312
M. A. Czarnecki et al.
can be attributed to the approximations inherent to the underlying electronic structure
theory. Unfortunately, modeling an explicit solvation in CCl 4 dramatically increases
the computational demand with each solvent molecules adding 74 electrons to the
calculation. This is in contrast to the comparably inexpensive PCM implicit model.
As benchmarked, it adds only miniscule cost to VPT2 calculations of NIR spectra
[38]. Conveniently, the strongly local character of the ν(OH) mode opens a way for a
feasible simplification applied within the Numerov approach. Accordingly, the r OH
distance can be varied only by changing the coordinate of the associated oxygen and
hydrogen atom, thus serving as a simple approximation to the normal coordinate.
Hence, the harmonic analysis step may be skipped in this approach. Interestingly,
the VPT2 calculations by Schuler et al. [37] led to the best results with implicit
solvation. This result confirmed previous findings by Be´ c et al. [39] which have
not been benchmarked vs. higher level anharmonic computations, but instead were
based on comparisons with the experimental spectra. Nevertheless, it is advised to
examine carefully the frequencies predicted by VPT2 calculations in conjunction
with an implicit solvation model. The studies discussed here reveal the need to take
spectral shifts in the calculated frequencies into account, which depend on the chosen
theoretical method and the examined solute as well.
Finally, one should highlight the key importance in spectroscopy of anharmonic
approaches to molecules in aqueous environment [40]. As it was mentioned earlier,
water serves as the essential medium for biochemical processes. Therefore, the
detailed understanding of the vibrational spectra of hydrated molecules is crucial for
progressing the potential of vibrational spectroscopy in the monitoring of biological
samples. However, water creates a polar solvation environment with high permittivity, a high mobility of solvent molecules as well as directional interactions, e.g.
change dipole interactions and hydrogen bonding. It is challenging to properly
account for the related effects, which calls for reliable high-quality computational
solvation treatments. The considerations toward feasible approaches to this problem
have been recently reviewed [3]. Typical examples are sophisticated stationary point
calculations incorporating implicit solvation models or explicitly considered solvent
molecules, as well as applications of ab initio molecular dynamics (MD) [3]. Because
of the high computational cost of the latter, one needs to consider the accuracy level
necessary to describe inner and outer solvation layers and the impact of the necessary approximations on the vibrational analysis. Such considerations have been made
by Lutz et al. in their methodological study of the vibrational spectrum of aqueous
glycine [40]. The authors compared the spectra of hydrated glycine simulated using
diverse approaches to solvation modeling with subsequent anharmonic treatments
[40]. Simplified MD simulations indicated that an accurate quantum–mechanical
treatment that is applied solely to the hydrated molecule, leads to an inadequate
description of the hydration. An adequate account for the influence of hydration
requires stepping beyond a simplified QM/MM scheme, in which the coupling is realized via empirical Coulombic and non-Coulombic interaction potentials. Further, MD
simulations with electrostatic embedding offered only a slightly improved description but still did not reproduce the preferred charge configuration of the solvated
glycine. In contrast, Hartree–Fock-based QMCF-MD simulation that included a QM
M. A. Czarnecki et al.
can be attributed to the approximations inherent to the underlying electronic structure
theory. Unfortunately, modeling an explicit solvation in CCl 4 dramatically increases
the computational demand with each solvent molecules adding 74 electrons to the
calculation. This is in contrast to the comparably inexpensive PCM implicit model.
As benchmarked, it adds only miniscule cost to VPT2 calculations of NIR spectra
[38]. Conveniently, the strongly local character of the ν(OH) mode opens a way for a
feasible simplification applied within the Numerov approach. Accordingly, the r OH
distance can be varied only by changing the coordinate of the associated oxygen and
hydrogen atom, thus serving as a simple approximation to the normal coordinate.
Hence, the harmonic analysis step may be skipped in this approach. Interestingly,
the VPT2 calculations by Schuler et al. [37] led to the best results with implicit
solvation. This result confirmed previous findings by Be´ c et al. [39] which have
not been benchmarked vs. higher level anharmonic computations, but instead were
based on comparisons with the experimental spectra. Nevertheless, it is advised to
examine carefully the frequencies predicted by VPT2 calculations in conjunction
with an implicit solvation model. The studies discussed here reveal the need to take
spectral shifts in the calculated frequencies into account, which depend on the chosen
theoretical method and the examined solute as well.
Finally, one should highlight the key importance in spectroscopy of anharmonic
approaches to molecules in aqueous environment [40]. As it was mentioned earlier,
water serves as the essential medium for biochemical processes. Therefore, the
detailed understanding of the vibrational spectra of hydrated molecules is crucial for
progressing the potential of vibrational spectroscopy in the monitoring of biological
samples. However, water creates a polar solvation environment with high permittivity, a high mobility of solvent molecules as well as directional interactions, e.g.
change dipole interactions and hydrogen bonding. It is challenging to properly
account for the related effects, which calls for reliable high-quality computational
solvation treatments. The considerations toward feasible approaches to this problem
have been recently reviewed [3]. Typical examples are sophisticated stationary point
calculations incorporating implicit solvation models or explicitly considered solvent
molecules, as well as applications of ab initio molecular dynamics (MD) [3]. Because
of the high computational cost of the latter, one needs to consider the accuracy level
necessary to describe inner and outer solvation layers and the impact of the necessary approximations on the vibrational analysis. Such considerations have been made
by Lutz et al. in their methodological study of the vibrational spectrum of aqueous
glycine [40]. The authors compared the spectra of hydrated glycine simulated using
diverse approaches to solvation modeling with subsequent anharmonic treatments
[40]. Simplified MD simulations indicated that an accurate quantum–mechanical
treatment that is applied solely to the hydrated molecule, leads to an inadequate
description of the hydration. An adequate account for the influence of hydration
requires stepping beyond a simplified QM/MM scheme, in which the coupling is realized via empirical Coulombic and non-Coulombic interaction potentials. Further, MD
simulations with electrostatic embedding offered only a slightly improved description but still did not reproduce the preferred charge configuration of the solvated
glycine. In contrast, Hartree–Fock-based QMCF-MD simulation that included a QM