Hi all. I'm trying to fit partial charge set for new molecules with the methods used in the following papers. (J. Phys. Chem. B 2009, 113, and 11463–11478, Article Polarizable and Nonpolarizable Force Fields for Alkyl Nitrates
). In these papers, the authors fitted their charge sets on not only ESP on evenly spaced grid (such as ChelpG method) but also permanent multipoles by given weights (1/Ngrid for potentials, 0.1 for dipoles, 0.01 for quadrupoles). To construct matrices for least-square minimization (Ax = b), A for ESP part (defined as q/r, where q is partial charge and r is distance between atoms and grid points) could be derived as Ajk=\sum_i{(1/rik)·(1/rij)} and b for ESP part as bk = \sum_i{Vi/rik}. So, by the definition of molecular dipole moment, \mu = \sum_i{qiRi} (Ri is coordination vector of nuclears of atoms), I thought Ajk = Rk·Rj and bk = \mu · Rk. However, the charge set resulted in those fitting procedure was overestimated. While I removed the dipole moment part, the charge set seems to be right. Can you give any comments about it? Thanks in advance!
Tian Lu
Fitting for dipole moment is redundant. It is well known that common ESP fitting charges (e.g. MK, CHELPG) have very good reproducibility to dipole moment, see Table 3 of http://doi.org/10.3866/PKU.WHXB2012281 for comparison, so considering dipole moment during fitting never detectably improves the fitted charge.
3 votes
1 thanks
More Seulwoo Kim's questions See All
Similar questions and discussions