Modeling electron density distributions from X-ray diffraction to derive optical properties: Constrained wavefunction versus multipole refinement

The rational design of next-generation optical materials requires an understanding of the connection between molecular structure and the solid-state optical properties of a material. A fundamental challenge is to utilize the accurate structural information provided by X-ray diffraction to explain the properties of a crystal. For years, the multipole refinement has been the workhorse technique for transforming high-resolution X-ray diffraction datasets into the detailed electron density distribution of crystalline material. However, the electron density alone is not sufficient for a reliable calculation of the nonlinear optical properties of a material. Recently, the X-ray constrained wavefunction refinement has emerged as a viable alternative to the multipole refinement, offering several potential advantages, including the calculation of a wide range of physical properties and seeding the refinement process with a physically reasonable starting point. In this study, we apply both the multipole refinement and the X-ray constrained wavefunction technique to four molecules with promising nonlinear optical properties and diverse structural motifs. In general, both techniques obtain comparable figures of merit and generate largely similar electron densities, demonstrating the wide applicability of the X-ray constrained wavefunction method. However, there are some systematic differences between the electron densities generated by each technique. Importantly, we find that the electron density generated using the X-ray constrained wavefunction method is dependent on the exact location of the nuclei. The X-ray constrained wavefunction refinement makes smaller changes to the wavefunction when coordinates from the Hartree-Fock-based Hirshfeld atom refinement are employed rather than coordinates from the multipole refinement, suggesting that coordinates from the Hirshfeld atom refinement allow the X-ray constrained wavefunction method to produce more accurate wavefunctions. We then use the experimentally derived wavefunctions to calculate the molecular dipole moment, polarizability, hyperpolarizability, and refractive index and show that these are in good agreement with the values calculated using ab initio methods. Thus, this study shows that experimental wavefunctions can be reliably generated from X-ray diffraction datasets, and that optical properties can be reliably calculated from these wavefunctions. Such a concerted interplay of experiment and computation via the X-ray constrained wavefunction refinement stands to enable the molecular engineering of tailor-made next-generation optical materials. (C) 2013 AIP Publishing LLC.