Giant thermovoltage in single InAs-nanowire FETs

The quest for solid-state thermoelectric (TE) conversion has been one of the prime driving forces behind semiconductor research before the discovery of the transistor effect. The achievement of efficient TE devices indeed requires a non-trivial trade-off between interdependent material characteristics and can be expressed in terms of the maximization of the figure of merit ZT = S2T/, where S is the Seebeck coefficient,  and  are the electrical and thermal conductivity, T the average operation temperature. Large ZT values, though, have proven elusive over the past decades despite the great design flexibility offered by semiconductor materials1. Here, we present results on thermovoltages in single InAs nanowire (NW) field effect transistors. Thanks to a buried heating scheme (see Fig. 1), we achieved both a large thermal bias, T > 10K and a strong field-effect modulation of electric conduction through nanostructures randomly positioned on oxidized silicon. This experimental arrangement allows a detailed mapping of S(; T) and the comparison with classic models for thermoelectric transport in degenerate semiconductors. The adopted approximations lead to a novel estimate of the electron mobility e~ 11000 cm2/Vs in the NW. Such value is significantly larger than the one obtain by field-effect, e, FE, on the same wire. The discrepancy can be understood in terms of history effects in the NW free charge filling by electric field, as a consequence of the slow surface and trap charge dynamics. On the other hand, the S vs  dependence is largely independent of spurious charge screening and hysteresis. Our results also indicate that special care should be taken in the interpretation of transport results based on field-effects in these nanostructures.