Low temperature chemical vapor deposition of aluminosilicate thin films on carbon fibers

Article abstract of DOI:10.1111/j.1551-2916.2005.00248.x

We report the deposition of aluminum oxide and aluminosilicate thin films onto carbon fiber substrates, at temperatures of 200 and 250°C, respectively. For aluminosilicate films, the Al/Si ratio of the resultant film varied concomitantly with the composition of the liquid precursor mixture. The growth rate for the oxide films was 15-17 A/min, comparable with other methods carried out at higher temperatures. Cross-section SEM images indicate that the deposited films are conformal, following the complex topography of the carbon fiber substrate. Preliminary gas-phase IR analysis suggests that the coatings decompose the nerve agent simulant dimethyl methylphosphonate at temperatures as low as 35°C, suggesting the utility of the reported methodology for the design/fabrication of actively protective fabrics and clothing.

Full text of DOI:10.1111/j.1551-2916.2005.00248.x

J. Am. Ceram. Soc., 88 [7] 1973–1976 (2005)
DOI: 10.1111/j.1551-2916.2005.00248.x
ournal
J
Low Temperature Chemical Vapor Deposition of Aluminosilicate Thin
Films on Carbon Fibers
Vernal N. Richards, Jason K. Vohs, and Bradley D. Fahlmanw
Department of Chemistry, Central Michigan University, Mount Pleasant, Michigan 48859
Geoffrey L. Williams
Department of Biology, Central Michigan University, Mount Pleasant, Michigan 48859
We report the deposition of aluminum oxide and aluminosilicate
thin films onto carbon fiber substrates, at temperatures of 2001
and 2501C, respectively. For aluminosilicate films, the Al/Si ra-
tio of the resultant film varied concomitantly with the compo-
sition of the liquid precursor mixture. The growth rate for the
aluminosilicate films, systems utilizing one or two precursors
have been used; the first single-source precursor was the mono-
siloxide complex, Al(O Pr) (OSiMe ). However, the use of this
i
6
2
3
precursor required temperatures exceeding 9001C in order to get
films with an even distribution of aluminum and silicon. Al-
though attempts have been made to control Al:Si ratios in alu-
minosilicate films, such stoichiometric control remains difficult
˚
oxide films was 15–17 A/min, comparable with other methods
carried out at higher temperatures. Cross-section SEM images
indicate that the deposited films are conformal, following the
complex topography of the carbon fiber substrate. Preliminary
gas-phase IR analysis suggests that the coatings decompose the
nerve agent simulant dimethyl methylphosphonate at tempera-
tures as low as 351C, suggesting the utility of the reported
methodology for the design/fabrication of actively protective
fabrics and clothing.
7
to control, for both single-source and dual-source systems.
II. Results and Discussion
Our CVD system for the growth of aluminum oxide and alu-
minosilicate films featured the controlled hydrolysis of water-
sensitive precursor liquids. To prevent extensive gas-phase reac-
tions and a granular coating, water and precursor vapors were
introduced through separate tubes, and allowed to come into
I. Introduction
contact only in the immediate vicinity of the substrate (Fig. 1).
ITH the constant threat of terrorist attacks, much attention
has been focused on the development of materials that
i
Al( OPr)]
For aluminum oxide films, a volatile precursor, [Me
2
2
,
W
was used alone; for aluminosilicate films, tetraethoxysilane was
added to the above aluminum-containing liquid. It should be
will protect both military personnel and civilians against the re-
lease of chemical and biological warfare agents. Solid metal ox-
ides, such as aluminum oxide, have been shown to adsorb and
decompose organophosphorus-based chemicals at temperatures
1
13
noted that both H and C NMR confirmed that the simple
physical mixing of aluminum and silicon liquids did not result in
reaction by-products, only a homogeneous one-phase solution.
The effective deposition temperature for aluminosilicate and
alumina thin films was found to be substrate-dependent. On Si
1
not far removed from ambient. The deposition of thin films of
these materials onto substrates such as clothing fibers and fab-
rics would represent an intriguing utility for actively protective
clothing, for both military and civilian use. For the deposition
of coatings onto thermally sensitive substrates with such com-
plex topography, it is imperative to have a low-temperature
method that is still able to produce conformal thin films. The
ultimate methodology would further enable one to control
the stoichiometry of the growing film. To this end, we report
the low-temperature deposition of conformal aluminum oxide
and aluminosilicate thin films on carbon fiber substrates, with
facile control over Al:Si ratios for the latter films. Our temper-
atures represent the lowest-reported temperatures to date, al-
lowing other possible applications for temperature-sensitive
substrates.
(
100), the lowest temperatures yielding a homogeneous coating
of aluminum oxide and aluminosilicate were 1251 and 2001C,
respectively (Fig. 2). However, when carbon fibers were used as
substrates, no observable coating was observed at temperatures
below 2001 and 2501C for aluminum oxide and aluminosilicate,
respectively. These values indicate that a higher temperature is
required for the migration/nucleation of surface species over the
complex topography of an irregular carbon fiber, relative to a
silicon wafer. The increased thermal conductivity of Si (100)
relative to carbon would also explain the increased migratory
efficiency of intermediates at lower temperature on silicon, caus-
ing nucleation and growth of films at lower temperatures. The
temperature regime used in our system is far below the reported
deposition temperature (5001–6001C) for low-pressure chemical
vapor deposition (LPCVD) using the above aluminum pre-
Common precursors used for the chemical vapor deposition
2
3
4
(
CVD) of Al O films include AlCl , AlMe , Al(acac) , and
2 3 3 3 3
5
various aluminum alkoxides. Alkoxides are the precursors
of choice, producing films with little carbon incorporation at
low temperatures even in the absence of an oxidizing gas. For
8
9,10
cursor alone or associated with metal acetylacetonates.
To our knowledge, our study also represents the first use of
[
i
2 2
Me Al( OPr)] for atmospheric-pressure chemical vapor deposi-
tion (APCVD) growth of aluminum oxide-containing films,
allowing for facile industrial scale-up relative to LPCVD
techniques.
L. C. Klein—contributing editor
The aluminosilicate films grown on the carbon fiber are con-
formal, perfectly following the rough terrain of the fiber surface
(Fig. 3). Because of the low temperatures used for deposition,
the as-deposited films are amorphous, confirmed by powder
X-ray diffraction. Auger spectroscopy suggests that the alumi-
Manuscript No. 11255. Received August 6, 2004; approved November 9, 2004.
Supported by the State of Michigan and Central Michigan University by a Research
Excellence Fund award for this and continuing work. Also supported by Dendritic Nano-
technologies and the Department of Defense (Army Research Laboratory).
w
Author to whom correspondence should be addressed. e-mail: fahlm1b@cmich.edu
1
973

1
974
Communications of the American Ceramic Society
Vol. 88, No. 7
Fig. 1. Schematic of the chemical vapor deposition system used for deposition of aluminum oxide and aluminosilicate thin films.
num oxide films are stoichiometric Al
icon environment in aluminosilicate films is intermediate be-
tween SiO and SiO(OH) . Our observed growth rates for
2
O
3
. By contrast, the sil-
films are comparable with APCVD using metal acetylacetonate
precursors.
7
2
2
Figure 4 shows elemental dot-maps of C, O, Al, and Si con-
centrations for a cross-section of a coated carbon fiber. Both Al
and Si are observed to completely encapsulate the fiber, indi-
cating a high degree of surface mobility of reactive intermediates
during deposition. It should be noted that a randomly inter-
twined tow of fibers is placed in the CVD chamber for de-
position runs, rather than an individual carbon fiber. The
conformality exhibited at such a low temperature will allow
for facile scale-up of this procedure; a tow of fibers may be
reeled through the deposition furnace at a specific rate, with
coated fibers being rewound onto a new spool.
˚
˚
aluminum oxide (17 A/min) and aluminosilicate (15 A/min)
The Al:Si ratio of the deposited film was found to vary di-
rectly with the relative ratio in the precursor solution (Table I).
i
Al( OPr)]
This is counterintuitive, as the volatility of [Me
2
2
is
much greater than Si(OEt) , suggesting that a solution contain-
4
ing both liquids would simply yield silicon-free aluminum oxide
films until the Al precursor has been vaporized. This trend was
only observed when the fraction of the aluminum precursor ex-
ceeded that of the silicon component. When there is an excess of
the relatively involatile component corresponding to a relatively
low overall solution volatility, the vapor pressure of the alumi-
num precursor is suppressed, resulting in less at.% Al in the
growing film.
Fig. 2. Elemental dot-map of the aluminum (blue) and silicon (red)
concentrations on Si (100). Shown are (a) aluminum oxide deposited
at 1251C and (b) aluminosilicate grown at 2001C (2:1 Al:Si precursor).
The penetration-depth intensity from the silicon substrate has been
subtracted.
Fig. 3. Cross-section SEM image of a carbon fiber coated with an alu-
minosilicate thin film. Scale bar is 2 microns.

July 2005
Communications of the American Ceramic Society
1975
Table I. Aluminum/Silicon Ratios of Precursor and Deposited
Filmw
Al:Si precursor ratio
Al:Si film ratio
0
1
1
2
2
.85
.00
.30
.00
.50
w
0.46
1.06
1.23
1.74
2.32
Deposition temperature 5 2501C; flow rates: H
2
O 5 10 mL/min and precursor
5
40 mL/min.
there is still a trace amount of methanol that is inherent in the
liquid from its synthesis. This corresponds to the observation of
a n(O–H) peak from the vaporization of DMMP in the presence
of an uncoated fiber (Fig. 5(a)). It should be noted that the in-
tensity of this peak is identical for the blank experiment when no
fiber is present in the gas IR cell. However, at temperatures
ꢀ
351C there is a marked increase in the concentration of gas-
phase methanol that is present in the cell when a coated fiber is
present, relative to uncoated carbon fiber. It is especially note-
worthy that there was no observable difference in the low-tem-
perature decomposition behavior between aluminum oxide and
aluminosilicate-coated fibers, or for fibers with varying film
thicknesses. This suggests that both films possess a similar con-
centration of surface hydroxyl groups, which are proposed to be
11,12
influential in the decomposition pathway.
The low deposi-
Fig. 4. Cross-section elemental dot-maps of a coated carbon fiber (1:1
Al:Si precursor). Shown are concentrations of (a) carbon, (b) oxygen,
tion temperatures used in our system are most desirable for such
catalytic applications, as the low crystallinity as well as copious
surface defects and ÀOH moieties provide increased sites
for adsorption/decomposition of gaseous organophosphorus
agents. We are continuing our work in this area by examining
other metal oxide films, as well as more sensitive surface analysis
techniques to determine the decomposition mechanism and ef-
fectiveness of the deposited films.
(c) aluminum, and (d) silicon.
In order to study the potential catalytic activity of the oxide
films toward the decomposition of chemical warfare agents,
the organophosphorus compound dimethyl methylphosphonate
(
DMMP) was used. This compound has been used as a simulant
in previous studies, as it mimics the structure of a number of
deadly agents. The use of the deposited oxide layer as a catalyst
for the decomposition of DMMP vapor at ambient tempera-
tures was studied using gas-phase IR spectroscopy (Fig. 5). One
of the products of DMMP decomposition is methanol, indicated
in IR by the presence of a strong n(O–H). Although DMMP was
distilled in vacuo overnight and stored over molecular sieves,
III. Experimental Procedure
Trimethylaluminum, tetraethoxysilane (Strem Chemicals, Inc.,
Newburyport, MA), and aluminum isopropoxide (Aldrich, St.
Louis, MO) were purchased and used without further purifica-
Fig. 5. Gas-phase IR spectrum of (a) an uncoated carbon fiber exposed to dimethyl methylphosphonate (DMMP) vapor and (b) an aluminosilicate-
coated carbon fiber exposed to DMMP vapor.

1
976
Communications of the American Ceramic Society
Vol. 88, No. 7
tion. These reagents are air/moisture sensitive and were handled
in a glove box (He-63-P pedatrol, Vacuum Atmospheres, Haw-
thorne, CA) under an inert atmosphere of argon. Dimethyl
methylphosphonate (Aldrich, 97%) was distilled under vacuum
for 24 h and stored over molecular sieves prior to use.
ing a Filmetrics F20 film thickness analyzer (Filmetrics, Inc, San
Diego, CA).
Acknowledgment
To prepare the aluminum precursor, an amount of aluminum
isopropoxide (11.73 g, 575 mmol) and trimethylaluminum
We give special thanks to Dr. Donald Tomalia for his continuing support and
many fruitful discussions related to this and other research projects currently in
progress.
(
8.273 g, 114.9 mmol) were weighed into separate Schlenk flasks
inside the glove box. The trimethylaluminum was subsequently
transferred to the flask containing the aluminum isopropoxide
powder via cannula at À781C. The reactants were left to stir for
References
2
4 h, after which the product was separated from the remain-
1
1
V. N. Sheinker and M. B. Mitchell, ‘‘Quantitative Study of the Decomposition
of Dimethyl Methylphosphonate (DMMP) on Metal Oxides at Room Tempera-
ture and Above,’’ Chem. Mater., 14 [3] 1257–68 (2002), and references therein.
B. Lux, H. Colombier, H. Altena, and K. Sternberg, ‘‘Preparation of Alumina
Coatings by Chemical Vapour Deposition,’’ Thin Sol. Films, 138 [1] 49–64
(1986).
M. D. Groner, F. H. Fabreguette, J. W. Elam, and S. M. George, ‘‘Low-Tem-
ing solid by distillation. H NMR (300 MHz, C
H CH -Al, 1.1 6H CH(CH , 3.9 1H CH(CH
300 MHz, C , d): 3.0, 25.3, 66.9.
In order to minimize vapor-phase mixing, the precursor and
6
D
6
, d): À0.6
1
3
6
(
3
)
3 2
3
)
2
:
C NMR
2
D
6 6
water vapor were carried into the CVD chamber through sep-
arate lines. A tow of carbon fibers (ca. 10–20 cm in length) was
rinsed with acetone and dried prior to use. Silicon wafers were
cleaned prior to use by dipping in a 10% solution of HF, fol-
lowed by rinsing with distilled water and air drying. Before the
precursor was allowed to enter the CVD chamber, the substrate
was placed at the base of the quartz tube, and the system was
purged with argon until the desired reaction temperature was
reached. Initial trials were run at a reaction temperature of
3
2 3
perature Al O Atomic Layer Deposition,’’ Chem. Mater., 16 [4] 639–45
(2004).
4
O. B. Ajayi, M. S. Akanni, J. N. Lambi, H. D. Burrow, O. Osasona, and B. P.
Podor, ‘‘Preparation and Optical Characterization of Pyrolytically Deposited Thin
Films of Some Metal Oxides,’’ Thin Sol. Films, 138 [1] 91–5 (1986).
S. Blittersdorf, N. Bahlawane, K. Kohse-Hoinghaus, B. Atakan, and J. Muller,
¨
¨
‘‘CVD of Al O Thin Films Using Aluminum Tri-isopropoxide,’’ Chem. Vapor
Depos., 9 [4] 194–8 (2003).
F. Chaput, F. A. Lecomte, A. Dauger, and J. P. Boilot, ‘‘Preparation and
Structure of Aluminosilicate Aerogels,’’ Chem. Mater. 1 [2] 199–201 (1989).
A. R. Barron, ‘‘Chemical Vapor Deposition of Insulating Thin Films,’’ in The
CVD of Nonmetals, Edited by W. S. Rees Jr. VCH, Weinheim, Germany, 1996.
W. Koh, S.-J. Ku, and Y. Kim, ‘‘Chemical Vapor Deposition of Al O Films
2 3
Using Highly Volatile Single Sources,’’ Thin Sol. Films, 304 [1–2] 222–4 (1997).
B. D. Fahlman and A. R. Barron, ‘‘CVD of Chromium-Doped Alumina
Ruby’ Thin Films,’’ Chem. Vapor Depos., 7 [2] 62–6 (2001).
10
B. D. Fahlman and A. R. Barron, ‘‘Substituent Effects on the Volatility of
Metal B-Diketonates,’’ Adv. Mater. Opt. Electron., 10 [3–5] 223–32 (2000).
M. K. Templeton and W. H. Weinberg, ‘‘Adsorption and Decomposition of
5
2
3
6
5
001C, and the temperature was gradually reduced in subse-
7
quent trials as to ascertain the lowest temperature required for
deposition. Trials with varying substrate positioning and pre-
cursor flow rates were run to determine the precursors along
with the water vapor introduced at flow rates of 40 and 10 mL/
min, respectively. The flow was stopped and the furnace turned
off after 1.5 h for all trials. Characterization of thin films on
silicon wafers and carbon fibers was performed using a JEOL
JSM-840A scanning electron microscope (JEOL USA, Peabody,
MA) equipped with a WinEDS detector, capable of detecting
elements ꢀ B. Film thickness measurements were obtained us-
8
9
‘
11
Dimethyl Methylphosphonate on an Aluminum Oxide Surface,’’ J. Am. Chem.
Soc., 107 [1] 97–108 (1985).
M. K. Templeton and W. H. Weinberg, ‘‘Decomposition of Phosphonate Esters
Adsorbed on Aluminum Oxide,’’ J. Am. Chem. Soc., 107 [4] 774–9 (1985).
12
&