U.S. patent application number 11/962259 was filed with the patent office on 2008-05-08 for silicon nitride passivation with ammonia plasma pretreament for improving reliability of algan/gan hemts.
Invention is credited to Steven C. Binari, Andrew P. Edwards, Jeffrey A. Mittereder.
Application Number | 20080105880 11/962259 |
Document ID | / |
Family ID | 38138404 |
Filed Date | 2008-05-08 |
United States Patent
Application |
20080105880 |
Kind Code |
A1 |
Edwards; Andrew P. ; et
al. |
May 8, 2008 |
SILICON NITRIDE PASSIVATION WITH AMMONIA PLASMA PRETREAMENT FOR
IMPROVING RELIABILITY OF AlGaN/GaN HEMTs
Abstract
This invention pertains to an electronic device and to a method
for making it. The device is a heterojunction transistor,
particularly a high electron mobility transistor, characterized by
presence of a 2 DEG channel. Transistors of this invention contain
an AlGaN barrier and a GaN buffer, with the channel disposed, when
present, at the interface of the barrier and the buffer. Surface
treated with ammonia plasma resembles untreated surface. The method
pertains to treatment of the device with ammonia plasma prior to
passivation to extend reliability of the device beyond a period of
time on the order of 300 hours of operation, the device typically
being a 2 DEG AlGaN/GaN high electron mobility transistor with
essentially no gate lag and with essentially no rf power output
degradation.
Inventors: |
Edwards; Andrew P.; (Cary,
NC) ; Binari; Steven C.; (Annandale, VA) ;
Mittereder; Jeffrey A.; (Annandale, VA) |
Correspondence
Address: |
NAVAL RESEARCH LABORATORY;ASSOCIATE COUNSEL (PATENTS)
CODE 1008.2
4555 OVERLOOK AVENUE, S.W.
WASHINGTON
DC
20375-5320
US
|
Family ID: |
38138404 |
Appl. No.: |
11/962259 |
Filed: |
December 21, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11311592 |
Dec 9, 2005 |
7338826 |
|
|
11962259 |
Dec 21, 2007 |
|
|
|
Current U.S.
Class: |
257/76 ;
257/E21.407; 257/E29.246; 257/E29.253 |
Current CPC
Class: |
H01L 29/66462 20130101;
H01L 29/7787 20130101; H01L 29/2003 20130101 |
Class at
Publication: |
257/076 ;
257/E29.246 |
International
Class: |
H01L 29/778 20060101
H01L029/778 |
Claims
1. A high electron mobility transistor, comprising: a substrate; a
GaN buffer layer disposed on the substrate; an AlGaN barrier layer
disposed on the GaN buffer layer such that a 2 DEG channel is
disposed at the interface of the AlGaN barrier layer and GaN buffer
layer; and a passivation layer disposed on the AlGaN barrier layer,
wherein an upper surface of the AlGaN barrier layer has been
pre-treated with an ammonia plasma prior to a deposition of the
passivation layer.
2. The transistor of claim 1, wherein the transistor has
essentially no measurable drain current collapse for up to about
176 hours of dc bias stress.
3. The transistor of claim 1, wherein the transistor has
essentially no measurable gate lag for up to about 176 hours of dc
bias stress.
4. The transistor of claim 2, wherein the dc bias stress is
characterized by a drain current of 100-300 mA/mm and a drain
voltage of about 23-30 volts.
5. The transistor of claim 2, wherein the dc bias stress is
provided by a microwave signal having an input power P.sub.in of
5-20 dBm at a frequency of 2-12 GHz.
6. The transistor of claim 1, wherein the AlGaN barrier layer is
about 150 A to 300 A thick.
7. The transistor of claim 1, wherein the GaN buffer layer is about
5000 .mu.m to 15,000 .mu.m thick.
8. The transistor of claim 1, wherein the transistor has
experienced essentially no decrease in rf power output (P.sub.out)
up to about 70 hours.
9. The transistor of claim 1, wherein the AlGaN barrier layer is
made from Al.sub.xGa.sub.1-xN, wherein
0.15.ltoreq.x.ltoreq.0.40.
10. The transistor of claim 1, wherein the rate of degradation of
microwave power output, while under continuous microwave operation,
is at least 100 times smaller than a conventional high electron
mobility transistor that is not pre-treated with ammonia
plasma.
11. The transistor of claim 1, wherein the transistor has electron
mobility in excess of 1000 cm.sup.2/V sec.
12. The transistor of claim 1, further comprising a AlN buffer
layer disposed between the GaN buffer layer and the substrate.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a divisional application of U.S.
application Ser. No. 11/311,592, filed Dec. 9, 2005.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] This invention pertains to improving reliability of
heterojunction transistors, particularly high electron mobility
transistors, with an ammonia plasma pretreatment prior to
passivation.
[0004] 2. Description of Related Art
[0005] Prior methods which have been used to prepare processed
AlGaN/GaN high electron mobility transistors (HEMTs) for reliable
operation in the past have included using no passivation at all, or
direct deposition of a variety of electrically insulating materials
intended to passivate surface states. These material films can be
deposited by such processes as plasma enhanced chemical vapor
deposition. Silicon nitride (SiN) is one of the most commonly used
surface passivating films for AlGaN/GaN HEMTs and can be deposited
onto the device by the aforementioned method. However, silicon
nitride passivation, while an improvement over no passivation at
all, still results in a decrease in performance of the device after
dc and rf bias stress. This is a significant limitation and
disadvantage in the reliability of AlGaN/GaN HEMT electronic
devices.
[0006] AlGaN/GaN high electron mobility transistors have shown
exceptional microwave power output densities, with a recently
reported continuous wave power density of 30 W/mm and 50% power
added efficiency at a frequency of 8 GHz, In addition, a 36-mm
gate-width GaN HEMT has been demonstrated with a total power output
of 150 W and a power added efficiency of 54%. However, device
reliability remains a major concern for III-N HEMTs. In AlGaN/GaN
HEMTs, degradation of the dc, transient, and microwave
characteristics are often seen after relatively short periods of
normal device operation. Although reliability is improving,
microwave power output typically degrades by more than 1 dB in less
than 1000 hours of operation.
OBJECTS AND BRIEF SUMMARY OF THE INVENTION
[0007] It is an object of this invention to pretreat an AlGaN/GaN
heterojunction field effect transistor with ammonia plasma prior to
passivation in order to improve reliability thereof.
[0008] Another object of this invention is to pretreat an AlGaN/GaN
transistor with a low-power ammonia plasma prior to silicon nitride
deposition for high power applications.
[0009] Another object of this invention is to improve reliability
of an AlGaN/Gan transistor adapted for use in radars and
communication equipment.
[0010] Another object of this invention is to improve reliability
of AlGaN/GaN HEMTS characterized by the presence of 2 DEG
channel.
[0011] Another object of this invention is prevention of drain
current collapse caused by electron traps.
[0012] Another object of this invention relates to reduction of
surface and bulk traps in high power and high electron mobility
AlGaN/GaN heterojunction transistors that can potentially operate
at voltages exceeding 100 volts and with electron mobility in
excess of 1000 cm.sup.2/v sec.
[0013] These and other objects of this invention can be attained by
pretreatment of a 2 DEG AlGaN/GaN heterojunction transistor with a
low-power ammonia plasma prior to passivation.
BRIEF DESCRIPTION OF THE DRAWINGS
[0014] FIG. 1 is a cross-sectional schematic representation of a
high mobility transistor with AlGaN, GaN and doped GaN layers
deposited on a silicon carbide substrate.
[0015] FIGS. 2 (a) and (b) show induced current collapse of
unpassivated HEMT in (a) before stress and after stress and in (b)
of passivated HEMT devices with silicon nitride only and with
silicon nitride passivation, stressed for 60 hours and 176 hours,
respectively.
[0016] FIG. 3 is a plot of normalized Drain Current versus Time,
showing drain current response to pulsed gate voltage, in (a) with
no passivation; in (b) for the same device after 64-hour stress; in
(c) silicon nitride passivated device after 80-hour stress; and in
(d) a device pretreated with ammonia plasma and passivated with
silicon nitride, after 176-hour stress.
[0017] FIG. 4 is a plot of power _P.sub.out versus Time showing rf
output degradation of (a) passivated device with silicon nitride
only and (b) pre-treated with ammonia plasma and coated or
pre-treated with silicon nitride wherein stress conditions were
20v, 200 mA/mm operated at 2 GHz at 1 dB compression of the
gain.
DETAILED DESCRIPTION OF THE INVENTION
[0018] The purpose of this invention, in a preferred embodiment, is
to improve the reliability of aluminum gallium nitride/gallium
nitride (AlGaN/GaN) high electron mobility transistors by
incorporating an ammonia (NH.sub.3) plasma pre-treatment prior to
silicon nitride (SiN) passivation of the heterojunction transistors
after all other processing has been completed.
[0019] This invention pertains to an electronic device and to a
method for making it. The device is a heterojunction transistor,
particularly a high electron mobility transistor characterized by
the presence of a 2 DEG channel. Transistors of this invention
contain an AlGaN barrier and a GaN buffer, with the channel
disposed, when present, at the interface of the barrier and the
buffer. The method pertains to treatment of the device with ammonia
plasma prior to passivation to extend reliability of the device
beyond a period of time on the order of 300 hours of operation, the
device typically being a 2 DEG AlGaN/GaN high electron mobility
transistor with essentially no gate lag and with essentially no rf
power output degradation.
[0020] Pursuant to one embodiment of the method, an in-situ ammonia
plasma treatment is used before a silicon nitride deposition on an
Al.sub.x Ga.sub.1-xN/GaN (where x is 0.20 to 0.30), after all other
processing has been completed. The ammonia plasma pretreatment and
the silicon nitride deposition can both be performed in a plasma
enhanced chemical vapor deposition system. The substrate
temperature is maintained at 250.degree. C. for both. For the
ammonia plasma pretreatment, a relatively low 35 W power level is
typically used. An example of resulting process parameters are 200
mT chamber pressure, 400 sccm N.sub.2+SiH.sub.4 (95:5) gas flow, 9
sccm ammonia gas flow, 35 W ICP power, and OW RIE power. A silicon
nitride film thickness of 750 A is adequate, with an optical index
(n) of approximately 2.0, which indicates the approximate
composition of the Si.sub.3N.sub.4 passivation layer.
[0021] The nitride deposition process was performed after all of
the processing steps and was followed by etching openings to the
metal contacts and deposition and patterning of a Ti--Au overlay
metal. The ammonia plasma pretreatment and the silicon nitride
deposition were performed in an inductively coupled plasma (ICP)
configured plasma-enhanced chemical vapor deposition system with a
bottom electrode diameter of 8 inches. The silicon nitride film,
using about 5% by volume SiH.sub.4 in a balance of N.sub.2, was
formed using a SiH.sub.4:NH.sub.3:N.sub.2 plasma recipe
(N.sub.2+SiH.sub.4 of 300 to 400 sccm; NH.sub.3 of 9 to 16 sccm).
The substrate temperature was maintained at 250.degree. C. for both
the ammonia plasma pretreatment and the silicon nitride deposition
processes. The silicon nitride film thickness and the refractive
index were measured by ellipsometry, and were 740-800 A and
2.03-2.09, respectively, for different device runs. For the
pretreatment, the chamber pressure was 50 mT and the duration was
180 seconds. The ICP power was set to 35 W at 13.56 MHz, while the
bottom electrode was set to 0 W.
[0022] The invention can be described in connection with FIGS. 1-4
wherein FIG. 1 shows a cross-sectional schematic transistor 10
containing barrier layer 12 disposed on channel 14, which is
followed by channel/buffer layer 16, which in turn is disposed on
buffer layer 17, and which in turn is disposed on substrate 18. The
substrate supports all of the layers of the transistor. The barrier
layer is AlGaN and the buffer/channel layer is GaN with the channel
layer being at the interface of the barrier and the buffer/channel
layers. The channel has a thickness on the order of 50 A to 100 A.
It is the 2 DEG channel layer that characterizes a high electron
mobility transistor, with electron mobility being in excess of 1000
cm.sup.2/V sec. There is a layer of AlN between the buffer/layer
and the substrate layers. The 2-DEG is formed at the interface of
the buffer/channel and the barrier layers.
[0023] Completing the schematic transistor of FIG. 1 is contact 20
which provides electrical connection to source 24, barrier layer
12, and buffer/channel layer 16. Likewise, contact 22 provides
electrical connection to drain 26, barrier layer 12, and
buffer/channel layer 16. Source 24, drain 26, and gate 28 are
disposed in top layer 30. Purpose of the source is to provide
electrons to the transistor device. These electrons flow through
the interface 2 DEG between the barrier layer and the
buffer/channel layer to the drain. The gate controls the flow of
these electrons through the 2 DEG. The source, the drain and the
ohmic contacts 20, 22, 24 and 26 are typically Ti--Al--Ni--Au and
the gate is typically Ni--Au. Typical barrier layer thickness is
200 A 300 A, typical channel layer thickness is 50 A to 100 A,
typical buffer/channel layer thickness is 1 .mu.m to 3 .mu.m,
typical buffer layer thickness is 500 A to 1500 A and typical
substrate thickness is 1 .mu.m to 2 .mu.m.
[0024] Shown in FIGS. 2 (a) and (b) are the measured drain current
for three consecutive sweeps of V.sub.DS from 0 to 30V with
V.sub.GS held at 0 V. The characteristics in FIG. 2 (a) are for an
AlGaN/GaN HEMT without silicon nitride passivation prior to dc bias
stress. For the non-stressed device (solid lines), the three traces
are nearly coincident, as they should be. But after dc bias stress
for 64 hours with V.sub.DS=30V and I.sub.DS=200 mA/mm (dashed
lines), the second and third traces depart significantly from the
first trace. This reduction in drain current for the second and
third traces is due to trapped charge in the structure as a result
of hot electron injection caused at high drain voltages during the
first sweep. This effect is present in the device after stress due
to the generation or activation of defects during the 64 hour dc
bias stress. It can be see in FIG. 2 (b), that a silicon nitride
passivation layer does little to suppress this effect (dashed
lines). But the key result, shown in the solid lines, is that an
ammonia plasma treatment prior to passivation completely eliminates
this effect for at least 176 hours of bias stress.
[0025] Corresponding results are seen with gate lag measurements as
shown in FIGS. 3 (a-d). The gate lag measurements are for
V.sub.DS=1 V and V.sub.GS=V.sub.TH-2 V, pulsed to V.sub.GS=0 V. The
drain current pulse shown in each figure is normalized to the
steady state value of drain current. The ideal response
characteristic is for the pulsed current to rise to the steady
state value, as shown in FIG. 2 (d). Only the ammonia plasma plus
silicon nitride processed device shows the ideal characteristic
after bias stress. These characteristics again indicate that
additional trapping levels are present after dc bias stress and
that the ammonia plasma step is effective in suppressing their
generation.
[0026] The radio frequency (rf) degradation rate, in dB/hour, is
also improved by more than 100 times for the ammonia pretreated
sample as compared to devices on the same split wafer with silicon
nitride only. FIG. 4 shows the change in Pout, normalized to its
initial value, versus hours of rf bias stress. The pre-treated
device shows little drop to 70 hours with a Pin of 7 dBm, while the
silicon nitride only device is degrading in the first 16 hours at
Pin of 5 dBm, then even more rapidly after Pin was increased to 7
dBm.
[0027] An added benefit of the invention is believed to be that the
ammonia plasma provides hydrogen (H) atoms at the AlGaN surface.
Passivation of surface defects by hydrogen has been used
extensively in the past for silicon and gallium arsenide
technologies. Hydrogen has been found to penetrate well into the
AlGaN and reduce the density of bulk n-AlGaN deep level traps.
These traps are thought to play a leading role in the current
collapse phenomena.
[0028] Ammonia plasma ionizes gas into charged particles that cause
the surface to be cleaned and/or charged, as is well known. The
chamber used herein for ammonia plasma treatment was a typical for
semiconductor equipment manufacturers. Treatment duration was 3
minutes, although it can be higher or lower, but is typically in
the range of 3-5 minutes. Other parameters that may be used to
produce the desired ammonia plasma include power of 10-35 watts,
ammonia flow rate of 30 to 70 sccm. Plasma frequency is 13.56
MHz.
[0029] Rf stress conditions included conditions for dc bias and a
microwave signal that goes into the device. The input power of this
signal is typically 10-15 dBM at a frequency of 2-12 GHz. The dc
stress conditions include drain voltage typically of 20-30 volts
and a drain current typically of 100-300 mA/mm.
[0030] In conclusion, reliability of AlGaN--GaN high eletron
mobility transistors that exhibit induced trapping effects due to
extended dc bias or microwave operation has been improved by the
incorporation of an ammonia plasma treatment prior to passivation.
This processing step suppresses increases in current collapse and
eliminates gate lag reductions after extended dc bias and
significantly lessens degradation under microwave operation. It is
believed that the interaction of the plasma with the exposed
surface and H ions and/or atoms diffusion into the epitaxial layers
are responsible for the improved device characteristics after
extended dc bias and microwave operation.
[0031] While presently embodiments of the invention have been shown
of the novel transistors and treatment with ammonia plasma, and of
the several modifications discussed, persons skilled in this art
will readily appreciate that various additional changes and
modifications can be made without departing from the spirit of the
invention as defined and differentiated by the following
claims.
* * * * *