Fig 4. Equipartition of ΔHvap estimates.
Systematic differences between ΔHvap estimates from different gas-phase setups for a range of linear alcohols from methanol to octanol are due to violation of equipartition. The data is reported numerically in Table E in S3 Text. The compound abbreviations are listed in Table E in S3 Text and in Ref. [43]. The left plot depicts the differences between two MD protocols using the VR and the WC thermostats, respectively, to the SD protocol. Compared to the SD setup, the WC setup is found to yield estimates that are consistently too low, while the VR setup yields estimates that are too high. The deviation between the WC setup and the SD setup lies outside the sum of the standard errors for the compounds with alkyl chains of three or more carbon atoms. The deviation between the VR and SD setup lies within the sum of the standard errors. The right plot depicts the internal temperatures averaged over the simulation runs for all three gas-phase setups, as well as for the liquid setup. The total temperature is kept at 298K in all cases. The internal temperatures of the SD setup are very close to the target value and virtually indistinguishable from the liquid simulations. The VR results show a significantly increased internal temperature, while the WC results show significantly too cold internal temperatures.