Volatile imido-hydrazido compounds of the refractory metals niobium, tantalum, molybdenum, and tungsten

Article abstract of DOI:10.1021/ic062435e

Volatile 1,1-dimethyl-2-(trimethylsilyl)hydrazido(1-) complexes of niobium, tantalum, molybdenum, and tungsten have been synthesized and fully characterized for use as precursors in their chemical vapor deposition to metal nitrides. Different reaction pat

Full text of DOI:10.1021/ic062435e

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/materials | Journal of Materials Chemistry
Aerosol assisted chemical vapour deposition of MoO3 and MoO2 thin films
on glass from molybdenum polyoxometallate precursors; thermophoresis
and gas phase nanoparticle formation
Sobia Ashraf, Christopher S. Blackman, Geoffrey Hyett and Ivan P. Parkin*
Received 24th May 2006, Accepted 24th July 2006
First published as an Advance Article on the web 8th August 2006
DOI: 10.1039/b607335b
Aerosol assisted chemical vapour deposition (AACVD) of molybdenum polyoxometallates
dissolved in acetonitrile or water yielded adhesive thin films of molybdenum oxides on glass. At
substrate temperatures of 300–350 uC single phase MoO3 was obtained, from 350–500 uC mixed
phases of MoO3–MoO2 were formed and at 500–550 uC single phase MoO2 was observed. The
morphology of the as-deposited molybdenum oxide films was found to be dependent upon a
number of factors including the nature of the precursor used, the deposition temperature and
position of the film within the reactor. Needles, spheres, agglomerates and platelets formed
depending on the conditions. The films with a needle-like microstructure displayed enhanced
hydrophobicity to water droplets (125u contact angle). X-Ray diffraction showed that the MoO3
˚
films had typical cell constants of a = 3.96, b = 13.85, c = 3.69 A and the MoO2 films had typical
˚
cell constants of a = 5.62, b = 4.84, c = 5.56 A, b = 119.32u. The MoO2 films were readily
converted to MoO3 by annealing in air for 30 minutes at 600 uC. The MoO3 films functioned as
gas sensors showing a linear change in electrical resistance upon exposure to trace amounts of
ethanol vapour in air.
been used in aerosol-assisted depositions of thin films.10,11 In
Introduction
this paper we present the deposition, analysis and functional
properties of molybdenum oxide films from AACVD using
polyoxometallates, large charged clusters formed by the early
transition metals in high oxidation states. Polyoxometallates
are the antithesis of conventional CVD precursors, which
tend to be molecular, volatile non-ionic complexes. The main
advantage of polyoxometallates as CVD precursors is their
relatively low temperatures required for clean decomposition
to the corresponding metal oxide (300–500 uC). Furthermore
molybdenum polyoxometallates containing hetero-atoms, such
as phosphorus, vanadium, niobium and tungsten, are easy to
prepare, and provide a convenient route to incorporating a
secondary element into the molybdenum oxide films.12,13
We have shown previously in a communication that poly-
tungstates can be used as AACVD precursors for the
deposition of tungsten oxide films.11 Here we show that the
approach can be extended to include polyoxomolybdates and
hence demonstrate the versatility of polyoxometallates as
AACVD precursors.
Molybdenum oxide thin films have applications in a diverse
range of fields including electrochromic devices, catalysis and
as gas sensor components.1–5 The gas sensing properties arise
due to changes in electrical conductivity upon adsorption of
trace gases onto the semiconductor surface.1 Molybdenum
oxide thin films have been deposited from a variety of
techniques including sol–gel,2 spin-coating,3 electrochemical
deposition,4 spray pyrolysis,5 flash and thermal evaporation,6
physical vapour deposition (PVD),7 combustion CVD
(CCVD)1 and hot filament metal oxide deposition
(HFMOD).8 All of these methods involved the use of small
molecular precursors typically containing a single molybde-
num atom. Molybdenum oxide thin films have been deposited
from dual-source CVD reactions of molybdenum hexa-
carbonyl and oxygen and from single source precursors such
as molybdenum pentacarbonyl 1-methylbutylisonitrile.9 The
structure, phase and properties of the resulting films are
strongly dependent upon the deposition technique employed.
Aerosol assisted chemical vapour deposition (AACVD) is a
variant of the CVD process involving the use of liquid–gas
aerosols to transport soluble precursors to a heated substrate.
The method has traditionally been used when a conventional
atmospheric pressure CVD precursor proves involatile or
thermally unstable. However, by designing precursors specifi-
cally for AACVD, the restrictions of volatility and thermal
stability are eliminated, and new precursors and films can be
investigated. Ionic precursors such as sodium fluoride have
Experimental
The polyoxomolybdates [nBu4N]4[Mo8O26], [nBu4N]2[Mo2O7],
[nBu4N]4[Mo8O26] and [nBu4N]3[PMo12O40] were synthesised
according to the literature procedure12 and [NH4]6[Mo7O24]
was purchased from Aldrich and used without further
refinement. Nitrogen (99.99%) was obtained from BOC and
used as supplied. Coatings were deposited on 150 mm 6
45 mm 6 4 mm sheets of SiO2 coated float-glass (the SiO2 acts
a blocking layer preventing diffusion of ions from within
the glass into the deposited film), which was supplied by
Department of Chemistry, University College London, 20 Gordon Street,
London, UK WC1H OAJ
This journal is ß The Royal Society of Chemistry 2006
J. Mater. Chem., 2006, 16, 3575–3582 | 3575

View Article Online
Pilkington Glass Plc. The glass was cleaned prior to use by
washing with acetone and propan-2-ol and then dried in air.
The glass substrate was heated on a graphite block containing
a Whatman cartridge heater in a horizontal-bed cold-wall
AACVD reactor. The temperature of the graphite block was
monitored by Pt–Rh thermocouples. Measurements indicated
that temperature gradients of less than 25 uC at 500 uC were
noted across the glass substrates. All gas-handling lines were
made of stainless steel and were of 1/4 in internal diameter,
except for the inlet to the mixing chamber and the exhaust line
from the apparatus, which were 1/2 in i.d.
angle incidence beam (5u). Scanning electron microscopy
(SEM) was carried out on a field emission JEOL 6301F
instrument. Wavelength dispersive X-ray analysis (WDX) was
carried out on a Philips ESEM with an Oxford INCA system.
Samples were coated with gold to improve conductivity
and images were recorded digitally. X-Ray photoelectron
spectroscopy (XPS) measurements were carried out on a VG
ESALAB 220i XL instrument using focussed (300 mm spot)
monochromatic Al-ka X-ray 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 surface elemental carbon at 284.6 eV. Raman
spectra were acquired on a Renishaw Raman System 1000
using a helium–neon laser of wavelength 632.8 nm. The
Raman system was calibrated against the emission lines of
neon. Differential scanning calorimetry (DSC)/thermal gravi-
metric analysis (TGA) was performed using a NETZSCH1
STA 449 instrument from 20 uC to 550 uC at a heating rate
of 10 K min21, with the samples stored under nitrogen in
aluminium crucibles. Hardness scratch tests were conducted
with a stainless steel scalpel. Contact angle measurements for
water droplets were determined by measuring the spread of a
known volume of water on the surface.
General synthesis procedure
The polyoxometallate precursors (0.25 g) were dissolved in
50 ml of acetonitrile (except [NH4]6[Mo7O24] which was used
in water) and an aerosol was generated at room temperature
using a PIFCO ultrasonic humidifier. Nitrogen was passed
through the aerosol mist, forcing the aerosol droplets laden
with precursor into the reactor chamber (Table 1). The exhaust
from the reactor was vented directly into the extraction system
of a fume cupboard. Deposition experiments were conducted
by heating the horizontal bed reactor to the required
temperatures before diverting the nitrogen line through the
aerosol into the reactor. The gas flow was continued until all
the precursor mix had passed through the reactor, typically
30–60 minutes. Films were cooled to room temperature in situ
under a flow of nitrogen and after cooling were stored and
handled in air.
Gas sensor measurements
The sensor substrates were made of alumina tiles with
interdigitated gold electrodes on the top and a platinum strip
on the back, which serves as a microheater. The platinum
heater track of each sensor also formed one arm of a
Wheatstone bridge, which allowed the resistance, and hence
the heater temperature, to be both regulated and programmed.
The gas sensing setup comprised a stainless steel test
chamber to which gases (BOC special gases) were streamed
Film analysis
X-Ray diffraction (XRD) patterns were recorded using a
Bruker D8 discover reflection diffractometer, with Cu-Ka
˚
radiation (l = 1.5406 A) in reflection mode with a glancing
Table 1 Deposition temperatures, gas flow rates and phases seen by XRD and Raman for the AACVD of molybdenum polyoxometallate
precursors on glass. Phases separated by commas indicate that sections of the substrate contained separate phases, entries such as MoO2/MoO3
indicate that both phases coexisted in the same region of the substrate
Precursor
Substrate temp./uC
N2 Flow rate/l min21
XRD
Raman
[nBu4N]4[Mo8O26
[nBu4N]4[Mo8O26
[nBu4N]4[Mo8O26
[nBu4N]4[Mo8O26
[nBu4N]4[Mo8O26
[nBu4N]4[Mo8O26
[nBu4N]2[Mo2O7]
[nBu4N]2[Mo2O7]
[nBu4N]2[Mo2O7]
[nBu4N]2[Mo2O7]
[nBu4N]4[Mo6O19
[nBu4N]4[Mo6O19
[nBu4N]4[Mo6O19
[nBu4N]4[Mo6O19
[nBu4N]4[Mo6O19
[nBu4N]4[Mo6O19
]
]
]
]
]
]
550
500
450
400
350
300
500
450
400
350
550
500
450
400
350
300
450
400
550
500
450
450
400
350
1
MoO2
MoO2
MoO2
MoO2
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
2
2
1
1
2
1
1
1
MoO3, MoO2, MoO2/MoO3
MoO3, MoO2/MoO3
MoO3, MoO2/MoO3
MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO3
MoO2/MoO3
MoO2/MoO3
MoO2
MoO3, MoO2 MoO2/MoO3
MoO3
MoO3,MoO2/MoO3
MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO2/MoO3
MoO3
]
]
]
]
]
]
[NH4]6[Mo7O24
[NH4]6[Mo7O24
]
]
MoO2/MoO3
MoO2/MoO3
MoO3
[nBu4N]3[PMo12O40
[nBu4N]3[PMo12O40
[nBu4N]3[PMo12O40
[nBu4N]3[PMo12O40
[nBu4N]3[PMo12O40
[nBu4N]3[PMo12O40
]
]
]
]
]
]
MoO3/MoP2O7
MoO3/MoO2/MoP2O7
MoO3/MoP2O7
MoO3/MoP2O7
amorphous
MoO2
MoO2/MoO3
MoO2/MoO3
MoO3
MoO3
3576 | J. Mater. Chem., 2006, 16, 3575–3582
This journal is ß The Royal Society of Chemistry 2006

View Article Online
through PC controlled mass flow controllers (Tylan General).
Two terminal resistance measurements in varying gas atmo-
spheres were performed using an in-house constructed test-rig
that has been described previously.13 The devices were initially
allowed to stabilize for 24 hours at their operating tempera-
tures before any measurements were taken. Automated
resistance readings were made every 60 s with a multiplexed
Keithley Digital Multimeter.
temperatures of 350–500 uC mixed phase MoO2/MoO3 films
were obtained. These mixed phase films showed good coverage
of the substrate, however for some depositions, in particular
those using [nBu4N]4[Mo8O26], the substrate was comprised of
at least two distinct regions: an adherent white portion
closest to the precursor inlet which had a Raman spectrum
characteristic of only MoO3 and a dark blue region further
away from the precursor inlet which displayed additional
bands due to MoO2. This variation in the ratio of MoO2 to
MoO3 was also observed for the entirely mixed phase films
listed as MoO3/MoO2 in Table 1, with the MoO2 phase
predominating with increasing distance from the precursor
inlet. The Raman spectra for the films deposited at 400 uC
from the AACVD reactions of [nBu4N]4[Mo8O26] and
[nBu4N]4[Mo6O19] are shown in Fig. 1. The film deposited
from [nBu4N]4[Mo8O26] has a Raman pattern characteristic of
orthorhombic MoO3 with bands at 282, 337, 665, 822 and
996 cm21. The Raman spectrum for the film obtained from
[nBu4N]4[Mo6O19] displayed additional bands due to the
Results and discussion
Aerosol assisted chemical vapour deposition of five poly-
oxometallate
precursors:
[(n-C4H9)4N]4[Mo8O26];
[NH4]6-
[(n-C4H9)4N]4[Mo2O7];
[(n-C4H9)4N]4[Mo6O19];
[Mo7O24] and [(n-C4H9)4N]3[PMo12O40], using acetonitrile or
water as a solvent and transport medium were studied at
substrate deposition temperatures of 300–550 uC. The coatings
were deposited exclusively on the upper substrate of a cold
wall CVD reactor.11 The upper substrate (not directly heated)
was placed 4 mm above and was measured to be 50–75 uC
colder than the bottom plate. The temperature measurements
given in this work, Table 1 and the text of this paper, relate to
the upper substrate temperature. This somewhat unusual
deposition site arises from the relatively large size of the
polyoxometallate clusters coupled with gas phase nucleation
effects. Gas phase particles are subject to a thermophoretic
force when exposed to a temperature gradient.14 This force is
directed away from the hot surface, so larger particles act as if
repelled from a hot surface and attracted towards a colder one.
In effect the larger particles cannot diffuse through the thermal
boundary layer at the surface of the hotter substrate. Since the
flow of gas in the reactor is laminar rather than turbulent,
thermophoresis is usually the dominant force in determining
the deposition location of particles; hence coatings were
obtained exclusively on the top plate. Similar effects have
been observed in AACVD using nanoparticles.15 At deposition
temperatures in excess of 500 uC deep blue films were formed
whereas lower reaction temperatures resulted in the deposition
of white coatings. The films were adherent passing the
Scotch tape test. At 350–500 uC and irrespective of the
polyoxometallate used the films were comprised of multiple
regions, a well adhered deep-blue film, an adherent white
coating and a powdery black/blue surface deposit.
monoclinic MoO2 phase at 474, 562 and 681 cm21 16
.
X-Ray diffraction. Glancing angle X-ray diffraction (XRD)
was used to analyse the films. By using a small focused X-ray
spot (1–2 mm2) at low incident angle is was possible to analyse
separate regions of the films and obtain X-ray profiles of the
substrate at different points along its length. This variation in
the oxide phase deposited along the length of the substrate is
illustrated well by the film deposited from [nBu4N]4[Mo8O26] at
450 uC and 0.5 l min21. The film is comprised of approxi-
mately six regions varying in colour. The XRD patterns of the
different regions are shown in Fig. 2. The region closest to the
precursor inlet is comprised solely of MoO3 and that furthest
away of MoO2. The intermediate regions consist of a mixture
of these two phases representing a transition from the trioxide
to the dioxide. Films deposited from [nBu4N]4[Mo8O26] at
Complete coverage of the substrate was obtained at deposi-
tion temperatures below 400 uC with less coverage observed at
higher temperatures, the films having full-width coverage but
being localised to the central third of the substrate. This form
of deposition often relates to depletion of the precursor in the
carrier gas stream due to faster reaction rates at higher
temperature, in effect a mass transport limited deposition.
Fig. 1 Raman spectra for films deposited from [nBu4N]4[Mo8O26], a,
and [nBu4N]4[Mo6O19], b, in acetonitrile at 450 uC and 0.5 l min21. The
spectrum of a is characteristic of molybdenum trioxide (hollow circles)
whereas that of b contains additional peaks due to the dioxide phase
(filled circles). Spectrum c shows the Raman pattern of the film
deposited from [nBu4N]4[Mo8O26] at 550 uC which is characteristic of
molybdenum dioxide, with the exception of the broad peak in the
region 800 cm21, which may be a consequence of the high laser power
or due to a more oxidised surface compared with the bulk of the film.
Film characterisation
Raman analysis. The Raman spectra recorded for the
molybdenum polyoxometallate precursors showed similar
variations in phase with increasing substrate temperature. At
300 uC films were comprised solely of the MoO3 phase whereas
at the highest deposition temperatures of 500–550 uC single
phase MoO2 was observed. At intermediate substrate
This journal is ß The Royal Society of Chemistry 2006
J. Mater. Chem., 2006, 16, 3575–3582 | 3577

View Article Online
Fig. 2 Powder XRD patterns for the film deposited by the AACVD
reaction of [nBu4N]4[Mo8O26] in acetonitrile at 450 uC and 0.5 l min21
as a function of distance from the precursor inlet. The stack plot
illustrates the predominance of the molybdenum dioxide phase (filled
circles) over the molybdenum trioxide phase (hollow circles) with
increasing temperature.
Fig. 3 Powder XRD patterns for the films obtained from the
AACVD reactions of [nBu4N]3[PMo12O40] in acetonitrile at varying
substrate temperatures. The stack plot illustrates the predominance of
the molybdenum dioxide phase (filled circles) over the molybdenum
trioxide phase (hollow circles) with increasing deposition temperature.
The filled squares represent the peaks arising due to the MoP2O7
phase.
400 uC also have variable composition (Table 1). In contrast
the films obtained at 500 uC and 550 uC consisted solely of the
MoO2 phase. The single phase MoO3 and MoO2 films showed
no preferential orientation, whereas the MoO3 components of
the mixed phase films did show some degree of preferred
orientation. However due to the overlap of peaks it was
not possible to determine the plane of orientation. Preferred
orientation arises when the constraints of a thin film are
molybdenum oxide formation and the crystallisation isotherms
as assessed from the TGA/DSC data (see below). Films
deposited from [nBu4N]3[PMo12O40] at 400 and 450 uC
contained additional peaks in their XRD spectra which could
be assigned to cubic MoP2O7. A stack plot showing the
variation in phase as a function of deposition temperature is
shown in Fig. 3 for films deposited using [nBu4N]3[PMo12O40].
imposed upon
a material, and is a feature commonly
associated with CVD films.
Thermal gravimetric analysis, differential scanning calorime-
try. The TGA/DSC data shows that the polyoxometallates all
follow the same general decomposition pattern, Fig. 4, 5. The
initial step involves loss of the tetrabutylammonium attendant
counter ion to yield the core cluster that upon further
heating decomposes to MoO3. The final step involves the
decomposition of MoO3 to MoO2. This is illustrated for
[nBu4N]3[PMo12O40] (Fig. 4) which showed a total mass loss at
500 uC of 46% that corresponds to decomposition to MoO2
(theoretically a 45% mass loss). Four distinct decomposition
steps are present, and these are interpreted as follows. The first
mass loss at 20–100 uC corresponds to removal of solvent from
the polyoxometallate cluster. From 240–355 uC a 28% mass
The X-ray diffraction data is consistent with the Raman
study; at substrate temperatures of 550 uC and 500 uC the
diffraction patterns are characteristic of monoclinic MoO2,
at 300 uC and 350 uC of orthorhombic MoO3 whereas those
obtained at intermediate temperatures relate to a mixture
of the two oxide phases. However the observation of a
predominantly MoO2 phase for the films deposited at 450 uC
from [nBu4N]4[Mo8O26] contradicts with the Raman data, in
which MoO3 was also observed. This difference may be due to
oxidation of the films at high laser powers or a consequence of
the surface sensitivity of the technique compared with XRD.
The diffraction patterns for MoO3 indexed in the Pbnm space
group with typical cell parameters of a = 3.96, b = 13.85, c =
˚
3.69 A; this compares with literature values of a = 3.92, b =
17
13.94, c = 3.66 A, whereas the MoO2 diffraction pattern
˚
indexed in the P1211 space group with typical cell parameters
˚
of a = 5.62, b = 4.84, c = 5.56 A, b = 119.32u which again
compares very well with typical literature values of a = 5.61,
b = 4.84, c = 5.53 A, b = 119.62u.18 The coatings obtained from
˚
[nBu4N]4[Mo8O26] at a given deposition temperature were the
most crystalline in nature (as calculated from XRD line
broadening studies) followed by those deposited from
[nBu4N]3[PMo12O40] and those from [nBu4N]2[Mo2O7] and
[nBu4N]4[Mo6O19] were the least. For example the coatings
deposited from [nBu4N]4[Mo8O26] and [nBu4N]3[PMo12O40] at
˚
400 uC yielded an MoO3 phase with crystallite sizes of 310 A
and 200
˚
A respectively (determined using the Scherrer
Fig. 4 TGA/DSC data for [nBu4N]3[PMo12O40]. Step 1 represents
loss of [nBu4N]4 to yield PMo12O40, step 2 decomposition to MoO3 and
step 3 the conversion to MoO2.
equation). This difference in crystallinity to a certain extent
correlates with the temperature changes associated with
3578 | J. Mater. Chem., 2006, 16, 3575–3582
This journal is ß The Royal Society of Chemistry 2006

View Article Online
Fig. 5 TGA/DSC data for [nBu4N]4[Mo8O26]. Step 1 represents
removal of solvent from the complex, Step 2 represents loss of
[nBu4N]4 to yield Mo8O26, step 3 decomposition to MoO3 and step 4
the conversion of MoO3 to MoO2.
loss occurs, corresponding to the loss of the three tetrabutyl-
ammonium ions (theoretically 28%). From 355–360 uC, a 5%
mass loss occurs, corresponding to the decomposition of the
[PMo12O40]32 core to 12MoO3 (theoretically 5%). This is
associated with a sharp exotherm at 360 uC, indicating the
formation of MoO3. From 360–500 uC, a 6% mass loss occurs,
corresponding to the decomposition of MoO3 to MoO2
(theoretically 7%). This final change is accompanied by a
broad exotherm.
Scanning electron microscopy and wavelength dispersive
analysis by X-rays. Scanning electron microscopy (SEM) was
used to examine the surface morphology of the films and
measure the film thickness. The microstructure of the
deposited films varied greatly as a function of the precursor,
the deposition temperature and position of the film on the
substrate. Deposition from [nBu4N]4[Mo8O26] at 400 uC
afforded a material that is comprised of spherical agglomerates
of crystallites up to 1 mm in diameter, intertwined with needle
agglomerates, Fig. 6, images 1 and 2. The film deposited
from the same precursor at 500 uC yielded a coating with a
similar microstructure; however there appears to be a greater
concentration of needle agglomerates (images 3 and 4). These
morphologies are consistent with the formation of multi-phase
films, with the spherical agglomerates representing the trioxide
phase, whereas the needle agglomerates correlate to MoO2.
This trend was observed for many of the coatings deposited
from [nBu4N]4[Mo8O26]; regions of the films in which the
dioxide phase predominated contained a higher concentration
of the needle agglomerates. Annealing the films in air at 550 or
600 uC resulted in a film microstructure composed entirely of
spherical agglomerates or of a condensed continuous film.
Deposition from [nBu4N]2[Mo2O7], [nBu4N]4[Mo6O19] and
(NH4)6[Mo7O24] at 500 uC (images 6, 7 and 8 respectively)
afforded a material comprised of agglomerates of crystallites
up to 1 mm in diameter, with a greater degree of agglomeration
observed for the films deposited from [nBu4N]2[Mo2O7]
and [nBu4N]4[Mo6O19], whereas the coating obtained from
[nBu4N]3[PMo12O40] at the same temperature was composed of
needles and platelets up to 1 mm in length fused together
(image 9). These morphologies obtained in this study are
Fig. 6 Scanning electron micrographs of molybdenum oxide films
deposited from AACVD reactions of polyoxometallate precursors.
Images 1 and 2 show the surface morphology of the coatings deposited
from [nBu4N]4[Mo8O26] at 400 uC, images 3 and 4 at 500 uC and
image 5 following annealing in air at 550 uC. Images 6, 7, 8 and 9
show the films deposited from [nBu4N]2[Mo2O7], [nBu4N]4[Mo6O19],
[NH4]6[Mo7O24] and [nBu4N]3[PMo12O40] respectively prepared at
550 uC
unlike those seen in literature for films deposited from
molybdenum hexacarbonyl and oxygen by APCVD on silicon
wafers by molybdenum pentacarbonyl 1-methylbutylisonitrile
via PECVD on glass and molybdenum pentachloride by means
of spray pyrolysis, all of which have a uniform and compact
microstructure, similar to that observed upon sintering the
polyoxometallate deposited films.8,9 The range of different
morphologies obtained in this study has important implica-
tions for the functional properties of the films, in particular the
gas-sensing properties which can be enhanced by an open
This journal is ß The Royal Society of Chemistry 2006
J. Mater. Chem., 2006, 16, 3575–3582 | 3579

View Article Online
porous structure and the hydrophobicity of the films which is
dependent upon a needle-like microstructure. Cross-section
SEM showed that all the film thicknesses were of the order
500 nm, which correlates to a growth rate of approximately
(recorded under a nitrogen atmosphere), which shows that the
MoO3 phase is formed within a narrow temperature range and
readily undergoes oxygen loss. However all films are converted
to MoO3 following annealing in air at 600 uC, which implies
that the oxygen loss is a dynamic process. The variation in
phase composition within an individual film is a consequence
of the slightly elevated substrate and gas flow temperatures
nearer the back of the reactor. This is because the reactor is of
a cold wall design and only the substrate plate is heated. The
variation in colour from pale yellow to blue with increasing
distance from the precursor inlet represents the transition from
the MoO3 phase to MoO2. The observation of the pale yellow
regions is consistent with the presence of MoO3, however
MoO2 is grey. This discrepancy in colour suggests that the blue
regions analysed as MoO2 also contain Mo5+ centres such as
MoO2(OH), which although they are present only in trace
amounts as indicated by the lack of any diffraction or
spectroscopic evidence have a very intense blue colour and
hence dominate over the grey MoO2.
15 nm min21
.
Wavelength dispersive analysis of the molybdenum oxide
films deposited from [nBu4N]4[Mo8O26], [nBu4N]2[Mo2O7],
[nBu4N]4[Mo8O26] and [NH4]6[Mo7O24] showed only the
presence of molybdenum and oxygen in the expected ratios
(MoO3, MoO2 or a mixed phase). In addition to molybdenum
and oxygen significant levels of phosphorus were detected
in the WDX spectra for the films obtained from
[nBu4N]3[PMo12O40] deposited at 400–500 uC with a Mo : P
elemental ratio of 12 : 1, which mirrored the ratio of the
starting material. This implies that other polyoxometallates
with mixed element cores are likely to serve as suitable
precursors for molybdenum oxide films with controlled
incorporation of a secondary element.
X-Ray photoelectron analysis. The films obtained from the
AACVD reaction of [nBu4N]4[Mo8O26] in acetonitrile at 400 uC
and carrier gas flow of 0.5 l min21 were comprised of two
distinct regions; a white MoO3 region closest to the precursor
inlet (region A) and a deep blue mixed MoO3 and MoO2 phase
further away (region B). XPS of region A revealed a doublet
corresponding to Mo6+ 3d5/2 and Mo6+ 3d3/2 photoelectrons at
binding energies of 233.1 and 236.3 eV respectively and an O 1s
ionisation at 531.0 eV. These molybdenum and oxygen
chemical shifts are in good agreement with previous studies
of MO3.16 The absence of any splitting or broadening of the
Mo 3d doublet indicates that the Mo was present in a single
environment; this is consistent with the Raman and XRD data
which showed that the region was comprised solely of the
MoO3 phase. In contrast the Mo 3d doublet for region B was a
composite of two peaks: a shoulder is observed at binding
energies of 229.4 eV and 232.6 eV, indicative of the presence
of MoO2, which is again in agreement with the Raman and
XRD analysis.
Functional properties. The molybdenum oxide films were
hydrophobic in nature, with water contact angle values in the
range 75–125u. The highest contact angles were measured for
the films which were comprised of MoO2 or a mixture of the
two oxide phases. SEM showed that these films had a needle-
like microstructure, which is indicative of a rough surface (see
Fig. 6, image 3). The surfaces were adhesive in that water
droplets tended not to be displaced even when placed vertically
or inverted. As such they are examples of a Wenzel hydro-
phobic surface19 where the water droplet penetrates the surface
structure rather than a Cassie–Baxter surface20 where the
water droplet is held by the surface and in effect is supported
on a cushion of air. We have recently reported water contact
angles as high as 145u on a WSe2 surface, this surface was also
very spiky in nature.21 Whilst the molybdenum oxide films are
not quite as hydrophobic they are easier to produce and do not
involve the use of potentially toxic selenium. It is unusual to
have contact angles in excess of 110u, in fact commercial lenses
for spectacles typically have contact angles of 95u.21 The
generation of very hydrophobic thin films such as the ones
reported here are unsuitable for use in self-cleaning or anti-
mist applications. However they could find application in
biology for micropipetting and in microfluidics.
The XPS surface scan of the film deposited from
[nBu4N]3[PMo12O40] at 450 uC and carrier gas flow of
1 l min21 was similar to that of region B, containing multiple
Mo 3d environments at binding energies of 233.0 and 236.1 eV
and 232.0 eV and 235.0 eV indicating the presence of a mixed
MoO3 and MoO2 surface, and O 1s ionisations at 530.9
and 532.0 eV. In addition a P 2p ionisation was observed at
134.2 eV consistent with the presence of MoP2O7, which is in
agreement with XRD analysis. The Mo : P ratio was again
found to be 12 : 1, the same as in the precursor. Therefore,
using [nBu4N]3[PMo12O40] as a precursor has yielded molyb-
denum oxide composite thin films containing a molybdenum
phosphorus oxide, in which the molybdenum to phosphorus
ratio of the initial polyoxometallate precursor has been
preserved.
The molybdenum oxide films were also examined as gas
sensitive resistors for the detection of ethanol. The molybde-
num oxide films were deposited as previously described except
that they were laid down on a gas sensor substrate, which
consisted of a series of interdigitated gold electrodes on an
alumina tile. The molybdenum oxide coating covered both the
electrodes and the alumina tile. The thin film responded well to
pulses of ethanol (100 ppm level, 50 minutes on and then off)
in dry air showing a significant change in resistance (Fig. 7)
and recovered to baseline resistance upon removal of the
ethanol pulse. The magnitude of the response did not change
significantly with each pulse of ethanol with the exception of
the initial pulse which was approximately half that of
subsequent pulses. We have shown previously that CVD can
be used to form gas sensors of tungsten oxide and chromium
titanium oxide.23 The molybdenum oxide sensors formed here
Phase variation. Raman and XRD analysis show that low
deposition temperatures are favourable for the formation of
MoO3 whereas at high deposition temperatures MoO2 is the
dominant phase. This variation in phase as a function of
deposition temperature correlates well with the TGA data
3580 | J. Mater. Chem., 2006, 16, 3575–3582
This journal is ß The Royal Society of Chemistry 2006

View Article Online
further vaporised as is the case with conventional AACVD. In
conventional aerosol CVD, the solvent is evaporated to leave a
small molecular unit that itself is vaporised. Furthermore the
effects of thermophoresis are typically negligible. It would
appear that formation of nanoparticles in the gas phase and
then agglomeration are important which then undergo further
reaction (or sintering) as a secondary more CVD like process.
The ease of polyoxometallate formation, the majority of
which are standard preparations as reported in Inorganic
Synthesis,12 coupled with their clean decomposition to
molybdenum oxide makes them desirable for use as novel
CVD precursors. The use of the phosphorus doped poly-
oxometallate, [nBu4N]3[PMo12O40], resulted in the deposition
of phosphorus doped molybdenum oxide films in which the
Mo : P ratio of the core has been retained. This indicates that it
is the polyoxometallate core that determines the secondary
element level in the film and could be readily harnessed by
using a mixture of polyoxometallate precursors. Furthermore
the attendant organic counter ions do not introduce any
detectable carbon or nitrogen in the films and correlate well
with the TGA plots that show mass losses equivalent to the
counter ion removal at ca. 250 uC. In addition the main
solvent used in these experiments, acetonitrile, must evaporate
cleanly as the deposited films are free of carbon con-
tamination. Polyoxometallates are good single source pre-
cursors to molybdenum oxide films, since no additional
oxygen is available from either the carrier gas stream or the
aerosol solvent (in the case of acetonitrile) then the only
source of oxygen for the growing films is from the
polyoxometallate core.
Fig. 7 The change in electrical resistance upon repeated exposure
to 100 ppm of ethanol vapour for a gas sensor deposited from
[nBu4N]3[PMo12O40] in acetonitrile at 550 uC.
show similar responses to those made previously. One
emerging feature of CVD prepared gas sensors is that they
tend to have a much denser, less porous microstructure
compared with sensors made by screen printing13 consequently
they tend to have lower overall responses. Advantages that
CVD does have over screen printing for making gas sensors
include the ease of manufacture and the fact that these films
are readily integrable into silicon microfabrication.
Use of an aerosol in CVD
The use of an aerosol in CVD has modified the requirements
of the precursor. In traditional molecular organic chemical
vapour deposition precursors are designed to be volatile and
are often single source such that they contain the bonds
required in the growing film in the precursor. For example
metal oxide films are typically laid down from metal alkoxides
such as Ti(OiPr)4 and metal nitride films from Ti(NMe2)4.22 In
aerosol assisted chemical vapour deposition the primary
requirement of the precursor is solubility in a suitable solvent
and the ability of the solvent to form a stable aerosol mist. The
precursors used in this work are large ionic complexes with
very high molecular masses ranging from 788 to 2549 g mol21
that would be unsuitable for use as conventional CVD
precursors due to their lack of volatility and thermal stability.
However they are readily transported in acetonitrile or water
aerosols to the reaction site where the aerosol droplet
evaporates depositing the precursor on the surface. It is
possible that the formation particularly of the MoO3 phase
occurs partially in the gas phase; this is consistent with the
SEM studies which revealed a spherical morphology for this
phase. Such a morphology would be expected upon evapora-
tion and condensation of an aerosol droplet, which is in effect
a spherical microreactor which decreases in size following
evaporation of the solvent. Interestingly the size of these
aerosol droplets is ca. 15 microns based on the piezoelectric
crystal used and the diameter of the MoO3 particles is in the
range between 0.5–1 microns.23 In this instance however it is
some what debatable whether film growth is due to a CVD
process. The basis of the growth appears to be evaporation of
solvent to produce polyoxometallate clusters which undergo
agglomeration, rather than a small molecular unit which is
Conclusion
We have demonstrated the use of large charged clusters as
single-source precursors in CVD reactions, where it is the
convention to use small, neutral monomers. This has been
achieved by taking advantage of the aerosol vaporisation
technique, whereby the precursor only has to be soluble in a
suitable solvent, and volatility is no longer a requirement, as is
the case with conventional CVD processes. AACVD reactions
of [nBu4N]4[Mo8O26], [(n-C4H9)4N]4[Mo2O7], [(n-C4H9)4N]4-
[Mo6O19], [NH4]6[Mo7O24] and [nBu4N]3[PMo12O40] resulted
in the deposition of films comprised of multiple regions
corresponding to MoO3, MoO2 and a mixture of these oxides,
with the MoO2 phase predominating with increasing tempera-
ture and distance from the reactor inlet. Annealing in air for
30 minutes at 600 uC resulted in the conversion of all films to
orthorhombic MoO3. The resulting film microstructures were
varied, dependent upon both the starting precursor and the
deposition temperature and unlike those observed previously
in literature reports. The range of film morphologies obtained
has important implications for the functional properties,
namely the gas sensing and contact angles, upon which film
microstructure impacts strongly. The molybdenum oxide films
are hydrophobic and function as gas sensitive resistors for
detecting traces of ethanol vapour in air.
Previous reports detailing the CVD of molybdenum oxide
films have either described dual-source routes or involved the
initial deposition of molybdenum metal followed by sintering
This journal is ß The Royal Society of Chemistry 2006
J. Mater. Chem., 2006, 16, 3575–3582 | 3581

View Article Online
in an oxygen atmosphere to yield the oxide.9 In contrast
polyoxometallates provide a single-source route to the forma-
tion of molybdenum oxide films; furthermore they enable
controlled levels of a dopant to be incorporated into the film,
which may modify its functional properties.
7 M. S. Burdis and J. R. Siddle, Thin Solid Films, 1994, 237, 320.
8 M. A. D. Moraes, B. C. Trasferetti, F. P. Rouxinol and R. Landers,
Chem. Mater., 2004, 16, 513–520.
9 T. Ivona, K. A. Gesheva, G. Popkirov, M. Genchevard and
T. Tzoetova, Mater. Sci. Eng., B, 2005, 119, 232.
10 R. Binions, C. J. Carmalt and I. P. Parkin, Thin Solid Films, 2004,
469, 416.
11 W. B. Cross and I. P. Parkin, Chem. Commun., 2003, 1696–1697.
12 Inorganic Synthesis, ed. A. P. Ginsberg, J. Wiley and Sons,
Chichester, 1990, vol. 27, p. 86.
Acknowledgements
The EPSRC is thanked for support. Mr D. Knapp, Mr J.
Nolan and Mr D. Morphett are thanked for technical
assistance in constructing the CVD rig. Mr R. Palgrave, Dr
R. Binions and Mr K. Reeves are thanked for their assistance
with XPS, WDX and SEM analysis respectively. Professor K.
C. Molloy is thanked for useful discussions. Professor Parkin
thanks the Royal Society/Wolfson Trust for a research merit
award.
13 G. Chabanis, I. P. Parkin and D. E. Williams, Meas. Sci. Technol.,
2003, 14, 76; G. A. Shaw, K. E. Pratt, I. P. Parkin and
D. E. Williams, J. Mater. Chem., 2005, 15, 149; G. A. Shaw,
K. E. Pratt, I. P. Parkin and D. E. Williams, Sens. Actuators, B,
2005, 104, 151–162.
14 E. Radkov and R. E Beer, Polyhedron, 1995, 4, 2139.
15 R. G. Palgrave and I. P. Parkin, J. Am. Chem. Soc., 2006, 128,
1587.
16 K. Gesheva, A. Cziraki, T. Ivanova and A. Szekeres, Thin Solid
Films, 2005, 492, 322; J. G. Choi and T. Thompson, Appl. Surf.
Sci., 1996, 93, 143.
17 N. Wooster, Z. Kristallogr. Kristallgeom. Kristallphys.
Kristallchem., 1931, 80, 504–512.
References
18 A. Magneli, Ark. Kemi, Mineral. Geol., 1947, 24, 1–11.
19 R. N. Wenzel, Ind. Eng. Chem., 1936, 28, 988.
20 A. B. D. Cassie and S. Baxter, Trans. Faraday Soc., 1944, 40,
546.
21 I. P. Parkin and R. G. Palgrave, J. Mater. Chem., 2005, 15, 1689.
22 J. R. Roos and H. Veltrop, Thin Solid Films, 1990, 193, 547.
23 G. A. Shaw, K. E. Pratt, I. P. Parkin and D. E. Williams, Sens.
Actuators, B, 2005, 104, 151; G. A. Shaw, K. E. Pratt, I. P. Parkin
and D. E. Williams, J. Mater. Chem., 2005, 15, 149.
1 M. J. Davis, G. Benito, D. W. Sheel and M. E. Pemble, Chem. Vap.
Deposition, 2004, 10, 29.
2 K. D. Lee, Thin Solid Films, 1997, 302, 84.
3 K. Hinokuma, K. Kishimoto and T. Kudo, J. Electrochem. Soc.,
1994, 141, 876.
4 A. Guerfi and L. A. Dao, J. Electrochem. Soc., 1989, 136, 2435.
5 J. Scarminio, A. Lorcenco and A. Gorenston, Thin Solid Films,
1997, 302, 66.
6 M. R. Tubbs, Phys. Status Solidi A, 1974, 21, 253.
3582 | J. Mater. Chem., 2006, 16, 3575–3582
This journal is ß The Royal Society of Chemistry 2006