ABSTRACT
Thermodynamics does not place any upper limit on the dimensionless thermoelectric figure-of-merit,
ZT, but for many years it was impossible to find values significantly greater than unity. Empirically, one
can predict an upper bound by combining the best electronic properties of any known crystalline
material with a phonon conductivity that is typical of glass. By this means, one arrives at a highest value
of ZT equal to about four. Recently, however, it has been demonstrated that certain low-dimensional
structures can have values of the figure-of-merit that exceed those found in bulk materials. This allows us
to upgrade our predicted limit for ZT to, perhaps, 20. If this value were eventually achieved,
thermoelectric energy converters would have no less than about 50% of the efficiency of a Carnot cycle.
The performance would then be comparable with the best that is likely to be achieved using converters
that are based on the vacuum diode.