(Each question is worth 50 points for scientific analysis, and 17 points for grammer/style/ability to be concise! Don't feel compelled to use both pages unless you have large handwriting...)
Note: The test is based on a literature analysis of French and Dowd J. Comp. Chem. 15 (1994) 561 and Dory et al. JACS 117 (1995) 518. This key is from a student with a perfect score.
1. French and Dowd employ melecular mechanics to study psicose, and Dory et al. use semiempirical MO theory to examine pericyclic reactions. Explain why each set of authors chose the level of theory they did. Would the other level of theory have been appropriate? Why or why not?
In both papers, the level of theory employed was appropriate for the type of study that was being undertaken. French and Dowd were doing conformational studies of furanoses and pyranoses, and in general, MM is good for conformational analysis. MM3 gives accurate, fast (inexpensive) results for geometries and conformational analysis. Semiempirical methods such as PM3 or AM1 would have been a poor choice for this conformational study because they do poorly with furanoses, predicting them to be too flat. Ab initio calculations would have been outrageously expensive considering the many thousands of conformations that were being tested, and it is not necessary to use such a high level of theory since MM works well for this type of study.
Dory et al. used semiempirical MO theory, AM1 and PM3, for calculating transition state structures
and energies. In general, semiempirical theory is good for calculating transition state structures and
energies. Molecular mechanics is not used to calculate transtition states because it is not designed to
deal with partially formed/partially broken bonds that are found in transition states. Ab initio would
give good results in this type of study, but AM1 and PM3 are much faster and cheaper. Dory et al. use
precedent, in part, to justify their choice of methods. As noted in the paper, AM1 had been used
previously, and had given results which agreed with experimental data for the study of [1,5] shift
and Diels-Alder addition of simple substrates.
2. How do the two sets of authors judge the quality of their theoretical predictions? What results do they derive from computation that were either not known or incorrectly known from experiment (if any)?
French and Dowd compare their calculated equilibrium values to NMR data for psicose. Their MM3 calculations predict a pyranose, furanose equilibrium which agrees well with their experimental data. (Calculations predict the equilibrium ratio of furanose:pyranose to be 53:47; NMR results show that the actual ratio is 54:46.) Their results suggest the presence of some additional conformers not indicated by NMR analysis. (These conformers would be present in very low relative concentrations, making their detection by NMR very difficult.)
Dory et al. use their AM1 and PM3 calculations to predict which products are energetically accessible
from the given substrates. They then compare the structures of the predicted products with the structures
of the ovserved products. Their preliminary calculations reproduce, at least qualitatively, the experimental
results: four of the five calculated Diels-Alder adducts were detected experimentally. Their preliminary
calculations, which showed one of their experimentally observed products to be energetically inaccessible,
prompted them to reevaluate their original structural assignment. Upon reevaluation, they found an error
in their original structural assignment. Upon completion of their calculations, they were able to define three
"zones," which allowed them to determine the theoretical accessibility of various products. Their
experimental data for formation of the products was in excellent agreement with theory. The experimental data
also verifies that the theoritical "configurational barriers" are close in energy.
3. What are the imprtant implications, if any, for future computational studies that arise from the results of the two sets of authors? That is, how have they added (or not) to the body of knowledge about these computational methods in terms of strengths and weaknesses in application?
The French and Dowd paper makes several important contributions to the computational field. Previous modelling studies of sugars had always included assumptions about side chain orientation. This article presents the first exhaustive search of side group orientations for fuanoses and a thorough, though not exhaustive, search for pyranoses. The authors show that this is, in fact, a viable approach. The paper also indicates that MM3 is capable of accurately modeling solution state conformations using only a bulk dielectric constant. However, the paper suggests that there is an inherent problem with MM3 that leads to a calculation of a strong overpreference for b-forms (over the a-forms) of these sugars. The overall success of this study suggests that the same method could be used for conformational studies of other sugars (substituted furanoses and pyranoses), and perhaps even more complex carbohydrates.
The paper by Dory et al. demonstrates that AM1 and PM3 can be used to accurately solve selectivity problems involving competing mechanisms. The calculations can also be used to correct ambiguous structural assignments and to predict the outcome of Diels-Alder reactions. Clearly, this type of study could be extended to other types of competing reactions. Further work can also be done using a wider variety of Diels-Alder substrates for prediction of Diels-Alder products.