Comparison of quantum chemical parameters and Hammett constants in correlating pK(a) values of substituted anilines.

Historically, Hammett constants have been extremely effective in describing the influence of substituents on chemical reactivity and other physical and chemical properties, whereas variables derived from quantum chemical calculations have generally been less effective. Taking the experimental pK(a)s of substituted anilines as a representative physicochemical property, five ab initio quantum chemical indices are compared for effectiveness as one-parameter regression descriptors for pK(a). All of the tested descriptors performed well for a set of 19 mono-, 13 di-, and 4 trisubstituted anilines, and two performed somewhat better than the traditional Hammett sigma constants. Among the calculated quantities, the best representation of the aniline pK(a)s is produced by the minimum average local ionization energy on the molecular surface.

Molecular structure-property relationships for alkenes.

Structure-property relationships were obtained for 11 physical and chemical properties (boiling points (bp), melting points (mp), molar refractions (MR), molar volumes (MV), heats of combustion (HCKJ), molar heats of vaporization (HVMOL), flashpoints (FLASHK), second virial coefficients (VIRC2), critical temperatures (Tc), critical pressures (Pc), and viscosities (VISC)) for a data set consisting of 162 C4-C9 monoalkenes. Both molecular connectivity indices and ad hoc descriptors were tested as structural descriptors, and both produced high-quality regression equations for most of the properties. As was observed in an earlier study of alkanes [J. Am. Chem. Soc. 110 (1988) 4186], mp were not well described by either descriptor set. For most properties, the mass/bulk of the molecule was found to be the most important structural feature determining the property, suggesting that dispersion forces play a dominant role in determining those properties influenced by intermolecular interactions. The amount of branching in the molecule and the nature of the double bond environment were also found to be influential features.

Cellular automata models of chemical systems.

This paper describes the use of kinematic, asynchronous, stochastic cellular automata to model liquid properties, solution phenomena and kinetic phenomena encountered in complex biological systems. Cellular automata models of dynamic phenomena represent in silico experiments designed to assess the effects of competing factors on the physical and chemical properties of solutions and other complex systems. Specific applications include solution behavior, separation of immiscible liquids, micelle formation, diffusion, membrane passage, first- and second-order chemical kinetics, enzyme activity and acid dissociation. Cellular automata is thus considered as providing an exploratory method for the analysis of dynamic phenomena and the discovery and understanding of new, unexpected phenomena.

Explorations of molecular structure-property relationships.

The problem of the relationship between the structure of a molecule and its physical, chemical, and biological properties is one of the most fundamental in chemistry. Three molecular structure-property studies are discussed as illustrations of different approaches to this problem. In the first study the carcinogenic activities of polycyclic aromatic hydrocarbons and their derivatives are examined. Molecular orbital calculations of the presumptive activation steps and species for these compounds (based on the "bay region" theory of activation) are seen to yield a surprisingly good guide to the observed carcinogenic activities. Both activation and deactivation steps are considered. The second study reviews structure-property work on the tissue solubilities of halogenated hydrocarbons. Relatively simple structural descriptors give a good account of the solubilities of these compounds in blood, muscle, fat, and liver tissue. With the aid of principal components analysis it is shown that there are two dominant dimensions to this problem, which can be interpreted in terms of solubilities of the compounds in lipid and saline environments. The final study, which examines the boiling points of aliphatic alcohols, illustrates the value of using more than one descriptor set. The (perhaps surprising) conclusion is that a theoretical model can sometimes be more accurate than the data upon which it is based. Moreover, two models are better than one.

Modeling the tissue solubilities and metabolic rate constant (Vmax) of halogenated methanes, ethanes, and ethylenes.

Experimental solvent:air and tissue:air partition coefficients for 25 halogenated methanes, ethanes, and ethylenes in saline solution; olive oil; and rat blood, muscle, liver, and fat tissues have been examined using theoretical molecular modeling techniques. The metabolic rate constant, Vmax, was also investigated by these techniques for 19 chlorinated compounds in this group. Two graph theoretical approaches (the distance method of Wiener and the connectivity index method of Randic, Kier, and Hall) and an approach utilizing ad hoc molecular descriptors were employed. Satisfactory regression models for solubility were obtained with both the Randic-Kier-Hall approach and the ad hoc descriptors approach. Fluorine substituents decrease tissue solubilities, whereas both clorine and bromine substituents increase tissue solubilities, with the relative influence being Cl less than Br. Tissue solubilities can also be represented conveniently in terms of contributions from oil and saline solubilities, a procedure reinforced by factor analysis of the data. Equations derived by these methods adequately estimated the solubilities for eight additional compounds. No approach could successfully model Vmax for all 19 compounds, but a subset of 16 compounds was modeled using the connectivity indices. The equation is limited in its use but indicated future modeling directions for Vmax.

Relationships between carcinogenicity and theoretical reactivity indices in polycyclic aromatic hydrocarbons.

Theoretical reactivity indices have been used to examine the metabolic reactions presumed, on the basis of recent biochemical evidence, to be responsible for the transformation of polycyclic aromatic hydrocarbon precarcinogens to ultimate carcinogens. Of a large number of indices examined, several show strong correlations with carcinogenic activity in a set of 25 representative compounds. The results support the belief that specific transformations involving dihydrodiol, "bay-region" epoxide, and carbonium ion intermediates are responsible for the carcinogenic activity of these compounds. Additional implications of the results are discussed, including the suggestion that this type of analysis might provide a rapid and simple means for prescreening compounds for potential carcinogens.