H. J. Goldsmid

Although there is no thermodynamic limit to the dimensionless thermoelectric figure of merit, ZT, a number of authors have reached the conclusion that this quantity will never exceed 4. This prediction was made before the advent of low-dimensional systems and when it seemed to be very difficult to find values of ZT in excess of unity, particularly at ordinary temperatures. Now there are reliable reports of ZT substantially greater than unity for special configurations (e.g. nanostructures) of established thermoelectric materials at room temperature and for conventional thermocouples made from some of the newer compounds at elevated temperatures. It therefore seems to be an appropriate time to review the prediction of an upper limit for ZT.

There seems to be no reason to increase the bound on ZT for bulk materials but a significantly higher value of, perhaps, 20 may be possible for superlattices, quantum wires and quantum dots, through an improvement in the power factor. It is possible that the efficiency or coefficient of performance may rise to about half the Carnot cycle limit.

Both a high power factor and a low thermal conductivity are important in a thermoelectric material, but the latter quantity in the solid state cannot be less than the lattice contribution. Transport in a vacuum is, therefore, a highly attractive proposition. There now seems to be a reasonable chance that the problems associated with vacuum thermionic converters at moderate temperatures may eventually be overcome. If this happens, these devices should be as good as the best that one can now predict for thermoelectric converters. It is envisaged that energy converters of the future may combine solid-state and vacuum technologies.