DFT-based Vibrational Spectra for THz-Spectroscopy and Defect Fingerprinting in Molecular Crystals and Solids.

Abstract

Spectroscopic techniques based on atomic vibrations provide a powerful tool for the atomic scale characterization of solids. Unfortunately, the translation of their spectra into atomistic structures tends to be an inverse-problem, as a structural model is required to assign the observed spectral peaks. This is further complicated by the fact that the exact position of the latter is sensitive to the precise underlying atomic structure. This results in the need for very accurate models. With the steady growth of computational resources, the calculation of vibrational spectra for extended and periodic systems has become more attainable at the level of quantum mechanical calculations. In this work, we first present the example of the THz vibrational spectrum of lactose-monohydrate (LM), and use our results to identify the spectral lines of the observed spectra of different phases, obtained experimentally by heating the LM sample.1 The accompanying water loss induces two phase transitions. According to our results, all phases, including the starting high purity commercial sample, are mixtures of different phases. We discuss the impact of both structural—such as water content and orientation— and methodological—such as Pulay stresses, periodic boundaries, and supercell sizes—aspects on the calculated spectra, and show that DFT-based spectra under periodic boundaries can be matched with experimental data. The importance of an extended periodic system for obtaining an accurate vibrational spectrum is also shown in studying defects in diamond. However, here, we show that the qualitative picture of the defect character of each atom in the system is independent of the system size, allowing for small periodic cells to determine the relevant defect atoms at much reduced computational cost.2 Defects tend to be very localized, resulting in atomic modes.3 Therefore, an often-used strategy for selecting the contributing atoms considers only their relative position with regard to the defect center. Using the atomprojected vibrational spectrum, we present a quantitative method for determining the defect character of each atom in the system, allowing for a rational incremental improvement of the defect spectrum. This method is then applied on several simple defects in diamond.