Documente Academic
Documente Profesional
Documente Cultură
6 JUNE 2000
938
INVITED PAPER
Special Issue on Advanced Optical Devices for Next Generation Photonic Networks
1.
Introduction
The authors are with NTT Network Innovation Laboratories, Yokosuka-shi, 239-0847 Japan.
a) E-mail: ishi@aecl.ntt.co.jp
939
Fig. 2
tion layer, and is usually frequency dependent. Because holes have a much lower drift velocity than electrons, their transport dominates the space charge eect
and saturation behavior in a pin-PD. The uni-travelingcarrier photodiode (UTC-PD) was designed to oer
higher operation current densities [6]. The benets of
UTC-PDs are created from their use of electrons as the
only active carriers. Here, hole transport does not directly aect the diode response and output saturation
mechanisms. InP/InGaAs UTC-PDs have been fabricated with various diode dimensions and demonstrated
to produce signicantly higher output.
This paper reviews the operation, design, and performance of the UTC-PD. We discuss how its photoresponse diers from that of the pin-PD. The application
of InP/InGaAs UTC-PDs as optoelectronic drivers is
also presented.
2.
Band diagrams are schematically shown for a UTCPD and a pin-PD in Fig. 2. The active layer of the
UTC-PD consists of a neutral (p-type) light absorption layer and a widegap (depleted) carrier-collection
layer. Carriers are photogenerated in the absorption
layer, and minority electrons diuse/drift into the collection layer. Here, the diusion block layer gives
electrons unidirectional motion toward the collection
layer. The quasi-eld formed by the band gap grading and/or doping grading is very eective in reduc-
ing electron traveling time. Since holes are the majority carriers, their transport behaves depending on the
electron current. Consequently, the photoresponse of a
UTC-PD is determined by the electron transport in the
whole structure. In a typical UTC-PD structure with
similar absorption layer thickness WA and collection
layer thickness WC , the absorption layer traveling time
A is dominant because the diusive velocity is usually smaller than drift velocity in the collection layer
with high electric eld. As in the case of heterostructure bipolar transistors [7], electrons represent velocity overshoot in the collection layer, where the velocity
(vos 4 107 cm/s) is far greater than hole saturation
velocity (vs 5 106 cm/s).
In a pin-PD, on the other hand, both electrons
and holes in the depleted absorption layer contribute
to the response. When electrons and holes travel independently, the output response is sum of both current
components. Here, due to the signicant dierence in
carrier velocities, the 3-dB-down bandwidth is determined by the hole transport.
Since the carrier generation and collection are separated in space, the total carrier travel distance of a
UTC-PD is relatively long. Lets compare a pin-PD
with the absorption layer thick ness of WA to a UTCPD with WA = WC . Under uniform carrier generation
over the absorption layer, the average carrier traveling
distance is 0.5 WA in the pin-PD, while it is 1.5 WA
in the UTC-PD. In spite of this drawback, the total
carrier travel time of a UTC-PD, which is determined
by the electron transport, can be similar to or a little
shorter than that of a pin-PD. This condition is realized in InP/InGaAs UTC-PDs designed for a long wavelength operation. Here, the minority-electron mobility
e (diusivity De ) is a key parameter. We should note
that the eective electron diusion velocity, 3De /WA
in InGaAs, can be higher than the hole saturation velocity vs . At a limiting case where the collection layer
travel time C (= WC /vos ) dominates (or WA = 0), the
bandwidth f3dB is simplied to 2.8/(2Celectron ) for a
UTC-PD, while it is given by 3.5/(2Chole ) for a pinPD (Chole = WC /vs ). Due to the large dierence in
carrier velocities, the UTC design can nally provide a
shorter C and superior frequency response.
3.
940
Fig. 4
(1b)
jR Je (x)
[Je (x) + Jh (x)] =
x
1 + jR x
(3)
where R is dielectric relaxation time / in the absorption layer. In deriving Eq. (3), we ignore the hole
diusion current and take Jh (x) to be E(x). Equation (3) just indicates that the continuity of particle
current can be maintained at a frequency 1/R
and, to achieve this condition, the absorption layer doping has to be high enough. On the other hand, electron
current in the collection layer is regarded as current
induced in a parallel capacitor. The frequency photoresponse is the total (short circuit) photocurrent Jtot
941
942
Fig. 7
of WA .
i-InGaAsP Spacer
16
1.0
i-InP Spacer
6
7
1.0E18
n+ -InP Cli.
n-InP Collector
263
1.0E16
50
5.0E18
n+ -InP Subcollec.-2
n+ -InGaAs Etch Stop
10
1.5E19
500
1.5E19
n+ -InP Subcollec.-1
i-InGaAs Etch stop
10
De = kT e /q. The inuence of the maximum diusive velocity is not negligible as WA decreases. For
WA values larger than 100 nm, A tends to change as
WA2 indicating classical diusive transport. However,
this diusive mode becomes less valid as WA falls; A
changes more gradually as WA /vth determines it, not
the electron diusivity (doping level).
5.
InP/InGaAs UTC-PDs have been fabricated with MOVPE (metal-organic vapor-phase-epitaxy) grown epitaxial materials. A typical structure uses p+ -InGaAs
and n -InP as the absorption- and collection-layers,
respectively (Table 1). In layer design, an important
point is the interface between absorption/collection
layers. Due to the conduction band discontinuity,
240 meV at the InP/InGaAs heterointerface, the InP
potential spike can produce current blocking unless
the band prole is designed carefully. This problem can be most eectively eased by adopting compositional grading combined with planar doping [10],
[11]. Additional intermediate layers of i-InGaAs/iInGaAsP(EG = 1.0 eV)/i-InP/n+ -InP are inserted between the p+ -InGaAs absorption- and n-InP collectionlayers. The thickness of each layer and the doping level
for n+ -InP are optimized by numerical simulation so
that a smooth conduction band change is realized at
the junction voltage used. Figure 8 presents simulated
band proles for structures with and without compositional grading. It is clearly seen that the use of the
i-InGaAsP layer can reduce not only the potential spike
height, but also the potential notch depth.
A double-mass structure is used for fabricating
InP/InGaAs UTC-PDs. In processing back-illuminated
planer devices, conventional metal lift-o for Ohmic
contacts, wet chemical etching, and electrode-isolation
by polyimide lm have been used. Edge-illuminated
waveguide devices have also been reported [12]. The
responsivity is generally the same as that of a pin-PD
as far as each absorption layer thickness is equal. When
applying UTC-PDs to photoreceivers, they are integrated with a 50- coplanar waveguide (CPW) line,
a bias capacitor, and a matched resistor. Figure 9 illustrates a planar device used for pulse-photoresponse
943
Fig. 9 Device geometry of a planar device used for pulsephotoresponse characterization by means of the electro-optic
sampling (EOS) technique.
Photoresponse of UTC-PDs
944
(6)
945
the Ipeak -Vop curve to the negative Vop side, which results in a narrower active region. However, this eect
was negligible in the fabricated device because RS was
only a few Ohms. The maximum peak current obtained
was 184 mA (averaged 460 kA/cm2 ) with an FWHM of
4.8 ps (Vout = 4.6 V on a 25- load).
6.5 On the Bias Dependence of Electron Velocity in
InP
Electron velocity overshoot plays an essential role in
UTC-PD operation, especially in reducing the space
charge in the collection layer and elevating saturation
output. Because an input optical pulse is very short
(subpicosecond), EOS measurements on UTC-PDs can
give more direct information for the overshoot than
microwave measurements on InP/InGaAs/InP doubleheterostructure bipolar transistors. Figure 16 represents the pulse photoresponse curves at various bias
voltages measured for a UTC-PD with WA = 80 nm
and WC = 400 nm [16]. The optical input level was
946
The features of UTC operation mode can be summarized as: high current density, broad bandwidth, and
low operation voltage. In conguring a high bit-rate
broadband photoreceiver without electrical ampliers,
for example, a basic requirement is a short enough CR
time constant (determined by junction capacitance and
load resistance). UTC-PDs permit us to use a smaller
junction size for generating a certain output signal level
(1 Vpp /20 mApp for driving digital circuits), which
means that the possible bandwidth can be expanded
(e.g. Ipp = 30 mApp at f3dB = 174 GHz, [17]). Because
the operation voltage can be low, close to 0 V, a large
voltage swing is allowed even at a low bias voltage.
Furthermore, in order to achieve stable and reliable operation, a low power dissipation (a low junction temperature) is needed. Capability of the low operating
voltages can also answer this requirement.
A UTC-PD module has been used to directly drive
a D-type ip-op circuit as the receiver of a 40-Gbit/s
time-division multiplexing system [18][20]. Eye diagrams of receiver outputs are shown in Fig. 17 for
40 and 80-Gbit/s signals [18], [21]. Compared to the
case of post amplier utilization [22], [23], the input
sensitivity has been much improved. Millimeter-wave
generation with a UTC-PD has been demonstrated
at 40 and 60 GHz [24], [25]. Optical signals produced
by mode-locked laser diodes have been eciently converted to millimeter-wave signals. The highest power
was 12 dBm at 60 GHz with a photocurrent of 30 mA.
Fig. 17
947
[2]
8.
[3]
Conclusion
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Acknowledgement
The authors would like to thank N. Watanabe for epitaxial growth, Y. Sato for his assistance in device fabrication, and Y. Matsuoka, S. Mitachi, K. Yamazaki,
Y. Ishii, H. Toba and H. Kyuragi for their continuous
support.
[16]
[17]
References
[18]
[1] K. Hagimoto, Y. Miyamoto, T. Kataoka, H. Ichino, and
O. Nakajima, Twenty-Gbit/s signal transmission using
948
Tadao Ishibashi
received the B.S.,
M.S. and Ph.D. degrees in applied physics
from Hokkaido University in 1971, 1973,
and 1986, respectively.
Since joining
NTT Laboratories, Musashino, Tokyo, in
1973, he has been involved in research
of semiconductor devices and related material processing.
His work included
submillimeter-wave Si IMPATT diode oscillators, LPE growth of InP/InGaAs materials and their application to eld effect transistors, MBE growth of MQW laser diodes, and GaAsbased/InP-based heterostructure bipolar transistor ICs. He is
currently working on ultrahigh-speed optoelectronic devices and
their integration. During 1991 to 1992, he stayed at Max-PlanckInstitute, Stuttgart, as a visiting scientist. Dr. Ishibashi is a
member of the Japan Society of Applied Physics, and the Institute of Electrical and Electronics Engineers (IEEE). He received
the Ichimura Award in 1992 for the development of ballistic collection transistors.
Tomofumi Furuta
was born in Tokyo, Japan, on March 10, 1958. He received the B.S. and M.S. degrees in electrical engineering from Tokyo University
of Agriculture and Technology, Japan,
in 1981 and 1983, respectively, and the
Ph.D. degree from the University of Tokyo in 1986. In 1986, he joined the NTT
LSI laboratories, Kanagawa, Japan. He
has been engaged in the research of semiconductor physics and devices. Dr. Furuta is a member of The Japan Society of Applied Physics and
The American Institute of Physics.
949
Hiroshi Fushimi
was born in Tokyo, Japan, on May 27, 1964. He received B.S. and the M.S. degrees in electrical engineering from the Nagaoka University of Technology, Nagaoka, Niigata,
Japan, in 1987 and 1989, and the Ph.D.
degree in electrical engineering from Keio
University, Yokohama, Kanagawa, Japan,
in 1998, respectively. He joined NTT
Transmission Systems Laboratories, Nippon Telegraph and Telephone Corporation, Yokosuka, Kanagawa, Japan, in 1989, where he was engaged
in research on high-speed GaAs ICs and coherent lightwave transmission systems. In 1993, he moved to NTT LSI Laboratories,
Atsugi, Kanagawa, Japan, where he has been engaged in research
on defect physics and electronically enhanced defect reactions in
semiconductors. He works now at NTT Photonics Laboratories,
Atsugi, Kanagawa, Japan. Dr. Fushimi is a member of the Japan
Society of Applied Physics, the Surface Science Society of Japan,
and the Japanese Association of Crystal Growth.
Satoshi Kodama
was born in Ohtsu,
Japan, in 1968. He received the B.S. degree in information engineering, and M.S.
and Ph.D. degrees in electrical engineering from Hokkaido University, Sapporo,
Japan, in 1991, 1993, and 1996, respectively. In 1996, he joined LSI laboratories,
Nippon Telegraph and Telephone Co., Atsugi, Japan. He is currently a Research
Engineer at Photonics Laboratories, Nippon Telegraph and Telephone Co., Atsugi, Japan. His research interest is integrated advanced optelectronics devices. Dr. Kodama is a member of the Japan Society of Applied Physics.
Hiroshi Ito
received the B.S. and
M.S. degrees in physics, and Ph.D. degree in electrical engineering, from Hokkaido University, Japan, in 1980, 1982,
and 1987, respectively. Since joining NTT
Laboratories in 1982, he has been involved
in research on growth and characterization of III-V compound semiconductors
using MBE and MOCVD, and their applications to devices such as heterojunction bipolar transistors (HBTs), eld effect transistors, lasers, and photodiodes. From 1991 to 1992, he
was at Stanford University as a visiting scientist. His current
research interests focus on ultrafast photonic devices and III-V
materials. Dr. Ito is a member of IEEE, the Physical Society of
Japan, and the Japan Society of Applied Physics.
Tadao Nagatsuma
received the
B.S., M.S. and Ph.D. degrees in electronic engineering from Kyushu University, Fukuoka, Japan, in 1981, 1983, and
1986, respectively. He joined the Nippon Telegraph and Telephone Corporation (NTT), Atsugi Electrical Communications Laboratories, Kanagawa, Japan,
in 1986, and currently is a distinguished
technical member, senior research scientist, supervisor at NTT Telecommunications Energy Laboratories. His current research involves highspeed electronics and millimeter-wave photonics, and their application to sensors and communications. Dr. Nagatsuma is a member of the Optical Society of America, the IEEE, and the Japan
Society of Applied Physics. He was the recipient of the 1989
Young Engineers Award from IEICE, the 1992 Andrew R. Chi
Best-Paper Award from IEEE, the 1997 Okochi Memorial Award,
and the 1998 Japan Microwave Prize.
Naofumi Shimizu
was born in
Osaka, Japan, in September 1962. He received the B.E. and M.E. degrees in engineering physics from Kyoto University,
Kyoto, Japan, in 1986 and 1988, respectively. In 1988, he joined NTT LSI Laboratories, Kanagawa, Japan. He was engaged in research and development on IIIV high-speed devices. Since 1998, he has
been with Network Innovation Laboratories, where he has been engaged in research on high-speed lightwave transport systems.
Yutaka Miyamoto
was born in Tokyo, Japan, on December 8, 1963. He
received the B.E. and M.E. degrees in
electrical engineering from Waseda University, Tokyo, Japan, in 1986 and 1988,
respectively.
In 1988, he joined the
NTT Transmission Systems Laboratories,
Yokosuka, Japan, where he engaged in
research and development on high-speed
optical communications systems including
the 10-Gbit/s terrestrial optical transmission system called FA-10G system. He is currently a senior research engineer of NTT Network Innovation Laboratories, Yokosuka, Kanagawa, Japan. His current research interest includes
ultra-high-speed communication systems and their related devices. Mr. Miyamoto is a member of the Institute of Electrical
and Electronics Engineers (IEEE). He received the best paper
award of the rst optoelectronics and communication conference
(OECC96) in 1996 from the IEICE.