Vapor phase epitaxy of Hg1−xCdxTe on CdTe heteroepitaxial substrates

,. . . . . . . .

ELSEVIER

CRYSTAL GROWTH

Journal of Crystal Growth 149 (1995) 35-44

Vapor phase epitaxy of Hgl_xCdxTe on CdTe heteroepitaxial substrates N . V . Sochinskii

a,d,e, S.

Bernardi b, E. D i 6 g u e z a,. , P. Franzosi c, S.V. Kletskii d

Departamento de F[sica de Materiales, Universidad Aut6noma de Madrid, ES-28049 Madrid, Spain b Consorzio CREO, Via Pile 60, 1-67100 L'Aquila, Italy c CNR-MASPEC Institute, Via Chiavari 1 8 / A , 1-43100 Parma, Italy d Institute for Semiconductor Physics, Pr. Nauki 45, 252650 Kiev, Ukraine e New Semiconductors, P.O. Box 222, 254210 Kiev, Ukraine

Received 9 June 1994; manuscript received in final form 8 December 1995

Abstract

To grow Hg l_xCd~Te layers for infrared (IR) detector fabrication, the new technique of vapor phase epitaxy (VPE) on CdTe heteroepitaxial substrates (HS) has been studied in detail. A theoretical simulation of the VPE growth process has been carried out. Hgl_~CdxTe layers were grown by VPE at 530°C in evacuated ampoules. CdTe/sapphire and CdTe/GaAs HS obtained by metalorganic chemical vapor deposition (MOCVD) technique were employed as substrates. The melt solution of Hgl_rTey was used as a VPE source, and the dependence of the VPE growth rate on the source composition in the range y = 0.65-0.85 was studied. Various experimental methods were used to characterize the effect of VPE growth conditions on the properties of Hg I ~CdxTe layers like component distribution, surface morphology, structural quality, optical and electrical properties. Our results show a great potential for the VPE on HS technique for the epitaxy of IR grade Hg l_~CdxTe.

1. I n t r o d u c t i o n

In the past, several techniques like liquid phase epitaxy (LPE) [1-3], metalorganic chemical vapor deposition ( M O C V D ) [4,5], molecular b e a m epitaxy (MBE) [6,7], etc. have been employed to grow HgTe-based layers (Hgl_xCdxTe, Hg~_ xMnxTe and H g l _ x Z n x T e ) for infrared (IR) detectors. The substrates used are either bulk substrates (BS) like C d T e and C d l _ x Z n x T e

* Corresponding author.

monocrystals or heteroepitaxial substrates (HS), like an epitaxial layer of CdTe grown on sapphire, GaAs or Si. Since the use of BS has several limitations [8], it is advantageous to grow large area layers on HS for I R arrays [9]. Even though vapor phase epitaxy (VPE) of Hgl_xCdxTe on BS has been intensively studied in the past [10], it is no longer used for the fabrication of I R detectors. This is due to strong compositional inhomogeneity of the VPE layers arising from intensive interdiffusion at the H g ~ _ x C d x T e / C d T e interface during the prolonged growth at high temperature [10-12]. This nonuniformity drastically deteriorates the I R

0022-0248/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0022-024 8(94)01018-8

36

N.V. Sochinskii et aL /Journal of Crystal Growth 149 (1995) 35-44

spectral selectivity of the VPE layers. Moreover, it becomes difficult to control the electrical properties by post-growth annealing in Hg vapor, which is usually done on Hgt_,CdxTe with a constant x-value [13]. Thus nowadays, VPE of Hgl_xCdxTe on BS has been mainly employed for growing heterostructures with x-graded composition, e.g. for ohmic contacts on CdTe [14,15] or Cdl_~ZnxTe [16]. To overcome the drawback of the VPE layers grown on BS, several researchers have applied the VPE technique for growing Hg t _xCdxTe layers on HS (see Refs. [17,18], and references therein) with excellent morphology and structural perfection. Kay et al. [17] demonstrated the VPE growth of a HgTe layer on HS, followed by the transformation of the HgTe/CdTe double layer heterostructure into a Hg~_xCdxTe layer, by the interdiffusion of Hg and Cd. This method was further applied for growing a Hgl_xCd~Te/sapphire heterostructure with an extremely uniform composition of Hgl_xCd~Te [17]. Similarly, Tanaka et al. [18] were able to obtain a Hgl_xCd~Te layer from a MOCVD grown CdTe layer on a sapphire substrate. Hence this approach is highly promising for the fabrication of IR grade Hgl_xCdxTe epitaxial layers. In this paper, we present the results of detailed investigations on the VPE growth of Hgl_~Cd~Te on CdTe/sapphire and C d T e / GaAs. The VPE growth conditions have been evaluated with respect to growth temperature and source composition. Various characterization techniques have been employed for evaluating the quality of the Hg~_xCdxTe VPE layers. The dependence of the HS on the defect formation in Hg l_xCd~Te VPE layers has been discussed. The results obtained have been compared with LPE, MOCVD, and MBE grown Hgl_xCdxTe on HS.

the transport of Te 2 and Hg from the source, and (2) the transformation of the HgTe/CdTe double layer heterostructure into Hgt_xCdxTe of uniform composition by interdiffusion of Hg and Cd. Thus to obtain a Hgt_xCdxTe layer with a desired x value, it is only necessary to have a homogeneous HgTe/CdTe heterostructure with particular thicknesses of the HgTe and CdTe layers. For example, two layers of HgTe and CdTe with thicknesses of 20 and 5 /~m, respectively produces a layer of the Hg0.sCd0.2Te of 25 /zm.

I 1.00

T= 530°C v= 2.5 ~ m / h

I--

[a)

z u

0.5C

"u

0.2 ~ "

1

Z 0

U

~

B

0

5

10 15 20 TH]CKNE 5S ( l,'- m )

25

T:530°C v : 0,25 .,.U,mI h

1.00

oZ

~<-~-0.75 wz ! 1 o X 0 <.9

u

0.50 -

(b)

0.2 5

8

0

5

10

15

20

25

30

1

35

THICKNESS (A~ rn)

2. Theoretical simulation

According to Fleming and Stevenson [10] and Kay et al. [17], the growth of Hg l_xCd~Te on HS consists of two processes: (1) VPE growth of a HgTe layer on the surface of a CdTe layer due to

Fig. 1. D i s t r i b u t i o n of Cd t h r o u g h o u t the t h i c k n e s s of the H g 1 xCdxTe layer grown by V P E o n H S at different mom e n t s of t i m e (t - time of V P E growth, t ' - time of postg r o w t h annealing). (a) Curve 1: t = 0 h, curve 2 : 0 . 5 h, curve 3: 2 h, curve 4 : 4 h, curve 5 : 6 h, curve 6 : 8 h, curve 7: t + t' = 8 h + 4 h, curve 8: t + t' = 8 h + 2 0 h; (b) curve h t = 0 h, curve 2 : 7 h, curve 3 : 2 0 h, curve 4 : 3 0 h, curve 5 : 4 0 h, curve 6 : 6 0 h, curve 7 : 8 0 h, curve 8 : 1 2 0 h.

N.V. Sochinskii et al. /Journal of Crystal Growth 149 (1995) 35-44

To predict favorable experimental conditions for VPE growth of Hgl_xCdxTe on HS, we have carried out a theoretical simulation of the growth process following the model of VPE reported elsewhere [14]. The procedure has been previously applied to the case of LPE [19]. For the calculation, the following parameters have been used: the thickness of the CdTe layer in HS is 5 /xm; the growth temperature (T) is kept constant at 530°C; growth rates (v) of the HgTe layer over the CdTe layer have been taken as 2.5 and 0.25 /xm/h. These numerical values of growth parameters are typical for VPE of Hgl_xCdxTe [10,14,17,18,20]. The interdiffusion coefficient of Cd and Hg can be expressed in the form

D(x,T)

= 3.15 × 101° ×

10 -3"53x

× e x p ( - 2.24 ×

104/kT)

(in /~m2/s), which was obtained experimentally for VPE of Hgl_xCdxTe over bulk CdTe substrates [20]. Also other known expressions of D(x,T) for the HgTe/CdTe heterosystem gave nearly the same results for temperatures above 450°C which is important for VPE [21]. Fig. 1 presents the results of the theoretical simulation. A few remarkable features of the growth process have been revealed. When v is high (Fig. la), the interdiffusion rate of Hg and Cd is the main factor which decides the time of Hg~_~CdxTe formation. During the HgTe growth (Fig. la, curves 2-6) the interdiffusion of Hg and Cd produces a Hg~_xCdxTe x-graded layer with a thickness of 25/zm. Then, to convert this layer to one of uniform composition with x = 0.2, the growth of HgTe has to be stopped, and annealing should be done in order to have interdiffusion of the components (Fig. la, curves 7 and 8). Such kinetics of Hgl_xCdxTe layer formation can be experimentally achieved in an open-tube system (see e.g. Ref. [22]) where the growth of HgTe could be easily stopped by removing the VPE source and subsequently in situ annealing can be carried out. It is also worth noting that due to a strong concentration dependence of D(x,T), the Cd atoms diffuse deeper into the HgTe layer than Hg atoms do in CdTe. Such "asymmetry" is known for the HgTe/CdTe system [23,24] and

37

this implies that at the beginning of growth (Fig. la, curves 2-5) the VPE on HS does not differ from VPE on BS [14]. However, when v is low (Fig. lb), a different growth kinetics takes place. In this case, the VPE growth rate of HgTe is the main factor deciding the time required for the formation of the Hg~_xCdxTe layer throughout the growth process except at the starting period (Fig. lb, curves 2-5). Therefore, an Hg I _~CdxTe layer of uniform composition can be obtained without stopping of HgTe layer growth and an additional annealing step, as shown in Fig. lb (curves 6-8). The only parameter which determines the value of x is the growth time. Obviously, the VPE process with this kinetics can be realized in a sealed ampoule more easily in contrast to the previous process where an additional annealing step is required after the growth is finished.

3. Experimental procedure VPE growth has been carried out with the growth conditions used in the above numerical simulation. The Hgl_xCdxTe layers were grown on HS in evacuated quartz ampoules at 530°C. The melt solution of Hgl_yTey with y = 0.8 was used as VPE source in most of the experiments [14]. Source compositions in the range of y = 0.65-0.85 were also used. The Hg~_yTey source and the HS were placed at an optimum distance of 3 cm as predicted by Fleming and Stevenson [10] with no temperature gradient between the source and the HS. The growth time was varied in the range of 70-120 h depending on the thickness of the CdTe layer in the HS. No post-growth annealing of the Hg~_xCd~Te layer was carried out. Two kinds of HS were used in our experiments: namely, MOCVD grown CdTe/sapphire and CdTe/GaAs. CdTe layers of 7-15/zm thickness were grown on monocrystalline wafers of sapphire (0001) and GaAs(100) of 2" in diameter. The growth was done in a horizontal MOCVD reactor at 360-380°C using dimethyl-cadmium and diisopropyl-telluride as precursors. The asgrown heterostructures were divided into four

38

N.V.. Sochinskii et aL /Journal of Crystal Growth 149 (1995) 35-44

parts and each of them was used as HS for separate experiments on VPE of Hgl_,CdxTe. The details of the MOCVD apparatus and growth are described elsewhere [25]. The surface morphology of the as-grown CdTe and Hgl_,Cd~Te layers and the cleaved surfaces of HS and the H g l _ , C d , T e / s a p p h i r e and Hgt_xCdxTe/GaAs VPE heterostructures were analyzed by a scanning electron microscope (SEM, Philips XL30). The distribution of the components over the heterostructure thickness were measured using a calibrated energy dispersive analysis of X-rays (EDAX) in SEM. The structural quality of the CdTe and H g t _ . C d . T e layers were characterized by high resolution X-ray diffractometry ( H R X R D ) using a Philips goniometer equipped with a two-crystals four-reflections m o n o c h r o m a t o r (Ge, 220). C u K a I radiation and the symmetric 333 reflection (0 B = 38.2 °) were used; the corresponding penetration depth was approximately 4.6/xm. The sample area irradiated was about 1 × 4 m 2, the smaller dimension being parallel to the diffraction plane. Under these experimental conditions, the full width at half maximum (FWHM) of the (004) Bragg peak of a perfect Ge crystal was 8 arc see.

The Fourier transformed IR transmission (FTIR) spectra of H g l _ x C d . T e layers were recorded by a Perkin-Elmer 1710 spectrometer at room temperature (RT). Electrical characteristics of Hg 1_ . C d . T e layers were evaluated by conventional Hall effect measurements using a Van der Pauw geometry in the temperature range from 20 K to RT.

4. Results and discussion

4.1. Effect of source composition As depicted from the theoretical simulation (Fig. lb), the kinetics of the H g t _ , C d x T e layer formation depends strongly on the growth rate of the HgTe layer on the HS. Thus, a precise control of v during the VPE growth is an important prerequisite for determining the Hg and Cd dis-

/ O.4

/

i

/

/o

/ O.3

o/ o//

o: "I0.2 p-

/

o c~ 0.1

!

/

!

/ °o I

Hg 1-y Tey ]

0.2

I

I

I

0.4

Hg CONCENTRATION(l-y)

Fig. 2. Dependence of the growth rate of the Hgi_xCdxTe VPE layer on the Hgl_yTey source composition (growth temperature: 530°C, source to substrate distance: 3 cm).

tribution over the H g l _ , C d x T e layer thickness. In previous studies devoted to the VPE growth of Hgl_xCdxTe on BS [10-12] and on HS [17,18], it was shown that tJ depends on the composition of the VPE source. Nevertheless, the reported results did not coincide as pointed out by Fleming and Stevenson [10], and are not sufficient for determining the type of source required and the value of v corresponding to a definite source composition. We believe that the discrepancies observed are due to the deviation from stoichiometry in the solid VPE sources which were used in the previous studies. Hence the possibility of using a melt solution of Hgl_yTey (y = 0.65-0.85) as VPE source was investigated. The source preparation procedure was similar to the one used for the preparation of the LPE growth solutions [2]. Fig. 2 shows the dependence of u on the Hgl_yTey source composition that was observed in the VPE experiments. Decrease of the Hg concentration decreases the HgTe growth rate as predicted from Fig. lb (e.g. u = 0.25 / z m / h corresponds to y = 0.8). It is also worth noting that the results shown in Fig. 2 are in a qualitative agreement with the one reported by Kay et al. [17] for a solid VPE source of Hg0.s_t.05Te composition.

N. V. Sochinskii et aL /Journal of Crystal Growth 149 (1995) 35-44

39

Fig. 3. Cleaved surface of a VPE heterostructure (SEM). (a) Hg I tCdxTe/sapphire, (b) Hg 1 xCdxTe/GaAs (arrows indicate the places in which the local E D A X measurements were done, see text for details).

4.2. Effect of HS While the cleaved surfaces of various HS (CdTe/sapphire and CdTe/GaAs) did not show any significant difference, unexpectedly a different structure of Hgl_xCdxTe VPE layers grown on various HS was observed. Fig. 3 represents the typical cleaved surfaces of the Hg 1_xCdxTe/sa pphire (a) and Hgl_xCd,Te/GaAs (b) VPE heterostructures. When a CdTe/sapphire HS was used (Fig. 3a), the VPE heterostructure has an exceptionally sharp interface, and the Hgl_xCdxTe layer is homogeneous with no spe-

(a)

At2 03, Hg 1-xCdxTe

584

cific defects that could be attributed to the VPE process. However, when a CdTe/GaAs HS was used (Fig. 3b), the VPE heterostructure has a rough interface, and the Hg~_xCdxTe layer has a polycrystalline structure near the interface, with a strongly non-planar surface, and with voids up to 5/zm in size (arrow 1). Fig. 4 shows the distribution of components over the thickness of the VPE heterostructures represented in Fig. 3. It is shown that the use of both types of HS provides a uniform Cd composition in the Hgl_xCdxTe layers as was theoretically predicted (Fig. lb). Nevertheless, there exist

fin As

.& o

H g1-x Cdx Te

(b)

384 ~2

I vO 4 3 8

tAt

<

6_ j 288

I

Z w

I--

uZ

~z292 w z

,

146-

; z.,

{

HV

.

K

{ L

2

I

4

iI

~

I

6 8 10 12 THICK N E S S ( ~ m )

I

O

Ld

50

~

35

~ I

14 16

18

u~ 192 Z

--

.........{.#.£.....

50 (To) o < o~

-

As

i

20(Hg)

w 0

96

lO(Cd) u 17(Go)

15 t

10

17 20

I "ec~r

30

40

-- ~'r ....

50

I-

I

60

70

-~t-J 3(A s )

80

TH[C K NESS (/u.m )

Fig. 4. Distribution of components throughout the thickness of the VPE heterostructures (EDAX). (a) Hg l_xCdxTe/sapphire , (b) Hg I _xCdxTe/GaAs. Arrow indicates the place in which the component concentrations were determined in at%. Markers on the curves represent the intensity dispersion in counts per seconds.

40

N.V. Sochinskii et al. /Journal of Crystal Growth 149 (1995) 35-44

some differences. The interdiffusion of the layer-substrate components in the Hg~_xCdxTe/ sapphire heterostructure (Fig. 4a) is practically negligible (the validity of the EDAX technique should be taken into account having an accuracy of about 1 at%). In a Hgl_xCdxTe/GaAs heterostructure (Fig. 4b), the interdiffusion of layer-substrate components was found to be very strong in the metallic (Cd, Hg-Ga) and chalcogenic (Te-As) sublattices. The lengths of interdiffusion regions are comparable with the thickness of the Hg~_xCdxTe layer (it is to be noted that this observation was

done on a thick Hg~_xCdxTe layer, Figs. 3b and 4b). The Ga atoms diffuse deeper into the Hgl_xCd~Te layer than the As atoms. This is because of the higher diffusion rate in the metallic sublattice of Hgl_~CdxTe than in the Te sublattice. Local analysis of the component concentration shows Ga enrichment (about 1.5 times) in the vicinity of voids in comparison with a void-free part of the Hg~_~Cd~Te layer (Fig. 3b, arrows 1 and 2, respectively). Fig. 4 also shows that there is a thin HgTe-rich layer near the Hgl_xCdxTe layer surface. We attribute the appearance of this surface layer to

Fig. 5. Surface morphology of epitaxial layers (SEM). (a) CdTe grown by MOCVD on sapphire, (b) and (c) Hg I xCdxTe grown by VPE on CdTe/sapphire MOCVD HS.

N. V. Sochinskii et aL / Journal of Crystal Growth 149 (1995) 35-44

the post-growth cooling process during which the HgTe VPE growth was still continuing while the interdiffusion of Hg and Cd had already stopped. Thus, based on the comparison of Fig. 3a and Fig. 4a, one could reliably conclude that a CdTe/sapphire heterostructure is an appropriate HS for Hgl_~CdxTe VPE growth under the above-specified growth conditions. Also, Figs. 3b and 4b prove that it is inconvenient to use CdTe/GaAs heterostructures as HS for the prolonged VPE growth of Hgl_xCdxTe.

during the VPE process. From additional microprofile measurements, it was found that the maximum height of the surface relief changes from +1.2 /zm for a CdTe layer to +0.5 /zm for a Hgl_~CdxTe layer. Specific details of the Hgl_xCdxTe layer morphology can be seen in Fig. 5c. The main surface defects are craters of up to 2 ~m in diameter (arrow 1), which are obviously due to the substrate relief. The depth of the craters decreases during the VPE growth (arrow 2), which indicates a predominance of the tangential component of the VPE layer growth rate over the normal one. The submicrometer size white spots (arrow 3) were identified by local EDAX measurements as HgTe microdrops. They may be related to vapor condensation during the post-growth cooling. Thus, our studies on the transformation of surface relief shows improvement of the growth surface planarity during the Hg~_xCd~Te VPE growth on HS.

4.3. Surface morphology One of the practical requirements of HgTe containing layers grown by different epitaxial techniques is the surface planarity [1,2,7,12,14,17]. It is known that a remarkable feature of the Hg~_xCd~Te layers grown by VPE on BS under optimal growth conditions [11,12,14] is the good quality of the growth surface. To check this possibility for the VPE growth on HS, the morphology of the growth surface was carefully studied on the CdTe/sapphire HS and the Hg~_xCdxTe layers grown on them. Fig. 5 shows the typical surface morphology of a CdTe MOCVD layer (a) and a Hgl_xCdxTe VPE layer (b,c). It is seen that on a macro-scale (Figs. 5a and 5b) the surface relief of both layers have good uniformity. In SEM it was always possible to get a better contrast for a Hgl_xCd~Te layer (Fig. 5b) than for a CdTe layer (Fig. 5a). This indicates a decrease in surface roughness

4

4.4. X-ray characterization The HRXRD profiles after each step of the Hg l_xCdxTe/sapphire growth process are shown in Fig. 6. They show the narrow Bragg peak for the sapphire substrate (a) and broader ones for the CdTe MOCVD layer (b) and the Hg 1_xCdxTe VPE layer (c). The FWHM values (given in arc sec) demonstrate the good crystal quality of all the materials investigated; the larger values observed in the layers are due to the high density of

0.61--

10

J FWHM

o

lit 6" It ii ii ii i I

(5. d

~2

(23 U

(a)

l I

k \

FWHM 6in

,4 I

\

66''

\

I t

\

FWHM '~

i

\

%

%

I

0,3t-

\

I I

I

l

I

t

45.73

I

/

I 1

II I I i l I I

41

/

(b) .J

45.74

OL---L ~ I 38.92

/

\

/

\

t

/ \

/ I 38.94

t

\

\

I "" 38.96

\

/

(c) /

\

/

\ % I

38.24

I

38.26

I

I "~38.28

ANGLE(deg.)

Fig. 6. H R X R D rocking curves of (a) a sapphire substrate, (b) a CdTe layer grown by M O C V D on sapphire and (c) a Hg t _xCdxTe layer grown by V P E on C d T e / s a p p h i r e M O C V D HS.

N.V. Sochinskii et aL /Journal of Crystal Growth 149 (1995) 35-44

42

defects located at the CdTe/sapphire or Hgl_~CdxTe/sapphire interfaces which are due to large lattice mismatch. Comparison of curves (b) and (c) is very useful. It is seen that they have a similar symmetric shape and nearly the same FWHM value. This proves the structural uniformity of the CdTe and Hgl_xCd~Te layers and demonstrates that the structural quality of the Hg l_xCd~Te VPE layers is limited by that of the HS. In other words, the VPE process does not create additional defects.

10 3

E u >._I

0 3600

I 0

I 10

I

I I l 20 30 1000 IT (K -7)

I 40

10 2o

z 0

6O

~ 20

(a)

~- 0.1

~

The FTIR technique is usually used for determining the composition and thickness of Hg~ _xCd~Te layers grown by different techniques (see e.g. Ref. [26]). Typical results of the FTIR measurements of the Hg 1_~Cd~Te/sapphire VPE heterostructures are shown in Fig. 7. The asgrown Hg 1_xCd~Te layer with the HgTe-rich skin layer on the surface (Fig. 4a) has a FTIR spectrum with a low transmittance and a very smooth high-wavenumber edge (Fig. 7, curve 1). After removing the HgTe-rich skin layer by a brief etching process with Br 2 (5 vol%)-buthanol solution followed by washing with ethylene-glycol and isopropanol [14], the FTIR spectrum shows an increase in the transmittance by more than 50%, with a well-pronounced interference and an exceptionally sharp high-wavenumber edge (Fig. 7, curve 2). This kind of FTIR spectrum supports the good optical uniformity of the Hgl_xCd~Te

~

10

0

4.5. Optical characterization

Hgo'7 Cdo'3Te T : 300K

102

~AI~AI/~

A

2

2800 2000 WAVENUMBER (cm -1)

]Fig. 7. FTIR spectra of a Hg 1 ,CdxTe/sapphire VPE heterostructure. Curve 1: as grown, curve 2: after chemical etching (see text for details).

.~ 10 ~s I-z kd ~ 1018 Z (2,

(b)

1017 < u

I

1016 0

I 10

I

I l I 20 30 1000 / T ( K -I)

[ 40

Fig. 8. Temperature dependence of (a) the Hall mobility and the carrier concentration of a Hgo.7Cdo.3Te layer grown by VPE on CdTe/sapphire M O C V D H S .

(b)

VPE layers and is comparable with the best Hg I _xCd~Te layers grown by other epitaxial techniques. 4.6. Electrical characterization The as-grown chemically-etched Hg]_xCdxTe/sapphire VPE heterostructures were electrically characterized. Etching was done as mentioned above and only small islands (about 2 /zm in diameter) of the HgTe-rich skin layer were left for the deposition of In contacts. Fig. 8 shows the results of Hall effect measurements on a typical Hgl_xCdxTe VPE layer. Figs. 8a and 8b show the temperature dependencies of the Hall mobility (/z) and the carrier concentration (p), respectively for an as-grown p-type Hgl_xCdxTe. The numerical values for these samples are typically /.t = 200 cm2/V • s. and p = 1 × 1017 cm -3 at 77 K which values are similar to those of bulk

N.V. Sochinskii et al. /Journal of Crystal Growth 149 (1995) 35-44

43

Hgl_xCdxTe crystals and LPE layers of similar composition [13]. Thus, the electrical characterization data confirm the E D A X measurements (Fig. 4a) about the negligible interdiffusion in the Hg l _ x C d x T e / sapphire V P E heterostructures (in the opposite case, AI which is a known donor in Hgl_xCd~Te [13] could modify the electrical parameters of the V P E layers). Therefore, one can expect the possibility to control the electrical properties of V P E layers either by the conventional post-growth low-temperature annealing in Hg vapor or by in situ doping.

Hg I _~CdxTe VPE layers was found to be limited by one of the HS. The F T I R spectra of the Hgl_~CdxTe V P E layers show a good transmittance with a well-pronounced interference and a sharp band edge. The as-grown Hg I _xCdxTe V P E layers were of p-type with a carrier concentration p = 1 × 1017 cm -3 and a Hall mobility /~ = 200 c m 2 / V , s. at 77 K. The results obtained by our technique agree well with the results obtained with other epitaxial techniques like LPE, M O C V D , MBE. Hence this technique can be promising for the epitaxy of I R grade Hg I _xCdxTe.

5. Conclusions

Acknowledgements

The new technique of Hgl_xCdxTe VPE on CdTe HS has been theoretically and experimentally studied. Theoretical simulation of the V P E process has been carried out to evaluate the dependence of the growth kinetics with time and the V P E growth rate for producing Hgl_xCdxTe layers of uniform x composition at 530°C. A post-growth annealing was found to be necessary when the growth rate is high (e.g. 2.5 / z m / h ) , while this process is not needed when the growth rate is low (e.g. 0.25 t z m / h ) . Hg~ _xCdxTe layers were grown on C d T e / s a p phire and on C d T e / G a A s M O C V D substrates in evacuated quartz ampoules. The melt solution of Hg l_yTey has been used as a V P E source, and a V P E growth rate in the range of 0 . 0 5 - 0 . 4 5 / x m / h could be obtained by using a source composition in the range y = 0.65-0.85. The effect of the HS on the defect formation and component distribution in Hgl_xCdxTe layers was studied for two cases. The C d T e / s a p p h i r e heterostructure seems to be an appropriate HS. The C d T e / G a A s heterostructure showed strong interdiffusion during the prolonged V P E growth. The properties of the Hgl_xCdxTe VPE layers grown on C d T e / s a p p h i r e HS were investigated. It was found that HgTe-rich thin layers are created on the surface of the Hgl_xCdxTe VPE layers during the post-growth cooling. Improvement of the surface planarity was revealed during VPE growth, The structural quality of the

This work has been partially supported by the Spanish "Comision Interministerial de Ciencia y Tecnologla" C I C Y T under projects Esp92-1345-E and Mat92-0911-C02-01. One of us (N.V.S.) thanks the fellowship from the Spanish Ministerio de Educaci6n y Ciencia.

References [1] S. Bernardi, C. Bocchi, C. Ferrari, P. Franzosi and L. Lazzarini, J. Crystal Growth 113 (1991) 53. [2] N.V. Sochinskii, Proc. Advanced Infrared Technology and Application, Florence, 1992 (1993) p. 29. [3] C.A. Castro and J.H. Tregilgas, J. Crystal Growth 86 (1988) 138. [4] S. Murakami, T. Okamoto, K. Maruyama and H. Takigawa, Appl. Phys. Letters 63 (1993) 899. [5] M. Funaki, A.W. Brinkman, T,D. Hallam and B.K. Tanner, Appl. Phys. Letters 62 (1993) 2983. [6] S. Sivananthan, X. Chu and J.P. Faurie, J. Vacuum Sci. Technol. B 5 (1987) 694. [7] J.M. Arias, J.G. Pasko, M. Zandian, S.H. Shin, G.M. Williams, L.O. Bubulac, R.E. DeWames and W.E. Tennant, Appl. Phys. Letters 62 (1993) 976. [8] N.V. Sochinskii, M.D. Serrano, S. Bernardi and E. Di~guez, Proc. Advanced Infrared Technology and Application, Capri, 1993 (1994) p. 43. [9] D. Amingual, Proc. Advanced Infrared Technology and Application, Florence, 1992 (1993) p. 57. [10] J.G. Fleming and D.A. Stevenson, J. Crystal Growth 82 (1987) 621. [11] S.B. Lee, L.K. Magel, M.F.S. Tang, D.A. Stevenson, J.H. Tregilgas, M.W. Goodwin and R.L. Strong, J. Vacuum Sci. Technol. A 8 (1990) 1098.

44

N.V. Sochinskii et al. /Journal of Crystal Growth 149 (1995) 35-44

[12] S.B. Lee, D. Kim and D.A. Stevenson, J. Vacuum Sci. Technol. B 9 (1991) 1639. [13] P. Capper, J. Vacuum Sci. Technol. B 9 (1991) 1667. [14] N.V. Sochinskii, V.N. Babentsov, S.V. Kletskii, M.D. Serrano and E. Di6guez, Phys. Status Solidi (a) 140 (1993) 445. [15] N.V. Sochinskii, M.D. Serrano, V.N. Babentsov, N.I. Tarbaev, J. Garrido and E. Di6guez, Semicond. Sci. Technol. 9 (1994) 1713. [16] M. Koralewski, N.V. Sochinskii, M.D. Serrano, E. Di6guez, J. Garrido, G. Lifante, B. Noheda and J.A. Gonzalo, Appl. Phys. Commun. 13 (1994) 69. [17] R.E. Kay, H. Chan, F. Ju and B.A. Bray, US Patent No. 4846926 (July 11, 1989). [18] M. Tanaka, K. Ozaki, K. Yamamoto, H. Ebe and Y. Miyamoto, J. Crystal Growth 117 (1992) 24.

[19] G.I. Zhovnir, S.V. Kletskii and N.V. Sochinskii, Phys. Status Solidi (a) 115 (1989) K31. [20] K. Zanio and T. Massopust, J. Electron. Mater. 15 (1986) 103. [21] D.A. Stevenson and M.F.S. Tang, J. Vacuum Sci. Technol. B 9 (1991) 1615. [22] S. Bernardi, J. Crystal Growth 87 (1988) 365. [23] S.V. Kletskii, N.V. Sochinskii and G.I. Zhovnir, Ukr. Fiz. Zh. 38 (1993) 894. [24] A. Tardot, A. Hamoudi, N. Magnea, P. Gentile and J.L. Pautrat, Appl. Phys. Letters 62 (1993) 2548. [25] S. Bernardi, to be submitted. [26] J. Thompson, P. Mackett, L.M. Smith, D.J. Cole-Hamilton and D.V. Shenai-Khatkhate, J. Crystal Growth 86 (1988) 233.