Assessment of new DFT methods for predicting vibrational spectra and structure of cisplatin: which density functional...

Ten different DFT methods, including several recently developed functionals have been tested for their performances in prediction of infrared and Raman spectra and molecular structure of cisplatin. The assessed DFT methods cover the range from meta-GGA to hybrid, double hybrid and long-range corrected hybrid models (M06-L, M06, M06-2X, PBE0, mPW1PW, B3LYP, B2PLYP, CAM-B3LYP, ωB97XD and LC-ωPBE). The calculated structural parameters and theoretical spectra have been compared to the corresponding experimental data. It is shown that the LC-ωPBE scheme is superior to other DFT methods in predicting the geometry of cisplatin. Unfortunately, the M06-L, M06-2X and B3LYP functionals are deficient in the evaluation of the strength of two Pt←NH3 coordination bonds in cisplatin (the calculated bond lengths are too long and the Pt-N stretching frequencies are underestimated). Both the PBE0 and mPW1PW functionals, in conjunction with the LanL2TZ(f) basis set for Pt give very similar theoretical results and seem to be the best methods for predicting the IR and Raman spectra of cisplatin. The long-range corrected functionals (LC-ωPBE, ωB97XD and CAM-B3LYP) have shown good performances in predicting the frequencies of Pt-ligand vibrations and are promising new tools for theoretical study of novel platinum(II) compounds.