S.Schenker, C. Schneider, S. B. Tsogoeva, T. Clark Journal of Chemical Theory and Computation 2011, 7, 3586 (Paywall)
While the main focus of the article is on DFT methods, this is the first systematic study that I have come across that indicates that PM6 is significantly better at reproducing ab initio TS structures and barrier heights (or in this case barrier height differences).
For example, for the five aldol reactions considered in this paper both AM1 and PM3 predicted a completely different (multistep) reaction mechanism than PM6 and the ab initio methods. The difference in barrier heights were predicted to within 2 kcal/mol of CC2/TZVP//M06-2X/TZVP results.
The performance of PM6 is a bit more disappointing for seven nitro-Michael reactions where the difference in barrier heights were underestimated by as much as 9 kcal/mol. However, the authors note "Remarkably, the geometries optimized with PM6 are quite accurate in some cases. The RMSD of all interatomic distances relative to the MP2-optimized geometries is 0.42 Å for 4a and 0.61 Å for 4b, comparable to or better than most DFT optimizations for these systems." Unfortunately, barrier-differences computed using ab initio single point calculations based on these geometries were not reported.
These findings are very interesting given the speed of PM6 which makes automatic high throughput searches for TSs feasible.
Acknowledgements: Thanks to Martin Hediger for alerting me to this paper.
While the main focus of the article is on DFT methods, this is the first systematic study that I have come across that indicates that PM6 is significantly better at reproducing ab initio TS structures and barrier heights (or in this case barrier height differences).
For example, for the five aldol reactions considered in this paper both AM1 and PM3 predicted a completely different (multistep) reaction mechanism than PM6 and the ab initio methods. The difference in barrier heights were predicted to within 2 kcal/mol of CC2/TZVP//M06-2X/TZVP results.
The performance of PM6 is a bit more disappointing for seven nitro-Michael reactions where the difference in barrier heights were underestimated by as much as 9 kcal/mol. However, the authors note "Remarkably, the geometries optimized with PM6 are quite accurate in some cases. The RMSD of all interatomic distances relative to the MP2-optimized geometries is 0.42 Å for 4a and 0.61 Å for 4b, comparable to or better than most DFT optimizations for these systems." Unfortunately, barrier-differences computed using ab initio single point calculations based on these geometries were not reported.
These findings are very interesting given the speed of PM6 which makes automatic high throughput searches for TSs feasible.
Acknowledgements: Thanks to Martin Hediger for alerting me to this paper.
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
No comments:
Post a Comment