American Chemical Society, Journal of Physical Chemistry B (Soft Condensed Matter and Biophysical Chemistry), 28(118), p. 8135-8147, 2014
DOI: 10.1021/jp5011692
Full text: Download
The two-dimensional infrared spectrum of NMAH and NMAD in H$_2$O and D$_2$O is computed based on force field parametrizations ranging from standard point charge (PC) to more elaborate multipolar (MTP) representations of the electrostatics. For the latter, the nonbonded parameters (MTP and van der Waals) were optimized to reproduce thermodynamic data. The frequency trajectory and frequency-frequency correlation function (FFCF) are determined from explicit frequency calculations on $∼ 10^6$ snapshots without using a more traditional ``mapping'' approach. This allows us to both sample configurations and compute observables in a consistent fashion. In agreement with experiment, the FFCF shows one very rapid time scale (in the 50-fs range) followed by one or two longer time scales. In case of three time scales, the intermediate one is $≈ 0.5$ ps or shorter whereas the longest time scale can extend up to 2 or 3 ps. All interaction models lead to 3 time scales in the FFCF when fitted to an empirical parametrized form. When two time scales are assumed -- as is usually done in the analysis of experimental data --and the short time scales is fixed to the $τ_1 = 50$--100 fs range, the correlation time $τ_c$ from the simulations ranges from 0.7 to 1 ps which agrees quite well with experimentally determined values. The major difference between MTP and point charge models is the observation that the later decay times in the FFCF are longer for simulations with MTPs. Also, the amplitude of the FFCF is reduced when simulations are carried out with MTPs. Overall, however, PC-based models perform well compared to those based on MTPs for NMAD in D$_2$O and can be recommended for such investigations in the context of peptide and protein simulations.