Transient Nuclear-Magnetic-Resonance Study of the Conduction Band of MetallicNaxWO3:W183Relaxation

Transient nuclear-magnetic-resonance measurements reported previously for the ${}_{}{}^{23}{}_{}{}^{}\mathrm{Na}$ resonance in metallic sodium tungsten bronzes ${\mathrm{Na}}_x\mathrm{W}{\mathrm{O}}_3$ have been extended to the ${}_{}{}^{183}{}_{}{}^{}\mathrm{W}$ resonance in these compounds. Measurements of the ${}_{}{}^{183}{}_{}{}^{}\mathrm{W}$ spin-lattice relaxation time $T_1$ and the spin-echo phase-memory time $T_2$ have been carried out for $x=0.56, 0.65, 0.73, 0.81, \mathrm{and} 0.89$ in the temperature range $1\le T\le 4°$K. Whereas $T_{1}T$ is at least 4400 sec °K for the ${}_{}{}^{23}{}_{}{}^{}\mathrm{Na}$ resonance, $T_{1}T$ was found to have values between 6 and 16 sec °K (depending on the $x$ value) for the ${}_{}{}^{183}{}_{}{}^{}\mathrm{W}$ resonance. An analysis of these relaxation times provides conclusive evidence that the conduction band in these compounds has no significant sodium $3s$ or $3p$ character, but instead is based on tungsten $5d$ states as postulated initially by Sienko. Tungsten $6s$ states play a relatively minor role in the conduction band, contributing less than 10% to the total density of states $\mathrm{N}\left({\zeta }_0\right)$ at the Fermi surface. Core polarization provides the largest contribution to the ${}_{}{}^{183}{}_{}{}^{}\mathrm{W}$ resonance shift, whereas the orbital hyperfine interaction probably provides the most important of the conduction-electron contributions to nuclear relaxation. The orbital magnetic susceptibility in ${\mathrm{Na}}_x\mathrm{W}{\mathrm{O}}_3$ is estimated to be small, having an upper limit of 10×${10}^{-6\mathrm{}}$ emu/mole for $x=0.56\mathrm{}$ and 18×${10}^{-6\mathrm{}}$ emu/mole for $x=0.89\mathrm{}$. These results should be contrasted with tungsten metal, for which the orbital susceptibility is large (≥66×${10}^{-6\mathrm{}}$ emu/mole); the orbital mechanism predominates in the resonance shift,...