Interpreting Raman and IR vibrational spectra in complex organic molecules lacking symmetries poses a formidable challenge. In this study, we propose an innovative approach for simulating vibrational spectra and attributing observed peaks to molecular motions, even when highly anharmonic, without the need for computationally expensive ab initio calculations. Our approach stems from the time-dependent stochastic self-consistent harmonic approximation to capture quantum nuclear fluctuations in atom dynamics while describing interatomic interaction through state-of-the-art reactive machine-learning force fields. Finally, we employ an isotropic charge model and a bond capacitor model trained on ab initio data to predict the intensity of IR and Raman signals.

在缺乏对称性的复杂有机分子中解释拉曼和红外振动光谱是一项艰巨的挑战。在这项研究中，我们提出了一种创新的方法，用于模拟振动频谱并将观察到的峰值归因于分子运动，即使在高度非谐波的情况下，也不需要计算昂贵的从头开始计算。我们的方法源于瞬态随机自洽谐波近似，用于捕获原子动力学中的量子核涨落，同时通过最先进的反应式机器学习力场描述原子间相互作用。最后，我们采用从头数据训练的各向同性电荷模型和键合电容器模型来预测 IR 和拉曼信号的强度。