You are on page 1of 4

ME.

l Comparison Between InP and Other Semiconductor, Materials for the Realization of Millimeter Wave Two Terminal Devices
P.A. ROLLAND, M.R. FRISCOURT, C. DALLE, D. LIPPENS

Centre Hyperfrequences e t Semiconducteurs - U.A. 287 CNRS - U.S.T.L.F.A. 59655 VILLENEUVE D'ASCQ CEDEX - FRANCE Abstract This paper reviews the advance made in the performance of m m . wave two terminal devices, mainly Gunn, IMPATT and resonant tunneling diodes, on the basis of the semiconductor material systems used (Si, GaAs, InP).
1. Introduction

Most recent semiconductor growth techniques now allow the fabrication of quite sophisticated multiple heterojunction three terminal devices that begin to put the heat on two terminal devices even i n the ".wave range where they have established their place. So it appears interesting to review the recent advance in the performance of these ".wave two terminal devices made by using new materials or applying band-gap engineering when possible. Gunn, IMPATT and resonant tunneling diodes will be considered.

II.GUNN DIODES

- The so called G u n n diode is a Transferred-Electron Device which has been proven to be a reliable microwave and millimeterwave solid state frequency source. Its operating mechanism basically relies on the Transferred-Electron Effect which depends on the bulk properties of the semiconducting material from which it is grown.
The intrinsic 3 dB cut-off frequency of the negative differential mobility is of about 4 0 GHz for GaInAs, 1 0 0 GHz for GaAs and 200 GHz for InP which clearly shows the interest of InP for the realization of fundamental mode TED'S in the millimeter wave range.
A conventional Gunn diode consists of a N + N N + structure. In the mm-wave range only 50 % of the N zone support the propagation of the accumulation layers that are responsible for RF power generation. The existence of a lossy dead zone has not only a drastic influence on the RF performance since it dissipates nearly half of the generated power but in addition increases the bias voltage and temperature sensitivity of the device [l].

The best experimental results published so far for these conventional devices are 150 m W output power with 3.6 % efficiency for InP at 94 GHz [2] and 96 m W with 2.7 % efficiency at 94 GHz also for GaAs [3]. InP devices were operated in a fundamental mode while GaAs diodes were operated as second harmonic generators. These results confirm the better power-frequency behaviour of InP but also indicate that GaAs is a good competitor for the 1 0 0 m W level in the W band. However the full benefit of GaAs and InP can only be realized by removing the dead zone. This can be achieved by several techniques : graded heterojonctions [41, planar doped barriers [5], Camel diodes [6], Schottky barrier [7] and tunnel injection 181. Camel diode injectors were successfully used for GaAs second harmonic devices and yield 7 0 mW output power and 2 % conversion efficiency in the W band [9]. Further improvements will be probably achieved with this technique which can be applied to InP as well.
80

Graded GaAlAs launchers were also successfully used for GaAs G u n n devices [ l O J (111. Significant improvement was observed in efficiency of second harmonic GaAs diodes using this technique. A maximum output power of 80 mW was achieved at 94 GHz with maximum efficiency of 2.4 % [12]. Such a forward biased launcher is not possible in the InP material system but hot electron injection can be achieved using reverse-biased GaInAs/lnP heterojunction or M / l n P current limiting cathodes. The poor conduction bandgap of the GalnAs/InP system (AEc - 0.15 0.2 eV) limits the interest of such a graded heterojunction limiting cathode contact while the low barrier height of a Schottky barrier contact on InP allows a range of current limiting values which can approach the optimum 3 0 % current injection ratio. Despite the difficulty t o engineer Schottky barrier height this kind of current limiting cathode is theoretically the most interesting one with maximum predicted RF performance of 300 mW and 9 % a t 1 0 0 GHz. Only Varian has succeeded today in this complex contact processing and has obtained significant results : output power in excess of 150 mW and 4.7 % efficiency. GaAs and mainly InP TED's are far from having reached their ultimate limits. Bandgap engineering and planar doped techniques are just begining for hot electron injection in ".wave TED's. Further significant improvements can be expected in the next future with GaAs devices covering the 1 0 0 m W requirements and InP the 2 0 0 mW medium power applications.
IMPATT DIODES

IMPATT diodes are important for mm-wave systems since the remain the most powerful1 solid state device. Conventional devices are mostly built as P+PNN+ structure. As opposed t o G u n n diodes no characteristic cut-off frequency of the bulk properties can be a priori pointed out when comparing the basic materials from which IMPATT diode can be grown, mainly Si, GaAs and InP. So a realistic comparison of the possible materials must account for all the self-consistent operating effects (transport properties, thermal limitation, electronic limitation, circuit matching...). In addition ".wave IMPATT's are vertical highly submicronic devices which renders their optimum design and fabrication critical. Most of the efforts devoted to IMPATT's have tend to minimize the voltage drop across the avalanche zone which ideally does not contribute to the RF power. This can be achieved by keeping the avalanche zone width very thin as compared to the transit ones, or by using a low-field ionizing material for the avalanche zone. In the first case a Double Read structure can be realized using doping modulation technique, in the second case heterostructure can be grown using GaInAs/InP or Si/Ge material system for example. But two important effects limit this confined avalanche zone concept, i n Read structures and heterostructures as well. The P and N doping level of the drift zones must be kept at rather low values in order to maintain a sufficient electric field in these drift zones. This limits both the current density and RF voltage swing because of space charge effects. In addition if the avalanche zone is strictly confined by heavily doped spikes in Read structures or by the use of heterostructures, then the carriers generated by impactionization leave the avalanche zone well before the half period of the terminal RF voltage. This yields a premature collapse of the avalanche process and a drastic reduction of the RF performance. Self consistent theoretical comparisons between Si, GaAs and InP Double Drift IMPATT diodes RF performance in the ".waverange have shown that Silicium emerges as the best material for ".wave IMPATT's. GaAs diodes are strongly electronically limited by their low saturated drift velocity at high temperature and high fields, while InP diodes are more drastically thermally limited due t o their poor ionization rate [13].
81

Figure 1 presents state-of-the-art performance experimentally achieved with Gunn and IMPATT diodes. It confirms t h e preeminence of Silicium IMPATT's as high power solid state sources and that of InP as medium power local oscillators in the ".waverange as long a s v e r t i c a l c u r r e n t flow t r a n s i s t o r s t r u c t u r e s will come against technological problems

10

10
0

PULSED
0

RESONANT TUNNELING DIODES

z
K
W

%Am-

The resonant tunneling diode first n lo-' demonstrated in 1974 by Esaki consists of two potential barriers in series separated by a potential well. The barriers are thin enough that electrons can tunnel through them into and out of the quantum well. This structure can be 10-2 compared to a Fabry-Perrot optical interferometer. The requirements for ".wave operation are :
10-3

o Si
aGa

. IMPATT

O' I ' ,

\
t '

mInP

OGaAs

GUNN

a
I
1 10

1
100

I 1000

high peak to valley ratio high peak current density - low resistance ohmic contacts.
-

FREQUENCY (GHz)

Pin. 1 : State of the art performance of GUNN and IMPATT diodes.


The need of high current density imposes very thin barriers (10 to 4 0 allows the use of lattice matched or pseudomorphic materials. The crucial parameters for RTD's are :

A)

which

- high conduction band discontinuity (0.5 to 1 eV) - low effective mass - high quality growth.
Table 1 below summarizes the main characteristics of the most interesting RTD's that have been grown so far. This table indicates that Double-Barrier tunnel devices have been a subject of great interest these lost years and that if GaInAs/AlAs have exhibited the highest peak to valley ratio at room temperature new promissing InAs/AlSb structures grown on GaAs have been recently proposed.

82

RESONANT TUNNELING DIODES STRUCTURES GoAs/GoAIAs InPIGaInAs

I $~~~~~~~~~~~~ I
AEc AEc

PERFORMANCES

- 0.3 eV

I
COMMENTS First RDN at 300'K 1 4 11

JdJ.

= 4 a t 300'K

- 0.15to -0.2eV

Jv'J, = 1.2 10 kA/cm: Tunneling for holes I151

Alo 4dno s2Ad Cao 4 7 h 53As

m'horrier =

AEc = 0.5eV 0.072 I I I * ~ ~ I= 0.04 I

JdJ,

JdJv- 6 at 300"K
J

- 30 at 77K

Direcl-semiconductors latlice-matched Unsymetrical barriers on n + I n P substrate [161

In0 ~ ~ ( G ~ I - ~ A I ~ AEc 5 0.5eV ) As/lno.szGao 42AS x < l

- 100 kA/cmz

AlAs/Cao,47In0 52As

m*&G-s= 0,042
AEc = 0.6 GalnAs-InAs

AEB = 2.27 x 0.6 m A I A ~- 0.14

II
J,
J&J, = 5 P = LJl- Ilk)

Pseudomorphic RTD

I
-3

InP substrate I171

AlAdCalnAslInAs

105 Akmz

InAdAISB

AEg = 2.32-0.41 1 9 = .

Lattice-matc ed a = 6.081 new materia alnp = 5.87 on GaAs substrate (191

Table 1
REFERENCES
[l] M.R. FRISCOURT et al. IEEE Trans. o n Electron Devices, Vol. ED-30, no 3, March 8 3 , pp.

[2]

223-229. M.A. DI FORTE-POISSON et al. Proc. of t h e 1st i n t e r n a t i o n a l c o n f e r e n c e on InP a n d r e l a t e d m a t e r i a l s f o r advanced e l e c t r o n i c s and o p t i c a l devices. 20-23 March 89, Norman, O k l a o m a (USA). [3] S.J. JEROMETENG e t al. IEEE Electron Device L e t t e r s , Vol. 10, no 9, Sept. 8 9 , pp. 4 1 2 414.

[4] T. J U D D et al. Applied Physics L e t t e r s 4 9 , p. 1652, 1986. [5] R. MALIK et al. Electronics L e t t e r s , 16, p. 836, 1980. 163 J.M. SHANNON. Applied Physics L e t t e r s 35, p. 63, 1979. [71 C.R. CROMWELL and S.M. SZE. Solid State Electronics 9 , p. 1035, 1966. [8] M. HEIBLUM et al. Physics Review L e t t e r s 55, p. 2200, 1985. 191 R.B. BEAL et al. E l e c t r o n i c s L e t t e r s , 13, p. 871, 1989. [ l o ] N.R. C O U C H et al. Solid State Electron. 1988, 31, p. 613. Ill] Z. GREENWALD et al. Solid S t a t e Electron, 1988, 31, p. 1211. [12] H. SPOONER et al. GEC J. Research, Vol. 7, 1989, p. 34. [13] D. LIPPENS et al, 1 0 t h Intern. Conf. on Infrared and millimeter wave, Florida, 9-12 Dec. 1985. [14] C.I. HUANG et al. Appl. Phys. L e t t . 51, 1987, p. 121. [15] M. RAZEGHI e t al. Electron. Lett. 23, 1987, p. 116. [ l 6 ] Y. SUGIYAMA et al. Appl. Phys. L e t t . 52, 1988, p. 314. [17] T. INATA et al. J a p a n e s e Journal of Appl. Phys. 26, 8, 1987, p. L1332. [18] T. BROEKAERT et al. Appl. Phys. Lett. 53, 1988, p. 1545. [19] J.R. SODERSTROM et al. J. Appl. Phys. 66(10), 1989, p. 5106.
83

You might also like