PICVib: an accurate, fast and simple procedure to investigate selected vibrational modes and evaluate infrared intensities

Abstract

The generalization of the PICVib approach [M. V. P. dos Santos et al., J. Comput. Chem., 2013, 34, 611] for calculating infrared intensities is shown to be successful and to preserve all interesting features of the procedure such as easiness of implementation and parallelization, flexibility, treatment of large systems and at high theoretical levels. It was tested and validated for very diverse molecular systems: XH3 (D3h), YH4 (D4h), conformers of RDX, S(N)2 and E2 reaction product complexes, the [W(dppe)2(NNC5H10)] complex, carbon nanotubes, and hydrogen-bonded complexes (H2O···HOH, MeHO···HOH, MeOH···OH2, MeOH···OHMe) including the guanine-cytosine pair. The PICVib shows an excellent overall performance for calculating infrared intensities of localized normal modes and even mixed vibrations, whereas care must be taken for vibrations involving intermolecular interactions. DFT functionals are still the best combination with high level ab initio methods such as CCSD and CCSD(T).

PICVib: an accurate, fast and simple procedure to investigate selected vibrational modes at high theoretical levels and evaluate Raman intensities

“…For normal modes involving the OH stretching, the PICVib(MP2 : M06‐2X) approach provided the most consistent and systematic results for all complexes with an average error of −1.2%, whereas the errors of the remaining methods MP2 : wB97XD, MP2 : PBE1PBE and MP2 : B3LYP were −1.9%, −3.2% and −3.4%, respectively. It is noteworthy that the Raman intensities calculated with the PICVib procedure are more accurate (smaller relative percentage errors) than those for the infrared intensities, even when the Raman intensities have smaller numerical values.…”

“…In addition, Raman intensities calculated with the PICVib method are more accurate than the calculated infrared intensities. Notice, however, that the NO modes used in the validation of the PICVib approach for infrared intensities are more complex than the CH modes used here. Because the depolarization and unpolarized ratios are smaller than 1 and sometimes of the order of 0.1, it is expected that ρ α ( P ) and ρ α ( U ) would have larger errors than the Raman intensities when computed with the PICVib procedure.…”

“…More specifically, each component ( r , s = x , y , z ) is determined by the linear coefficient, , of a 5‐point values expansion of the polarizabilities, , along the normal coordinates Q n , that is, where the coefficients and are obtained by fitting. Notice that the Raman intensities have a linear fitting process similar to the PICVib approach for infrared intensities . However, the Raman intensities calculated with the PICVib procedure have a second‐order computational demand, while the infrared intensities require single‐point dipole (expectation value) calculations.…”

“…Aiming at these goals, we have proposed, implemented and tested an approach denoted as procedure for investigating categories of vibrations (PICVib) that is capable of providing fast and accurate results for vibrational wavenumbers [31] and infrared intensities. [32] This approach is quite simple to implement and to use because it basically involves the calculation of the vibrational infrared and Raman spectra using a low demanding method from which the normal modes of interest can be selected for a better description at a higher level. Most importantly, the PICVib approach involves only single-point calculations with the more demanding high-level method at a few structures obtained along the selected normal modes.…”

“…convergence criteria, iterative subspace size, preconditioner and quantum chemical method for calculating analytical gradients and analytical/numerical Hessians, the mode-tracking approach can be more computationally intensive than the PICVib procedure, especially for Raman intensities with highly correlated methods. [31,32,34] In addition, it requires (meta)programming, which is freely available (AKIRA code) for several computational quantum chemistry programs. [34] The PICVib approach is very simple to implement, quite general, can be used with any program that provides normal modes with a lower level method (e.g.…”