Development of wafer annealing for semi-insulating InP

Substrate

Feature

Development of wafer annealing for semiinsulating InP by M. Uchida, Japan Energy Corporation I

Semi-insulating (SI) InP is becoming more and more important not only for high frequency devices such as HEMTs and HBTs but also for opto-electronic devices such as laser diodes and photodetectors in optical fibre communication systems with high data transmission rate exceeding 40 Gbs-1. This trend needs high-quality SI InP, especially with extremely low Fe concentrations. Japan Energy Corporation (following on the tradition of Nippon Mining) has developed a commercially applicable wafer annealing process to prepare SI InP for this purpose.

S

emi-insulating InP is a very important material for high-frequency electronic devices such as HEMTs and HBTs. These high-frequency InP-based devices are competing with those based on GaAs, and are believed likely to become the principal devices operating in the very high frequency range exceeding 60 GHz. The applications of these high frequency devices are expected in the field of automobile anti-collision systems and broadband integrated service digital network systems (B-ISDN) and so on. These electronic devices are also indispensable for the realization of OEICs for optical fibre communication systems. It is also likely that SI InP may replace the conductive suhstrates presently in use for o p t o e l e c t r o n i c devices such as laser diodes (LDs) and photo diodes (PDs) in optical fibre communication systems. In these examples, high data transmission rates exceeding 40 Gbs -1are necessary in order to realize multimedia communicat i o n s w i t h m o v i e s , a u d i o and documents. Hitherto, LDs and PDs for optical fibre communication systems have been based on conductive InP such as S-, Sn- or Zn-doped InP substrates. These substrates cannot be

32

0.0002 l0 9 10 8 10 7 10 6

0.02

waferannealingunder ~ ~ ! phosphorus vapor pressure

10 5 10 4 10 3 ~D

0.002

102 I 101 10 o 10-1 1013

i

• •

2.0 (ppmw)

02 OI !

.~.

[

.

'

undopedsemiinsulating InP

conventionali,

1

i carrierconcentrationof :,~ undopedInP i

1014

:

1015

[ii i i 81ii~i Eii!iiiiii!i!!~!

1016

1017

Fe concentration (cm -3) Figure 1. Relationship between resistivities and Fe concentrations.

applied for devices with the transmission rates higher than 40 Gbs -1due to the large junction capacitance at the interface between epitaxial layers and substrates. The importance of SI InP is thus increasing because it may become the main material in InP for all applications in electronic and optoelectronic devices. SI InP is industrially produced by compensating shallow donors with deep acceptor impurities, Fe, with concentrations in the range of 1.0-10

III-Vs Review • Vol.10 No.2 1997 0961-1290/97/517.00 ©1997, Elsevier Science Ltd

× 1016cm -3.When the Fe concentration is less than 1.0 x 1016 cm 3, semiinsulating p r o p e r t i e s c a n n o t be achieved. However, Fe-doped InP has various disadvantages:•

nonuniformity of electrical properties along crystal growth axis due to Fe segregation [1]

•

the displacement of Fe by Mg or Zn dopants diffusing into Fe doped InP substrates during the MOCVD process [2]

•

low electrical activation efficiency after ion implantation due to residual Fe atoms in the implanted layers [3].

In the case of InP, it had been believed that undoped semi-insulating InP can not be realized. Since the band gap of InP is 1.4 eV, it is necessary that the carrier concentration due to impurities and defects must be lower than 107 cm -3. This low level impurity concentrations however is in reality too

pure to realize by the-state-of-the-art technologies. Therefore, the intrinsic SI InP is substantially impossible to prepare. There are neither any deep levels as EL2 in GaAs, which is indispensable for undoped SI GaAs in its compensating mechanism. It was found, however, that undoped InP can be converted to semiinsulating by heat treatment. Undoped high resistive InP was reported for the first time in 1986 by Klein et al. [4].

Regardless of whether we "re talking about cellular p,hone or space communication, silicon carbide makes it possible to develop more robust components for higher power, higher frequencies, a.~d for more powerful aj~lications.

Figure 2. Photograph of the high pressure annealing furnace.

You can see the opportunities offered by silicon

1010 !

,

10 9 }

,

25atm

carbide. We listen to your

,

needs and tailor make the

latm

equipment you need for

10 7F~ 106 10 5

, , , ~ --

' . . I I ' •

tomorrow's silicon carbide electronics.

-----1-. -

Contact us and we will tell

• 1o4|

you more. Start by reading

10 3 [ 102[

.I.

J"

our World Wide Web pages.

.

101[-

10 o[ 10"10

5

950

C,,40hrl0Annealing15

m m m

Run Figure 3. Reproducibility of SI InP preparation after wafer annealing under phosphorus vapour pressure.

Ideon Science & Technology Park S=-223 70 Lund, Sweden Telephone +4646 16 89 80 Fax +46 46 16 89 81 www http://www.epigress.se

Substrate Feature

109

I

'

I

i

I

'

I

i

I

decreased to the level less than the concentration of activated Fe [9].

i

•

Shallow donors due to the complex defect (hydrogen-indium vacancy) can be eliminated by annealing to a lesser extent than the concentration of residual Fe [10].

•

Indium-related defects which act as shallow acceptors are created u n d e r low p h o s p h o r u s v a p o u r pressure annealing and they compensate residual shallow donors

E

1o

L

~

~

A

i

A V

._~ 7 •~

10

GZ

Resistivity Variation = 8.3 % 10

6

,

I

i

I

-20

-30

i

-10

I

i

0

I

i

10

I

[lt].

i

20

30

Distance from the Center (mm) 7000

,

I

'

I

i

I

'

I

'

I

'

o

~5000 E

'

3000 Mobility Variation = 2.9 %

0

1000 -30

I

-20

,

I

I

-10

I

0

i

I

10

t

I

20

30

Distance from the Center (mm) Figure 4. Typical electrical properties of 50 mm diameter SI InP after MWA (950°C, 1 atm, 40 hrs + 807°C, 30 atm, 40 hrs) process.

They reported that undoped high resistivity InP (3.6 x 10s ohm cm) was o b t a i n e d after a n n e a l i n g at 900940°C u n d e r a p h o s p h o r u s v a p o u r pressure of 6 bar for about three weeks. In 1989, H o f m a n n etal. [5] reported that the resistivity h i g h e r than 107 ohm cm could be obtained by annealing undoped conductive InP at 900°C under a phosphorus vapour pressure of 5 atm, and Kainosho etal. [6] obtained wholly SI 50 mm InP by annealing u n d o p e d conductive InP at 900°C u n d e r a p h o s p h o r u s v a p o u r pressure of 15 arm. However, the reproducibility of the preparation of SI InP was poor in that time. The origin of the poor reproducibility was finally ascribed to be due to unintentional contamination, mainly by Cr and Ni during heat treatment

34

III-Vs Review • Vol.lO No.2 1997 0961-1290/97/$17.00 ©1997, Elsevier

[7]. The prevention of this contamination made possible reproducible production of SI InP. It was also found that a slight amount of Fe contamination also occurred during annealing. The amount of this unintentional Fe contamination is however very low, compared with conventional industrial Fe-doped InE It was also shown that the reproducible semi-insulating properties can be achieved when undoped InP wafers were annealed under vacuum [8]. From these results, the mechanism o f the semi-insulating behaviour has been discussed by various researchers as follows:• Slight amount of Fe is activated by annealing and the concentration of shallow donors which may be related to phosphorus vacancies is

Science

Ltd

In any case, it became clear that the small amount of Fe is indispensable for SI InP, being independent of it if it is incorporated intentionally or unintentionally. The annealing procedure has an e f f e c t in a c t i v a t i n g this small amount of Fe and in reducing the concentration of shallow donors. From these conclusions, the stable production of SI InP with very low Fe concentrations can be performed by annealing extremely slightly Fe-doped InP as shown in Fig. 1. In this method, very lightly Fe-doped InP grown by LEC which is conductive in its asgrown state can be converted to be semi-insulating by wafer annealing at 950°C for 40 hrs. T h e annealing is p e r f o r m e d by using a high pressure annealing furnace as show in Fig. 2. This annealing furnace consists o f a high-pressure container with multiple heater zones. T h e evacuated quartz ampoule, in which as-cut and etched InP wafers and high purity red phosphorus are sealed, is set in this furnace where argon gas pressure is controlled to balance the phosphorus vapour pressure. It became possible to reproducibly produce SI InP as shown in Fig. 3, but it still had the disadvantage that the u n i f o r m i t y o f electrical p r o p e r t i e s was unsatisfactory, that is, 20-60% for resistivity and 6-30% for mobility. For improving this nonuniformity, Japan Energy Corporation has developed a Multiple-step Wafer Annealing (MWA) technology [12] for InP, which is well known for preparing high quality GaAs wafers [13]. In this MWA process, wafers are annealed at first at 950°C for 40 hrs under a phosphorus vapor pressure of 1 atm and then rean-

nealed at 807°C for 40 hrs u n d e r a phosphorus vapor pressure of 30 atm. Wafer annealing is p e r f o r m e d by using a high pressure annealing furnace as shown in Fig. 2. It was found that MWA is effective in i m p r o v i n g the uniformity of electrical properties, that of resistivity from 50 to 10% and that of mobility from 10 to 2%. The first step annealing is for realizing SI behavior and the second step annealing is for improving the uniformity, which is ascribed to the reduction of native defects such as phosphorus vacancies. In Fig. 4, the uniformity of electrical properties after the MWA process is shown. It can be known that the uniformity upto 8.3% for resistivity and 2.9% for mobility is realized. Japan Energy thus developed two grades of SI InP with low Fe concentrations, one is single-step wafer annealed (SWA) wafers and the other is MWA wafers. For both wafers, the Fe concentration is limited to a very low level, which cannot be achieved by the conventional Fe-doped crystal growth method. SWA wafers are expected for epitaxial growth applications, where the lower Fe concentrations are desired while the resistivity uniformity of 2060% is accepted as far as the resistivity whole over the wafer is larger than 1.0 x 10v ohm cm. MWA wafers are required for ion implantation based electronic devices where low Fe concentrations are essential for wafer to wafer consistency of the activation efficiency. Japan Energy is distributing these wafer annealed SI I n P s a m p l e s for d e v i c e applications. For further information, the readers can contact freely the addresses below.

[3] K. Kainosho, H. Shimakura, H. Yamamoto, T, Inoue and O. Oda, Proc. First Int. Conf. on InPand Related Materials (1989) 312. [4] R B. Klein, R. L. Henry, T. A. Kennedy and N. D.Wilsey, Defects in semiconductors (edited by H. J. v. Bardeleben), MaterialScience Forum 10-12 (1986) 1259. [5] D. H o f m a n n , G. Miiller and N. Streckfuss, Appl. Phys. A48 (1989) 3154. [6] K. Kainosho, H. Shimakura, H. Yamamoto and O. Oda, Appl. Phys. Len. 59 (1991) 932. [7] K. Kainosho, M. Ohta, M. Uchida, M. Nakamura and O. Oda, J. Electronic Materials 25(3) (1996) 353. [8] R. Fornari, A. Brinciotti, E. Gombia, R. Moscam and A. Sentiri, Mat. Sci. Eng. B28 (1994) 95. [9] K. Kainosho, O. Oda, G. Hirt and G. Mtiller, Mat. Res. Soc. ,fymp. Proc. 325 (1994) 101.

efficient engines, advanced information and communication solutions - the demands

[11] R. Fornari, A. Brinciotti, E. Gombia, R. Moscam and A. Sentiri, Mat. Sci. Eng. B28 (1994) 95.

for the de

t of new,

envlronmenta/ly adapted technical solutions are many

[12] M. Uchida, K. Kainosho, M. Ohta and O. Oda, Proc. of Sth Int. Conf. on InP and Related Materials (1996) 43.

and diverse. Silicon carbide, as the basis of

[13] O. Oda, H.Yamamoto, M. Seiwa, G. Kano, T. Inoue, M.Mori, H. Shimakura and M. Oyake, Semicond. Sci. Technol. 7 (1992) A215-223.

semiconductor

ts.

creates a whole new range of opportunities for monitoring and controlling processes in

Contacts: (Technical Information) M. Uchida, Materials & Components Lab., JAPAN ENERGY CORPORATION, 3-17-35, Niizo-Minami, Toda, Saitama, 335 Japan. Tel:+81-(0)48- 433-2051. Fax: + 81-(0)48 - 445 -5400. e-mail:[email protected]

an automobile engine; not k=.astbecause it can operate at tempera~res up to 600 °C. Let us tell you more about what silicon carbide can do. Why not start by reading our World Wide Web pages?

References

(Marketing Department) S. Satoh,

[1] A. Seidl, F. Mosel, J. Friedrich, U. Kretzer and G. Miiller, Mat. Sci. Eng. B21 (1993) 321.

Compound Semiconductor Marketing Dept., JAPAN ENERGY CORPORATION, 10-1, Toranomon 2-chome, Minato-ku, Tokyo, 105 JAPAN. Tel: +81-(3)-5573-6592. Fax: +81-(3)-5573-6779. Email:a640a@ hq.j-energy.co.jp

[2] C, Blaauw, B. Emmerstorfer, R. A. Bruce and M. Benzaquen, Proc. 6th Conf. on Semi-insulating H I - V Materials, Toronto (OUP, 1990) 137.

Better performance, more

[10] D. E Bliss, G. G. Bryant, D. Gabble, G. Iseler, E. E. Hailer and E X. Zach, Proc. of 7th Int. Conf. on InPand RelatedMaterials (1995) 678.

m m m Icteon Science & Technology Park SE-223 70 Lund, Sweden TeLephone +4646 16 89 80 Fax +46 46 16 89 81 ~ww http:1/www.epigress.se