Predicting Infrared Spectra of Nerve Agents Using Density Functional Theory

• DOI: 10.1007/s10812-016-0347-5
• Publication Date: 2016-09-05T03:57:18Z
• Authors: Y.-P. Zhang; H.-T. Wang; W.-P. Zheng; C. Sun; Y. Bai; X.-D. Guo; H. Sun
• Keywords: Mean square

Vibration frequencies of four nerve agents and two simulators are calculated using B3LYP coupled with ten basis sets. To evaluate the accuracy of calculated spectra, root mean square error (RMSE) and weighted cross-correlation average (WCCA) are considered. The evaluation shows that B3LYP/6-311+g(d,p) performs best in predicting infrared spectra, and polarization functions are found to be more important than diffusion functions in spectra simulation. Moreover, B3LYP calculation underestimates frequencies related to the P atom. The WCCA metric derives 1.008 as a unique scaling factor for calculated frequencies. The results indicate that the WCCA metric can identify six agents based on calculated spectra.