U.S. patent application number 14/419608 was filed with the patent office on 2015-07-23 for apparatus and method for optically initiating collapse of a reverse biased p-type-n-type junction.
This patent application is currently assigned to Applied Physical Electronics, L.C.. The applicant listed for this patent is Applied Physical Electronics, L.C.. Invention is credited to William C. Nunnally.
Application Number | 20150207015 14/419608 |
Document ID | / |
Family ID | 50068679 |
Filed Date | 2015-07-23 |
United States Patent
Application |
20150207015 |
Kind Code |
A1 |
Nunnally; William C. |
July 23, 2015 |
Apparatus and method for optically initiating collapse of a reverse
biased P-type-N-type junction
Abstract
An optical method of collapsing the electric field of an
innovatively fabricated, reverse-biased PN junction causes a
semiconductor switch to transition from a current blocking mode to
a current conduction mode in a planar electron avalanche. This
switch structure and the method of optically initiating the switch
closure is applicable to conventional semiconductor switch
configurations that employ a reverse-biased PN junction, including,
but not limited to, thyristors, bipolar transistors, and insulated
gate bipolar transistors.
Inventors: |
Nunnally; William C.;
(Austin, TX) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Applied Physical Electronics, L.C. |
Austin |
TX |
US |
|
|
Assignee: |
Applied Physical Electronics,
L.C.
Austin
TX
|
Family ID: |
50068679 |
Appl. No.: |
14/419608 |
Filed: |
July 31, 2013 |
PCT Filed: |
July 31, 2013 |
PCT NO: |
PCT/US13/53005 |
371 Date: |
February 4, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61679686 |
Aug 4, 2012 |
|
|
|
Current U.S.
Class: |
250/214SW ;
257/77 |
Current CPC
Class: |
H01L 31/167 20130101;
H01L 31/1136 20130101; H01L 31/105 20130101; H01L 31/111 20130101;
H01L 31/1113 20130101; H01L 31/0288 20130101; H01L 31/1105
20130101 |
International
Class: |
H01L 31/105 20060101
H01L031/105; H01L 31/167 20060101 H01L031/167; H01L 31/0288
20060101 H01L031/0288 |
Claims
1. A semiconductor switch, comprising: a. a first layer comprising
a P-type semiconductor material; b. a second layer comprising a
semiconductor material doped with a mid-band dopant, wherein the
second layer is in contact with the first layer; c. a third layer
comprising an Intrinsic-type or slightly N-type semiconductor
material, wherein the third layer is in contact with the second
layer thereby forming a semiconductor junction interface between
the second layer and the third layer; and d. a fourth layer
comprising an N-type semiconductor material, wherein the fourth
layer is in contact with the third layer; wherein the semiconductor
switch is configurable in a reverse-bias circuit such that in a
first switch state the semiconductor junction interface is reverse
biased thereby blocking current flow through the circuit in the
absence of optical energy from an optical energy source, and
wherein optical energy applied to the switch from the optical
energy source produces a second switch state in which a sufficient
electrical charge is produced in the switch to cause an electric
field present at a depletion region in the switch to exceed a
breakdown level at the semiconductor junction interface, thereby
enabling a flow of current through the switch.
2. The semiconductor switch of claim 1, wherein an energy level of
the optical energy applied to the switch from the optical energy
source is less than a band-gap energy of a basic material forming
the switch, and further wherein the energy level of the optical
energy applied to the switch from the optical energy source is
sufficient to ionize the mid-band dopant.
3. The semiconductor switch of claim 2, wherein the basic material
comprises silicon carbide.
4. The semiconductor switch of claim 3, wherein the mid-band dopant
comprises vanadium or zinc or a combination of vanadium and
zinc.
5. The semiconductor switch of claim 2, wherein the fourth layer
and third layer are configured to allow at least some of the
optical energy from the optical energy source to pass through, and
the second layer is configured to absorb the optical energy passing
through the fourth and third layer.
6. The semiconductor switch of claim 2, further comprising a
metal-Ohmic contact forming an electrical connection between the
fourth layer and the reverse biasing circuit.
7. The semiconductor switch of claim 6, wherein the metal-Ohmic
contact comprises a contact opening, and wherein the metal-Ohmic
contact is positioned to receive optical energy from the optical
energy source through the contact opening.
8. An optically-activated switching apparatus, comprising: a. a
semiconductor device, comprising: i. a first layer comprising a
P-type semiconductor material; ii. a second layer comprising a
material doped with a mid-band dopant, wherein the second layer is
in contact with the first layer; iii. a third layer comprising an
Intrinsic-type or slightly N-type semiconductor material, wherein
the third layer is in contact with the second layer thereby forming
a semiconductor interface between the second layer and the third
layer; and iv. a fourth layer comprising an N-type semiconductor
material, wherein the fourth layer is in contact with the third
layer; b. a reverse biasing circuit electrically connected to the
semiconductor device to apply a reverse bias to the semiconductor
device; and c. an optical energy source configured to selectively
direct optical energy onto the semiconductor device.
9. The optically-activated switching apparatus of claim 8, wherein
the semiconductor device comprises silicon carbide.
10. The optically-activated switching apparatus of claim 9, wherein
the mid-band dopant comprises vanadium.
11. The optically-activated switching apparatus of claim 10,
wherein the photon energy of the optical energy source is at least
2.1 eV.
12. The optically-activated switching apparatus of claim 11,
wherein the optical energy source is a green light source.
13. The optically-activated switching apparatus of claim 12,
wherein the optical energy source comprises a frequency-doubled YAG
laser.
14. The optically-activated switching apparatus of claim 11,
wherein the optical energy source comprises a green laser
diode.
15. The optically-activated switching apparatus of claim 8, wherein
the optical energy source is configured to apply the optical energy
at the fourth layer.
16. A method for optically activating a semiconductor switch,
wherein the switch comprises a first layer comprising a P-type
semiconductor material, a second layer comprising a semiconductor
material doped with a mid-band dopant, a third layer comprising an
Intrinsic-type or slightly N-type semiconductor material, and a
fourth layer comprising an N-type semiconductor material, wherein
the method comprises the steps of: a. reverse biasing the switch by
applying a source potential comprising a positive lead and a
negative lead, wherein the positive lead is electrically connected
to the fourth layer and the negative lead is electrically connected
to the first layer; b. directing an optical energy source toward
the semiconductor switch; and c. activating the optical energy
source to apply sufficient optical energy to cause an electric
field at a depletion layer adjacent an interface within the
semiconductor switch to exceed a breakdown level, thereby enabling
a flow of current.
17. The method of claim 16, wherein the directing an optical energy
source toward the semiconductor switch step comprises the step of
directing the optical energy source toward the fourth layer of the
semiconductor switch.
18. The method of claim 16, wherein the activating the optical
energy source step comprises the application of photon energy of at
least 2.1 eV.
19. The method of claim 16, wherein the activating the optical
energy source step comprises the application of green optical
energy to the semiconductor switch.
Description
TECHNICAL FIELD
[0001] The present invention relates to electronic semiconductor
switches and, more particularly, to all semiconductor switches that
employ P-type:N-type junctions for blocking conduction, including
but not limited to diodes, all types of thyristors, insulated gate
bipolar transistors, and bipolar transistors.
BACKGROUND ART
[0002] Advanced power switching technologies are needed to enable
future defense and commercial power control capabilities and
concepts, but capabilities are presently limited by the
availability of precisely controllable, high voltage (10-100 kV),
high current (1-100 kA) switches. For example, presently available
semiconductor switches are limited in per device parameters on the
order of 1 kV and several tens of amps with switching times of
several hundred nanoseconds. Conventional high power semiconductor
switches rely on the blocking electric field produced by the charge
accumulation at the depletion region between P-type material and
N-type material which is termed a PN junction. In order to change
the switch impedance from a large value to a small value, the PN
junction electric field must be collapsed through electron
avalanche processes or flooded with sufficient electrical charge to
reduce the blocking electric field magnitude. Most semiconductor
switches including all types of Thyristors, Bipolar Transistors
(BJTs), Insulated Gate bipolar Transistors (IGBTs), and other
devices that employ a reverse biased PN junction to block current
flow in the switch "open" condition. Switching from the blocking
state to the conduction state is accomplished by injecting charge
in the reversed biased junction, by increasing the electric field
in the depletion region to exceed the dielectric strength of the
junction, and/or injecting avalanche seed electrons into the
electric field through capacitive coupling. These processes are
spatially and temporally dependent upon the mechanisms required to
inject charge, the time required to raise the electric field, or
the transverse spreading velocity of the conducting plasma across
the entire cross section of the device to reduce the switch
impedance. Therefore, most P-N junction-based switches employing
charge injection triggering result in closure times of hundreds of
nanoseconds to multiple microseconds that severely limit high
power, high frequency operation. More importantly, operating these
types of switches in circuits in which the current rise time is
much less than the impedance or voltage fall time of the switch
result in excessive power dissipation in the switch which further
relates to system efficiency and switch lifetime limitations.
[0003] The other approaches to high voltage, high power, high
frequency power switches are optically based. Specifically, the
three main optically controlled semiconductor switches are (1)
linear photo-conductive switches, (2) non-linear optically
initiated electron avalanche switches, and (3) optically gated PN
junction devices. The most common, linear photo-conductive switches
as illustrated in FIG. 1, uses photon energy greater than the
semiconductor band gap energy or above band photons from a source
(70) to uniformly illuminate a semiconductor (72) between
electrodes (73). In a linear photo-conductive switch, the photon
energy of the controlling optical source is greater than the
semiconductor band gap energy such that the energy is absorbed in
the absorption depth (75) of the semiconductor. Ideally, each
absorbed photon generates an electron (76) hole (77) pair to reduce
the semiconductor resistivity and transition the initial large
resistance that limits current flow to a small resistance that
permits conduction. The increase in the density of electrons and
holes between the switch electrodes reduces the switch resistance
in the time the optical energy is delivered to the switch, which
can be nanoseconds when a high power optical source such as a laser
is employed. The photo-conductivity produced by the optical source
decays as determined by the semiconductor recombination time such
that the switch will return to a high resistance state when the
incident optical energy is terminated. For example, the
recombination time in silicon can be as large as milliseconds while
the recombination time in gallium arsenide (GaAs) can be less than
nanoseconds. Alternatively, the conductivity can be maintained by
reducing the optical intensity to the value that compensates for
the loss of holes and electrons to recombination.
[0004] A second type of linear photo-conductive switch that employs
photon energy less than the semiconductor band gap energy or
sub-band photons, is illustrated in FIG. 2. In this embodiment, a
semiconductor (80) is sandwiched between two electrodes (81).
Sub-band photon energy is injected (83) at the edge of the
semiconductor. The sub-band photon energy is less than the band gap
energy, but sufficient to be absorbed and ionize mid-band dopants.
The energy absorbed by mid-band dopants in the semiconductor
produce electrons (85) and holes (86) to increase the conductivity
of the semiconductor (82) and change the switch impedance from
blocking to conduction. In the case of sub-band photons, the
effective optical absorption depth can be several cm, for example
in the case of silicon carbide, which enables most of the injected
optical energy to penetrate to the region between the electrodes
(84). A sufficient quantity of sub-band photons is injected and
absorbed in the region between the electrodes to increase the
semiconductor conductivity and change the switch resistance from a
large blocking resistance to an appropriate conduction resistance.
In this case, after the optical pulse terminates, the conductivity
also returns to the off state with the recombination time of the
host material.
[0005] Linear photo-conductive switches have demonstrated
capability to switch high voltages (10-100 kV) and conduct high
currents (1-20 kA) with precise temporal control (sub nanosecond).
However, linear photo-conductive switches are limited by the
requirement for a substantial laser system that makes them suited
only to laboratory systems or fast systems which cannot be provided
by other means. Thus linear photo-conductive switches are
applicable in high power, precise control or fast rise time pulse
generation systems.
[0006] The second optically controlled switch technology is based
on extensive work at Sandia National Laboratories in Albuquerque
(SNLA) in the development of optically initiated electron avalanche
switches in Gallium Arsenide (GaAs). The SNLA approach, illustrated
in FIG. 3, is commonly fabricated on the surface of a
semi-insulating GaAs wafer (103) by depositing dopants and metals
to form contacts (104) and (105). An above band optical source is
configured to inject tens of nano Joules of optical energy via
optical fiber or optical components at multiple sites (106) near
the cathode electrode (104) to initiate electron avalanche
streamers (107) that cross the switch to the anode electrode (105).
This GaAs approach is termed a non-linear or high gain approach
since the absorbed photons are employed only to initiate an
electron avalanche streamer that produce a much larger number of
electron-hole pairs through electron impact ionization or avalanche
and reduce the optical energy required by up to 4 orders of
magnitude when compared to the linear photo-conductive switch
approach. A unique feature of the non-liner photo switch in GaAs is
that conduction continues after the optical pulse is terminated.
This continued conduction is termed "lock on" mode. A major
limitation of the non-linear GaAs switch is that the current in
each conducting filament is limited to 20-50 amps with the lifetime
of the switch inversely related to the filament current. A second
limitation of multiple filament non-linear GaAs photo switches is
that the filaments initiated at multiple sites tend to coalesce
into one conduction path, also illustrated in FIG. 3. The coalesced
current density has shown to damage the contacts and the GaAs
substrate, which also limits the switch capability.
[0007] In order to increase the current capability of the
non-linear switches, much additional work has shown that the path
of the individual streamers (107) can be controlled by illuminating
lines (108) between the cathode electrode (104) and the anode
electrode (105) across the switch (108) as illustrated in FIG.
4.
[0008] A third optically controlled switch is the optically gated
PN junction switch, illustrated in FIG. 5, and an optically gated
Bipolar Junction Transistor, illustrated in FIG. 6, as in a reverse
biased Silicon Carbide (SiC) P-type, intrinsic, N-type (PIN)
diode.
[0009] The optically gated PIN diode of FIG. 5 may be fabricated by
depositing contacts (110) on the P-type SiC (112) formed on one
side of an intrinsic or slightly N-type SiC wafer (113). A heavily
doped n-type layer (114) is formed on the opposite side of the
wafer to interface to the negative contact (115). To change the
reverse biased PIN diode from blocking current flow to conducting
current, above band photons (111) (photon energy greater than the
band gap energy) are deposited in the p-type surface where the
energy is absorbed in much less than 1 mm. The photons absorbed in
the p-type layer (112) produce electron (116) hole (117) pairs. The
holes move toward the negative terminal (110) and the electrons
drift into the PIN diode intrinsic wafer (113) to provide seed
electrons for avalanche impact ionization and result in current
flow through the reverse biased PIN diode. The current conduction
continues while the optical energy is present and conduction ceases
after the optical energy is terminated and any residual electrons
have been swept out of the device. In the optically gated bipolar
junction transistor (BJT) of FIG. 6, above band photons (121) are
projected on the surface of the base region through a Silicon
Dioxide layer (122) and absorbed in a thin layer of the base (124).
The absorbed band photons produce electron-hole pairs (128) and
(129) in the base region that move in the electric field to the
emitter (123) and exit through contact (120) while the electrons
are injected into the base (124) collector (125) junction and
amplified by the BJT gain to exit to the collector contact
(126)/(127). This device is critically dependent upon the optical
energy being injected such that current is terminated after the
optically generated electrons are swept from the device.
Limitations of Existing Switch Technologies.
[0010] Conventional high-voltage semiconductor switches have
limited performance portfolios of the combination of operating
voltage, operating current, transition or switching time. For
example, thyristors can handle moderate voltages (several
kilovolts), very large currents (100s of kA), but turn on very
slowly (microseconds) while field effect transistors or FETs will
support moderate voltages (several kilovolts), turn on very fast
(ns) but handle only small currents (tens of amps). Major
applications require a switch that will handle large voltages (tens
of kV), high currents (several kAs), and transition or turn on
rapidly (ns) and operate at high average powers or switching rates.
More importantly, handling high power (voltage times current) in
high frequency circuits requires that the switch inductance be
small which further requires a compact switch. A semiconductor
switch that can provide all the necessary parameters, without
assembling a large array of switches, does not exist at the present
time.
[0011] The GaAs switches conduct through optically initiated,
electron avalanche current filaments in which the current is
limited to about 20-40 amps to prevent GaAs bulk material and
contact damage. This feature of non-nonlinear GaAs switches thus
requires a very large number of conducting filaments in order to
operate in the kA current range. Previous work has demonstrated the
ability to initiate multiple conducting filaments using multiple
optical fibers (FIG. 3) to initiate multiple triggering points to
increase the total switch current. However, multiple conducting
filaments, initiated near the cathode, tend to coalesce as the
avalanche streamers progress between the switch electrodes at a
depth of 50-100 microns. This accumulation of current filaments
also damages the contacts and bulk GaAs material. Further work has
demonstrated channeling or separation of the conducting filaments
using photo-conducting lines across the switch using additional
optical energy, as illustrated in FIG. 4. This optical boundary
generation, that requires a rather complicated optical arrangement,
has permitted the non-linear GaAs switches to be used at higher
total currents, but the total current is still less than required
for a number of applications, and the optical complexity of the
trigger system hinders wide usage. Furthermore, the total optical
energy required to control the multiple parallel filaments
approaches the quantity required to close a linear photo-conductive
switch.
[0012] One more feature of the GaAs switches is that the switches
maintain conduction or "lock-on" after the conducting filaments are
formed via electron avalanche streamers, much like gas discharges,
that persist until the driving voltage is removed and current
ceases. This is not the case for a linear photoconductive switch,
using GaAs or other semiconductors, which opens after the optical
pulse has terminated with the material recombination time or
several tens of ns to return to the multi-mega Ohm resistance. In
both types of photo conductive and photo-initiated conduction
switches, the conduction voltage is sufficiently large to limit the
application of these devices to pulse generation applications with
a limited duty cycle.
[0013] The optical injection of electrons into both the reverse
biased PIN diode, FIG. 5, and the BJT, FIG. 6, using above band
photons is similar to the linear photoconductive switch in that a
large quantity of optical energy is required to initiate and
sustain conduction. This shortcoming thus requires a large laser
which hinders wide application, and in the case of these devices,
requires a large UV laser wavelength which is hindered by the
transmission of UV photons through optical fibers.
SUMMARY OF INVENTION
[0014] In accordance with the present invention, there is provided
an optical method of initiating the collapse of the electric field
of an innovatively fabricated, reverse biased PN junction to cause
a semiconductor switch to transition from a current blocking mode
to a current conduction mode in a planar electron avalanche
fashion. This method of fabricating and optically initiating the
switch closure is applicable to conventional semiconductor switch
configurations that employ a reverse biased PN junction, including,
but not limited to, thyristors, bipolar transistors, and insulated
gate bipolar transistors. This invention is also directed to a
method of initiating (triggering) a planar, electron avalanche
closure of these types of semiconductor switches with a small
quantity of optical energy. In addition, the switch according to
the present invention closes or transitions in nanoseconds rather
than the tens to hundreds of nanoseconds closure of power
semiconductor switches.
BRIEF DESCRIPTION OF DRAWINGS
[0015] A complete understanding of the preferred embodiments of the
present invention may be obtained by reference to the accompanying
drawings, when considered in conjunction with the subsequent,
detailed description, in which:
[0016] FIG. 1 is a front perspective view of a linear photo
conductive switch.
[0017] FIG. 2 is a cross-section view of a second configuration of
a linear photo-conductive switch.
[0018] FIG. 3 is the basic non-linear or high gain photo-conductive
switch.
[0019] FIG. 4 is an improved non-linear or high gain
photo-conductive switch.
[0020] FIG. 5 is an optically gated PIN diode.
[0021] FIG. 6 is an optically gated BJT.
[0022] FIG. 7 is an illustration of a reverse biased PN junction in
a PIN diode configuration that illustrates the basic principle of
the present invention.
[0023] FIG. 8 is a graphical illustration of the energy band
structure of silicon carbide (SiC).
[0024] FIG. 9 is a diagram depicting the basic structure of a
preferred embodiment of the present invention.
[0025] FIG. 10 is a diagram of a 10 kV PIN diode switch,
semiconductor physics model indicating model dimensions, model
circuit parameters, materials and doping densities.
[0026] FIG. 11 is a diagram of the semiconductor physics model
waveform outputs for PIN diode voltage, PIN diode current density,
and optical input.
DESCRIPTION OF PREFERRED EMBODIMENTS
[0027] FIG. 7 is an illustration of the principle of the present
invention, showing a reverse biased PN junction in a PIN
configuration, where the P-type (180): intrinsic type (181): N-type
(182) structure is reverse biased near voltage breakdown by source
(184) in series with load resistor (183). The PN junction electric
field, which results in the electric field spatial distribution
shown in the plot (185) with a peak value that is less than the
breakdown electric field level (186), blocks current flow. In the
blocking mode electric field plot (185) the electric field (187) is
at a maximum near the P-Intrinsic material interface, but less than
the avalanche breakdown electric field value (186). The blocking
electric field is a result of the depletion region charge (188).
The object of this invention is to initiate an avalanche collapse
of the electric field produced by the charges in the reverse-biased
depletion region. Prior means of collapsing the blocking electric
field have employed an external electric source to raise the
reverse voltage on the PN junction to exceed the avalanche
breakdown electric field (186).
[0028] In the preferred embodiment of the present invention,
sub-band optical energy (190) is introduced into the structure to
produce electron-hole pairs (192) that move in the electric field
(191). The more mobile electrons leave the structure while the
slower holes add charge to the intrinsic side of the P-Intrinsic
junction. The increase in positive hole charge (193) induces
additional negative charge (194) to further increase the electric
field (191) to exceed the breakdown level (186) shown, and initiate
the collapse of the depletion region through electron avalanche.
Therefore, it may be seen that instead of applying a fast rising
voltage for the purpose of exceeding the breakdown voltage of a PN
junction, the preferred embodiment of the present invention changes
the electric field in the depletion region through absorbing
sub-band optical energy, near the reverse-biased P-N interface in
the structure.
[0029] A simple calculation of the additional charge required to
increase the PN junction electric field (187) to exceed the
breakdown electric field (186) can be used to estimate the
equivalent optical energy that is required to produce the electric
charge. TABLE 1 is a simple estimation of the optical energy
required to overvolt three PIN diodes. For example, to overvolt a
reverse biased, 10 kV PIN diode to 13 kV requires an optical energy
of less than 100 nJ per square cm, assuming unity quantum
efficiency. In the preferred embodiment, the common PIN structure
of FIG. 7 is modified to provide the mid-band dopant sites capable
of absorbing the sub-band optical energy. Specifically, an
additional layer is added to the PN junction interface to
preferentially absorb the sub-band optical energy. The new layer
can be added via an epitaxial deposition or implantation. The
energy level of the dopants in the new layer added to the common
PIN structure as part of the preferred embodiment must be mid-band
acceptors and/or donors. Therefore, the energy levels of various
dopants in the base material must be known.
TABLE-US-00001 TABLE 1 Calculation of Over Voltage Charge and
Optical Requirements Parameter Symbol 100 kV PAS 10 kV PAS 10 kV
PAS Unit Comments Reverse Bias Voltage 100 10 10 kV Over Voltage
100 10 3 kV Total Applied Voltage 200 20 13 kV PIN Structure Cross
Section Ac 1.00E-04 1.00E-04 1.00E-04 m2 Blocking Voltage Vb
1.00E+05 1.00E+04 1.00E+04 V SiC Dieletric Strength Ebd 3.00E+08
3.00E+08 3.00E+08 V/m Breakdown Thickness tbd 3.33E-04 3.33E-05
3.33E-05 m2 p+ doping NA 1.00E+25 1.00E+25 1.00E+25 m-3 n doping ND
7.00E+19 7.00E+19 7.00E+19 m-3 Semi-insulating SiC pn voltage Vo
3.00E+00 3.00E+00 3.00E+00 V Blocking Voltage Va 1.00E+05 1.00E+04
1.00E+04 V Bias SiC Dielectric Constant 9 9 9 Depletion Region
Blocking Voltage Wafer thickness 1.50E-03 3.50E-04 2.50E-04 m
Blocking Voltage 1.00E+05 1.00E+04 1.00E+04 V Depletion Region
Width - Blocking Wjb 1.19E-03 3.77E-04 3.77E-04 m xp 8.35E-09
2.64E-09 2.64E-09 m xn 1.19E-03 3.77E-04 3.77E-04 m Electric Field
At Interface Epn 8.39E+07 2.65E+07 2.65E+07 V/m Blocking E-field
SiC Dielectric Constant 9 9 9 Depletion Region - Blocking Cb
6.68E-12 2.11E-11 2.11E-11 F for 1 sq cm area Blocking charge Qb
6.68E-07 2.11E-07 2.11E-07 C Bias Overvoltage Depletion Region
Overvoltage DV 1.00E+05 1.00E+04 3.00E+03 V Total Applied Voltage
2.00E+05 2.00E+04 1.30E+04 V Risetime Tr 1.00E-09 1.00E-09 1.00E-09
s Voltage change rate dV/dt 1.00E+14 1.00E+13 3.00E+12 V/s 2X over
voltage depletion region Wjov 1.69E-03 5.33E-04 4.30E-04 m
Avalanche Electric Field E 1.19E+08 3.75E+07 3.02E+07 V/m ce
Depletion Region - Overvoltage Cov 4.72E-12 1.49E-11 1.85E-11 C 2X
Overvoltage Charge Qa 9.44E-07 2.99E-07 2.41E-07 C Delta Charge
Qa-Qb 2.77E-07 8.74E-08 2.96E-08 C net charge required Wavelength
5.32E-07 5.32E-07 5.32E-07 m doubled YAG Frequency 5.64E+14
5.64E+14 5.64E+14 Hz Photon Energy 3.74E-19 3.74E-19 3.74E-19 J
Number Electron-hole pairs 1.73E+12 5.47E+11 1.85E+11 ea holes
provide charge rgy to produce same delta charge 6.46E-07 2.04E-07
6.91E-08 J 1 electron-hole/photon 646.26 204.30 69.09 nJ per square
cm Number of Photons 1.73E+21 5.47E+20 1.85E+20 ea indicates data
missing or illegible when filed
[0030] FIG. 8 is an illustration of the energy band structure of
silicon carbide (SiC). The SiC band gap between the conduction band
energy (200) and the valence band energy (201) determines the band
gap energy (202). The photon energy required to ionize the mid-band
dopants (206) must be less than the band gap energy (202) as
illustrated in FIG. 8. Acceptors with energy levels near the
valence band and donors with energy levels near the conduction band
are ionized at room temperature Thus dopants with mid-band energy
levels, such as Vanadium and/or Zinc, near the middle of the SiC
band gap provide absorption sites for the sub-band optical energy
with photon energy sufficient photon energy (206). The relationship
of the mid-band dopant energy (205) and the conduction band energy
(200) determine the photon energy (206) required to ionize the
mid-band dopants. The selection of vanadium as the target
absorption requires that the photon energy (206) be greater than
the energy required to ionize the vanadium, which is the difference
between the conduction band energy (203) and the vanadium acceptor
energy (205). For example, the Vanadium energy level is about 1.1
eV and the 4H SiC band gap energy is about 3.2 eV such that the
sub-band photon energy should be greater than 2.1 eV. The photon
energy of various optical sources, calculated in TABLE 2, indicates
a green optical source such as doubled YAG laser or a green
laser-diode would be sufficient.
TABLE-US-00002 TABLE 2 Photon Wavelength Energy Source Nd: YAG 2
.times. Nd: YAG 3 .times. Nd: YAG Nitrogen Unit Photon Wavelength
1064 532 355 337 nm Photon Frequency 2.82E+14 5.64E+14 8.45E+14
8.90E+14 Hz Photon Energy 1.87E-19 3.74E-19 5.60E-19 5.90E-19 J
1.17 2.34 3.50 3.69 eV
[0031] FIG. 9 is an illustration of the basic structure of a
preferred embodiment of the present invention according to the
principles explained above. Specifically, a PN-type junction with
an additional layer at the PN-junction interface, in a PIN
configuration is reverse biased by source voltage (228) in series
with load resistor (229). The metal-Ohmic contact (220) is
connected to the load resistance (228) and the voltage source (229)
that reverse biased the PN junction. The PIN structure consists of
a P-type layer (227) that interfaces to a critical part of this
preferred embodiment, a mid-band dopant layer (226). The mid-band
dopant layer is formed from an Intrinsic-type material or slightly
N-type material doped with a mid-band dopant, preferably including
vanadium if a SiC semiconductor implementation is used. This
mid-band dopant layer is followed by the Intrinsic layer or
slightly N-type layer (222), then the heavily doped N-type layer
(224), and the N-type metal-Ohmic contact (220). The reverse biased
junction results in a blocking electric field (234), with peak
value (23) less than the breakdown electric field (231) in the
depletion region (221). The Sub-band optical energy (223) is
injected into the structure (through the N-type contact is one
option) and is absorbed preferentially in the mid-band dopant layer
(226), to produce electrons and holes. The additional holes
increase the electric field (235) to exceed the avalanche threshold
electric field (232), which results in avalanche breakdown of the
SiC in the region (233) near the PN interface. This highly
conducting, plasma region then expands to move toward the N-type
(224) contact while compressing and increasing the applied electric
field to speed the avalanche process in a regenerative manner. As
the conducting region reaches the N-type contact, the blocking
field and blocking voltage rapidly disappears to allow current to
flow through the conducting avalanche plasma.
[0032] FIG. 10 is a diagram of the preferred embodiment of a 10 kV
PIN diode, semiconductor physics model with the densities,
dimensions, and circuit parameters used to generate the operational
waveforms shown in FIG. 11.
[0033] FIG. 11 is a set of plots of the SiC PIN diode voltage, PIN
diode current density, and optical input power for the
semiconductor model of FIG. 10 in which the optical power was 0.5
W/cm.sup.2 and the initial reverse bias was 9500 V. The avalanche
gain in the semiconductor model was adjusted to match reverse
self-breakdown voltage of a known SiC PIN diode to calibrate the
model.
[0034] It will be understood that semiconductor materials other
than SiC may be used in the implementation of the invention in
various alternative embodiments. Such materials include, without
limitation, silicon, gallium arsenide, gallium nitride, and
aluminum nitride.
[0035] Since other modifications and changes varied to fit
particular operating requirements and environments will be apparent
to those skilled in the art, the invention is not considered
limited to the example chosen for purposes of disclosure, and
covers all changes and modifications which do not constitute
departures from the true spirit and scope of this invention.
[0036] Having thus described the invention, what is desired to be
protected by Letters Patent is presented in the subsequently
appended claims.
* * * * *