Effect of pressure on impurity-impurity interactions in dilute Au:Mn, Cu:Mn, and Au:Fe spin-glass alloys

The electrical resistivity of the dilute spin-glass alloys Au-0.10-at.% Mn, 0.15-at.% Mn, 0.10-at.% Fe, 0.13-at.% Fe, and Cu-0.15-at.% Mn has been measured from 1.2 to 40 K at pressures to 100 kbar. In these alloys the cooperative "locking-in" of the impurity spins at a temperature $T_0$ leads to a resistivity maximum at $T_{max}$. Application of pressure is found to shift $T_{max}$ in a manner which is strongly system dependent: $\frac{d{T}_{max}}{dP}>0\mathrm{}$ for Au:Mn, $\frac{d{T}_{max}}{dP}\approx 0$ for Cu:Mn, and $\frac{d{T}_{max}}{dP}<0\mathrm{}$ for Au:Fe. These results are shown to be clearly incompatible with the widely held belief that $k{T}_{max}\approx {\Delta }_{\mathrm{R}\mathrm{K}\mathrm{K}\mathrm{Y}}$, where ${\Delta }_{\mathrm{R}\mathrm{K}\mathrm{K}\mathrm{Y}}\approx \frac{c{J}^{2}S(S+1\mathrm{})\mathrm{}}{E_F}$ is the average strength of the Ruderman-Kittel-Kasuya-Yosida interaction, and indicate that $T_{max}$ is a function of both ${\Delta }_{\mathrm{R}\mathrm{K}\mathrm{K}\mathrm{Y}}$ and $T_K$, the Kondo temperature. This expectation is confirmed in a recent theory of Larsen who introduces an impurity-impurity interaction strength ${\Delta }_c$ into the parquet approximation of the Kondo resistivity and obtains an explicit expression for $T_{max}=T_{max}({\Delta }_c,T_K)$. It is shown that both the sign and the magnitude of $\frac{d{T}_{max}}{dP}$ for the systems studied here are a natural consequence of both increasing $T_K$ and ${\Delta }_c$ in all systems and depend on the relative magnitudes of ${\Delta }_c$ and $T_K$. In particular, one would expect to find $\frac{d{T}_{max}}{dP}<0\mathrm{}$ in systems such as Au:Fe with relatively high Kondo temperatures. A further result of the analysis is that