calculation of the r0a product in n+–n–p and p+–p–n gainassb infrared detectors
TRANSCRIPT
Infrared Physics & Technology 45 (2004) 181–189
www.elsevier.com/locate/infrared
Calculation of the R0A product in nþ–n–p and pþ–p–nGaInAsSb infrared detectors
Tian Yuan a,1, Soo-Jin Chua a,b,*, Yixin Jin c
a Department of Electrical and Computer Engineering, Center for Optoelectronics, National University of Singapore,
Singapore 119260, Singaporeb Institute of Materials Research and Engineering, National University of Singapore, Singapore 119260, Singapore
c Changchun Institute of Physics, Chinese Academy of Sciences, Changchun 130021, PR China
Received 29 April 2002
Abstract
In this paper, the zero-bias resistance areas product R0A is calculated in the nþ–n–p and pþ–p–n Ga0:8In0:2As0:81Sb0:19infrared detectors, on the base of the material parameters in the three layers. The calculated results show that pa-
rameters in the heavily doped layer in the different structures have different influences on R0A. Moreover, R0A in the nþ–n–p structure is higher than that in the pþ–p–n structure because the higher carrier concentration in the nþ-region for
the nþ–n–p structure improves R0A whereas the one in the pþ-region for the pþ–p–n structure reduces R0A.� 2003 Elsevier Ltd. All rights reserved.
Keywords: R0A; Photodetectors; GaxIn1�xAs1�ySby; Parameters
1. Introduction
In recent years, the quaternary alloys GaIn-
AsSb lattice matched to GaSb substrates have
been increasing interest in infrared (IR) optoelec-tronic devices for spectral range 2–5 lm [1]. Such
devices have many important applications, in-
cluding gas spectroscopy analysis, remote sensing
* Corresponding author. Address: Department of Electrical
and Computer Engineering, Center for Optoelectronics, Na-
tional University of Singapore, Singapore 119260, Singapore.
Tel.: +65-68744784; fax: +65-67797454.
E-mail address: [email protected] (S.-J. Chua).1 Current address: Microelectronic Centre, School of Electri-
cal and Electronic Engineering, Nanyang Technological Uni-
versity, 639798 Singapore.
1350-4495/$ - see front matter � 2003 Elsevier Ltd. All rights reserv
doi:10.1016/j.infrared.2003.09.003
of gas pollutants, light–wave communication sys-
tem using fluoride glass fibers, as well as a number
of medical applications. GaInAsSb alloys have
already shown their capabilities to achieve efficient
lasers [2] and detectors [3–5]. One of the importantfigures of merit in the detectors is the zero-bias
resistance areas product R0A because the value ofR0A directly affects the detector performance ex-pressed by the detectivity D�. Lin et al. [6] reported
the experimental results of R0A less than 100 X cm2
in GaInAsSb detector. However, very little infor-
mation can be provided from literature about the
theoretical analysis of R0A for these material de-tectors.
In this paper, the zero-bias resistance areas
product R0A is calculated based on the differentdevice structures (nþ–n–p and pþ–p–n) and the
ed.
182 T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189
related material parameters. The results show:
which is the optimal structure out of the two
structures and what are junction depths and car-
rier concentrations that would maximize R0A. Thetwo different structures are quantitatively com-
pared and R0A is checked by the influence of thecarrier concentrations and the junction depths in
the three layers of each structure, as well as the
surface recombination velocities on the two sur-
faces for each structure. Furthermore, on the basis
of the theoretical calculation, some optimum
conditions are proposed to maximize R0A for de-tectors. These results are important for the fabri-
cation of detectors.
2. Device structure and theoretical model
We consider the structures of nþ–n–p and
pþ–p–n Ga0:8In0:2As0:81Sb0:19 detectors shown in
Fig. 1, in which the lattice of Ga0:8In0:2As0:81Sb0:19alloys is matched to GaSb [7]. In the p–n junction,four kinds of noise mechanism: Auger, Radia-
tive, Generation–recombination (G–R) and tun-
neling, are considered and expressed by R0A. Thesecontributions to R0A all add in parallel and theexpressions are given in the previous paper [8]. In
isotype nþ–n and pþ–p junctions, the basic as-
sumptions are made as follows:
(1) The nþ- or pþ-region is heavily doped such that
the depletion region for the nþ–n or pþ–p
dttn +
(a)
rpS
reSe
rnn Snn
rpn+
SpWpWn
W npWnn
pnn+
Fig. 1. The two structures fo
junction exists in the low-doped region [9].
They are shown in Fig. 1. G-R noise in the
nþ–n (or pþ–p) depletion region is considered.
(2) The effective interface recombination velocities
Spp and Snn located at the pþ–p and nþ–n junc-tions are introduced. They are derived to take
into account the minority-carrier behavior in
the heavily doped pþ- and nþ-layers [10,11].
(3) Compared to the major carrier concentration
in the heavily doped layer, the minority-carrier
concentration is very low; additionally the
minority-carrier diffusion mechanism is deter-
mined by the interface recombination velocity.Therefore, Auger and radiative noise mecha-
nisms in the heavily doped regions are ne-
glected.
(4) The tunneling mechanism across nþ–n and
pþ–p junctions is neglected.
The interface recombination velocities Spp andSnn are valid for the arbitrary pþ–p and nþ–n high-low junctions [11]:
Spp ¼pDepþpþLepþ
repþ coshdpþLepþ
� �þ sinh dpþ
Lepþ
� �
repþ sinhdpþLepþ
� �þ cosh dpþ
Lepþ
� � ð1Þ
Snn ¼nDpnþnþLpnþ
rpnþ coshtnþLpnþ
� �þ sinh tnþ
Lpnþ
� �
rpnþ sinhtnþLpnþ
� �þ cosh tnþ
Lpnþ
� � ð2Þ
where D ¼ KTl=q, L ¼ ðDsÞ1=2, r ¼ SL=D.
dp+dt
(b)
Wn
re p+
Se
rpp Spp
pWp
Wnp Wpp
p+n p
r a GaInAsSb detector.
T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189 183
3. Calculated results and discussion
The calculations have been performed on the
nþ–n–p and pþ–p–n Ga0:8In0:2As0:19Sb0:81 infrared
detectors operated at 300 K. The basic parametersare shown in Table 1 and the related parameters
have been evaluated in the same way as in the
previous papers [7,8]. For practical calculation, the
r and Nf in the p–n and pþ–p (or nþ–n) junc-tions are assumed to be the same. Parameters
Table 1
Basic parameters for the nþ–n–p and pþ–p–n structures
Basic parameters T ¼ 300 K, x ¼ 0:8, Nf ¼ 1014 cm�3
nþ–n–p structure
nþ-region n-region p-
Carrier concentration (cm�3) 5· 1018 1018 10
Width (lm) 0.5 2 5
Mobility (cm2/V s) 1000 1000 24
Surface recombination
velocity (m/s)
0 0
Nf and r are respectively the trap density and capture cross-section in
1013 1014 1015 1016 1017 1018100
101
102
103
104
(R0A)Auger
(R0A)rad
(R0A)GRnp
(R0A)GRpp
(R0A)total
p+ =n=1018 cm-3
Nf σ s =0.1cm-1
N f σs =0.01cm- 1
R0A
(ohm
-cm
2 )
p(cm- 3 )(a) (
Fig. 2. The dependence of ðR0AÞtotal and its components on: (a) thecentration for the pþ–p–n structure.
except for mobilities, such as carrier concentra-
tions, widths, surface recombination velocities,
have been calculated to show how these parame-
ters affect R0A.
3.1. R0A in the pþ–p–n structure
Fig. 2 shows R0A and its components in the pþ–p–n structure as functions of p- and n-side carrier
concentrations, with other parameters in Table 1
, rs ¼ 10�15 cm2
pþ–p–n structure
region pþ-region p-region n-region
17 1018 1017 1018
0.5 5 2
0 240 240 1000
0 0
the depletion region.
1015 1016 1017 1018 1019101
102
103
(R 0A)Auger
(R 0A)rad
(R 0A)GRnp
(R 0A)GRpp
(R 0A)total
p+ =1018 cm- 3 p=1017 cm-3
R0A
(ohm
-cm
2 )
n(cm- 3 )b)
p-side carrier concentration (p) and (b) the n-side carrier con-
184 T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189
keeping constants. Moreover, ðR0AÞGRpp , ðR0AÞGRnpand ðR0AÞtotal with Nf ¼ 1013 cm�3 and r ¼10�15 cm2 are also plotted in Fig. 2(a). In Fig. 2(a),
all of R0A components contribute to ðR0AÞtotal inNf ¼ 1014 cm�3 and r ¼ 10�15 cm2, especially in
p > 1016 cm�3. In the range of p < 1016 cm�3,ðR0AÞtotal is mainly determined by the G-R noise inthe p–n junction, while other noise components
have a little contribution. Because of ðR0AÞGR /1=rNf [8], ðR0AÞGR and ðR0AÞtotal are increased if rand Nf are reduced, which is similar to that in then–p GaInAsSb detector [8]. In Fig. 2(a), the
maximum value of ðR0AÞtotal are obtained at p ¼1017 cm�3. Based on this condition, Fig. 2(b) showsthe dependence of ðR0AÞtotal and its components onthe n-side carrier concentration. ðR0AÞtotal dependson all the components in n > 1016 cm�3; in n <1016 cm�3, ðR0AÞtotal is mainly limited by ðR0AÞGRnp.With increasing n, ðR0AÞtotal rises up and keeps aconstant in n > 1018 cm�3. No tunneling mecha-
nism appears in Fig. 2 because the direct tunneling
mechanism needs a necessary factor that bothn- and p-type semiconductor materials must be
degenerate. For the Ga0:8In0:2As0:19Sb0:81 alloy,
the degeneracy state is in p > 1018 cm�3 and n >1016 cm�3.
Fig. 3 shows the dependence of ðR0AÞtotal on thep-side carrier concentration with the pþ-side car-
rier concentration (pþ), width (dpþ) and surfacerecombination velocity (Se) as parameters, underthe condition of other parameters in Table 1
keeping constants. ðR0AÞtotal is obviously reducedwith increasing pþ, dpþ and Se. The pþ-side pa-rameters affect ðR0AÞAuger and ðR0AÞrad in the p-sidethrough Spp and ðR0AÞtotal. With increasing the p-side carrier concentration, the maximum value of
ðR0AÞtotal is obtained at p ¼ 1017 cm�3, therefore
the carrier concentration in the pþ-side should behigher than 1017 cm�3. In the p–n homojunction
structure, ðR0AÞtotal, associated with the Auger andradiative mechanisms [8], is strongly reduced by
the p-side surface recombination velocity, In the
triple-layer structure, it is hoped that the interface
recombination velocity can be reduced with a
heavily doped layer. In Fig. 3(c), although
ðR0AÞtotal markedly decreases in the range ofSe > 104 m/s, with the increasing of Se from 0 to
102 m/s, ðR0AÞtotal almost does not change. This
indicates that the Spp is limited by the parametersin the heavily doped layer more or less. In addi-
tion, ðR0AÞtotal in Fig. 3(a) strongly decreases atpþ ¼ 1021 cm�3 and in p > 1018 cm�3 because the
tunneling mechanism appears in this part.
Fig. 4 shows ðR0AÞtotal with the change of thep-side width (d). ðR0AÞtotal decreases with the thickerwidth in the p-side only in the range of p > 1016
cm�3. This result is caused by Auger and radiative
noise mechanisms because only both of them are
related with p-side width [12]. As shown in Fig. 2,
these two noise mechanisms affect ðR0AÞtotal in therange of p > 1016 cm�3. Fig. 5 shows the depen-
dence of ðR0AÞtotal on the n-side carrier con-centration with the n-side width (t) and surfacerecombination velocity (Sp) as parameters. Al-though ðR0AÞtotal has a little decreasing with in-creasing t and Sp, its optimum value is not affectedby these two parameters in n > 1018 cm�3. There-
fore, if the n-side carrier concentration is higher
than 1018 cm�3, the influence of these two para-
meters in the n-side on ðR0AÞtotal can be neglected.This result is advantageous of the device fabrica-
tion because the limitation of these two parameters
is canceled.
3.2. R0A in the nþ–n–p structure
Fig. 6 shows ðR0AÞtotal and its components in thenþ–n–p structure as functions of p- and n-sidecarrier concentrations with other parameters in
Table 1 keeping constants. Moreover ðR0AÞGRnnand ðR0AÞtotal with the different nþ are also plottedin Fig. 6(b). The different ðR0AÞGRnn and ðR0AÞtotalare marked as the number 1, 2 and 3. In Fig. 6(a),
in the range of 1016 cm�3 < p < 1018 cm�3,
ðR0AÞtotal is dominated by all components and itsmaximum value is obtained. In p < 1016 cm�3,ðR0AÞGRnp controls ðR0AÞtotal, while in p > 1018
cm�3, ðR0AÞtunnel controls ðR0AÞtotal and the tun-neling mechanism strongly reduces ðR0AÞtotal. InFig. 6(b), with the change of n, all of the R0Acomponents contribute to ðR0AÞtotal. The tunnelingmechanism does not appear because the the p-side
material is in the nondegeneracy state with p lessthan 1017 cm�3. With the change of nþ, ðR0AÞAugerand ðR0AÞrad should be changed through Snn, butthese two mechanisms do not. With increasing nþ,
1013 1014 1015 1016 1017 1018 1019
1
10
100 n=1018 cm-3
p+ =1018 cm-3
1019
1021R0A
(ohm
-cm
2 )
p(cm-3)(a)1013 1014 1015 1016 1017 10180
20
40
60
80
100
120
140 p+ =n=1018 cm-3
dp+ =0.5 µm
1 5
R0A
(ohm
-cm
2 )
p(cm-3)(b)
1013 1014 1015 1016 1017 101810- 3
10- 2
10- 1
100
101
102
103
102
Se=0(m/s)
108
104
p+ =n=1018 cm-3
R0A
(ohm
-cm
2 )
p(cm-3)(c)
Fig. 3. The dependence of ðR0AÞtotal on the p-side carrier concentration with the pþ-side: (a) carrier concentration (pþ), (b) width (dpþ )and (c) surface recombination velocity (Se) as parameters for the pþ–p–n structure.
T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189 185
the peak of ðR0AÞtotal rises and shifts to the high nbecause the peak of ðR0AÞGRnn changes in that way.That is, the high nþ-side carrier concentration is
conducive to improve ðR0AÞtotal. With the variety
of Sp and tnþ , ðR0AÞtotal never changes and theirinfluence on ðR0AÞtotal will be neglected. The in-fluence of the parameters in the nþ-side for the nþ–
n–p structure is completely opposite to that of the
1013 1014 1015 1016 10170
20
40
60
80
100
120
140 p+=n=1018cm-3
d=5µm
8
10
R0A
(ohm
-cm
2 )
p(cm- 3 )
Fig. 4. The dependence of ðR0AÞtotal on the p-side carrier con-centration with the p-side width (d) as a parameter for the pþ–p–n structure.
1015 1016 1017 101820
40
60
80
100
120
140
160p+=1018cm-3 p=1017cm-3
t=0.5µm 5
R0A
(ohm
-cm
2 )
n(cm-3)(a)
Fig. 5. The dependence of ðR0AÞtotal on the n-side carrier concentrativelocity (Sp) as parameters for the pþ–p–n structure.
186 T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189
parameters in the pþ-side for the pþ–p–n structure.
We will discuss later. Because the influence of the
n-side width on ðR0AÞtotal in the nþ–n–p structure issimilar to that in the pþ–p–n structure, it is ne-
glected here.
Fig. 7 shows the influence of the p-side pa-rameters (Se and d) on ðR0AÞtotal for the nþ–n–pstructure. The parameters Se and d are only relatedwith Auger and radiative mechanisms. The high Seand d reduce ðR0AÞAuger and ðR0AÞrad [12], thereforeðR0AÞtotal decreases. Moreover, the peak of
ðR0AÞtotal shifts to the high p in Fig. 7(a), on thecontrary to the low p in Fig. 7(b). Compare Fig.7(a) with Fig. 3(c), except that ðR0AÞtotal in the nþ–n–p structure is a little higher than that in the pþ–
p–n structure at Se ¼ 0, ðR0AÞtotal in the pþ–p–nstructure is obviously higher than that in the nþ–
n–p structure. Therefore, it is useful to add a
heavily doped layer on the p-side surface for
weakening the influence of the surface recombi-
nation on ðR0AÞtotal.
p+=1018cm-3 p=1017cm-3
1015 1016 1017 10180
20
40
60
80
100
120
140
160
Sp=0(m/s)
102
R0A
(ohm
-cm
2 )
n(cm-3)(b)
on with the n-side: (a) width (t) and (b) surface recombination
1014 1015 1016 1017 1018 1019 102010-2
10-1
100
101
102
103
104
105
n+=5 X1018cm-3 n=1018 cm-3
(R 0 A)Auger
(R 0 A)rad
(R 0 A)GRnp
(R 0 A)GRnn
(R 0 A)tunnel
(R 0 A)total
R0A
(ohm
-cm
2 )
R0A
(ohm
-cm
2 )
p(cm-3)(a)1015 1016 1017 1018 1019
101
102
103
3
2
1
1 n+ =1018cm-3
2 5X1018
3 1019
p=1017cm-3 3
2
1
(R 0 A)Auger
(R 0 A) Rad (R 0 A)
GRnp
(R 0 A) GRnn (R 0 A) total
n(cm-3)(b)
Fig. 6. The dependence of ðR0AÞtotal and its components on: (a) the p-side carrier concentration (p) and (b) the n-side carrier con-centration (n) for the nþ–n–p structure as well as ðR0AÞGRnn with the variety of the nþ-side carrier concentration (nþ).
1014 1015 1016 1017 1018 1019 102010-4
10-3
10-2
10-1
100
101
102
103
108
104
102
Se=0(m/s)
n+=5X1018cm-3 n=1018cm-3
R0A
(ohm
-cm
2 )
p(cm-3)(a)1014 1015 1016 1017 1018 1019 102010-2
10-1
100
101
102
103 n+=5X1018cm-3 n=1018cm-3
d=2µm
5
10
R0A
(ohm
-cm
2 )
p(cm-3)(b)
Fig. 7. The dependence of ðR0AÞtotal on the p-side carrier concentration with the p-side: (a) surface recombination velocity (Se) and(b) width (d) as parameters for the nþ–n–p structure.
T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189 187
10-1 100 101 102 103 104 105 106 107 108101
102
103
104
105
Snn
Spp
Spp(
or S
nn) (
m/s
)Se(or Sp)(m/s)
Fig. 8. The dependence of the interface recombination veloci-
ties Spp and Snn on the Se and Sp for the pþ–p–n and nþ–n–pstructures respectively.
188 T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189
4. Discussion
Now we discuss why the opposite results are
shown in Figs. 3(a) and 6(b), and why Sp and tnþhave not any influence on ðR0AÞtotal in the nþ–n–pstructure. At nþ ¼ 5� 1018 cm�3, Lhnþ is to be
0.0136 lm, and it will continue decreasing with theincreasing of nþ, which can be considered as
Lhnþ tnþ . Therefore Snn in Eq. (2) can be ap-proximated by Snn ¼ nLhnþ
nþsnþwhich is not related to Sp
and tnþ in the nþ–n–p structure (snþ is the minor-ity-carrier lifetime in the nþ-side). Thus, Sp and tnþcould not change ðR0AÞAuger and ðR0AÞrad throughSnn and thereby ðR0AÞtotal could not be affected forthe nþ–n–p structure. In addition, based on Ref.
[8], it can be calculated that snþ is determined bys1Auger in the n-type material [13,14], so that Snn is
further approximated by Snn / n
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1þ n2i
ðnþÞ2
r. In the
range of nþ > 1018 cm�3, there exists nþ ni andSnn / n. This result shows that nþ almost does notaffect Snn, and thereby not affect ðR0AÞAuger andðR0AÞrad in the nþ–n–p structure. However withincreasing the nþ-side carrier concentration,
ðR0AÞGR and ðR0AÞtotal are increased in Fig. 6(b).On the contrary, in the pþ–p–n structure, Lepþ isequal to 31.2 lm at pþ ¼ 1018 cm�3 and to 0.032
lm at pþ ¼ 1021 cm�3. This shows that with in-
creasing pþ, the variety of the relationship between
Lepþ and dpþ is changed from Lepþ dpþ to Lepþ dpþ . Therefore, in the pþ–p–n structure, the influ-ence of each parameter in the pþ-side on ðR0AÞAugerand ðR0AÞrad associated with Spp could not be ne-glected, and correspondingly, ðR0AÞtotal is changedby the variety of the parameters in the pþ-side.
Fig. 8 shows the variety of Spp and Snn with Seand Sp respectively. As above discussions, Snn isnot affected by Sp, which shows a horizontal line inFig. 8. However Spp increases in 10 < Se < 107 m/s.The two limiting values of the surface recombi-
nation velocity are S ¼ 0 and S ¼ 1. The formerimplies an ideal ohmic contact whereas the latter
means an ideal majority-carrier contact. In prac-
tice, the contact at the pþ-region surface of a
pþ–p–n structure is a nonideal majority-carrier
contact and is found to have a value of Se that isfinite and greater than zero. In Fig. 8, although Spp
has a high value (>10 m/s) in Se < 10 m/s, Spp isobviously lower than Se in Se > 10 m/s. This resultdemonstrates that adding a layer will effectivelyreduce the influence of the surface recombination
on GaInAsSb photodetectors. In Ref. [6], the au-
thors have mentioned that the surface leakage
current is an important factor to influence R0A,which increases as the surface recombination is
suppressed. Basing on our simulation results, we
provide a suitable structure to limit the surface
recombination. In addition, utilizing the high-lowjunction can minimize minority-carrier recombi-
nation losses [15]. Therefore the detectors proper-
ties are improved for triple-layer structures.
5. Conclusion
In this paper, four kinds of noise mechanismsare considered for R0A in the nþ–n–p and pþ–p–nGa0:8In0:2As0:81Sb0:19 infrared photodetectors.
ðR0AÞtotal with these noise mechanisms are calcu-lated based on the material parameters in the three
T. Yuan et al. / Infrared Physics & Technology 45 (2004) 181–189 189
layers. The main conclusions can be drawn from
the above study as follows:
(1) In the pþ–p–n structure, the parameters in the
pþ-side affect ðR0AÞAuger and ðR0AÞrad throughSpp and hence ðR0AÞtotal. The high ðR0AÞtotal isobtained with the low carrier concentration
and the surface recombination velocity as well
as the thin width in the pþ-side. In addition,
the thick width in the p-side reduces
ðR0AÞtotal. However, although ðR0AÞtotal has alittle decreasing with the thick width and the
high surface recombination velocity in the
n-side, the optimum value of ðR0AÞtotal couldnot be affected by these two parameters in
the n-side carrier concentration higher than
1018 cm�3.
(2) In the nþ–n–p structure, the nþ-side carrier
concentration is required to be as high as pos-
sible in order to obtain the high ðR0AÞtotal andat the same time, the carrier concentration in
the n-side should be increased, while the widthand the surface recombination in the nþ-side
and n-side almost do not influence ðR0AÞtotal.In addition, the low surface recombination
velocity and the thin width in the p-side is
required to obtain the high ðR0AÞtotal.(3) Compared the influence of the surface recom-
bination velocity in the pþ-side in the pþ–p–n
structure with that in the p-side in the nþ–n–p structure on ðR0AÞtotal, it is found that, exceptthat ðR0AÞtotal in the nþ–n–p structure is a littlehigher than that in the pþ–p–n structure at
Se ¼ 0, ðR0AÞtotal in the pþ–p–n structure is ob-viously higher than that in the nþ–n–p struc-
ture. This result indicates that it is useful to
add a heavily doped layer on the p-side surface
for weakening the influence of the surface re-
combination on ðR0AÞtotal.
References
[1] B. Zhang, Y. Jin, T. Zhou, H. Jiang, Y. Ning, S. Li, Mar.
Res. Soc. Symp. Proc. 415 (1996) 31.
[2] H.K. Choi, S.J. Eglash, Appl. Phys. Lett. 59 (1991)
1165.
[3] A. Aardvark, G.G. Allogho, G. Bougnot, J.P.R. David, A.
Giani, S.K. Haywood, G. Hill, P.C. Klipstein, F. Mansoor,
N.J. Mason, R.J. Nicholas, F. Pascal-Delannoy, M. Pate,
L. Ponnampalam, P.J. Walker, Semicond. Sci. Techno. 1S
(1993) S380.
[4] A.K. Srivastave, J.C. Dewinter, C. Caneau, M.A. Pollack,
J.L. Zysdind, Appl. Phys. Lett. 48 (1986) 903.
[5] B. Zhang, T. Zhou, H. Jiang, Y. Ning, Y. Jin, Electron.
Lett. 31 (1995) 830.
[6] C. Lin, Y.L. Zheng, A.Z. Li, J. Cryst. Growth 227–228
(2001) 605.
[7] Y. Tian, T. Zhou, B. Zhang, Y. Jin, Y. Ning, H. Jiang, G.
Yuang, Opt. Eng. 37 (1998) 1754.
[8] Y. Tian, T. Zhou, B. Zhang, Y. Jin, H. Jiang, J. Phys. D 31
(1998) 3291.
[9] S. Singh, P. Singh, IEEE Trans. Electron. Dev. 38 (1991)
337.
[10] J. Hauser, P. Dunbar, Solid-State Electron. 18 (1975)
15.
[11] J. Alamo, J. Meerbergen, F. D�Hoor, J. Nijs, Solid-StateElectron. 24 (1981) 533.
[12] Y. Tian, T. Zhou, B. Zhang, H. Jiang, Y. Jin, IEEE Trans.
Electron. Dev. 46 (1999) 656.
[13] M. Takeshima, J. Appl. Phys. 43 (1972) 4114.
[14] A. Beattie, G. Smith, Phys. Status Solidi 19 (1967)
577.
[15] A. Lugue, A. Cuevas, J. Eguren, Solid-State Electron. 21
(1978) 793.