Solid Solubility and Diffusion Coefficients of

Boron in Silicon
G. L. Vick and K. M. Whittle
Instrument~Controls Division, Conrac Corporation, Duarte, California

ABSTRACT
The solid solubility and diffusion coefficients of boron in silicon have been
determined as a function of t e m p e r a t u r e over the range of 700~176 by
anodically sectioning diffused layers. The solid solubility was found to v a r y
from 1.6 x 1019 a t o m s / c m 3 at 700~ to 2.4 x 1020 a t o m s / c m ~ at 1151~ The
diffusion coefficients for i m p u r i t y levels below l0 TM a t o m s / c m 3 m a y be r e p r e -
sented by D = 6.0 x 10 -T exp (--38600/RT). The diffusion coefficients above
10TM a t o m s / c m 3 were found to be d e p e n d e n t on the i m p u r i t y level.

I n view of the importance of boron as a dopant in formed to d e t e r m i n e w h e t h e r there was precipitation

the technology of silicon electronic devices, there is of the boron such as that found for phosphorus by
a r e m a r k a b l e paucity of i n f o r m a t i o n regarding the T a n n e n b a u m (6). Diffused wafers were cooled from
solid solubility of boron in silicon. I n T r u m b o r e ' s re- 1150~ to room t e m p e r a t u r e in times r a n g i n g from 1
view of solid solubilities of elements in g e r m a n i u m rain to 1 hr w i t h o u t preceptible change i n profile,
and silicon (1), boron is represented b y only two leading to the conclusion that precipitation was not
points, a eutectic point and a single point based on occurring to a n y m e a s u r a b l e amount.
diffusion data. Sectioning of the diffused i m p u r i t y profiles was ac-
This paper describes a series of experiments in complished by anodically growing and then r e m o v i n g
which boron is diffused into silicon at a n u m b e r of layers of silicon dioxide. Anodic oxide growth was
different t e m p e r a t u r e s b e t w e e n 700 ~ and 1151~ from carried out in a solution consisting of 3 parts e t h y l e n e
an effectively infinite source. The resulting diffusion glycol and 1 part phosphoric acid at 100V. The oxide
profiles are anodically sectioned to determine the solid was removed with HF and the anodization repeated.
solubility of boron in silicon as a function of t e m p e r a - After e v e r y fifth a n o d i z a t i o n - H F cycle a m e a s u r e m e n t
ture. The diffusion coefficients of boron in silicon are was made of sheet resistivity and depth of silicon re-
also calculated as a function of t e m p e r a t u r e and boron moval. Sheet resistivity was m e a s u r e d by a standard
concentration. 4-point probe method. Silicon removal was measured
by m a s k i n g off a region of the wafer d u r i n g anodiza-
Experimental Procedure t i o n - H F cycles a n d m e a s u r i n g the height of the r e s u l t -
A boron rich glass (B203) is pyrolytically deposited ing plateau by means of interference fringes. I r w i n ' s
on the surface of 3.0 to 6.5 o h m - c m n - t y p e wafers at (7) plot of resistivity vs. i m p u r i t y concentration was
relatively low t e m p e r a t u r e (688~ The wafers are used to determine the i m p u r i t y concentration of suc-
t h e n diffused at higher t e m p e r a t u r e s in an nonoxidiz- cessively removed layers of boron concentration.
ing atmosphere (Nf). U n d e r these conditions the boron
glass is reduced by the silicon, m a k i n g elemental boron Accuracy
available for diffusion into the silicon. The deposition The diffusion t e m p e r a t u r e s were controlled to _ 2 ~
and diffusion systems have previously been described and diffusion times were controlled to --+0.05 rain.
by Whittle and Vick (2). The reduction of the B20~ The 4-point probe m e a s u r e m e n t s on a given sample
makes boron atoms available at a rate greater than were consistent from r e a d i n g to reading w i t h i n 1%.
the rate at which they are diffused away into the The depth of silicon r e m o v a l was d e t e r m i n e d b y
silicon. This results in a t h i n layer of a b o r o n - r i c h calibrating the silicon removal per anodization cycle
phase such as that discussed b y Busen et al. (3). This against depth m e a s u r e m e n t s made by means of i n t e r -
boron rich phase will act as an effectively infinite ference fringes. The calibration was made to a depth
source of boron for diffusion into the silicon. Similar of 4~ with an accuracy of _--_~/4 fringe, giving a possible
phases m a y be observed in phosphorus diffusion and consistent error of about 7%. I n addition, the r a n d o m
have been used by Kooi (4) in d e t e r m i n i n g solubility errors, represented as deviations of data points from
of phosphorus in silicon. the calibration curve a m o u n t e d to about --+5% rms.
Experiments have been performed which show that The plots of i m p u r i t y concentration vs. depth shown
the surface concentrations are i n d e p e n d e n t of all in Fig. 1 and 2 involve the first derivative of the curve
deposition parameters over a wide range [ref. (2)] of conductivity vs. depth which was e x p e r i m e n t a l l y
and are i n d e p e n d e n t of diffusion time for all the differ- determined. The deviation of data points from comple-
ent values of diffusion time reported in this paper m e n t a r y error functions in Fig. 1 a n d from the best
and are dependent only on diffusion temperature. The smooth curves in Fig. 2 are -+ 13 % rms.
work of Owen and Schmidt (5) could be i n t e r p r e t e d A significant source of inaccuracy in the values of
as predicting a surface concentration i n d e p e n d e n t of diffusion coefficient at t e m p e r a t u r e s above 1000~ lay
time u n d e r certain conditions. This work is not directly in graphical integration and differentiation of the i m -
applicable here, however, because the rate of chemi- p u r i t y profiles. These procedures are somewhat sub-
cal decomposition of the B20~ controls the rate at jective in n a t u r e and it was found that variations in
which boron is made available for diffusion. This is diffusion coefficient as large as -+100% could be
demonstrated b y the fact that the doping will be i n - achieved b y differing estimations of the slopes. The
hibited by the presence of oxygen in the atmosphere. values reported in Fig. 4 represent the best j u d g m e n t
The presence of the b o r o n - r i c h phase ensures an ade- of the authors as to the i n t e r p r e t a t i o n of existing data.
quate supply of boron so that surface concentration At t e m p e r a t u r e s less t h a n 1000~ the diffusions are
will be d e t e r m i n e d b y the solid solubility. too shallow to justify accuracy greater t h a n -+100%.
In view of the departures from c o m p l e m e n t a r y
error function form of the profiles obtained at the Solid Solubility
higher t e m p e r a t u r e diffusions, experiments were per- I n Fig. 1 a n d 2 are plotted the curves of i m p u r i t y
K e y w o r d s : Solid solubility, diffusion aoefflcient, b o r o n , silicon. vs. depth. The curves for diffusions below 1018~ are

1142

Downloaded on 2017-04-17 to IP 14.139.128.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 V o l . 116, N o . 8 SOLUBILITY AND COEFFICIENTS OF BORON 1143
I I I I I I ~ r i i i
The solid solubility data indicate a lower solubility
of boron in silicon t h a n had previously been reported.
There is, however, little actual data to support a higher
value of solubility. Diffusion profiles having larger sur-
face concentrations have been reported. These, how-
ever, have been calculated on the assumption of a
pxio~ c o m p l e m e n t a r y error function distribution which is
shown to be false for t e m p e r a t u r e s above 1000~
Diffusion Coefficients
For those diffusion profiles which m a y be r e p r e -
sented by the familiar c o m p l e m e n t a r y error function,
I X 1019 ~ , , . ~I Cx = Cserfc ( x / 2 ( D t ) ' / 2 ) , calculation of diffusion co-
efficients is s t r a i g h t f o r w a r d since C~ is the c o n c e n t r a -
o tion of the diffusant at distance x after time t and C~
is the extrapolated surface concentration. All the nec-
essary data can be obtained from the diffusion profiles
i 42 HRS AT 700~
i n Fig. 1.
I X IO ts I I J The profiles which do not fit a c o m p l e m e n t a r y error
0 O-I 0,2 O,S function must be dealt with differently. The possi-
DEPTH FROM SURFACE, MICRONS bilities of source depletion or of i m p u r i t y precipitation
as explanations for the shape of these curves have
Fig. 1. Boron concentration vs. depth from surface been ruled out as described earlier. It m a y be assumed,
then, that the profiles result from a concentration
I XI021 I I I I dependent diffusion coefficient similar to that found
by T a n n e n b a u m (6) for phosphorus. The diffusion
coefficients m a y be calculated as a function of i m p u r i t y
concentration b y a method presented by C r a n k (8)
using the equation
, o o
1 d x ~oCl
I X IO 20
Dc=cl = ~ ~ xdC
2t dc
w h e r e D is the diffusion coefficient at a c o n c e n t r a t i o n
03
C, and x and t are depth and time, respectively.
o~ A p p l y i n g this equation to the curves in Fig. 2 results
AT 1151~ in the dependence of diffusion coefficient on i m p u r i t y
o= IXIO 19 concentration shown in Fig. 4. The slope d x / d c and the
oc
a r e a f : l XdC were determined graphically so that the
o= curves in Fig. 4 should be considered to show a q u a l -
itative t r e n d only. A n accurate d e t e r m i n a t i o n of dif-
o
fusion coefficient as a function of i m p u r i t y concentra-
tion could be made by subjecting the i m p u r i t y profile
IXlO J8 data to computation by a computer. These curves in
Fig. 4 are seen to approach a m i n i m u m value of the
diffusion coefficient at each t e m p e r a t u r e at which the
diffusion coefficient is i n d e p e n d e n t of i m p u r i t y con-
centration.
A plot of the logarithm of the diffusion coefficient vs.
IXlO 17 I inverse t e m p e r a t u r e should yield a straight line since
0 1.0 2.0 3.0 4.0 5.0 6.0 the t e m p e r a t u r e dependence of the diffusion coefficient
DEPTH FROM SURFACE , MICRONS follows the form
Fig. 2. Boron concentration vs. depth from surface D = Do exp ( E J R T )
where Ea is the activation energy for diffusion, R is
found to fit well the c o m p l e m e n t a r y error function. the gas constant, and T is the absolute temperature.
For diffusion at 1018~ and higher the curves are sim-
ply the smooth curve which best fits the data. Solid
solubilities are obtained b y extrapolating the diffusion Ic~IO I IIIIIII I I1[11111 I IIIIIHI I lllllll:
profiles to zero depth and are plotted vs. t e m p e r a t u r e
in Fig. 3. /

~lXlO ~ I I I I f

Ij
~ J~7~2
E
1
~ IXIO 20 o

w "I j I / ~
~U- IOI3
U.
|

g
1(514 I I IIl]lll I II~ll III i i ilitlll i i i i ill
I X JO 19 1017 jo ~8 i0 19 i0 20 Io 21
600 700 800 900 I000 I100 1200
TEMPERATURE I~ BORON CONCENTRATION , ATOMS/ CM 2

Fig. 3. Boron solid solubility vs. temperature Fig. 4. Diffusion coefficient vs. boron concentration

Downloaded on 2017-04-17 to IP 14.139.128.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 1144 J. Electroehem. Soc.: S O L I D S T A T E S C I E N C E A u g u s t 1969
I X l O II i I
tion are represented on Fig. 5 by a data point indicat-
ing the value of the concentration i n d e p e n d e n t coeffi-
cient, together with a vertical line indicating the r a n g e
of values which hold at higher concentrations.

tXt(~ 12 Summary
Values for the solid solubility of boron as a function
of t e m p e r a t u r e i n silicon are presented in this paper
for the first time.
Values of diffusion coefficients for boron in silicon
are given over a r a n g e of t e m p e r a t u r e almost 300~
1•
greater t h a n was previously available (700~176
(9-11). Concentration d e p e n d e n t diffusion coefficients
of boron in silicon are reported.
|
CI | 0 =6.0 X I0 "7 EXP (-38,600) Manuscript s u b m i t t e d Oct. 7, 1968; revised m a n u -
RT script received April 11, 1969.
,3
ixr6"
A n y discussion of this paper will appear in a Dis-
3 cussion Section to be published in the J u n e 1970
JOURNAL.
REFERENCES
1. F. A. Trumbore, Bell Systems Tech. J., 39, 205
i X1(5~5 (1960).
2. K. M. Whittle and G. L. Vick, This Journal, 116,
645 (1969).
3. K. M. Busen, W. A. Fitzgibbons, and W. K. Tsang,
ibid., 115, 291 (1968).
4. E. Kooi, ibid., 111, 1383 (1964).
[XIO 16 I J I I 5. A. E. Owen and P. F. Schmidt, ibid., 115, 548
.6 ,7 ,8 .9 1.0 I.I (1968).
t03/TEMPERATURE ~ 6. E. T a n n e n b a u m , Solid State Electronics, 2, 123
(1961).
Fig. 5. Boron diffusion coefficient vs. temperature 7. J. C. Irwin, Bell Systems Tech. J., 41, 387 (1962).
8. J. 'Crank, "The Mathematics of Diffusion," p. 232,
The diffusion coefficients are plotted against the in- Oxford U n i v e r s i t y Press (1956).
verse of t e m p e r a t u r e in Fig. 5 and can be shown to 9. E. C. Williams, This Journal, 1{)8, 795 (1962).
10. A. D. K u r t z and R. Yee, J. Appl. Phys., 31, 303
follow the equation D = 6.0 x 10 -7 exp(--38600/RT). (1960).
Those diffusion coefficients (above 1000~ for which 11. C. S. F u l l e r and J. A. Ditzenberger, ibid., 27, 544
the coefficients are dependent on i m p u r i t y concentra- (1955).

~|,,,
Technca]l
The Estimation of Bending Stresses from
Flexure Measurements
R. E. Pawel*
Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee

The flexure technique, based on the physical defor- flexure experiments on various kinds of specimens.
mation observed w h e n a suitably prepared specimen (This particular problem is, of course, entirely separate
is reacted on one side, has recently been employed in from the u s u a l l y more tedious one of i n f e r r i n g the
several investigations in which the characterization complete stress distribution from the b e n d i n g stress
and m e a s u r e m e n t of stresses induced by surface or values.) While it might be argued that this u n c e r t a i n t y
n e a r - s u r f a c e reactions were of interest. Originally the is not a serious d r a w b a c k because of the m a n y other
tool of the electroplater, this t e c h n i q u e is now being sources of error involved with the technique, such a
used in other disciplines to collect special types of possibility nevertheless w a r r a n t s consideration, and it
i n f o r m a t i o n which are otherwise difficult to obtain. is the i n t e n t of the present note to express some
Since the principal e x p e r i m e n t a l m e a s u r e m e n t s i n thoughts a n d the results of an e x p e r i m e n t which ap-
flexure studies are ones of strain, the problem of t r a n s - pear p e r t i n e n t to the m a n n e r i n which b e n d i n g stresses
f o r m i n g or assigning appropriate stress values must be are d e t e r m i n e d from b e n d i n g strains.
faced. The general methods for doing this have existed Since the stress system imposed on the specimen m a y
in the l i t e r a t u r e for some time, b u t some u n c e r t a i n t y be considered to be uniform, it is helpful for purposes
has arisen recently concerning the extent and i m p o r - of this p r o b l e m to t h i n k only in terms of b e n d i n g
tance of the Poisson (transverse) contributions d u r i n g stresses a n d to visualize the surface reaction that cre-
* Electrochemical Society Active Member. ates them, w h a t e v e r it m a y be, as resulting in simple

Downloaded on 2017-04-17 to IP 14.139.128.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).