Reconciliation of the microcrystalline and the continuous random network model for amorphous semiconductors
- PDF / 613,432 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 4 Downloads / 171 Views
Reconciliation of the microcrystalline and the continuous random network model for amorphous semiconductors. J.K.Bording and J.Tafto Department of Physics, University of Oslo. P.O.Box 1048, Blindern, 0316 Oslo Norway E-mail: [email protected] We show by molecular dynamics simulations, that the radial distribution function of an amorphous material does not change significantly by introducing a considerable volume fraction of nanocrystals. The nanocrystals, embedded in a continuous random network, ensure a certain degree of medium range order in the amorphous material. Our simulations, which are on germanium, show that microcrystals smaller than 2 nm can comprise at least 20 % of the volume without significantly changing the radial distribution function from that of pure continuous random network. By increasing the size of the crystals, without altering the crystal to amorphous volume ratio, the radial distribution changes. The molecular dynamics simulations show that the nanocrystals are unchanged at low temperature. At higher temperature the mobility and critical size of the grains increase, transforming the sub-critical crystalline grains into the surrounding continuous random network matrix. INTRODUCTION The main criterion for a good model of amorphous semiconductors is a radial distribution function in close agreement with experiments. The radial distribution function is obtained by Fourier transformation of diffraction intensities. A limitation, however, is that different atomic arrangements may exhibit similar radial distribution functions. An early idea about the structure of amorphous semiconductors was the microcrystalline model [1,2]. The term microcrystals, in this context, refers to crystals on the nanometer scale, and so nowadays the term nanocrystals are often used. By analogy with traditional polycrystals, the idea was that by reducing the size of the grains, the sharp rings in diffraction experiments would turn into the halos that are characteristic of an amorphous phase. The fact that small coherent volumes were readily observed in the materials by transmission electron microscopy [1,3,4] supports this model. A major limitation, however, was that the radial distribution function would exhibit too much crystallinity unless the crystals were smaller than 1 nm. This is so small that they can no longer be considered crystals. In particular it was the crystalline third nearest neighbor distance that proved hard to remove. Thus, the continuous random network model (CRN) [5] was adapted to amorphous semiconductors, and amorphous silicon and germanium in particular [6]. This model has an atomic arrangement with virtually no medium or long range order, but very strong short-range order. Because hand-built [6] and computer-built models exhibited radial distribution functions in good agreement with experiments [14] this model was favored by most researchers in the field [7,8]. While the first computerbuilt models used specialized algorithms [9,10], recent models have evolved by molecular dynamics simulatio
Data Loading...
