Improved Electrical, Thermal, and Thermoelectric Properties Through Sample‐to‐Sample Fluctuations in Near‐Percolation Threshold Composite Materials

Effective medium theories (EMT) are powerful tools to calculate sample averaged thermoelectric material properties of composite materials. However, averaging over the heterogeneous spatial distribution of the phases can lead to incorrect estimates of the thermoelectric transport properties and the figure of merit ZT in compositions close to the percolation threshold. This is particularly true when the phases’ electronic properties are rather distinct leading to pronounced percolation effects. The authors propose an alternative model to calculate the thermoelectric properties of multi‐phased materials that are based on an expanded nodal analysis of random resistor networks (RRN). This method conserves the information about the morphology of the individual phases, allowing the study of the current paths through the phases and the influence of heterogeneous charge transport and cluster formation on the effective material properties of the composite. The authors show that in composites with strongly differing phases close to the percolation threshold the thermoelectric properties and the ZT value are always dominated exclusively by one phase or the other and never by an average of both. For these compositions, the individual samples display properties vastly different from EMT predictions and can be exploited for an increased thermoelectric performance. A thermoelectric random resistor network model is used to show that sample‐to‐sample fluctuations in near‐percolation composite materials can be exploited to achieve a larger thermoelectric Figure of Merit ZT than predicted by effective medium theory. This effect emerges through the existence of percolation paths that dominate charge carrier transport in the material.