Unexplained discrepancies between theoretical and experimental vibrational properties of LiH

Determine the physical origin of the remaining discrepancies between lithium hydride phonon dispersions and vibrational densities of states predicted by density functional theory within the local density approximation and by coupled-cluster CCSD(T) theory and the available experimental measurements, specifically ascertaining whether lattice expansion effects and/or non-adiabatic effects account for these differences.

Background

The study develops machine-learned force fields trained via a Δ-learning approach to approximate higher-level wavefunction methods (HF, MP2, CCSD, CCSD(T)) for solids and applies them to diamond and lithium hydride. For LiH, the computed phonon dispersions and densities of states at both DFT-LDA and CCSD(T) levels show systematic deviations from experimental results, particularly in the optical mode region. The authors also compute vibrational densities of states using velocity autocorrelation functions to assess anharmonic effects, finding only small anharmonic corrections, which suggests that anharmonicity alone does not resolve the mismatch.

They note that while DFT-LDA aligns better with lower optical modes and CCSD(T) with higher ones in H-projected spectra, an overall consistent agreement with experiment is lacking. Potential contributing factors mentioned include lattice expansion and non-adiabaticity, but their relative roles remain unresolved.