Large-band-gap SIC, Ill-V nitride, and II-VI ZnSe-based semiconductor
device technologies
H. MorkoG, S. Strite,a) G. B. Gao, M. E. Lin, B. Sverdlov, and M. Burns
University of Illinois, Materials Research Laboratory and Coodinated Science Laboratory
104 South Goodwin Avenue, Urbana, Illinois 61801
(Received 13 January 1993; accepted for publication 17 March 1994)
‘-
In the past several years, research in each of the wide-band-gap semiconductors, Sic,. GaN, and a
ZnSe, has led to major advances which now make them viable for device applications. The merits
of each contender for high-temperature electronics and short-wavelength optical applications are
compared. The outstanding thermal and chemical stability of Sic and GaNshould enable them to
operate ai high temperatures and in hostile environments, and ‘also make them attractive for
high-power operation. The present advanced stage of development of Sic substrates and ’
metal-oxide-semiconductor technology makes Sic the leading contender for high-temperature and
high-power applications if ohmic contacts and interface-state densities can be further improved.
GaN, despite fundamentally ,-superior electronic properties and better ohmic contact resistances,
must overcome the lack of an ideal substrate material and a relatively advanced Sic infrastructure
in order to compete in electronics applications. Prototype transistors have been fabricated from both
Sic and GaN, and the microwave characteristics and high-temperature performance of Sic
transistors. have been studied. For optical emitters and detectors, ZnSe, Sic, and GaN all have c.
demonstrated operation in the green, blue, or ultraviolet (UV) spectra. Blue Sic light-emitting ” ~~
diodes (LEDs) have been on the market fo;-several years, joined recently by UV and blue
GaN-based LEDs. These products should find wide use in full color display and other technologies.
Promising prototype UV photodetectors have been fabricated from both Sic and GaN. In laser
development, ZnSe leads the way with more sophisticated designs -having further improved
performance being rapidly demonstrated. If the low damage threshold of Z&e continues to limit
practical laser applications, GaN appears poised to become the semiconductor of choice for
short-wavelength lasers in optical memory and other applications. For further development of these
materials to be realized, doping’densities (especially p type) and-ohmic contact technologies have
to be improved. Economies of scale need to be realized through the development of larger Sic
substrates. Improved substrate materials, ideally GaN itself, need tb be aggressively pursued to
furthe; develop the GaN-based material system-and enable the fabrication of lasers. ZnSe material
quality is already outstanding and now researchers must focus their attention on addressing the short
lifetimes of ZnSe-based lasers to determine whether the material is sufficiently durable for practical
laser applications. The problems related to these three wide-band-gap semiconductor systems have
moved away from materials science toward the device arena, where their technological developmtint
can rapidly be brought to maturity.
TABLE OF CONTENTS
I. Introduction:. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1364
II. Sic....................................... 1364
J. Sic bipolar transistors. . . . . . . . . . . . . . . . . . . . . 1374
K. Sic light emitting diodes. . . . . . . . . . . . . , . . . . . 1375
L. Sic photodiodes. . . . . . . . . . . . . . . . . . . . . . . . . . 1376
M. Sic ‘rectifiers. . . . . . . . . . . . . . . . . . . . . . . . . . . .’ . 1377
A.
B.
C.
D.
E.
F.
G.
H.
I.
Sic for high power, high temperature N. Remaining issues in Sic and summary ........ 1377
electronics. .............................. 136-q III. GaN based III-V nitride semidonductors. ....... 1379
A.
B.
C.
D.;
Polytypisti in SIC. ....................... 1365
Sic substrate crystal growth. ............... 1367
Sic thin film epitaxy. .. : .................. 1367
Dopant considerations. .......... - ......... 1368
Oxidation of Sic. ......................... 1368
Sic etching. ............ :. ............... 1369
Ohmic contacts to Sic. .................... 1370
Sic field effect transistors. ................. 1371
1. Sic power MOSFETs. .................. 1371
2. Sic JFETs ............................ 1372.
3. Sic MESFETs ........................ 1373
E.
F.
G.
H.
1.
J.
‘IPresent address: BM Research Division, Ziirich Research Laboratory,
Siumerstrasse 4, CH:8803 Riischlikon, Switzerland.
K.
L.
Fundamental properties of G’aN, AlN, and InN 1380
Nitride crystal growth. .. .- ................. 1380
Substrates for nitride epitaxy. ............... -1381
Buffer layers for nitride heteroepitaxy on
sapphire. ... .I ........................... 1382
Polytypism in the III-V nitrides. .............. 1382
Electrical properties of undoped nitride thin
films ................................... 1382
Properties of doped GaN. .................. 1382
Thermal stability and etching of.GaN. ........ 1384
Ohmic contacts to GaN: ................... 1384
Properties of nitride alloys. ......... t ....... 1385
GaN based LEDs. ..................... . .. F 1386
GaN FETs. .............................. 1388
J. Appl. Phys. 76 (3), 1 August 1994 0021-8979/94/76(3)/i363/36/$6.00 0 1994 American institute of Physics 1383
Downloaded 24 Jun 2010 to 159.226.35.205. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
Administrator
高亮
Administrator
高亮
Administrator
下划线
M.TowardsaGaNlaser.. ..... - ..;. .... /. ..... 1389:
N. Negative electron affinity devices. ........... 1389
IV ZnSe based II-VI semiconductor lasers .......... 1389
A. ZnSe crystal growth ....................... 1390
B. Doping ................................. 1390
C. Ohmic contacts ........................... 1390
D. Laser structures. .......................... 1391
E. Outlook for ZnSe as a laser material. ........ 1394
. Summary and discussion. ..................... 1394
I. INTRODUCTION
The recent surge of activity in wide-band-gap semicon-
ductors has arisen from the need for electronic devices ca-
pable of operation at high power levels, high temperatures,
and caustic environments, and separately, a need for optical
materials, especially emitters, which are active in the blue
and ultraviolet (UV) wavelengths. Electronics based on the
existing semiconductor device technologies of Si and GaAs
cannot tolerate greatly elevated temperatures or chemically
hostile environments due to the uncontrolled generation of
intrinsic carriers and their low resistance to caustic chemi-
cals. The wide-band-gap semiconductors SIC and GaN, and
perhaps sometime in the future, diamond, with their excel-
lent thermal conductivities, large breakdown fields, and re-
sistance to chemical attack, will be the materials of choice
for these applications. In the optical device arena, the ever-
increasing need for higher-density optical storage and full
color display technologies are driving researchers to develop
wide-band-gap semiconductor emitter technologies which
are capable of shorter-wavelength operation.
Much progress has been made in Sic for high-
temperature and high-power devices applications due to the
availability of high-quality Sic substrates, advances in
chemical-vapor-deposition (CVD) growth of epitaxial struc-
tures, and the ability to easily dope the material both n and p
type. The large Si-C bonding energy makes Sic resistant to
chemical attack and radiation, and ensures its stability at
high temperatures. In addition, Sic has a large avalanche
breakdown field, excellent thermal conductivity, and a high
electron saturation velocity, all of which make it ideal for
high-power operation. Metal-semiconductor and metal-
oxide-semiconductor transistors with outstanding high-
temperature performance have already been demonstrated.
With recent advancements in the growth of the 3C and 4H
polytypes, which have higher electron mobilities than the 6H
polytype, Sic is certain to attract continued attention for
high-power electronics applications.
sensors would reduce complexity and increase reliability.’
Hydraulic systems, a fire hazard in aircraft, and heat radia-
tors in satellites could then be greatly reduced in size and
number yielding in considerable weight reductions.
In the field of optical devices, several trends are pushing
research into new materials. The ever-increasing need for
denser optical storage media is driving the development of
shorter-wavelength semiconductor laser technologies be-
cause the diffraction-limited optical storage density increases
quadratically as the probe laser wavelength is reduced. To-
wards this end, yellow lasers based on InG&AlP heterostruc-
hues have been successfully demonstrated; however, this
material system is limited to 650 nm. Toward still shorter
wavelengths, the recently demonstrated ZnSe-based laser
technology is capable of operation in the green and blue
wavelengths, and GaN can potentially lase in the UV.
Wide-band-gap emitters are also bringing semiconductor
technology to full color displays. For the first time, all three
primary colors can be generated using semiconductor tech-
nology, which promises to allow the reliability, compactness,
and other desirable attributes of semiconductors to be ap-
plied to this important technological market.
In this review, we concentrate on device-oriented re-
search and applications of Sic, GaN, and ZnSe. Section II
covers Sic and describes recent progress in Sic substrates,
epitaxy, processing, and the devices to which these advances
have led. Theoretical work detailing the high-frequency and’
high-power performance of Sic and other wide-band-gap
semiconductors is also described. Section III examines GaN
along with its AlN and InN alloys, covering advances in
epitaxial techniques, doping and device technology. ZnSe is
reviewed in Sec. IV, with a primary focus on p-type doping
with activated nitrogen and the design and fabrication of
lasers.
Il. SIC
Sic was one of the first semiconductors discovered and
its large cohesive energy caused some to mistake it for an
element. Its high electron saturation velocity, wide band gap,
and high thermal conductivity, among other properties, make
it highly attractive for high-temperature and high-power ap-
plications. Despite its indirect band gap, Sic blue LEDs are
commercially available. Below, we describe the material
properties, crystal growth, doping, oxidation, etching, ohmic
contacts, and device technology of Sic.
Industries such as the aerospace, automotive, petroleum,
and others have continuously provided the impetus pushing
the development of fringe technologies which are tolerant of
increasingly high temperatures and hostile environments.
Sic and the III-V nitride devices will be capable of improved
high-power and -temperature operation due to their large
band gaps. GaN may prove superior since it has lower ohmic
contact resistances and is predicted to have larger electron
saturation velocities. A suitable high-temperature semicon-
ductor technology could allow bulky aircraft hydraulics and
mechanical control systems to be replaced with heat-tolerant
in situ control electronics. On-site electronics, actuators, and
A. Sic for high-power, high-temperature electronics
An appreciation of the potential of Sic for electronic
applications can be gained by examining Table I which com-
pares the relevant material properties of Sic and GaN with
Si and GaAs, the two most popular semiconductor device
technologies, and GaP and diamond, two other contenders
for high-temperature applications. Most notable are the large
thermal conductivities, breakdown voltages, and saturation
velocities of Sic, GaN, and diamond. The device maximum
operating temperature parameter is calculated as the tem-
perature at which the intrinsic carrier concentration equals
5XlOl’ cme3 and is intended as a rough estimate of the
1364 J. Appl. Phys., Vol. 76, No. 3, 1 August 1994 Morkop et a/.
Downloaded 24 Jun 2010 to 159.226.35.205. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
Administrator
下划线
Administrator
打字机
光学存储密度随着
波长的减小二次方增加
TABLE? I. Comparison of important semiconductor properties for high-temperature electronics.
Properly Si GaAs GaP
3C Sic
(6H SIC) Diamond GaN
Band gap (eV)
at 300 K
1.1
Maximum operating
temperature (K)
. --
Melting point (K)
Physical stability Good
Electron mobility
RT, cm’/V s
Hole mobility
RT, cm2/v s
Breakdown voltage
Eb , lO’/V/cm
Thermal conductivity
cr. W/cm
Sat. C . elec. drift vel.
r&at), 10’ cm/s
Dielectric const. K
600?
1690
1400
600
0.3
1.5
1
11.8
1.4
760?
1510
Fair
8500
400
0.4
2.3
1250?
1740
Fair
350
100
0.5 0.8
2
12.8 11;l
2.2
(2.9)
1200
(1580)
Sublimes
>2100
Excellent
1000
(600)
40
4
5
2
9.7
5.5 3.39
1400(?)
Phase
change
Very good
2200
Good
900
1600 lSO?
10 51
_.
20 1.3
2.7 2.7
5.5 9
band-gap iimitation on device operation. More important for
the eventual maximum operating temperature is the physical
stability of the material.
Figure 1 shows the electron velocity field relationship
for Si, GaAs, InP, and SiC.2 The large saturation velocities in
Sic are attributable to its larger optical phonon energies. It is
clear that Sic can also function well at much higher electric
fields than the conventional semiconductors. In Sec. III we
discuss theoretical predictions of an even larger electron
saturation velocity in GaN.
108
-i 10’
ic......................
.#
4
:
25
106
- ld 104 16 106 10'
Electric Field, V/cm
FIG. 1. Velocity-field relationships of Si, GaAs, InP, 3C and 6H SIC illustat-
ing the superior high-field electron drift velocities and breakdown field of
SiC. 3C Sic has a higher saturation velocity than 6H Sic due to reduced
phonon scattering.
J. Appl. Phys., Vol. 76, No. 3, 1 August 1994 Morkoq eta/. 1365
B. Polytypism in Sic
‘Sic is the most prominent of a family of close-packed
materials that exhibit a one-dimensional polymorphism
called polytypism. The Sic polytypes are differentiated by
the stacking sequence of the tetrahedrally bonded Si-C bilay-
ers, such that the individual bond lengths and local atomic
environments are nearly identical, while the overall symme-
try of the crystal is determined by the stacking periodicity.
Ashorthand has been developed to catalogue the literally
infinite number’of possible polytype crystal structures. Each
Sic bilayer, while maintaining the tetrahedral bonding
scheme of the crystal, can be situated in one of three possible
positions with respect to the lattice. These are each arbitrarily
assigned the notation A, B, or C. Depending on the stacking
order, the bonding between Si and C atoms in adjacent bi-
layer planes is either of a zinc-blende (cubic) or wurtzite
(hexagonal) nature. Zinc-blende bonds are rotated 60” with
respect to nearest neighbors while hexagonal bonds are mir-
ror images (Fig. 2). Each type of bond provides a slightly
altered atomic environment making some lattice sites in-
equivalent in polytypes with mixed bonding schemes and
reducing the overall crystal symmetry. These effects are im-
portant when considering the substitutional impurity incor-
poration and electronic transport properties of Sic.
If the stacking is ABCABC..., the purely cubic zinc-
blende structure, commonly abbreviated as 3C Sic (or beta
Sic) is realized (Fig. 3). The number 3 refers to the three
bilayer periodicity of the stacking and the letter C denotes
the overall cubic symmetry of the crystal. 3C Sic is the only
possible cubic polytype. The purely wurtzite ABAB... stack-
Downloaded 24 Jun 2010 to 159.226.35.205. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
Administrator
高亮
Zincblende
k------c Wurtzite
FIG. 2. Zinc-blende and wurtzite bonding between Si and C atoms in adja-
cent planes. The three tetrahedral bonds are 60” rotated in the cubic case and
mirror images in the hexagonal case.
ABCABC
ing sequence is denoted as 2H SIC reflecting its two bilayer
stacking periodicity and hexagonal symmetry. All of the
other polytypes are mixtures of the fundamental zinc-blende
and wurtzite bonds. Some common hexagonal polytypes
with more complex stacking sequences are 4H and 6H Sic
(Figs. 4 and 5). 4H Sic is composed equally of cubic and
hexagonal bonds, while 6H SIC is two-thirds cubic. Despite
the cubic elements, each has overall hexagonal crystal sym-
metry. The family of hexagonal polytypes is collectively re-
ferred to as alpha Sic. Rhombehedral structures such as 15R
and 21R have also been documented.3
The different polytypes have widely ranging physical
properties. 3C Sic has the highest electron mobility and
saturation velocity because of reduced phonon scattering re:
sulting from the higher symmetry. The band gaps. differ
widely among the polytypes ranging from 2.3 eV for 3C Sic
to 2.9 eV in 6H Sic to 3.3 eV for 2H Sic. In general, the
greater the wurtzite component, the larger the band gap.
Among the Sic polytypes, 6H is most easily prepared and
best’studjed, while the 3C and 4H polytypes are attracting
--
: more attention for their superior electronic properties. The
polytypism of SiC makes it nontrivial to grow single-phase
material, but it also offers some potential advantages- if
crystal-growth methods can be developed sufficiently to
capitalize on the possibility of polytype homo/hetero-
junctions. Figure 6 shows one such possibility, a 6H/3C Sic
interface. Such a junction incorporates the advantages of hec
c
B
A
C
B
A
A
FIG. 3. Crystal structure of the purely cubic 3C Sic polytype: Each lattice
site (k, representing cubic symmetry) is’equivalent.
1366 J. Appl. Phys., Vol. 76, No. 3, 1 August 1994 Morkop et al.
V
ABCTBCA
FIG. 4. Crystal structure of the 4H SIC polytype. Half of the atomic sites are
hexagonally bonded (h) while half are cubic (k).
erojunction band offsets while maintaining a completely
nonpolar, lattice matched, and coherent interface.
L. Powell et aL4 have shown that 3C Sic can be grown on
well-oriented basal plane (0001) 6H substrates. In this case,
the heteroepitaxy is terrace controlled, and the adatoms are
free to choose the most energetically favorable stacking se-
quence’in the direction perpendicular to the substrate surface.
Proper control of the growth conditions gives rise to 3C Sic
heteroepitaxy. When the substrate is miscut several degrees
away from the basal plane toward (1120) .! steps ‘are prevalent
1.. _. __.
.~ . .
7
B
C
A
C
B
-A
B “.-
c
A’
C
B
A-
ABCABCA
FIG. 5. Crystal ~structure of 6H SiCpolytype, the most prevalent polytype.
The lattice is two thirds cubic (k, and ka) and one third hexagonal (h t). The
two cubic sites are inequivalent and are expected fo have slightly dierent
binding energies for substitutional impurities.
Downloaded 24 Jun 2010 to 159.226.35.205. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
Administrator
高亮
Administrator
高亮
Administrator
下划线
f
6H Sic
(0001)
0 :Si
o-c
.- Heterointc
I
*-
3C Sic
(111)
:rface
.
FIG. 6. Schematic of a 3U6H SIC polytype heterojunction. When the plane
of the interface is normal to the stacking direction, a coherent, lattice-
matched heterojunction is possible.
on the growth surface. If the step spacing is: less than the
adatom surface diffusion length, the growth is step con-
trolled, and the