Spin frustration and magnetic ordering in the $S=\frac{1}{2}$ molecular antiferromagnet ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$

We have investigated the low-temperature magnetic state of face-centered-cubic (fcc) ${\mathrm{Cs}}_3{\mathrm{C}}_{60}$, a Mott insulator and the first molecular analog of a geometrically frustrated Heisenberg fcc antiferromagnet with $S=1/2$ spins. Specific heat studies reveal the presence of both long-range antiferromagnetic ordering and a magnetically disordered state below $T_N=2.2$ K, which is in agreement with local probe experiments. These results together with the strongly suppressed $T_N$ are unexpected for conventional atom-based fcc antiferromagnets, implying that the fulleride molecular degrees of freedom give rise to the unique magnetic ground state.

Figure 1

(a) Temperature dependence of the specific heat, $C$, for ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$ (triangles) and A15-${\mathrm{Cs}}_3{\mathrm{C}}_{60}$ (circles). The solid line represents the phonon specific heat $C_{\mathrm{ph}}$ for A15-${\mathrm{Cs}}_3{\mathrm{C}}_{60}$, which is estimated from a fit above $T_N$ using Eq. (1). Inset: Temperature dependence of the magnetic specific heat, $C_m$, for A15-${\mathrm{Cs}}_3{\mathrm{C}}_{60}$ obtained by subtracting the phonon background contributions from total $C$. (b) Low-temperature part of $C/T$ for ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$, plotted as a function of $T^2$. Arrows indicate the reported Néel temperature, $T_N=2.2$ and 46 K in fcc- and A15-${\mathrm{Cs}}_3{\mathrm{C}}_{60}$, respectively.

Figure 2

Temperature dependence of the specific heat, $C/T\left(T\right)$, for antiferromagnetic ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$ and superconducting Rb${}_{0.35}$${\mathrm{Cs}}_{2.65}{\mathrm{C}}_{60}$, (a) below 35 K and (b) below 5 K, respectively. The arrow indicates the superconducting transition temperature $T_c=27$ K for Rb${}_{0.35}$${\mathrm{Cs}}_{2.65}{\mathrm{C}}_{60}$. Inset: Normalized electronic specific heat $C_e/{\gamma }_n{T}_c$ calculated using $T_c=27$ K in the two distinct limits with ${\Delta }_0/k_{\mathrm{B}}{T}_c=1.76$ (solid line) and 2.50 (dashed line).

Figure 3

(a) Magnetic specific heat divided by temperature, $C_m/T$, as a function of temperature. $C_m$ was obtained by subtracting $C$ for superconducting Rb${}_{0.35}$${\mathrm{Cs}}_{2.65}{\mathrm{C}}_{60}$ from $C$ for ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$. (b) $C_m$ plotted as a function of $T$. (c) Low-temperature part of $C_m/T$ as a function of $T^2$. $C_m/T$ obtained by the two procedures is shown. Solid and dashed lines represent the linear fit and the estimated magnon contribution for A15-${\mathrm{Cs}}_3{\mathrm{C}}_{60}$, respectively. Arrows represent the reported $T_N$.

Figure 4

(a) Temperature evolution of the ${}^{133}$Cs NMR spectrum in ${\mathrm{fcc}-\mathrm{Cs}}_3{\mathrm{C}}_{60}$ measured at $H=8.0018$ T. (b) Temperature dependence of the nuclear magnetization decay curves for the fast-relaxing component. Solid curves are the fits assuming stretched exponential decay, $1-M\left(t\right)/M_0\propto exp[-{(t/T_1)}^{\alpha }]$, where $\alpha$, $M_0$, and $M\left(t\right)$ are a stretch exponent, the nuclear magnetizations at thermal equilibrium and at a time $t$ after saturation comb pulses, respectively. (c) Temperature dependence of $1{/}^{133}{T}_1$. Filled circles (crosses) were obtained by the fits using the stretched exponential model (simple exponential model). Inset: Temperature dependence of the stretch exponent for the fast-relaxing component. Arrows indicate $T_N=2.2$ K.