Comment on "On the accuracy of force fields for predicting the physical properties of dimethylnitramine"

Zheng and Thompson have recently reported a comparison of three atomistic force fields for prediction of physical properties of dimethylnitramine (DMNA) from molecular dynamics (MD) simulations.1 Specifically, they compared the rigid molecule force field by Sorescu, Rice, and Thompson (SRT);2 the generalized AMBER- (with partial charges assigned by Zheng and Thompson); and a fully atomistic quantum chemistry (QC)-based force field developed by Smith, Bharadwaj, Bedrov, and Ayyagari (SBBA)3 for prediction of unit cell parameters, bulk modulus, and melting temperature for crystal DMNA, as well as volumetric properties of liquid DMNA. The authors found that simulations using the SBBA force field predict DMNA properties that significantly deviate from experimental data. This was particularly apparent for the melting temperature of DMNA, which was found to be 70 K lower than the experimental value. The two other force fields investigated overestimated the melting temperature of DMNA, but to a lesser extent. In an attempt to improve prediction of the DMNA melting temperature, Zheng and Thompson have adjusted the AMBER dihedral potential, arguing that the original potential predicts dihedral barriers that are off a factor of 3 from predictions of QC calculations and experiments.