SiGe Raman spectra vs. local clustering/anticlustering : Percolation scheme and ab initio calculations

TL;DR

This paper develops a 1D percolation model to interpret Raman spectra of SiGe alloys, linking local clustering to spectral features, and validates it with experimental and ab initio data.

Contribution

It introduces a novel 1D percolation scheme for analyzing Raman spectra in SiGe alloys, improving understanding of local clustering effects.

Findings

The model accurately predicts Raman intensity variations with local clustering.

Ab initio spectra confirm the model's predictions.

The approach clarifies the relationship between local structure and spectral features.

Abstract

We formalize within the percolation scheme, that operates along the linear chain approximation, namely at one dimension (1D), an intrinsic ability behind Raman scattering to achieve a quantitative insight into local clustering or anticlustering in an alloy, using SiGe as a case study. For doing so, we derive general expressions of the individual fractions of the six SiGe percolation-type oscillators [1(Ge-Ge), 3(Si-Ge), 2(Si-Si)], which monitor directly the Raman intensities, via a relevant order parameter k. This is introduced by adapting to the 1D oscillators of the SiGe diamond version of the 1D percolation scheme, namely along a fully consistent 1D treatment, the approach originally used by Verleur and Barker for the three-dimensional (3D) oscillators of their 1D cluster scheme applying to zincblende alloys [H.W. Verleur and A.S. Barker, Phys. Rev. 149, 715 (1966)], a somehow problematic one in fact, due to its 3D vs. 1D ambivalence. Predictive k-dependent intensity interplays between the SiGe (50 at.%Si) Raman lines are confronted with existing experimental data and with ab initio Raman spectra obtained by using large (32 atom) disordered supercells matching the required k values, with special attention to the Si-Ge triplet and to the Si-Si doublet, respectively.