Chasing Anharmonicities in Polarization-Orientation Raman Spectra of Acene Crystals with Machine Learning

TL;DR

This work develops a first-principles machine-learning framework to investigate anharmonic effects in polarization-orientation Raman spectra of molecular crystals, using anthracene and naphthalene as case studies. By coupling ML interatomic potentials with equivariant polarizability models and a Gamma-point RGDOS approach, the authors enable large-scale, temperature-dependent simulations that capture lattice-expansion and anharmonic dynamics beyond the harmonic approximation. They demonstrate that low-frequency PO-Raman signals largely follow quasi-harmonic behavior with detectable anharmonic signatures such as mode mixing and broadening, though the simulated PO-pattern changes are subtler than reported experimentally in anthracene, partly due to peak overlaps and tensor-level limitations. The study highlights the value of mode-resolved decomposition for interpreting PO-Raman maps and provides a robust framework for linking theory and experiment in complex molecular crystals with significant anharmonicity.

Abstract

We present a first-principles machine-learning computational framework to investigate anharmonic effects in polarization-orientation (PO) Raman spectra of molecular crystals, focusing on anthracene and naphthalene. By combining machine learning models for interatomic potentials and polarizability tensors, we enable efficient, large-scale simulations that capture temperature-dependent vibrational dynamics beyond the harmonic approximation. Our approach reproduces key qualitative features observed experimentally. We show, systematically, what are the fingerprints of anharmonic lattice dynamics, thermal expansion, and Raman tensor symmetries on PO-Raman intensities. However, we find that the simulated polarization dependence of Raman intensities shows only subtle deviations from quasi-harmonic predictions, failing to capture the pronounced temperature-dependent changes that have been reported experimentally in anthracene. We propose that part of these inconsistencies stem from the impossibility to deconvolute certain vibrational peaks when only experimental data is available. This work therefore provides a foundation to improve the interpretation of PO-Raman experiments in complex molecular crystals with the aid of theoretical simulations.

Figures (8)

• Figure 1: a) Comparison of percentage (%) error on $\Gamma$-point harmonic phonon frequencies of anthracene for MACE-MLIP and HDNNP-MLIP with respect to DFT (PBE+MBD) reference values. b) Bar plot showing the absolute errors on $\Gamma$-point harmonic phonon frequencies for Raman active intermolecular modes of anthracene (labeled by their symmetry) with MACE-MLIP and HDNNP-MLIP.
• Figure 2: Correlation plots of the different components of the polarizability tensor as predicted by MACE-$\alpha$ (top row) and SA-GPR (bottom row) with respect to the DFPT values. Predictions are obtained from a test set of 150 independent structures sampled from configurations at three different temperatures (100 K, 150 K, 295 K) and different lattice parameters.
• Figure 3: a) PO-Raman spectrum of anthracene at 100K obtained with the "full-ML" framework. b) PO-Raman spectrum of anthracene at 100K obtained with the $\Gamma$RGDOS-ML framework. Intensities are normalized to the highest peak amplitude in both cases. c) Difference between the PO-Raman spectra obtained with the two frameworks. Negative values correspond to features that are more intense in full-ML than in $\Gamma$RGDOS-ML and viceversa. d) Difference between the unpolarized Raman spectra obtained with the two frameworks.
• Figure 4: Temperature dependence of the parallel configuration PO-Raman spectra of anthracene (a,c,e) and naphthalene (c,d,f) in the intermolecular motion range of 10 to 170 cm$^{-1}$, obtained with the $\Gamma$RGDOS-ML method. The intensity is normalized to the maximum intensity of the 100K spectrum.
• Figure 5: a-c) PO-Raman spectra of anthracene at 100 K and 295 K obtained with the $T_e$-PIGS method musil_quantum_2022 to include nuclear quantum effects (NQEs). Intensity is normalized to the highest value at 100 K. b-d) Unpolarized Raman spectra of anthracene at 100 K and 295 K obtained with classical MD and $T_e$-PIGS. Zoom-in on the region around around 750 cm$^{-1}$ of the unpolarized Raman spectra, showing the redshift of the peaks due to NQEs.