GaAs(110) interface formation

Surface Science132 (1983) 505-512 North-Holland Publishing Company

505

EFFECT OF TEMPERATURE ON THE Ge/GaAs(llO) FORMATION *

INTERFACE

Ping CHEN **, D. BOLMONT *** and CA. SBBENNE Luborataire de Physipe des Salides, assaci~ au Centre National de la Recherche UniversitP Pierre et Marie Curie, 4 Place JussieK F-75230 Paris Cedex 05, France Received

20 October

1982; accepted

for publication

17 December

~~ieat~~~e,

1982

The initial growth, by a molecular beam epitaxy technique, of a Ge overlayer on a clean cleaved GaAs(l10) heated substrate is studied in ultrahigh vacuum by low energy electron diffraction, Auger electron spectroscopy and photoemission yield spectroscopy. The effect of the substrate temperature on the crystallography, the composition and the electronic structure at both the Ge/GaAs interface and the Ge surface is analysed with particular emphasis on the 350°C range where segregation is observed upon interface formation.

1. Introduction

The heterostructure Ge/GaAs( 110) has attracted much attention because of the perfect fitness of the lattice constants. However, an ideally abrupt junction is very difficult to obtain in practice. Bauer and M~Mena~n [I] have found that it can be the case only when the substrate temperature is controlled between 350 and 430°C. At higher temperatures, interdiffusion occurs and at lower temperatures, islands start to form. Recently Month and Gant [2] reported that, even at 295”C, As atoms are segregated towards the Ge overlayer surface and that interdiffusion takes place near 395°C. Month et al. [3] have demonstrated later that some Ga atoms were also segregated in the same temperature range. In an earlier article [4] we have shown that an ideally abrupt interface forms under epitaxial conditions at 16O’C on a GaAs(ll0) cleaved surface. The electronic structure changes, which have been carefully checked in that case, show that the Ge induced reconstruction of the GaAs( 110) surface is followed by the formation of a uniform Ge layer to which corresponds a continuous * Work supported in part by Direction GenQale des Tel&communications under Contract 81.35.053.00.790.92.45.BCZ. ** On leave from Fudan University, Shanghai, People’s Rep. of China. *** Gn leave from Conservatoire National des Arts et Metiers, Paris, France.

0039-6028/83/0000-0000/$03.00

0 1983 North-Holland

No.

506

P. Chen etal./ Effect of temperature on Ge/GaAs(llO) interfaceformatron

interface state band overlapping the GaAs valence band edge. The valence band discontinuity is determined to be 0.55 + 0.10 eV, in agreement with results obtained by XPS [5], for example. In this paper we study the growth of Ge overlayers on GaAs( 110) at a higher temperature. The combined measurements of low energy electron diffraction (LEED), Auger electron spectroscopy (AES) and photoemission yield spectroscopy (PYS) have proved successful to study interface formation on semiconductors. Indeed, this procedure enables the identification of and correlation between the interface geometry, its stoichiometry and its electronic structure [6,7]. It brings out a general idea of the growth process and we shall show that, at 35O”C, the deposition of Ge on GaAs(ll0) follows a different pattern compared to lower temperature, with segregation of both Ga and As atoms at the Ge surface.

2. Experimental The GaAs samples, the doping levels of which are n = 2 x 1016, 4 x 10” and 1.5 X lOI* cmm3 and p = 1.6 X lOI* cme3, were provided by Radiotechnique RTC (Caen, France). They were cleaved in an ultrahigh vacuum chamber where are mounted the LEED, AES and PYS equipments as described earlier [6]. The cleaved surface looks mirror-like with very few tear marks, as observed by optical microscope. Ge was evaporated from a Knudsen-type crucible shielded by a liquid-N, trap. The deposition rate was calibrated using a quartz microbalance and fixed usually at 0.01 ML/s. It was raised to 0.1 ML/s when “thick” layers were prepared, 1 ML being defined as 8.84 X lOI atoms per cm*, the atomic density in the GaAs(ll0) plane. A special furnace has been designed to heat the sample by direct radiation during deposition. Although the substrate temperature is very well controlled and reproducible, its absolute value is not accurately known. In the present case, this temperature has been determined with a drilled sample in which a thermocouple had been fixed near the surface. The uncertainty on the absolute value is estimated at about + 30°C. This point should be kept in mind when we compare to the substrate temperatures given by different authors. In order to keep a constant temperature during-deposition, the shutter in front of the crucible was opened after a preheating period of 15 min for the sample. The pressure remained always below 3 X lo-” Torr. In most cases, measurements of photoemission yield, Auger and LEED spectra were performed after each Ge deposition. However, in order to avoid too long cycles and possible damage effects from the primary electron beam, some runs were devoted either to the Auger or LEED or PYS measurements. No detectable difference was found between the runs.

P. Chen et al. / Effect of temperature on Ge/GaAs(llO) 3.

interface formation

507

Results and discussion

We shall focus here on the results obtained for a growth temperature of 350°C because, although the clean GaAs(ll0) surface properties are not modified at this temperature, just like at 160°C, a new stable growth process is observed different from that at 160°C. At the beginning of the deposition (O-3 ML), the LEED pattern keeps the original 1 x 1 unit mesh of GaAs( 110) with much brighter spots than in the 160°C grown structures, which implies that we now have a better epitaxy. Then, at about 3-6 ML, l/3 order spots appear in the [OOl] direction, which correspond to a 3 x 1 lattice as has been reported by Month and Gant [2]. However, we find also l/4 order streaks, parallel to [OOl], which appear upon further Ge deposition and become clearer at nearly 20 ML. These streaks possibly arise from a 1 x 4 structure with much disorder in the [OOl] direction and which we shall refer to as 0 x 4. This new LEED pattern (3 x 1 + 0 x 4).

Fig. 1. The LEED pattern of Ge/GaAs(llO) primary electron energy is at 52 eV.

grown at 350°C with a coverage of 50 ML. The

508

P. Chen et al. / Effect of temperature

on Ge/GaAs(llO)

interface formation

shown in fig. 1, remains stable at higher coverages up to about 100 ML, maximum coverage studied here. The ratio of the Auger peak intensities of Ga (54 eV) to As (31 eV) is reproducibly found equal to 0.5 on the clean surface. It does not change upon heating at 350°C. The relative intensity variations of several substrate Auger peaks obtained with the 35O’C grown structures are shown in fig. 2. The most remarkable feature in these variations is the fact that the low energy peaks of both Ga and As reach a limit with a higher value than the high energy peaks seem to do. This behaviour is utterly different from what was observed with the 16O’C grown structures [4]. It is considered as a criterion for growth with a segregated layer [2]. The main point coming out of our measurements is that the Ga atoms are segregated as well as As on the Ge surface, contrary to what Month and Gant [2] observed on their heterostructures. In the ideal case where the segregated Ga and As atoms form a uniform layer with the same atom density as in the bulk of GaAs, on top of a pure Ge layer deposited over a pure GaAs substrate, the relative intensity variations can be modellized as proposed by Month and Gant [2]: 1,/Z,=

1 -exp(-d,/X)+exp[-(Bd+d,)/X].

In this equation, d, is the segregation layer thickness, d is the interlayer distance (2 A for Ge/GaAs(l 10)) and X is the effective escape length which is estimated at 5 f 1 A and 16 k 1 A for the low and the high energy Auger electrons respectively [4], in accordance with those determined by Month et al.

131.

COVERAGE

Fig. 2. The relative Auger peak variations with respect to Ge coverages for Ge/GaAs( 110) grown at 35O’C. The dashed lines are the ideal segregation model curves defined in the text. A scheme of this model is shown in the insert.

P. Chen et al. / Effect of temperature on Ge/GaAs(llO)

interface formation

509

We find that the saturation values for both Ga and As are nearly the same and they give d, = 1 A. Therefore, in the model considered here, there is about 0.5 ML of GaAs at the Ge surface. These data alone are not sufficient to determine how the Ga and As atoms are placed in the segregation layer. However, they probably form an ordered two-dimensional compound with the Ge surface atoms, since a specific LEED pattern has been observed. This would have to be confirmed by photoemission measurements. The curves calculated from the above segregation model have been drawn in fig. 2 and we see that they deviate from the experimental data except in the saturation limit of the low energy peaks. Although the model gives a good qualitative idea of what happens, it is unrealistic for a closer quantitative fit. In particular, the out-diffusion of GaAs has been neglected. To verify this possibility, we have performed several annealings at 350°C after the deposition of 4 ML of Ge on a substrate held at 160°C a temperature at which no out-diffusion is assumed. The relative intensities are shown in table 1 for the low energy peaks. Their increase after annealing confirms the existence of out-diffusion. Meanwhile the relative intensities of the high energy Auger transitions remain constant. This would indicate that diffusion is limited to a few ML in the present experimental conditions. The picture of the growth process which comes out of the LEED and Auger results is confirmed by the variations of the photoemission yield spectra upon Ge epitaxy. In fig. 3, the effective densities of filled states after removal of the contribution of the GaAs substrate states are presented at several Ge coverages for the two substrate temperatures of 160 and 35O’C. The procedure followed to obtain these curves has been explained elsewhere [4]. In short, the subtracted GaAs contribution is the density of filled states at zero coverage, multiplied by an exponential term, exp( -&i/X), in order to represent the Ge overlayer attenuation effect, in which X is determined to be 12 1- 2 A [4]. The main features of the set of curves obtained at 35O’C are comparable to those at 16O’C which have been discussed in detail in ref. [4]. The double structure change in the valence band at very low coverage, labelled a-a’ in fig. 3, is observed at both temperatures. It is interpreted as a displacement towards

Table 1 Evolution of the Auger relative intensities of As and Ga after heat treatments of a sample with 4 ML of Ge

As (31 eV) Ga (54 eV)

Ge deposition at 16O“C

Annealing at 35O’C 15 min

+30min

+60 min

20% 20%

23% 20%

43% 31%

38% 31%

510

P. Chen et al. / Effect of temperature

on Ge/GaAs(llOJ

interface formation

the gap of filled surface states of GaAs. Simultaneously, the large band bending which is established at both temperatures, at low coverage on n-type samples is interpreted as a displacement into the gap of empty surface states of 1

Ge/GaAs(llO)

Fig. 3. Difference curves of the effective densities of filted states of Ge/GaAs( 110) grown 7’ = 160°C and at T = 350°C for different Ge coverages 8.

at

GaAs. Both effects are explained by an impurity induced change of the GaAs(l10) surface relaxation [6,8]. These displaced states, both empty and filled, are removed upon completion of the first Ge layer, when only interface states associated with Ge-GaAs bonds remain (see below). At 350°C as well as at 16O”C, the surface band of Ge, labelied c in fig. 3, appears at high coverages, and the Ge bulk states, labelled d, grow progressively with B beyond 1 ML. The main difference between the two sets of curves concerns the wide band shown schematically as a straight line and labelled b in fig. 3. It has been attributed, from the 16O*C heterostructures [4], to the states associated with the bonding between Ge atoms and the GaAs substrate. Indeed, the contribution of these states, b, increases regularly with 0 at submonolayer coverages, i.e. during the interface formation. for both substrate temperatures. Beyond, the evolution of b depends on the temperature. At 16O”C, the contribution of the interface states to photoemission decreases by the attenuation through a thicker and thicker Ge layer. However, at 350°C, the band of states b attributed to Ge-Ga and Ge-As bonds keeps on increasing well beyond 1 ML; it begins to decrease at coverages of a few monolayers and does not seem to vanish at very large 8. These observations on the behaviour of the states b are consistent with the

P. Chen et al. / Effect of temperature on Ge/GaAs(llO)

interface formation

511

effect of segregation which occurs at 350 and not at 160°C. At low temperature Ge-Ga and Ge-As bonds remain located at the interface while at high temperature such bonds are present at both the Ge-GaAs interface and the Ge surface where the segregated As and Ga atoms seem to form a mixed two-dimensional compound with Ge and keep the corresponding photoemission signal from vanishing upon increasing Ge coverage. It should be noticed that the discontinuity of the valence band edges at the interface is found to be 0.55 k 0.1 eV, independent of the substrate temperature. This seems to rule out an important intermixing under the present experimental conditions, in agreement with the Auger observations, since a diffuse heterojunction would decrease the band discontinuity. The present results confirm the segregation effect in Ge/GaAs( 110) epitaxial layers grown at least at 300°C as first reported by Miinch and Gant [2] and further studied in more details by MSnch et al. [3]. However, a noticeable difference in the quantitative values of segregated species shows up between the two sets of results, in spite of apparently close preparation conditions (clean cleaved substrates at similar temperatures, deposition rates roughly equal, same base pressure etc.). In the present experiments, at 350°C the segregation of roughly the same amount of As and Ga atoms, about l/4 of a monolayer each, is reproducibly observed in a coverage range of up to 100 ML of Ge. It is accompanied by the appearance of a 3 x 1 plus 0 x 4 LEED pattern and the existence of Ge-As and Ge-Ga bonding states which suggest the formation of a definite two-dimensional compound formed by Ga, Ge and As atoms, the structure and chemical composition of which cannot be precisely given now. The specific LEED pattern reported here seems to be associated with the presence of Ga atoms at the surface. For example, the post-annealing of a 40 ML structure grown at 160°C brings As enrichment of the Ge surface without any trace of Ga and no specific LEED pattern besides the 1 x 1 diagram and blurred spots of a 8 x 10 superstructure. In contrast, the post-annealing of a 4 ML structure grown at 160°C induces both Ga and As segregation and the 3 x 1 plus 0 X 4 LEED diagram is observed. Further studies are certainly needed to obtain a better understanding of the Ge/GaAs( 110) system regarding the detailed characteristics of both the interface and the mixed surface compound as well as the segregation process which appears quite sensitive to the experimental conditions.

Acknowledgments It is a pleasure to thank Dr. F. Proix for many helpful discussions comments. The able assistance of B. Helie is gratefully acknowledged.

and

512

P. Chen et al. / Effect of iemperature on Ge/GaAs(lIO)

interface formalion

References [1] [2] [3] [4] [5] [6] [7] [8]

R.S. Batter and J.C. McMenamin, J. Vacuum Sci. Technol. 15 (1978) 1444. W. Miinch and H. Gant, J. Vacuum Sci. Technol. 17 (1980) 1094. W. Mbnch, R.S. Bauer, H. Gant and R. Murschall, J. Vacuum Sci. Technol. 21 (1982) 498. P. Chen, D. Bolmont and C.A. Sebenne, J. Phys. C (Solid State Phys.) 15 (1982) 6101. E.A. Kraut, R.W. Grant, J.R. Waldrop and S.P. Kowalczyk, Phys. Rev. Letters 44 (1980) 1620. D. Bolmont, P. Chen, C.A. Sebenne and F. Proix, Phys. Rev. B24 (1981) 4552. D. Bolmont, P. Chen, F. Proix and C.A. Sebenne, J. Phys. C (Solid State Phys.) 15 (1982) 3639. D. Bolmont, P. Cher., V. Mercier and C.A. Sebenne, Physica 117/l 18B (1983) 816.