Thermoelectric power of 1:2 tetracyanoquinodimethanide (TCNQ) salts

Abstract

A distinctive feature of a number of 1:2 tetracyanoquinodimethanide (TCNQ) salts, such as triethylammonium-${(\mathrm{TCNQ})}_{2}$ [TEA-${(\mathrm{TCNQ})}_{2}$], quinolinium-${(\mathrm{TCNQ})}_{2}$, and acridinium-${(\mathrm{TCNQ})}_{2}$ is a constant thermopower $Q\ensuremath{\simeq}\ensuremath{-}60$ \ensuremath{\mu}V/K over a wide temperature range, \ensuremath{\gtrsim} 100 K, along the highly conducting or TCNQ chain direction. These salts are characterized by $\ensuremath{\rho}$, the fraction of filled sites on the TCNQ chain \ensuremath{\simeq} \textonehalf{}. Past calculations of Beni et al. have shown that the extended Hubbard model with very strong on-site Coulomb repulsion and zero bandwidth can account for the magnitude and temperature variation of $Q$ for quinolinium-${(\mathrm{TCNQ})}_{2}$ and acridinium-${(\mathrm{TCNQ})}_{2}$. I discuss the approximation of zero bandwidth and conclude that it is not justifiable. I then show that a near-constant value close to -60 \ensuremath{\mu}V/K for temperatures \ensuremath{\gtrsim} 100 K can be obtained using a model (although not the extended Hubbard model) with two bands having nearly identical widths and scattering times and retaining the feature of very strong on-site repulsion. Near identity of the two bands is reasonable for $\ensuremath{\rho}\ensuremath{\simeq}\frac{1}{2}$. This model can also account for the magnitude and temperature variation of the conductivity of these salts over the same large temperature range.