# Black-Box Simulations of Anharmonic Vibrational Chiroptical Spectra: Problems with Property Third Derivatives and the Solvent

## TL;DR

This paper explores challenges in simulating anharmonic chiroptical spectra and suggests ways to improve accuracy using VPT2 and better solvent models.

## Contribution

The paper identifies issues with solvent models in calculating higher-order molecular properties and proposes solutions for reliable simulations.

## Key findings

- PCM solvent models can introduce errors in third derivatives of polarizabilities and dipole moments.
- Explicit cavities for hydrogen atoms cause incorrect signs in ROA and VCD bands.
- Combining different approximation levels or using non-explicit cavities improves simulation accuracy.

## Abstract

Chiroptical methods,
including vibrational circular dichroism (VCD)
and Raman optical activity (ROA), reveal details about molecular structure.
For three model molecules, α-pinene, camphor, and fenchone,
we show that increased sensitivity of modern spectrometers makes it
possible to record even fine spectral features, such as overtone and
combination bands. However, understanding, interpretation, and simulation
of them require relatively expensive computations, going beyond the
harmonic approximation. For this purpose, vibrational perturbation
theory at the second order (VPT2) has proven to provide an excellent
price-performance balance. As it becomes more common, inconsistencies
in electronic structure calculations, hidden by error compensation
at the harmonic level, emerge. In particular, while trying to interpret
the spectra, we found that the commonly used polarizable continuum
models (PCM) of solvent may introduce erroneous perturbations to the
higher derivatives of dipole moments and polarizabilities needed to
simulate spectral intensities. We therefore analyze the experimental
spectra on the basis of the simulations and explore parameters allowing
for a “black-box” VPT2 application. In particular, explicit
cavities used for the hydrogen atoms resulted in excessively large
third derivatives of molecular polarizabilities and sometimes led
to incorrect signs of ROA and VCD bands, even for fundamental transitions.
This could be partially rectified by a combination of different approximation
levels used for the calculation of different properties, or by using
PCM cavities not explicitly adapted for hydrogen atoms. Under these
conditions, VPT2 combined with a proper treatment of resonances appears
as an excellent tool to simulate and understand the spectra, including
the assignment of weak anharmonic bands.