Electron trapping and detrapping in thermally nitrided silicon dioxide
K. Ramesh, A. N. Chandorkar, and J. Vasi
Citation: J. Appl. Phys. 65, 3958 (1989); doi: 10.1063/1.343362
View online: http://dx.doi.org/10.1063/1.343362
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v65/i10
Published by the American Institute of Physics.
Related Articles
The impact of Ge codoping on the enhancement of photovoltaic characteristics of B-doped Czochralski grown Si
crystal
J. Appl. Phys. 111, 043707 (2012)
Understanding ultrafast carrier dynamics in single quasi-one-dimensional Si nanowires
Appl. Phys. Lett. 100, 071104 (2012)
High temperature annealing of n-type 4H-SiC: Impact on intrinsic defects and carrier lifetime
J. Appl. Phys. 111, 033515 (2012)
Effect of nickel contamination on high carrier lifetime n-type crystalline silicon
J. Appl. Phys. 111, 033702 (2012)
Boron-oxygen defect in Czochralski-silicon co-doped with gallium and boron
Appl. Phys. Lett. 100, 042110 (2012)
Additional information on J. Appl. Phys.
Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
Electron trapping and detrapping in thermally nitrided silicon dioxide
K. Ramesh, A. N. Chandorkar; and J. Vasi
Department 0/ Electrical Engineering, Indian Institute o/Technology, Bombay, Bombay-400 076, India
(Received 1 August 1988; accepted for publication 4 January 1989)
Thermal nitridation ofSi02 was carried out at 25% and 40% ofNH 3 for various times and
temperatures. Flat-band voltage and interface-state densities were studied as a function of
processing conditions. Avalanche injection was used to study electron trapping. The density of
electron traps in nitrided oxides depended on the nitridation conditions in the same way as the
negative flat-band voltage (positive voltage). A consistent model for negative flat-band voltage
and electron traps is presented to explain the results obtained. Thermal emission oftrapped
electrons dominates the detrapping mechanism. Energy depths of these traps were found to be
1.3-1.6 eV.
I. INTRODUCTION
Thermal nitridation of silicon dioxide has been of interest recently. Thermal nitridation ofSi0 2 in.an ammonia ambient improves the properties of Si02 , resulting in a higher
dielectric constant, less sensitivity to ionizing radiation, and
lower incidence of low-field breakdown. I - 3 However, various authors have reported that thermal nitridation introduces a large number of electron traps in the films.4,5 It is
well known that the preserice of electron traps usually degrades the performance of the device, giving rise to drifts in
threshold voltage and transconductance. 6 In nitrided oxides,
it has been suggested that the electron traps improve the
breakdown voltage distribution and radiation hardening. 4,7
In the literature thereisdisagreement concerning the
characteristics and the origin of electron traps in nitrided
oxides. Capture cross sections of both 10- 17 and 10- 14 cm 2
have been reported from studies of avalanche injection and
high-field current. 4.5,8.9 Recently Severi and Impronta 10 in
their investigations found the capture cross section of the
electron traps' to be 10- 16 cm 2 • A low generation rate of
interface states II, 12 has been observed for thermallynitrided
Si02 when subjected to electron injection. Thermal emission
of electrons from the electron traps has been reported by
Terry et al. 9
In this paper, we report the electrical characteristics of
thermally nitrided silicon dioxide prepared under different
nitridation conditions. The parameters which were varied
were the ammonia concentration in the ambient (25% and
40%), the time of nitridation (30--120 min) and the temperature of nitridation (1000, 1050, and 1100 ·C). Negative
flat-band voltage and interface-state densities of the nitrided
oxides were studied. Electron trapping in nitrided oxides
was studied using the avalanche injection technique. The
capture cross section of the electron traps was found to be
10- 16 crn: 2. We found from our measurements that the density of electron traps in nitrided oxides follows the trend of
the negative flat-band voltage. Detrapping of the trapped
electrons was also studied in detail. Energy depths of these
traps were evaluated. We suggest that detrapping is mainly
due to thermal emission.
II. DEVICE FABRICATION
Dry oxides of thickness t ox 30--40 nm were thermally
grown at 1000·C on p-type (lOO)-oriented Si substrates
3958
J. Appl. Phys. 65 (10),15 May 1989
with resistivities of 0.125-0.25 n cm. This was followed by
an anneal in nitrogen at 1000 ·C for 30 min. Thermal nitridation was then done on some dry oxides. Nitridation ambient
composition ratios of 25% and 40% ammonia in nitrogen
were used. Nitridation was done at three temperatures,
1000, 1050, and 1100·C for times ranging from 30 to 120
min. Post-nitridation anneal in nitrogen was done for some
samples at the temperature of nitridation for 30 min. For
gate contaCt, aluminum was deposited in an e-beam evaporation system. Gate contacts of 1 mm 2 were defined by photolithography. Finally, post-metallization anneal was carried
out in hydrogen at 450 ·C for 30 min.
.
III. EXPERIMENTAL METHODS
This seCtion briefly describes the experimental techniques used in this study.
.
A. High-frequency C- V
High-frequency capacitance-voltage (HFCV) measurements at 1 MHz were made using a PAR 410 C- V plotter
and HP 9826 instrument controller. The flat-band voltage
V FB and fixed charge Nf for nitrided oxides and control dry
oxides were evaluated from a high-frequency C- V curve. 13
B_ Quasistatic C- V
Interface-state densities Dit were measured by the quasistatic technique. 14 A slow linear ramp (typically 30 m V/s)
was applied across the MOS capacitor. The displacement
current through the MOS capacitor was measured with a
Keithley 617 electrometer and data was transferred to the
HP 9826 instrument controller via the IEEE 488 instrumentation bus. The measured current is proportional to the lowfrequency capacitance. The interface-state density was calculated from the quasistatic C- V curve using a standard
procedure. 14
C. Electron trapping
Hot electrons were injected into the oxide from Si by the
avalanche injection technique. The circuit used to inject hot
electrons from silicon into the insulator was similar to that
published by Nicollian et al. 15 The MOS capacitor is driven
into deep depletion momentarily by applying a sine wave of
ISO-kHz frequency and 30--40 Vpop amplitude. The injected
current in the MOS capacitor was maintained constant dur-
0021-8979/89/103958-05$02.40
@) 1989 American Institute of Physics
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3958
ing electron injection by a constant current feedback circuit.
The high-frequency C- V curves were measured on all the
capacitors prior to electron injection. The shift in flat-band
voltage due to electron trapping was measured by interrupting the electron injection at fixed intervals oftime. The capture cross section (T and the effective densities Neff of electron
traps were calculated using the first-order capture process
for a single trap level,16 assuming no detrapping, and also
assuming that the traps were located close to the Si-Si0 2
interface.
-2.0
~1000·C
~
UJ
\!)
~ -1.5
~
o
~"'T
·~100·C
~
5 -1.0
l1.
,
,,
.
1050 C
-0.5 L...---'-_-'---'-_...L...--L_...L.---'
D. Field emission of trapped electrons
Field emission measurements were carried out by applying negative and positive voltages to the gate electrode at
room temperature. Shifts in V FB . due to field emission of
trapped electrons were measured as a function of applied
field for fixed time.
E. Thermal emission of trapped electrons
Thermal detrapping measurements were used to find
the energy levels of the electron traps in the insulator. Thermal detrapping measurements were carried out as a function
oftime (10-100 min) for a constant temperature and as a
function of temperature (100-350 DC) for fixed times. No
electric field was applied during detrapping. HFCV measurements were used to measure the shift in flat-band voltage
due to thermal detrapping.
IV. RESULTS AND DISCUSSION
A. Negative flat-band voltage
The HFCV curves show a larger negative flat-band voltage for thermally nitrided oxides compared to the control
dry oxides. The amount of negative flat-band voltage in nitrided oxides was found to be a function of nitridatiori time,
temperature, and the NH3 content. Negative flat-band voltage for nitrided oxides improves by 0.1-0.2 V with postnitridation anneal for 30 min in nitrogen. Chen, Tseng, and
Chang l ? have also observed an improvement in negative flatband voltage with a post-nitridation anneal.
Figure 1 shows the flat-band voltage as a function of
nitridation time for different nitridation temperatures at
40% NH 3. From Fig. 1, we clearly observe that the trend is
for a smaller negative flat-band voltage with longer nitridation times and also for higher nitridation temperatures. The
, bars in the plot indicate the spread in V FB over a 2-in. wafer.
It was also observed that the negative flat-band voltage improves by a small amount for 40% NH3 when compared to
25% NH3 at different nitridation temperatures and times.
However, at 1050 DC for 120 min it improves by 0.3-0.4 V
and hence results in a crossover with the 1100 DC curve for
40% NH3 nitridation as shown in Fig. 1.
Several authors have explained the negative flat-band
voltage. First, Ito, Nakamura, and Ishikawa I have suggested
that it is due to active hydrogen atoms dissolving Si-O
bonds and generating positively charged silicon ions, pro-
o
20
40
60· 80
100 120 140
NITRIDATION TIME (min)
FIG. 1. Plot of negative fiat·band voltage vs nitridation time for different
nitridation temperatures at 40% NH 3 •
gas in Si02 with increasing temperature is responsible for the
observed smaller shift in V FB for higher temperatures. In
another model, they have suggested that it is a result of indiffusion of radicals into the Si02 films causing generation of
Si dangling bonds. 18 The possible radicals could be O-H, H,
and N-H. The formation ofSi-N bonds during the reaction
process reduces or redistributes the Si dangling bonds and
results in a decrease of negative flat~band voltage. Vasquez
and Madhukar l9 pointed out that the flat-band voltage is
dependent on the nitrogen distribution and the strain created at the interface. At sufficiently high nitrogen concentrations, nitrogen incorporation leads to healing out of the initial strain and thus reduces the flat-band voltage.
At present, the exact nature of the nitridation-induced
shifts in flat-band voltage is stillundear. We speculate in our
model that the dissociated hydrogen gas from NH3 during
nitridation breaks the Si-O bonds as suggested by Chen and
co-workers l ? and trivalent silicon defects (O=Si') are
created. These trivalent silicon defects act as fixed charge
and contribute to the larger negative flat-band voltage in
nitrided oxides. We suggest that the relaxation of V FB with
increasing temperature, time, and NH3 content during nitridation is not due to the decreasing·solubility of hydrogen gas
in Si02 as pointed out by Chen and co-workers 17 but rather
due to a larger incorporation of nitrogen into the oxide
which is bonded with the already created trivalent Si, thereby reducing the flat-band voltage at higher temperatures or
times. These centers are similar to the centers that act as
electron traps as described in Sec. IV C.
B. Interface-state densities
Interface-state densities of the nitrided and control dry
oxides were measured. In nitrided oxides they were found to
be a function of nitridation time, temperature, and percent of
NH3 content. We have observed that for smaller times of
nitridation (30 min), D jl in midgap is higher compared to a
l20-min nitridation as shown in Fig. 2. The trend in interface-state density as a function of nitridation time is quite
similar to that observed for the flat-band voltage shift. We
can explain the above results in the same fashion as that of
tons, and dangling bonds. Chen, Tseng, and Chang l ? have
fiat-band voltage shift. At the Si-Si0 2 interface, trivalent sili~.
also invoked the dissociation of Si-O bonds by hydrogen
gas and suggested that the decreasing solubility of hydrogen
con (Si=Si' ) gives rise to interface-state densities according
to Svensson. 20
3959
J. AppL Phys., Vol. 65, No.1 0, 15 May 1989
Ramesh, Chandorkar, and Vasi
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3959
lJ4~
________________________________
~
~
>
L&J
10
Cl
!:;
0
>
N~
8
0
z
Ie
ld
u
"i
1050 ·C.
e(
13
10
e(
2
aJ
'<
~
6
)(
1100 ·C
...
..J
0-
~
11
10
...
10
10 ~--~~--~~__~~__~~____~~
0.0
0.2
0.3
0.6
0.8
1.0
'"
4
I-
J:
Ev
2
E (eV)
40 'I. NH3
0
10
30
20
40
50
60
TIME (min)
FIG. 2. The effect of nitridation time on interface-state densities for nitrided
oxide at HOO'C, 40% NH 3. (0) Control dry oxide, (a) 3D-min nitridation,
(b) 6O-min nitridation, and.(c) l20-min nitridation.
C. Electron trapping
In thermally nitrided oxides the density of electron traps
has been found to be larger than for the control dry oxide
samples. 4 We measured the control samples as well as the
25% and 40% nitrided oxides. The results indicated that
V FB shifts due to avalanche injection are more for 25% than
for 40% nitrided oxides at 1100 ·C. Figure 3 shows the V FB
shift due to injection as a function of injection time for the
control oxides and nitrided oxides. Electron trapping in nitrided oxides is also a function of nitridation temperature as
shown in Fig. 4 for 40% NH 3 • The density of electron traps
was less for 1100 ·Cnitrided oxide compared to 1050·C ni-
~........~--<,..--o-- 1100·C
8
7
L _ - - - - - - - - - - - - - 1 1 0 0 ·C
a
FIG. 4. Plot of Hat-band voltage shift due to electron trapping vs injection
time for different nitridation temperatures. The average injected current
density was 2.2x 10- 5 A.
trided oxide. In oxides nitrided at 1100 ·C, a hump was observed due to electron trapping as shown in Fig. 3. The exact
nature of this behavior is not clear but it should be mentioned that before the "turnaround," the t:.. V FB for 25% and
40% follow each other closely. The trend in V FB shifts due to
electron trapping with change in nitridation temperature is
similar to that observed for negative flat-band voltage. Figure 5 shows the V FB shifts due to
electron trapping in 40%
\
nitrided oxide as a function of nitridation temperature.
Interface states were generated due to electron injection
in nitrided oxides. Figure 6 shows the Dit -Eg curves for the
nitrided oxide grown at 1100 ·C for 60 niin. The stretchout in
HFCV curves due to interface states generated by electron
injection is small and therefore the V FB shifts were assumed
t6 be due to electron traps only.
.
Electron capture cross section (7 of these traps was evaluated in a similar way to that of Nicollian et al. 15 from the
slope of the plot In (1 - t:.. V FB / t:.. V FBO) vs time where
t:.. V FBO is the saturation voltage of t:.. V FB • The capture cross
sections werefound to be 10- 15_10- 16 cm 2 • These values are
larger than that of water related traps in oxides (10- 1710- 18 cm 2 ). These traps do not seem to be due to OH or H 2 0
defects as suggested by Lai, Dong, and Hartstein. 4
~
z
~.4
\!I".S
~
'<
25 ./. N H3
)1----01
40'/, NH3
...-..... Ory oxide
...
0----0
..J
.............
g10.S
........
Z
5?
9.5
[j
8.5
8
g
"
\
\
\
\\
\
\
~ 7.5
a
~
> 6.5
--Ir--o.
60 min
120 min
\
\
\.
<l
o
.10
20
30
40
50
60
TIME (min)
FIG. 3. Plot of Hat-band voltage shift due to avalanche injection vs injection
time fcir 25% and 40% NH3 at 1100 ·C. The average injected current density was 2.2X 10- 5 A.
3960
J.Appl. Phys., Vol. 65, No. 10,15 May 1989
5.5
950
1000
1050
1100
NITRIDATION TEMPERATURE
1150
'c
FIG. 5. Plot of the Hat-band voltage due to electron trapping as a function of
nitridation temperature at 40% NH"
Ramesh, Chandorkar, and Vasi
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3960
8.20
ld4 __----------------------------~--,
I
~ 8.00
i
i
w
~ 7.60
I
~
I
i
i..... ________________ --
a
~
a 7.60
.'
---
z
4:
/
~
..... _....... _.-.-
(D
,/
~
..J
~.-.,; ... /
t..
~
o
7.40
7.20
l-
t..
:E 700
11\
ld°L-____L-____L-____
6.80
L -____L -_ _ _ _~~
0.0
0.2
0.4
0.6
E~
0.8
50
100
150
200
250
DETRAPPlNG TIME(hr)
Ec.
E(eV)
FIG. 6. Interface-state density distribution for oxide nitrided for 60 min, at
40% NH, and llOO'C. (0) Initial, (a) after avalanche injection ofelectrons, ami (b) after thermal detrapping at 350 'C.
Based on our results, the variation of electron traps in
nitrided oxides with nitridation temperature or NH3 content
has a similar behavior to that of a negative flat-band voltage.
The larger number of electron traps in nitrided ox~des may
be due to trivalent Si defects located 100 A or so away from
the interface. These are similar to the trivalent Si defects
which produce the fixed charge except that the latter are
very close to the interface (within about 20 A). The similar
trends observed in negative flat-band voltage and electron
traps are due therefore to similar kinds of defects. It is also
possible that some of the electron traps might be due to nitrogen bonded to trivalent silicon (O=Si' ). Nitrogen acts as a
substitutional impurity and replaces oxygen sites in the
Si-O bond structure and thus also acts as an electron trap.21
D. Field emission
During field emission, it was observed that the rate of
detrapping of trapped electrons was significantly dependent
on the polarity of voltage applied at the gate. All the trapped
electrons could be emptied out in 25 min at a negative field of
7 MV Icm as shown in Fig. 7. However, with positive fields
only 30% of the trapped electrons could be removed. This
implies that most of the trapped electrons are located near
the Si-Si02 interface, and can tunnel to the silicon under
negative bias.
E. Thermal emission
Some of the trapped electrons were released at room
temperature within 3-4 h of the avalanche injection. Typi-
100
FIG. 8. Shift in fiat-band voltage due to electron detrapping as Ii function of
detrapping time at room temperature.
cally, this resulted in a 0.5-1.0 V V FB shift. But later, even
after 100 h, the shift in V FB is only 1 V as shown in Fig. 8.
Chang, Johnson, and LyonS have also reported room-temperature detrapping in nitrided oxides, and in their case V FB
had fully recovered to its original value after 18 h, whereas in
our case it is only a partial recovery. We consider that this
detrapping of electrons may be due to a tunnel emission
mechanism from shallow traps. A plot of flat-band voltage
shift due to detrapping versus logarithmic time (at 25 ·C)
shown in Fig. 9 is a straight line. According to Yamabe and
Miura22 this straight line indicates the existence of tunnel
emission.
The remainder of the trapped electrons were either detrapped at varying temperatures up to 375 ·C for fixed times
(10-20 min) or at fixed temperatures (between 200 and
350 DC) for times up to 100 min. At temperatures of 300375 ·C we were able to detrap almost all the trapped electrons. After the detrapping experiment, we could fill these
traps by repeating the avalanche injection and the V FB shift
due to electron trapping is the same as before. It was also
possible to detrap all the filled traps a second time. This
indicates that the traps are intrinsic to the insulator, and
probably not created by the avalanche process. Dit at mid gap
after detrapping was reduced nearly to the initial value before injection as shown in Fig. 6. The possible reason for
reduction in Dil after thermal detrapping is that the interface
states created during avalanclIe injection are annealed during thermal detrapping due to possible restructuring of the
lattice at the interface.
~10.00
D-
w
o
o
0
1.0
!of
Time= 25min
5> 9.00
FIG. 7. The percentage recovery of!!. V FB I!!. V FBO vs applied
field as a result of field emission
at room temperature.
a
FIG. 9. Flat-band voltage
shift due to electron detrapping at room temperature as
a function of logarithmic
time.
z
~ 8.00
~
?:100
l-
t..
l:
L..............................................._~i~
0_ 10 -6 -6 -4 -2 0
2
4 6 6
FIELD (t-~.v/cm)
3961
J. Appl. Phys., Vol. 65, No. 10,15 May 1989
1Il 6.00'L..__~_
-4.00 -2.00
0.00
2.00
4.00
LOGARITHMIC TIME
Ramesh, Chandorkar, and Vasi
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3961
trapped charge at room temperature is due to tunnel emission from shallow traps near the interface. Later, thermal
emission dominates the detrapping process from the (energetically) deeper traps in the nitrided oxides.
550,------------------. 0.25
... .A'~ ...
.....
A""-----0.20
m
o
z
0.15
~ 5.00
~
~ 4.80
>"-
~
v. CONCLUSIONS
0
m
0.10
~
...
>"<I
c:
...J
~ 450
I
If)
I
0.05
4.40 ,:
,
420L-__~____~__~____~ 0.00
40
o
20
60
80
OETRAPPING TIME(min)
FIG. 10. (a) Flat-band voltage shift as a function of detrapping time at
200 ·C. (b) Plot of log normalized fiat-band voltage shift as a function of
detrapping time.
The energetic location of the (deeper) electron traps
was found by thermal detrapping studies at constant temperature for different times. Figure lO(a) shows the a v FB vs
detrapping time at· 200°C. The nonlinear variation of
In(.6.v FBol a v FB) with time, as shown in Fig. lO(b), indicates that the traps do not have a single time constant, but
are distributed in energy.23 It may be noted that earlier,
while calculating (7, we had assumed a single trap level. Although it is now seen that the traps are distributed, the value
of (7 calculated is approximately correct.
To determine the energy distribution of electron traps a
procedure similar to the thermal emission of electrons from
interface states to the conduction band of silicon was used. 24
The rate of emission of trapped electrons per cm 2 from the
traps at energy ET is given by
kTDo, (ET)v
dnT
(1)
dt
vt + 1
where'v is the escape frequency of electrons and DOl (E T ) is
the electron trap density!cm 2 eV. The trap level ET is given
by
ET = kTln(vt
+ 1) .
(2)
Now,
aVFB=qnT(tox -X)I€ox,
where X is the centroid of the charge. For t~ 10( 1 ) can be written as
da VFB
dt
kTDo, (E T )q(tox -
X)
t€ox
11
(3)
s, Eq.
(4)
The slope of the curve da V FD I dt vs 1/t gives the electron
trap density with trap location determined by Eq. (2). The
traps are located at energy levels between 1. 3 and 1.6 eV. The
corresponding DOl (E T ) values were found to be 2-7 X 10 121
cm 2 eV, assuming X is near to the interface. Chang and coworkers 5 have suggested tunnel emission for detrapping and
estimated the depth of dominant traps to be ;;;.2 eV below the
cond1,lction band. Emission of trapped electrons by tunneling at high fields by Terry et al. 9 yields a trap depth of 3.5-4
eV. From our observations, the electron traps in nitrided
oxides are of two types having different detrapping mechanisms and are distributed in energy. Initially, detrapping of
3962
J. Appl. Phys., Vol. 65, No.1 0, 15 May 1989
Our in vestigations show that the characteristics of thermally nitrided silicon dioxide are dependent on nitridation
time, temperature, and on the NH3 content used. The negative flat-band voltage decreases with increasing nitridation
time, temperature, and NH3 content. The density of electron
traps strongly increased upon nitridation. The exact number
is dependent on the processing conditions. The variation
with processing conditions of V FB and electron trap density
is similar. This had led us to a model of trivalent silicon
defects being responsible for both. Capture cross section of
these electron traps were found to be 10- 15_10- 16 cm 2. Energy depths ofthese traps were evaluated to be 1.3-1.6 eV.
ACKNOWLEDGMENT
This work was partially funded by the Ministry of Human Resource Development, Government of India, under
its thrust area program in microelectronics.
'T. Ito, T. Nakamura, and H. Ishikawa, J. Electrochem. Soc. 129, 184
( 1982).
2T. Ito, H. Arakawa, T. Nozaki, and H. Ishikawa, J. Electrochem. Soc.
127,2248 (1980).
3R. K. Pancholy and F. M. Erdmann, IEEE Trans. Nuc!. Sci. NS-30, 4141
(1983).
4S. K. Lai, D. W. Dong, and A. Hartstein, J. Electrochem. Soc. 129, 2042
( 1982).
5S. T. Chang, N. M. Johnson, and S. A. Lyon, App!. Phys. Lett. 44,316
(1984).
6S. M. Sze, Physics of Semiconductor Devices, 2nd ed. (Wiley, New York,
1981).
7R. Sundaresan, M. M. MatIoubian, and W. E. Baiely, IEEE Trans. Nuc!.
Sci. NS·33, 1223 (1986).
"A. Yankova, L. DoThanh, and P. Balk, Solid-State Electron. 30, 939
( 1987).
9F. L. Terry, Jr., P. W. Wyatt, M. L. Naiman, B. P. Mathur, e. T. Kirk, and
S. D. Senturia, J. App!. Phys. 57, 2036 (1985).
10M. Severi and M. Impronta, App!. Phys. Lett. 51, 1702 ( 1987).
lIS. S. Wong and W. G. Oldham, IEEE Trans. Electron Devices ED-32, 978
(1985 ).
12S. K. Lai, J. Lee, and V. K. Dham, Tech. Dig. IEDM 83,190 (1983).
13E. H. Nicollian and J. R. Brews, MOS Physics and Technology (Wiley,
New York, 1982).
14M. Kuhn, Solid-State Electron. 13, 873 (1970).
15E. H. Nicollian, e. N. Berglund, P. F. Schmidt, and J. M. Andrews, J.
App!. Phys. 42,5654 (1971).
16D. J. DiMaria, in The Physics of Si02 and Its Interfaces, edited by S. T.
Pant elides (Pergamon, New York, 1978), p. 160.
17C. Chen, F. Tseng, and C. Chang, J. Electrochem. Soc. 131, 875 (1984).
'"p. Pan, J. App!. Phys. 61, 284 (1987).
19R. P. Vasquez and A. Madhukar, App!. Phys. Lett. 47, 998 (1985).
2I'e. M. Svensson, in The Physics of Si02 and Its Interfaces, edited by S. T.
Pante1ides (Pergamon, New York, 1978), p. 328.
21S. T. Pantelides, Thin Solid Films 89,103 (1982).
12K. Yamabe and Y. Miura, J. App!. Phys. 51, 6258 (1980).
"It is also possible that the traps are distributed spatially in the Si0 2and it is
tunneling from these traps which causes detrapping. However, the strong
temperature dependence of the detrapping from the deeper states indicates a thermal process, and therefore leads to the conclusion of energetically distributed traps.
24L, Manchanda, J. Vasi, and A. B. Bhattacharyya, Solid-State Electr~m.
22,29 (1979).
Ramesh, Chandorkar, and Vasi
Downloaded 27 Feb 2012 to 202.75.203.73. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
3962