A study of hydrogen microstructure in amorphous silicon via inversion of nuclear magnetic resonance spectra.

We present an inverse approach for studying hydrogen microstructure in amorphous silicon. The approach consists of generating a prior distribution (of spins/hydrogen) by inverting experimental nuclear magnetic resonance (NMR) data, which is subsequently superimposed on a network of amorphous silicon. The resulting network is then relaxed using a total-energy functional to obtain a stable, low-energy configuration such that the initial spin distribution is minimally perturbed. The efficacy of this approach is demonstrated by generating model configurations that not only have the correct NMR spectra but also satisfy simultaneously experimental structural, electronic and vibrational properties of hydrogenated amorphous silicon.