Amorphous semiconductors have been a major research theme for me for more than 20 years, now. In collaboration with Barkema, I have produced some of the best models of amorphous silicon (Barkema and NM, PRB 2000) that remain a reference 13 years after their publication and are still used by groups around the world for various studies (most recently on the topic of hyperuniformity - Xie et al, PNAS 2013).
Using a combination of methods such as the Wooten-Winer-Weaire algorithm, ART, ART nouveau, kinetic ART and molecular dynamics, and in collaboration with various colleagues, I have characterized in details the diffusion and relaxation mechanisms (for example,; Kerrache et al, 2011; Joly et al, PRB 2013) and the evolution of the energy landscape in amorphous and disordered silicon, in excellent agreement with X-ray, calorimetry and nanocalorimetry experiments (for example, Kallel et al, PRL 2010; Béland et al, PRL 2013). Among others, we have demonstrated clearly that, contrary to a common interpretation of experimental experimental, vacancies do not diffuse as a whole in a-Si, they either remain fixe or rapidly dissociate as they jump (Joly et al, 2013).
We have also recently, through simulation on a few systems, suggested that glasses, at low temperature at least, relax through a two-step replenish-and-relax mechanism, where high barriers are necessary to move from one energy basin to another. Similarly, we have shown that coordination defects are not necessary for relaxation, even at relatively low-temperature (Joly et al, PRB 2013). We demonstrated also that the existence of a low-density liquid phase in computer simulations was very sensitive to the choice of forcefield, suggesting that very careful simulations were needed to ensure that validity of this hypothesis (Beaucage and NM, 2005).
J. - F. Joly, L. K. Béland, P. Brommer, N. Mousseau, Contribution of vacancies to relaxation in amorphous materials: A kinetic activation-relaxation technique study, Physical Review B 87, 144204 (2013).Abstract: The nature of structural relaxation in disordered systems such as amorphous silicon (a-Si) remains a fundamental issue in our attempts at understanding these materials. While a number of experiments suggest that mechanisms similar to those observed in crystals, such as vacancies, could dominate the relaxation, theoretical arguments point rather to the possibility of more diverse pathways. Using the kinetic activation-relaxation technique, an off-lattice kinetic Monte Carlo method with on-the-fly catalog construction, we resolve this question by following 1000 independent vacancies in a well-relaxed a-Si model at 300 K over a timescale of up to one second. Less than one percent of these survive over this period of time and none diffuse more than once, showing that relaxation and diffusion mechanisms in disordered systems are fundamentally different from those in the crystal.
A. Kerrache, N. Mousseau, L. J. Lewis, Amorphous silicon under mechanical shear deformations: Shear velocity and temperature effects, Physical Review B 83, 134122 (2011).Abstract: Mechanical shear deformations lead, in some cases, to effects similar to those resulting from ion irradiation. Here we characterize the effects of shear velocity and temperature on amorphous silicon (a-Si) modeled using classical molecular-dynamics simulations based on the empirical environment-dependent interatomic potential (EDIP). With increasing shear velocity at low temperature, we find a systematic increase in the internal strain leading to the rapid appearance of structural defects (fivefold-coordinated atoms). The impacts of externally applied strain can be almost fully compensated by increasing the temperature, allowing the system to respond more rapidly to the deformation. In particular, we find opposite power-law relations between the temperature and the shear velocity and the deformation energy. The spatial distribution of defects is also found to depend strongly on temperature and strain velocity. For low temperature or high shear velocity, defects are concentrated in a few atomic layers near the center of the cell, while with increasing temperature or decreasing shear velocity, they spread slowly throughout the full simulation cell. This complex behavior can be related to the structure of the energy landscape and the existence of a continuous energy-barrier distribution.
H. Kallel, N. Mousseau, F. Schiettekatte, Evolution of the Potential-Energy Surface of Amorphous Silicon, Physical Review Letters 105, 045503 (2010).Abstract: The link between the energy surface of bulk systems and their dynamical properties is generally difficult to establish. Using the activation-relaxation technique, we follow the change in the barrier distribution of a model of amorphous silicon as a function of the degree of global relaxation. We find that while the barrier-height distribution, calculated from the initial minimum, is a unique function that depends only on the level of relaxation, the reverse-barrier height distribution, calculated from the final state, is independent of global relaxation, following a different function. Moreover, the resulting gained or released energy distribution is a simple convolution of these two distributions indicating that the activation and relaxation parts of the elementary relaxation mechanism are completely independent. This characterized energy landscape can be used to explain nanocalorimetry measurements.
G. T. Barkema, N. Mousseau, High-quality continuous random networks, Physical Review B 62, 4985-4990 (2000).Abstract: The continuous random network (CRN) model is an idealized model for perfectly coordinated amorphous semiconductors. The quality of a CRN can be assessed in terms of topological and configurational properties, including coordination, bond-angle distributions, and deformation energy. Using a variation on the sillium approach proposed 14 years ago by Wooten, Winer, and Weaire, we present 1000-atom and 4096-atom configurations with a degree of strain significantly less than the best CRN available at the moment and comparable to experimental results. The low strain is also reflected in the electronic properties. The electronic density of state obtained from ab initio calculation shows a perfect band gap, without any defect, in agreement with experimental data.