J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The ElectrochemicalSociety, Inc.

1523

Manuscript submitted Feb. 12, 1990; revised manuscript Denki Kagaku, 57, 829 (1989).
received Dec. 4, 1990. 4. J. Mizusaki and H. Tagawa, in "Proceedings of SOFC-
Nagoya," p. 65 (1989).
Osaka Gas Company, Limited, assisted in meeting the 5. T. Higuchi, M. Miyayama, H. Yanagida, S. Ohtoshi, H.
publication costs of this article. Ohnishi, S. Sasaki, K. Ishimaru, and M. Ippommatsu,
To be published.
6. J. H. Kuo, H. U. Anderson, and D. M. Sparlin, J. Solid
REFERENCES State Chem., 83, 52 (1989).
1. T. Nakanishi, Ceram. Bull. Soc. Jpn., 25, 614 (1990). 7. D. P. Karim and A. T. Aldred, Phys. Rev. B, 20, 2255
2. J. Shimoyama, J. Mizusaki, and K. Fueki, in "Proceed- (1979).
ings of 53th Chemical Society of Japan," 1-263 (1988). 8. M. Kertesz, I. Riess, and D. S. Tannhauser, J. Solid State
3. H. Yokokawa, N. Sasaki, T. Kawata, and M. Dokiya, Chem., 42, 125 (1989).

A Mathematical Model for Chemical Vapor Deposition

Processes Influenced by Surface Reaction Kinetics: Application
to Low-Pressure Deposition of Tungsten
Rikhit Arora* and Richard Pollard**
Department of Chemical Engineering, University of Houston, Houston, Texas 77204-4792

ABSTRACT
A model for the simultaneous reaction kinetics and transport processes in chemical vapor deposition (CVD) reactors
has been extended to treat deposition of materials that have a broad range of surface characteristics, e.g., sites with multi-
ple dangling bonds and adsorbates with multiple b o n d i n g configurations. The model uses the nature of the surface to de-
termine the elementary processes that can take place during growth. Rate constants for these processes are calculated
from first principles using statistical thermodynamics, transition state theory, and bond dissociation enthalpies. I n this
way, deposition rates are determined without either assuming the reaction mechanism or arbitrarily choosing any kinetic
parameter values. The utility of the approach is illustrated by modeling low-pressure CVD of tungsten from tungsten
hexafluoride and hydrogen. The treatment considers 14 species and eight reactions in the gas together with 21 species and
65 processes at the surface. The calculations indicate that, for the range of operating conditions considered, the process is
controlled by surface reaction kinetics and that gas-phase reactions are unimportant. Deposition rates and surface fluxes
predicted by the model show quantitative agreement with available experimental data. I n addition, the major reaction
pathways and rate-limiting steps are identified. This information is used to develop simplified rate expressions (still with-
out using any fitted rate constants) that give reasonable predictions for the growth rate. Sensitivity studies are performed
to assess the impact of uncertainties in species properties and rate constants on the theoretical results.

Many models have been developed to help improve our ways are predicted rather than assumed and local compo-
u n d e r s t a n d i n g of chemical vapor deposition (CVD) pro- sitions, deposition rates, and rate-limiting factors are de-
cesses. Examples range from analytic expressions for dif- termined without arbitrarily choosing the values for any
fusion across a b o u n d a r y layer (1) to computer codes for rate constants. Previously, application of the model to
multidimensional heat, mass, and m o m e n t u m transport metalorganic vapor-phase epitaxy of GaAs provided con-
(2). Some studies (3) have included elementary gas-phase siderable insight into the system behavior and yielded
reactions but, up until now, the rates of surface reactions growth rates that agree quantitatively with available ex-
have been determined empirically, i.e., using overall or perimental data (4, 5).
lumped reactions and choosing kinetic parameter values I n this paper, the treatment of elementary surface pro-
to give deposition rates that match experimental results. cesses in the model is generalized to give rate equations
This approach to the surface kinetics may be adequate for that can be applied to a broader spectrum of CVD systems.
systems that are controlled primarily by rates of diffusion For example, effects associated with the availability of
or gas-phase decomposition. However, for processes more than one dangling b o n d per surface site are taken
where surface reactions play a significant role, questions into consideration as well as the ability of a given species
arise concerning the u n i q u e n e s s of an empirical kinetic to adsorb in several different bondJing configurations and
model and its ability to describe the system behavior out- the possibility of steric hindrance. The utility of the ex-
side the range of operating conditions for which data are tended CVD model is illustrated by considering low pres-
already available. Therefore, a more fundamental ap- sure deposition of tungsten using Wle6 and H2 as reactants.
proach is needed to describe the kinetics of heterogeneous Interest in this material for contacts and interconnects in
reactions in CVD, especially since there is a trend towards VLSI circuits stems from its high conductivity and high
operating at lower pressures and temperatures where sur- melting point and its compatibility with silicon (6, 7).
face processes tend to be more important. The model for CVD of t u n g s t e n includes 14 species and
Recently, a CVD model has been developed that uses eight reactions in the gas phase and 21 species and 65 ele-
statistical mechanics, transition state theory, and bond dis- mentary processes at the surface. Theoretical predictions
sociation enthalpies to obtain the rate constants for ele- are compared with experimental data for a wide range of
mentary processes in the gas phase and at the growing sur- operating conditions and the rate..controlling steps and
face (4, 5). The calculations treat the m a n y plausible d o m i n a n t intermediate species are identified. Sensitivity
simultaneous chemical reactions and the finite rates of ad- studies are used to assess the impact of uncertainties in
sorption and desorption on different types of sites as well species properties and rate constants on the predicted dep-
as the fluid flow of the gas mixture and m u l t i c o m p o n e n t osition rates. Furthermore, results from the complete
heat and mass transport. I n this way, major reaction path- model are used to obtain simplified kinetic expressions
that yield accurate predictions for the deposition rate and
* Electrochemical Society Student Member. that still use parameter values determined from first prin-
** Electrochemical Society Active Member. ciples.
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1524 d. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

Table I. Standard heats of formation AH~,z98, standard entropies S~98, and heat capacities Cp for gaseous species in CVD of tungsten using WF 6
and H2 as reactants. Values for a, ~, % and ~ were obtained by curve-fitting heat capacities for the temperature range 300-700 K.

AH~,298 S~~ C. = a + ~T + ~/T2 + 5T -2 (cal tool -1 K -1)

Species (kcal mo1-1) (cal mo1-1 K -l) a 10313 10s~ l'0-s~ Ref.

H2 0.0 31.2 7.66 - 1.76 0.14 0.32 (9)

H 52.1 27.4 4.97 0.0 0.0 0.0 (9)
F2 0.0 48.4 6.32 5.35 0.30 -0.15 (9)
F 18.9 37.9 6.00 -1.88 0.10 -0.08 (9)
HF -65.1 41.5 7.20 -0.83 0.83 -0.06 (9)
WF 6 -411.5 81.5 28.1 20.0 -1.20 -4.05 (9, 10)
WF5 -309.1 81.2 27.5 13.1 -0.78 -3.38 (10)
WF4 -222.0 77.7 19.3 13.6 -0.82 -1.91 (10)
WF 3 - 121.2 70.0 15.7 8.68 0.53 - 1.81 (10)
WF 2 -20.6 67.3 12.9 6.32 -0.38 -0.92 (10)
WF 92.3 59.9 7.35 3.33 -0.19 -0.37 (9)
W2F 8 -572.2 231.0 39.8 27.0 -1:63 -5.36
W2F10 -746.2 238.0 48.4 34.0 -2.06 -6.78
HWFs -374.! 81.4 23.6 24.7 -1.34 -3.59

M o d e l Description c o u n t s for t h e p o s s i b i l i t y t h a t d i f f e r e n t t y p e s o f s u r f a c e
T h e r a t e o f p r o d u c t i o n R~ o f e a c h s p e c i e s i i n t h e gas s i t e s exist, e.g., l e d g e sites a n d p l a n a r sites (4), a n d t h a t
p h a s e is g i v e n b y each type m may have a different total surface concentra-
t i o n CTm a n d a d i f f e r e n t n u m b e r ~m o f d a n g l i n g b o n d s avail-
nrg a b l e p e r site. If~m > 1, m o r e t h a n o n e s p e c i e s c a n a d s o r b at
R[ = - ~ (Av)ilr~ [1] a g i v e n site a n d , t h e r e f o r e , e a c h d a n g l i n g b o n d s h o u l d b e
l=l r e g a r d e d as a v a c a n c y .
I n g e n e r a l , s o m e o f t h e a d s o r b a t e s m a y b e a b l e to e x i s t i n
w h e r e (Av)u = Vr,il - [Link]. T h e n e t r a t e r] o f r e a c t i o n 1 is b a s e d s e v e r a l d i f f e r e n t b o n d i n g c o n f i g u r a t i o n s , i.e., to l i n k w i t h
o n t h e l a w o f m a s s a c t i o n , viz. d a n g l i n g b o n d s at d i f f e r e n t n u m b e r s (nv)jm o f a d j a c e n t
a t o m s i n t h e solid. F o r e x a m p l e , o n c e r t a i n m a t e r i a l s , a h y -
r] ~, ~t ( , ~ , ~ f -i l,
d r o g e n a t o m c a n e i t h e r b o n d to a s i n g l e site or f o r m a
---- -- piVb,il [2]
'~bl ~ pll_ll pl i=1 / b r i d g e b o n d w h e r e t h e e l e c t r o n f r o m h y d r o g e n is s h a r e d
b y t w o a d j a c e n t s u r f a c e a t o m s o f t h e solid (20-22). T h e s e
T h e m a g n i t u d e s of t h e e q u i l i b r i u m c o n s t a n t s K~I are cal- a d s o r b a t e s h a v e d i f f e r e n t p r o p e r t i e s a n d , t h e r e f o r e , it is
c u l a t e d (8) u s i n g t h e t h e r m o c h e m i c a l p r o p e r t y v a l u e s of n e c e s s a r y to r e g a r d t h e m as i n d e p e n d e n t s p e c i e s i n t h e
t h e g a s - p h a s e s p e c i e s (see T a b l e I). W h e r e v e r p u b l i s h e d c a l c u l a t i o n s . N o t e also t h a t a m o d e l w h i c h o n l y c o n s i d e r s
t h e r m o c h e m i c a l d a t a a r e n o t available, t h e s t a n d a r d h e a t s t h e t o t a l n u m b e r o f d a n g l i n g b o n d s p e r u n i t area, i.e., t h a t
of formation are evaluated using bond dissociation enthal- t r e a t s t h e p r o d u c t C ~ m as o n e q u a n t i t y , w o u l d i n c o r r e c t l y
p i e s (see T a b l e II) a n d t h e h e a t c a p a c i t i e s Cp a n d s t a n d a r d a l l o w a n a d s o r b a t e to f o r m a b r i d g e b o n d to a s i n g l e sur-
e n t r o p i e s S~98 a r e c a l c u l a t e d u s i n g s t a t i s t i c a l m e c h a n i c s face a t o m .
(13-16). I n t h i s p r o c e d u r e , t h e v i b r a t i o n a l c o n t r i b u t i o n s to I f ~m > 1 a n d a n a d s o r b a t e f o r m s b o n d s w i t h m o r e t h a n
Cp a n d S~98 a r e d e t e r m i n e d f r o m a set o f c h a r a c t e r i s t i c fre- o n e a d j a c e n t s u r f a c e a t o m , t h e a c t i v i t i e s of a n y r e m a i n i n g
q u e n c i e s (see T a b l e III) a n d t h e r o t a t i o n a l c o n t r i b u t i o n s v a c a n c i e s or a n y s i n g l y b o n d e d a d s o r b a t e s t h a t are p r e s e n t
a r e o b t a i n e d f r o m m o m e n t s o f i n e r t i a w h i c h , i n t u r n , de- o n t h o s e sites a r e r e d u c e d b e c a u s e t h e i r p o s i t i o n s b e c o m e
p e n d o n m o l e c u l a r g e o m e t r y , b o n d l e n g t h s , a n d b o n d an- " f r o z e n " and, h e n c e , t h e y c a n n o l o n g e r i n t e r a c t w i t h m o r e
gles (9-11, 17, 19). T h e e l e c t r o n i c c o n t r i b u t i o n is c a l c u l a t e d t h a n a s i n g l e n e a r e s t - n e i g h b o r site. U n d e r t h e s e c i r c u m -
u s i n g t h e n u m b e r o f u n p a i r e d e l e c t r o n s (13). s t a n c e s , t h e p r o b a b i l i t i e s fvm a n d fjm t h a t d a n g l i n g b o n d s
A t t h e s u r f a c e , t h e n e t r a t e o f p r o d u c t i o n Rk o f s p e c i e s k and singly bonded adsorbates are oriented in positions
is g i v e n b y t h a t will a l l o w r e a c t i o n s to o c c u r b e c o m e less t h a n u n i t y
(see below). F u r t h e r m o r e , s t e r i c h i n d r a n c e m a y also pre-
nrs v e n t c e r t a i n r e a c t i o n s f r o m t a k i n g p l a c e a n d t h i s c a n b e ac-
Rk = ~ (AS)klrl [3] c o u n t e d for b y m o d i f y i n g t h e s e s t a t i s t i c a l factors. W i t h t h e
l=l
introduction of such probabilities, three different relation-
s h i p s are n e e d e d to d e s c r i b e t h e a c t i v i t i e s ak i n Eq. [4]: (i)
w h e r e (hS)kl = [Link] -- Sb,~ a n d
for a g a s e o u s s p e c i e s , ak = Pi w h e r e Pi is t h e p a r t i a l p r e s -

rl=kbl~lmCTm(Kal[]k akSf'kl--~k akSb'kl) [4]

Table III. Vibrational frequencies for gas-phase species.
I n Eq. [4], t h e s u b s c r i p t k r e f e r s t o v a c a n c i e s as w e l l as spe- Degeneracies are given in brackets. For W2Fs, W2F10,and WFsH,
cies i n t h e gas a n d o n t h e surface. T h i s r a t e e x p r e s s i o n ac- bending frequencies are estimated using COFWFand ratios of reduced
masses (13, 17), coww is assumed to be equal to the stretching
frequency for Wz, and C~wHis taken to be the same as the frequency
for WH. Stretching frequencies for W2 and WH are, in turn,
Table II. Gas-phase bond dissociation enthalpies D~ used to calculated using techniques appropriate for diatomic molecules (18).
evaluate standard enthalpies of formation for intermediate species.
The value for W-H is calculated using Pauling's method (11).
Species toL (cm -1) Ref.
Bond Y-Z D~ (kcal mo1-1) Refi H2 3916 (9)
F2 917.9 (9)
H-H 103.3 (9) HF 4138 (9)
WFs-F 121.0 (10) WF6 769, 670(2), 712(3), 256(3), 322(3), 216(3) (10)
WF4-F 106.0 (10) WFs 800, 700, 650(2), 600, 300, 280(2), 270(2), 160(2) (10)
WF3-F 120.0 (1O) WF4 730(3), 700, 180(3), 160(2) (10)
WF2-F 120.0 (10) WF3 700, 650(2), 300, 250(2) (10)
WF-F 132.0 (10) WF2 735, 670, 160(2) (10)
H-F 135.9 (9) WF 726.5 (10)
W-H 74.1 W~F8 707(8), 244, 194(4), 261(9)
W-W 128.0 (12) W2Flo 707(10), 244, 194(5), 261(12)
F-F 37.9 (9) HWF5 2133, 823, 670(2), 712(3), 256(3), 322, 216(3)

Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 d. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1525

Table V. Vibrational frequencies and degeneracies for major surface

species. Values far 1H(a) and 2H(a) (21, 22, 31, 32) are
experimental. The frequencies for 1F(a), zt'(a), 4F(a), and the surface-
adsorbate bonds in 2WF~(a) (2 -< x -< 6), are estimated using
empirical relations (21, 25). For the 2WF~(a) species, stretching and
bending frequencies of W--F bands not in contact with the surface
are taken to be equal to the corresponding gas-phase values.

Species ~L.(cm -I)

4 1H(a) 2133, 804(2)
~H(a) 1050, 1290, 645.2
ZWF6(a) 707(4), 534(2), 314(2), 261(4), 227(5), 129(2), 81(2)
2WF~(a) 707(3), 534(2), 314(2), 261(3), 227(4), 129(2), 81(2)
2WF4(a) 707(2), 534(2), 314(2), 261,227(4), 129(2), 81(2)
zWF~(a) 707, 534(2), 314(2), 22'7(3), 129(2), 81(2)
iF(a) 707, 194(2)
Fig. i. Schematic diagram of (100) surface for CVD of tungsten from 2F(a) 538, 679,331
WF 6 and H2. Some typical adsarbates are shown; clear, black, and 4F(a) 707(3)
shaded circles denote, W, H, and F atoms, respectively. Fluorine atoms
can occupy on-top, bridge, and hollow sites whereas hydrogen can only
occupy on-top and bridge sites (20-22, 25). A tungsten atom is present half-plane b e l o w m i g h t be t h o u g h t of as ledges. H o w e v e r ,
at the apex of each pyramid. The lines represent bonds; broken lines c o n s i d e r a t i o n of t h e i n t e r a t o m i c d i s t a n c e s and the posi-
are used if the bonds have been covered by a tungsten atom in the half- tions of t h e d a n g l i n g b o n d s s h o w s that n o n e of t h e s e l e d g e
plane above. Individual dangling bonds or orbitals are not shown. Note sites offer a n y a d v a n t a g e for d e p o s i t i o n o v e r planar sur-
that, for typical sets of processing conditions, minor vertical displace- face sites. Therefore, for CVD of (100) t u n g s t e n , t h e solid
ments of the surface atoms may occur (26) but the overall extent of sur- interface can be m o d e l e d u s i n g only one t y p e of site and
face reconstruction is not significant (20). t h e s u b s c r i p t m can be d r o p p e d .
A d s o r p t i o n / d e s o r p t i o n of e a c h species at t h e g r o w i n g
surface is treated as an e l e m e n t a r y process w i t h a rate ex-
sure in t h e gas p h a s e i m m e d i a t e l y a d j a c e n t to the surface; p r e s s i o n g i v e n b y Eq. [4]. E q u i l i b r i u m c o n s t a n t s for these
(it) for an adsorbate, ak = fjmOjmw h e r e Ojmis t h e n u m b e r of steps are d e t e r m i n e d u s i n g statistical m e c h a n i c s and b o n d
b o n d s a t t a c h e d to species j d i v i d e d by the m a x i m u m n u m - dissociation e n t h a l p i e s (see T a b l e IV). T h e s e calculations
b e r of b o n d s t h a t can be f o r m e d , i.e., qm/~mCTm; (iii) for dan- r e q u i r e f r e q u e n c i e s for t h e vibrational m o t i o n s of the ad-
gling b o n d s (i.e., vacancies), ak = f ~ % m w h e r e 0vm is the sorbates a n d characteristic v a l u e s are p r e s e n t e d in Table
ratio of t h e n u m b e r of d a n g l i n g b o n d s to the m a x i m u m V. It s h o u l d be n o t e d that the mode]L c o n s i d e r s a d s o r p t i o n
n u m b e r of b o n d s that can be formed. This fractional cover- f r o m the gas p h a s e onto surface sites b u t not onto ad-
age of v a c a n c i e s O~m is related to the fractional c o v e r a g e s sorbates. In this context, it is n e c e s s a r y to specify that a
0jm of t h e a d s o r b a t e s b y d e p o s i t i n g t u n g s t e n a t o m is only r e g a r d e d as part of the
solid after it has f o r m e d b o n d s to four t u n g s t e n atoms in
t h e half-plane b e n e a t h it. E q u i l i b r i u m c o n s t a n t s for sur-
o face p r o c e s s e s o t h e r t h a n a d s o r p t i o r d d e s o r p t i o n are deter-
m i n e d by c o m b i n i n g r e l e v a n t i n f o r m a t i o n on gaseous spe-
T h e m a g n i t u d e s of ~m, CT~,J~m,f~o, and (nv)jm d e p e n d on cies and a d s o r p t i o n e q u i l i b r i u m constants, as d e s c r i b e d in
t h e c r y s t a l l o g r a p h i c characteristics of t h e g r o w i n g layer. A p p e n d i x A.
F o r the t e m p e r a t u r e s a n d p r e s s u r e s u s e d in CVD of The m o d e l for CVD of t u n g s t e n f r o m WF6 and H2 treats
t u n g s t e n , t h e d e p o s i t has a b o d y - c e n t e r e d cubic (bcc) 14 species and eight reactions in t h e gas p h a s e (see Tables I
s t r u c t u r e (23). M o r e o v e r , t h e (100) p l a n e c o n t a i n s t h e high- and VI). I n principle, several additional reactions could
est c o n c e n t r a t i o n of available d a n g l i n g b o n d s and it is t h e h a v e b e e n i n c l u d e d , b u t results for o t h e r d e p o s i t i o n pro-
m o s t c o m m o n l y o b t a i n e d surface d u r i n g polycrystalline cesses (5) s u g g e s t e d that gas-phase reaction rates w o u l d be
g r o w t h (24). F o r this orientation, e a c h surface t u n g s t e n n e g l i g i b l e for t h e tow surface t e m p e r a t u r e s and reactor
a t o m m a k e s f o u r b o n d s w i t h a t o m s in t h e solid (see Fig. 1) p r e s s u r e s typically u s e d in t h e t u n g s t e n process. More-
a n d has f o u r d a n g l i n g b o n d s available for surface pro- over, reactions [G1] and [G6] in T a b l e VI are the first steps
cesses, i.e., ~h = 4. At first sight, j u n c t i o n s b e t w e e n W in plausible r e a c t i o n s e q u e n c e s , and it was f o u n d that sub-
a t o m s in t h e u p p e r m o s t p l a n e and a d j a c e n t W a t o m s in t h e s e q u e n t steps do not n e e d to be c o n s i d e r e d unless the first

Table IV. Adsorption equilibrium constants, K,q,o = Ko,Texp (Qo{O,,}/RT), for major species on a (100) tungsten surface. Values of Ko,T are
obtained for a standard state of 1 atm. using statistical mechanics with partition functions of adsorbed and gas-phase species (15, 27). The
parameters A, B, and C in the relation for Ko,T are obtained by curve-fitting over the range 300 -< T(K) ~< 1200. Each adsorbate is taken to be
immobile and, hence, the translational contribution to the partition function is negligible. Also, none of the adsorbates listed have any rotational
contribution since (i) they have no external rotational degrees of freedom (28) and (ii) there are no internal rotational degrees of freedom for
either multi-atomic species involving bridge bonds or single atoms. The vibrational contribution to the partition function is obtained using
characteristic frequencies. It is assumed that the electronic contribution (from unpaired electrons) is negligible for adsorbates on a metal surface.
The heat of adsorption (Qo)i is equated to the total strength of the bond(s) formed by species j to the surface, e,g., (Qo)i = 1.7 x D~ - j] for
(n,) i = 4 and (Q~ = 1.5 x D~ - j] for (nv)i = 2 (25). Experimental values are used for D~ D~ (29, 30). The
superscript preceding each species denotes the number of sites to which it is bonded.

go, T = Ko,298exp (A + BT + CT 2) Qa{0v}= Qao - G(1 - 0v)

Species Ko,298 A 103B 106C 10-3 Qao]R (K) IO-3G/R(K)

1H(a) 8.61 • 10-6 1.88 -7.25 3.04 24.7 4.6

aH(a) 8.72 x 10-6 1.84 -7.14 3.08 36.9 6.9
2WF6(a) 4.22 • 10-11 -0.71 2.49 -0.43 24.9 10.1
2WFs(a) 9.64 x 10-12 -0.44 1.44 0.01 24.9 10.1
2WF4(a) 1.40 x 10-n -0.17 0.37 0.51 24.9 10.1
~NF3(a) 2.05 x 10-n -0.57 1.95 -0.21 24.9 10.1
2WF2(a) 3.41 x 10-1~ -0.22 0.62 0.24 24.9 10.1
IF(a) 2.85 x 10 -7 0.78 -3.11 1.50 33.2 0.0
2F(a) 1.43 • 10 v 1.23 -4.85 2.25 49.8 0.0
~F(a) 1.12 • 10 7 1.60 -6.27 2.82 56.4 0.0
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1526 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

Table VI. Rate constants for elementary gas-phase processes. The backward rata constant k~, is given by k~{T,p} exp (-E'o]RT). Values of the
pre-exponential factor ko' (in s-] and mo1-1 cm ~ s -1 for uni- and hi-molecular reactions, respectively) for o given temperature T and pressure p are
calculated using the thermodynamic formulation of transition state theory (13, 28, 34). For unimolecular decomposition reactions [G1]-[G5], the
activation energy E', is taken to be the dissociation enthalpy of the breaking bond (34) and the pressure dependence of k~{T, p} is evaluated
using the quantum Rice-Ramsperger-Kassel (QRRK) technique (13, 35-37). For metathesis reactions ([G6] and [G7]), E',, values are calculated by
the bond energy bond order (BEBO) method (38, 39) and, for the fluorine exchange reaction [G8], E~, is estimated using the Szab6 method (28)
with (x set to unity.

k~ {600,p}
No. Reactions p = 10-3 atm p = 10-~ atm 10 ~ • EER (K)

[G1] WF5 + F ~- WF6 2.07 x 1017 2.40 • I0 TM 60.4

[G2] 2H ~ H2 6.83 • l0 s 6.83 • I04 52.0
[G3] 2F ~.~F2 6.44 x l0 s 6.44 • I0~ 19.1
[G4] 2WF8 ~ W2F10 3.94 • 10~ 3.93 x 10~3 64.4
[G5] 2WF~ ~ W3F~ 1.67 x 10~ 1.67 x 10~3 64.4
[G6] HWF5 + H ~- WF8 + H2 1.41 • 10~~ 1.41 x 10~0 45.1
[G7] H + HF + WF8 ~ WF~ + H2 3.63 x 1013 3.63 x 101~ 44.0
[GS] WF~ + WF~ ~ 2WF~ 1.56 x 109 1.56 • 1Os 2.0

o n e is f a v o r a b l e . F o r e x a m p l e , t h e a c t i v a t i o n e n e r g i e s for t h e statistical f a c t o r s fj m a n d fv~ for t h e (100)W e x a m p l e are

u n i m o l e c u l a r d e c o m p o s i t i o n o f WF~ (1 -< x --- 6) are all d e r i v e d i n A p p e n d i x B.
q u i t e l a r g e ( w i t h v a l u e s g i v e n b y t h e b o n d s t r e n g t h s in I n t h e s t e a d y state, t h e r a t e s o f t h e v a r i o u s h e t e r o g e -
T a b l e II) a n d t h e p r e - e x p o n e n t i a l f a c t o r s are also similar. n e o u s r e a c t i o n s are s u c h t h a t
H e n c e , if t h e x = 6 r e a c t i o n is u n f a v o r a b l e , t h e x < 6 s t e p s
a r e e x p e c t e d t o b e e v e n less likely. R e a c t i o n s [G4]-[G6] a n d Rj = 0 [6]
[G8] a r e i n c l u d e d t o a c c o u n t for t h e p o s s i b i l i t y t h a t WF5 for e a c h s u r f a c e s p e c i e s j. F o r s p e c i e s t h a t p a r t i c i p a t e in
a n d WF~ g e n e r a t e d at t h e s u r f a c e (see b e l o w ) d e s o r b a n d t h e s u r f a c e p r o c e s s e s a n d t h a t also e x i s t in t h e gas p h a s e
t h e n r e a c t i n t h e gas p h a s e .
A t t h e s u r f a c e , t h e ability o f c e r t a i n s p e c i e s to a d s o r b in Ri = Ni,s [7]
different bonding configurations increases the number of T h e flux N~,~o f s p e c i e s i f r o m t h e s u r f a c e is a f f e c t e d b y t h e
i n d e p e n d e n t s p e c i e s t h a t m u s t b e c o n s i d e r e d f r o m 8 (H, F, i n l e t gas c o m p o s i t i o n a n d b y t r a n s p o r t o f s p e c i e s f r o m t h e
a n d WF= w i t h 1 -< x - 6) to 21. F o r t h e s e s p e c i e s , e l e m e n - b u l k gas to t h e d e p o s i t i o n surface.
tary surface reactions are generated based on the structure For deposition of t u n g s t e n u n d e r typical sets of operat-
of the growing surface and on w h e t h e r or not activated i n g c o n d i t i o n s , t h e ratio o f t h e m e a n f r e e p a t h in t h e gas to
complexes with meaningful bond angles and bond lengths t h e c h a r a c t e r i s t i c d i m e n s i o n o f t h e r e a c t o r r a n g e s f r o m ap-
c a n b e f o r m e d (13). H e r e it is i m p o r t a n t t o c o n s i d e r all p o s - p r o x i m a t e l y 10 -~ to 10 -~. U n d e r t h e s e c i r c u m s t a n c e s , con-
sibilities s i n c e o n e c a n n o t tell a p r i o r i w h i c h r e a c t i o n s will t i n u u m m e c h a n i c s p r o v i d e s a r e a s o n a b l e d e s c r i p t i o n for
b e m o s t i m p o r t a n t . R a t e c o n s t a n t s f o r t h e s e s t e p s are t h e n t h e t r a n s p o r t p r o c e s s e s a n d , t h e r e f o r e , N~.~c a n b e l i n k e d t o
calculated using statistical m e c h a n i c s and transition state f l u x e s i n t h e gas p h a s e t h r o u g h m a t e r i a l b a l a n c e s o f t h e
t h e o r y . With t h i s a p p r o a c h , a total o f 65 e l e m e n t a r y pro- form
c e s s e s are o b t a i n e d for t h e t u n g s t e n d e p o s i t i o n e x a m p l e
V . N~ = R~ [8]
(17). T h e s u r f a c e r e a c t i o n s t h a t p l a y a s i g n i f i c a n t role in t h e
p r o c e s s are l i s t e d in T a b l e VII t o g e t h e r w i t h t h e i r r a t e c o n - I f ng g a s e o u s s p e c i e s are p r e s e n t in t h e s y s t e m , n g - 1 m a -
s t a n t s . A c t i v a t e d c o m p l e x e s for f o u r o f t h e m a j o r e l e m e n - terial b a l a n c e e q u a t i o n s are n e e d e d , t o g e t h e r w i t h t h e rela-
t a r y s t e p s are d e p i c t e d s c h e m a t i c a l l y in Fig. 2. V a l u e s o f tionship b e t w e e n the fluxes

Table VII. Rate constants for major surface processes involved in deposition of tungsten by hydrogen reduction of tungsten hexafluoride. The
backward rate constant kb, is given by ~ c, ko{T} exp (-EoI{O,}/RT) with kb, in real cm -2 otto -~ s-I where n is unity for reactions [$1] and [$8] and
is zero otherwise. The partition function formulation of transition state theory (28) is used to estimate/(0{7"} and values are listed for "Is =
300~ Also, the coefficients A and 13 in the approximate relation ~1 c, ko = A ~ are given for 300 -< T,(~ -< 600 where ~1 = 4 and c, =
1.673 • 10 -9 real cm -2. The structure of each activated complex is postulated by applying o coherent set of ground rules, e.g., the complex is
mobile/tight for gas-surface reactions and immobile/tight for surface-only reactions (40). The activation energy for the backward step Eo is set
equal to AHI + hHphr, where AH, is the standard enthalpy AHR, of reaction for endothermic reactions and is zero otherwise. The quantity hHphy,
accounts for interactions in the physically adsorbed layers and it is taken to be 4 kcal mol -~ (41). The different types of elementary processes
considered include associative desarptian IS I], adsorption [$2]-[$7], gas/solid reaction [$8], and surface-only reactions [$9]-[S 16]. Note that
reactions [$8] and [$13] are catalyzed by vacancies, viz., two additional vacancies are required for these processes to proceed in either
direction. Reactions [$11] and [$12] appear to be identical but they proceed via different activated complexes, as illustrated in Fig. 2. Note that
an additional 49 reactions were included in the analysis (17), but they are not listed here since they were found to be unimportant over the
range of operating conditions considered.

Csko = A T ~ Ea = C + D(1 - 0v)

No. Reaction A [~ ~ Csko{300} C D
[$1] 2 2H(a) ,~ H2 + 4v 4.72 x 102 -0.96 1.97 • 10~ 2.0 0.0
{$2] WF6 + 2v ~ ~WF6(a) 1.06 x l0 T 0.68 4.99 x 108 26.9 -I0.I
[$3] WF5 + 2v ~ 2WFs(a) 1.26 x 108 0.05 1.65 x 108 26.9 -10.1
[$4] WF4 + 2v ~- 2WF4(a) 1.39 x 10a [Link] 1.42 • l0 s 26.9 -10.1
[$5] HF + 3v ~ 1H(a) + 2F(a) 1.63 • 105 0.89 2.46 x 107 8.3 -5.9
[$6] HF + 3v ~ 2H(a) + 1F(a) 1.01 • l0 s 0.50 1.69 x 107 2.0 0.0
[$7] HF + 6v ~ 2H(a) + aF(a) 5.35 x 104 1.11 2.94 x 107 26.7 -6.9
[$8] HF + 2H(a) ,~ H2 + 2F(a) 7.93 x 102 -0.33 1.21 x 103 2.0 0.0
[$9] 2F(a) + 2WFs(a) ~,~2WFs(a) + 2v 5.27 x 103 1.42 1.67 x 106 13.2 0.0
IS10] 2F(a) + ~WF4(a) ~ 2WFs(a) + 2v 1.47 • 103 1.70 2.31 x 106 5.5 0.0
[Sll] 2F(a) + 2WF3(a) ~ zWF4(a) + 2v 3.18 x 102 1.36 7.47 • 108 12.5 0.0
[S12] 2F(a) + 2WF3(a) ~,~2WF4(a) + 2v 1.73 • 10a 0.91 3.09 • l0 s 12.5 0.0
[S13] W(s) + 2F(a) + 2 1F(a) ~ 2WF~(a) + 2v 7.30 x 103 0.83 8.01 • l0 s 2.0 0.0
[S14] 1H(a) + v ~ 2H(a) 2.22 x 10s 0.75 1.62 x l0 s 14.2 2.3
[S15] 'F(a) + v ~-2F(a) 2.12 • 103 0.80 2.02 • 108 18.6 0.0
[$16] 2F(a) + 2v ~- 4F(a) 4.49 x 103 0.62 1.53 x 108 8.5 0.0

Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. E/ectrochern. Sac., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1527
ng Table VIII. Lennard-Jones force constants for gas-phase species.
MiNi = cv [9] These quantities are used in the evaluation of viscosities, thermal
i=]
conductivities, binary diffusion coefficients, thermal diffusion
For an ideal gas mixture, the fluxes of the species are re- coefficients, and pressure corrections for unimolecular reaction rates
(36, 37, 42, 43, 45). For H2 and HF, values of G and e/k are based
lated to their mole fractions x~ in the gas phase using n g - 1
on experimental data. For the other species, values are estimated
Stefan-MaxweU equations
using the methods presented in Ref. (46), or by related techniques.
(XiSk XkNi)
VXi -- kiTV In T = ~ [10] Species ~ (A) eJk (K) Ref.
k~ C~ik
H2 2.915 38.0 (45)
together with the requirement that the mole fractions sum H 2.708 37.0 (46)
to unity. Far from the surface, the species inlet composi- F~ 3.357 112.6 (46)
tions x~ are specified as the b o u n d a r y conditions. At the F 2.968 112.6 (46)
3.148 330.0 (46)
surface, Eq. [3]-[7] and [t0] are c o m b i n e d to obtain the cam- HF
WF~ 5.159 343.0
positions 0i and xi. WF5 5.003 452.0
Temperature variations from the inlet gas to the heated WF4 4.847 561.0
substrate are determined using the thermal energy equa- WF3 4.691 670.0
tion. I n this equation, heat transferred by convection is WF2 4.535 779.0
W2F8 5.159 343.0
balanced by the effects of conduction and heat generated W2F~0 5.159 343.0
by interdiffusion of species and gas-phase reactions. The HWF5 5.003 452.0
terms associated with viscous energy dissipation and the
Dufour effect are not included since these effects are negli-
gible for typical sets of operating conditions (42). Radia- Ts and the inlet gas temperature T| are specified as bound-
tion from the substrate may be included readily (17) b u t it ary conditions for the thermal energy equation.
is not considered here because the surface temperature T~ The equations presented above can be used to describe
is held at a set value during a n experiment. Therefore, both any steady-state, thermal CVD process. Furthermore, the
rate constants determined for each elementary process in
CVD of t u n g s t e n are valid for any reactor configuration. It
Reactants Activated Complex Products is only when the equations of continuity and motion are in-
troduced to calculate the velocity for Eq. [9] that one must
specify a particular reactor geometry. A n impinging-jet
H system is considered here since it has the advantage that
the transport processes are i n d e p e n d e n t of radial position
in the region close to the center of the flow field and,
hence, the governing equations for heat, mass, and mo-
m e n t u m transport can be transformed into ordinary differ-
ential equations (43). Details of the transformation for lam-
inar flow of an isobaric, Newtonian fluid have been
SI:H~z+4V ~ 2 zH(a) presented previously (42-44). Relationships for the tem-
perature and composition dependences of physical param-
eters needed for the model are based on the Chapman-
Enskog kinetic theory of dilute gases (45) and values are
calculated using experimental or theoretical Lennard-
Jones force constants (see Table VIII).
Results and Discussion
Figure 3 shows t u n g s t e n deposition rates for a range of
surface temperatures. Theoretical predictions that include
2v surface reaction rates based on kinetic parameters calcu-
$8: H2+ 2F(a) ~ HF + ZH(a)
lated from first principles agree closely with the experi-

T s (~

400 350 300

, =

10 4

Sl 1 : WF~(a) + 2 v @ WF3(a) + iF(a)

.E
E
E
10 2

100 1 I I
1.5 1.6 1.7 1.8
Sl 2: WF 4 (a) + 2 v ~ WF3(a) + iF(a)
103ITs (K)
Fig. 2. Postulated reaction trajectories for four elementary processes
on a (100)W surface. The schematic diagrams show the reactants, the Fig. 3. Dependence of tungsten deposition rate r, on surface tem-
activated complex, and the reaction products, with solid lines repre- perature T,. 9 - data (47). Operating conditions: reactor pressure,
seating normal bonds and dotted lines representing bonds that are p= = 1 Tort; ratio of partial pressures of recfctants, p~2/p~vF6= ] 5. Solid
being either farmed or broken. The clear, black, and shaded circles de- line - predicted using the complete CVD model. Dashed line ~ calcu-
note, W, H, and F atoms, respectively. Kinetic parameters values for lated assuming that local equilibrium is attained for the surface reac-
the reactions are presented in Table VII. tions.
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1528 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

mental data and show an overall activation energy of ap- lated as a rate-limiting reaction (48), will not occur because
proximately 20 kcal mol-L The model and data give values hydrogen does not exist on the surface as a molecule; dis-
for rs that are more than two orders of magnitude smaller sociative adsorption to give 1H(a) or 2H(a) is energetically
than the hypothetical m a x i m u m (dashed line), a limit that more favorable (49, 50). Similar energy considerations for
could only be achieved if the rate constants for all the sur- HF suggest that this molecule will also be dissociated on
face reactions were sufficiently large to establish local the surface. Reactions of the type
equilibrium at the gas/solid interface. In this limit, the dep- 2WFx(a) + YH(a) ~ 2WF=-I(a) + HF(g) + y v
osition rate is dictated by (i) the rate of mass transport of
the reactants to the surface and (ii) the surface composi- 4 <_x<_6, y = l o r 2
tions (which are, in turn, constrained by chemical ther- have large activation energies for y = 2 and are also unfa-
modynamics). The m a x i m u m deposition rate is almost in- vorable for y = 1 because values of the fractional surface
d e p e n d e n t of the surface temperature T~ because, as T~ is coverage 0,H are too small. Adsorption of WFx (4 -< x -< 6)
raised, the larger species diffusion coefficients are offset through a single W--F--Ws bond is energetically less favor-
by thicker b o u n d a r y layers. The substantial difference be- able than reactions [$2]-[$4] and adsorption steps for these
tween actual deposition rates and the m a x i m u m possible species involving direct W--W bonds have very small pre-
indicates that, for the operating conditions considered, the exponential factors (17). All the reactions that can take
process is controlled by surface kinetics. Hence, the place between adsorbed hydrogen and fluorine atoms, v i z .
growth rate is not sensitive to the transport processes, e.g.,
r~ does not depend markedly on the gas flow rate. More- HF(g) + (m + y) v ~ "~H(a) + YF(a)
over, the fractional utilization of WF~ is quite low in the ex- m= 1, y = 1, 2, or 4, and m = 2, y = 1 o r 4
periments considered and the reactant compositions are
fairly uniform across the b o u n d a r y layer (see below). were included in the calculations. However, with the ex-
U n d e r these circumstances, it is meaningful to compare ception of reaction [$6] these processes were found to be
theoretical and experimental results even though the reac- u n i m p o r t a n t because either the activation energy was too
tor configurations are not the same in the two cases. large or the fractional coverages of the atoms were too
For the range of temperatures and pressures considered, small.
gas-phase reaction rates are negligible and only a small The deposition rate is controlled by removal of 2F(a)
proportion of the 65 surface processes included in the from the surface (reaction [$8]) and by conversion of
model are found to be significant. The d o m i n a n t reaction 2WF4(a) to 2WF3(a) (reactions [Sll] and [S12], which are
pathways predicted b y the model are presented in Fig. 4. identical except for the structure of the activated complex,
T u n g s t e n hexafluoride is transported from the inlet to the as illustrated in Fig. 2). The activation energies are 4.0 kcal
surface and is adsorbed to form aWF6(a), which has two mo1-1 for [$8] and 24.9 kcal mol ~ for [Sll]/[S12], and this
fluorine bridge bonds, as shown in Fig. 1. This species gives an overall temperature dependence for r S which
loses its fluorine atoms sequentially to give 2WF~(a), agrees with the data, as illustrated in Fig. 3. Reaction [$8]
~WF~(a), and 2WF~(a). The adsorbed WF~ reacts almost in- influences the deposition rate even though it is not part of
stantaneously to give t u n g s t e n solid by reaction [S13]. the t u n g s t e n decomposition pathway because 2F(a) is in-
Note that only a small fraction of each ~r162 species de- volved in the m a i n elementary processes. I n particular,
sorbs rather than decomposing further on the surface. faster removal of 2F(a) by reaction [$8] would tend to lower
Table VII lists the major surface processes, b u t it is 02F which would increase both the thermodynamic driving
worthwhile to consider some of the reactions that were force for production of 2WF4(a) and the fraction 0v of vacant
found to be u n i m p o r t a n t since m a n y of them seemed quite sites, and these changes would make the driving force for
likely before the rate constants were calculated. For exam- reaction [S 11]/[S 12] larger.
ple, fluorine exchange reactions between gaseous Deposition rates predicted for various inlet gas composi-
t u n g s t e n fluorides and the surface, v i z . tions agree closely with experimental data available in the
literature (see Fig. 5 and 6). A dependence ofr~ on p ~ is ex-
2WF=(a) + y v ~- ~WF~_~(g) + ~F(a) pected because hydrogen participates in reaction [$8]. One
4<-x<-6, y = 1, 2, or 4 might also expect P~F6 to affect the deposition rate since
reaction [S 11]/[S 12] involves t u n g s t e n atoms that originate
are possible, b u t they have relatively large activation ener- from WF6. However, the dependence of p ~ is fairly weak,
gies and small preexponential factors (17). Dissociation of even at small values of P~w6(see Fig. 6), because changes in
molecularly adsorbed hydrogen, which has been postu- surface-compositions are attenuated relative to changes in
the adjacent gas phase especially at larger values of P~F6
Gas-Phase
(see Fig. 7). This behavior can be attributed to local equi-
spedes librium being established for adsorption of WF6 and for
conversion of aWF6(a) to 2WF4(a) via 2WFs(a). Under these
conditions, the surface compositions that appear in each
C~s-Phase rate expression can be interrelated through the corre-
S~es sponding t h e r m o d y n a m i c relationship, v i z . OvcFo ~ Ka2
.~ljacent to
the Sur~c~ Pw~6(O~)2, OwF~~ 0WF~(0~)2/K~,~O2F,and 0w~4 ~ 0wFs(0v)~/K~,I~02F-
Therefore, an increase in PWF~would not alter 0wF6as much
since it would be offset by changes in %. Similarly,
changes in 0vcF4and 0wF5are attenuated by changes in 0~
Adsorbed
Species and [Link] a result, the deposition rate is fairly insensitive
to P~VF6.
The fractional surface coverage of WF4 is found to be
Solid
greater than that for WF5 because the W--F bond strength
is smaller for WF5 (see Table II). Note that dispropor-
tionation of 2WF~(a) is not important because the probabil-
Fig. 4. Principal reaction pathways for low-pressure CVD for ity of adsorbed WF~ molecules interacting with each other
tungsten. The operating conditions ore T~ = 300~ P~2 = 0.75 Torr, is extremely small when 0WE5is SOlow. Also, the rate of dis-
and p~VF6 = 0.1 Torr. The bracketed numbers are fluxes of fluorine proportionation of WF5 in the gas phase (reaction [G8] in
atoms in units of 10 ~oreal cm -2 s-'. These fluxes do not always appear Table VI) is found to be negligible for the range of operat-
to balance exactly because (i) the values have been rounded off and (ii) ing conditions considered. The prediction that 0WF4> 0WF5
several minor pathways are not shown. For example, some 2WF~(a) is consistent with mass spectroscopic data obtained dur-
is lost from the surface but the fraction that desorbs is 4 x 10 -s, ing temperature programmed desorption studies (51).
6 • 10 -4, and 2 x 10 -8 for x = 5, 4, and 3, respectively. The deposi- The surface compositions predicted for 16 of the species
tion rate in this example is 9.1 • 10 -9 real cm -2 s-1. Arrows: included in the calculations are not shown in Fig. 7 be-
(~) rate-limiting steps; ( ~ ) steps that do not signifi- cause their fractional coverages are less than 10-7. How-
cantly affect the deposition rate; (-~) steps close to equilibrium. ever, the m a g n i t u d e of 0~for a given species should not be

Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Sac., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1529

10 2 10 ~

/ []

T: .sE

10 1 , , , , , , , , , , 10 0 , , , , , , , , I

10 q 10 0 10 q 10 0

P~12(torr) P~2(t~

Fig. 5. Dependence of tungsten deposition rate on the inlet partial pressure of H2. Operating conditions: (a, left) T, = 378~ p~vr6= 0.3 Tort, ( I )
data (47); (b, right) 7", = 300~ p~w6 = 0.1 Torr, (R) data (48). Solid lines ~ predicted using the complete CVD model.

102 . . . . . . . . ,

#-.
41
E
F=
=E

3 I I t I I 101 , , , , , , , , I

0.0 0.02 0.04 0.06 0.08 0.10 10 -1 10 o

P~NF6(tOrr) P~NFs(torr)

Fig. 6. Dependence of tungsten deposition rate on the inlet partial pressure of WF 6. Operating conditions: (a, left) T, = 300~ P~2 (Torr): (O)
0.75; (1~) 1.0; data (48). (b, right) T, = 392~ P~2 = 0.5 Torr, ( I ) data (47). Solid lines ~ predicted using the complete CVD model.

regarded as a measure of its importance in the process. For (i.e., O4F ~ 0.1, as shown in Fig. 7) yet the pathway diagram
example, all the tungsten is deposited via 2WF3(a) but 0w~3 (Fig. 4) indicates that these q u a d r u p l y b o n d e d fluorine
is of O(10 -12) because reaction [S12] is very fast. Also, 0~Fis atoms are only p r o d u c e d by a side reaction. These results
of O(10 -~3) even though ~F(a) represents a major pathway imply that experimental identification of the dominant ad-
for fluorine atoms (see Fig. 4) and 01 s is of O ( 1 0 -8) despite sorbed species will not necessarily' provide insight into the
the participation of ~H(a) in formation of H F (see Fig. 8). On major reaction pathways.
the other hand, a p p r o x i m a t e l y 40% of the available dan- For the gas phase adjacent to the growing surface, the
gling b o n d s are attached to fluorine atoms in hollow sites mathematical m o d e l calculates local velocities, tempera-

0.10
&
0.4
8 0.08 jJJ
pWF s 4 J
1=
jS j
0.3
0.06
.o 8
SSJJ
ss S 0.2
"c" 0.04
m
ss S 0m
s" OWFs

s IJ ~ o.1
p 0.02
o_ lOaewF -
1038=F

I 104OWFs 0.0
0.0 0.02 0.04 0.06 0.08 0.10
0.0
0.0 0.02 0.04 0.06 0.08 0.10
P~vF6(torr)
p~%(torr)

Fig. 7. Influence of inlet partial pressure of WF6 on the fractional occupancy of sites at a growing (IO0)W surface. Operating conditions: T, =
300~ P~2 = 0.75 Torr; (a, left) comparison of surface coverages for ZWFx(a) (4 <- x -< 6) with the WF 6 partial pressure immediately adjacent to the
surface (dashed line); (b, right) surface coverages for other major species.
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1530 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

Gas-Phase p~(torr)
Species
0.0 0.4 0.8 1.2 1.6 2.0

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Ts = 623K

10 -7
Gas-Phase
Species o
Adjacent to
the Surface z:
Q.

10 -8
Adsorbed
Spedes

Fig. 8. Principal reaction pathways involving hydrogen for low-pres- 0.0 0,02 0.04 0.06 0.08 0.10
sure CVD of tungsten. Operating conditions are as given in Fig. 4. The
bracketed numbers are fluxes of hydrogen atoms in units of 10 -m tool p~%(torr)
cm -z s=1. The arrows have the same significance as in Fig. 4.
Fig. 10. Influence of operating conditions on the sum of the partial
pressures of WF s and WF 4 immediately adjacent to the growing
tures, and compositions. Typical partial pressures for the tungsten surface. The effect of P~F6 is shown for P~2 = 0.75 Torr and
most a b u n d a n t species are presented in Fig. 9. Profiles for the effect of P~2 is shown for p~w6 = 0.1 Torr.
the reactants are fairly uniform, in keeping with the pre-
dicted slow growth and kinetic control for the process.
Gas-phase intermediates and reaction products diffuse energy for deposition. The model also predicts the com-
away from the surface and their compositions decrease bined flux N ~ of WF4 and WF~ from the surface. For the
monotonically with the distance z. Thermal diffusion, P~VF6values given in Fig. 10 and T~ = 300~ N4/~ranges from
which has been included in the calculations, hinders mo- 1.2 • 10t2-3.7 • 10TM molecules cm -2 s -1, which can be com-
tion of large molecules to the hot substrate and vice versa. pared with the reported value (51) of 2 • 1012 molecules
For example, with the operating conditions specified in cm -2 s -1 obtained by extrapolation of high-temperature
Fig. 4 and 9, thermal diffusion lowers PwF~immediately ad- data. Overall, the results suggested that, if these interme-
jacent to the surface b y 23.0% and raises the corresponding diates are indeed responsible for u n w a n t e d nucleation and
value of PH2 by 2.4%; the effect is smaller for H~ since it has growth of t u n g s t e n on silica or other surfaces, and if the
a larger partial pressure. Despite these changes, thermal gas-phase composition adjacent to the main growth sur-
diffusion reduces r~ by less than 1% for the range of operat- face provides a meaningful indication of the overall partial
ing conditions considered because the shifts in composi- pressure of subfluorides in the reactor, slow growth of
tion offset each other. Note that this balance is obtained t u n g s t e n would be preferable. However, even if the magni-
even though the effect on PWF6is m u c h larger than that on tude of (PwF4 + PwF~)dictates the selectivity, other factors,
P~2 because the process is relatively insensitive to the WF~ e.g., productivity, deposit uniformity, and grain size, are
composition, as illustrated b y Fig. 6. likely to impose additional constraints that m u s t be taken
An important issue in CVD of t u n g s t e n is deposit selec- into consideration when optimizing the process.
tivity, and it has been suggested that loss of selectivity is For typical operating conditions, the total gas flow rate
directly related to the amounts of t u n g s t e n subfluorides has a relatively small effect on the rate of t u n g s t e n deposi-
such as WF 4 and WF 5 generated at the growing surface (51). tion because the process is controlled primarily by reac-
Figure 10 summarizes theoretical predictions for the effect tion kinetics. For example, the model shows that doubling
of several operating conditions on the combined partial the inlet mass flow rate for the lower curve in Fig. 6a would
pressures of these species in the region adjacent to the sur- increase rs less than 8% and that doubling it again would
face. The quantity (PwF4 + PWFs)is smallest at low values of have even less effect, increasing r~ by approximately 5%.
Ts, P~w6,and P~2. The temperature dependence yields an ef- This result suggests that uncertainties in the flow rate are
fective activation energy for subfluoride desorption of ap- not critical in the t u n g s t e n CVD process, l~evertheless, in
proximately 20 kcal tool -1 (22 kcal mo1-1 for WF5 alone), general, it is beneficial to stipulate the flow rate because
which agrees closely with experimental observations (51). the relative importance of mass transport and kinetics de-
Broadly, this value reflects the difference between the heat pends on the operating conditions. Furthermore, the
of desorption of the subfluorides and the overall activation method used to i m p l e m e n t changes in processing condi-
tions may be important, especially if a large range of pro-
cessing variables is considered. For example, in Fig. 6b,
the calculations assume [based on information provided
10 0 on the mass flow controllers used in the experiment (47)]
that the H2 flow rate is fixed at 75 sccm for P~F6 < 0.75 Torr
WF6 and that the WF6 flow rate is set at 112.5 sccm for P~F6 >
0.75 Torr. If, instead, the H2 or WF6 flow rates are fixed for
all values of p~cF~,the shape of the rs vs. p~a~ diagram would
10 .4 be slightly different (compare curves A and B in Fig. 11).
O
I n these cases, the total mass flow rate changes across the
diagram. A third possibility is to keep the total flow rate
the same and vary the flow rates of both reactants simulta-
10 -8
neously (see curve C). At a given value o f p ~ 6, the only dif-
WF 4 HF ference between cases A, B, and C is the total mass flow
rate of gas. The largest difference in rs that results is 25%,
and this is obtained in the hydrogen-starved regime (on
0.0 0.5 1.0 1.5
the right-hand side of the diagram), which would not nor-
z (cm) mally be chosen in practice.
For the range of operating conditions considered, the
Fig. 9. Composition profiles for major gas-phase species. Operating rate-limiting factors and d o m i n a n t reaction pathways for
conditions are as given in Fig. 4. The distance z is measured from the CVD of t u n g s t e n do not vary appreciably. I n general, how-
growing tungsten surface. ever, changes in operating conditions do tend to change
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1531

A p p r o x i m a t e models that i m p l e m e n t assumption (1), as-

60 sumptions (1) and (2), and assumptions (1) to (3) are re-
ferred to as models AMI, AMII, and AMIII, respectively.
Deposition rates obtained with model AMI differ from
values calculated with model DM by less than 2%. Com-
parison of results for model DM with those for models
40
AMI, AMII, or AMIII (e.g., see Fig. 12) suggests that any of
these approaches would be adequate for typical sets of op-
erating conditions. At very small values of P~F6, adsorption
of WF6 moves away from equilibrium and this makes the
dependence of rs on P~vF6 slightly stronger. Hence, under
these conditions, the slope on the 'rs-pwF6 ~ diagram is some-
what less for AMIII than AMII. Note that, with either AMII
or AMIII, the compositions of several surface species can
be eliminated from the calculation and, in each case, a sin-
gle, albeit complicated, expression for the deposition rate
10-1 100 can be used (see A p p e n d i x C). These expressions can help
to isolate which factors are primarily responsible for the
P~VFs(t~ observed trends in behavior, e.g., Eq. [C-1] and [C-4] both
show that r~ is sensitive to the magnitudes of 0v and 02F.
Fig. 11. Influence of the gas flow rate on the relationship between The presence of 02Fin the simplified rate expressions em-
/~vF6 and r~. Operating conditions: T~ = 392~ P~2 = 0.5 Torr; A ~ H 2 phasizes that it would be inappropriate to use Langmuir-
flow rate = 75 scorn; B ~ total flow rate = 150 sccm; C ~ WF 6 flow Hinshelwood or Langmuir-Rideal kinetic models (28) for
rate = 112.5 sccm. In each case, it is implicitly assumed that a valve is this system since they would involve replacing all the sur-
available to control the total pressure and, hence, to maintain a con- face compositions with the corresponding gas-phase par-
stant value of P~2- tial pressures. Furthermore, the surface reaction probabil-
ity (SRP) for species i, defined as the flux of i reacting at
the surface divided by the flux of i ,entering the reactor, de-
the relative importance of different aspects of the process pends markedly on operating conditions, e.g., as p~VF6in-
and, therefore, care is needed when interpreting data ob- creases on the lower curve in Fig. 6a, the S RP for WF6 falls
tained under new sets of conditions. For example, a recent from 2.8 • 10 -4 to 4.3 • 10 -5 and the S R P for H2 rises from
study (52) uses a very dilute quantity of WF6 with argon in 1.1 • 10 ~to 1.7 • 10 -~. Therefore, use of SRPs to model the
excess hydrogen. In this extreme, the model indicates that tungsten system would necessitate fitting their composi-
diffusion and thermal diffusion of WF6 to the surface have tion d e p e n d e n c e to available data in an empirical manner.
more influence on the deposition rate than reaction kinet- Similar problems have also been encountered in metalor-
ics (40). This result emphasizes that, over a sufficiently ganic CVD of GaAs (54) and in deposition of silicon from
wide range of operating conditions, a given CVD process is silane (55), and these difficulties reflect the inability of re-
likely to exhibit a n u m b e r of regimes with different rate- action probabilities to account for the complexity of the
limiting factors (5). Therefore, various interpretations of underlying elementary steps.
process behavior postulated for different sets of condi- Model AMIII can be simplified further because mass
tions may not be contradictory. transport is relatively u n i m p o r t an t in this system. Specifi-
A unique feature of the CVD model is that it considers cally, the flux of H F from the surface can be approximated
all the plausible elementary processes and predicts which b y DHFPHF/~, w h er e DHF is an effective diffusion
of t h e m significantly affect the system behavior and which coefficient, and 3 is the boundary layer thickness. The
do not. For example, calculations show that the theoretical magnitude of 5 can be estimated u.~ing a mass-transfer cor-
predictions for low-pressure deposition of tungsten are not relation (1, 45, 53). Incorporation of this approach into Eq.
affected in any way if the complete kinetic model (with [C-4] yields
eight gas-phase reactions and 65 surface processes) is re-
placed by a model with no gas-phase reactions and only r~ A(Ov)4(PH2)213(pwF6)I/3/k~
= [12]
the nine species and 16 surface reactions shown in Tables where k = (1 + B(0v)4k/PH2)1/3, A = (kb4, + kb,12)(C/C22) 1/3,
IV and VII. (Therefore, this combination of species and re- B = 6(kb,ll + kb.12)C3RT~/C2DHF,and
actions will subsequently be referred to as the detailed
model, DM.) There are many options for simplifying the ki-
netics further because not all 16 reactions are equally im-
portant or need to be treated in the same detail. The de-
Ov = 1 -- C60v~/~Hz -- {C4PwF6
sired approach will depend, at least in part, on the
information sought, and some possibilities are outlined
below.
(CIC2pvr
+2MI+Cs(Ov)2)[-P--P-~-2] ~131(0
) )2 v [13]
1. If one is only concerned with calculating the deposi-
tion rate, an eight species/12 reaction m o d e l may be used
because (i) 1H(a) and the reactions in which it participates Definitions for C1-C6 are given in A p p e n d i x C. If the reac-
([$5] and [S14]) can be neglected since these reactions are tant partial pressures are assumed to be uniform across the
relatively m i n o r (Fig. 8) and (it) reactions [$3] and [$4] can boundary layer, as indicated by Fig. 9, the deposition rate
be excluded since their rates are quite low. A direct conse- for a given inlet gas composition can be calculated directly
quence of item (it) is that the deposition rate can be linked using Eq. [12]. Comparison of the exponents for Pw~6 and
directly to the species fluxes, i.e. PH2 in Eq. [12] with the slopes in Fig. 5 and 6 might suggest
that Eq. [12] was incorrect. However, ~ increases as P~2 is
r~ = --NwF6.s = --3NH~.~ = 6NHF.s [11] raised and % becomes smaller as P~6 is increased, and
these dependences m a k e the effective exponents smaller
2. Rate expressions for reactions [S1], [$9], [S10], and (see curves for model AMIV in Fig. 12).
[S16] can be replaced by definitions of the corresponding Overall, the analytic expression above approximates rs
equilibrium constants since local equilibrium is estab- reasonably well considering that it uses kinetic parameter
lished for these steps. values evaluated from first principles together with a sim-
3. The rate expression for reaction [$2] can be replaced ple representation of the mass transport. For the range of
by the definition of the corresponding equilibrium con- operating conditions considered, an approximate value for
stant because this reaction only moves away from equilib- is adequate because diffusion and thermal diffusion are
rium at small values of P~6. Also, the rates of the forward u n i m p o r t an t relative to reaction kinetics, and r~ does not
steps for reactions [Sll]-[S13] can be neglected relative to change appreciably with the gas flow rate. In principle,
their backward rates. Eq. [12] and [13] can be applied to any reactor configuration
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1532 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.
i ! i ! {
1
10
DM,AM I

E .'_=
E
E E
vc

3 I I I I I 10~ 1 i , f i i i i,i I I ,

0.0 0.02 0.04 0.06 0.08 0.10 101 10 0

P~I2 (tOrr)
P~NF6(torr)

Fig. 12. Dependence of tungsten deposition rate on inlet gas composition for various kinetic models. Operating conditions and data far (o, left)
and (b, right) are as in Fig. 6a and 5b, respectively. Curves: DM is the detailed model; AMI is the eight species/12 reaction model; AMII is
Eq. [C-1]-[C-3]; AMIII is Eq. [C-4]-[C-6]; AMIV is Eq. [12] and [13] with ~ ~ 2.0 Sc-I/3 V~v/a (45, 53).

provided that the operating conditions give kinetic control program is used to model operating conditions outside the
and values for PWF~a n d PH2 are available, e.g., PwF~ ~-- P~ws range which was used to fit these parameters, the rate
and p H 2 ~ P~2 for the reactors used to obtain the-data in equation may no longer be valid and, hence, the calculated
Fig. 5 and 6 because only a small fraction of the tungsten trends in behavior may be incorrect. For example, it would
entering the system is incorporated into the growing film. be inappropriate to use the relation rs = k(PH2)l/2to assess
If the fractional utilization of the reactants is significant, the impact of local WF6 depletion on deposit uniformity
the reactant compositions in Eq. [12] and [13] cannot be re- because it omits the dependence of r~ on PwF6that mani-
placed by their inlet partial pressures. U n d e r these circum- fests itself w h e n Pw~8 is small. Naturally, when sufficient
stances, it would be more appropriate to describe the pro- data are available, an empirical relation could be extended
cess using the equations for simultaneous heat, mass, and to account for such effects. However, if experimental data
m o m e n t u m transport together with kinetic model DM, are already available for this regime, the need for the
AMI, AMII, or AMIII. I n this situation, if one wished to model might not be so great. Moreover, uncertainties
monitor depletion of reactants in the streamwise direction would arise again if the processing conditions were
or b a c k m i x i n g of gases within the deposition chamber, it changed further and, for each new combination of reac-
would be necessary to model the fluid flow and transport tants one wished to evaluate, a series of experiments
p h e n o m e n a in multiple directions. A multidimensional would be needed to survey the range of possible operating
model would also be needed if one was concerned with in- conditions and to obtain enough data for another empiri-
complete premixing of the reactants (which could lead to cal kinetic model. The fundamental method for predicting
uncertainties in local values for P~2 and PWF~,ffrespective deposition rates in CVD processes presented in this paper
of the conversion efficiency). At present, however, the avoids m a n y of the problems associated with the empirical
computer time required to solve such multidimensional approach to the reaction kinetics. In particular, its pre-
problems using the complete kinetic model would be pro- dictive capability minimizes the need for experimental
hibitive. Therefore, a n effective strategy would be to use a data and offers the potential for assessing the relative mer-
one-dimensional model of the type presented here to iden- its of various process alternatives on the computer in an ef-
tify the d o m i n a n t pathways and subsequently to incorpo- ficient manner.
rate a simplified kinetic model into the multidimensional
calculations. The choice of simplified kinetics will depend Sensitivity Studies
on the application. For example, one might eliminate reac- The kinetic parameter values for the CVD model are de-
tions [$3] and [$4] because they do not influence r~ signifi- termined from statistical mechanics and transition state
cantly, i.e., only small proportions of the 2WFs(a) and theory using a coherent set of ground rules (40). This tech.
zWF4(a) formed at the surface desorb rather than being in- nique has the advantage that reaction pathways, rate-
corporated into the growing film (see Fig. 4). On the other limiting steps, and deposition rates are predicted from first
hand, one might include these reactions since the magni- principles and it is not necessary to arbitrarily choose the
tudes ofpwF5a n d Pw~4could play a role in loss of deposit se- magnitudes of rate constants to match theory and experi-
lectivity. The behavior of the subfluorides can be deter- ment. However, the approach does involve m a n y uncer-
m i n e d directly by i m p l e m e n t i n g model DM (which has tainties, and it is important to assess the impact of these
nine surface species and 16 elementary surface processes) uncertainties on the theoretical results. Here, it should be
or indirectly by using models AMII or AMIII and back-cal- noted that recording the relative effects of changing indi-
cu-lating the rates of reactions [$3] and [$4] from the rate vidual rate constants usually just serves to confirm which
constants given in Table VII and the appropriate equilib- steps have the greatest impact on the deposition rate. It is
rium relationships. perhaps even more meaningful to determine how inaccu-
When working with a multidimensional code, one might racies in a basic quantity such as a vibrational frequency
ask whether or not a simple overall rate expression based or a bond strength affect all the other thermochemical
on the observed dependence of deposition rate on inlet gas properties and rate constants and then to use the model to
composition would be able to represent the deposition find out how these changes influence the results for the
process adequately. With an empirical approach of this CVD process u n d e r investigation. Information of this type
type, the magnitudes of any kinetic parameters involved not only quantifies our level of confidence in the theoreti-
are invariably chosen by fitting theoretical deposition cal predictions but also indicates directly which additional
rates to an existing set of data. Therefore, if the computer m e a s u r e m e n t s would help clarify our u n d e r s t a n d i n g of
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1533

system performance. Some examples of tests performed 10 I

for the t u n g s t e n example are described below. Most of
these studies use variations in the deposition rate at small
04
values o f p ~ 6 as a reference since the trends in behavior in
this region were found to be the most sensitive to changes
in property values. Also, attention was focused on proper-
ties that were expected to have a substantial effect on the 10 0 ev
system behavior rather than those used to determine rate 0.3
constants for the m i n o r reactions. Insight into the most im-
portant equilibrium constants and rate constants can be
obtained by inspection of the coefficients A~-A9 or C~-C8
defined in Appendix C.
Uncertainties in the characteristic frequencies of gas- i 0 -I I I 0.2
phase species are not expected to be critical because (i) 22 24 26
other contributions to the partition function are so large
Qao/R (K)
that the vibrational terms are usually only significant at
very high temperatures, and (i~) gas-phase reactions are Fig. 13. Dependence of tungsten deposition rate on the heat of ad-
u n i m p o r t a n t for deposition of t u n g s t e n using typical reac- sorption Qoo at zero coverage for 2WFx(a), 2 -~ x -~ 6. Deposition condi-
tor pressures and temperatures. In contrast, vibrational tions are as in Fig. 4. Note that the heat of adsorption ~ a for 2WF~(a)
contributions can be significant on the surface since trans- can be calculated using the corresponding value for ~ and the relation-
lation is either absent or small (restricted motion in a 2-d ship given in Table IV.
box) and there are no external rotations. For example, the
W-F-W~ b e n d i n g frequency (COwry = 129 cm -~) affects the
absolute magnitude of the growth rate, e.g, for the lower the deposition rates too low. Note that, despite the sensi-
curve in Fig. 6a, a 25% increase in this frequency lowers r~ tivity of r~ to Qao, the magnitude of this quantity does not
by approximately 17 and a 25% reduction raises r~ by significantly affect the trends predicted by the model.
23-34%, with the greatest change being at small values of Calorimetry and desorption experiments have demon-
P~6. However, these changes do not alter rate-limiting strated that the heat of adsorption for hydrogen decreases
steps, major reaction pathways, or trends in behavior as the surface coverage becomes larger (29, 56). This de-
(such as the shapes of the curves in Fig. 3, 5, and 6). The pendence was included in the calculations by choosing a
shifts in r~ are caused by modifications to the values of K~] value for G in the relation Qa{ev}= Q~o - G(1 - e~) to give Qa
for [$2], [$9], and [S10] which are calculated using values of values that agree with desorption data (see Table IV). The
S ~ and Cp for 2WF~(a), 2 -< x -< 6. Note that ~WFWis also used influence of % on Q~ was taken to be somewhat stronger
to obtain K~ for reactions [$3], [$4], and [$11]-[S13], b u t [$3] for zWF=(a), 1 -< x -< 6, since these adsorbates are larger.
and [$4] are u n i m p o r t a n t from the standpoint of growth Figure 14 shows that use of a constant Q~ value for these
rate and the forward rates for [S-11]-[S13] are negligible. species would give a slightly greater dependence of r~ on
Also, the magnitude of ~wFw does not affect the rate con- P~6 b u t that the overall shape of the curve would not be
stants for reactions [$2]-[$4] and [$9]-[S13] because the fre- changed markedly. For adsorbed fluorine, values of G
quency appears in the partition functions of both the acti- were set to zero because no experimental data on the effect
vated complex and the reactant, and hence it cancels out. of surface coverage were available for highly electronega-
The frequency does alter the pre-exponential factor in the tive species. However, even when G is taken'to be 5.0, 7.5,
rate constant for WF6 adsorption, but r s is insensitive to and 8.5 kcal mol -~ for ~F(a), 2F(a), and 4F(a), respectively,
this parameter because reaction [$2] is relatively close to the position of curve DM in Fig. 14 is not altered.
equilibrium. The relative heats of formation of adsorbed fluorine
Growth rates obtained using the approximate kinetic atoms are evaluated using bond order conservation theory
model AMIII seem to agree more closely with the data in (25), e.g., the total bond strengths of ~F(a), 2F(a), and 4F(a)
Fig. 12a than values calculated using the more detailed ki- are taken to be in the ratio 1:1.5:1.7. A reduction in the sec-
netics (model DM). Since AMIII assumes adsorption/ ond ratio to 1.68 is equivalent to reducing the total bond
desorption equilibrium for WF6 but DM does not, this re- strength for 4F(a) by 1.2%. With this change, the surface
sult suggests that the value for kb.2 used in the detailed coverage of 4F(a) becomes smaller and the resulting higher
model may b e t o o small. The magnitude of kb.2, as with the concentration of vacancies leads to a faster growth rate
other rate constants, is not k n o w n precisely and, therefore, (see Fig. 14). On the other hand, use of a larger ratio stabi-
it is worthwhile to assess how a change in its value would
affect the model predictions. For example, if kb,2is taken to i i i i
be twice as large as the value listed in Table VII, the reac- C
tion becomes very close to equilibrium and the r~-p~z 6
curve for model DM becomes as good as the result for
model AMIII. Adsorption of WF6 still tends to move away
from equilibrium as P~w6 is lowered. However, further in-
creases in kb,2 do not affect r~ noticeably over the range of E
compositions considered. E
The equilibrium constants for reactions [$2]-[$4] and
[$9]-[$13] are calculated using heats of formation of the
zWFx(a) species and these quantities are, in turn, deter-
m i n e d from heats of adsorption Qa. I n principle, different
Q~ values could be used for each species in the series
2 -< x -< 6 but one can only justify choosing a single value
because no experimental data on these heats of adsorption 2 i I i t i
are available. Figure 13 shows that the heat of adsorption 0.0 0.02 0.04 0.06 0.08 0.10
Qao at zero coverage m u s t lie within a small range of values
to obtain deposition rates comparable to observed results, P~F6(t orr)
and the n u m b e r listed in Table IV represents the middle of
the acceptable range. With qao > 26.5, the 2WF~(a) species
are so stable that the fraction of vacant surface sites is too Fig. 14. Sensitivity studies on the infl,,lence of/~vF~ on the tungster
small to permit growth (see Fig. 13), i.e., the surface be- deposition rate. Operating conditions and data are the same as the
comes passivated because there is insufficient space avail- lower curve in Fig. 6a. Curves: DM is the detailed model; A uses Q~ =
able for the adsorbates to decompose. On the other hand, 18.4 for 2WFx(a), 2 ~ x -~ 6 (value chosen to give rsA ~ rs.~ at P~vF6=
with small values of Q~o, the thermodynamic driving force 0.1 Tort); B uses the ratio of bond strengths for "~F(a) and 2F(a) =
for adsorption of WF~ becomes too small, and this makes 1:1.51; C uses the ratio of bond strengths for IF(a) and 4F(a) = 1:1.68.
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1534 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

lizes 4F(a) and lowers rs, e.g., rates for the lower curve in quently, the level of sophistication of the kinetic
Fig. 6a fall by approximately 55% with 1:1.5:1.72. A reduc- calculations presented here is believed to represent a prac-
tion in t h e first ratio has two main effects: (i) the equilib- tical and meaningful m e t h o d for treating the large n u m b e r
rium constant for reaction [$8] becomes smaller, which of plausible elementary steps.
tends to increase rs, and (ii) the heat of formation of 2F(a)
be c om es less negative, which tends to lower rs, because it Conclusions
increases the activation energy for the backward step in re- A mathematical m o d e l has been developed for CVD pro-
action [S11]/[S12] [Ea is equated to (AH~,298)wF3(~+ (AH~,298)2~) cesses that are influenced by surface reaction kinetics. The
o .{_
- (hHf,298)WF4(a) AHphys].The net effect is that r~ is reduced, approach uses statistical mechanics, transition state
e.g., rates for the lower curve in Fig. 6a are 34% less w h e n theory, and bond dissociation enthalpies to determine the
the ratio is 1:1.49. The impact of increasing the ratio to 1.51 rate constants for elementary processes in the gas phase
is shown in Fig. 14. These results indicate that the absolute and at the growing surface. Results from these calcula-
ma gn i t u d e of the deposition rate is sensitive to the bond tions are c o m b i n e d with the governing differential equa-
strength ratios but, despite this uncertainty, the overall tions for fluid flow and heat and mass transport to describe
trends in behavior are not altered appreciably. the system behavior. In this way, deposition rates and rate-
The composition profiles given in Fig. 9 indicate that limiting steps are predicted without either assuming a re-
there will be m o r e WF4 in the gas than WFs. However, the action m e c h a n i s m or arbitrarily choosing any kinetic pa-
amounts of these gas-phase intermediates are sensitive to rameter values.
their heats of formation. For example, if AH~,~98for WF~ is The m o d el considers all plausible species and elemen-
m a d e 5 kcal m o l - ' m o r e n e g a U v e , a larger fraction of the tary processes since one does not k n o w a priori which in-
2WFs(a) desorbs and the ratio of partial pressures PWFs/PwF4 termediates and reaction pathways are most favorable.
adjacent to the surface changes from 0.4 to 21.4. Therefore, The fractional surface coverages of individual adsorbates
at this stage, one cannot specify unequivocally w het h er are monitored since these quantities cannot, in general, be
WF4 or WF5 is the d o m i n a n t tungsten-containing interme- replaced by products of adsorption equilibrium constants
diate in the system. Nevertheless, the 5 kcal mo1-1 change and partial pressures in the gas phase adjacent to the sur-
in hH~,298 for WF5 reduces r~ by less than 0.1% and it does face. Moreover, i f a species can adsorb in different bonding
not alter the trends in behavior for t h e total partial pres- configurations, it is necessary to regard each possibility as
sure o f WF4 and WF 5 given in Fig. 9. an i n d e p e n d e n t species because the properties and reac-
The theoretical calculations assume that, as a result of tivity of the adsorbate will be different in each case. Fur-
steric hindrance, WF~ (2 -< x --< 6) cannot adsorb onto thermore, meaningful concentrations of vacancies and
tungsten sites wh er e only one dangling bond remains (i.e., species at the surface can only be obtained if the model ac-
f ~ for reactions [$2]-[$4] is taken to be 0.5625, see Ap- counts for the possibility that several adsorbates may
pendix A). However, the surface coverages in Fig. 7 sug- bond to a given surface site. In general, the activities of ad-
gest that the a m o u n t of %VF~(a) present is too small to sorbates and vacancies that appear in the rate equations
cause significant steric hindrance. U n d e r these circum- should be related to fractional coverages through statisti-
stances, the statistical factor f~, should be replaced by cal factors which account for the directionality of the
0.625. Theoretical results obtained using this value for f~m bonds and the possibility of steric hindrance.
are essentially unchanged, e.g., r~ values for the lower Theoretical predictions for low-pressure CVD of
curve in Fig. 6a are increased by less than 2%. Further- tungsten agree closely with available experimental data.
more, rs is unaffected by the choice off~m for reaction [S13] Th e calculations indicate that, for the range of operating
because this step occurs almost instantaneously. The conditions considered, the growth rate is controlled by
choice of probability factors to describe the interactions surface reaction kinetics and the extents of gas-phase reac-
a mon g adsorbates, i.e., determination of appropriate tions are negligible. U n d er these circumstances, diffusion
values for fjm, often presents difficulties (28). In CVD of and thermal diffusion across the boundary layer are unim-
tungsten, however, these factors do not affect the process portant, and it is only necessary to perform multidimen-
behavior, e.g., r~ is altered by less than 10-3% iffj~ is set to sional calculations to describe the transport processes if
unity for all surface species. the fractional utilization of reactants is significant or if the
In the light of the uncertainties described above one reactants are not properly mixed. Furthermore, only a
migh t argue that prediction of reaction rates in CVD proc- small subset of the 65 surface processes considered play a
esses should only be attempted if a complete set of accu- significant role in the process and the principal reaction
rate eq u i l i b ri u m constants and rate constants are avail- pathway involves sequential removal of fluorine atoms
able. Indeed, with this philosophy, one might envisage a from adsorbed WF6. The rate-limiting steps .are found to be
series of detailed molecular dynamical calculations to ob- conversion of WF4 to WF 3 and removal of adsorbed fluo-
tain m o r e definitive kinetic parameters (28). Unfor- rine by Ha to give H F in the gas. This result demonstrates
tunately, procedures of this type, even if they were possi- that more than one elementary process can significantly
ble, would be exceedingly time consuming. Furthermore, affect system behavior at a single set of operating condi-
the im p o rt an t elementary steps in a given process are not tions.
k n o w n a priori and, therefore, considerable effort would The CVD m o d el provides information on the amounts of
u n d o u b t e d l y be e x p e n d e d o~ reactions that, in the end, the various adsorbates present at the growing surface. Re-
turn out to be u n i m p o r t a n t from a practical standpoint. sults for deposition of tungsten indicate that the growth
For example, uncertainties in property values that only af- rate is sensitive to the fractional coverage of vacant sites
fect reactions involved in m i n o r pathways do not have a and that this is dictated primarily by the a m o u n t of ad-
significant effect on the theoretical predictions. Therefore, sorbed fluorine. The extent to which fluorine covers the
m o r e detailed values w o u l d not be helpful in theses cases. surface is affected by the rates of side reactions as well as
On the other hand, uncertainties in quantities associated the rates of the tungsten subfluoride decomposition steps.
in some way with the rate-limiting steps were found to af- In general, adsorbates involved in the main reaction se-
fect the absolute m a g n i t u d e of the growth rate. In this re- q u e n c e m i g h t only occupy a small fraction of the surface
gard, the calculations pinpoint the thermochemical prop- sites and, hence, the importance of a reactive intermediate
erties and rate constants that influence the deposition rate is not necessarily linked to its surface coverage. The calcu-
the most and, hence, they indicate w h e r e additional meas- lations also suggest that rates of desorption of tungsten
u r e m e n t s or more detailed calculations would be most subfluorides can be minimized by operating under condi-
beneficial. It should be noted that, for the few sensitivity tions that give low growth rates.
studies where the tungsten deposition rate was changed Many of the reactions in the principal tungsten deposi-
substantially, the effect was so dramatic it indicated that tion pathway achieve local equilibrium and several other
the quantity being considered m u s t lie within a very nar- steps can be regarded as irreversible. As a result, it has
row range, and the values in this range were always found been possible to develop several simplified kinetic models
to be physically realistic. Moreover, even these uncertain- which still use fundamental values for the rate constants
ties did not alter the predictions for major reaction path- and which predict growth rates that agree fairly well with
ways, rate-limiting steps, or trends in behavior. Conse- experimental data. In addition to the computational bene-
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1535

fit, these models provide insight into the factors that con- II
trol the deposition rate and its dependence on inlet gas Ag(g) ~ B(g)
composition. For example, the equations show that the
growth rate is sensitive to the surface concentration of va- Ill v IIIJl 2,
1
cancies and the fractional coverage of bridge-bonded fluo- A(a)+, ~ B(a)
rine. The results also demonstrate that use of Langmuir-
Hinshelwood kinetics or surface reaction probabilities to The magnitude of K~ can be equated to the product
describe CVD of t u n g s t e n would be inappropriate. KIKnKm because enthalpy, entropy, and free energy are
additive (8). Values of AH~,298and S~98for the relevant gas-
The values of bond strengths and entropies influence phase molecules are given in Table I and the temperature
rate constants as well as equilibrium constants and it is dependences of these thermodynamic properties are ob-
found that uncertainties in properties of this type or in the tained by integration of the heat capacities, Cp. Also, the
rate constants themselves can affect the absolute magni- changes in enthalpy and entropy on adsorption/desorption
tudes of the calculated deposition rates. These uncertain- of reactant(s) and product(s) are estimated using the quan-
ties are only important for quantities that are either di- tities in Eq, [A-4] and [A-5]. With this approach, one obtains
rectly or indirectly associated with the rate-controlling
steps. However, the changes do not alter the predictions
for trends in behavior, d o m i n a n t species, or major reaction hHm = (AH~.298)B(g)- (Q~)B + f$s98(Cp)B(g)dT
pathways. Furthermore, it is found that statistical factors
and steric hindrance effects do not play a significant role in
the t u n g s t e n process. Overall, the theoretical predictions {(AH~,298)A(g)--(Q,)A+ fill (Cp)A(g)dT} [A-6]
for d o m i n a n t reaction pathways, taken in conjunction
with the sensitivity studies, pinpoint the properties and and
rate constants for which more detailed calculations or ex-
perimental measurements would be most beneficial from a o + (T~ (Cp)B(g)dT
practical standpoint. ASm = (Sr,a98)B(g) R In (KoT~)B +
The success of the theoretical predictions for low-pres- ' ~298 T
sure deposition of t u n g s t e n demonstrates that a mathe-
matical model which calculates rate constants for elemen- -
o

(Sf,298)A(g)+ R In (Ko,T). + J298

[r~(C,)A(~)
T dT
} [A-7]
tary processes in the gas phase and at the surface from first
principles can be used to provide insight into the factors
that control system behavior. An advantage of this ap- Substitution of Eq. [A-6] and [A-7] into Eq. [A-1] gives the
proach over CVD models with empirical treatments of sur- equilibrium constant.
face reaction kinetics is that it establishes a fundamental
framework for evaluating alternative sets of processing A P P E N D I X lq
conditions and for assessing the relative merits of new Estimation of the Statistical Factors f,. and fi.
combinations of reactants. The statistical factors fvmfor vacancies are calculated on
the basis that an empty surface site on the (100) plane of a
Acknowledgment body-centered cubic structure has four dangling bonds
This work was supported by the National Science Foun- available. In reaction [S1], for example, both atoms in H2
form two bonds to the surface by linking to dangling
dation (Grant No. CBT-8451112). bonds on adjacent sites (see Fig. 2). However, for this ad-
sorption process to occur, it is expected that the dangling
Manuscript received Sept. 7, 1990. bonds must be oriented in the proper position, i.e., in the
same orientation as the final bonds themselves. The quan-
APPENDIX A tity 0v accounts for the overall concentration of dangling
Evaluation of Equilibrium Constants for Surface Reactions bonds present on the surface whereas fvm represents the
likelihood that the dangling bonds required are in these fa-
The equilibrium constant K,j for surface reaction 1 is vorable positions. For a site having n dangling bonds avail-
given by
able, the probability of a given dangling bond being in one
K~ = exp {(-AHm + T~ASRt)/RT~} [A-l] particular position is 0.25n if multiply bonded adsorbates
predominate. Therefore, on average, the statistical proba-
where the enthalpy AHm and entropy hSm of reaction are bility fvm that the dangling bond at a site is correctly ori-
defined as ented is En (0.25 n Ln) where Ln is the fraction of the sites
with n dangling bonds available and 1 -< n -< 4. If it is as-
ng+na sumed that Ln = 0.25, one obtains fvm = 0.625. Note that this
~k/-/RI = - ~ (AS)k,I(AH~,T)k [A-2] derivation must be modified for reaction [S13] because it
/r involves a dangling bond on the same site as the adsorbate.
I n this case, the site can have a m a x i m u m of three dan-
and gling bonds andfvm = (0.33 + 0.67 + 1.00)/4 = 0.5.
ng+na
If the surface coverage of large adsorbate molecules
such as ~ F = ( a ) is fairly high, some of the remaining dan-
ASm = - k=l
~ (hS)k,l(aS~)k [A-3] gling bonds will be screened from interacting with the gas
phase. U n d e r these circumstances, the value offvm should
The standard enthalpy AH~,Tof formation of any surface be reduced to reflect the lower reaction probability. For
species j is given by example, if adsorption steps [$2]-[$4] are sterically hin-
dered at sites where only a single dangling bond is avail-
(AH~,T)j = ( A H ~ T ) i - Q, [A-4] able, the statistical factor for these elementary processes
becomesfv~ = (0.0 + 0.5 + 0.75 + 1.0)/4 = 0.5625. Similarly,
where the subscript i refers to the same species in the gas if adsorption onto sites with two dangling bonds is also
and Q, is the heat of adsorption, equivalent to the total sterically hindered, fvm is reduced to 0.4375.
strength of the bond(s) formed by species j to the surface For singly bonded adsorbates, i.e., 1H(a) and ~F(a), the
(33). Similarly, the standard entropy of the surface species probability f~m that it is in a position favorable for reaction
can be obtained from with a dangling b o n d or adsorbate on a neighboring site is
taken to be 0.25. For adsorbates that form bonds with mul-
(S~)j = (S~)i + R In (Ko,T) [A-5] tiple sites, the activities are equated to the fractional sur-
where values for Ko,wmay be calculated using the results face coverages, i.e., fj~ is set to unity.
presented in Table IV.
Consider a surface reaction 1 of the type A(a) + v ~ B(a) APPENDIX C
where A and B represent adsorbates. The equilibrium con- Rate Equations for Simplified Kinetic Models
stant K~a for this reaction is determined by constructing a For model AMII, it is assumed that local equilibrium is
thermodynamic cycle that contains the corresponding gas- established for surface reactions [S1], [$9], [S10], and [S16].
phase reaction, viz. Therefore, the rate expressions for these steps can be re-
Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 1536 J. Electrochern. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc.

placed by definitions of the equilibrium constants for the kbl backward rate constant for heterogeneous
reactions. Combination of these equations with the gen- reaction 1.
eral rate equation, Eq. [4], for reactions [$2], [$8], and [Sll]- k~ forward rate constant for heterogeneous reaction 1
[S13] yields k~l backward rate constant for homogeneous
reaction 1
(%)~ r (0v) 6 . 02F -] k~ forward rate constant for homogeneous reaction 1
[C-l] K m u l t i c o m p o n e n t thermal diffusion ratio (42)
Kal = k~/kbl; equilibrium constant for heterogeneous
where reaction 1
Keq,a adsorption equilibrium constant
02~ = [r~ + A~pHr(0~)~X/-~] (AhpH2(0~)2)-~ [C-2] K~l = k'n/k'~l; equilibrium constant for homogeneous
reaction 1
and Eq. [5] can be replaced by L. fraction of sites with n dangling bonds available,
l_<n_<4
Ov = 1 - A6(ATPttF(O~)~H2H2)
Mi molecular weight of species i
-- 202F(1 + A8(O~)2) - Ag0.~H~H~ [C-3] na total n u m b e r of adsorbed species
ng total n u m b e r of gas-phase species
The coefficients A~ - A9 i n the impYmit relations for r~ are nrg total n u m b e r of homogeneous reactions
combinations of fundamental rate constants and equilib- nrs total n u m b e r of heterogeneous reactions
rium constants for the d o m i n a n t elementary surface pro- n u m b e r of adjacent atoms at sites of type m to
cesses, viz., A~ =(kb,ll + kb,~2)ga,2/([Link]), A~ = (kb,n + kb,12)/ which an adsorbate j forms bonds
(kb,2Ka,[Link]) , A~ = Ka,11/12(kb,11 + kb,12)/kb,13, A4 = [Link],8/ NA Avogadro's n u m b e r , 6.023 x 1023mo1-1
Ni flux of species i in the gas phase, mol cm -2 s
(4~), A5 = kb.~/4,A6 = 2/kb,2, Av = Ka,2kb,~,As = Ka,~6/2,and normal flux of species i from the surface, mol cm -2
Ni,s
A9 = 2/x/~,~. With this approach, Eq. [C-1]-[C-3] can be S-1
combined with Eq. [9] and [10] at the surface, and the P pressure in the gas phase, Torr
growth rate calculated directly. Note that the other surface Pi = pxi; partial pressure of species i, Torr
concentrations can be determined from the values ob- pO = px~ inlet partial pressure of species i, Torr
tained for 0~ and 02F using the kinetic and thermodynamic Q~ = Qao - G(1 - 0~); heat of adsorption, kcal m o l - '
parameter values given in the tables. Qao heat of adsorp~lon at zero surface coverage, kcal
I n model AMIII, WF~ adsorption (reaction [$2]) is also as- mol-1
sumed to be at equilibrium and the forward rates for reac- rl net rate of heterogeneous reaction 1
tions [S11]-[S13] are neglected. U n d e r these circumstances, rf net rate of homogeneous reaction 1
Eq. [C-l] can be simplified further to yield r~ deposition rate, n m m i n -~
R universal gas constant, 1.987 cal mo1-1 K - '
rate of production of species j on the surface, mol
(~ PWF~
r~ = (kb.n + kb,~)C~(0-~)2 [C-4] Era-2 s-1
RI rate of production of species i in the gas phase, tool
cm-3 s-1
where stoichiometric coefficient of species k in backward
Sb,kl
step of heterogeneous reaction 1 (Sbkl -----0)
02~ = C1C2 -(0v)~
- PWF6+ C3 PHF
- -(0v)2 [C-5] Sf,kl stoichiometric coefficient of species k in forward
(02~)2 P ~ step of heterogeneous reaction 1 (sf,~ --- 0)
s~ standard entropy of species i at temperature T, cal
and mol-, K-1
ASR1 standard entropy change for reaction 1, cal mo1-1
% = 1 - C4(Ov)2pw~6- 202F[1 + C~(O~)2] -C60~X/~H~ [C-6] K-1
The coefficients C~-C6 are combinations of rate constants T temperature, K
and equilibrium constants for the major reactions, viz. T~ surface temperature, K
C1 = Ka,2/(K, gK~o), C2 = 4 (kbn + k ~ _ / k b a , Ca = K ~ . 8 / ~ , T| inlet gas temperature, K
y vacancy
C4 = 2Ka,2, C5 = 2K,46, and C6 '= 2/VK~,I. ' V mass-average velocity, cm s - '
LIST OF SYMBOLS Xi mole fraction of species i in the gas phase
a h y d r o d y n a m i c parameter, s -~ (43) x? inlet mole fraction of species i
ak activity of species k at the surface (ak is equal to Pi Z distance from the surface, cm
for gas-phase species, fjm0jm for adsorbates, and
fvm0w for vacancies) Greek
AM approximate model Ot parameter in Szab5 equation (28)
c = 3.184 • 10-5 p / R T , total concentration in the gas characteristic energy of interaction for Lennard-
phase, mol cm -3 Jones potential, cal
Cv molar heat capacity of gas-phase species, cal mol- "qm n u m b e r of dangling bonds/site at site of type m
K-1 0jm = Cjm/~mCwm; the n u m b e r of bonds attached to spe-
cjm surface concentration of species j, mol cm -2 cies j divided by the m a x i m u m n u m b e r of bonds
Cwm total surface concentration of sites of type m, mol that can be formed (at sites of type m)
Era-2 0vm = c~/~mCwm; ratio of the n u m b e r of dangling bonds
D~ gas-phase dissociation enthalpy of bond Y-Z, kcal to the m a x i m u m n u m b e r of bonds that can be
mol -~ formed (at sites of type m)
~k diffusion coefficient for binary interactions, cm 2 Pb,il stoichiometric coefficient of species i in backward
S-1 step of homogeneous reaction 1 (Vb,~>--0)
DM detailed model Pf,il stoichiometric coefficient of species i in forward
E~ activation energy of the backward step for hetero- step of homogeneous reaction 1 (vf,~-> 0)
geneous reaction 1, keal mol -~ Er collision diameter, A
E~ activation energy of the backward step for homo- oJI vibrational frequency of bond(s), 1, cm -1
geneous reaction 1, keal m o l - '
fjm probability that adsorbate j at site of type m is ori-
ented in a position that will allow reaction to occur
fvm probability that dangling bonds at site of type m Subscripts
are oriented in a position that will allow reaction to i gas-phase species
occur J adsorbate
G parameter that describes the dependence of heats k i and j
of adsorption on 0v (see Table IV), kcal mol 1 reaction 1
AH~,T standard enthalpy of formation at temperature T, m type of surface site
kcal mo1-1 s surface
hHra enthalpy change for reaction 1, kcal mol -~ v vacancy
k Boltzmann constant, 3.299 x 10 -24 cal K -I o0 in the bulk fluid

Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 138, No. 5, May 1991 9 The Electrochemical Society, Inc. 1537

Superscripts 26. A. J. Melmed, R. T. Tung, W. R. Graham, and G. D. W.

1 singly b o n d e d adsorbate Smith, Phys. Rev. Lett., 43, 1521 (1979).
2 bridge-bonded adsorbate 27. K. J. Laidler, S. Glasstone, and H. Eyring, J. Chem.
4 q u a d r u p l y b o n d e d adsorbate Phys., 8, 659 (1940).
28. K. J. Laidler, "Chemical Kinetics," Harper and Row,
REFERENCES New York (1987).
1. G. Stringfellow, Rep. Prog. Phys., 45, 469 (1982). 29. A. Couper and C. S. John, J. Chem. Soc., Faraday
2. H. K. Moffat and K. F. Jensen, This Journal, 135, 459 Trans., 77,950 (1977).
(1988). 30. G. Bolbach and J. C. Blais, Surf. Sci., 137, 327 (1984).
3. M. E. Coltrin, R. J. Kee, and J. A. Miller, ibid., 133, 1206 31. W. Ho, R. F. Willis, and E.W. Plummer, Phys. Rev.
(1986). Lett., 40, 1463 (1978).
4. M. Tirtowidjojo and R. Pollard, J. Cryst. Growth, 93, 32. M. R. Barnes and R. F. Willis, ibid., 41, 1729 (1978).
108 (1988). 33. A. W. Adamson, "Physical Chemistry of Surfaces,"
5. M. Tirtowidjojo and R. Pollard, ibid., 98, 420 (1989). J o h n Wiley & Sons, Inc., New York (1982).
6. R. T. Kurnik, Chem. Eng. Prog., 81, 30 (1985). 34. V. N. Kondrat'ev, "Chemical Kinetics of Gas Reac-
7. S. Sachdev and R. Castetlano, Semiconductor Interna- tions," P e r g a m o n Press, Inc., New York (1987).
tional, 8(5), 306 (1985). 35. L . S . Kasse], J. Phys. ~hem., 32, 1065 (1928).
8. K. Denbigh, "The Principles of Chemical Equilib- 36. P. J. Robinson and K. A. Holbrook, "Unimolecular Re-
rium," Cambridge University Press, Cambridge actions," Wlley-Interscience, L o n d o n (1972).
87. A.M. Dean, J. Phys. Chem., 89, ,i600 (1985).
(1983). 88. H. S. J o h n s t o n and C. Parr, J. Am. Chem. Soc., 85, 2544
9. D. R. Stull and H. Prophet, " J A N A F Thermochemical
Tables," N S R D S - N B S 37, U.S., NBS, Washington, (1963).
DC (1971). 39. R.D. Gilliom, J. Chem. Phys., 65, 5027 (1976).
10. D. L. Hildenbrand, J. Chem. Phys., 62, 3074 (1975). 40. R. Arora and R. Pollard, In preparation.
11. L. Pauling, "The Chemical Bond," Cornell University 41. J. M. Thomas and W. J. Thomas, "Introduction to the
Press, Ithaca, NY (1967). Principles of Heterogeneous Catalysis," Academic
12. R. Haque, M. Pelino, and K . A . Gingerich, J. Chem. Press, Ltd., L o n d o n (1967).
Phys., 71, 2929 (1979). 42. J. P. J e n k i n s o n and R. Pollard, This Journal, 131, 2911
13. S.W. Benson, "Thermochemical Kinetics," John Wiley (1984).
& Sons, Inc., New York (1976). 43. M. Michaelidis and R. Pollard, ibid., 131, 860 (1984).
14. G. J. Janz, " T h e r m o d y n a m i c Properties of Organic 44. R. Pollard and J. Newman, ibid., 127, 744 (1980); 127,
Compounds," Academic Press, Inc., New York 2756 (1980).
(1967). 45. R. B. Bird, W. E. Stewart, and E. N. Lightfoot, "Trans-
15. D. A. McQuarrie, "Statistical Mechanics," Harper and port Phenomena," J o h n Wiley & Sons, Inc., New
Row, New York (1976). York (1960).
16. M. Tirtowidjojo and R. Pollard, J. Cryst. Growth, 77, 46. R. A. Svehla, N A S A Tech. Report R-132, Lewis Re-
200 (1986). search Center, Cleveland, OH (1962).
17. R. Arora, Ph.D. Thesis, University of Houston, Hous- 47. C. M. McConica and K. Krishnamani, This Journal,
ton, TX (1990). 133, 2542 (1986).
18. K. M. Guggenheimer, Discuss. Faraday Soc., 9, 221 48. P. van der Putte, Philips J. Res., 42, 608 (1987).
(1950). 49. D. D. Eley, Discuss. Faraday Soc., 8, 34 (1950).
19. T. Moeller, "Inorganic Chemistry," J o h n Wiley & 50. M.-C. Desjonqueres, J.-P. Jardin, and D. Spanjaard, J.
Sons, Inc., New York (1982). Chim. Phys., 85, 831 (1988).
20. M. K. Debe and D. A. King, Surf. Sci., 81, 193 (1979). 51. J. R. Creighton, This Journal, 136, 271 (1989).
21. U. A. Jayasooriya, M. A. Chesters, M.W. Howard, 52. C. R. Kleijn, C. J. Hoogendoorn, A. Hasper, J. Holle-
S. F. A. Kettle, D. B. Powell, and N. Sheppard, ibid., man, and J. Middelhoek, in Mat. Res. Soc. Syrup.
98, 526 (1980). Proc., "VLSI-V," S. S. Wong and S. Furukawa, Edi-
22. R. Biswas and D. R. Hamann, Phys. Rev. Lett., 56(21), tors, p. 109, Materials Research Society, Pittsburgh,
2291 (1986). P A (1990).
23. R. E. Reed-Hill, "Physical Metallurgy Principles," 53. H. Schlichting, " B o u n d a r y Layer Theory," McGraw-
Wadsworth, Boston, MA (1973). Hill, Inc., New York (1979).
24. J. E. J. Schmitz, A. J. M. van Dijk, and M. W. M. Graef, 54. M. Tirtowidjojo and R. Pollard, Mater. Res. Soc. Symp.
in "Tenth International Conference on Chemical Proc., 131, 109 (1989).
Vapor Deposition," G.W. Cullen, Editor, P V 87-8, 55. R. J. Buss, P. Ho, W. G. Breiland, and M. E. Coltrin, J.
p. 625, The Electrochemical Softbound Proceedings Appl. Phys., 63, 2808 (1988).
Series, Pennington, NJ (1987). 56. V. J. Mimeault and R. S. Hansen, J. Chem. Phys., 45,
25. E. Shustorovich, Surf. Sci. Rep., 6, 1 (1986). 2240 (1966).

Downloaded on 2014-11-09 to IP [Link] address. Redistribution subject to ECS terms of use (see [Link]/site/terms_use) unless CC License in place (see abstract).