Note: Descriptions are shown in the official language in which they were submitted.
1313571
METAL OXIDE SEMICONDUCTOR FIELD-EFFECT
TRANSISTOR FORMED IN SILICON CARBIDE
Field of the Invention
The present invention relates to metal
oxide semiconductor field effect transistors
(MOSFETs), and in particular to such transistors
formed in silicon carbide.
Background of the Invention
The growth in the use of semiconductor
devices for electrical applications has resulted in
a number of different devices which have particular
application in the creation of circuits and
electrical components. One type of device is known
as a metal-oxide-semiconductor field-effect
transistor (MOSFET) which is named after its three
main components. In broader terms, such a device
can be referred to as a metal-insulator-
semiconductor field-effect transistor (MISFET), but
as the most common applications use an oxide as the
insulating layer, the oxide designation will be used
primarily throughout this applicat;on. It will be
understood, however, that other insulating materials
can be appropriately used and referred to.
A field effect transistor differs somewhat
from a junction transistor. Junction transistors,
which historically were the first developed, are
formed when two p-n junctions are placed in close
proximity with one another and share a small portion
of the semiconductor material known as the base. A
junction transistor controls the flow of current
131357~
--2--
from a portion of semiconductor material adjacent
one of the junctions (the collector) through the
base and then to and out from the semiconductor
portion adjacent the other junction (the emitter) by
controlling the applied voltage on the base.
~ field effect transistor works on a
somewhat different principle. Typically, current
enters a field effect transistor through a region of
semiconductor material known as the source, and
exits the semiconductor material from another region
of semiconductor material known as the drain. The
source and drain are separated from each other by
yet another region of semiconductor material which
is known as the gate. ~hen an appropriate voltage
of either positive or negative bias (depending upon
the type of transistor) is applied to the active
region through the gate, current can be controlled.
In particular, if the semiconductor material in the
gate is an n-type material through which current
would normally flow, applying a negative bias to the
gate depletes electrons from the active region,
making the conducting channel smaller and thereby
hindering the flow of electrons from the source to
the drain. Such a device is referred to as a
depletion mode MOSFET. ~lternatively, where the
semiconductor material in the gate is a p-type
material which is normally nonconductive, applying a
positive bias voltage to the gate depletes holes
from the region, making it more conductive from the
resulting excess of electron carriers.
In order to passivate the surface of the
source and the drain and isolate the gate contact
from the gate semiconductor portion, an insulator
material is positioned between these respective
portions. Because silicon is presently the
semiconductor material most commonly used in
MOSFETs, the insulating portion most commonly is
~3~3571
formed from silicon dioxide. The gate contact,
which can be a metal or other conductive material,
the insulator (usually silicon dioxide), the
semiconductor material and the device's method of
operation give the MOSFET its name.
The MOSFET has gained wide acceptance as
an appropriate device since its introduction. As is
the with all other semiconductor devices, however,
some of the characteristics of a MOSFET will be
limited by the characteristics of the semiconductor
material from which it is formed. Because silicon
has some inherent limitations for certain
applications, corresponding MOSFETs formed from
silicon will also have inherent limitations.
Accordingly, it has long been recognized
that one method of improving the performance of
devices is to attempt to form them on materials
having superior characteristics. One such material
having a number of superior characteristics is
silicon carbide (SiC). Silicon carbide has some
excellent semiconductor properties: a wide bandgap,
a high thermal conductivity, a high melting point, a
high breakdown electric field strength and a high
saturated electron drift velocity. The wide bandgap
gives silicon carbide advantages over semiconductor
materials with narrower bandgaps. Additionally, its
high thermal conductivity and better temperature
stability mean that devices made from silicon
carbide can be packed more closely together without
risk of destroying each other from dissipated heat
energy, and devices made from silicon carbide can
operate at significantly higher temperatures than
can those devices made from narrower bandgap
semiconductors.
Accordingly, a number of attempt~ have
been made to form devices, and specifically M~SFETs,
on silicon carbide. Silicon carbide is, however, a
` 1313571
--4--
difficult material to work with. To date,
successful production of crystalline silicon carbide
of an appropriate chemical purity and low defect
level has remained a somewhat elusive goal, and the
growth of single crystal thin films and large
crystals have heretofore been difficult to
accomplish. Additionally, successful introduction
and activation of the necessary dopant ions into
silicon carbide for device manufacture has likewise
proved difficult.
These problems have recently been
successfully addressed as described in several
issued patents and applications assigned to the assignee
of the present ivnention. These include "Growth of
Beta-Sic Thin Films and Semiconductor Devices Fabricated
Thereon", U.S. patent No. 4,912,063 (see Canadian
Application Serial No. 581,147 filed October 25, 1988);
"Homoepitaxial Growth of Alpha-Sic Thin Films and
Semiconductor Devices Fabricated Thereon", U.S. patent
No. 4,912,064 tsee Canadian Application Serial No.
581,144 filed October 25, 1988) and "Implantation and
Electrical Activation of Dopants into Monocrystalline
Silicon Carbide", (see Canadian application Serial
No. 581,148 filed October 25, 1988). The advances
in forming silicon carbide thin films, silicon
carbide single crystals, and in successfully doping
silicon carbide according to these methods have
rekindled interest in producing commercial-quality
devices from silicon carbide, including transistors.
A number of attempts have been made to
produce various junctions, diodes, rectifiers and
other contacts on silicon carbide. More
specifically, however, in the patent literature, U.S.
Wallace No. 3,254,280 discusses a method of forming
a junction transistor in silicon carbide. According
to Wallace, the appropriate necessary single
crystals of silicon carbide can be "grown in
1 31 3571
--5--
accordance with any suitable procedure known to
those skilled in the art," and doping can take place
using "any . . . method known in the art." Although
the production of doped single crystal silicon
carbide is rather easily dismissed in Wallace's
discussion, in practice forming doped single
crystals of appropriate purity and defect level is
quite difficult and commercial devices based on
Wallace's teaching have not been observed.
Hall U.S. patent No. 2,918,396 also discusses a
method of forming a junction type transistor in silicon
carbide in which the primary technique suggested by
the disclosure is that of placing an alloy formed of
silicon along with an nactivator~ (dopant) on the
surface of a single crystal of silicon carbide,
raising the silicon carbide to a temperature below
its melting point but sufficient to cause the alloy
to melt and dissolve a surface portion of the
silicon carbide. When the materials are cooled, a
p-n junction hopefully results. According to Hall,
appropriate crystals can be prepared using the Lely
technique. As is known to those familiar with
silicon carbide technology, however, the Lely
process represents an unseeded sublimation technique
which has generally failed to overcome the inherent
difficulties in producing device quality single
crystals of silicon carbide.
Other researchers have made specific
attempts to produce workable MOSFETs on silicon
carbide. For example, in Inversion-TyPe MOS Field
Effect Transistors Usina CVD Grown Cubic SiC on Si,
Jap. J. Appl. Phys. 23, L862 (1984), Shibahara et
al. discuss their attempts to produce an inversion
type, n-channel MOSFET on cubic silicon carbide
grown on the (100) face of silicon by chemical vapor
deposition (CVD). Shibahara's work is also
discussed in Novel Refractorv Semiconductors,
1 3 1 3~7 1
Materials Research Society symposium ~roceedings,
edited by T. Aselage, ~. Emin, and c. Wood
(Materials Research Society, Pittsburgh, PA, 1987),
Vol. 97, p. 247.
In spite of these described methods, the
devices fabricated by Shibahara et al. have never
been demonstrated to have successfully operated
above room temperature. As discussed earlier,
operation of devices at very high temperatures is
one of the particular reasons for seeking to form
devices on silicon carbide. Devices formed on
silicon carbide which~cannot operate at temperatures
different from those upon which devices formed on
silicon can operate offer no particular advantage.
Kondo et al. also describe an experimental
MOSFET produced on beta silicon carbide in
ExPerimental 3C-SiC MOSFET, IEEE Electron. Device
Lett., EDL-7, 404 (1986). According to this
discussion, Kondo first grew beta silicon carbide
film epitaxially on a p-type silicon (100) substrate
using CVD. Xondo fabricated a depletion mode
MOSFET; nevertheless, the resulting device showed no
current saturation, no threshold cutoff, and no high
temperature capability. Therefore, the techniques
disclosed in Xondo's study must be deemed
unsuccessful.
Accordingly, this invention seeks to
produce a metal-oxide-semiconductor
field-effect transistor (MOSFET), fabricated from
silicon carbide.
Further the invention seeks to
provide both inversion mode and depletion mode
MOSFETs fabricated from silicon carbide.
Still further this invention seeks
to provide a MOSFET formed on silicon carbide which
can operate at temperatures as high as 650
centigrade.
1 31 3571
--7--
Summary of the Invention
The invention in one aspect pertains to the method of
forming a metal-oxide-semiconductor field-effect transistor
suitable for operation at temperatures of at least 650
centigrade and high radiation densities, the method comprising
oxidizing a single crystal silicon carbide substrate having a
first conductivity type to form a silicon dioxide surface layer,
selectively applying gate contact material to the silicon dioxide
surface layer, forming a doped source and a doped drain of a
desired conductivity type by directing an ion beam of dopant ions
onto the single crystal silicon carbide substrate in which the
silicon carbide substrate is maintained at a temperature of
between about 600K and about llOOK and applying source and drain
contacts.
The invention also pertains to a method of forming an
inversion mode metal-insulator-semiconductor field-effect
transistor suitable for operation at temperatures of at least
650 centigrade and high radiation densities, the method
comprising forming a doped source and a doped drain having a
first conductivity type in a doped portion of silicon carbide
having an opposite conductivity type by directing an ion beam of
dopant ions onto the silicon carbide substrate in which the
silicon carbide substrate is maintained at a temperature of
between about 600K and about llOOK.
Further the invention provides a method of forming a
depletion mode metal-insulator-semiconductor field effect
transistor suitable for operation at temperatures of at least
650 centigrade and high radiation densities, the method
comprising forming a doped source and a doped drain in a silicon
carbide semiconductor portion having the same conductivity type
as the doped source and doped drain by directing an ion beam of
dopant ions onto the silicon carbide substrate in which the
silicon carbide substrate is maintained at a temperature of
between about 600K and about llOOK.
Still further the invention provides a method of
forming a metal-oxide-semiconductor field-effect transistor
suitable for operation at temperatures of at least 650
1 31 3571
centigrade and high radiation densities which method comprises
forming an insulating surface layer upon a silicon carbide
subs-trate, selectively applying gate contact material to the
insulating surface layer, forming a doped source and a doped
drain of a desired conductivity type in the silicon carbide
substrate by directing an ion beam of dopant ions onto the
silicon carbide substrate in which the silicon carbide substrate
is maintained at a temperature of between about 600K and about
llOOK and applying source and drain contacts.
Further stlll the invention provides an inversion mode
metal-oxide-semiconductor field-effect transistor comprising a p-
type single crystal silicon carbide substrate, a source formed of
n-type silicon carbide, a drain formed of n-type silicon carbide,
a p-type silicon carbide gate and source and drain contacts
formed of tantalum silicide.
Further another aspect of the invention provides a
depletion mode metal-oxide-semiconductor field-effect transistor
comprising a single crystal p-type alpha silicon carbide
substrate, a p-type beta silicon carbide epitaxial layer upon the
p-type alpha silicon carbide substrate for electronically
bordering the depletion region, an active epitaxial layer of n-
type beta silicon carbide upon the p-type beta silicon carbide
epitaxial layer, a more heavily doped n-type source region in the
active epitaxial layer of n-type beta silicon carbide, a more
heavily doped n-type drain region in the active epitaxial layer
of n-type beta silicon carbide and a gate region in the active
epitaxial layer of n-type beta silicon carbide and defined by the
portion of the active epitaxial layer which is positioned between
the more heavily doped source and drain regions.
Other aspects and advantages of the invention and the
manner in which the same are accomplished will be set forth in
the accompanying detailed description which illustrates exemplary
and preferred embodiments, and in the following drawings in
which:
Description of the Drawings
Figures 1 - 5 illustrate several of the steps and the
resulting structure of an n-channel inversion mode metal-
1313571
insulator-semiconductor field-effect transistor formed according
to the present invention;
Figure 6 is a cross-sectional view of an n-channel
depletion mode metal-insulator-semiconductor field-effect
transistor according to the present invention;
Figure 7 is a plot of drain current versus drain
voltage at a temperature of 296K for an n-channel depletion mode
MOSFET according to the present invention;
Figure 8 is another plot of drain current versus drain
voltage at a temperature of 573K for the same MOSFET as Figure 7
according to the present invention; and
Figure 9 is yet another plot of drain current versus
drain voltage at a temperature of 923K for the same MOSFET as
Figures 7 and 8 according to the present invention.
Detailed Description of the Invention
Briefly the invention in one aspect pertains to a
method of forming a metal-oxide-semiconductor field-effect
transistor suitable for operation at temperatures of at least
650 centigrade, and at high radiation densities and at high
power levels. This method preferably comprises oxidizing a
single crystal silicon carbide substrate having a first
conductivity type to form a silicon dioxide surface layer, then
selectively applying gate contact material to the silicon dioxide
surface layer. A doped source and a doped drain of a desired
conductivity type are formed by high temperature implantation of
doping ions, following which source and drain contacts are
applied. The implantation achieved by directing an ion beam of
dopant ions onto the single crystal silicon carbide substrate in
which the silicon carbide substrate is maintained at a
temperature of between about 600K and about 1100K. Other aspects
of the invention as noted above will become evident herein.
As an example of the invention, depletion mode n-
channel metal-oxide-semiconductor field-effect transistors were
fabricated on n-type beta silicon carbide (111) thin films
epitaxially grown by chemical vapor deposition on the (0001) face
of 6H alpha silicon carbide single crystals. The gate oxide was
thermally grown on the silicon carbide, the source and drain were
1313571
-9A-
doped n by nitrogen ion implantation at 823K. Stable saturation
and subthreshold current was achieved at drain voltages (VDs)
exceeding 25 volts. Transconductances as high as 11.9 mS/mm were
achieved. Stable transistor action was observed at temperatures
as high as 923K, the highest temperature reported to date for a
transistor in any material.
Historically, research on electrical devices formed on
silicon carbide has been rather limited, mainly because of the
difficulty in obtaining high quality silicon carbide films.
Recently, however, and as described in the co-pending patent
applications to the same assignee described earlier, success in
growing both alpha and beta silicon carbide thin films on silicon
carbide substrates, as well as success in growing large single
crystals of silicon carbide, have provided a better foundation
for device research than has previously ever existed. The MOSFET
devices described herein were formed using some of these
successful new techniques.
Additionally, and as also referred to earlier, a novel
and successful method of adding dopant ions to silicon carbide
has also been recently developed. As therein, it has been
1313571
--10--
discovered that attempts to dope silicon carbide
with ion implantation techniques at both room
temperature (e.g. 2s8K) and low temperatures (e.g.
77K) have been unsuccessful, even following
annealing. When, as discussed in the above
application, the implantation is carried out at
relatively high temperatures (e.g. 623K, 823K,
1023K), however, initial damage to the lattice is
minimized and an annealing step at a more moderate
temperature (1200C) than is usually necessary
sufficiently activates the dopant ions.
Figures 1-5 show some of the steps used in
forming an n-channel inversion mode MOSFET according
to the present invention and its resulting
structure. Figure 5 is a cross-sectional view of
the finished device which has a concentric ring
structure which will be described further herein.
In Figure 5, the source is indicated at 10 and is
n-type. The drain is indicated at 11 and is also
n-type, both formed by high temperature implantation
in a p-type silicon carbide substrate 12. The gate
is indicated at 13. The insulating layer of silicon
dioxide is indicated at 14, the source contacts at
15, and the drain contacts at 16, with both source
and drain contacts being formed of tantalum silicide
(TaSi2), which is a novel use of this material. As
set forth earlier, the gate 13 has a gate contact 17
formed of polysilicon. In a preferred embodiment of
the invention, the concentric gate ring had a 20
micrometer (um) wide connecting strip which extended
to a 100 um X 100 um contact pad. The source
contact 15 is an outer concentric semicircle that
surrounds the gate ring 13 and the gate contact
except for the gate's connection strip. The source
ring also has a connecting strip to the 100
micrometer diameter contact pad 15.
1313571
Figure 1 illustrates some of the initial
steps in forming a MOSFET according to the present
invention. The silicon dioxide layer 14 is added
through normal oxidation procedures to the p-type
substrate of silicon carbide 12. A layer of
phosphorous doped polysilicon 17 is deposited over
the oxide layer 14. A photoresist material 20 is
applied and patterned as shown. After the exposed
polysilicon is etched away and the photoresist
removed, the substrate 12 and oxide layer 14 have
the appearance shown in Figure 2 in which the only
remaining polysilicon is that which will form the
gate contacts 17. Nitrogen is added by high
temperature ion implantation (823K) through the
oxide layer 14 into the p-type silicon carbide 12 to
form n-type wells for the source 10 and the drain
11 .
In Figure 3, additional photoresist 20 has
been added and patterned in order to etch windows in
the exposed oxide layer 14 above the source and
drain. In Figure 4, tantalum silicide has been
sputter deposited over the photoresist and over the
exposed silicon carbide surface exposed by the
openings in the oxide. When the photoresist 20 is
lifted off, the only tantalum silicide which remains
is that on the previously exposed silicon carbide
surfaces adjacent the source and drain (Figure 5).
Figure 6 shows a depletion mode MOSFET
formed according to the present invention. In
Figure 6, a p-type alpha silicon carbide substrate
is shown at 21 which carries a p-type beta silicon
carbide layer 22 and an n-type beta silicon carbide
layer 23. A more heavily n-doped source 24 and
drain 25 are also illustrated along with the gate
26, Figure 6 being a cross-sectional view as
discussed earlier. As in the inversion mode MOSFET
described earlier, source and drain contacts 27 and
1313571
-12~
30 are formed of tantalum silicide while the gate
contacts 31 are formed of polysilicon and are
superimposed on the oxide layer 32.
In particular em~odiments, the silicon
carbide films were first polished using 0.1 um
diamond paste, oxidized to remove polishing damage
and etched in hydrofluoric acid (HF) to remove the
oxide film. The gate oxide was subsequently grown,
preceded by a three-step cleaning process using hot
sulfuric acid (H2S04) for five minutes, a one-to-one
mixture of hot ammonium hydroxide (NH40H) and
hydrogen peroxide (H202) for five minutes, and HF for
one minute, followed by a rinse with deionized
water.
A 500 nm thick film of polysilicon was
deposited on the prepared oxide via low pressure
chemical vapor deposition at 893K which was
degenerately doped by phosphorous (P) diffusion at
1173K for five minutes, then patterned as shown in
the drawings to form the gate contacts.
The n-type doped source and drain areas
were then formed by high temperature ion
implantation of nitrogen through the oxide. The
implantation was carried out at 773K, 70 keV at a
dosage of 5.0 X 1014 cm~2.
Figure 7 shows the drain current versus
drain voltage characteristics measured at a
temperature of 296K for a MOSFET formed according to
the present invention in beta silicon carbide. The
particular device upon which the measurements were
made had a gate length of 7.2 um and a gate width of
390 um, with a source to drain contact distance of
24 um. As indicated in Figure 7, this device showed
very stable drain current saturation out to a drain
source voltage of 25 volts. This trend actually
continued to a source drain voltage of 30 volts at
which point the oxides began to break down.
1 3 1 357 1
-13-
Accordingly, this is the first time stable
saturation has been reported for drain source
voltages of greater than 5 volts for any field
effect transistor formed in beta silicon carbide.
The threshold voltage was a gate voltage
(VG) of -12.9 V, as determined from a plot of the
square root of the drain-source current versus VG~
The leakage current at a VDS of 25 volts in this
device was 3.75 microamps (uA) in the off-state (VG =
-15V). The transconductance of this device at room
temperature with V~s fixed at 20V was 5.32 mS/mm at
VG = 2.5V.
Figure 8 is another plot of drain current
versus drain voltage for the same device, but with
the device heated to 573K and allowed to stablize
for 15 minutes at that temperature. Despite the
increase in temperature, the drain current
saturation was still very stable up to 25 volts.
The leakage current at a drain source voltage of 25
volts and a gate voltage of -15 volts increased to
22 uA and the threshold voltage shifted negatively
to VG = - 13.3V.
Figure 9 is yet another plot of drain
current versus drain voltage for the same de~ice,
but measured at a temperature of 923K. The
transconductance decreased with this further
increase in temperature. The lower transconductance
of the device as measured in Figure 9 at 923K is
demonstrated by the lower current at a zero gate
voltage as compared with Figure 7 and Figure 8.
Although the transconductance at 923K became erratic
above a gate voltage of 1 volt, it reached a maximum
of about 4.8 mS/mm at a gate voltage of 8 volts and
a drain source voltage of 20 volts. This decrease
in transconductance at higher temperatures is due to
increasing lattice scattering at the higher
temperatures.
1313571
-14-
In Figure 9, the threshold voltage again
shifted negatively to a gate voltage of -14.8 volts
at 923K. The leakage current increased to 128 uA at
a gate voltage of -15 volts and a drain source
voltage of 25 vslts. At 973K, the device showed
similar current saturation but the gate oxide
experienced breakdown. Therefore, the current was
being injected at the gate and the device could not
be cut off.
In the description and drawings, there
have been set forth preferred and exemplary
embodiments of the invention which have been set
forth by way of example and not of limitation, the
scope of the invention being that set forth in the
following claims.