U.S. patent application number 15/194774 was filed with the patent office on 2016-10-20 for semiconductor device with junction termination extension.
The applicant listed for this patent is GENERAL ELECTRIC COMPANY. Invention is credited to Stephen Daley Arthur, Alexander Viktorovich Bolotnikov, Kuna Venkat Satya Rama Kishore, James William Kretchmer, Peter Almern Losee, Kevin Sean Motocha, Richard Joseph Saia, Ljuibisa Dragolijub Stevanovic, Zachary Matthew Stum.
Application Number | 20160307997 15/194774 |
Document ID | / |
Family ID | 48614126 |
Filed Date | 2016-10-20 |
United States Patent
Application |
20160307997 |
Kind Code |
A1 |
Arthur; Stephen Daley ; et
al. |
October 20, 2016 |
SEMICONDUCTOR DEVICE WITH JUNCTION TERMINATION EXTENSION
Abstract
A semiconductor device may include a substrate comprising
silicon carbide; a drift layer disposed over the substrate doped
with a first dopant type; an anode region disposed adjacent to the
drift layer, wherein the anode region is doped with a second dopant
type; and a junction termination extension disposed adjacent to the
anode region and extending around the anode region, wherein the
junction termination extension has a width and comprises a
plurality of discrete regions separated in a first direction and in
a second direction and doped with varying concentrations with the
second dopant type, so as to have an effective doping profile of
the second conductivity type of a functional form that generally
decreases along a direction away from an edge of the primary
blocking junction.
Inventors: |
Arthur; Stephen Daley;
(Glenville, NY) ; Bolotnikov; Alexander Viktorovich;
(Niskayuna, NY) ; Losee; Peter Almern; (Clifton
Park, NY) ; Motocha; Kevin Sean; (El Dorado, AR)
; Saia; Richard Joseph; (Niskayuna, NY) ; Stum;
Zachary Matthew; (Niskayuna, NY) ; Stevanovic;
Ljuibisa Dragolijub; (Niskayuna, NY) ; Kishore; Kuna
Venkat Satya Rama; (Hyderabad, IN) ; Kretchmer; James
William; (Ballston Spa, NY) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
GENERAL ELECTRIC COMPANY |
Schenectady |
NY |
US |
|
|
Family ID: |
48614126 |
Appl. No.: |
15/194774 |
Filed: |
June 28, 2016 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14396852 |
Oct 24, 2014 |
9406762 |
|
|
15194774 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H01L 29/1608 20130101;
H01L 29/7395 20130101; H01L 21/0465 20130101; H01L 29/36 20130101;
H01L 29/8611 20130101; H01L 29/6606 20130101; H01L 29/0646
20130101; H01L 29/7811 20130101; H01L 29/0638 20130101; H01L
29/8613 20130101; H01L 29/0619 20130101 |
International
Class: |
H01L 29/06 20060101
H01L029/06; H01L 29/16 20060101 H01L029/16; H01L 29/861 20060101
H01L029/861; H01L 29/66 20060101 H01L029/66; H01L 29/739 20060101
H01L029/739; H01L 29/78 20060101 H01L029/78; H01L 29/36 20060101
H01L029/36; H01L 21/04 20060101 H01L021/04 |
Claims
1. A semiconductor device comprising: a substrate comprising
silicon carbide; a drift layer disposed over the substrate doped
with a first dopant type, so as to have a first conductivity type;
an anode region disposed adjacent to the drift layer, wherein the
anode region is doped with a second dopant type, so as to have a
second conductivity type; and a junction termination extension
disposed adjacent to the anode region and extending around the
anode region, wherein the junction termination extension has a
width w.sub.jte and comprises a plurality of discrete regions
separated in a first direction and in a second direction and doped
with varying concentrations with the second dopant type, so as to
have an effective doping profile of the second conductivity type of
a functional form that generally decreases along a direction away
from an edge of the primary blocking junction, wherein the width
w.sub.jte is less than or equal to a multiple of five (5) times the
width of the one-dimensional depletion width
(W.sub.depl.sub._.sub.1D), and wherein a charge tolerance of the
semiconductor device is greater than 1.0.times.10.sup.13 per
cm.sup.2.
2. The semiconductor device of claim 1, wherein the anode region
comprises a material disposed atop the drift layer.
3. The semiconductor device of claim 1, wherein the effective
doping profile of the junction termination extension is a
monotonically decreasing function N(x) of the distance x away from
the edge of the primary blocking junction.
4. The semiconductor device of claim 3, wherein the monotonically
decreasing function N(x) that governs the effective doping profile
of the junction termination extension varies as x.sup.1/2.
5. The semiconductor device of claim 3, wherein the monotonically
decreasing function that governs the effective doping profile of
the junction termination extension is:
N(x)=N.sub.max+(N.sub.min-N.sub.max)(x/w.sub.jte).sup.1/2, wherein
N.sub.max is the average dopant concentration at the edge of the
primary blocking junction, and wherein N.sub.min is the average
dopant concentration at an outer edge of the junction termination
extension.
6. The semiconductor device of claim 1, wherein neighboring ones of
the discrete doped regions are separated from their nearest
neighbors by a spacing in a range of about 0 to about
2.5.lamda..
7. The semiconductor device of claim 6, wherein the minimum
effective doping is no smaller than 15% of the full JTE dose.
8. The semiconductor device of claim 1, wherein the silicon carbide
substrate has a n+ conductivity type, wherein the first dopant type
is n-type, such that the first conductivity type is n-type, and
wherein the second dopant type is p-type, such that the second
conductivity type is p-type.
9. The semiconductor device of claim 1, wherein the silicon carbide
substrate has a p-type conductivity type, wherein the first dopant
type is n-type, such that the first conductivity type is n-type,
and wherein the second dopant type is p-type, such that the second
conductivity type is p-type.
10. The semiconductor device of claim 1, wherein the silicon
carbide substrate has a n+-type conductivity type, wherein the
first dopant type is p-type, such that the first conductivity type
is p-type, and wherein the second dopant type is n-type, such that
the second conductivity type is n-type.
11. The semiconductor device of claim 1, wherein at least a portion
of the junction termination extension is formed in the anode
region.
12. The semiconductor device of claim 11, wherein the portion of
the junction termination extension formed in the anode region
physically connects the anode region to the junction termination
extension via a same dopant type.
13. A semiconductor device comprising a semiconductor substrate
having a first surface and a second surface; an active region
formed on the substrate; and an edge region surrounding the active
region and having a width W.sub.edge, wherein the edge region
comprises: a plurality of discrete corner regions having impurities
of a second conductivity type; and a plurality of discrete straight
regions having impurities of the second conductivity type, wherein
at least one of the straight regions adjoins respective ones of the
corner regions, wherein the effective impurity concentration of the
second conductivity type decreases along a direction away from an
interface between the edge region and the active region, wherein
the shape of the corner regions differs from the shape of the
straight regions, and wherein the width of the edge region
W.sub.edge is less than or equal to a multiple of five (5) times
the one dimensional depletion width (W.sub.depl.sub._.sub.1D).
14. The semiconductor device of claim 13, wherein at least one of
the straight regions has a rectangular shape, and wherein at least
one of the corner regions has a trapezoidal shape.
15. The semiconductor device of claim 13, wherein at least one of
the straight regions is square.
16. The semiconductor device of claim 15, wherein each of the
corner regions is trapezoidal.
17. The semiconductor device of claim 13, wherein the edge region
has a charge tolerance of at least about
1.0.times.10.sup.13/cm.sup.2.
18. The semiconductor device of claim 17, wherein the charge
tolerance suffices to accommodate an interface charge density of at
least about 1.times.10.sup.12/cm2.
19. A semiconductor device comprising a semiconductor substrate
having a first surface and a second surface; an active device
region formed on the substrate comprising a primary blocking
junction; and an edge region adjacent the primary blocking junction
having a width W.sub.edge, wherein the edge region comprises a
plurality of discrete regions having a plurality of impurities of a
first conductivity type, wherein an effective impurity
concentration of the first conductivity type in the edge region
decreases along a direction away from an interface between the
primary blocking junction and the edge region, and wherein the
width of the edge region W.sub.edge is less than or equal to a
multiple of five (5) times the one dimensional depletion width
(W.sub.depl.sub._.sub.1D).
20. The semiconductor device of claim 19, wherein the edge region
has a charge tolerance of at least about
1.0.times.10.sup.13/cm.sup.2.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of U.S. application Ser.
No. 14/396,852, filed on Oct. 24, 2014, which is a National Stage
Entry of PCT Application PCT/US13/41073, filed on May 15, 2013,
which claims priority to and the benefit of U.S. Provisional Patent
Application 61/648,149, filed on May 17, 2012. All of the prior
applications are hereby incorporated by reference herein.
BACKGROUND
[0002] The invention relates generally to semiconductor devices
and, more particularly, to silicon carbide-based devices that
utilize a junction termination extension.
[0003] Breakdown voltage of the reverse-blocking junction typically
serves to limit the maximum reverse voltage that a semiconductor
device formed with a p-n junction can withstand. Such a blocking
junction may comprise, for example, a p-n junction of a thyristor,
a diode, a bipolar transistor, an insulated-gate transistor, or a
corresponding junction in a metal-oxide-semiconductor field-effect
transistor (MOSFET). Avalanche breakdown occurs in such a device at
a voltage substantially less than the ideal breakdown voltage
because excessively high electric fields are present at certain
locations ("high field points") in the device under reverse bias. A
high field point of a blocking junction under reverse bias usually
occurs slightly above the metallurgical junction along a region of
curvature, such as that at the end of the junction.
[0004] In particular, breakdown voltage is critical for high power
devices, such as silicon carbide (SiC) devices, and related
properties, such as robustness to active dose and interface charge
variation, are more significant in SiC devices, than in silicon
(Si) based devices.
[0005] Semiconductor devices may utilize any of various structures
and methods to achieve an increase in the breakdown voltage of a
p-n junction, for example close to p-n junction entitlement. For
example, junction termination extension (JTE) regions may be
utilized near terminated portions of the p-n junction. In general,
a JTE region may be considered as a more lightly doped extension of
a heavily doped semiconductor region that adjoins a semiconductor
region of the opposite conductivity type and which is usually
lightly doped, to form the foregoing p-n junction. The principal
function of the JTE region is to reduce the high concentration of
electric fields that would otherwise exist in the vicinity of the
non-terminated portion of the p-n junction, and especially at the
high field points (which are typically near the corners of the
locally doped regions), by laterally extending the blocking
junction.
[0006] In addition to breakdown voltage, the design of the JTE
affects a number of critical properties of the semiconductor
device, including reliability, fabrication process complexity, and
charge tolerance, and many of the affected properties have complex
interrelationships.
[0007] It would therefor be desirable to provide a JTE design that
improves the critical properties of silicon-carbide based
semiconductor devices, such as break down voltage, charge tolerance
and reliability.
DRAWINGS
[0008] These and other features, aspects, and advantages of the
present invention will become better understood when the following
detailed description is read with reference to the accompanying
drawings in which like characters represent like parts throughout
the drawings, wherein:
[0009] FIG. 1 is a cross-sectional view of an example diode
configured in accordance with aspects of the present invention;
[0010] FIG. 2 is a partial top view, showing the varying dopant
profile for an example junction termination extension (JTE);
[0011] FIG. 3 is a cross-sectional view of the termination for a
blocking junction for an IGBT configured in accordance with aspects
of the present invention;
[0012] FIG. 4 is a cross-sectional view of an example mesa diode
configured in accordance with aspects of the present invention;
[0013] FIG. 5 illustrates three example JTE effective doping
profiles (single zone, x.sup.1/2, and x.sup.2) across the length of
the JTE;
[0014] FIG. 6 shows the resulting breakdown voltage sensitivity to
JTE peak dose variation curves for the three doping profiles used
in FIG. 5;
[0015] FIG. 7 is a plan view of dimensional definitions for a
diode, which is generalizable to a generic vertical power
device;
[0016] FIG. 8 schematically shows an example layout of a brick
structure G-JTE mask with a varying lateral JTE dose;
[0017] FIG. 9 is a plan view of dimensional definitions for a
generic vertical power device;
[0018] FIG. 10 illustrates example graded JTE (G-JTE) lateral
doping profiles;
[0019] FIG. 11 illustrates an example unit cell for a brick
structure G-JTE layout; and
[0020] FIG. 12 illustrates the generation of unit cells around
device corners where rectangular bricks become trapezoids, with
heights determined by radial increases in steps of 2.
DETAILED DESCRIPTION
[0021] A technique for terminating high voltage SiC junctions is
described below which can achieve blocking voltages extremely close
to the one dimensional parallel-plane break down voltage (1-D
BV.sub.PP limit) and offers improved robustness to active dose and
interface charge variations which are more significant in SiC than
Si power device applications. This new brick structure Graded
Junction Termination (G-JTE) design, which for particular
configurations, uses only one masking level, has been implemented
with 1.2 kV SiC MOSFETs and representative test diodes. The
blocking voltage (BV) of the test diodes reached the 1-D BV.sub.PP
limit of approximately 1.6 kV with 11 .mu.m, n-type 4H-SiC drift
layers, doped ND=9.times.10.sup.15/cm.sup.3, where ND is the donor
concentration. The robustness to charge sensitivity has been
experimentally verified, maintaining BV>1.2 kV over a wide range
of implanted JTE doses (2.times.10.sup.13/cm.sup.2 to over
4.times.10.sup.13/cm.sup.2), greatly outperforming the conventional
single-zone JTE designs. This single masking step, brick structure
G-JTE termination design lends itself especially well to even
higher voltage device applications (>3 kV), where implementing
multiple-zone JTEs, requiring multiple (up to four) masking levels,
are routinely used. It should be noted that the JTE design was
verified up to 8 kV.
[0022] A semiconductor device 200 is described with reference to
FIGS. 1-3 and FIGS. 5-10. It should be noted that although the
contacts are not shown in FIG. 1 for ease of illustration, the
semiconductor device will include contacts, as known in the art. As
indicated in FIG. 1, the semiconductor device 200 includes a
substrate 202 comprising silicon carbide. A drift layer 214 is
disposed over the substrate 202 and comprises a drift region 214
doped with a first dopant type, so as to have a first conductivity
type. The semiconductor device 200 further includes a second region
216 adjacent to the drift region 214 and proximal to a surface 204
of the drift layer 214. The second region 216 is doped with a
second dopant type, so as to have a second conductivity type.
[0023] The semiconductor device 200 further includes a junction
termination extension (JTE) 220 disposed adjacent to the second
region 216. As indicated, for example, in FIGS. 1 and 2, the
junction termination extension 220 has a width w.sub.jte, and
comprises a number of discrete regions 221 separated in a first
direction 272 and in a second direction 274 (FIG. 2) and doped with
varying concentrations of the second (p-type) dopant type, so as to
have an effective doping profile of the second conductivity type of
a functional form that generally decreases along a direction away
from the edge of the primary blocking junction 230. It should be
noted that by "varying concentration", it is meant that the density
of the regions varies, and it is this varying density that defines
the effective JTE dose that varies. Typically, all regions will
have the same dose/doping. As used here, "effective doping" is the
fraction of the JTE area that is open to receive the JTE implant
dose versus the total area being sampled, and thus it is the
equivalent of dosing the JTE at (fraction)*(full JTE dose) for that
region. Some example effective doping levels include 15%, 50% and
90% of the full JTE dose. The width w.sub.jte (FIG. 1) is less than
or equal to a multiple of five (5) times the width of the
one-dimensional depletion width W.sub.depl.sub._.sub.1D, and the
charge tolerance of the semiconductor device 200 is greater than
1.0.times.10.sup.13 per cm.sup.2. For particular configurations,
the interface (or field oxide) charge density may be about
1.times.10.sup.12/cm.sup.2, or even in a range of about 40-60% of
1.times.10.sup.12/cm.sup.2. Beneficially, the charge tolerance
suffices to accommodate the interface charge density.
[0024] It should be noted that what is shown in FIG. 1 as
W.sub.depl.sub._.sub.1D is the thickness of the epi layer t.sub.epi
and generally, W.sub.depl.sub._.sub.1D is not equivalent to
t.sub.epi. As used here, charge tolerance is defined as the span of
the JTE dose range for a particular JTE type where the breakdown
voltage is greater than the design voltage. See, for example, Table
1.
TABLE-US-00001 TABLE 1 Charge Tolerance for three example JTE
effective doping profiles (single zone, x.sup.1/2, and x.sup.2)
W.sub.jte/ JTE_dose_min JTE_dose_max Charge Design
W.sub.depl.sub.--.sub.1D (#/cm.sup.2) (#/cm.sup.2) tolerance Single
4.4 1.1 .times. 10.sup.13 1.6 .times. 10.sup.13 0.5 .times.
10.sup.13 zone Quad 4.4 1.1 .times. 10.sup.13 2.6 .times. 10.sup.13
1.5 .times. 10.sup.13 design SQRT 4.4 1.6 .times. 10.sup.13 6.0
.times. 10.sup.13 4.4 .times. 10.sup.13 design
[0025] As used here, "single zone" (indicated by reference numeral
520 in FIG. 5) refers to a constant dose. As discussed below, the
effective doping profile (or effective impurity concentration) may
be tailored depending on the specific device characteristics that
are desired. For particular configurations, the effective doping
profile of the junction termination extension 220 is a
monotonically decreasing function N(x) of the distance x away from
the edge of the primary blocking junction 230. Still more
particularly, the monotonically decreasing function N(x) that
governs the effective doping profile of the junction termination
extension (220) varies as x.sup.1/2, and still more particularly,
is:
N(x)=N.sub.max+(N.sub.min-N.sub.max)(x/w.sub.jte).sup.1/2
where N.sub.max is the average dopant concentration at the edge of
the primary blocking junction 230, and where N.sub.min is the
average dopant concentration at an outer edge 232 (332 in FIG. 4)
of the junction termination extension 220 (320 in FIG. 4).
[0026] FIG. 5 shows three example doping profiles across the length
of an example JTE, and FIG. 6 shows the corresponding dose
sensitivity curves. The JTE dose may be viewed as the net JTE dose.
In addition to the square-root functional form (which is indicated
by reference numeral 510 in FIG. 5) discussed above, FIG. 5 shows a
monotonically decreasing function N(x) that governs the effective
doping profile of the junction termination extension 220 which
varies as x.sup.2 (which is indicated by reference numeral 500 in
FIG. 5) and, more particularly, which is
N(x)=N.sub.max+(N.sub.min-N.sub.max)(x/w.sub.jte).sup.2,
where N.sub.max is the average dopant concentration at the edge of
the primary blocking junction 230, and where N.sub.min is the
average dopant concentration at an outer edge 232 of the junction
termination extension 220.
[0027] As indicated in FIG. 6, although the single zone (reference
numeral 520 in FIG. 5) and quadratic (reference numeral 500 in FIG.
5) effective doping profiles can achieve similar peak breakdown
voltage (BV) values, the square-root (reference numeral 510 in FIG.
5) effective doping profile provides the broadest charge control
range. For example, for a 1200 volt device and assuming that the
design breakdown specification for the 1200 volt device is 15%
greater than the device rating (.gtoreq.1380 volts), the BV SQRT
distribution provides the widest dose range which meets this
constraint, (.about.1.8.times.10.sup.13 to 6.0.times.10.sup.13).
The design voltage is indicated by reference numeral 600 in FIG. 6.
If the JTE dose is centered at 3.5.times.10.sup.13/cm.sup.2, it can
theoretically accommodate charge variations larger than
+/-1.7.times.10.sup.13. By comparison the single zone JTE can
achieve similar peak BV values, but the charge control range is
much narrower, about +/-2.times.10.sup.12/cm.sup.2 centered about
the peak BV dose.
[0028] Returning now to FIG. 1, the semiconductor device 200
further includes a passivation layer 206 disposed over the drift
layer 214. For particular configurations, the passivation layer 206
comprises a multilayer structure. A variety of materials for the
passivation layer 206 may be used and include, without limitation,
silicon oxide, phosphor doped silicate glass layer (PSG), silicon
nitride, and polyimide. It should be noted that these are example
materials for the passivation layer, and the passivation layer may
take other forms depending on the specific implementation.
[0029] Charge tolerance is an important aspect of the semiconductor
devices of present invention. To determine its significance, the
present inventors posed the following rhetorical question: what are
the sources and the magnitudes of charges available in the
termination region for SiC? Listed in Table 2.2 below are estimates
compared to the silicon case. Gate oxides are not directly relevant
to termination considerations but are added as a lower bound for
control of charge density for each technology. Charge densities in
the table are treated as effective charge at the interface (as
modeled in the charge sensitivity curves) and assumed positive in
polarity however the charge sensitivity curve suggests that proper
placement of the target JTE can accommodate either polarity.
TABLE-US-00002 TABLE 2.2 Estimates of the magnitudes of charge
densities in silicon & SiC terminations: silicon SiC gate oxide
2 .times. 10.sup.10/cm.sup.2 2.5 .times. 10.sup.11-1.3 .times.
10.sup.12/cm.sup.2 Field oxide 1-3 .times. 10.sup.11/cm.sup.2
unknown PECVD nitride 1.0 .times. 10.sup.12/cm.sup.2 1.0 .times.
10.sup.12/cm.sup.2 effect Dopant activation ~100% 40-70%
[0030] The effects that are most important are those that
contribute charge densities that are a significant fraction of that
density required to support the peak E-field at breakdown, for
silicon
Q.sub.0.sub._.sub.silicon.about.1.3.times.10.sup.12/cm.sup.2, for
SiC Q.sub.0.sub._.sub.sic.about.1.3.times.10.sup.13/cm.sup.2. For
SiC, the possible variation of the activated implant dose is the
largest apparent contributor creating an uncertainty of
.about.1.times.10.sup.13/cm.sup.2 when the target JTE dose is set
at 3.5.times.10.sup.13/cm.sup.2. This is probably larger than the
other effects; e.g. the field oxide charge density is unknown and
speculated to be in the 10.sup.12/cm.sup.2 range.
[0031] Having a termination that can accommodate a large range of
charge variations without significant extra area being consumed is
a major advantage as the process control of SiC charges (dynamic
and static) is relatively immature compared to silicon. However,
competing factors associated with the design of the junction
termination extension can have various effects on the charge
tolerance. Because of its overall importance to the performance and
reliability of the semiconductor device, the JTE design is selected
to ensure sufficient charge tolerance.
[0032] In addition to the functional form of the effective doping
profile, the spacing and geometry can affect the charge tolerance
and overall performance of the junction termination extension. FIG.
7 is a plan view of dimensional definitions for a diode with an
active area 920, which can be generalized to a generic vertical
power device. FIG. 2 shows an example "brick" structure for a
junction termination extension. As used here, a "brick" 221 (FIG.
1) comprises a small, discrete region 221 of either an opening in
an implantation mask (implanted dose) or a masked region (no dose).
By varying the amount of open vs. masked areas using these
"bricks", the continuous lateral doping profile simulated in the
previous section could be approximated by a number of discrete
regions. Referring to FIGS. 1 and 2, "G-JTE" is the graded JTE
structure. "LDG" is the lateral depletion gap 240 (FIG. 1), 340
(FIGS. 3 and 4), which is the region where the epi reaches the
surface plane. The depletion edge 260 is adjacent to the depletion
gap 240. The fieldstop 250 (FIG. 1), 350 (FIG. 3), is a heavily
doped region, doped with the same type as the epi (N type for the
example shown in FIG. 1). The purpose of the fieldstop 250 is to
stop any electric field from reaching the saw street, which
separates the chips. The "saw street" is where the dicing saw cuts
through the substrate to singulate devices from the wafer. A half
saw street 270 is shown in FIG. 2. The semiconductor is damaged by
the sawing, offering a leaky or resistive surface if any electric
field is allowed to reach it. The fieldstop 250 halts any depletion
expansion ensuring that the electric-field does not reach the saw
street. Geometrically, the outside periphery can occupy a
significant fraction of the chip area.
[0033] The minimum "brick" size (referred to herein as .lamda.) is
practically limited by lithographic and other microelectronic
process steps, but in the limiting case wherein .lamda..about.0
(defined below), the brick structure G-JTE becomes a continuous
structure. FIG. 8 schematically shows an example layout of a brick
structure G-JTE mask. The dose is varied from x=0 (100% implanted
dose) by x.sup.-1/2 with a discrete brick size of .lamda..about.1.3
.mu.m (illustrated)
[0034] For particular configurations, neighboring ones of the
discrete doped regions 221 are separated from their nearest
neighbors by a spacing in a range of about 0 to about 2.5.lamda.,
assuming the minimum effective doping is no smaller than 15% of the
full JTE dose. However, for other examples, the minimum effective
doping may be less than 15% of the full JTE dose. For particular
physical models, .lamda. may be defined as:
.lamda..ltoreq.( 1/10)*W.sub.depl.sub._.sub.1D,
That is, .lamda. is defined as being less than or equal to one
tenth of the width of the one dimensional depletion width of the
blocking junction at its breakdown voltage. Higher voltage
structures can use larger .lamda.'s, as their 1D depletion widths
are larger. For example: for BV-1000 volts,
W.sub.depl.sub._.sub.1D.about.10 .mu.m in 4H SiC, therefore
.lamda..about.1.0 .mu.m, for B.about.3000 volts,
W.sub.depl.sub._.sub.1D.about.30 .mu.m and .lamda..ltoreq.3.0
.mu.m. As noted above, there is a limit to how well the small
structures can be printed, and this will bound the lower dimension
of k.
[0035] The graded "brick" pattern can be formed using a graded,
hard mask, for example a SiO.sub.2 mask. With the graded JTE mask,
the pattern goes from open islands to small holes. When designing
the graded mask, a corrective factor may be used to achieve the
desired functional form for the effective doping profile. That is,
the algorithm used to generate the pattern of islands and holes in
the SiO.sub.2 mask may be corrected for proximity effects.
[0036] For particular configurations, the ratio of the active area
of the junction termination extension 220 to the total area of the
semiconductor device 200 is greater than about sixty percent (60%),
and more particularly, is greater than about sixty five percent
(65%), and still more particularly, is greater than about seventy
percent (70%). Referring now to FIGS. 7 and 9, the active area 920
is that portion of the vertical power device that contributes
directly to the vertical current flow. For example, for a PIN
diode, the active area 920 would be the anode area (FIG. 7).
Similarly, for switches, the active area is the area of the cells
that control the current flow. For the power device shown in FIG. 9
(a field effect transistor, or FET, or an insulated gate bipolar
transistor, or IGBT), the active area 920 now excludes the gate pad
and the gate runners. More generally, areas of the vertical power
device that are not part of the active area include: the areas
reserved for the saw streets, the fieldstop regions, or any area
taken up by the termination (LDG and JTE), and areas consumed by
overhead items e.g. gate contact pads 900 or gate runners 910 (as
shown in FIG. 9). The total device area is the area of the chip
from corner to corner (usually power chips are squares or
rectangles). For particular examples, JTEs may be formed using
relatively small bricks (.about.1 .mu.m), with 15%, 50% and 90%
open areas. Beneficially, by achieving a relatively high active to
total chip area ratio (while still achieving the necessary
performance for the JTE), the current conduction when the device is
in the "on" state is improved. This can be achieved, for example,
by ensuring that the termination length (G-JTE+LDG+fieldstop)
should be as short as possible, while still meeting all the design
goals, as the termination occupies the outside periphery of the
device.
[0037] It should be noted that the device structure indicated in
FIG. 1 is applicable to a number of device types, depending on the
specific doping. For example, for particular configurations, the
silicon carbide substrate 202 has a n+ conductivity type, where the
first dopant type is n-type, such that the first conductivity type
is n-type, and where the second dopant type is p-type, such that
the second conductivity type is p-type. For this configuration, the
p-type second region 216 and the n-type drift layer 214 form a p-n
junction, such that the semiconductor device 200 comprises a diode.
This example diode configuration is illustrated schematically in
FIG. 1. For other configurations, the substrate and the drift layer
are p-type, and the second region is n-type.
[0038] Similarly, for the configuration shown in FIG. 3, the
silicon carbide substrate 202 has a p-type conductivity type, and
the first dopant type is n-type, such that the first conductivity
type is n-type. For this configuration, the second dopant type is
p-type, such that the second conductivity type is p-type. For this
configuration, the p-type second region 216 and the n-type drift
layer 214 form a p-n junction, and the substrate 202 and the drift
layer 214 form another p-n junction, such that the semiconductor
device 200 comprises a transistor, for example a thyristor or an
IGBT. This example transistor configuration is illustrated
schematically in FIG. 3. It should be noted that although FIG. 3
shows a non-punch-through IGBT, the JTE is equally applicable to
punch-through IGBTs as well, which include a buffer layer (not
shown) of the same type as the drift layer.
[0039] Similarly, FIG. 3 also schematically illustrates a
configuration, for which the silicon carbide substrate 202 has a
n+-type conductivity type, and the first dopant type is p-type,
such that the first conductivity type is p-type. For this
configuration, the second dopant type is n-type, such that the
second conductivity type is n-type. For this configuration, the
n-type second region 216 and the p-type drift layer 214 form a p-n
junction, and the substrate 202 and the drift layer 214 form
another p-n junction, such that the semiconductor device 200
comprises a transistor, for example a thyristor or an IGBT. Other
aspects of the device 200 shown in FIG. 3 are similar to those
described above with reference to FIG. 1. For example, the junction
termination extension 220 has a width w.sub.jte and comprises a
number of discrete regions 221, which are doped with varying
concentrations of the second (n-type) dopant type, so as to have an
effective doping profile of the second conductivity type of a
functional form that generally decreases along a direction away
from the edge of the primary blocking junction 230. The outer edge
232 of the junction termination extension 220 is indicated in FIG.
3. The depletion edge 360 is adjacent to the depletion gap 340. The
fieldstop 350 (FIG. 3), is a heavily doped region, doped with the
same type as the epi (N type for the example shown in FIG. 3). As
noted above, the purpose of the fieldstop 350 is to stop any
electric field from reaching the saw street, which separates
neighboring devices.
[0040] Further, the device 200 shown in FIG. 3 further includes a
passivation layer 206 disposed over the drift layer 214. The
passivation layer is described above with reference to FIG. 1.
[0041] Another semiconductor device 300 embodiment is described
with reference to FIGS. 2 and 4-10. As shown in FIG. 4, the
semiconductor device 300 includes a substrate 302 comprising
silicon carbide. A drift layer 314 is disposed over the substrate
302 and is doped with a first (n-type) dopant type, so as to have a
first conductivity type. An anode region 316 is disposed adjacent
to the drift layer 314 and is doped with a second (p-type) dopant
type, so as to have a second conductivity type. The semiconductor
device 300 further includes a junction termination extension 320
disposed adjacent to the anode region 316 and extending around the
anode region 316. As indicated in FIGS. 2 and 3, the junction
termination extension 320 has a width w.sub.jte and comprises a
number of discrete regions 321 separated in a first direction and
in a second direction and doped with varying concentrations of the
second (p-type) dopant type, so as to have an effective doping
profile of the second conductivity type of a functional form that
generally decreases along a direction away from the edge of the
primary blocking junction 330. The width w.sub.jte (FIG. 3) is less
than or equal to a multiple of five (5) times the width of the
one-dimensional depletion width W.sub.depl.sub._.sub.1D, and the
charge tolerance of the semiconductor device 300 is greater than
1.0.times.10.sup.13 per cm.sup.2. This example mesa diode
configuration is illustrated schematically in FIG. 4.
[0042] For particular configurations, the anode region 316
comprises a material that was epitaxially grown on the drift layer
314 and was subsequently partially etched to form the anode region
316. The JTE implant into the anode mesa sidewall is indicated by
reference numeral 322 in FIG. 4 and physically connects the P+
anode to the P-type JTE pattern via the same type doping as the JTE
that is located on the lower etched plane.
[0043] For particular configurations, the semiconductor device 300
further includes a passivation layer 306 disposed over the drift
layer 314. The passivation layer is described above. In addition,
the fieldstop region 350 and saw line are in the actual device.
However, the saw line is not shown in FIG. 4 for ease of
illustration.
[0044] The JTE 320 is similar to the JTE 220 described above. In
particular, it should be noted that the JTE 220, 320 correspond to
relatively high values of the following figure of merit (FOM):
JTEFOM = ( Wdepl 1 D Wjte ) * ( Qtol QEcritical ) * ( BVpk BV 1 D )
* ( Ereliable Epk_oxide ) ##EQU00001##
[0045] As used here, W.sub.JTE=width of the JTE from main junction
edge toward the saw street, and W.sub.1Ddepl=depletion width of the
one dimensional vertical doping profile on the lightly doped side,
so a narrower JTE gives a larger first term. For particular
configurations, the first term should be in a range of about
0.2-1.0.
[0046] Turning now to the second term in this expression for the
JTE FOM, Qtol=the charge tolerance range of the design in
#/cm.sup.2 (same as JTE dose), as calculated from the dose
sensitivity curve, and QEcritical=Charges/cm.sup.2 required to
balance the critical electric field (from Gauss's law). Thus, the
second term in the FOM is the ratio of the JTE charge tolerance
divided by the charge required to create the critical 4H-SiC
breakdown field derived from Gauss' law, e.g. see FIG. 3.5 in
Fundamentals of Power Semiconductor Devices, B. Jayant Baliga,
Springer-Science, 2008. The critical field is only a weak function
of the doping on the lightly doped side of the junction and is
.about.3.1.times.10.sup.6 volts/cm for 9.times.10.sup.15/cm3 doped
material common to 1200 volt devices. This gives
QE.sub.critical=.di-elect
cons.*E.sub.critical=(9.7)*(8.85.times.10.sup.-14
F/cm)*(3.1.times.10.sup.6 V/cm)/(1.6.times.10.sup.-19 coul/charge),
.about. 1.7.times.10.sup.13 charges/cm.sup.2. Although this value
was estimated for a 1200 V design, one skilled in the art will
recognize that QE.sub.critical is a function of the device rating.
The charge tolerance is taken as the dose width of the BV vs JTE
dose curve above the design voltage, typically >15% above rated
BV [design voltage.gtoreq.(BV rating)*1.15]. Q.sub.tol is given in
units of # of unit charges/cm.sup.2 (e.g. units of implant dose).
For particular configurations, Q.sub.tol>1.0.times.10.sup.13,
and may be greater. For example, the data presented in Table 1
shows Q.sub.tol of 1.5.times.10.sup.13 for Quad design and
4.4.times.10.sup.13 for the SQRT design, so the
Q.sub.tol/QE.sub.critical ratio may be 0.6 for
Q.sub.tol=1.0.times.10.sup.13, 0.88 for the Quad design, and 2.6
for the SQRT designs for the 1200 volt device data shown.
[0047] The third term of the JTE FOM is the ratio of the peak
breakdown voltage (BV) achievable (BV.sub.pk, due to the
termination design) vs. the 1D BV entitlement, given by calculating
the avalanche BV for the 1D doping profile of the main blocking
junction. This ratio should be in the range 0.80 to 1.0 (>80% of
1D entitlement), and for particular configurations>90%
(ratio>0.9).
[0048] The fourth term in the JTE FOM is the ratio of maximum peak
electric field strength in the passivation layer directly covering
the termination at the rated voltage (e.g. 1200 volts) considered
acceptable for long term reliability, E.sub.reliable, to the
calculated peak field in the oxide layer, E.sub.pk.sub._.sub.oxide,
for a given design and surface charge. The design goal is to keep
E.sub.pk.sub._.sub.oxide<E.sub.reliable so the passivation has
long term reliability. This ratio should never be less than 1.0,
and may be larger (ratios of 1.0-2.0 are typical). As an example,
for silicon oxide E.sub.reliable.about.4.times.10.sup.6 V/cm is
commonly quoted as the value below which silicon dioxide has
extended long term reliability.
[0049] It should be noted that the device structure indicated in
FIG. 4 is applicable to a number of device types, depending on the
specific doping. For example, for particular configurations, the
silicon carbide substrate 302 has a n+ conductivity type, and the
first dopant type is n-type, such that the first conductivity type
is n-type. For this configuration, the second dopant type is
p-type, such that the second conductivity type is p-type. For this
configuration, the p-type anode region 316 and the n-type drift
layer 314 form a p-n junction, such that the semiconductor device
300 comprises a mesa diode. This example mesa diode configuration
is illustrated schematically in FIG. 4.
[0050] Similarly, for other configurations the silicon carbide
substrate 302 has a p-type conductivity type, and the first dopant
type is n-type, such that the first conductivity type is n-type.
For these configurations, the second dopant type is p-type, such
that the second conductivity type is p-type. For this
configuration, the p-type anode region 316 and the n-type drift
layer 314 form a p-n junction, and the substrate 302 and the drift
layer 314 form another p-n junction, such that the semiconductor
device 300 comprises a transistor, for example a thyristor or an
IGBT.
[0051] Further, for other configuration, the silicon carbide
substrate 302 has a n+-type conductivity type, and the first dopant
type is p-type, such that the first conductivity type is p-type.
For these configurations, the second dopant type is n-type, such
that the second conductivity type is n-type. For this
configuration, the n-type anode region 316 and the p-type drift
layer 314 form a p-n junction, and the substrate 302 and the drift
layer 314 form another p-n junction, such that the semiconductor
device 200 comprises a transistor, for example a thyristor or an
IGBT.
[0052] A semiconductor device is also described with reference to
FIGS. 1, 2, 11 and 12. As indicated in FIG. 1, the semiconductor
device includes a semiconductor substrate (for example, a SiC
substrate) having a first surface and a second surface; an active
region formed on the substrate; and an edge region surrounding the
active region and having a width W.sub.edge.
[0053] As indicated in FIGS. 2 and 12, the edge region includes a
number of discrete corner regions having impurities of a second
conductivity type. As indicated in FIGS. 2 and 11, the edge region
further includes a number of discrete straight regions having
impurities of the second conductivity type. As indicated in FIG. 2,
at least one of the straight regions adjoins respective ones of the
corner regions. The effective impurity concentration (or effective
doping profile, as described above) of the second conductivity type
decreases along a direction away from an interface between the edge
region and the active region. As indicated in FIGS. 11 and 12, the
shape of the corner regions differs from the shape of the straight
regions. The width of the edge region W.sub.edge is less than or
equal to a multiple of five (5) times the one dimensional depletion
width (W.sub.depl.sub._.sub.1D).
[0054] For particular configurations, at least one of the straight
regions has a rectangular shape, and at least one of the corner
regions has a trapezoidal shape. FIG. 11 shows example square
(rectangular) straight regions, whereas FIG. 12 shows example
trapezoidal corner regions. For more particular configurations,
each of the straight regions is rectangular, and each of the corner
regions is trapezoidal. For the specific arrangement shown in FIG.
11, the straight regions are square. As will be recognized by those
skilled in the art, the "square", "rectangular", and "trapezoidal
regions will not typically be perfectly square, rectangular or
trapezoidal but rather will typically have somewhat rounded corners
as a result of the inherent limitations of the processing
techniques involved. In addition, these shapes (rectangular,
square, and trapezoidal) are only possible example shapes for the
doped JTE regions, and the regions may have other shapes, as well,
for example circular.
[0055] A semiconductor device is described with reference to FIGS.
1-3 and 5-8. As indicated in FIGS. 1 and 3, for example, the
semiconductor device includes a semiconductor substrate (for
example, a SiC substrate) having a first surface and a second
surface; an active device region formed on the substrate comprising
a primary blocking junction; and an edge region adjacent the
primary blocking junction having a width W.sub.edge.
[0056] As indicated in FIGS. 1 and 2, for example, the edge region
comprises a number of discrete regions having a number of
impurities of a first conductivity type, where an effective
impurity concentration of the first conductivity type in the edge
region decreases along a direction away from an interface between
the primary blocking junction and the edge region. The width of the
edge region W.sub.edge is less than or equal to a multiple of five
(5) times the one dimensional depletion width
(W.sub.depl.sub._.sub.1D). More particularly, the edge region has a
charge tolerance of at least about
1.0.times.10.sup.13/cm.sup.2.
[0057] Beneficially, the above-described junction termination
extensions are area efficient, using a minimum chip area to achieve
a large BV/BV.sub.pp ratio, thus maximizing the allowable chip
active area, for example resulting in area efficiencies for the
semiconductor device that exceed seventy percent (70%). In
addition, the above-described junction termination extensions have
designs that are scalable for higher and lower voltages. Another
significant benefit of the above-described junction termination
extensions is their charge tolerance, that is the resulting
semiconductor devices can accommodate relatively large swings in
surface charge, for example in the passivation layer above the
junction termination extension or corresponding to doping
activation variability for silicon carbide (SiC). This improved
charge tolerance is particularly important for SiC devices, where
the interface charge is unknown and may be dynamic.
[0058] Yet another benefit of the above-described junction
termination extensions is their reliability, namely, the electric
field in the dielectric above the termination is within acceptable
limits. For example, modeling results for the present designs show
peak static fields<1 MV/cm. Other benefits include providing a
passivation scheme over the termination with improved mechanical
immunity from scratches, moisture, and ionic transport. Further,
the above-described junction termination extensions are practical
to implement, in that they require relatively simple processing and
are compatible with FET process and materials. In addition, the
above-described junction termination extensions possess good
capability under high dV/dt.
[0059] Although only certain features of the invention have been
illustrated and described herein, many modifications and changes
will occur to those skilled in the art. For example, although the
invention is described with reference to specific device
structures, it is equally applicable to other vertical device
structures, including without limitation, Schottky devices,
junction barrier JBS Schottky devices, MPS and bipolar junction
transistors. Similarly, although many of the above-described
examples include a junction termination extension, a depletion
region and a field stop, the above described JTE designs are
equally applicable to semiconductor devices that do not include
field stops. It is, therefore, to be understood that the appended
claims are intended to cover all such modifications and changes as
fall within the true spirit of the invention.
* * * * *