RHEED oscillations for InAs morphology control

J.H.G.Owen, R.S.Ross, and J.J.Zinck

HRL Laboratories, 3011 Malibu Canyon Rd. Malibu, CA 90265, USA

Abstract

RHEED oscillation amplitude decay rates can be varied by control of the Gr. III, Gr. V flux or the sample temperature. Surface morphology of quenched samples has been compared to the RHEED oscillations, and consistent trends in sample morphology with growth parameters have been found. In the growth regime studied, the morphological changes were found to be dominated by the substrate step edges, and opposite trends in surface morphology with RHEED oscillation decay rates are postulated for singular and vicinal surfaces.

Introduction

During typical MBE layer-by-layer growth, the intensity of the specular beam from a Reflection High Energy Electron Diffraction (RHEED) pattern oscillates with a period of one monolayer. These oscillations decay, and the RHEED intensity reaches a constant value after a certain number of oscillations. We have been conducting a series of experiments, looking at the decay rates of these oscillations as a function of growth rate, substrate temperature and the ratio between the As and In fluxes (the V/III ratio).
The purpose of these experiments is to find out what effect the various 'controls' available in an MBE system, i.e. flux and temperature, have upon the decay constants of RHEED oscillations. This data can then be used to validate the Kinetic Monte Carlo (KMC) models, aid in the construction of control models, and allow us to optimise our growth conditions.
Experimental details may be found below in Appendix 1.

Results

A typical RHEED oscillation is shown in Fig.1. The period of the oscillations is 4.5s, giving a growth rate of 0.22 ML/s. The RHEED oscillations decay, and the maxima and minima of the curve may be fitted to an exponential decay, as shown by the red lines. These give us two parameters: t_max and t_min.

Fig.2a is typical of growth in the As-rich regime, with relatively large decay constants t_max =16 ML and t_min =18 ML. Under As-deficient conditions, the surface will undergo a transition from (2x4) to (4x2), which has a very distinctive RHEED signature (Fig.2b). At the beginning, the growth proceeds as normal, but the oscillations damp very quickly, and there is a general upward shift in specular intensity as the surface reconstruction becomes (4x2). As a result of the shift in the average specular intensity t_max drops sharply, and t_min increases.

The trends of t_max and t_min with substrate temperature and with the In flux (i.e. the growth rate) have been studied in order to compare with the KMC simulations. These are varied by varying D/F in the simulations, where D is the diffusion rate (controlled by T), and F, the In flux. In all cases, we find that by increasing D/F, we decrease the number of oscillations observed, and hence decrease the values of t _max and t_min. The effect of the substrate temperature is seen in Fig.3.

Moreover, the effect of the As flux is to reduce the decay constants, equivalent to increasing D/F. This is contrary to conventional wisdom, which holds that excess As increases the roughness of the surface. The flux trends may be seen in Fig. 4. The top left chart shows the range of III and V fluxes used, and the three graphs correspond to the arrows, with the direction of the arrows showing the direction of larger decay constants. Either the Group III or Group V fluxes may be varied, while holding the other constant, or else both may be varied so that the ratio remains constant.

The change in decay constants is having a large effect on the surface morphology, in agreement with the trends seen in RHEED. Large values of t_max and t_min corresponds to small islands, and relatively inactive steps. Small values of t_max and t_min corresponds to a surface closer to step flow mode. STM images taken at three different Group III and V fluxes are seen in Fig. 5.

Conclusions

We explain the trends in terms of a novel growth model, shown schematically in Fig.6. The pre-existing steps on the wafer are quite straight, but as growth proceeds, and the step moves forward, it merges with islands, and so a rough step is formed. The rough step then reduces the area where islands will nucleate (since more of the terrace is close to a step edge, until ultimately, the steps are so rough that no islands nucleate, and the surface is in a step-flow mode. Thus the oscillations die out, without multilayer roughness being induced.

Appendix 1

An EPI MBE chamber was used in these experiments. The As source is an SVT valved cracker, enabling the As flux to be varied continuously, as well as shuttered on and off. 0.1 mm InAs wafers are used for the substrate with a nominal miscut of 0.1 degrees. A 1000 nm buffer layer is grown to ensure a smooth initial surface and a sharp RHEED pattern confirmed before experiments begin. The RHEED pattern is captured using a CCD camera connected to a PC and the intensity of the specular beam (and other beams) is selected out and measured in real time (at about 10fps) using a homebuilt program. Typical growth conditions are 450-500 C and 0.2-1.2 ML/s. V/III ratios range from 0.9:1 to 4:1. The In flux is measured from the period of the RHEED oscillations, and the As flux is meaured by the method of uptake oscillations.
To obtain uptake oscillations, a large amount of In (5-10 ML) is deposited onto the surface with the As shutter closed. This In forms into small droplets. When the In shutter is closed, and the As shutter reopened, the In droplets are consumed by the As, with the As flux rate-limiting, and thus oscillations in the RHEED intensity are obtained which are a measure of the As incorporation rate under those conditions. These oscillations are often weak, as the process of In deposition roughens the surface, and in our system only 3-4 are obtained before the amplitude dies away, so the statistical error is larger than for In oscillations. The error is largest with large As flux measurements (shortest period), which corresponds to a large V/III ratio, where fortunately accuracy is not so important.