Hydrated electrons in water clusters: inside or outside, cavity or non-cavity?

We compare the applicability of three electron-water molecule pseudopotential models in modeling hydrated electron physical properties. The analysis is based on a) simple quantum mechanical model calculations, and b) one-electron mixed quantum-classical molecular dynamics simulations of an excess electron in size selected water cluster anions. The quantum mechanical calculations illustrate that the recently suggested Larsen-Glover-Schwartz (LGS) model predicts a too attractive potential in the vicinity of the oxygen. As a result, the LGS ground state eigenvalue and the asymptotic behavior of the model wave function are inaccurate. The Turi-Borgis (TB) potential used for comparative purposes reproduces these properties satisfactorily. Mixed quantum-classical molecular dynamics simulations on negatively charged water clusters provide an ideal test case for further testing the potentials. In addition to the LGS and TB models, we also investigated a modified form of the LGS model (m-LGS) that were introduced to correct the huge overbinding of the electron in bulk LGS simulations. While the LGS and m-LGS models predict non-cavity hydrated electron structure in clusters at room temperature, the TB potential prefers the traditionally accepted cavity structure. As another major difference, the electron exclusively localizes in the interior of the clusters in LGS based simulations, while two possible isomers (interior vs. surface state isomers) emerge from TB calculations. The computed associated physical properties are also analyzed and compared to available experimental data. We found that the LGS and m-LGS potentials provide results that are inconsistent with the size dependence of the experimental data. In particular, LGS simulations fail to reproduce the trends of the radius of the excess electron and the position of the absorption spectra with cluster size. The simulated TB tendencies are qualitatively correct. In conclusion, we observe that the cavity preferring pseudopotential model results physical properties in significantly better agreement with experimental data than the models predicting non-cavity structure for the hydrated electron