Correlation effects in hopping conductivity

We introduce an alternative mapping of the hopping problem onto a random resistor network, which takes into account correlations among the occupation probabilities of different sites. The effect of correlations on the low-temperature conductivity of one-dimensional chains in the nearest-neighbor and variable-range-hopping cases is investigated by approximation of the network conductances by the critical conductance at the percolation threshold. Correlation effects are found to lead to increased activation energies and a mesoscopic behavior which is much closer to experimental data than that predicted by mean-field theory.