Approaches for obtaining accurate rate constants for hydrogen abstraction by a chlorine atom

We have assessed computational methodologies for calculating the rate constants for hydrogen abstraction by Cl(?) for a selection of 12 reactions. For the conventional approach of calculating higher-level [B2K-PLYP/aug'-cc-pV[(T+d),(Q+d)]Z] single-point energies at lower-level [BH&H-LYP/6-31+G(d,p)] stationary points, large deviations from experimental rate constants are found in a number of cases in which the activation energy is very low. These discrepancies are due largely to deviations in the calculated activation energies and can be further traced to the inability of the low level to adequately locate the transition structures. We have examined several alternative approaches for calculating rate constants, namely, IRCmax, IRCmax at 0 K (ZK-IRCmax, with zero-point vibration energies (ZPVEs) incorporated), variational transition-state theory (VTST), and VTST with the inclusion of an Eckart tunneling correction (VTST+E). We find that the low level gives reasonable values for the ZPVEs and thermal enthalpy and entropy corrections that are required in such approaches. While the VTST+E approach yields the closest agreement with experimental rate constants for the systems considered, we find that the simpler IRCmax approach gives adequate values and is able to avoid the major shortcomings of the conventional approach in a cost-effective manner.