolation of [TiCl4(LЈ)] [where LЈ = dppm (8), dppe (9) and
dppp (10)]. Analytical and spectroscopic data for (8–10) were
consistent with the proposed formula (Scheme 2) and an X-ray
crystal structure was determined for 9.
chloride ligands over the plane of the chelate ring.
those seen in titanium and hafnium octahedral complexes with
1,2-bisphosphinoethane8 having the two axial chloride ligands
directed over the phosphine chelate ring–here the Cl(1)–Ti–
Cl(2) angle is 161.79(8)Њ. Other distortions include a folding by
0.39 Å of one of the phosphorus atoms [P(1)] out of the basal
Ti/Cl(3)/Cl(4)/P(2) plane (which is coplanar to within 0.02 Å),
an enlargement of the Cl(3)–Ti–Cl(4) angle to 107.74(7)Њ and a
ligand bite angle of 76.66(5)Њ. All of these parameters are com-
parable to those seen in related structures.8 The Ti–P and Ti–Cl
distances are again unexceptional (Table 2). The packing of the
molecules is dominated by aromatic ؒ ؒ ؒ aromatic edge-to-face
interactions with the edges of the C(20) and C(8) containing
phenyl rings being directed orthogonally into the faces of
the C(8) and C(14) containing rings respectively of neighbour-
ing molecules; the associated ring centroid ؒ ؒ ؒ ring centroid
separations are 5.05 and 5.13 Å.
The octahedral coordination geometry around the titanium
metal atom found in complexes 5 and 9 is unexceptional.
Titanium tetrachloride readily forms adducts of the type
TiCl4Ln with donor ligands.9 The coordination numbers are
generally five or six but in a few rare cases eight coordination
has been found.10 Reaction of C2H4(PMe3)2 with TiCl4 in the
presence of diethylzinc produced TiCl4[C2H4(PMe3)2]2 with a
distorted square antiprismatic geometry.10 Reid and co-workers
have thoroughly investigated the coordination geometry of
titanium(IV) halides with chelating phosphine and arsine
ligands.11 They noted that both six- and eight-coordinate com-
plexes could be obtained. Notably the eight-coordinate
complexes were formed with phosphines and arsines with rela-
tively small cone angles at the metal; o-C6H4(PMe3)2 and o-
C6H4(AsMe3)2 whereas the comparable six-coordinate species
were formed with the more sterically demanding Ph2PCH2PPh2
and o-C6H4(PPh3)2 species.11 Considering the comparatively
large size of PCy2H and dppe it is not surprising that
octahedral geometry is found for 5 and 9. A search of the
Cambridge structural data base revealed that 5 is the first
crystallographically characterised TiCl4(PRx)2 structure with-
out a chelating phosphine ligand. The trans arrangement found
for 5 does not indicate that all unidentate phosphines adopt this
Scheme
2
Ph2P(CH2)nPPh2: n = 1, 1,2-bis(diphenylphosphino)-
methane (dppm) (8); n = 2, 1,2-bis(diphenylphosphino)ethane (dppe)
(9); n = 3, 1,2-bis(diphenylphosphino)propane (dppp) (10).
In the solid state complex 9 is seen to have approximate C2
symmetry about an axis passing through the bisector of the
equatorial Cl(3)–Ti–Cl(4) and P(1)–Ti–P(2) angles (Fig. 2).
The geometry at titanium exhibits a distortion comparable to
J. Chem. Soc., Dalton Trans., 2002, 2702–2709
2703

View Article Online
geometry. Attempts at elucidating the geometries of the other
complexes by IR spectroscopy, considering Ti–Cl stretches
and P–H bands was difficult, primarily as the samples were
exceedingly air sensitive and partially decomposed during the
analysis. However preliminary IR analysis indicates that 4 has a
trans geometry (two or less Ti–Cl stretches).
In order to study the decomposition pathway of compounds
1–10 and to assess their suitability as CVD precursors, thermal
analyses were carried out. The TGA/DSC (thermal gravimetric
analysis/differential scanning calorimetry) of 4 (10 ЊC minϪ1
from 20 to 500 ЊC) shows a melt at 120 ЊC by DSC followed by
a decomposition with an onset temperature of 200 ЊC. The
decomposition of 4 is clean and shows a weight loss of 50%.
This behaviour indicates an incomplete decomposition to TiP
up to 500 ЊC. All the other complexes show similar behaviour
with a weight loss of between 50 and 60%.
Chemical vapour deposition studies
Low pressure chemical vapour deposition of 1–10 were investi-
gated.12 No films were deposited from compounds 1, 2, 3 or 10
because the precursor either decomposed before volatilisation,
or it carried through the hot zone without depositing a film.
From compounds 4, 5, 6/7, 8 and 9, two types of film were
deposited – as shown by an initial survey study of the film
formed on the curved surface of a hot-wall glass-tube and a
thorough depth-composition analysis of the films formed on
small glass plates (5 × 0.5 cm) placed inside of the glass tube.
The films produced by CVD varied in colour, as shown in
Table 3. The titanium phosphide films produced from 4 and
5 are resistant to attack by common solvents, for example
toluene, acetone and tetrahydrofuran. However, concentrated
nitric acid digested the film from 4 and partly digested the film
from 5, after four weeks immersion. Concentrated hydrochloric
acid completely digested the films after one week’s immersion.
All the films passed the Scotch tape test. The films were also
hard and could not be scratched with a brass stylus and in
some cases with a stainless steel scalpel. The silver and gold TiP
films formed from 4 and 5 were electrically conducting with
sheet resistance measurements of 2–7 Ω ᮀϪ1. The black films
were more electrically insulating. All of the films had contact
angles for the spread of a water droplet in the region of 48–90Њ.
These values are very close to that of uncoated glass of 70Њ and
indicate a hydrophobic surface.
The films produced on curved glass were analysed by Energy
Dispersive Analysis by X-rays (EDAX) and Scanning Electron
Microscopy (SEM), electron probe and Raman spectroscopy.
The EDAX data showed the presence of titanium and phos-
phorus in the films produced from compounds 4, 5, 6/7 and 8,
but only titanium (no P) in the film from 9. Significant break-
through of the excitation volume through the coating to the
underlying glass was observed in most cases and so accurate
quantitative analysis by EDAX was difficult. However, electron
probe measurements showed that the composition of the
titanium phosphide films deposited appeared to change from
Ti1.1P to Ti3P (Table 3, vide infra).10 The formation of Ti1.1
P
from compound 4 is in contrast with a previous report,
where TiP1.1 films were grown from 4 in the temperature range
350–600 ЊC.7
The Raman spectrum of the film grown on curved glass from
4 (Fig. 3) is very similar to that obtained for bulk TiP with major
bands occurring at 234 and 305 cmϪ1. The film grown from
compound 6/7 produced a different Raman spectrum with
broad peaks occurring at 432 and 631 cmϪ1 and a weak peak
at 151 cmϪ1. The film grown from 9 showed the formation of
Ti1.6P from EDAX measurements. Despite this, the Raman
spectrum of this film indicates that the anatase phase of TiO2
has been produced. It is possible that a TiP film was formed
during the CVD reaction and surface oxidation has occurred on
exposure to air, or that an oxide was formed during the CVD
2704
J. Chem. Soc., Dalton Trans., 2002, 2702–2709

View Article Online
Fig. 3 Raman spectrum of the film grown from compound 4.
process. By Scanning Electron Microscopy (SEM) the film
flat glass enabled a more detailed analysis by depth profiling
grown from 5 shows spherical particles of size 200 nm which are
aggregated together to form a film (Fig. 4). The Ti1.1P film
grown from compound 4 had similar spherical particles, indi-
cative of some degree of gas-phase nucleation (Fig. 5). Because
techniques.
Film analysis on flat glass; depth profiling studies
Reactions were conducted in the same way for the curved glass
samples with the exception that small pieces of glass were
inserted into the glass tube. Deposition was noted on both the
top and bottom surfaces of the glass substrates. Use of flat
substrates enabled more detailed analysis by XPS, Rutherford
backscattering and X-ray powder diffraction. Notably, the flat
surface depositions were some 50–100 ЊC lower in temperature
than the curved wall experiments for the same set oven temper-
ature of 550 ЊC.
Electron probe results on curved glass samples showed
almost stoichiometric TiP was formed from 4. Raman showed a
TiP spectrum. EDAX analysis of film deposited onto flat glass
showed a slightly titanium rich phase. The film was amorphous
by powder XRD. Rutherford backscattering spectra (RBS)
showed a surface layer of mixed products with a film of Ti1P1.05
underneath (Fig. 6). The surface layer was 140 nm thick and
Fig. 4 SEM of the film produced by the decomposition of 5.
Fig. 6 Rutherford backscattering spectrum of the surface layer of the
film formed from the CVD of 4.
Fig. 5 SEM of the film produced by the decomposition of 4.
had composition Ti1.0P0.5O0.5C0.5Cl2, the bulk of the film was
1700 nm thick of formula Ti1.00P1.05 with almost undetectable
levels of carbon and chlorine contamination. Film thickness
from RBS gave a growth rate during the CVD process of
ca. 400 nm hϪ1. The X-ray photoelectron spectra of the film
of the somewhat contradictory nature of the film analysis
on curved glass (e.g. TiO2 by Raman and Ti1.6P by EDAX) a
second set of depositions was conducted on flat glass using
the same CVD procedure. In this case formation of a film on
J. Chem. Soc., Dalton Trans., 2002, 2702–2709
2705

View Article Online
although, complete etching through to the substrate was not
achieved. Surface carbon contamination reduced over sub-
sequent etches to almost nothing. The surface layers show a Ti
2p3/2 binding energy of 458.8 eV, indicative of oxide or phos-
phate, although there is almost no phosphorus detected at the
surface suggesting an overlayer of TiO2. There was significant
oxygen contamination in the layers examined however the
second layer shown by RBS was not etched into. This is because
the etching-gun could only remove ca. 300 nm in a meaningful
timescale (8 h). The O 1s peak was extremely broad and
encompassed two peaks, one at 531 eV (oxide)12 and 532.8 eV
(phosphate).13 This peak remained through all the etched layers.
However, after etching the only visible P 2p peak was centred at
128.6 eV (TiP)14 and the only visible Ti 2p3/2 was at 454.4 eV
although the peak shape was poor. The Ti 2p3/2 and P 2p peaks
show evidence of oxide/phosphate and phosphate presence
respectively but these are extremely minor compared to the
phosphide peaks.
Fig. 7 X-Ray photoelectron spectroscopy depth-profile analysis of the
film formed from the CVD of 4.
Electron probe analysis of the film produced from 8 on
curved glass gave a composition of Ti3P. Raman analysis
showed a low quality TiO2 spectrum. EDAX analysis of the
film deposited on flat glass gave a stoichiometry of Ti2.1
P
from 4 was fully consistent with the RBS results. It can be
seen from the elemental composition by XPS that there is a
surface layer containing mostly titanium and oxygen (Fig. 7).
Sputtering reduces the carbon in this portion of the film and
phosphorus begins to be revealed. There is an abrupt junction
where the film composition changes to containing only
titanium and phosphorus. The atomic concentration for
titanium is slightly misleading due to problems in accurately
integrating the Ti 2p3/2 peaks. Analysis of the binding energies
showed that at the surface the Ti 2p3/2 binding energy was 459.0
eV (TiO2 is 458.7 eV)12 although the shift for TiPO4 is in an
almost identical location13), suggesting oxide or phosphate. The
O 1s peak is centred on 531.2 eV (TiO2 is 530.6 eV and PO43Ϫ is
although the amount of titanium and phosphorus was very
low with significant breakthrough to the underlying glass sub-
strate. EDAX also revealed that the film had considerable
amounts of carbon although the amount of chlorine present
was low (ca. 1%). This amount of carbon probably accounts for
the colour of the film. RBS analysis indicated a relatively thin
film (ca. 160 nm) with a fixed composition of TiP. X-ray photo-
electron spectroscopy showed no surface titanium or phos-
phorus peaks and showed only the presence of a carbon film.
Sputtering reduced the amount of carbon in the film but it
remained significant until the substrate was reached. Sputtering
also revealed a Ti 2p3/2 peak with a binding energy of 454.7 eV
(TiP)12 and a phosphorus peak at 128.8 eV (TiP).14 There was
little sign of contamination with oxygen in either oxide or
phosphate form. Combination of this data is probably best
interpreted by the formation of the TiP phase that was con-
taminated with TiO2 at the surface and by carbon throughout
the film.
3Ϫ
532.4 eV but for titanium PO4 it is difficult to differentiate
from oxide).13 The P 2p peak is invisible on the surface but
3Ϫ
sputtering reveals a small peak centred on 133.8 eV (PO4 is
133.9 eV)13 for the first few sputtered layers. This suggests that
the surface is TiO2 and the first few sputtered layers are mainly
TiO2 with a small amount of phosphate present (from atomic
concentrations). The phosphorous atomic concentration is
growing at around level 8 as a large P 2p peak centred on 128.8
eV is revealed (TiP is 128.6 eV).14 The Ti 2p3/2 peak through
these levels is broad and poorly shaped suggesting a mixture of
TiO2, TiPO4 and TiP. After the abrupt junction at level 10 only
a Ti 2p peak at 454.6 eV is visible (TiP 454.6 eV) and a P 2p
peak at 128.8 eV (TiP 128.6 eV). In the TiP layer there was
undetectable chlorine contamination and almost undetectable
carbon contamination.
Electron probe analysis of the film produced from vapour
phase experiments of 9 gave a composition indicative of a TiO2
film. The Raman spectrum obtained for this sample was that
of TiO2. EDAX analysis of the film on flat glass showed a
Ti1.6P stoichiometry with heavy carbon contamination (40%)
although only slight chlorine contamination (2%). XPS showed
surface Ti 2p3/2 binding energy of 458.8 eV (TiO2 is 458.7 eV
although TiPO4 is similar), a P 2p binding energy of 133.4 eV
3Ϫ
(PO4 is 133.9 eV) and an O 1s binding energy of 532.2 eV
3Ϫ
(PO4 is 532.4 eV). This would suggest a titanium phosphate
Winter has reported a similar LPCVD reaction for the
formation of TiP films from 4 on silicon and glass substrates.7
Notably the films obtained were X-ray amorphous. Analysis by
XPS and RBS revealed a composition of TiP1.1 with no chlorine
or carbon contamination. This compares very favourably with
the results that were obtained in this study.
terminated surface. The first etched layers reveal a Ti 2p3/2
binding energy principally centred on 454.8 eV (TiP 454.6 eV)
and a P 2p energy of 128.4 eV (TiP 128.8 eV). Further etching
rapidly reduced the P 2p peak and reveals a broad poorly
formed Ti 2p peak indicative of oxide. The O 1s peak observed
at this region is centred at 530.6 eV (TiO2 530.6 eV).
Electron probe results from the vapour phase experiment on
curved glass on 5 showed a stoichiometry of Ti1.7P. EDAX
analysis of the film on flat glass showed a slightly more phos-
phorus rich stoichiometry (Ti1.44P) with more carbon present
than detected in the film produced from CyPH2 4. Raman spec-
troscopy of 5 showed a characteristic TiP spectrum. The film
was amorphous by XRD. RBS analysis of the film showed two
distinct regions. The top ca. 500 nm was of variable com-
position but was consistently metal rich with significant oxide
and carbon contamination. A second region was ca. 800 nm
thick and was stoichiometric TiP. Oxide and carbon contamin-
ation were reduced in the second layer but not eliminated.
There was almost undetectable chlorine contamination in either
layer. XPS analysis was used to examine the film further,
Summary of the depth profile analysis
The depth profiling studies of 4, 5, 8 and 9 all reveal that the
samples show a change in composition with depth. The bulk
film in most cases is TiP. The top layers of the films contain the
most oxide. This is fully consistent with the films undergoing
some post reaction oxidation upon storage in air. If an ingress
of oxygen occurred during the CVD process then the oxygen
content would be expected to be uniform with depth through-
out the sample. The formation of the oxide coating on the
surface probably accounts for why these films are so resistant
to solvents and acids. Notably none or negligible amounts of
chlorine were detected in the films indicating that it is fully
2706
J. Chem. Soc., Dalton Trans., 2002, 2702–2709

View Article Online
removed during the deposition stage. This is relatively unusual
especially for the studies of 8 and 9 where no active hydrogen
is present on the phosphine ligands. Notably the amount of
carbon seen in some of the films was considerable. This could
be directly correlated with the nature of the phosphine ligand.
The higher the carbon content of the starting phosphine on the
adduct and the lower its number of active hydrogens the more
carbon was found in the film. Interestingly even in the highly
carbon contaminated films the titanium and phosphorous seem
to present as stoichiometric TiP from XPS measurements.
The seemingly contradictory film analysis from the different
techniques is easily reconciled. XPS interrogates only the first
ten or so atomic layers of the surface. Hence, by combining
XPS analysis with sputtering to remove ca. 20 nm thickness
segments from the surface individual layers can be examined,
much like looking at different layers in an onion. Thus XPS
revealed changes in composition with depth. The RBS func-
tions in a somewhat similar fashion. The EDX and electron
probe measurements rely on ejected X-rays for analysis. At
the accelerating voltages used in the experiments typical pene-
tration depths were 1–2 µm. Hence the EDAX and electron
probe measurements consistently gave titanium rich films in a
number of cases. This was because the total titanium count was
made up from the surface titanium oxide as well as the bulk
titanium phosphide phases. The phosphorus count was made
up primarily from the titanium phosphide although some
surface phosphate was also possible. Thus it is likely that only
TiP was formed as the Ti phase in these reactions and the
analysis that indicated Ti2P etc. on the curved glass samples was
misleading as to the true nature of the films.
siderably less complicated apparatus. However, the prior syn-
thesis and isolation of an air sensitive precursor are notable
problems with the single source route. Further the use of other
secondary or tertiary phosphines, although somewhat success-
ful, produces films with extensive carbon contamination. Thus
the strategy of employing primary phosphines in an extension
of this work to other metals would be the most promising
avenue to pursue.
The degree of oxidation noted at the surface of all of the
films appears to happen quite rapidly after exposure of the film
to air. However the processes seems to be self-limiting because
the oxide layer formed on the films from 4 and 5 did not seem to
progress further into the solid on storage in air for six months
as the optical properties were invariant with time and reanalysis
indicated an unchanged material.
Conclusions
A range of complexes of the type [TiCl4(L)2] (L = PhPH2,
Ph2PH, PPh3, CyPH2, Cy2PH, PCy3) and [TiCl4(LЈ)] (LЈ =
dppm, dppe or dppp) have been synthesised and characterised.
Two of the complexes, namely [TiCl4(Cy2PH)2] and [TiCl4-
(dppe)] have been structurally characterised. Vapour phase
thin-film studies indicate that [TiCl4(L)2] (L = CyPH2, Cy2PH
and PCy3) and [TiCl4(dppm)] are effective titanium phosphide
precursors whereas [TiCl4(L)2] (L = PhPH2, Ph2PH and PPh3)
and [TiCl4(dppp)] did not produce a film under the same con-
ditions. Interestingly, [TiCl4(dppe)] produced a film containing
only titanium (no P) under similar conditions. These results
suggest that the complexes with primary or secondary phos-
phines (CyPH2 and Cy2PH) are superior titanium phosphide
precursors.
All of the films formed on flat glass were amorphous by
X-rays. This is probably not too surprising as the surface looks
to have been oxidised in all cases. Further the temperature
of deposition on the flat glass plates was some 50 ЊC lower than
the walls of the glass that was analysed in the curved glass
experiments.
Experimental
General procedures
All manipulations were performed under a dry, oxygen-free
dinitrogen atmosphere using standard Schlenk techniques or in
a Mbraun Unilab glove box. All solvents were distilled from
appropriate drying agents prior to use (sodium/benzophenone
for toluene, THF and hexanes; CaH2 for CH2Cl2). All other
reagents were procured commercially from Aldrich and used
without further purification. Microanalytical data were
obtained at University College London (UCL).
Aspects of film analysis
We have shown previously that titanium dioxide surfaces
show photo-assisted hydrophillicity.15 This can be assessed by
determination of the contact angle of a water droplet on the
coated surface. In the titanium phosphide films formed in these
experiments the initial films gave contact angles in the range
48–80Њ, notably the contact angles did drop on photo-
irradiation by 5–10Њ. This is not as great as that expected for a
pure TiO2 that can drop by 60–70Њ but is still in line with a TiO2
surface. Notably the film from 6/7 had the most readily identi-
fied TiO2 surface from Raman analysis and the lowest contact
angle measurement.
The attempted CVD of 1, 2, 3 or 10, all of which contained a
phenyl substituent on the phosphorus atom, resulted in failure.
In most cases transport of the precursor through the hot zone
of the CVD experiment did occur. However, an oily material
was seen in the colder sections of the reaction tube past the
hot zone. It is thus possible that the temperatures employed in
these experiments and the residence time of the precursor in the
reaction tube is such that film growth does not occur. It is
tempting to speculate that in the Cy derivatives 4 and 5
cyclohexene is eliminated in the reaction pathway, indeed this
has been observed by Winter in a related reaction.7 In the
corresponding phenyl cases a similar pathway via a benzyne
intermediate would be expected to be less favoured.
Physical measurements
1H and 31P NMR spectra were recorded on a Brüker AMX400
spectrometer at UCL. The NMR spectra are referenced to
CDCl3, which was degassed and dried over molecular sieves
1
prior to use; H chemical shifts are reported relative to SiMe4
(0.00 ppm); 31P chemical shifts are reported relative to external
85% H3PO4. IR spectra were recorded on a Nicolet 205 instru-
ment. EDAX/SEM results were obtained on a JOEL 35-CF
instrument using ISIS software (Oxford Instruments). Raman
spectra were acquired on a Renishaw Raman System 1000 using
a helium–neon laser of wavelength 632.8 nm. The Raman sys-
tem was calibrated against the emission lines of neon. Electron
microprobe analyses were obtained on a Jeol 8600 instrument
and referenced against phosphorus and titanium standards.
TGA of the compounds were obtained from the Thermal
Methods Laboratory at Birkbeck College (ULIRS). Rutherford
back scattering was obtained in association with Professor
Colligon at the University of Salford. X-Ray powder diffrac-
tion patterns were measured on a Philips X-pert diffractometer
using unfiltered (CuKα1 λ = 1.5045 Å, Kα2 λ = 1.5443 Å) radi-
ation in the reflection mode using glancing angle incidence.
Samples were indexed using the Unit Cell program and com-
pared to database standards. SEM/EDAX was obtained on
a Hitachi S570 instrument using the KEVEX system. X-Ray
A dual source CVD route to TiP films on glass has been
reported from the reaction of TiCl4 and ButPH2.6 The product
was of high purity and equivalent in colour, composition and
physical properties to that observed here from the single source
CVD of 4. Notably in both cases a monoalkylphosphine was
employed in the CVD experiment. The single source route
has some appeal in that the reaction uses inexpensive and con-
J. Chem. Soc., Dalton Trans., 2002, 2702–2709
2707

View Article Online
photoelectron spectra were recorded with a VG ESCALAB
220i XL instrument using focussed (300 µm spot) mono-
chromatic Al-Kα radiation at a pass energy of 20 eV. Scans
were acquired with steps of 50 meV. A flood gun was used to
control charging and the binding energies were referenced to an
adventitious C 1s peak at 284.8 eV. Depth profile measurements
were obtained by using argon beam sputtering. Contact angle
measurements of selected glass samples were determined by
measuring the spread of a 7.5 µl droplet of water and applying
a suitable program. Electrical conductivity was determined by a
four-probe device.
Found C, 54.50; H, 4.26. 31P{1H} NMR (CDCl3): δ Ϫ14.60 (s),
JH–P = 10 Hz. 1H NMR (CDCl3): δ 3.98 (t, 2H, CH2), 7.34–7.66
(m, 20H, Ph).
[TiCl4(dppe)] (9). TiCl4 (2.5 cm3, 1.0 M solution in toluene),
dppe (0.99 g, 2.5 mmol), toluene (25cm3). An orange solid
resulted. Anal. Calc. For C26H24Cl4P2Ti: C, 53.10; H, 4.11.
Anal. Calc. For C26H24Cl4P2Ti(CH2Cl2)0.5: C, 50.47; H, 4.00.
Found C, 49.57; H, 4.06. 31P{1H} NMR (CDCl3): δ 20.80 (s).
1H NMR (CDCl3): δ 2.92–3.02 (m, 2H, CH2), 7.19–7.73 (m,
20H, Ph). Crystals suitable for X-ray crystallography were
obtained by solvent diffusion of hexanes into a concentrated
CH2Cl2 solution of 9 at room temperature over a number of
days.
Preparations
[TiCl4(PhPH2)2] (1). PhPH2 (0.25 cm3, 2.50 mmol) was added
to a solution of TiCl4 (1 cm3, 1.0 M solution in toluene) in
hexane (25 cm3). A yellow/orange solid began to precipitate
immediately. The precipitate was isolated by filtration, washed
with hexanes (3 × 10 cm3) and dried in vacuo (0.27 g, 66%).
Anal. Calc. For C12H14Cl4P2Ti: C, 35.16; H, 3.44. Found C,
[TiCl4(dppp)] (10). TiCl4 (2.5 cm3, 1.0 M solution in toluene),
dppp (1.02 g, 2.5 mmol), toluene (25 cm3). An orange solid
resulted. Anal. Calc. For C27H26Cl4P2Ti: C, 53.86; H, 4.35.
Found C, 52.75; H, 4.00. 31P{1H} NMR (CDCl3): δ Ϫ1.85 (s).
1H NMR (CDCl3): δ 2.16–2.23 (m, 2H, CH2), 2.68–2.71 (m, 4H,
CH2), 7.29–7.65 (m, 20H, Ph).
33.26; H, 3.46. 31P{1H} NMR (CDCl3): δ Ϫ72.35 (s), JH–P
308 Hz. H NMR (CDCl3): δ 4.96 (d, 4H, PhPH2), 7.37–7.51
(m, 10H, PhPH2).
=
1
Chemical vapour deposition experiments
A similar method to that described above for 1 was adopted
for the synthesis of all the other complexes. The quantities used
in each preparation are given below.
CVD experiments were conducted inside of a glass tube of
length 50 cm and internal diameter 1 cm. The tube was evacu-
ated to ca. 10Ϫ2 Torr. Small glass plates (5 cm × 0.5 cm) made
from (BDH standard) microscope slides were inserted inside
of the glass tube. The glass tube was placed inside of a tube
furnace. The central portion of the tube was heated to 550 ЊC
and the end containing the precursor heated to 100–150 ЊC. The
experiments were typically conducted for 4 h. A film was
formed in the central portion of the glass tube, both on the
walls of the tube and on the inserted microscope slides. After
the deposition, the glass was allowed to cool to room tem-
perature, the vacuum was turned off and the system exposed to
air. The walls of the glass tube were broken and the film formed
on the inside surface of the glass was examined, as was the film
that had formed on the microscope slides.
[TiCl4(Ph2PH)2] (2). TiCl4 (3 cm3, 1.0 M solution in toluene),
Ph2PH (1 cm3, 6.00 mmol), hexanes (40 cm3). An orange solid
resulted. Anal. Calc. For C24H22Cl4P2Ti: C, 51.24, H 3.91. Anal.
Calc. For C24H22Cl4P2Tiؒ(C7H8)0.5: C, 54.31; H, 4.31. Found C,
54.04; H, 3.72. 31P{1H} NMR (CDCl3): δ Ϫ5.39 (s), JH–P
330 Hz. H NMR (CDCl3): δ 5.57 (d, 2H, PhPH2), 7.28–7.56
(m, 20H, PhPH2).
=
1
[TiCl4(PPh3)2] (3). TiCl4 (5 cm3, 1.0 M solution in toluene),
Ph3P (2.62 g, 1.00 mmol), hexanes (50 cm3). A red solid resulted
(3.1 g, 87%). Anal. Calc. For C36H30Cl4P2Ti: C, 60.53; H, 4.23.
Found C, 57.68; H, 4.01. 31P{1H} NMR (CDCl3): δ 7.53 (s).
1H NMR (CDCl3): δ 7.36–7.71 (m, 20H, PhPH2).
X-Ray crystallography
[TiCl4(CyPH2)2] (4). TiCl4 (5 cm3, 1.0 M solution in toluene),
CyPH2 (1.3 cm3, 1.00 mmol), hexanes (40 cm3). A yellow solid
resulted (2.0 g, 96%). Anal. Calc. For C12H26Cl4P2Ti: C, 34.15;
H, 6.21; Cl, 33.65. Found C, 34.08; H, 5.87; Cl, 34.94. 31P{1H}
NMR (CDCl3): δ Ϫ39.71 (s), JH–P = 314 Hz. 1H NMR (CDCl3):
δ 4.07 (d, 4H, CyPH2), 1.19–2.18 (m, 22H, CyPH2). IR θ
Crystals of 5 and 9 were grown CH2Cl2–hexanes mixtures
at room temperature. X-Ray analysis used the SHELXTL PC
system for structural refinement.16
Crystal data for 5. C24H46P2Cl4Ti, M = 586.3, monoclinic,
P21/c (no. 14), a = 10.982(1), b = 10.169(1), c = 13.332(1) Å, β =
103.42(1)Њ, V = 1448.2(1) Å3, Z = 2 (Ci symmetry), Dc = 1.344 g
cmϪ3, µ(Cu-Kα) = 7.02 mmϪ1, T = 203 K, orange/red plates;
(Ti–Cl) 350 cmϪ1
.
[TiCl4(Cy2PH)2] (5). TiCl4 (2.5 cm3, 1.0 M solution in
toluene), Cy2PH (1 cm3, 5.00 mmol), hexanes (25 cm3). A pink
solid resulted (1.47 g, 100%). Anal. Calc. For C24H46Cl4P2Ti:
C, 49.17; H, 7.91. Found C, 48.80; H, 8.36. 31P{1H} NMR
2153 independent measured reflections, F 2 refinement, R1
=
0.048, wR2 = 0.124, 1899 independent observed absorption
corrected reflections [|Fo| > 4σ(|Fo|), 2θ ≤ 120Њ], 147 parameters.
CCDC reference number 185882.
1
(CDCl3): δ 20.33 (s), JH–P = 318 Hz. H NMR (CDCl3): δ 3.99
(d, 2H, Cy2PH ), 1.16–2.66 (m, 44H, Cy2PH). Crystals suitable
for X-ray crystallography were obtained by solvent diffusion
of hexanes into a concentrated CH2Cl2 solution of 5 at room
temperature over a number of days.
Crystal data for 9. C26H24P2Cl4Ti, M = 588.1, monoclinic,
P21/c (no. 14), a = 12.229(2), b = 14.560(2), c = 16.047(4) Å, β
= 111.25(1)Њ, V = 2662.8(7) Å3, Z = 4, Dc = 1.467 g cmϪ3
,
µ(Cu-Kα) = 7.66 mmϪ1, T = 183 K, orange platey needles; 3294
independent measured reflections, F 2 refinement, R1 = 0.055,
wR2 = 0.131, 2429 independent observed absorption corrected
reflections [|Fo| > 4σ(|Fo|), 2θ ≤ 114Њ], 251 parameters.
CCDC reference number 185883.
[TiCl4(PCy3)] (6)/[TiCl4(PCy3)2] (7). TiCl4 (1 cm3, 1.0 M solu-
tion in toluene), PCy3 (0.56 g, 2.00 mmol), hexanes (40 cm3). A
cream solid resulted (0.51 g, 68%, based on the formation of 7).
Anal. Calc. For C36H66Cl4P2Ti: C, 57.61; H, 8.86. Anal. Calc.
For C18H33Cl4P2Ti: C, 45.99; H, 7.07. Found C, 49.28; H, 7.63.
31P{1H} NMR (CDCl3): δ 80.06 (s), 20.01 (s). 1H NMR
(CDCl3): δ 1.18–2.57 (m, PCy3).
See http://www.rsc.org/suppdata/dt/b2/b203613b/ for crystal-
lographic data in CIF or other electronic format.
Acknowledgements
[TiCl4(dppm)] (8). TiCl4 (2.5 cm3, 1.0 M solution in toluene),
dppm (0.94 g, 2.5 mmol), toluene (25 cm3). An orange solid
resulted. Anal. Calc. For C25H22Cl4P2Ti: C, 52.30; H, 3.86.
The EPSRC is thanked for grant GR/M98623 (I. P. P. and
C. J. C.) and GR/M98630 (K. C. M.). C. J. C. is also grateful
to the Royal Society for a Dorothy Hodgkin fellowship and
2708
J. Chem. Soc., Dalton Trans., 2002, 2702–2709

View Article Online
8 F. A. Cotton, C. A. Murillo and M. A. Petrukhina, J. Organomet.
research grant. Pilkington Glass are thanked for the glass sub-
strates and Epichem Ltd for phosphines. Professor R. J. H.
Clark is thanked for help with the Raman spectra and the
EPSRC for grant GR/M82592 for purchase of the Raman
spectrometer.
Chem., 2000, 593, 1; F. A. Cotton, C. A. Murillo and M. A.
Petrukhina, J. Organomet. Chem., 1999, 573, 78; F. A. Cotton,
P. A. Kibala and W. A. Wojtczak, Acta Crystallogr., Sect. C, 1991,
47, 89.
9 B. Patel, W. Levason, G. Reid, V. A. Tolhurst and M. Webster,
J. Chem. Soc., Dalton Trans., 2000, 3001.
10 F. A. Cotton, J. H. Matonic, C. A. Murillo and M. A. Petrukhina,
Inorg. Chim. Acta, 1998, 267, 173.
11 R. Hart, W. Levason, B. Patel and G. Reid, Eur. J. Inorg. Chem.,
2001, 2927.
12 Auger and X-ray Photoelectron Spectroscopy (Practical Surface
Analysis), eds. D. Briggs and M. P. Seah, Wiley, Chichester, 1990,
vol. 1.
13 C. E. Myers, H. F. Franzen and J. W. Anderegg, Inorg. Chem., 1985,
24, 1822.
References
1 R. L. Ripley, J. Less Common Met., 1962, 4, 496; B. Nolang,
O. Eriksson and B. Johansson, J. Solid State Chem., 1990, 86, 300.
2 K. Komaki, Japanese Patent JP 02 248 079, 1990.
3 T. Lundstrom and P. O. Snell, Acta Chem. Scand., 1967, 21, 1343.
4 R. F. Jarvis, R. M. Jacubinas and R. B. Kaner, Inorg. Chem., 2000,
39, 3243.
5 S. Motojima, T. Wakamatsu and K. Sugiyama, J. Less Common
Met., 1981, 82, 379.
6 C. Blackman, C. J. Carmalt, S. A. O’Neill, I. P. Parkin, L. Apostilco
and K. C. Molloy, J. Mater. Chem., 2001, 11, 2408.
7 T. S. Lewkebandara, J. W. Proscia and C. H. Winter, Chem. Mater.,
1995, 7, 1053.
14 J. A. Rotole, K. Gaskell, A. Comte and P. M. A. Sherwood, J. Vac.
Sci. Technol. A., 2001, 19, 1176.
15 I. P. Parkin, S. A. O’Neill, A. Mills, N. Elliott and R. J. H. Clark,
J. Mater. Chem., 2002, submitted for publication.
16 SHELXTL PC, version 5.03, Siemens Analytical X-Ray
Instruments Inc., Madison, WI, 1994.
J. Chem. Soc., Dalton Trans., 2002, 2702–2709
2709