Potential error from using ZT to optimize thermoelectric performance

Despite known shortcomings, the dimensionless figure of merit, ZT, continues to be the most widely used material performance metric in thermoelectrics research. Here, we expose yet another potential drawback of using the ZT to evaluate new thermoelectric materials. Specifically, by comparison with more accurate methods of estimating thermoelectric material energy conversion efficiency, we find that using the ZT as the optimization criterion could possibly lead to erroneous conclusions about the optimal doping density when optimizing performance. Using a simple semiconductor model, we find that optimal doping densities as determined using the ZT could be in error by a factor of two or more, potentially contributing a relatively small but non-negligible loss in device efficiency that could be easily avoided by using modern methods for estimation of material conversion efficiency.