Abstract | Meier states that the small basis sets (6-31G**) used in our survey of 25 density functional theories (DFTs) applied to some dispersion-bound homomolecular dimers [1] raise �serious doubt on the outcome of the study� [2]. The binding energies and monomer separations we computed at the DFT/6-31G** level led us to the irrefutable conclusion that only VSXC/6-31G**, PW91/6-31G** and HCTH407/6-31G** predicted all the test dimers to be bound. Meier comments that experimental binding energies (BEs) were not quoted in our study. We compared our results with high-level ab initio data because there is an understandable lack of accurate experimental data available, in particular for different conformers of the benzene dimer. (It has been shown that some high-level ab initio methods can reproduce the properties of rare-gas dimers [3].) These high-level ab initio data give binding energies based on electronic energies and do not include zero-point energy (ZPE) corrections, which would reduce the BEs. In order to make meaningful comparisons between our results and the high-level data, we did not include ZPE corrections in our calculated binding energies. It is quite apparent from the data we presented that most DFTs underestimate the binding energies of the seven test dimers. (Exceptions to this, as stated in our original work, are most BEs for the Ne2 dimer, some BE values for the ethylene dimer and all of the BEs predicted using VSXC/6-31G**). The inclusion of counterpoise corrections would reduce the calculated binding energies. Meier reports [2] the results of calculations using B3LYP and B3PW91 on the Ne2 and (CH4)2 dimers that show that bond lengths increase and binding energies decrease when the basis sets are increased from 6-31G** to 6-311++G(2d,2p). These results support our original findings that indicate that basis set effects are important [1]. To further illustrate the effects of basis set on BEs and monomer separations, we performed a series of calculations with PW91 on the Ne2 and (CH4)2 complexes using GAUSSIAN-03 [4] with ultrafine integration grids. These data are collected in Table 1. According to the data in the Table, maximum binding in both dimers is achieved with the use of double-zeta basis sets. This is observed for all of the dimers studied in our original work [1]. Thus, the efficient modeling of the BEs for the seven dispersion-bound dimers in our original work comes with the use of double-zeta basis sets. Bigger basis sets are not better in the case of modeling dispersion interactions with DFT in the complexes studied in [1]. We are currently conducting a more thorough study of the application of DFTs to weakly bound complexes. |
---|