Charge transfer underestimation

Time-dependent density functional theory (TDDFT) becomes widely used as a simple method for rapid and accurate calculations of molecular excitation energies. It has, however, been reported that conventional TDDFT calculations underestimate Rydberg excitation energies, oscillator strengths, and charge-transfer excitation energies. 

McDowell et a/.80 calculated = (0) and Aa for a set of smaller molecules using different theoretical methods. These results are reproduced in Table 13. They show that the dipole moment in general is underestimated by the Hartree-Fock calculations and overestimated by the LDA and GGA calculations. This is somewhat surprising, since - as we have seen above - density-functional calculations tend to underestimate charge transfers whereas these are overestimated by Hartree-Fock calculations. The Table shows also that the polarizabilities, which essentially are the first derivatives of the dipole moment with respect to field strengths, show some more scatter. 

The major application of this technique, principally by Lindholm and co-workers (see Chapter 10), has capitalized on the above limitation in a study of charge-transfer processes, where the products may exhibit a thermal energy distribution. Even in this application, cross sections are difficult to obtain because the sampling volume is not well defined. Lindholm has been careful to quote only Q values which are estimates of the relative reaction efficiencies. There is another reason why any such cross section so measured may be unreliable. It is plausible, and indeed it has recently been demonstrated, that charge-transfer reactions may yield some products which are forward-scattered in the laboratory framework these would result from collisions with small impact parameters. To the extent that these products will not be detected in a transverse tandem machine, the measured cross section will be underestimated. 

---