ESP/MM results:
We have tested the ESP/MM approach during the environment minimization in the proton
transfer step using different core/environment partitions. A full TS search in the core until
convergence has been performed before minimizing again the environment. Unfortunately
most of the calculations are unable to reach convergence. The few ones that converge
require a huge amount of iterations and total CPU time. For instance, for the partition 138/1160 the
ESP/MM approach needs 947/4942 core/environment iterations and 52690 s of total CPU
time, in front of 206/765 iterations and 15551 s when the QM/MM calculations are carried
out. Probably the problems in the convergence of the micro-iterative method are due to the
fact that the TS search in the core and the environment minimization follow different potential
energy surfaces.
|
QM(1SCF)/MM results:
We present now the results of the series of tests in which only one SCF cycle has been
performed to approximate the QM/MM energy during the minimization of the environment. A
full TS search in the core until convergence has been performed before minimizing again the
environment. The cases corresponding to the proton transfer and carbon configuration
inversion steps are given in Tables 3.7 and 3.8, respectively. Although this 1SCF option has no
definite effect on the total number of iterations, it is clear that the total CPU time is
noticeably smaller. Indeed this is due to the fact that each 1SCF energy evaluation during the
environment minimization is faster than a complete QM/MM energy calculation. On the other
hand, the energy column shows that the 1SCF option leads to the same transition state
structures as the complete QM/MM calculations. These results confirm that this 1SCF
approach is reliable and provides a cheaper alternative to be used in the micro-iterative
method.
|
