Computational strategy for predicting the specific optical rotation values of large flexible molecules

Abstract

The topic discussed in this thesis concerns the challenges faced when one needs to find a theoretical value for the specific optical rotation ([α]D) for a relatively large and flexible molecule. A method was developed to find all of the low-energy conformers that may be present in solution and to calculate the [α]D values for these conformers without exceeding a reasonable computer time. Initially, the effects of changes in molecular geometry on [α]D values were investigated using several small and rigid molecules. (2S)-2-methyl oxirane was used to investigate the effect of rotating the methyl group on calculated [α]D values. The [α]D values were calculated using different basis sets which were found to produce similar [α]D values at the optimized geometry. However, for a non-equilibrium geometry the different basis sets gave very different [α]D values. This variation was attributed to the diffuse p-orbital basis functions on hydrogen atoms which are present in the aug-cc-pVDZ basis set and absent in the 6-3IG* basis set. 2-Propanol was used to show that even though it is an achiral molecule and thus has an [α]D value of zero, there are several low-energy conformers of 2-propanol which have non-zero calculated [α]D values. The three low-energy conformers were obtained by rotating the hydroxy group about the HOCH dihedral angle: the trans conformer has C1 symmetry and an [α]D value of zero. However, the two gauche conformers have C, symmetry and a non-zero [α]D values which have the same magnitude but opposite sign. The calculated [α]D values for each conformer were averaged using Boltzmann statistics to produce an average [α]D value of zero. The Monte Carlo (Me) program was written to search for all of the low-energy conformers of flexible molecules by randomly varying user-specified dihedral angles along the molecular backbone. The MC program used AMI semiempirical method in the GAMESS software for fast energy screening. These AMI low-energy conformers found using the MC program were re-optimized at the B3LYP/6-310* level of theory and then their [α]D values were calculated at the B3LYP/aug-cc-pVDZ level of theory. The effectiveness of the MC program at correctly finding the low-energy conformers was tested using the achiral molecule: 3-hydroxypentadial. The average [α]D value for this molecule converged to zero after two MCruns. The magnitude of error for the calculated [α]D value when the MC routine was used to find the low-energy conformers was found to be ±40° based on six MC runs using a chiral molecule: (S)-(+)-ethyl 3-hydroxybutanoic acid with a known experimental [α]D value. The absolute stereochemistry of 3,5-dimethoxyoctan-1-ol with two chiral centers is currently unknown. However, the relative stereochemistry of the two chiral centers is known to be syn. After seven MC runs, the average [α]D value converged to 61° for an (R,R) conformer. The experimental [α]D value is still needed to evaluate whether if the calculated value is correct within the expected error of ±40° and to assign the absolute stereochemistry for 3,5-dimethoxyoctan-l-ol. Using the methods developed in this work, it appears feasible to compute the [α]D values for even larger flexible molecules with multiple chiral centers. This tactic will be very useful in cases where other methods for elucidating the absolute stereochemistry, such as NMR and X-ray diffraction, do not work well.