Computational Details

In modified COSMO-SAC, two separate o-profiles have to be generated using COSMO file information more efficiently. The first step is identification of electronegative and hydrogen atoms. In the Gaussian03 COSMO output file, elements are identified by their respective atomic numbers (1 for hydrogen, 6 for carbon, 7 for nitrogen, 8 for oxygen, etc.). As carbon does not participate in hydrogen bonding, it is not taken into the matrix and the rest of the elements along with their respective positions (1, 2, 3, etc.) are stored in a matrix (matrix A). This kind of storing will be useful in identifying the segments of a particular atom.

Next, all hydrogen atoms do not participate in hydrogen bonding. As for hydrocarbon systems, the hydrogen that is bonded with carbon will not participate in hydrogen bonding. Based on geometry optimization, performed in Quantum Chemistry packages like Gaussian03, the C-H bond length (e.g. for octane optimized by HF/6-31G*, bond length varies from 1.085722 A to 1.088926 A ) is chosen. So, if the distance between a carbon and hydrogen comes within this bond length, the hydrogen is bonded with the carbon. So hydrogen is removed from matrix A. Now the matrix will constitute only the elements that are capable of hydrogen bonding. But all segments of these elements will not participate in hydrogen bonding. Positively charged segments from the surfaces of all nitrogen, oxygen and fluorine atoms are considered as Hb acceptors. Negatively charged segments from the surfaces of hydrogen atoms bound to any one of the three acceptor atoms, that is, nitrogen, oxygen and fluorine, are considered Hb donors. When those are identified, the average charge densities (calculated by Equation 5.8) and segment areas are separated. The segment areas are then discretized in 71 bins with their respective charge densities and o-profiles for Hb and non-Hb formed (Equations 5.14 and 5.15 are used). The interaction energy matrix is also modified because the hydrogen bonding interaction will exist only when one hydrogen bonding (Hb) discrete bin would interact with another Hb discrete bin; otherwise only misfit energy will be present. Note that, for the case of interaction energy calculation, average charge densities are replaced with the 71 discrete charge densities. This allows using the c-profiles (both Hb and non-Hb) in Equations 5.21 and 5.22. As proposed, two separate c-profiles (Hb and non-Hb) would be generated for each compound and c-profiles would be the summation of these. Alcohols and amines have Hb acceptor and donor segments and non-Hb segments, whereas ketones and ethers do not have any Hb donor segments. According to this concept only H, O, N, F will participate in hydrogen bonding. Chlorine (Cl) is not included in this list; so, chloroform will have no hydrogen bonding segments. These separate c-profiles are shown in Figure 5.2.

FIGURE 5.2 (Continued)

Hb and non-Hb o-profiles of hexylamine, hexanol, acetone and chloroform (o in e/A²).