1. Introduction and experimental The RHEED and its intensity oscillations at LT-GaAs growth have some particular behaviours. The intensity, preliminaries phase and decay of oscillations depend on the BEP ratio, excess As content and substrate temperature, too. We Recently, molecular-beam-epitaxial (MBE) growth of will here investigate the behaviour of oscillation decay GaAs at low temperature (LT) —around 200C — has during the growth of LT-GaAs. The investigated deposition an increasingly importance in the semiconductor research temperature and the range of the BEP ratio are 200C and technology [1]. The LT-GaAs growth has become and 0.9-1.3, respectively. This investigation is based on an expanding important method since it provides highly the measurement and the observed intensity oscillations of insulating films and contributes to the synthesis of magnetic RHEED which are described in Refs [8,9].

semiconductors. It was shown that growth at this LT leads to incorporation of excess As in the crystal. The high concentration of excess As in LT-GaAs results in 2. Results and discussion a number of novel properties. As-grown and annealed LT-GaAs layers exhibit extremely high electrical resistivity The temporal evaluations of RHEED specular intensity and very short lifetimes of photoexcited carriers [2]. Its during the LT-GaAs growth — where the BEP ratio is close electrical parameters can be analysed using the combined to unity — are shown in Ref [8], Fig. 1 and Ref [9], Fig. 2.

band and hopping conduction model [3]. Depending on It is observable on the figures that when the BEP ratio the growth parameters these layers may contain up to 1.5% moves off from unity, then the decay of oscillations becomes excess of As [4,5]. The majority of excess As is in antisite stronger. If the ratio is 1.3 then the oscillation intensity position, while the remaining atoms are interstitial As or Ga is very trifling so its evaluation is difficult. The decay of vacancies [6]. The uniqueness of LT-GaAs is its high the oscillations was determined as described in Ref [11].

density of midgap states resulting from excess As, while The amplitude decay of oscillations was investigated peak to the structure of matrix remains perfect [7].

peak. The intensity of oscillations minima is changed. The Recently, the use of reflection high-energy electron peak to peak series are determined with the subtraction of diffraction (RHEED) to control the growth of LT-GaAs the background of the oscillations which is determined by has been reported [8–10]. The authors observed RHEED the line of minima of oscillations. After the subtraction, oscillations of the specular spot intensity with a period corexponential function is fitted to determine the decay of responding to one monolayer of deposition, as is observed intensity with the help of least-squares method. The during the traditional high temperature growth process.

exponential approach is according to the first part of Eq (1) It is not unsophisticated event to observe oscillations of where d is a decay time constant. The determined decay RHEED intensity at LT growth. The RHEED oscillations constants vs BEP ratio are shown in Fig. 1.

are very strongly influenced by the growth parameters, It is known that the strain in the grown layer influences such as deposition temperature, ratio of beam equivalent the observed RHEED oscillations. If the strain is larger pressure (BEP) etc. The RHEED oscillations were found in the grown structure then the oscillations become faster fundamental in two regions of BEP ratio at LT. One of quiet. If the strain is smaller or absent then the RHEED these regions is near and another is far from the unity of intensity oscillates longer. This effect is demonstrated and BEP ratio. The strongest oscillations were observed when described in the case of InGaAs/GaAs heterostructures in the BEP ratio was nearly one [8,9]. Oscillations were also Refs [11,12].

found in region of 40-100 of BEP ratio [10].

We can observe very strong changing in the oscillation ¶ E-mail: nemcsics@mfa.kfki.hu decay depending on the BEP ratio at 20C (see Fig. 1).

1400 kos Nemcsics with increasing of the excess As content. The decay of oscillations have several reasons. The excess As gives rise to lattice mismatch, so also to strain, in the grown layer. This strain can influence the decay of intensity oscillations. At first, we investigate this effect because the strain influence is known and the influence of growth phenomenon is unknown. From the given parameters (see Fig. 1) the mismatch dependence of the oscillation decay can be determined. The variation of decay time constant vs lattice spacing is shown in Fig. 2. It is clear, that not only alone the mismatch is responsible for the decay but also the other growth conditions. Changes of the excess As modify not only the mismatch but the growth conditions, e. g.

growth rate, too [13]. So, both the mismatch and the growth parameters influence the behaviour of the oscillation decay.

We approximate this decay with an exponential function.

Furthermore, we suppose that the both effects such as the mismatch and the growth influence can be separated from Figure 1. left : The decay constant vs. BEP ratio at 200C, each other. In this way the decay phenomenon can be right : lattice spacing vs. BEP ratio at 200, 210 and 240C. The described by two time constants, as follows lines serve as guide to the eye only.

-t -t -t I(t) =B0 exp = B0 exp + d G M -t = B exp (1) M where G and M are the assumed time constants of the separated influences, such as growth and mismatch, respectively. B and B0 are the scaling factors which depend on the excess As and also on the d/d. The decay originated from the mismatch can be expessed as follows:

1 1 B0 = - ln M( d/d) d( d/d) B( d/d) t = - e( d/d)(2) d( d/d) where the factors are functions of /d and also of the BEP ratio. In the case of stoichiometric LT-GaAs growth ( /d = 0) there is no decay from mismatch. This means, that for /d = 0 the reciprocal value of the decay Figure 2. The decay time constant vs. lattice spacing derived time constant originates fully from the crystal growth from Fig. 1.

phenomenon (d(0) =G(0) =G0), that is the value of 1/M(0) is zero. The value of G0 is constant. The other part of G, G1 depends on BEP ratio (or /d). The Depending on the growth parameters these LT-GaAs layers whole G = G1(BEP) +G0. So the second term of the may contain many excess As atoms. The majority of 1/M( d/d) expession e( d/d) has also an independent excess As is in the antisite position. The lattice spacing and dependent part on BEP ratio (or /d). The replacing ( d/d) of LT-GaAs becomes greater than that of the between BEP ratio and /d may be only in the case of the stoichiometric crystal. The lattice spacing of the non- narrow range of growth parameters where these ratios are stoichiometric LT-GaAs was determined in Ref [7]. The proportional with each other.

functions of lattice spacing vs BEP ratio are depicted also We have separated the supposed strain effect from the in Fig. 1.

growth phenomenon which can be responsible for the decay We can observe (see Fig. 1) that the decay time constant of oscillation. In the case of InGaAs growth, we have of oscillations d decreases rapidly during the LT-GaAs supposed that the growth influence remains constant at low growth with increasing of the BEP ratio, that is, also In content region, because the one of the most important Физика и техника полупроводников, 2005, том 39, вып. Some Aspects to the RHEED Behaviour of LT-GaAs Growth growth parameters, the deposition temperature, remained the same during the experiment. With this supposition we have received good correspondence between the theoretical critical layer thickness and the threshold thickness, which is derived from the M decay constant [12]. In the case of InGaAs In substitutes Ga in the lattice. Both elements estabilish similarly strong sp3 type bonding in the lattice because the similar valence structure. The situation in the case of LT-GaAs is quite different. The excess As which substitutes Ga in the lattice has different and weaker bonding as sp3 hybrid because its valence structure is different. This property modifies locally the probability of chemisorbtion of As atoms so also the probability of the incorporation of the further excess As atoms in the crystal [14]. The concentration of excess As can be determined from the rate of chemisorbed As atoms. As atoms that are chemisorbed on the arsenic-terminated GaAs (001) surface serve as precursors of excess As, and hence, the concentration of Figure 3. The function of M vs. d/d which is derived from the excess As depends directly on the steady-state coverage of high temperature InGaAs growth. The calculated data originated the chemisorbed As atoms [7]. The excess As perturbs from the LT-GaAs growth.

the bonding behaviour in the crystal, that is, the energy distribution along the surface. We make a simple description for the changing of the unperturbed surface layer by layer.

At the first step, the unperturbed area A can be written be determined as follows: 1/M 1/d-1/G0-1/G1, as follows: A1 = Ab-Aa, where A is the whole area of the similarly as described in Ref [12]. The strain dependent time investigated surface. The factors b and a, which are less constant vs composition in the case of InGaAs is given [12].

than one, give the parts of surface which can be covered The composition independent variation of M vs /d can by chemisorbed As and which can be incorporated by be derived from the above mentioned dependence with the excess As, respectively. The second step can be described help of the modified Vegard’s law [17,18]. The material as follows: A2 =(Ab-Aa)b-Aa. The nth-layer we can independent variation is shown in Fig. 3. The calculated get by follow-up the former given algorithm. The size M data from LT-GaAs are depicted in this figure. The of the perturbed area depends also on the number of the fitting parameter of b was determined with the help of leastgrown layers. This dependence can be negligible if the squares method [14]. The unity of BEP ratio serves as a number of the layers is not large [14]. Among the surface reference point for the calculation of M. In this calculation reconstructions of the GaAs (001) surface, the c(4 4) we have taken into consideration also the BEP ratio of 1.3.

reconstruction occurs at low temperatures under high As The M determined from LT growth corresponds to the fluxes [15,16]. The value of b can be estimated because calculated dependence, but we have to note here that the the maximum coverage of chemisorbed As atoms may be ratio of 1.3 is very difficult to evaluate. We can estabilish 0.75 monolayers like in the case of this reconstruction. The that the separation of the growth and mismatch influences value of a can be estimated by the maximum excess As on the decay of RHEED oscillations can describe the LT content which is 0.015 [7]. It can be seen that the factor b growth only in a narrow range. The intensity oscillation at is larger than a, so we can get, after arrangement of the the BEP ratio of 1.3 is very uncertain to evaluate because expression A and neglecting small terms, the following the very trifling intensity. This drastical intensity damage can simple power function for nth-step: An = Abn. We suppose result not only from the mismatch joined with the reduction that the intensity of RHEED is proportional to the size of unperturbed area but it can be also explained the change of the unperturbed surface. A continuous description by of the sticking coefficient of the deposited species.

replacing of n by rt, yields I(t) =cA(t) =cbrtA, where r is the growth rate, t is the growth time and c is a constant characterizing the diffraction power. This can be written 3. Conclusion in the following form: I(t) =cA exp(-t/G1), where G1 is The LT-GaAs growth is very complicated process. The the decay time constant originated from growth phenomena, which depends on the BEP ratio, this is, G1 = -1/r ln b. decay and absence of the RHEED intensity oscillations The G0 and G1 dependence on b are depicted in Fig. 2. can origin from several effects e. g. change of sticking To justify our discussion we can compare M extracted coefficients, change of unperturbed area and change of from the oscillation decay of LT-GaAs growth and the strain. Here was found, that the separation of growth and material independent decay constant, which is originated strain influence of the RHEED oscillation decay in the case from the mismatch. The variation of M( d/d) should of LT-GaAs is possible in a narrow region of BEP ratio.

Физика и техника полупроводников, 2005, том 39, вып. 1402 kos Nemcsics Acknowledgements: The author is indebted to T. Wosiski for the careful reading of the manuscript. This work was supported by the (Hungarian) National Scientific Research Fund (OTKA) through Grants N T030426 and T037509, which is very acknowledged.

Материалы этого сайта размещены для ознакомления, все права принадлежат их авторам.
Если Вы не согласны с тем, что Ваш материал размещён на этом сайте, пожалуйста, напишите нам, мы в течении 1-2 рабочих дней удалим его.