44874-GB6
Ultrafast Infrared Spectroscopy as a Probe of Molecular Dynamics: A Molecular Modeling Study
Throughout the past year, this grant has supported three students working on individual projects; two of them were continuing their work from the previous year. Two of the three will be graduating this year and one of those is planning on attending graduate school in chemistry.
The first project combined classical simulations and electronic structure methods to examine the vibrational spectroscopy of carbonmonoxymyoglobin. Experiments performed on this system concluded that the vibrational frequency of the CO moiety (and thus the protein environment) fluctuates on two timescales: on the order of 100 fs and on the picosecond scale.[1] The question then becomes: what motion within the protein occurs on these timescales?
This can be addressed readily through molecular dynamics simulations. In order to perform such a simulation, one must relate the coordinates of the atoms to the frequency of the CO stretch. Previous simulations on this system assumed that this frequency is proportional to the external electric field.[1] To test this assumption, we performed 100 calculations of the CO frequency from electronic structure methods using atomic coordinates from the classical simulation. We found little relationship between these frequencies and the electric field. However, we were able to find an excellent correlation between the frequency and a linear combination of the Fe-C distance, two coordinates describing motion within the porphyrin ring, and the electric field. Using this correlation, we obtain an infrared spectrum that is similar to the experiment, but narrower. In this fit, the dominant term is the Fe-C distance. It appears that the previous simulations by Merchant et al. were successful despite assuming that the frequency is proportional to the field (which we have found to be incorrect) because the leading term (Fe-C distance) is damped out in the calculation of the vibrational spectrum. Thus, for that calculation, the electric field becomes much more important. At present, we are running the molecular dynamics simulation in order to improve our statistical sampling, at which point, we intend to publish our results.
The second project examines water evaporation through organic monolayers. Well-formed monolayers of long-chain surfactants can reduce the rate of water evaporation by as much as a factor of 10,000. With the discovery of the ubiquity of uncharacterized organic material in tropospheric aerosols, it stands to reason that these will affect the growth rates of the aerosols. However, a single layer of butanol molecules coating a solution of concentrated sulfuric acid has almost no effect on the rate of water evaporation.[2] Our goal is to understand this surprising result using computer modeling.
Our initial system consisted of pure water coated with an alcohol surfactant. Rather than calculate the rate of evaporation directly, we can obtain data much more rapidly measuring the converse process of condensation (since at equilibrium, the rate of both processes must be equal). With a butanol surfactant, the rate of condensation falls significantly as surfactant is added in contrast to the experiment which shows almost no effect. However, preliminary results using a bulk phase of sulfuric acid provide a value much closer to the experiment.
In the water/butanol system, we have concluded that without exception, if a molecule forms a hydrogen bond with an interfacial molecule, it will condense. Should it fail to form that hydrogen bond, it will be deflected from the surface. At low coverage, it is not surprising to find most of the molecules condense by evading the surfactant molecules. As coverage increases, fewer waters manage to find such gaps and more of the molecules will hydrogen bond to the alcohol and then gradually drift into the bulk. At a monolayer of coverage, these two mechanisms occur with nearly equal rates. We are presently writing a paper regarding our results for room temperature water. During the summer, we generated a large amount of data on the sulfuric acid system which we are analyzing now. We have also begun exploring the model dependence of our results.
The third project (which was started this summer) involved an examination of the structure and dynamics of solutions of methanol and carbon tetrachloride. Experiments using both NMR and IR have led to the conclusion that methanol will form small cyclic clusters at relatively low concentration. It is also believed that these clusters persist at higher concentrations.[3] Using simulation, we were hoping to test this interpretation. At present, we have tried several models for the molecular dynamics and have been unsuccessful in capturing the experimental trends in the NMR rotational dynamics. We are currently exploring models that include polarizability and hope that these increased levels of sophistication will improve the agreement with experiment.
