Chatzidimitriou-Dreismann et al. Reply: The point of the Comment by Torii [1] is that our results [2] of the Raman cross sections s H and s D of the OH and OD stretching vibrations in liquid H 2 O-D 2 O mixtures may have a simple conventional interpretation, provided by standard quantum chemical calculations in the frame of the conventional Born-Oppenheimer scheme. More concretely, Torii applied the package GAUSSIAN 94 at the Hartree-Fock (HF) level to the isolated (gas-phase) molecules H 2 O,

more »

... 2 O, and HDO, and he found (i) that the Raman activity of the OH stretching mode of HDO is smaller by DA H 23.9% than the average of the OH stretching modes of H 2 O and (ii) that the Raman activity of the OD stretching mode of HDO is larger by DA D 17.4% than the average of the OD stretching modes of D 2 O [1]. Torii then concludes [1] that these numerical values may "explain" qualitatively the experimental results Ds H ഠ 27.5% and Ds D ഠ 110%, for an equimolar H 2 In our opinion, such HF calculations are inappropriate for the physical context under consideration; e.g., the calculated values of bond lengths, bond angles, and vibrational frequencies depend much on the basis set chosen. Moreover, from the viewpoint of present-day theory, these values differ inacceptably from the corresponding experimental "gas-phase" values (see below). Nevertheless, we tested the calculations described in [1], applying the same package GAUSSIAN 94 at the HF level, and using the basis set 6-311G ‫ءء‬ (which includes polarization functions), and which is essentially similar to that of [1] . GAUSSIAN 94 has a built-in option to calculate Raman activities at the HF level. Recall that the "standard procedure" of quantum chemistry determines the bond lengths and angles that minimize the total electronic energy of the molecule, by considering the nuclei as classical mass points. However, HF calculations provide always an "incorrect" molecular geometry of the water molecule, e.g., in our case (basis set 6-311G ‫ءء‬ ): bond length 0.941 Å; bond angle 105.46 ± . As mentioned above, these values (and also those of [1]) are significantly-and, to present-day valid computational standards, unacceptably-different from the experimental values as follows. (1) Experimental gas phase geometry [3]: bond length 0.957 Å, bond angle 104.5 ± . Therefore, to improve somewhat the calculated geometry, we performed with GAUSSIAN 94 associated calculations using the DFT (with SVWN) method, and we obtained (2) the calculated (DFT) gas phase geometry: bond length 0.969 Å, bond angle 103.71 ± . These values (2) are closer to the experimental values (1) than the HF values given above. Note, however, that the DFT algorithm of GAUSSIAN 94 has no option for the calculation of Raman activities. Now, applying the considered numerical procedure (GAUSSIAN 94, HF͞6-311G ‫ءء‬ ), and using the offered option "1SCF" on the calculated geometry (1), we obtain 0031-9007͞00͞84(22)͞5237(1)$15.00