[AMBER] Generating RESP Charges For Tightly-Bound Complexes - Charges from the complexes or the fragments?

From: Damian G. Allis, Ph.D. <damian.somewhereville.com>
Date: Sat, 27 Jun 2015 22:08:43 -0400

Greetings all,

Was hoping someone could provide a definitive answer, direct me to relevant discussions, or provide any other input (even strong opinions).

In an example system, I’ve a family of cucurbit[N]uril (N = 5,6,7) cages containing a neutral para-carborane (C2B10H12). For the cucurbit[7]uril:para-carborane, the carborane cage sits with more-or-less van der Waals packing - the calculated charges for the isolated fragments and the full complex are very similar to one another. For the cucurbit[6]uril:para-carborane complex, the cage is quite tightly packed into the cururbituril (perhaps the pocket is too small for the cage experimentally, but that’s peripheral to the question).

In calculating the RESP charges with antechamber from a Gaussian calculation for the cucurbit[6]uril:para-carborane complex, I obtain the following charges for the complex (left column), isolated fragment atoms (ignoring H atoms here), and a difference between the complex-isolated cases (then the relevant atoms from the two fragments in the right column).

RESP Charge (excluding H-atoms):

complex isolated difference
                                  Cucurbituril Atoms
 0.3028 0.5121 -0.2092 C=O
-0.1117 -0.0630 -0.0487 N-C-N
-0.0081 -0.1183 0.1101 N
-0.4158 -0.4428 0.0269 O
-0.0529 -0.0812 0.0283 C-C
                                  Para-carborane Atoms
 0.0591 0.0271 0.0320 Boron
-0.2731 -0.0989 -0.1741 para-C

Some of the charge differences are small, some are significant (specifically the C=O carbon atoms and the C atoms in the carborane). One could imagine other complexes where the bound molecule is interacting even more strongly with the cage molecule without any real bonding occurring, thereby leading to even larger differences in charge between the isolated and complexed cases.

To the questions - is there any generally agreed upon way to deal with these tightly-bound situations? Are these differences small enough where one would still parameterize in isolation and use those charges? Should the charges be generated for each complex and used for each as special cases (so custom charges for the entire series)? If the complexes are bound very tightly and the RESP charges become very different from the isolated molecules (maybe +/- 0.50 e- or more), is there any reason to not trust the generated RESP charges - or is there a value for the difference in charge between the isolated and complexed cases where one would then consider the RESP charges suspect? I realize I’m not differentiating between calculable and experimentally-realizable cases, but am wondering generally at what point in charge generation would one have to reconsider how they parameterize their system. Any assistance is most welcome.

Thanks in advance,
Damian G. Allis, Ph.D.

"Frankly, we just don't know." - Sir Patrick Moore, The Sky At Night @ 50

AMBER mailing list
Received on Sat Jun 27 2015 - 19:30:02 PDT
Custom Search