Machine learning based modeling of disordered elemental semiconductors: understanding the atomic structure of a-Si and a-C

Abstract

Disordered elemental semiconductors, most notably a-C and a-Si, are ubiquitous in a myriad of different applications. These exploit their unique mechanical and electronic properties. In the past couple of decades, density functional theory (DFT) and other quantum mechanics-based computational simulation techniques have been successful at delivering a detailed understanding of the atomic and electronic structure of crystalline semiconductors. Unfortunately, the complex structure of disordered semiconductors sets the time and length scales required for DFT simulation of these materials out of reach. In recent years, machine learning (ML) approaches to atomistic modeling have been developed that provide an accurate approximation of the DFT potential energy surface for a small fraction of the computational time. These ML approaches have now reached maturity and are starting to deliver the first conclusive insights into some of the missing details surrounding the intricate atomic structure of disordered semiconductors. In this Topical Review we give a brief introduction to ML atomistic modeling and its application to amorphous semiconductors. We then take a look at how ML simulations have been used to improve our current understanding of the atomic structure of a-C and a-Si.

Experiment-Driven Atomistic Materials Modeling: A Case Study Combining X-Ray Photoelectron Spectroscopy and Machine Learning Potentials to Infer the Structure of Oxygen-Rich Amorphous Carbon

“…This first iteratively trained GAP is used in a series of quasi-production runs, with 210 512-atom simple-cubic structures at different densities (1.5 to 2.5 g cm –3 ) and compositions (2.5 to 60% O contents), to generate a-CO x structures using a slower quench process from 3500 K down to 300 K over 100 ps. The final configurations range in structure from a-C:O to a-CO x and even “burnt” systems, with lots of CO and CO 2 molecules spontaneously forming at very-high O content.…”

“…The same procedure is repeated several times, while slowly increasing the maximum O content, until the GAP generates low-temperature structures of arbitrary composition whose predicted energies are close enough to the DFT values. This first iteratively trained GAP is used in a series of quasiproduction runs, with 210 512-atom simple-cubic structures at different densities (1.5 to 2.5 g cm −3 ) and compositions (2.5 to 60% O contents), to generate a-CO x structures using a slower quench process 72 from 3500 K down to 300 K over 100 ps. The final configurations range in structure from a-C:O to a-CO x and even "burnt" systems, with lots of CO and CO 2 molecules spontaneously forming at very-high O content.…”

“…In particular, the challenge in a-CO x atomistic structure prediction is to generate metastable oxygenated a-C structures with very high O content. Although the most realistic strategy would be to carry out simulated deposition as has been done for a-C in the past, ,,, trying to mimic the experimental growth process, this is also the most sophisticated and time-consuming method, in terms of both human and computer time required. Therefore, we explore alternative, cost-effective strategies and compare them to the modified GCMC approach developed here.…”

“…For a-C, melt-quench MD simulations are arguably the most straightforward type of atomistic structure generation procedure, after a simple geometry optimization. ,,− Melt-quench simulations of a-C start out from a high-temperature liquid sample (“melt”) and rapidly cool it down (“quench”) to trap the atoms into metastable configurations. There is an intermediate high-temperature annealing step, where the system is allowed to equilibrate and find more energetically favorable configurations.…”

Novel fabrication techniques for ultra-thin silicon based flexible electronics

“…LTPS is the process of converting a-Si into poly-Si at temperatures below 450 • C, which is lower than that required for processing single-crystalline Si [106][107][108]. Finally, a-Si is formed by a continuous random network of disordered Si atoms lacking long-range order [109,110]. This absence of a well-arranged lattice structure, along with defects such as grain boundaries and a density of states within the band gap, disrupts the movement of electrons, resulting in relatively lower electron mobility (Approx.…”