Growth of InP single crystals by growth-rate controlled synthesis, solute diffusion technique

Growth of InP single crystals by growth-rate controlled synthesis, solute diffusion technique

Journal of Crystal Growth 68 (1984) 639—643 North-Holland, Amsterdam 639 LETTER TO THE EDITORS GROWTH OF InP SINGLE CRYSTALS BY GROWTH-RATE CONTROLL...

483KB Sizes 2 Downloads 61 Views

Journal of Crystal Growth 68 (1984) 639—643 North-Holland, Amsterdam

639

LETTER TO THE EDITORS GROWTH OF InP SINGLE CRYSTALS BY GROWTH-RATE CONTROLLED SYNTHESIS, SOLUTE DIFFUSION TECHNIQUE Eishi KUBOTA Ibaraki Electrical Communication Laboratory, Nippon Telegraph and Telephone Public Corporation, Tokai - mura, Ibaraki - ken 319- 11, Japan

and Kiyomasa SUGII Musashino Electrical Communication Laboratory, Nippon Telegraph and Telephone Public Corporation, Musashino-shi, Tokyo 180, Japan Received 21 February 1984; manuscript received in final form 5 June 1984

A growth-rate controlled synthesis-solute diffusion (GRC-SSD) technique is demonstrated for rapid growth of high purity lnP single crystals. Large-sized InP boules 34 mm in diameter and 80 mm in length were prepared at a rate of 10 mm/day by this technique, whose growth rate is three times larger than that of the conventional SSD technique. X-ray topography and Hall measurements were performed for the GRC-SSD grown InP single crystal.

In-situ growth of an InP single crystal from In and P elements is attractive from the point of view of obtaining high purity crystals. The synthesis-solute diffusion (SSD) technique [1] is one method which has been studied for this purpose [2,3]. For this technique, however, there have been difficulties in producing single crystals. In addition, the technique possesses the drawback that the growth rate is low. In the past, many investigators have improved the SSD technique so as to produce large-sized single crystals by means of temperature-gradient modification [4—6],crucible improvement [7] and seeded growth [4]. For the growth of single crystals by the SSD technique, it is important to eliminate constitutional supercooling and to keep a flat growth interface [5], since spontaneous nucleation at the growth interface results in polycrystalline growth. Elimination of constitutional supercooling [8] can be achieved by increasing the temperature gradient [4—6], or suppressing the phosphorus concentration at the growth interface by use of an improved crucible

[7]. On the other hand, few attempts have been so far made to maintain a smooth growth interface, We have proposed a growth-rate controlled synthesis-solute diffusion (GRC-SSD) technique for the growth of InP single crystals, in which constitutional supercooling can be eliminated by both increasing the temperature gradient at the growth interface and lowering the ampoule at the same speed as the growth rate, and the growth interface can be kept flat by rotating the ampoule. The high temperature gradient is also suitable for rapid crystal growth. The purpose of this letter is to report details of InP crystal growth using the GRC-SSD technique. Also, properties of the grown crystal are reported. The growth apparatus and growth temperature profile are shown in fig. 1. In this set-up, a temperature gradient large enough to suppress constitutional supercooling is achieved by use of two ring heaters added at the lower edge of the upper furnace. The driving unit is used for lowering and rotating the ampoule; the lowering rate is adjusted

0022-0248/84/$03.OO © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

640

E. Kuhota, K. Sugii

/

Growth of InP by growth -rate controlled .~rnthe.sr~

DRIVING UNIT 400

AMPOULE --

R

as

=

(7.80±3.70)x

102

dexp~——~--——-), 4~ / l.34x1O

whereIndium gradient ture. Gand (°C/cm) T(K) (6N) of is isthe up the average to 200 average ggrowth and temperature red temperaphos-

-

I

region from 800 to 1000°C was measured

500, 900 1000 PC)

TEMPERATURE

FURNACE

In-P SOLUTION InP

perature

~WTh

INTERFACE

and at the bottom of the ampoule, respectively. After vacuum phorus (6N) ofbaking 50 g were [2]. the charged ampoule in the wascrucible sealed and loaded in the furnace. The temperature at the growth interface was maintained at 950°C, and

Fig. 1. Growth-rate controlled synthesis-solute diffusion (GRCSSD) apparatus and its temperature profile.

to the growth rate to increase the grain size along the pulling direction [9], and the rotation improves bution on the growth interface, resulting in a flat the homogeneity of the lateral temperature distrigrowth interface. The lowering rate was chosen to be equal to the average growth rate for the normal SSD process. Fig. 2 shows the growth temperature dependence of the average growth rate normalized by the temperature gradient for the SSD growth. The average growth rate, R (cm/day), in the tern-

~

-*-———--

laSt-—to freeze LF) portion

TEMPERATURE (Cl 1000 I

-,

Ei~

900

800

w 0

I ~102 0

9j

O~ 0

-

o~

0 >0~_~.

________ ~—

w

N

z

10-

3

I,,

0.8

io3rr

0.9

I K~l

Fig. 2. Growth temperature dependence of the growth rate normalized by the temperature gradient for SSD growth. The growth temperature, growth rate and temperature gradient are the average values in a growth run,

first-- to--freeze (FF1 portion

2 twni boundary _____ 10 rnrtt Fig. 3. Longitudinal cross sections of GRC-SSD grown lnP single crystal.

E. Kubota, K. Sugii

/

Growth of InP by growth - rate controlled synthesis

641

striation

~i;

~

5 mm

wall

wall

I

__I

n

0

*1

_

5 mm Fig. 4. X-ray topographs of longitudinal cross sections. (a) First-to-freeze portion and (b) middle portion. Diffraction vector, g, is parallel to the growth direction in both topographs.

642

E. Kuhota. K. Sugu

/

Growth of JnP hr growth - rate controlled synthesis

the temperature gradient was 50°C/cm. The phosphorus temperature was 450°C, corresponding to that yielding a phosphorus vapor pressure of about 2 atm. The ampoule was lowered at a rate of 10 mm/day, and rotated at a rate of 4 rpm. InP crystals with a diameter of 34 mm and a length of up to 80 mm could be grown within a week by GRC-SSD technique, corresponding to a growth rate of more than 10 mm/day. This growth rate is about three times larger than that for the conventional SSD technique [9]. Fig. 3 shows longitudinal cross-sections of the first-to-freeze (FF) and the last-to-freeze (LF) portions for an InP crystal successfully grown using the GRC-SSD technique. The crystal was virtually a single crystal, except for two twin boundaries near the FF portion, and free from needle-like In inclusions along the growth direction [9] which were usually observed in the SSD-grown InP polycrystals. The growth direction was <111)B, which is the preferred direction for unseeded growth [1]. A seeded growth experiment is m two longitudinal cross-sec-

~—

/

~

1016

Kefl_l

-



~-‘o~ -

‘-...

K

eff

.

5

(a) 0

io’~

I

0.1

I

0.5

1.0

1g

iO~ -

‘~7K

-

~

¶ ~-

io~

~ ~

tions cut from the same crystal are shown in fig. 4; (a) for near the FF portion, and (b) for the middle ~ (b) portion. For both topographs, the diffraction vecI I tor is parallel to the growth direction. It can be 0 0.5 1.0 seen that the periphery near the crucible wall is Solidified Fraction, g considerably strained and the dislocation generaFig 5. Electrical properties of GRC-SSD grown lnP. (a) Cartion has occurred in this area. Cell structures [9] ncr concentration (300 K) and (b) Hall mobility profiles, arising from polygonization of dislocations were plotted as a function of the solidified fraction, g. observed only near the FF portion. The dislocation density in this region was too high for individual dislocations to be clearly observed. The cell undoped LEC-InP [10]. Hall mobilities at 300 and structures disappeared in the middle portion, and 77 K decreased gradually from 4400 and 55800 2/V’ s at the FF portion to 4100 and 27200 the dislocation density was reduced to io~—i~~cm cm2. Flat striations were also observed in the cm2/V’ s at the LF portion, respectively. For X-ray topograph of fig. 4a for the FF portion. This SSD-grown crystals, in contrast, the mobility deshows that the growth interface was kept flat creased markedly from the FF to LF portions. For during growth, due to the ampoule rotation, example, the mobilities at 300 and 77 K varied Fig. 5 shows the relationship between the dccfrom 4600 and 96700 cm2/V ‘ s at the FF portion trical properties and solidified fraction, g, for the to 1600 and 7700 cm2/V’ s at the LF portion, GRC-SSD grown lnP crystal shown in fig. 3. The respectively [3]. This corresponds to the grain size carrier concentration profile obeys the normal variation along the growth direction; that is. fairly freezing law, and the slope of the solidified-fraclarge (10 mm square) near the FF portion and tion dependence of the carrier concentration gives small (2 mm square) near the LF portion [3]. It the effective distribution coefficient as being K~ 1, can be concluded that the electrical properties = 0.45. This is slightly smaller than Krti = 0.53 for were greatly improved by increasing the grain size,

E. Kubota, K. Sugii

/

Growth of InP by growth -rate controlled synthesis

and that this is probably due to a reduction in grain-boundary scattering [11]. In summary, InP single crystal growth at a rate of 10 mm/day was achieved by the GRC-SSD technique. The advantage using this technique in the inP growth is a rapid growth of high purity and large-sized single crystal. With a single crystal grown by the GRC-SSD technique, the carrier concentration profile obeys the normal freezing law, with keff = 0.45. Moreover, Hall mobility was 2/V~s at almost constant and more than 4000 cm 300 K throughout the ingot.

643

References [1] K. Kaneko, M. Ayabe, M. Dosen, K. Morizane, S. Usui and N. Watanabe, Proc. IEEE 61(1973) 884. [2] E. Kubota and K. Sugii, J. AppI. Phys. 52 (1981) 2983. [3] E. Kubota, Y. Ohmori and K. Sugii, Inst. Phys. Conf. Sen 63 (1981) 31. [4] K. Gillessen and A.J. Marshall, J. Crystal Growth 32 (1976) 216. [5] H. Fabig and L. Hildisch, Acta Phys. Acad. Sci. Hung. 44 (1978) 5. [61F. Moravec and J. Novotn5i, J. Crystal Growth 52 (1981) 679.

[7] K. Kaneko, M. Ayabe and N. Iwasa, Japan. J. Appl. Phys. 18 (1979) 861.

The authors would like to thank Drs. y Suemune and H. Okamoto for their encouragement. They also thank Dr. A. Katsui for his critical reading of the manuscript.

[8] WA. Tiller, K.A. Jackson, Acta Met. 1 (1953) 428. [9] K. Sugii, E. (1979) 289.

Chalmers,

H. Iwasaki, J. Crystal Growth 46

Yamamoto, S. Shinoyama and C. Uemura, J. Electrochem. Soc. 128 (1981) 585. J.N. Roy, S. Basu and D.N. Bose, J. AppI. Phys. 54 (1983)

[10] A. [11]

Kubota and

J.W. Rutter and B.

847.