Williams, D. L.; Kuklenz, K. D. A QSAR Model for Predicting Solvents and Solvent Blends for Energetic Materials, Proceedings of the International Annual Conference of ICT, 40th (Energetic Materials), Karlsruhe, Germany, 2/1-2/11, (2009)

Researchers in the paint and polymer industry have shown that the Hansen solubility parameters (HSP) are useful for predicting suitable solvents for the filled-polymer formulation process. To apply this work to the high explosive formulation process, the HSPs of the various energetic materials must be determined or predicted.

A quantitative structure activity relationship (QSAR) was developed that is based upon the output of a density functional theory optimization and frequency calculation (B3LYP/6- 31G(d)//B3LYP/6-31G(d)) using the Gaussian 03 computational package. Structural parameters were extracted from the Gaussian output files of each molecular species. These consisted of the geometric mean of the exact polarizability tensors (α , Å3), the dipole moment (μ, Debye) the highest occupied molecular orbital energy (HOMO, Hartree), the number of each type of atom, and the delta charge (Δq) – defined as the difference between the most negative heteroatom and the most positive hydrogen in the molecule. The value of Δq = 0 was given to hydrocarbons by fiat. A stepwise linear regression was used to determine the correlation of these inputs and mathematical transformations of these inputs to the HSPs for a training set of 54 solvents and nitrated compounds. The resulting QSAR matrix was then applied to 23 energetic materials and precursors yielding the HSPs (δD, δP, δH) in MPa1/2.

The HSPs were also determined for HMX, RDX, PETN, and HNS using experimental solubility data and the group additivity methods of Van Krevelen and Stefanis. The QSAR model outperformed the group additivity methods in matching the experimentally determined HSPs using the Hansen distance parameter (Ra) as the figure of merit.

En route to the QSAR model, a very simple model of molar volume was developed wherein the molar volume is computed directly from the molecular formula CaHbNcOdSePfFgClhBri via the following equation: Vm = 12.53 + 8.77a + 3.96b + 4.87c + 6.12d + 17.22e + 19.45f + 9.70g + 18.66h + 20.74i. The correlation of this equation with the literature values of 183 molecules was 99.67% with an R2 = 0.9847 over a range of 400 cm3/mol.