Accurate thermochemistry of covalent and ionic solids from spin-component-scaled MP2

TL;DR

This study demonstrates that spin-component-scaled MP2 methods significantly improve the accuracy of predicting thermochemical properties of covalent and ionic solids compared to traditional MP2 and density functionals.

Contribution

The paper introduces the application and reparameterization of SCS-MP2 and SOS-MP2 methods for solid-state thermochemistry, showing enhanced accuracy over existing approaches.

Findings

SCS-MP2 and SOS-MP2 outperform MP2 in predicting lattice constants, bulk modulus, and cohesive energy.

Reparameterized spin scaling parameters are similar to those used in molecular chemistry, indicating good transferability.

Errors are reduced by at least a factor of two compared to density functional methods.

Abstract

We study the performance of spin-component-scaled second-order M{\o}ller-Plesset perturbation theory (SCS-MP2) for the prediction of the lattice constant, bulk modulus, and cohesive energy of 12 simple, three-dimensional, covalent and ionic semiconductors and insulators. We find that SCS-MP2 and the simpler scaled opposite-spin MP2 (SOS-MP2) yield predictions that are significantly improved over the already good performance of MP2. Specifically, when compared to experimental values with zero-point vibrational corrections, SCS-MP2 (SOS-MP2) yields mean absolute errors of 0.015 (0.017) {\AA} for the lattice constant, 3.8 (3.7) GPa for the bulk modulus, and 0.06 (0.08) eV for the cohesive energy, which are smaller than those of leading density functionals by about a factor of two or more. We consider a reparameterization of the spin scaling parameters and find that the optimal parameters for these solids are very similar to those already in common use in molecular quantum chemistry, suggesting good transferability and reliable future applications to surface chemistry on insulators.