The adiabatic electron affinities (EA ad's) of the neutral hydroquinone radicals of 1 and 2, i.e., 1 · and 2 · , have been investigated by several quantum chemical models. The B3LYP and B3PW91 methods, used with the 6-311G( d, p) basis set, underestimated the EA ad's by about 6–8 kcal/mol, but with the 6-31+G( d) and 6-311+G( d, p) basis sets, results that are in reasonable accord with experiment were obtained, indicating the importance of diffuse functions for anions in the hybrid HF/DF calculations of EA ad's. The B3LYP-calculated results, including those for 1 and 2 (Y.H. Mariam, L. Chantranupong, J. Comput.-Aided Mol. Design 11 (1997) 345), along with literature values for the EA ad's of the phenyl and phenylnitrene radicals, give the trend in EA ad's that follows the order: phenoxyl (50.6)> 1 · (44.7)> 1 (43.7)> 2 · (38.7)>anilino (36.5)> 2 (34.7)>phenylnitrene (33.4)>phenyl (27.7). Apparently, the EA ad's of both radicals 1 · and 2 · are, respectively, greater than the EA ad's of the parent quinone and iminoquinone 1 and 2. UHF, PUHF, UMP2 and PMP2 energy calculations using B3LYP/6-311+G geometries gave wavefunctions that were found to be highly spin contaminated, and the UHF, PUHF and PMP2 models underestimated the EA ad's, while the UMP2 and the AM1 half-electron results were overestimated. Spin annihilation (in the UHF and UMP2 formalisms) of the first contaminant (quartet) does not apparently remove all the spin contamination. The results indicate, in addition to gradient corrections to the Slater exchange and the correlation functionals (Y.H. Mariam, L. Chantranupong, J. Comput.-Aided Mol. Design 11 (1997) 345), correlation, spin annihilation and inclusion of diffuse functions are important in B3LYP calculations of electron affinities for the systems studied in this work. Since the 6-31+G( d) basis set, with considerably less basis functions, is computationally less costly than the 6-311+G( d, p) basis set, it is recommended for quinone systems of interest where the computational cost is a consideration. The transferability of this computational model to larger systems is not, however, warranted and further work is needed.
Read full abstract