U.S. patent application number 14/918774 was filed with the patent office on 2017-04-27 for planar triple-implanted jfet.
The applicant listed for this patent is United Silicon Carbide, Inc.. Invention is credited to Anup Bhalla, Zhongda Li.
Application Number | 20170117418 14/918774 |
Document ID | / |
Family ID | 57223784 |
Filed Date | 2017-04-27 |
United States Patent
Application |
20170117418 |
Kind Code |
A1 |
Bhalla; Anup ; et
al. |
April 27, 2017 |
PLANAR TRIPLE-IMPLANTED JFET
Abstract
A JFET is formed with vertical and horizontal elements made from
a high band-gap semiconductor material such as silicon carbide via
triple implantation of a substrate comprising an upper drift region
and a lower drain region, the triple implantation forming a lower
gate, a horizontal channel, and an upper gate, in a portion of the
drift region. A source region may be formed through a portion of
the top gate, and the top and bottom gates are connected. A
vertical channel region is formed adjacent to the planar JFET
region and extending through the top gate, horizontal channel, and
bottom gate to connect to the drift, such that the lower gate
modulates the vertical channel as well as the horizontal channel,
and current from the sources flows first through the horizontal
channel and then through the vertical channel into the drift.
Inventors: |
Bhalla; Anup; (Princeton
Junction, NJ) ; Li; Zhongda; (Somerset, NJ) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
United Silicon Carbide, Inc. |
Monmouth Junction |
NJ |
US |
|
|
Family ID: |
57223784 |
Appl. No.: |
14/918774 |
Filed: |
October 21, 2015 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H01L 29/66909 20130101;
H01L 21/0465 20130101; H01L 29/1066 20130101; H01L 21/26513
20130101; H01L 29/1095 20130101; H01L 29/0619 20130101; H01L 21/266
20130101; H01L 29/1058 20130101; H01L 29/0692 20130101; H01L
29/8083 20130101; H01L 29/1608 20130101; H01L 21/26586 20130101;
H01L 29/0843 20130101; H01L 29/66068 20130101; H01L 29/42316
20130101 |
International
Class: |
H01L 29/808 20060101
H01L029/808; H01L 21/265 20060101 H01L021/265; H01L 29/66 20060101
H01L029/66; H01L 21/266 20060101 H01L021/266; H01L 29/16 20060101
H01L029/16; H01L 29/10 20060101 H01L029/10 |
Claims
1. A method for creating a JFET, comprising: a. applying a first
mask to a wafer to form a first patterned hard masking layer to the
top of the wafer, the wafer comprising an upper drift region and a
bottom drain region; b. applying three implants to the unmasked
portion of the top of the wafer to form a lower gate region, a
horizontal channel region, and a top gate region in an upper
portion of the drift region; c. forming a source region in a
portion of the top gate region; d. connecting the top and bottom
gate regions.
2. The method of claim 1, further comprising: a. forming a vertical
channel region via implantation, the vertical channel region being
at a distance from the source region, the vertical channel region
being adjacent to, and extending through, the top gate region, the
horizontal channel region, and the bottom gate region, to connect
to the drift region.
3. The method of claim 2, further comprising: a. forming the
vertical channel region via angled implantation with the first
patterned hard masking layer in place.
4. The method of claim 2, further comprising: a. forming a lightly
doped region surrounding the source region via implantation, where
the lightly doped region is doped at ten percent or less of the
level of doping of the source region.
5. The method of claim 2, further comprising: a. forming a vertical
channel region between two regions each comprising an upper and
lower gate region.
6. The method of claim 2, further comprising: b. forming two
vertical channel regions, one at each end of a region comprising an
upper gate region and a lower gate region.
7. A JFET, comprising: a. a silicon carbide drift region atop a
backside drain; b. in a portion of the top of the drift region, a
top gate region having a shape in plan view with outer dimensions;
c. beneath the top gate region, a horizontal channel region having
the shape and outer dimensions in plan view, the horizontal channel
region being vertically aligned with the top gate region; d.
beneath the horizontal channel region, a bottom gate region having
the shape and outer dimensions in plan view, the bottom gate region
being vertically aligned with the topside gate region; e. a
connection from the top gate region to the bottom gate region; f.
within the outer dimensions in plan view of the top gate region, a
source region; and g. abutting the top gate region, horizontal
channel region, and bottom gate region to one side, a vertical
channel region, the vertical channel region extending vertically
below the bottom gate region to abut the drift region; h. where a
first, second, and third doping concentration of the drain, the
vertical channel region, and the horizontal channel region
respectively are selected such that the threshold of the JFET is
set by the horizontal channel region, i. such that, when activated
by the top gate region and bottom gate region, current flows from
source region into the horizontal channel region and then into the
vertical channel region, drift region, and the backside drain.
8. (canceled)
9. A JFET, comprising: a. a drift region atop a drain; b. in a
portion of the top of the drift region, a top gate region defined
by a patterned mask; c. beneath the top gate region, a horizontal
channel region defined by the patterned mask; d. beneath the
horizontal channel region, a bottom gate region defined by the
patterned mask; e. a connection from the top gate region to the
bottom gate region; f. in a portion of the top gate region, a
source region, such that current flows from source region in to the
horizontal channel region and then into the drift region; and g. a
ring of material around the source region, between the source
region and the top gate region, where the ring of material has a
doping concentration that is--at ten percent or less of the doping
concentration--of the source region.
10. (canceled)
11. (canceled)
12. The JFET of claim 7, wherein the top gate region abuts the
source region.
13. The JFET of claim 7, wherein the first connection from the
first top gate region to the first bottom gate region is a bridge
of implanted semiconductor material.
14. (canceled)
15. (canceled)
16. The JFET of claim 9, wherein the first connection from the
first top gate region to the first bottom gate region is a bridge
of implanted semiconductor material.
17. The JFET of claim 9, further comprising: a first vertical
channel region at a distance from the source region, the first
vertical channel region being adjacent to, and extending through,
the top gate region, the horizontal channel region, and the bottom
gate region, to connect to the drift region.
18. The JFET of claim 17, further comprising a. a second vertical
channel at a distance from the source region, the second vertical
channel region being adjacent to, and extending through, the top
gate region, the horizontal channel region, and the bottom gate
region, to connect to the drift region, b. wherein the first
vertical channel is located at a first end of the horizontal
channel and the second vertical channel is located at a second end
of the horizontal channel.
19. The JFET of claim 7, further comprising: a ring of material
around the source region, between the source region and the top
gate region, where the ring of material has a doping concentration
that is at ten percent or less of the doping concentration of the
source region.
Description
FIELD OF THE DISCLOSURE
[0001] Vertical junction field-effect transistors made from wide
bandgap materials such as silicon carbide are useful in power
electronic circuits, such as power factor correction (PFC)
circuits, DC-DC converters, DC-AC inverters, and motor drives.
SUMMARY
[0002] Junction field-effect transistors (JFETs) and methods of
constructing JFETs are described herein. A JFET having vertical and
horizontal elements may be made from a semiconductor material such
as silicon carbide (SiC) by a process using a triple implantation
to form a horizontal planar JFET region comprising a lower gate, a
horizontal channel, and an upper gate, all above a drift region
resting on a drain substrate region. A source region may be formed
through a portion of the top gate, and the top and bottom gates are
connected. A vertical channel region may be formed adjacent to the
planar JFET region and extending through the top gate, horizontal
channel, and bottom gate to connect to the drift, such that the
lower gate modulates the vertical channel as well as the horizontal
channel, and current from the sources flows first through the
horizontal channel and then through the vertical channel into the
drift region.
[0003] This Summary is provided to introduce a selection of
concepts in a simplified form that are further described below in
the Detailed Description. This Summary is not intended to identify
key features or essential features of the claimed subject matter,
nor is it intended to be used to limit the scope of the claimed
subject matter. Furthermore, the claimed subject matter is not
limited to limitations that solve any or all disadvantages noted in
any part of this disclosure.
BRIEF DESCRIPTION OF THE DRAWINGS
[0004] A more detailed understanding may be had from the following
description, given by way of example in conjunction with the
accompanying figures. The figures are not necessarily drawn to
scale.
[0005] FIG. 1 provides, for reference, a cross section of an active
cell of a prior art vertical JFET.
[0006] FIG. 2 is a perspective view of a section of an active cell
of a first example triple-implanted JFET.
[0007] FIG. 3 is a doping profile graph of a triple-implanted JFET
such as the first example triple-implanted JFET.
[0008] FIG. 4 is a vertical cross section of the first example
triple-implanted JFET.
[0009] FIG. 5 is a perspective view drawing of an active cell of a
second example triple-implanted JFET.
[0010] FIG. 6 is a vertical cross section of the second example
triple-implanted JFET.
[0011] FIG. 7 is a vertical cross section showing the
interconnection of gate structures in an example triple-implanted
JFET.
[0012] FIGS. 8 and 9 are vertical cross sections of termination
regions of example triple-implanted JFETs showing how the same set
of implants may be used to form the gate pickup areas as well as
the termination guard rings.
[0013] FIGS. 10-18 illustrate an exemplary process for fabricating
triple-implant JFETs.
[0014] FIG. 10 is a vertical cross section of a starting substrate
comprising a lower drain region and upper drift region.
[0015] FIG. 11 is a vertical cross section of the substrate with
triple implant, shown with an implant mask in place.
[0016] FIG. 12 is a vertical cross section showing the addition of
vertical channel implant, shown with the vertical channel implant
mask in place.
[0017] FIG. 13 is a vertical cross section showing the addition of
a source implant, shown with the source implant mask in place.
[0018] FIG. 14 is a perspective view of the active cell of an
example triple-implanted JFET in process, showing the surface
contact region for the lower gate connection.
[0019] FIGS. 15 and 16 are further perspective views of an example
cell in process.
[0020] FIGS. 17 and 18 are vertical cross sections of an example
cell in process
[0021] FIG. 19 is a top view showing an exemplary layout of the
metal layers of a triple-implanted JFET, including both the active
and termination regions.
[0022] FIG. 20 is a graph of exemplary drain conduction versus
drain-source voltage triple-implanted JFET devices versus prior art
devices.
[0023] FIG. 21 is a graph of short-circuit transient performance of
an exemplary triple-implanted JFET device versus the performance an
exemplary prior art device.
[0024] FIG. 22 is a vertical cross section of a third example
triple-implanted JFET with a vertical channel region formed by
angled implant using the same mask as used for the triple
implant.
DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS
[0025] Junction field-effect transistors and methods of
constructing JFETs are described herein. A JFET having vertical and
horizontal planar elements may be made from a high band-gap
semiconductor material such as silicon carbide (SiC) by a process
using a triple implantation on a substrate comprising an upper
drift region and a lower drain region. The triple implantation is
used to form a horizontal planar JFET region comprising a lower
gate, a horizontal channel, and an upper gate, all above a drift
region resting on a drain substrate region. A source region may be
formed through a portion of the top gate, and the top and bottom
gates are connected. A vertical channel region is formed adjacent
to the planar JFET region and extending through the top gate,
horizontal channel, and bottom gate to connect to the drift, such
that the lower gate modulates the vertical channel as well as the
horizontal channel, and current from the sources flows first
through the horizontal channel and then through the vertical
channel into the drift.
[0026] When making transistors in a wide-bandgap semiconductor
material, such as SiC, GaN, AlN, or diamond, JFET-type structures
are often preferred over MOS-type structures. This is due to the
difficulty of obtaining the same quality of oxide interface as seen
with Si/SiO2. This in turn makes it difficult to achieve
transistors with the same level of reliability in MOS surface
conduction channels, which are prone to degradation. JFETs avoid
this issue because JFETs are bulk devices, unlike MOS devices, and
do not require surface conduction channels. This is particularly
advantageous for high power devices, since surface conduction
channel reliability issues are exacerbated at higher operating
temperatures.
[0027] FIG. 1 shows a vertical cross section of the active cell of
a prior art vertical trench JFET. The channel 102 and the gate
regions 105 form a mesa, with the channel 102 at the center of the
mesa. Atop the mesa, the upper source region 103 is in contact with
the channel 102. The source contact 110 connects to the source 103.
The gate regions 105 are connected to the gate contacts 109. At the
bottom of the mesa, the channel 102 connects to the drift region
106. The drift region 106 serves as a voltage blocking layer. The
drift region may be grown epitaxially on the substrate drain region
107. The drain 107 is connected to the drain contact 108. The
channel 102 may be formed, for example, through epitaxial growth or
by implanting at an angle between the gate regions 105 through
either or both sides of the mesa.
[0028] Precise control of the threshold voltage (Vth) in such a
trench vertical JFET may be challenging since Vth varies with the
amount and the position of charge in the channel 103. These may be
strongly affected by process variations. For example, the mesa
width may fluctuate due to variations in the photo process, the
etch process, and the etch slope. This may lead to large variations
in Vth. These variations may be reduced to some extent by forming
the channel 102 by implantation. However, some variation in mesa
dimensions and sidewall angles will persist. For example, in order
to maintain good blocking capability in the off-state, especially
for normally-off devices and normally-on devices with very negative
Vth voltages, the vertical structure requires the use of very long
channel regions 102. This in turn adds the complexity of deep
trench etches, which are quite difficult to control in wide bandgap
materials.
[0029] FIG. 2 shows the active cell of a first example of a
triple-implanted planar JFET. At the center of the top surface is
the source region 103. At the bottom, the drain contact 108 is in
contact with the substrate drain region 107. The voltage-blocking
drift region 106 is on top of the drain 107. Again, the drift 106
may be formed as an epitaxial layer on top of the substrate drain
region 107. The triple-implanted JFET structure is formed by three
gate regions 105a, 105b, and 105c, along with an implanted
horizontal channel region 104. The horizontal channel 104 leads to
two implanted vertical channel regions 102a and 102b. There are
separate top gate regions 105a and 105c, and a common bottom gate
region 105b. Current from the source 103 flows through the
horizontal channel 104 to the vertical channels 102a and 102b. The
vertical channels 102a and 102b bring the current to the drift
region 106, from where it flows to the drain contact 108 via the
drain 107.
[0030] The gate regions 105a, 105b and 105c are of the opposite
doping type to the drain 107, drift 106, vertical channels 102a and
102b, horizontal channel 104, and source 103. Lightly doped regions
101a and 101b of either doping polarity may be used to increase
gate-source breakdown voltage and/or add a source ballast
resistance. For example, the lightly doped regions 101a and 101b
may be doped at ten percent or less of the doping level of the
adjacent source 103 or gates 105a and 105c. Such added source
ballast resistance may benefit device operation in short circuit
mode, and is particularly useful when the JFET is employed as a
current limiting device. The gate connection region 105d is the
same polarity as the gate regions 105a, 105b and 105c, but a
heavier implant is used to short the top and bottom gates and to
provide a sufficient surface doping for good ohmic contact
formation.
[0031] FIG. 3 shows an example doping profile for a
triple-implanted n-channel JFET. For a device constructed as shown
in FIG. 2, the profile of FIG. 3 may correspond to cross section
AA' of FIG. 2. Referring now to FIG. 3, at the right is the N+
doping of the substrate drain region 107. To the left of this is
the N- doping of the drift region 106.
[0032] At the left of FIG. 3 are the doping profiles of the three
regions that may be implanted into the drift region 106 using a
common hard mask. At the far left, corresponding to the top of the
device, is a top gate 105a. Below this is a channel region 104, and
then a bottom gate 105b. The top gate 105a junction depth is quite
shallow, and doped enough to not deplete fully when the device is
operated at its negative maximum gate-source voltage. The top gate
doping is light enough to be easily compensated by the source
contact implant which is of the opposite polarity type. The channel
implant 104 usually has a peak doping lower than the top gate 105a,
and is disposed deeper than top gate 105a. The bottom (or deep)
gate region 105b is formed by the highest energy implant to
position it below the channel implant. Again, the charge in this
region must be at least sufficient for this region to not be fully
depleted when the device is operating in drain to source breakdown
with maximum reverse gate-source voltage. On the other hand, this
doping level is preferably not made too high, in order to minimize
implant damage that could affect carrier mobility in the channel
region 104.
[0033] FIG. 4 shows vertical a cross section of the device of FIG.
2, taken along section BB' where the gate contact is made. The top
gate regions 105a and 105c are connected to the bottom gate region
105b by the gate connect region 105d. Region 105d is relatively
heavily doped at the surface for good ohmic contact, and more
moderately doped in the region between the two gates to compensate
the channel region dopant. As a result, along cross section BB'
region 105d splits the horizontal channel region into regions 104a
and 104b to the left and right of 105d. However, away from section
BB' these regions 104a and 104b are connected as shown in FIG. 2 as
region 104.
[0034] FIG. 5 shows the active cell of a second example of a
triple-implanted JFET. In this device, the current from the central
source region 203 flows to the left through the horizontal channel
region 204 between the upper gate regions 205a and 205b and lower
gate region 205c into vertical channel regions 202a. Current also
flows from the source 203 to the right between upper gate regions
205e and 205f and lower gate region 205c to vertical channel region
202b. From the vertical channel regions 202a and 202b, the current
flows to the drift region 206, then through the substrate 207 to
the drain contact 208. As in the first example JFET of FIG. 2, here
in FIG. 5 the transistor has both top gates regions 205a, 205b,
205e, and 205f, and a common bottom gate 205c. Here the gate ohmic
contact regions 205d and 205g are disposed along the channel, and
may be much more heavily doped, and shallower than the top gate
regions 205a, 205b, 205e, and 205f. Lightly doped regions 201a and
201b of either doping polarity may be used to add source ballast
resistance, and/or improve gate-source breakdown voltage. The
lightly doped regions 201a and 201b may be doped, for example, at
ten percent or less of the level of doping of the adjacent source
203, or of the gates 205b and 205e. The gate contact zones 209a and
209b connect to both the gate ohmic contact regions 205d and 205g
and the lower gate regions 205c.
[0035] FIG. 6 shows a cross section of the device of FIG. 5 along
section DD'. An additional mask and implant may be used to form the
deeper gate contact zones 209a and 209b that connect to the lower
gate region 205c. Note how here the horizontal channel region is
divided into regions 204a, 204b, and 204c along section BB' by the
contact zones 209a and 209b, whereas away from section BB' region
204 is contiguous.
[0036] FIG. 7 shows the active cell of the triple-implanted JFET
shown in FIG. 4, where the vertical channel region 302 is at the
center of the cell. This device in effect has two series connected
JFETs, one that is lateral, and one that is vertical. The lateral
JFET is split into two parts. On the left side, the current flows
from the source region 303a laterally via the horizontal channel
region 304a to the common vertical channel region 302, passing
between the gate regions 305a and 305c. Similarly, on the right
side, the current flows from source 303b laterally via the channel
region 304b to the common vertical channel region 302, passing
between the gate regions 305b and 305d.
[0037] The vertical JFET is formed by the combination of bottom
gate regions 305c and 305d and the vertical channel region 302. The
current from lateral channels 304a and 304b flows down the vertical
channel 302 between the gate regions 305d and 305d, and then into
the drift region 306, and from there into the substrate drain
region 307 and ultimately to the drain contact 308.
[0038] The lateral JFET sets the device threshold voltage, Vth. The
vertical JFET is useful in shielding the lateral channels 304a-b
from the effects of high drain bias once the region between 305c
and 305d has been depleted. In making devices where low
on-resistance and short channel lengths are required, this feature
allows the designer to make the lateral channels quite short
without suffering from drain bias induced barrier lowering and the
loss of blocking capability. This is particularly useful in making
normally-off JFETs. In the case of 4H-SiC normally-off JFETs, the
device Vth must ideally be kept close to 1V with +/-0.2V control.
This is because the gate drive voltage should be kept between
2.5-3V to limit the maximum gate current that results from forward
biasing the gate-source junction. Given that the overdrive voltage
available is just 2.5V-1V=1.5V, a short channel is needed to get a
high on-state current. In the trench structure, this is not
possible without degrading the blocking capability of the device,
so a high cell repeat density must be used. The shielding effect of
the lower gate may be employed to minimize channel length. By
disposing the peak of the channel implant close to the top gate
junction, a higher channel charge is also possible for the same
Vth, further allowing a high transconductance for normally-off
devices.
[0039] While connecting the bottom gate regions 305c and 304d to
the gate potential instead of to the source potential increases the
gate-drain capacitance, it serves the purpose of depleting both the
lateral channels 304a and 304b and the vertical channel 302
channels much more effectively. This in turn allows a greater
doping level in the lateral and vertical channel 302, leading to
lower specific on-resistance.
[0040] FIG. 8 shows how the gate bus and gate pad pickup areas may
be formed adjacent to the termination zone of triple-implanted JFET
devices, e.g., for the device shown in FIG. 2. In FIG. 8, a
passivation dielectric 110 lies above the termination zone. Implant
114 bridges the top gate 105a and bottom gate and 105c layers
through the intervening horizontal channel implant region section
104a. Implant 114 has a heavy surface doping for good ohmic
contact. A contact gap 115 may be used to connect to the gate bus
that distributes the gate signal around the device. The termination
may use floating guard rings 111, 112 and 113 by the same material
that is used to form the same gate bridging implant 114. Of course,
other known techniques of junction termination may be independently
applied to the device.
[0041] FIG. 9 is similar to FIG. 8. The difference in the example
of FIG. 9 is that the guard rings are formed only using the heavier
implants used to form the bridge between the top and bottom gates,
but the actual top and bottom gate implants are not used.
[0042] FIGS. 10 to 18 show an example of a process for the
manufacture of a triple-implanted JFET in a high-bandgap
semiconductor material such as 4H-SiC. For purposes of
illustration, FIGS. 10 through 18 show the construction of the
first example triple-implanted JFET as described in reference to
FIG. 2. However, it will be appreciated that this process, and
variations thereof, may also be used to create a variety of
triple-implanted devices including, but not limited to, the second
example JFET discussed in reference to FIG. 5. Similarly, the
device may be fabricated with variants of this flow in different
materials to achieve the similar structures.
[0043] FIG. 10 is a vertical cross section of the shows starting
material, which consists of the drift region 606 on top of the
substrate 607. Both are of a first doping type. The drift 606 may
be grown epitaxially on the substrate 607.
[0044] FIG. 11 shows the start of the process. A thick patterned
mask 608 capable of blocking all the subsequent implants is formed
on the surface. Three implants, 1101, 1102, and 1103 are applied
and affect the unmasked areas of the drift 606. Top gate region
605a is created with a low energy implant 1101 of the second doping
type. Bottom gate region 605b is created with a high energy implant
1102 of the second doping type. With the same mask, the channel
implant 1103 is also performed to create the channel region 604.
Thereafter mask 608 may be removed. In the case of an n-JFET in
4H-SiC, mask 608 may be a thick oxide hard mask, for example, to
block high temperature Al implants for the top and bottom gates,
and an N implant for the channel.
[0045] In FIG. 12, a new mask 1202 is applied. Mask 1202 blocks the
implant in all regions outside of the active cell array, not shown.
This includes areas such as the gate bus, gate pad, and termination
regions. Mask 1202 is opened between the triple-implanted zones in
the active area. An implant 1201 of the second doping type is then
applied to form vertical channel regions 602a and 602b in the
exposed drift regions. This implant may be designed with an overlap
into the channel defined by the previous mask in order to mitigate
threshold voltage variations from edge effects caused by hard mask
slope and deep gate implantation ion scattering. The implant may
also be designed to have a depth deeper than the bottom gate in
order to optimize the trade-off between breakdown voltage and
reduced on-resistance from current spreading. Mask 1202 is then
removed. In a 4H-SiC n-JFET, Mask 1202 could be either a hard mask
or a photoresist mask, for example.
[0046] Mask 1202 may also be opened at the outer termination edge
for the creation of a channel stop, along with the subsequent
source implant.
[0047] In FIG. 13 a mask 1302 is applied to pattern an implant 1301
of the first dopant type. Implant 1301 is used to create the source
region 603, thereby splitting the gate top into regions 605a and
605c. Mask 1302 may then be removed. Mask 1302 may be, for example,
a hard mask or a photoresist mask.
[0048] The energy of implant 1301 is selected to balance design
considerations. Region 603 should be deep enough to connect to the
channel 604. However, it is preferable that implant 1301 does not
increase the doping at the junction of the channel 604 to bottom
gate 605b. This avoids or reduces any loss of gate-source breakdown
or any increased leakage in the gate-source junction. Of course,
the designer may also tune this depth. For example, a specific
zener clamp voltage may be desired between the gate and source.
[0049] FIG. 14 shows a top view of the wafer in process. A mask
with an opening in is applied, allowing gate contact region 605d to
be created through the opening via implantation. The implantation
may include both gate surface enrichment implant and a gate top to
gate bottom bridging implant. The former aids ohmic connection to
the gates. The latter connects the top and bottom gate layers. To
allow for mask alignment tolerances, a minimum gap 1404, e.g., 0.1
to 0.5 microns should be maintained between the heavy doped source
region 603 and gate ohmic region 605d. This aids in maximizing
gate-source breakdown and minimizing leakage.
[0050] FIG. 15 shows the appearance of the cell after all implants
are done and all surface films are stripped prior to implant
anneal. In the case of a 4H-SiC device, implant anneals may be
carried out at temperatures exceeding 1600 degrees C., with a
graphite cap to prevent surface degradation, or any other technique
commonly used in the art.
[0051] Once the implants are activated, the surface is cleaned and
a dielectric is deposited. FIG. 16 shows a 4H-SiC device with an
oxide mask, opened at the separate source and gate contacts. A
two-step silicidation process may be used to form silicided regions
609 and 613 with NiSi2. Alternatively, a two-mask approach may be
used to form separate silicides in the N- and P- regions by opening
each contact zone in turn.
[0052] FIG. 17 shows the device cross section though the source
contact along section FF' of FIG. 16. A first metal layer 610, such
as titanium topped by titanium-tungsten (Ti/TiW) is deposited on
the first dielectric 608, and patterned with a metal pattern mask,
not shown. Then a second dielectric 611 is deposited and patterned
with a via pattern mask, followed by the deposition and patterning
of the final metal layer 612. The final metal layer 612 may be
might be titanium topped by aluminum (Ti/Al.) A thick Al layer is
typically suitable for active area bonding for high current
devices.
[0053] FIG. 18 shows a vertical cross section in the gate pickup
region, along section EE' of FIG. 16. This shows the active area
after the steps described in reference to FIG. 17 are completed.
Referring to FIG. 18, the Ti/TiW metal layer 610 connects to the
gate silicide 613 that makes contact to the top and bottom gates
via the heavier doped region 605d. The metal 610 is patterned so
that the line running to the gate is isolated from the source and
serves to bring the gate signal to the gate bus at the device
periphery. The dielectric region 611 isolates the gate metal 610
from the source metal 612 everywhere in the active area, while the
dielectric region 608 keeps this thin gate metal isolated from the
underlying JFET cell.
[0054] FIG. 19 shows a top view of a simple stripe cell layout of
an example device. The regions 703 with dashed outlines are gate
metal lines formed in the first thinner metal layer. They connect
to the edge of the die on the left. Via 706 connects the gate metal
lines 703 to the final, thicker metal in region 701, shown with a
solid outline. Region 701 is the gate buss. Region 707 is a pad
where a gate bond wire may be placed. The source regions are
connected to a separate section 704 of the first, thinner metal
layer. Vias 705 open a path through the upper dielectric connecting
the source metal lines 704 to the source metal pad 702 formed in
the upper metal layer. It will be appreciated that many design
variations are possible to use the two metal layers to address the
gate and source current distribution in triple-implanted JFET
devices described herein.
[0055] FIG. 20 shows the output drain current (Id) versus
drain-source voltage (Vds) characteristics 801a of triple-implanted
JFET devices as described herein as compared with the Id-Vds
characteristics 802a of prior art devices. The flat current in
curve 801a after saturation, and tight control over saturation
current levels, are very useful for current limiting applications.
They result from a number of factors, including the shielding
effect of the lower gate and vertical channel region, the longer
channel length based on lateral cell dimensions now possible
without deep trench etching, and the ease with which a source
ballast resistance may be formed within the cell.
[0056] FIG. 21 shows the results of the use of the source ballast
resistance, e.g., a long channel length, on the device current
versus time in response to a surge such as a lightning strike.
Curve 801b is for a triple-implanted JFET device as described
herein, and curve 802b is for prior art devices. The new structure
lets through much lower energy, which may be estimated by
integrating the area under the square of the current versus time
curve. Another useful aspect of this structure is the low internal
gate resistance (Rg) possible with designs using two metal layers.
That in turn prevents any excess drain current due to dV/dt induced
turn-on caused by a displacement current through the internal
Rg.
[0057] FIG. 22 shows a third embodiment, where the vertical channel
JFET regions 102a and 102b are formed by angled implantation using
the same mask window as that used for the triple implant that forms
the top gate regions 105a and 105b, the bottom gate regions 105c
and 105d, and the horizontal channel regions 104a and 104b. Using
the same mask avoids any misalignment problems between the vertical
channel and triple implanted zones. However, since JFET regions
102a and 102b may have a higher peak concentration than the channel
regions 104a and 104b, it would require a significantly higher dose
for the deeper gate 105c and 105d to ensure conversion to p-type
behavior. On the other hand, the energy of the implants 102a and
102b may be selected to be high enough to form a current spreading
layer under the deep gate to further enhance on-resistance at the
expense of a lower device breakdown voltage.
[0058] In describing embodiments of the subject matter of the
present disclosure, as illustrated in the figures, specific
terminology is employed for the sake of clarity. The claimed
subject matter, however, is not intended to be limited to the
specific terminology so selected, and it is to be understood that
each specific element includes all technical equivalents that
operate in a similar manner to accomplish a similar purpose.
[0059] This written description uses examples to disclose the
invention, including the best mode, and also to enable any person
skilled in the art to practice the invention, including making and
using any devices or systems and performing any incorporated
methods. The patentable scope of the invention is defined by the
claims and may include other examples that occur to those skilled
in the art. Such other examples are intended to be within the scope
of the claims if they have structural elements that do not differ
from the literal language of the claims, or if they include
equivalent structural elements with insubstantial differences from
the literal languages of the claims.
* * * * *