Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
il~4785V
BACKGROUND OF THE INVENTION
The prior art method for forming polycrystalline
silicon on wafers is run in a hot-wall furnace using nitro-
gen gas as the carrier gas and silicon tetrahydride as the
source of silicon. The furnace is given a heat profile
which resembles a ramp beginning at the source end and
increasing towards the exhaust end of the tube. Each
furnace is independently profiled such that there is a rain-
out profile giving the most uniform front-to-back, and
side-to-side poly deposition as is possible within the
system.
According to the prior art practice, the quartz boat is
placed in the rain-out area of the furnace and wafers are
placed side-by-side with one broad surface on the boat such
that the deposition of the polycrystalline silicon occurs
over the opposing and upturned second major surface.
Because of the placement of the wafers on their flat sur-
faces, approximately 12 to 20 wafers fit within the deposi-
tion zone of the urnace at any one time. Normally, two
lines of wafers are placed on the boat. The deposition
profile of the polycrystalline material on the wafer appears
bell-shaped when taken on a straight line across the wafer
and perpendicular to the flow of gas. This means that in ~ -
the center of the wafer, the polycrystalline material is
thicke~t and at the edge of the wafer, it is thinnest.
Normally, an average thickness of 4500 Angstroms is chosen ~ ~-
with the thickest material in the center at 6000 Angstroms
and the edge thickness of the material at 3000 Angstroms.
In practice, the center portion of 6000 Angstroms thick ~;
material can be too thick for the manufacture of devices on
that wafer while the 3000 thick Angstroms of polycrystalline
silicon can be too thin for the successful fabrication of - -
- .
.
.
~0478S~
devices. Accordingly, some workable devices are fabricated
in the intermediate area where the polycrystalline thickness
is typically 4500 Angstroms. A third drawback in this
system is the wafers are placed into the furnace from the
exit end, the brown, powdered silicon material oftentimes
drops off the walls onto the wafer as they are being pulled
out of or put into the furnace. The buildup of such
powdered silicon material on the quartz tube is so rapid
that normally a maximum of ten to twenty runs can be made
using the same quartz tube. When the quartz tube is pulled
from the furnace to be cleaned, the coefficient of expansion
between the quartz and the silicon is so great that the
quartz tube breaks and a new center section must be fused
with the unbroken end sections as to reuse these two end
portions of the tube.
In summary, the prior art process for deposit poly-
crystalline on wafers has the problems of low throughput;
i.e., 12 to 20 wafers at a time, non-uniformity of deposi-
tion of material +1500 Angstroms across the surface of the
wafers, and the wafers are put in from the exit end of the
tube subjecting them to the flaking off of powdered silicon
material which then falls onto the wafers either prior to
polycrystalline silicon deposition or after such polycrys-
talline silicon deposition. Such deposition of granular
silicon renders the adjacent area unfit for the fabrication
of devices.
SUMMARY OF THE INVENTION
The present invention relates generally to a process
and product for the deposition of polycrystalline silicon
on a substrate in a heated tube, using a gaseous source and
- 3 -
1~347850
a vacuum and, more particularly, the present invention
relates to a process and product for depositing polycrystal-
line silicon on a substrate in a heated tube using silicon
tetrahydride as the source gas, and using a vacuum.
It is an object of the present invention to provide a
new improved method for depositing polycrystalline silicon
on a substrate using a heated tube and a vacuum.
It is an additional object of the present invention to ;
provide a hot-wall furnace tube having a flat temperature
profile in the hot zone of the tube which results in the
deposition of a uniform thickness of polycrystalline silicon
over the major portion of the surface of the wafer.
It is still a further object of the present invention
to provide a method for depositing polycrystalline silicon
from silane in a hot-wall furnace tube with improved uni-
formity as compared to that presently possible.
It is another object of the present invention to pro~
vide a means and a method for depositing polycrystalline
silicon from silane at a uniform rate over the wafer with a
tolerance of better than 500 Angstroms from edge to edge of
a single wafer.
It is still a further object of the present invention
to provide a method for depositing polycrystalline silicon
on wafers within a hot-wall furnace tube for increasing
the throughput capacity of the furnace. ~-~
It is a further object of the present invention to
provide a method for depositing polycrystalline silicon on
wafers in a hot-wall tube such that the major surface of the
wafers contain a uniform thickness of the polycrystalline ~
silicon over a~ great a region as possible. -
It is another object of the present invention to pro-
vide a method for depositing polycrystalline silicon on ~ - -
- 4 -
lQ47850
wafers in a furnace wherein the wafers are placed on edge
and their broad major surfaces are then placed perpendicular
to the flow of the source gas and the wafers are placed
closer together than previously thought possible.
It is an additional object of the present invention to
provide a method for depositing polycrystalline silicon on
many different substrates used in the manufacture of semi-
conductor devices at temperatures over 600C, such as sili-
con, germaniumj-sapphire, spinel, ceramic, silicon dioxide,
and refractory metals such-as tungsten,-molybdenum.
It is a stil-l further object of the present invention
to provide a method of depositing polycrystalline silicon in -
a heated tube, which method is independent of the means for
heating the tube and heating means such as RF, resistance or
radiant heat can be used.
It is another object-of the present invention to pro~
vide a method for depositing polycrystalline silicon on a
substrate using several gas sources such as silane, SiC12H2,
SiClH3, and SiC14.
It is a further object of the present invention to -
provide-a method for deposi-ti-ng polycrystalline silicon in
a heated evacuated tube under a vacuum on a substrate and
using a gaseous source, wherein the spacing of the wafers is
minimized for increasing the throughput of the method.
The invention of a method of forming a polycrystalline
silicon layer upon a plurality of silicon wafers, comprises the
steps of placing a plurality of silicon wafers in a closed
container; establishing a uniform temperature throughout a
portion of the closed container, said temperature being
less than 700C, and said wafers being placed within the
uniform temperature portion of the container; placing the
wafers on end perpendicular to the flow of gas and spaced one
from the other by a distance greater than 30 mils; and passing
~ '
~ _ 5 _
,
: .
1047850
a gas stream of silane over ~he heated wafers under the
motivation of a vacuum. Preferably the vacuum is established
between the range of 600 to 1600 millitorr.
More particularly, ~here is provided:-
a method of forming a polycrystalline silicon
layer upon a plurality of silicon wafers, comprising the steps
of:
placing a plurality of silicon wafers in a closed
container;
heating said wafers to a temperature less than
700C;
passing a gas stream of silane over the heated
wafers under the motivation of a vacuum established between
the range of 600 to 1600 millitorr; and
placing the wafers on end perpendicular to the
flow of gas and spaced one from the other by a distance greater
than 30 mils. -
There is also provided:-
a method of forming a polycrystalline silicon
layer upon a plurality of d licon wafers, comprising the teps
of:
placing a plurality of s~licon wafers in a closed
container;
heating said wafers to a temperature less than
700--C,
placing the wafers on end perpendicular to the flow
of gas and spaced one from the other by a distance600 to 1600
mlllitorr; and ~ ~ -
pass~ng 8 gas stream of silane over the heated --
w~fers under the motivation of a vacuum.
~ 5a
'~ .
1~47850
DESCRIPTION OF THE FIGURES
Fig. 1 shows a schematic view of a standard polycrys-
talline silicon deposition apparatus;
Fig. 2 shows the temperature profile used in the
apparatus shown in Fig. l; ~ ~
Fig. 3 shows a deposition curve normally associated ~ -
with the apparatus shown in Fig. l;
D~
,
1~)47850
Fig. 4 shows the top view of a typical two-inch wafer
shaded to show its non-uniform coating of polycrystalline
silicon in a system as shown in Fig. 1;
Fig. 5 shows a thickness profile taken along the line
5-5 of the wafer shown in Fig. 4 which is perpendicular to
the path of the gas flow;
Fig. 6 shows a plurality of thickness profiles taken
along the line 6-6 of the wafer shown in Fig. 4 which pro-
files are along the path of the gas flow at different posi-
tions in the tube;
Fig. 7 shows the schematic view of the apparatus of
the present invention;
Fig. 8 shows the temperature profile of the apparatus
shown in Fig. 7;
Fig. 9 shows the deposition profile at a selected
temperature;
Fig. 10 identifies a plurality of locations on a wafer -~
which is coated with different thicknesses of polycrystal-
line silicon when the wafers are closely spaced on the boat;
Fig. 11 shows the cross sectional profile of the poly-
crystalline film along the lines R-R, S-S, and T-T identi-
fied in Fig. 10, and the films are formed in the apparatus
shown in Fig. 7, wherein the wafers were spaced approxi-
mately 3000 mils apart;
Fig. 12 identifies a plurality of locations on a wafer
which is coated with a substantially uniform thickness of
polycrystalline silicon material over substantially all of
the area of the wafer, when the wafer is spaced from the
next adjacent wafer approximately 50 mils on centers using
20 mil thick wafers;
.
.
1047850
Fig. 13 shows the cross-sectional profile of the thick-
ness of polycrystalline film above the lines X-X and Y-Y
shown in Fig. 12 of a wafer placed in a furnace of Fig. 7
wherein the spacing between adjacent wafers is on 50 mil
centers and the wafers are 20 mils thick; --
Fig. 14 shows the maximum variations of deposited poly-
crystalline silicon over the wafer surface as a function of
the spacing of the wafers.
BRIEF DESCRIPTION OF THE INVENTION
In the prior art system of depositing polycrystalline
silicon, as shown with reference to Fig~ 1, there is shown
a source 1 of nitrogen gas which is the carrier gas for the
system and a source 3 of silicon tetrahydride which is a
source of silicon. The furnace tube 5 can be heated by
resistance heater coils 7 adjusted to give a temperature
profile as shown in Fig. 2. This temperature profile has
been chosen in combination with the deposition profile as
shown in Fig. 3 such that the highest degree of uniformity
of polycrystalline deposition is achieved on the wafers 9
which are placed within the fallout range of the tube as
indicated by the line 11 shown in Fig. 1. The fallout range
is that area of the tube 5 at which the polycrystalline
silicon deposits out of the gas flow through the tube. The
temperature within the furnace is such as to decompose the
silicon tetrachloride causing the silicon to rainout from
the gas stream onto the wafers positioned below. The top
view of the wafer having polycrystalline silicon deposited
thereon shown in Fig. 4 while a cross section through the
wafer taken on the lines 5-5 perpendicular to the gas flow
is shown in Fig. 5 and shows the variation in thickness
lQ47850
across a single representative wafer. The coverage of the
wafers is greatest at the center of the wafer and tapers off
t~ a thinnest portion on the edge of the wafers. Fig. 6
shows the variation in thickness of polycrystalline silicon
depending on the location of the wafer within the fallout
zone 11.
Referring again to Fig. 2, this view shows the tempera-
ture profile of the prior art deposition tube. The tempera-
ture is established as a ramp beginning at 625C at the
source end of the fallout zone identified as A. The central
portion B of the fallout zone is set at 650C, while the
exhaust end C of the fallout zone is held at 675C.
Referring again to Fig. 3, this view shows the deposi-
tion thickness profile as a measure of the position of the
wafer surface in the fallout zone. This figure shows a
variation of first a plus 3000 Angstroms and then a minus
3000 Angstroms along the fallout zone, The source end of
the fallout zone A shows a thickness of 3000 Angstroms,
while the central point B shows a thickness of 6000 Ang-
stroms and the exhaust end shows a thickness of 3000 Ang-
stroms. While it is possible to build devices with the
thickness of polycrystalline silicon over the entire range
as shown in Fig. 3, it is impractical from a commercial
viewpoint to identify and sort the individual die according
to thickness. It is not uncommon to have several hundred
die to a wafer. The actual identification and sorting of
these die is too costly. Again the profile is a typical
profile for a fixed run of 30 minutes, Larger or shorter
runs would give different numbers. Also other factors such
as flow rates and temperatures would give different numbers
from run to run if minute differences in such run parameters
occurred.
- 8 - ;
'
1~47850
Fig. 4 shows the top view of a typical two-inch wafer.
Larger or smaller size wafers would have similar shaped pro-
files in all the views, both in the prior art and in the new ,
system. The source gas is flowing from left to right in
Fig. 4.
Fig. 5 shows the variation of the polycrystalline depo-
sition across a single wafer along a line perpendicular to
the source gas flow. This figure shows that the target
thickness is identified as ~ Angstroms. This target thick-
ness is exceeded,by a figure of approximately 500 Angstroms
in the center Q of the wafer, and the actual thickness falls
short by about 1000 Angstroms at both edges P and R of the
wafer.
Fig. 6 shows the variation of polycrystalline silicon
thickness across the wafer taken along the direction of gas ,~
flow at the center N of a wafer and at both the first edge M
and txailing edge O depending upon the placement of the wafer
in the fallout zone 11. Curve C shows a generally decreas- '
ing thickness for a wafer placed at the exhaust end. Curve B
shows a concave variation'for the center of the fallout zone.
Curve A shows a generally increasing thickness for wafers
placed at the source end of the tube.
In operation, a medium thickness of 4500 Angstroms is
selected such that the thickest portion of the wafer along
the line 6-6 of Fig. 4 is approximately 6000 Angstroms thick,
Fig. 3, while the thinnest part of the wafer along the line
5-5 is at 3000 Angstroms thick, Fig. 3. This variation in
thickness guarantees that certain regions of the wafer have
an optimum thickness of the polycrystalline material at 4500
Angstroms. With the optimum thickness, certain usable devices
can be made on the wafer. However, it has been found that
1~47850
3000 Angstroms can be too thin and 6000 Angstroms can be too
thick for usable device performance.
Accordingly, it is desirable to form a polycrystalline
layer with a more optimum thickness over a greater portion
of the wafer. In the system shown in Fig. 1 only 12 to 20
wafers can be passed through the system at one time because
of the size of the tube and the size of the fallout zone. ~-
Since the wafers must lie on a major surface with the poly-
crystalline material raining out from the gas stream upon
the upper or opposite major surface of a substrate, the
physical limitation of the system is 12 to 20 wafers.
Another drawback on the system shown with reference to
Fig. 1 is the fact that the wafers are put in through the -
exhaust end 13 of the tube. When the wafers are being put
in, as well as when the wafers are taken out, some of the
powdered silicon material flecks off of the walls of the
tube at the end 13 and become deposited on the wafers. This
means that any polycrystalline material grown over that
powdered piece of silicon would be unsuitable for the forma-
tion of semiconductor devices. Also, any flecks of powdered -
silicon fallingon a newly grown polycrystalline layer
adversely affects the use of that area for an active device. ~-
Referring to Fig. 7 there is shown a schematic view of -~
the present system wherein the preferred source 20 of semi-
conductor material is silane in a gaseous form. Other ~
sources can be used as SiCl2H2, SiClH3 or SiC14. A flow- ~ -
meter 22 i8 provided for metering the correct amount of
silane gas flow into the tube and over the wafers. A first
source 24 of nitrogen is provided along with a nitrogen
flowmeter 26. This flow is normally used at a low flow
level to backflush any residual silane remaining within the
- 10 -
, , : ............................. ~ . : .
:, :' . - : : ~ .
:1~)478SO :
plumbing lines outside of the furnace since silane is explo- -
sive when above a certain temperature and exposed to air.
A second source 28 of nitrogen is provided along with a
flowmeter 30 for measuring the flow of nitrogen from the
source 30 into a tube 32. This source of nitrogen is used
for rapidly bringing the evacuated tube 32 up to atmospheric
pressure as well as aiding in the initial heating of the
wafers. While nitrogen is shown, any inert gas normally
used in the processing of semiconductor wafers can be used;
i.e., argon, etc. Best results are achieved when the
source gas 20 is used alone during the deposition of the ~
polycrystalline material. All gases flow in the direction ~-
of the arrow 34. An end cap 36 is in engagement with the
tube to provide a vacuum seal with the tube. The N2 and
SiH4 flows enter the tube 32 at the point where the line 38
passes through an appropriate fitting in the end cap 36. A
pressure sensor and vacuum gauge 40 is also attached to the ~ --
input line 38 for reading the pressure and vacuum at this
point. The furnace tube 32 is profiled to exhibit a flat ~
temperature profile as shown in Fig. 8 while the deposition - -
profile is shown with reference to Fig. 9. This means that
the flattened curve shown in Fig. 9 represents the variation
in the thickness of polycrystalline silicon material
deposited on a wafer when positioned at any location within
the entire heated zone of the furnace. The usable range of
the furnace proyides a thickness variation of only 500
Angstroms from the front to the back of the furnace. Refer- ~ -
ring to Fig. 13 briefly, this figure shows that for any one
wafer the thickness is substantially constant over the
entire wafer surface when the wafers are stood on edge
perpendicular to the gas flow. The embodiment shown with
, . . . . .
- , . :
.- , , ,.,,: . : '
. .
1~47850
reference to Fig. 7 provides this improved thickness control.
The profiling temperature for the furnace shown in
Fig. 7 can lie within a temperature range anywhere from
600-700 for giving practical results. At temperatures
lower than 600, the rate of deposition slows to the point
where the run takes too long. However, in those instances
where a slow deposition rate can be tolerated, temperatures
can be lowered to the minimum temperature at which the
silence decomposes. At the upper end of the temperature
spectrum; i.e., above 700C, crystalline imperfections are
formed on the surface of the wafers. Such imperfections or
outgrowths are formed in a deposition atmosphere in the
absence of hydrogen. Fig. 8 shows the preferred temperature
profile of the furnace 32 wherein the temperature of 600C
is established at the source end A, the center B and the
exhaust end C of the deposition zone indicated in Fig. 7 by
a line 41.
Fig. 9 shows the deposition profile of the system shown
in Fig. 7 when the tube is heated to 600C and the deposi-
tion run lasts for thirty minutes. The variation from the
source end A to the exhaust end B of the deposition zone is
500 Angstroms as indicated by a line 42. The deposition
profile within the preferred deposition zone of the tube
plu9 a leading and trailing edge is shown by the curve 43.
It has been found that the best results are achieved when
the maximum deposition thickness is set at the target thick-
ness and the variations occur on the downward side as shown
in Fig. 9. Similar deposition curves are achieved using a
target thickness other than 4500 Angstroms.
In some early experiments a cold trap cooled by liquid
nitrogen was used to remove the silane before it was vented
into the vacuum pump 44 shown in Fig. 7. This was to pre-
.
- 1~47850 -
vent damage to the vacuum apparatus. However, the cold trap
was allowed to warm after the deposition run was completed
it was damaged by the spontaneous burning of the silane as
it warmed and became exposed to air.
Accordingly, optical baffles 45 are attached at the
exhaust end 46 of the quartz tube 32 to trap out the
powdered silicon at this point. Wafers 47 are placed into
a quartz boat 49 and the loaded boat is loaded into the tube
through the source ~nd 51 of the quartz tube 32. In this
way contact with the deposited powdered silicon material at
the exhaust end 45 of the quartz tube is avoided. The
silicon boat carrying the wafers is placed within the pre-
ferred portion of the deposition curve, as discussed with
reference to Fig. 9. The wafers are placed on end and are
placed with their broad surface perpendicular to the gas
.... .... . .
flow.
In earlîer experiments, wafers were placed at greater
distances and wafers were manufactured having the top view
as shown in Fig. 10. The upper portion of the wafer indi-
cated by the line S-S has a uniform amount of material
deposited thereon, as shown by a comparison curve S-S' in -
Fig. 11, but the lower portion indicated by the line R-R was
substantially non-uniform and unusable as shown by the
comparison curve R-R' in Fig. 11. Additionally, a thickness
variation also occurred from the top to the bottom of the
wafer as indicated by the line T-T in Figs, 10 and 11. The
spacing which gave the results illustrated in Fig. 11 was
3000 mils~ It should be kept in mind that the length of the
spacing between wafers is that distances between midpoints - - -
of the thickness of adjacent wafers. When two wafers are
20 mils thick and the spacing is given as 50 mils, there is
actually a 30 mil open area between the rear surface of the
- 13 -
.
~V47850
first wafer and the front surface of the next wafer.
Accordingly, if thicker wafers are used, the spacings would
change also. This change would only be significant at the
upper and lower limits as in between it does not matter. It
is recommended that the actual spacing of 30 mils from
surface to surface should not be made smaller. At the upper
limit, minimum acceptable depositions on the top half of
wafers were achieved at actual back surface to front surface
spacing of 2980 mils.
Accordingly, many experiments were run to ascertain the
optimum spacing of the wafers side-by-side. This informa-
tion is shown in Fig. 14 by the line 53. This curve shows
the maximum variation across the wafer as a function of
wafer spacing. A preferable open space distance of 30 mils
between adjacent surfaces of adjacent wafers has been
selected as the preferable distance. A polycrystalline
silicon layer is formed on a wafer shown in Fig. 12 having
a deposition profile as shown in Fig. 13. A line 55 shows
the thickness variation in both the X-X~ and Y-Y~ directions
as shown in Fig. 12. This shows an essentially uniform
thickness of polycrystalline silicon material deposited
across the major portion of the wafer. It is only at the
edge points 61a and 61b of the curve 55 shown in Fig. 13
that a slight increase in thickness is found, It should be
emphasized that the thickness of the polycrystalline - -
material across the major surface between the lines is -
essentially uniform while the difference in thickness from ~ -
wafer to wafer from the source end of the deposition zone
to the exhaust end of the deposition zone differs by a total - -
of 500 Angstroms as shown with reference to Fig. 9.
In the impro~ed system as shown with reference to
Fig. 7, approximately 250 wafers can be placed on a 12-inch
- 14 -
.. . . .
1~47BSO
boat. This is a throughput greater than 10 to 1 as compared
with the prior art method of forming polycrystalline silicon
material.
The operation of the system shown in Fig. 7 has the
following special steps. The vacuum identified as the
preferred vacuum level lies within the range of 600 to 1600
millitorr. Nitrogen from source 24 is always used to purge
: ,, .
any residual SiH4 left in the system once the SiH4 is turned
off. Nitrogen from the source 28 is used to b~eak the
vacuum and establish atmospheric pressure within the tube 32.
Although specific embodiments of the invention have
been described herein, it is not intended to limit the
invention solely thereto but to include all of the varia-
tions and modifications which suggest themselves to one
skilled in the art within the spirit and scope of the
appended claims,
" - :"' '
:
: .
,.'
:' '
: .
- 15 - ~
~ ~`
- . , ,.,,, : " ~; , . ' . .~. ' .: