Cent. Eur. J. Phys.

9(4) 2011 1114-1121

DOI: 10.2478/s11534-011-0023-6
Central European Journal of Physics
Simulations of AlGaAs/GaAs heterojunction
phototransistors
Research Article
Beata ciana

, Marek Panek, Artur Borczuch, Marek Taczaa

Faculty of Microsystem Electronics and Photonics, Wrocaw University of Technology,
Janiszewskiego 11/17, 50-372 Wrocaw, Poland
Received 19 July 2010; accepted 21 February 2011
Abstract: Heterojunction bipolar phototransistors, based on GaAs technology, are widely used in optoelectronic in-
tegrated circuits. One of the methods to improve phototransistor performance is applying a delta-doped
thin base, which causes both a higher current gain and a better time response. There is a lack of infor-
mation about this kind of device in the literature. Our work presents the results of computer simulations of
AlGaAs/GaAs n-p-n phototransistors, with a bulk-doped and delta-doped base working as two- and three-
terminal devices. The characteristics and device parameters obtained clearly show that phototransistors
with delta-doped base have higher sensitivity and a better time response compared to the structures with
a bulk-doped base. Based on simulation results, we modied the epitaxial phototransistor structure with
a double delta-doped base that should improve device performance.
PACS (2008): 85.30.-z, 85.60.Dw, 07.05.Fb, 61.72.uj, 73.61.Ey
Keywords: phototransistor AlGaAs/GaAs heterostructure delta-doping spectral characteristics response time
Versita Sp. z o.o.
1. Introduction
Heterojunction bipolar phototransistors (HPTs) are a
promising alternative to such devices as PIN and
avalanche photodiodes. Recently, they have been widely
used in optoelectronic integrated circuits (OEIC) as op-
tical switches or preampliers [1, 2]. The HPT transis-
tor can operate without a base bias (oating base) be-
cause the optically generated carriers are responsible for
its work. So, the HPT structures occur as two-terminal
(optical base bias) or three-terminal (optical and electri-

E-mail: Beata.Sciana@pwr.wroc.pl
cal base bias) devices. The main disadvantages of pho-
totransistors are long response times and a small opti-
cal gain, especially at low optical excitation powers. To
solve these problems, dierent HPT constructions have
been proposed: a heterojunction phototransistor with a
punch-through base to reach high gain, high speed, and
low noise simultaneously [3, 4]; a punch-through photo-
transistor with avalanche enhancement [5, 6]; or a guard-
ring design around the active region [7]. In our earlier
papers [8, 9] the n-p-n AlGaAs/GaAs heterojunction pho-
totransistor with a thin 50 nm-thick Zn-delta-doped GaAs
base was described. The use of a delta-doped sheet lo-
cated at heterojunction bipolar transistor interfaces sup-
pressed the heterojunction potential spike, which resulted
in higher emitter injection eciency (current gain) and
1114
Beata ciana, Marek Panek, Artur Borczuch, Marek Taczaa
lower knee voltage. In this work we present the computer
simulations of AlGaAs/GaAs HPT structures, which are
focused on optimisation of the particular epilayer param-
eters (thickness, doping prole, and concentration). The
goal is to improve such device parameters as current gain,
sensitivity, and response times. At the rst step of the
modelling we simulated the phototransistor structure pre-
sented in our work [8], but with a bulk-doped base. The
optimised parameters of this structure were used in the
next step of simulations focused on AlGaAs/GaAs HPT
with a delta-doped base, working with optical base bias
as a two-terminal device or as a three-terminal device
with electrical and optical base bias. As a result of this
work, the modied AlGaAs/GaAs HPT structure with dou-
ble delta-doped base was proposed for fabrication and
experimental application.
2. Simulation assumptions
The software used for the phototransistor modelling is
the Atlas device simulator from Silvaco
1
. This software
simulates the electrical, optical, and thermal behaviour
of dierent semiconductor devices (transistors, light emit-
ters, detectors, solar cells). It allows DC/AC analysis and
time-domain responses for all semiconductor-based tech-
nologies in two and three dimensions. During the simu-
lations the following material and physical models were
used: Fermi-Dirac statistics, SHR (Shockley-Hall-Read)
and the bound radiative recombination, FLDMOB (Paral-
Figure 1. The simulated phototransistor structure.
1
Atlas Users Manual Device Simulation Software
vol. II, Silvaco International, Santa Clara, CA, Feb 2000
lel Electric Field Dependence) model. Device parameters
were calculated using the Gummel-Poon analytical model
for the HPT structures [10]. Spectral characteristics were
derived for wavelengths ranging from 100 nm to 900 nm.
The optical excitation power P
in
of 177 W and collector-
emitter bias U
C|
= 3 V were applied. The time-response
characteristics were determined using pulse-optical exci-
tation ( = 850 nm, P
in
= 177 W) with rise and fall times
0.5 ns and the pulse width 3 s. The simulated structure
is presented in Figure 1. We applied ring-shaped emitter
and base contacts; the collector contact was placed on the
bottom surface of the HPT structure. At rst we assumed
the epitaxial layer parameters from our work [8] (Table 1)
for the bulk-doped base design.
Table 1. Preliminary assumed parameters of the simulated photo-
transistor structure.
Layer Material Thickness (m)
Dopant
concentration
(cm
3
)
Emitter n-Al
0.3
GaAs
0.7
0.40 110
18
Base pGaAs 0.05 110
19
Collector nGaAs 0.55 110
18
3. Simulation results
3.1. Optimisation of the HPT structures with
the bulk-doped base
In the rst step of our simulations we analysed the in-
uence of the particular transistor layers features (thick-
ness, dopant concentration) on the output device param-
eters, such as current gain, response times, and cut-o
frequency. We considered bulk-doped base structures
(AlGaAs/GaAs HPT-bulk) and a classical three-terminal
transistor work without an optical excitation (U
C|
= 3 V,
U
B
= 1.5 V). The main aim of these investigations was
verication of the preliminary assumed parameters, which
in the next step of simulations were applied as the in-
put parameters for modelling of HPT structures with a
delta-doped base. The results indicated that the cur-
rent gain is more sensitive to the collector doping
concentration ^
C
for higher values of the emitter dop-
ing ^
|
. The highest current gain of 828 was achieved
for ^
|
= ^
C
= 1.5 10
18
cm
3
and the base dop-
ing ^
B
= 1 10
19
cm
3
. The inuence of the emitter,
base, and collector doping on the frequency performance
of the AlGaAs/GaAs HPT-bulk structures was analysed
by applying an AC base voltage signal with frequencies
ranging from 1 Hz to 100 GHz. The results indicated
1115
Simulations of AlGaAs/GaAs heterojunction phototransistors
that for constant base doping, higher values of the emitter
doping ^
|
increase the cut-o frequency |
T
of the sim-
ulated transistors. Additionally, the high frequency op-
eration can be improved by applying lower values of the
collector doping (^
C
= I 10
15
I 10
16
cm
3
). The
best values of |
T
= 7.4 GHz and K (current gain in dB)
52 dB were obtained for the following doping concen-
trations: ^
|
= 1 10
18
cm
3
, ^
B
= 1 10
19
cm
3
, and
^
C
= 5 10
15
cm
3
. Next, the inuence of the emitter,
base, and collector thickness (W
|
, W
B
, and W
C
) on the
output parameters of AlGaAs/GaAs HPT-bulk structures
was investigated. The current gain and cut-o frequency
|
T
increase with decreasing base thickness W
B
. It is con-
nected with lowering the base recombination current and
the discharge time of the base minority electron charge.
The best parameters were obtained for W
B
= 50 nm. In
the case of the emitter thickness W
|
, higher values of
this parameter cause current gain increases but the fre-
quency |
T
decreases. The emitter thickness should not
exceed 400 nm. Concerning the collector thickness, for
^
C
= 1.5 10
18
cm
3
, we observe no distinct relation-
ship between this value and device performance for the
simulated range of W
C
(250, 550, and 850 nm). On the
other hand, for a lower value of ^
C
= 5 10
15
cm
3
(bet-
ter high frequency work), the space charge region (SCR)
inside the collector layer increases, causing the absorp-
tion region also to increase. Thus, in this case we use
an optical excitation (P
in
= 177 W, = 850 nm) to op-
timise the W
C
value. The best device parameters are as
follows: current gain of 40 dB, current responsivity (P
|
) of
118 A/W, and the switching times (
on
,
o||
) of 33 ns and
41 ns, respectively, were obtained for W
C
= 2 m. For
higher values of W
C
we didnt observe any changes in the
collector current |
C
, which indicates that this thickness is
comparable with the SCR width.
3.2. Simulations of AlGaAs/GaAs HPT struc-
tures with a delta-doped base
3.2.1. Simulations of AlGaAs/GaAs HPT structures with
a delta-doped base without an optical excitation
At the next step of our investigations, the inuence of
the delta-doped thin base (50 nm) on the current gain K
and the cut-o frequency |
T
was analysed. Simulations
were carried out for the optimised transistor parameters:
^
|
= 1 10
18
cm
3
, ^
C
= 5 10
15
cm
3
, W
|
= 400 nm,
W
B
= 50 nm, W
C
= 2 m, and the bias U
B
= 1.5 V,
U
C|
= 3 V. We chose a low value of the collector doping
(^
C
= 5 10
15
cm
3
) to improve frequency performance.
To get a good ohmic collector contact, we introduced a
thin (W
SC
= 100 nm), high-doped (^
C
= 5 10
18
cm
3
)
subcollector layer at the bottom of the HPT structure. The
AlGaAs/GaAs HPT with a single- and double-delta doped
base were analysed. These structures were assigned as
AlGaAs/GaAs HPT-1 and AlGaAs/GaAs HPT-2, re-
spectively. The acceptor dopant distributions ^
^
= |()
inside the base region for both of the simulated structures
are presented in Figure 2 (Y is the direction normal to
the surface of the HPT structure).
a
b
Figure 2. Acceptor dopant distribution inside the base region for:
a) AlGaAs/GaAs HPT-1; b) AlGaAs/GaAs HPT-2.
We investigated the inuence of the Gauss doping dis-
tribution broadening W

and the distance between two

delta-doped regions W
2
(for AlGaAs/GaAs HPT-2 struc-
ture) on |
T
and K parameters. The results are listed in
Tables 2 and 3.
Data presented in Table 2 show that for AlGaAs/GaAs
HPT-1 the optimal value of the broadening of the
Gauss dopant distribution W

is 10 nm (|
T
= 5.45 GHz,
K = 39.4 dB). In the case of AlGaAs/GaAs HPT-2
structures, the best device parameters (|
T
= 5.39 GHz,
K = 39.4 dB) were obtained for W

= 8 nm and distance
between two delta-doped regions W
2
= 20 nm.
1116
Beata ciana, Marek Panek, Artur Borczuch, Marek Taczaa
Table 2. Inuence of the Gauss doping broadening W

on K and |
T
parameters for the AlGaAs/GaAs HPT-1 structure.
W

(nm) |
T
(GHz) K (dB)
8 breakdown
10 5.45 39.4
12 5.06 38.0
Table 3. Inuence of the Gauss doping broadening W

and sepa-
ration between two delta-doped regions W
2
on K and |
T
parameters for the AlGaAs/GaAs HPT-2 structure.
W

(nm) W
2
(nm) |
T
(GHz) K (dB)
8
30 4.73 38.5
10 5.02 37.9
20 5.39 39.4
10 20 4.70 38.4
3.2.2. Simulations of AlGaAs/GaAs HPT structures with
a delta-doped base under illumination
Spectral characteristics of the collector current |
C
were re-
alised for wavelengths ranging from 100900 nm, the opti-
cal excitation power P
in
of 1 W/cm
2
, and collector-emitter
bias U
C|
= 3 V. The time response characteristics were
determined using pulse optical excitation ( = 850 nm,
P
in
= 177 W) with the rise and fall times 0.5 ns and the
signal width 3 s. We analysed AlGaAs/GaAs HPT-1
and AlGaAs/GaAs HPT-2 structures working as the two-
and three-terminal devices (2T and 3T work). Spectral
characteristics of the resulting collector current |
C
for dif-
ferent ^
|
values determined for AlGaAs/GaAs HPT-1 2T
and AlGaAs/GaAs HPT-2 2T structures are presented in
Figures 3a, 3b. The estimated device parameters cur-
rent gain , collector current |
C
, responsivity P
|
, switch-on
and switch-o times (
on
and
o||
) are listed in Tables 4
and 5. The resulting data show distinctly that, in the case
of a delta-doped base, we can achieve the high current
gain, responsivity, and low values of response times for
Table 4. Inuence of ^
|
on the output parameters of AlGaAs/GaAs
HPT-1 2T structure.
^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
510
17
107 0.0317 179.6 273.0 76.0
110
17
59 0.0175 99.1 126.5 35.1
510
16
67 0.0199 112.7 87.6 30.1
110
16
69 0.0206 116.4 56.2 23.1
510
15
51 0.0151 85.7 50.2 19.2
a
b
Figure 3. Spectral characteristics of |
C
determined for dierent val-
ues of ^
|
and structures: a) AlGaAs/GaAs HPT-1 2T;
b) AlGaAs/GaAs HPT-2 2T.
Table 5. Inuence of ^
|
on the output parameters of AlGaAs/GaAs
HPT-2 2T structure.
^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
510
17
107 0.0317 179.5 256.2 87.3
110
17
105 0.0310 171.6 106.2 44.4
510
16
111 0.0329 186.2 84.2 40.0
110
16
48 0.0143 80.9 55.5 21.1
510
15
31 0.0092 52.3 51.0 16,5
low emitter doping (^
|
= I 10
15
10
16
cm
3
). This is
a main dierence from the AlGaAs/GaAs HPT-bulk struc-
tures, where achieving the high current gain and good time
response for a low value of ^
|
was impossible. Compari-
son of device parameters, estimated for the lowest values
of ^
|
used in simulations of the three types of HPT struc-
tures, is shown in Table 6.
The simulated Gummel characteristics show the changes
of the photocurrent |
pI
(optically generated at the B-C
junction) and the collector current |
C
(amplied |
pI
) ver-
1117
Simulations of AlGaAs/GaAs heterojunction phototransistors
Table 6. Comparison of device parameters of AlGaAs/GaAs HPT-
bulk 2T, AlGaAs/GaAs HPT-1 2T, and AlGaAs/GaAs HPT-
2 2T structures for low values of ^
|
.
HPT 2T ^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
bulk 510
16
11 0.0034 19.1 100.0 28.0
1 510
15
51 0.0151 85.7 50.2 19.2
2 510
15
31 0.0092 52.3 51.0 16.5
sus dierent optical excitation powers. Figures 4 and 5
present the Gummel and output characteristics for the
analysed structures. Gummel relations show a bene-
t of applying a delta-doped base. The gain of pho-
tocurrent appears at lower values of the optical power
P
in
= 6 30 nW than in the case of a bulk-doped base
(0.1 W for HPT-bulk) and achieves a reasonable value
( 30 40) for P
in
= 15 20 W (40 W for HPT-bulk).
a
b
Figure 4. a) Gummel and b) output characteristics of the Al-
GaAs/GaAs HPT-1 2T structure.
The highest value of responsivity was obtained for 820 and
850 nm. The saturation of |
C
occurs at U
C|
= 0.65 V, as in
the case of AlGaAs/GaAs HPT-bulk structures (Figures 4b
and 5b).
a
b
Figure 5. a) Gummel and b) output characteristics of the Al-
GaAs/GaAs HPT-2 2T structure.
At the nal step of our simulations, we investigated the
inuence of optical excitation on the performance of the
AlGaAs/GaAs HPT structures, with a delta-doped base
working as three-terminal devices (optical and electrical
base bias). The analysed structures were assigned as
AlGaAs/GaAs HPT-1 3T and AlGaAs/GaAs HPT-2 3T.
Data listed in Table 7 show the inuence of the emitter
doping ^
|
on the output parameters of the AlGaAs/GaAs
HPT-1 3T structure, estimated for the optical excitation
power P
in
= 53 W ( = 850 nm) and the base current
|
B
= 100 A. The electrical base bias distinctly improves
the response times. It seems that ^
|
= 1 10
16
cm
3
is
the optimal value of the emitter doping that guarantees
the high current gain and responsivity and low values of
response times simultaneously.
Figure 6a, 6b present the Gummel and output character-
istics of the analysed structures. In the case of Gum-
1118
Beata ciana, Marek Panek, Artur Borczuch, Marek Taczaa
Table 7. Inuence of the emitter doping ^
|
on the output parameters
of the AlGaAs/GaAs HPT-1 3T structure.
^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
510
15
44.3 0.0179 101.1 13.4 12.3
110
16
64.7 0.0232 131.4 14.0 14.9
510
16
50.5 0.0278 157.5 26.6 22.1
mel characteristics, determined for |
B
= 100 A and
= 850 nm, we observe two parallel lines connected to
the electrical base bias. The transistor gain ( 83) ap-
pears immediately with the electrical base bias, while for
the two-terminal device, a suitable value of the optical
a
b
Figure 6. a) Gummel and b) output characteristics (|
B
= 50, 100,
300 A; P
in
= 53 W, = 850 nm) of the AlGaAs/GaAs
HPT-1 3T structure.
power is required for the HPT work. Output characteris-
tics were simulated with and without an optical excitation
for dierent base currents (|
B
= 50, 100, 300 A) and for
the optical excitation power P
in
= 53 W ( = 850 nm).
The highest collector current enhancement by the opti-
cal excitation (about four times) is obtained for the lowest
value of |
B
= 50 A (Figure 6b).
Data listed in Table 8 show the inuence of the emit-
ter doping ^
|
on the output parameters of the Al-
GaAs/GaAs HPT-2 3T structure, estimated for the same
optical and electrical base bias. The highest values of
the current gain (109), responsivity ( 202 A/W) and
low values of response times ( 20 ns) were obtained
for ^
|
= 5 10
16
cm
3
, which is a better result than in
the case of AlGaAs/GaAs HPT-bulk 3T and AlGaAs/GaAs
HPT-1 3T structures. Comparison of device parameters,
estimated for ^
|
= 510
16
cm
3
for three simulated types
of HPT structures, is shown in Table 9.
Table 8. Inuence of the emitter doping ^
|
on the output parameters
of the AlGaAs/GaAs HPT-2 3T structure.
^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
510
15
24.1 0.0115 64.9 11.3 12.8
110
16
41.8 0.0165 93.2 13.4 13.3
510
16
109.0 0.0357 202.1 18.3 20.6
110
17
51.5 0.0377 213.6 25.8 26.5
510
17
69.1 0.0446 252.7 74.3 69.1
Table 9. Comparison of device parameters of AlGaAs/GaAs HPT-
bulk 3T, AlGaAs/GaAs HPT-1 3T, and AlGaAs/GaAs HPT-
2 3T structures for ^
|
= 5 10
16
cm
3
.
HPT 3T ^
|
(cm
3
) (A/A) |
C
(A) P
|
(A/W)
on
(ns)
o||
(ns)
bulk 510
16
9.6 0.0037 21.0 14.6 18.5
1 510
16
50.5 0.0278 157.5 26.6 22.1
2 510
16
109.0 0.0357 202.1 18.3 20.6
Figure 7a, 7b present the Gummel and output character-
istics determined for |
B
= 100 A and = 850 nm. The
electrical bias gives a constant gain ( 61) in the whole
investigated range of the optical powers. Output charac-
teristics were simulated for the same electrical and opti-
cal parameters mentioned earlier. We obtained a slightly
smaller value of the collector current ( 22 mA) compared
to AlGaAs/GaAs HPT-1 3T structures ( 32 mA), but
about three times higher than for AlGaAs/GaAs HPT-bulk
3T devices ( 8 mA).
1119
Simulations of AlGaAs/GaAs heterojunction phototransistors
a b
Figure 7. a) Gummel and b) output characteristics (|
B
= 50, 100, 300 A; P
in
= 53 W, = 850 nm) of the AlGaAs/GaAs HPT-2 3T structure.
The presented simulations conrmed our earlier suggestion that applying a delta-doped thin base in a construction of
an HPT transistor distinctly improves its device parameters. This kind of HPT structure is not suciently described in
the literature. Based on simulation results we modied the epitaxial phototransistor structure with a double delta-doped
base, which was described in our works [8, 9] and was not previously optimised. This modied structure is presented in
Figure 8. Lower values of the emitter and collector doping guarantee the high current gain and responsivity, as well as
short response times (lower values of E-B and B-C junction capacities).
Figure 8. The modied epitaxial structure of phototransistor with a double delta-doped base AlGaAs/GaAs HPT-2.
4. Conclusions
This work presents the results of computer simulations of
dierent AlGaAs/GaAs n-p-n phototransistor structures.
Two- and three-terminal devices with bulk- and delta-
doped base were investigated. The resulting character-
istics and calculated device parameters clearly show that
phototransistors with a delta-doped base have higher sen-
sitivity and better response times compared to bulk-doped
structures. They can work as two- and three-terminal de-
1120
Beata ciana, Marek Panek, Artur Borczuch, Marek Taczaa
vices and do not require the high emitter doping. Electri-
cal base bias improves the switch-on and switch-o times
of all investigated structures. The modied epitaxial struc-
ture of the AlGaAs/GaAs HPT-2 phototransistor, which
has not been considered so far, was proposed. It seems
that this will be the optimal structure to apply in fast op-
tical switches of optical communication systems and logic
circuits.
Acknowledgments
The work was partially supported by the Polish Min-
istry of Science and Higher Education under the grant
no. NN 515360436, and by the European Union within
the European Regional Development Fund through an In-
novative Economy grant (POIG.01.01.02-00-008/08); and
by the Wroclaw University of Technology statutory grant.
References
[1] H.T. Lin, D.H. Rich, A. Larsson, J. Appl. Phys. 79, 8015
(1996)
[2] L. Chien-Chung, M. Wayne, J.S. Jr. Harris, F. Sugi-
hwo, Appl. Phys. Lett. 76, 1188 (2000)
[3] Y. Wang, E.S. Yang, W.I. Wang, J. Appl. Phys. 74, 6978
(1993)
[4] H. Luo, H.K. Chan, Y. Chang, Y. Wang, IEEE Photonic.
Tech. L. 13, 708 (2001)
[5] J.C. Campbell, A.G. Dentai, G.-J. Qua, J.F. Ferguson,
IEEE J. Quantum Elect. QE-19, 1134 (1983)
[6] J.-W. Shi, Y.-S. Wu, F.-C. Hong, W.-Y. Chiu, IEEE
Electr. Device L. 29, 714 (2008)
[7] H. Dejun, L. Guohui, Y.F. Zhang, E.-J. Zhu, IEEE Pho-
tonic. Tech. L. 9, 1391 (1997)
[8] B. ciana et al., J. Cryst. Growth 310, 5227 (2008)
[9] J. Kovc et al., Microelectr. J. 40, 562 (2009)
[10] S.M. Frimel, K.P. Roenkera, J. Appl. Phys. 82, 3581
(1997)
1121