DERIVATION OF CLASS-II FORCE-FIELDS .2. DERIVATION AND CHARACTERIZATION OF A CLASS-II FORCE-FIELD, CFF93, FOR THE ALKYL FUNCTIONAL-GROUP AND ALKANE MOLECULES

A second generation Class II force field is derived for the alkyl grou p and alkane molecules. The Class II functional form is presented and force constants are given. The criteria that define this second genera tion force field are the following: (1) it accounts for the properties of both isolated small molecules, condensed phases, and macromolecula r systems and (2) the functional form is characterized by being anharm onic, with quartic stretching and quartic angle bending, and includes a variety of important intramolecular coupling interactions. It is als o characterized by a soft repulsion, either 9th power or exponential, rather than the more usual 12th power repulsion. The force field is de rived by scaling an analytical representation derived from a fit to a quantum mechanical energy surface. Only seven parameters were needed i n this scaling to reproduce 150 experimental observables. The resultin g Class II force field is shown to fit the structural, energetic, and dynamic properties of the alkane molecules comprising the training set well. These molecules include small acyclic alkanes, strained molecul es such as isobutane, and small rings including cyclopropane and cyclo butane. Thus the properties of highly strained molecules including sma ll rings are accounted for with one set of transferable parameters. Th e results are compared with those obtained from Class I diagonal quadr atic force fields commonly used in simulations of biological systems a nd the Class II functional form is shown to reproduce trends unattaina ble by the simpler forms where an harmonicity and coupling interaction s are not accounted for. Most dramatic is its ability to fit the small ring compounds, cyclopropane and cyclobutane, with the same transfera ble energy functions that account for larger rings and acyclic molecul es. This is the first energy surface able to achieve this range of app licability, a degree of transferability hypothesized in the literature to be unachievable. Finally, and perhaps most importantly it is point ed out that the methodology presented here provides a paradigm for the straightforward derivation of force fields for arbitrary molecules of interest even where experimental data is sparse or missing. The force field can be derived based on the techniques described here as long a s the quantum mechanical calculation can be carried out. The resulting quantum force field, at that point, could either be used on its own o r scaled to provide a pragmatic and reasonably accurate force field.