Vol. 129, No.

8 H Y D R O G E N - I M P L A N T E D SiN 1791

14. D. K. Brice, Radiat. EJf., 11, 227 (1971) ; and P r i v a t e
communication.
15. A. S. Grove, "Physics and Technology of Semicon-
ductor Devices," J o h n Wiley a n d Sons, Inc., New
York (1967).
16. H. J. Stein, P. S. Peercy, and D. S. Ginley, in "The
Physics of MOS Insulators," G. Lucovsky, S. T.
Pantelides, and F. L. Galeener, Editors, p. 147,
P e r g a m o n Press, New York (1980).
17. For details see P. Balk, in "Solid State Devices,
1973," The Institute of Physics, London and
Bristol, p. 51 (1974).
18. B. H. Yun, Appl. Phys. Lett., 27, 256 (1975).
19. H. J. Stein, J. Electron. Mater., 5, 161 (1976).
20. D. F r o h m a n - B e n t c h k o w s k y a n d M. Lenzlinger, J.
Appl. Phys., 40, 3307 (1969).
21. F. A. Sewell, Jr., H. A. R. Wegener, and E. T. Lewis,
Appl. Phys. Lett., 14, 45 (1969).
22. W. C. Johnson, Report NVL-0059-011, Dec. 1, 1979.
23. H. J. Stein, Abstract 216, p. 566, The Electrochemi-
cal Society Extended Abstracts, Atlanta, Georgia,
Oct. 9-14, 1977.

Grain-Growth Mechanisms in Polysilicon

Len Mei, 1 Michel Rivier, 2 Young Kwark, and Robert W. Dutton
Stanford Electronics Laboratories, Stanford University, Stanford, California 94305

ABSTRACT
The electrical characteristics of polycrystalline silicon layers are closely related to their grain structure. This paper de-
scribes a comprehensive study of the grain growth of polysilicon u n d e r a wide range of doping and processing conditions. A
grain-growth model has been developed and implemented in the SUPREM process simulator, and the simulated results are
compared to those obtained by transmission electron microscopy for As-, P- and B-doped polysilicon. These results indicate
that n-type dopants increase the growth rate whereas the p-type dopant has a negligible effect.

Polycrystalline silicon as a key technological i n n o - Experiments

vation has had a n increasing impact on the fabrication
of integrated circuits. Its applications include high-
value resistors, gate electrodes, interconnections, and
diffusion sources for both active devices and contacts
(1-3). Its electrical a n d material characteristics are
dependent on its metallurgical structure and segrega-
tion properties. Much work has been published con-
cerning the electrical conduction i n polysilicon layers
(4, 5), segregation of impurities at the grain boundaries
(6), and oxidation of doped polysilicon (7); however,
relatively few results have been reported on its m e c h -
anisms of g r a i n growth (8). In this paper, we report
a n extensive study of the grain structure of polysili-
con doped with the most c o m m o n l y used dopants (P,
As, and B) and processed at different times and t e m -
peratures. A g r a i n - g r o w t h model is also introduced to
d e t e r m i n e the degree of dopant e n h a n c e m e n t .
D u r i n g t h e r m a l processing, the increase in the a v e r -
age grain size of polysilicon is proportional to the
square root of time. This growth is believed to be a
diffusion-controlled m e c h a n i s m (8) w h e r e i n the pres-
ence of impurities plays two roles: an increase of va-
cancy concentration caused by a shift of the F e r m i
level and" the change of g r a i n - b o u n d a r y energy result-
ing from i m p u r i t y segregation at the boundaries. The
As and P dopants e n h a n c e grain growth to a variable
degree, and the B dopant has little effect. The As-
doped poly, for example, shows an initial increase in
grain size after a rise of concentration b e y o n d 1 X 1019
a t o m s / c m 3, however, this e n h a n c e m e n t reaches a m a x -
i m u m a n d drops at higher concentrations. This initial
e n h a n c e m e n t is believed to be the result of larger v a -
cancy concentration at higher doping levels. W h e n the
solid solubility is exceeded, however, the vacancy con-
centration no longer increases, instead, the segregated
impurities at the grain boundaries can h i n d e r the
b o u n d a r y migration. I n addition, the presence of clus-
ters restrains the b o u n d a r y migration, t h e r e b y contrib-
u t i n g to the reduced growth rate. The a p p a r e n t acti-
vation energy depends on vacancy formation, self-dif-
fusion of Si, g r a i n - b o u n d a r y segregation, and cluster
formation.
1present address: Fairchild Camera and Instrument Corpora-
tion, Palo Alto, California 94304.
Present address: Bendix Advanced Technology Center, Colum-
bia, Maryland 21045.
Key words: polysilicon, grain growth, grain boundary, doping
effect, diffusion.
All polysilicon layers discussed i n this paper were
deposited on t h e r m a l l y grown 1000A oxide on ~ 1 0 0 ~
single crystal silicon substrates by atmospheric pres-
sure chemical vapor deposition. The deposition rate at
776~ was 700 A / m i n . The film thicknesses were 2500,
5000, 6000, and 7500A. After deposition, a 1000A silox-
oxide was deposited at 450~ on the polysilicon to pre-
vent the escape of impurities d u r i n g s u b s e q u e n t high
t e m p e r a t u r e processing. Impurities were introduced
into the polysilicon layers b y ion i m p l a n t a t i o n with
sufficient energy for the peak concentration to appear
deep in the polysilicon layer.
Phosphorus-doped samples were studied to deter-
m i n e the effect of thickness on grain growth. The i m -
p u r i t y doses were 3 X 1014, 6 • 1014, and 9 X 1014
a t o m s / c m 2, respectively, for the 2500, 5000, and 7500A
films, and the average concentration was 1.2 X 1019
a t o m s / c m 3. These samples were a n n e a l e d at 950~ in a
n i t r o g e n a m b i e n t for 20, 60, 120, a n d 180 rain.
A second set of phosphorus-doped samples was used
to examine the effect of concentration on grain growth.
Average concentrations were 1.2 X 1019, 6 X 1019, and
1.2 X 1020 for the 2500_~ films and 1 X 1020 and 1.5 X
102o for the 6000@ films. This set of samples was a n -
nealed at I000 ~ and 1200~ for 60 rain.
The arsenic- and boron-doped samples all had film
thickness of 2500A with concentrations of 1.2 X 1019,
6 X 1019, 1.2 X 1020, and 6 X 1020 a t o m s / c m ~. T h e y
were a n n e a l e d at 900-~ 1000 ~ II00 ~ and 1200~ for 60
rain.
After t h e r m a l annealing, the samples were wet
etched from the back surface to the oxide/Si substrate
interface, then were t h i n n e d to a final thickness of ap-
p r o x i m a t e l y 2000A b y ion milling for several hours.
Samples with 2500A polysilicon were also prepared b y
an u n d e r c u t t i n g technique, in which the I-IF etehant
covered a mesa structure a p p r o x i m a t e l y 2 m m i n di-
ameter. Several hours were r e q u i r e d to complete the
u n d e r c u t to float off the polysilicon from the substrate.
A Philips EM400 transmission electron microscope was
used to determine the morphology and crystal struc-
ture of the polysilicon layers and g r a i n size was m e a -
sured b y the technique developed b y Wada and 1~ishi-
matsu (8).
 1792 J. Electrochem. Soc.: S O L I D - S T A T E S C I E N C E A N D T E C H N O L O G Y
Mechanism of Grain Growth in Polysilicon
The g r a i n - g r o w t h m e c h a n i s m in p o l y - S i is believed
to be controlled b y diffusion. This m o v e m e n t can be
accomplished b y a series of diffusion j u m p s of i n d i v i d -
ual atoms across the grain boundaries, and the g r o w t h
r a t e is t h e n d e t e r m i n e d b y the diffusion rate of the
atoms across each g r a i n b o u n d a r y . In the absence of a
d r i v i n g force, the n u m b e r of j u m p s in the opposite di-
rections is equal and, as a consequence, there is no net
m o v e m e n t of atoms. When a d r i v i n g force exists, the
free e n e r g y is r e d u c e d and atoms m i g r a t e from one
g r a i n to another, which contributes to the b o u n d a r y
m i g r a t i o n of b o u n d a r y and the r e s u l t a n t grain growth.
The d r i v i n g force has the following two components:
(i) e n e r g y difference b e t w e e n atoms at one side of
g r a i n b o u n d a r y and at the other and (ii) interracial
e n e r g y b e t w e e n two grains. Because of this i n t e r f a c i a l
energy, the g r a i n b o u n d a r y tends to minimize its area,
which m a k e s the d r i v i n g force i n v e r s e l y p r o p o r t i o n a l
to t h e b o u n d a r y curvature. The d r i v i n g force can be
expressed b y (9)
akb 2
r _ - -
L e n e r g y b e t w e e n the g r a i n b o u n d a r y and bulk. The acti-
w h e r e k is g r a i n - b o u n d a r y energy, b is the lattice con-
stant, L is the a v e r a g e grain size, and a is a geometric
factor.
The m i g r a t i o n rate of the b o u n d a r y is not only p r o -
p o r t i o n a l to the e x t e r n a l d r i v i n g force b u t also to
g r a i n - b o u n d a r y mobility. This m o b i l i t y measures atom
m i g r a t i o n u n d e r an e x t e r n a l d r i v i n g force and is d i -
r e c t l y p r o p o r t i o n a l to the Si self-diffusivity in a dif-
fusion-controlled g r o w t h regime (9). This leads to
g r e a t e r g r a i n - b o u n d a r y m o b i l i t y at h i g h e r v a c a n c y
concentrations. The m o b i l i t y is given b y
Dg
= [2]
kT
w h e r e Dg is the Si self-diffusion constant along a
boundary.
The b o u n d a r y m i g r a t i o n rate is the p r o d u c t of m o b i l -
ity and the d r i v i n g force
dL akb2D g
= ~F : - - [3]
dt LkT
which, w h e n r e a r r a n g e d and integrated, results in
L2=L02+2 Sota~b2 Dg dt
kT
If the g r a i n - b o u n d a r y e n e r g y and Si self-diffusivity
do not v a r y w i t h time, the final g r a i n size can be e x -
pressed in t e r m s of the initial grain size L0 and a g r a i n -
g r o w t h r a t e constant kr
L = y/L02 + kr2t
where kr is
x/2a~b2D g
kr-- v -~
W h e n the initial g r a i n size is small c o m p a r e d to "~/kr2t,
this reduces to L ~-~ krN/t-which is valid when the
p o l y - S i is deposited at t e m p e r a t u r e s m u c h lower than
t h e a n n e a l i n g t e m p e r a t u r e . Because the active con-
c e n t r a t i o n changes with the grain size and o u t - d i f f u -
sion, Eq. [3] m u s t b e used to d e t e r m i n e the grain
g r o w t h in a given t i m e step w i t h the n e w rate constant.
Silicon self-diffusivity at each g r a i n b o u n d a r y is a
function of the active i m p u r i t y concentration in t h e
g r a i n and is f u r t h e r described in the n e x t section. The
g r a i n - b o u n d a r y free e n e r g y ~. changes w h e n the grains
are in contact w i t h the p o l y / o x i d e interface and this
is discussed below.
August 1982
Diffusion Constant
Because of the e n e r g y balance, g r a i n b o u n d a r i e s are
the p r e f e r r e d sites for mobile de~ects, p r i m a r i l y the
vacancies. The Si self-diffusion constant at g r a i n
b o u n d a r i e s can be r e l a t e d to the b u l k self-diffusivity
b y (10)
/ 1 + AClexp [6]
w h e r e Ga is heat of segregation, A is the v i b r a t i o n a l
e n t r o p y factor, D 1 is the diffusivity in bulk, the sub-
script i denotes the intrinsic polysilicon, and Cl is the
i m p u r i t y concentration in an atomic fraction inside the
g r a i n d e t e r m i n e d b y the segregation of i m p u r i t i e s at
t h e boundary, which is a function of g r a i n size as dis-
cussed below. The ratio of diffusivity can also be e x -
pressed b y
Dig Di~ exp = "v exp [7]
Di1 -- Dio1 k'-~-" L'-~'-
[1] w h e r e AGb is the difference in the diffusion activation
vation e n e r g y for g r a i n - b o u n d a r y diffusion is a l w a y s
l o w e r t h a n t h a t for diffusion t h r o u g h grains, which
implies t h a t AGb should be positive (10). One sig-
nificant consequence is t h a t the r e l a t i v e i m p o r t a n c e of
g r a i n - b o u n d a r y diffusion becomes less at higher t e m -
peratures. The final diffusivity then becomes
AGb
exp \ ~ ]
1 + AClexp kT
Here, b u l k self-diffusivity as a function of doping eon-
centration can be d e t e r m i n e d from the consideration of
v a c a n c y coneentrations (11).
Effect of Polysilicon Thickness
When the average grain size in polysilieon becomes
c o m p a r a b l e to the thickness of the polysilieon layer,
m a n y grains are in contact with the oxide l a y e r a n d / o r
the Si s u b s t r a t e interface, the interfacial e n e r g y b e -
comes m o r e i m p o r t a n t in the g r o w t h mechanism.
Grains w i t h an orientation having lower interfacial
e n e r g y w i l l g r o w at the expense of grains w i t h high in-
t e r f a c i a l energy. This will result in p r e f e r e n t i a l orien-
[4]
tation w h e r e some grains can grow s e v e r a l times l a r g e r
than others and this is believed to be the s e c o n d a r y
recrystallization in polysilicon layers. This s e c o n d a r y
recrystallization has a quite different origin from that
in the m e t a l systems.
In this paper, the thickness effect is described by
m o d i f y i n g the g r a i n - b o u n d a r y e n e r g y t e r m according
to an e m p i r i c a l equation
[5]
~.--~.0/(1-t-h interfaeearea ) [9]
grain boundary area
w h e r e the constant h is to be d e t e r m i n e d from the e x -
p e r i m e n t a l results. The interface a r e a is a l w a y s 2 p e r
square of polysilicon, and the total g r a i n - b o u n d a r y
a r e a p e r square of polysilicon is calculated b y a s s u m -
ing that all grains are spherical and have the same size.
Impurity Segregation at the Grain Boundaries
Because Si self-diffusivity is a function of doping
concentration, the electrically active concentration
m u s t be d e t e r m i n e d at each t i m e step so as to calculate
the g r o w t h rate at t h a t instant in the simulation p r o -
gram. F o r the t e m p e r a t u r e s used in this study, we can
assume that the i m p u r i t y concentration segregated at
the g r a i n b o u n d a r i e s is a l w a y s in t h e r m a l e q u i l i b r i u m
w i t h the c o n c e n t r a t i o n in the grain, the ratio of ira-
Vol. 129, No. 8 GRAIN GROWTH IN POLYSILICON
p u r i t y concentration in the grain NG (NG = Nsi X CI)
to that of the total concentration'N1 is (6)
= exp 4- 1
NT LNsi
where Qs is the effective density of segregation sites at
the g r a i n b o u n d a r y ; its value, calculated from the re-
ported segregation data b y M a n d u r a h et al., is 2.64 •
1015 sites/cm 2. The e n t r o p y factor A changes with both
the dopant and its concentration: values reported b y
M a n d u r a h et al. are 3.02 for As and 2.46 for P both at
concentration of 2.0 • 10~9. For the grain size fre-
q u e n t l y observed after the usual process conditions, a
significant a m o u n t of As and P is segregated at the
grain boundaries, whereas boron shows no segregation
effect.
In the simulation model, the i m p u r i t y concentration
Ci in Eq. [8] at each time step is determined from the
segregation equation and the grain size is obtained
from the previous time step. The i n c r e m e n t of grain
size at a given time is thereby established from Eq. [4].
Results and Discussion
The effect of boron, arsenic, and phosphorus dop-
ants on g r a i n - g r o w t h e n h a n c e m e n t are to be analyzed
and compared. Boron was observed to have the least
e n h a n c e m e n t effect. Arsenic enhances grain growth at
a concentration above 1.0 X 10tg; however, the pres-
ence of clusters in the heavily arsenic-doped poly ap-
pears to p r e v e n t the g r a i n - b o u n d a r y migration, which
offsets e n h a n c e m e n t caused by the increase in vacancy
concentration. This is demonstrated by the decline of
the growth rate of heavily As-doped polysilicon. Phos-
phorus produces quite a different activation energy at
high concentrations, which is a t t r i b u t e d to the change
of the g r a i n - b o u n d a r y diffusion constant w h e n large
a m o u n t s of P are segregated at the bound,aries. Both
the variations of the diffusion activation er~ergy at the
grain boundaries and the g r a i n - b o u n d a r y energy as a
function of doping concentration can be d e t e r m i n e d by
a comparison o f simulated and e x p e r i m e n t a l l y mea-
sured results.
Grain growth o] B-doped polysilicon.--Boron doping
has little effect on the grain growth of polysilicon. Fig-
ure 1 shows the grain size as a f u n c t i o n of the reciprocal
of the processing t e m p e r a t u r e after 60 m i n of annealing.
The measured activation energy is a p p r o x i m a t e l y 0.5
eV at all three concentrations. The curves are the s i m u -
lated results obtained from the model discussed above.
The AGD value for this simulation is 3.7 eV. Because
the activation e n e r g y for Si self-diffusion is approxi-
mately 4.7 eV, the diffusion activation e n e r g y along the
g r a i n boundaries is d e t e r m i n e d to be a p p r o x i m a t e l y
1.0 eV. There is no a p p a r e n t variation of AGD with B-
doping concentrations. A s s u m i n g that diffusivity along
1.O I o 1.2 • 10 ~9 ANNEALING T I M E 6 0 rain
0.8! o 6 . 0 x 1020 POLY T H I C K N E S S 0 . 2 5 ~ m
x 1.2 x 10 21 DOPING ELEMENT 8
0.6i
:::L
0.4
UJ
N \
e3 "\. 8 -4:
Z Z
~0.2
0.1 , I
0.5 'o16 ' o.7' ' o 1 8 ' 1.0
l I T x 10.3 (OK)
Fig. I. The grain size v s . reciprocal of temperature of B-doped
poly after 1 hr of annealing.
1793
each grain b o u n d a r y is one to two orders of m a g n i t u d e
larger t h a n that in grains at 1000~ "~ is calculated to
be 1.0 • 10 - u . This would lead to a g r a i n - b o u n d a r y
[10] energy on the order of 1000 e r g / c m 2 which is c o m p a r -
able to m a n y such energies in the metal systems (12).
Recently, Makino and N a k a m u r a reported the grain
growth of boron-doped polycrystalline silicon (13).
Their micrographs show no enhanced grain growth for
alt concentrations at 1000~ i n a g r e e m e n t w i t h our
results. However, at 1100~ slight e n h a n c e m e n t was
observed for a concentration of 2 • 102~ while at a
higher concentration (3.3 • 10'~1) the grain growth is
suppressed as was observed for As-doped polysilicon
i n this work. They a t t r i b u t e d this suppressed grain
growth to the boron precipitations at the grain b o u n d -
ary. The discrepancy b e t w e e n e x p e r i m e n t a l results
from different sources can be caused b y various fac-
tors, such as the difference in deposition conditions
and doping techniques, which have not been consid-
ered here.
Grain growth o:f As-doped polysilicon.--As-doping
enhances grain growth at concentrations of 1.0 X 1019/
cm 3 and higher; however, this e n h a n c e m e n t is r e d u c e d
in very heavily doped polysilicon, as illustrated i n
Fig. 2. Clustering is believed to be responsible. This
effect is considered here b y d e t e r m i n i n g the electri-
cally active concentration according to
C1
- - - - 1 4- mKeCa m-1 [11]
Ca
where Ca is the electrically active concentration i n the
grain, and based on the results obtained by Antoniadis
et al. (14), m = 4 and Ke ~- 2.33 X 10 -67 exp 1.22 e V /
kT cm 9. The simulated results were plotted ir~ Fig. 2.
Figure 3 shows the g r a i n size vs. the reciprocal of the
processing temperatures. The m e a s u r e d activation e n -
ergies are b e t w e e n 0.4 and 0.5 eV. The AGb used to
obtain the agreeable s i m u l a t i o n results is 3.7 eV and
does not change significantly with doping. I n h e a v i l y
doped polysilicon, both the simulated and e x p e r i m e n t a l
results indicate that there is an a p p a r e n t drop in the
activation energy. Based on Eq. [8], this reduction is
the result of the segregation heat term w h e n the con-
centration becomes high.
Grain growth of P-doped polysilicon.--The effect of
phosphorus doping on grain growth varies from that
of boron and arsenic. At higher concentrations, doping
continues to enhance growth as illustrated in Fig. 4.
The presence of phosphorus changes both the g r a i n -
b o u n d a r y diffusion e n e r g y and the product "yiab2. At
present, it cannot be concluded which factor in the -~
product contributes this change. Empirically, the v a r i -
ation of "ykab2 and AGb as a function of doping con-
centration can be determined. This variation in grain-
b o u n d a r y diffusion activation e n e r g y results in a large
1.O
0.8 As DOPED POLY 0 . 2 5 ~ m
A N N E A L E D FOR I hour
0.6
&"
=L
~0.4
tu
N
"+-- ---- ~"- . . . .
.............................. ~ ~'.~
n ' o . 2 .z~
(.9 .......... o ...... s ......................... "~ ~" 1 2 0 0 ~
a """o .... ~ 1 1 0 0 ~
......... 1 0 0 0 ~
0.1 --v ~ ~ I I I ~ ~Jl ~ ~ ~ ~ ~ ~'~'4 90O~
, , 1019- 1020 1021
I.I
As DOPING C O N C E N T R A T I O N
Fig. 2. The variation of average grain size as a function of As
doping concentration annealed at four different temperatures for
1 hr.
1794 J. Electrochem. Sac.: S O L I D - S T A T E S C I E N C E A N D T E C H N O L O G Y August 1982

1.0 0.25 Fm POLY 1.0 0 . 2 5 p, rn POLY

0.8 As D O P E D POLY 0.8 P DOPED
\ o 1.2 x 1019 - -
[3 1.2 x 1019
0.6 0.6 ~ o 6.O x 1019 ............
o 6 . 0 x 1019 . . . . .
~. ,~ ~% X 1.2 x 1020
A 1.2 x 1020 .........
9. x ~ 0 6 . O x 1020 . . . .
+ 6.0 x 1020 ....
~0.4
lJl "-'& 0.4 .,o,, ",
N N \\ \
U) @ ",.\ \
Z z =-',\
<
r',- O. 2 r,. 0,2 ~ i
,,.o r ",.)~?\%

+
O. 1 I f I O. 1 I I I l IO"~. I I I I I
0.5 1.0 1.1 0.5 O16 O.7 0,8 0,g 1.0 1.1
l I T x 10 "3 (OK) lIT x 10"3(~

Fig. 3. The grain size vs. reciprocal af temperature of As-doped Fig. 5. The variation of grain size as a function of reciprocal
poly after 1 hr of annealing. temperature for P-doped poly annealing for 1 hr.

change in g r a i n - g r o w t h activation e n e r g y (Fig. 5). A t I IEXR I SIMULATION

0.31- PHOSPHORUS CONC. I IDATAI RESULTS
high concentrations, g r a i n size increases r a p i d l y as a n - [ INPOLY1"2• 1~ ~ I.......
nealing t e m p e r a t u r e is raised. L a r g e grains on the L ANNEALING TEMR: IO.50~.m I • /.......
o r d e r of several microns can be o b t a i n e d w i t h h e a v y P | 95ooc l~176 t
doping. ~o.2 F ~ - - ~_ . .-. . _:l
LU ~- ~.~-.-'~- ~ " . . . .
G r a i n size as a function of annealing time for four
polysilicon thicknesses has been i n v e s t i g a t e d e x p e r i -
m e n t a l l y and b y simulation, the results o b t a i n e d from
t h r e e thicknesses a r e p l o t t e d in Fig. 6. Despite the ef-
fect of a v a r i a b l e active concentration t h r o u g h o u t a n -
nealing, the i n c r e m e n t of g r a i n size follows a X/{ d e - o; , ~'o ' ,;o ' ,;o ' 2;o
TIME (In[n)
pendence. The v a r i a t i o n s are not significant. The e m -
pirical p a r a m e t e r h in Eq. [9] was d e t e r m i n e d to be Fig. 6. The change of average grain size as a function of an-
a p p r o x i m a t e l y 8 b y m a t c h i n g t h e s i m u l a t e d and e x - nealing time at 950~ for three different thicknesses of poly.
p e r i m e n t a l data.
Conclusion
Acknowledgment
The g r a i n - g r o w t h mechanisms of polysilicon w i t h
the t h r e e dopants (As, P, and B) have been i n v e s t i - This w o r k was s u p p o r t e d i n i t i a l l y b y ARO Contract
gated. E x p e r i m e n t a l results indicate t h a t the g r o w t h DAAG-29-77-C-0006, and then b y D A R P A Contract
rate is enhanced b y n - t y p e dopants, d e p e n d i n g on the DAAB-07-C-2684 u n d e r the direction of Dr. R i c h a r d
concentration levels. A m o d e l based on a diffusion- Reynolds. The authors a r e also g r a t e f u l to Dr. A n n
controlled m e c h a n i s m was introduced, and the results M a r s h a l l for h e r assistance in the TEM work, and Dr.
o b t a i n e d are in good a g r e e m e n t w i t h the e x p e r i m e n t a l J a m e s P l u m m e r and Dr. K r i s h n a S a r a s w a t for their
data. The different effects of As, P, and B a p p e a r to critical readings.
be a consequence of t h e i r difference in cluster f o r -
mation, segregation, a n d the change of g r a i n - b o u n d - M a n u s c r i p t s u b m i t t e d Jan. 8, 1981; revised m a n u -
a r y e n e r g y caused b y segregation. The values of g r a i n - script received M a r c h 16, 1981.
b o u n d a r y free e n e r g y and activation e n e r g y of g r a i n -
b o u n d a r y diffusion a r e in a g r e e m e n t w i t h the physical A n y discussion of this p a p e r will a p p e a r in a Discus-
reasoning, a l t h o u g h m o r e precise m e a s u r e m e n t s are sion Section to be p u b l i s h e d in the J u n e 1983 JOURNAL.
r e q u i r e d to d e t e r m i n e these parameters. The g r o w t h A l l discussions for the J u n e 1983 Discussion Section
r a t e d e p e n d e n c e on the p r e f e r r e d grain o r i e n t a t i o n as should be s u b m i t t e d b y Feb. 1, 1983.
a result of different deposition conditions in a d d i t i o n
to t h e g r o w t h m e c h a n i s m d u r i n g the oxidation h a v e REFERENCES
not been considered. F u r t h e r studies of these d e p e n -
i. P. M. Solomon, D. D. Tang, a n d R. D. Issac, lED_M,
dences should produce a d d i t i o n a l i n f o r m a t i o n concern- 510 (1979).
ing the g r a i n g r o w t h m e c h a n i s m discussed in this 2. L. G e r z b e r g and J. Meindl, IEDM Tech. Dig., 502
paper. (1979).
3. W. R. Hunter, L. Ephrath, W. D. G r o b m a n , C. M.
/ Osburn, B. L. Crowder, A. Cramer, and H. E.
5 Luhn, IEEE Trans. Electron Devices, ed-26, 853
+/1200 ~ (1979).
"7 4 4. J. Y. W. Seto, J. Appl. Phys., 46, 5247 (1975).
5. G. Baccarard, B. Ricco, a n d G. Spadini, ibid., 49,
uJ 5565 (1978).
N
~9 6. M. M a n d u r a h , K. Saraswat, R. C. Helm, a n d T.
Z .~ lOOO % ~ . ~ ~/~ Kamins, ibid., 51, 5755 (1980).
<{
rr .- o / / 7. H. Sunami, This Journal, 125, 892 (1978).
8. Y. W a d a and S. Nishimatsu, ibid., 125, 1499 (1978).
9. K. T. Aust and J. W. Rutter, in "Recovery and Re-
019 1020 1021 crystallization of Metals," L. Himmel, Editor, p.
DOPING CONCENTRATION ( l / c m 3)
131, Interscience Publishers, New Y o r k (1962).
10. D. Gupth, D. R. Campbell, and P. S. Ha, "Thin
F i l m s - - I n t e r d i f f u s i o n and Reaction," p. 161,
Fig. 4. The variation of average grain size of P-doped poly after W i l e y - I n t e r s c i e n c e Publication, New Y o r k (1980).
1 hr of annealing at three different temperatures. 11. C. P. Ha a n d J. D. P l u m m e r , This Journal, 126,
Vol. 129, No. 8 G R A I N G R O W T H IN P O L Y S I L I C O N 1795

1516 (1979). 13. T. Makino and H. N a k a m u r a , Solid State Electron.,

12. K. Lucke and H. P. Stuwe, in "Recovery and Re- 24, 49 (1981).
crystallization of Metals," L. Himmel, Editor, p. 14. D. A. Antoniadis and R. W. Dutton, IEEE J. Solid
171, Interscience Publishers, New York (1962). State Circuits, sc-14, 412 (1979).

Schottky Contact Fabrication for GaAs MESFET's

T. H. Miers*
Motorola, Incorporated, Semiconductor Research and Development Laboratories, Phoenix, Arizona 85008

ABSTRACT
An investigation of cleaning procedures prior to metal gate evaporation in GaAs MESFET fabrication was undertaken.
A photoresist residue was found to exist after development when using the lift-offtechnique to define the gate metal pat-
tern. This carbon residue was identified by AES depth profile measurements. Schottky diode electrical characteristics
were observed to deteriorate due to this residue. Of the cleaning solutions examined, only the NH40H solution removed
both C and O layers on the GaAs surface while stillbeing compatible with the fabrication process. The effects ofDI H20 rinse
time on Schottky diode electrical performance were also examined.

A n u m b e r of w o r k e r s have i n v e s t i g a t e d the p r o p - maintained. Also, the resist edge profile (negative

erties of S c h o t t k y contacts to GaAs (1-6). F o r the
most part, this w o r k has involved the e x a m i n a t i o n of
m e t a l l i z a t i o n schemes and surface cleaning p r o c e d u r e s
for GaAs. In general, these cleaning t r e a t m e n t s have
not been r e s t r i c t e d to a n y p a r t i c u l a r t y p e of f a b r i c a -
tion procedure, but m a i n l y have been a i m e d at r e -
moving oxides from the GaAs surface and leaving the
surface s t o i c h i o m e t r i c a l l y intact (4, 5, 7, 8). Other
w o r k e r s (3, 9-14) have c o n c e n t r a t e d on the effects of
t h e r m a l processing on the electrical and m e t a l l u r g i c a l
p r o p e r t i e s of S c h o t t k y b a r r i e r s on GaAs.
U n f o r t u n a t e l y , these h e a t - t r e a t m e n t cycles a n d
cleaning p r o c e d u r e s are not a l w a y s a d a p t a b l e to the
fabrication of GaAs MESFET's. F o r example, a con-
ventional fabrication p r o c e d u r e for M E S F E T ' s consists
of the following steps: (i) photoresist definition of
device mesa w i t h subsequent chemical etch of the
active l a y e r for device isolation; (if) photoresist defi-
nition of ohmic source and d r a i n contact pattern, m e t a l
evaporation, and m e t a l definition b y a lift-off process
w i t h subsequent ohmic alloy; and (iii) p h o t o r e s i s t
definition of S c h o t t k y gate p a t t e r n , m e t a l evaporation,
and m e t a l definition b y a lift-off process. There a r e
p r o c e d u r e s i n c o r p o r a t e d in each of these steps to
r e m o v e photoresist and oxide l a y e r s f r o m the GaAs
surface. Also, a p r e - e v a p o r a t i o n clean is u s u a l l y p e r -
f o r m e d a f t e r photoresist d e v e l o p m e n t to "clean up"
the surface p r i o r to deposition of the m e t a l contacts.
It iz'this p r e - e v a p o r a t i o n clean that is the subject of
this paper. The cleaning p r o c e d u r e is i m p o r t a n t for
both ohmic and S c h o t t k y contacts to GaAs. This w o r k
involved S c h o t t k y contacts, a l t h o u g h the results are
d i r e c t l y a p p l i c a b l e to ohmic contacts as well (15).
Hence the objective is to develop a p r e - e v a p o r a t i o n
clean t h a t is c o m p a t i b l e with GaAs M E S F E T f a b r i c a -
tion p r o c e d u r e s and yields S c h o t t k y contacts t h a t
have good electrical p r o p e r t i e s w i t h o u t e x t r a t h e r m a l
processing.
Requirements
In the f a b r i c a t i o n of GaAs MESFET's, there a r e
several strict r e q u i r e m e n t s to which the p r e - e v a p o r a -
tion clean p r o c e d u r e m u s t adhere. F i r s t the photoresist
must not be n o t i c e a b l y attacked. T h a t is, the resist
obviously cannot be r e m o v e d or lifted from the GaAs.
The photoresist l i n e / s p a c e w i d t h cannot be a l t e r e d
b y the cleaning procedure. This is especially i m p o r t a n t
because of the s m a l l (1 ~m) gate lengths that m u s t be
* Electrochemical Society Active M e m b e r .
Key words: gallium arsenide, Sehottky barrier, photoresist res-
Idue.
t a p e r or "lip") must be m a i n t a i n e d to achieve a clean,
s h a r p m e t a l lift-off w i t h o u t l e a v i n g r a g g e d edges or
m e t a l "wings" on the defined m e t a l p a t t e r n .
Second, the cleaning p r o c e d u r e should not a p p r e c i -
a b l y etch the GaAs. Since the device active l a y e r is
only 1000-2000A thick, v e r y little etching of t h e s u r -
face can be t o l e r a t e d without significantly reducing the
device pinch-off voltage. Third, the cleaning p r o c e d u r e
m u s t r e m o v e photoresist scum as well as surface
oxides and contaminants without l e a v i n g a n y residue
itself. Fourth, the contact that is p r o d u c e d m u s t e x -
h i b i t good m e c h a n i c a l (adhesion) and electrical p r o p -
erties. However, these two items are d e p e n d e n t on the
semiconductor and m e t a l p r o p e r t i e s as w e l l as the
cleaning procedure.
Experimental Procedure
The e x p e r i m e n t a l p r o c e d u r e was designed to m a t c h
the n o r m a l M E S F E T f a b r i c a t i o n p r o c e d u r e s as closely
as possible. A l l of the w o r k was p e r f o r m e d in the
same clean r o o m areas that are used for M E S F E T f a b r i -
cations. The same p r e c a u t i o n s and p r o c e d u r e s w e r e o'b-
served.
Bulk S t - d o p e d GaAs wafers (n ~ 2.5 • 101~ c m - ~ )
polished on one side were d e g r e a s e d in organics,
etched slightly, and cleaned in dilute HC1 and in
dilute NH4OH. A u G e / N i ohmic contacts were d e -
posited on the back (unpolished) side and a l l o y e d in
f o r m i n g gas at 450~ The wafers were then cleaned
again w i t h 10 H2D: 1HC1 and 10 H20: 1NH4OH followed
by a 30 sec DI H~O rinse. One piece of each w a f e r was
then cleaved off and designated as the control sample.
The r e m a i n d e r of each of the wafers was processed
t h r o u g h the s t a n d a r d M E S F E T gate photoresist p r o -
cedure. This consisted of the s p i n - o n application of
AZ1350J photoresi~t ( S h i p l e y C o r p o r a t i o n ) , soft-bake,
complete exposure, and d e v e l o p m e n t w i t h 1 AZ d e -
veloper: 1H20 followed b y a DI H20 rinse. The wafers
were then cleaved into four s a m p l e s w i t h each piece
receiving a different p r e - e v a p o r a t i o n clean procedure.
These p r e - e v a p o r a t i o n clean p r o c e d u r e s are outlined
in Table I.
The samples from o n e - h a l f of the wafers t h a t w e r e
treated as described in Table I r e c e i v e d no f u r t h e r
processing. These were stored in an N2 d r y b o x w i t h
the control sample from each w a f e r u n t i l A u g e r e l e c -
tron spectroscopy m e a s u r e m e n t s could be p e r f o r m e d .
S a m p l e s f r o m the r e m a i n i n g o n e - h a l f of the wafers
(along w i t h the control samples from these w a f e r s )
were i m m e d i a t e l y placed in an E - b e a m evaporator