Frustrated fcc antiferromagnet ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$: Structural characterization, magnetic properties, and neutron scattering studies

We report the crystal structure, magnetization, and neutron scattering measurements on the double perovskite ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$. The $Fm\overline{3}m$ space group is found both at 290 K and 3.5 K with cell constants $a_0=8.3541\left(4\right)$ Å and $8.3435\left(4\right)$ Å, respectively. ${\mathrm{Os}}^{5+} \left(5d^3\right)$ ions occupy a nondistorted, geometrically frustrated face-centered-cubic (fcc) lattice. A Curie-Weiss temperature $\theta \sim -700$ K suggests the presence of a large antiferromagnetic interaction and a high degree of magnetic frustration. A magnetic transition to long-range antiferromagnetic order, consistent with a type-I fcc state below $T_{\mathrm{N}}\sim 69$ K, is revealed by magnetization, Fisher heat capacity, and elastic neutron scattering, with an ordered moment of 1.65(6) ${\mu }_B$ on ${\mathrm{Os}}^{5+}$. The ordered moment is much reduced from either the expected spin-only value of $\sim 3 {\mu }_B$ or the value appropriate to $4d^3 {\mathrm{Ru}}^{5+}$ in isostructural ${\mathrm{Ba}}_2{\mathrm{YRuO}}_6$ of 2.2(1) ${\mu }_B$, suggesting a role for spin-orbit coupling (SOC). Triple-axis neutron scattering measurements of the order parameter suggest an additional first-order transition at $T=67.45$ K, and the existence of a second-ordered state. Time-of-flight inelastic neutron results reveal a large spin gap $\Delta \sim 17$ meV, unexpected for an orbitally quenched, $d^3$ electronic configuration. We discuss this in the context of the $\sim 5$ meV spin gap observed in the related ${\mathrm{Ru}}^{5+},4d^3$ cubic double perovskite ${\mathrm{Ba}}_2{\mathrm{YRuO}}_6$, and attribute the $\sim 3$ times larger gap to stronger SOC present in this heavier, $5d$, osmate system.

Figure 1

Left: Representation of the double perovskite structure of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ emphasizing the octahedral environment of ${\mathrm{Os}}^{5+}$. Right: The magnetic ${\mathrm{Os}}^{5+}$ ions form a frustrated fcc sublattice made of edge-sharing tetrahedra.

Figure 2

Room-temperature powder neutron diffraction patterns of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ recorded at 2.37 Å (a) and 1.33 Å (b). The 3.5 K neutron diffraction patterns at 2.37 Å (c) and at 1.33 Å (d) which show that there is no structural change from 290 K. Tick marks represent ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6,{\mathrm{Y}}_2{\mathrm{O}}_3$ ($\sim 1\%$ in mass), ${\mathrm{Ba}}_{11}{\mathrm{Os}}_4{\mathrm{O}}_{24}$ ($\sim 5\%$ in mass), and the magnetic phase of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ (c), respectively.

Figure 3

${}^{89}\mathrm{Y}$ MAS NMR signal intensity of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ as a function of frequency ($B_0=11.7$ T) at 345 K. [Inset: ${}^{207}\mathrm{Pb}$ MAS NMR of $\mathrm{Pb}\left({\mathrm{NO}}_3\right){}_2$ at 345 K.] The arrows define the region where a peak due to Y/Os antisite disorder would be expected based on results from ${\mathrm{Ba}}_2{\mathrm{YRuO}}_6$ [13].

Figure 4

(a) The temperature dependence of the inverse magnetic susceptibility of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ at 0.05 T and the field dependence of magnetization at 2 K (inset). The red line is a Curie-Weiss fit, performed for $150<T<300$ K. (b) Fisher heat capacity of ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$. A single peak is observable at 68.5 K, unlike the case of ${\mathrm{Ba}}_2{\mathrm{YRuO}}_6$. The inset is a zoom near the transition.

Figure 5

Neutron diffraction patterns on ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$ collected at 3.5 and 75 K and type-I fcc antiferromagnetic magnetic structure (inset). The magnetic reflections are indexed with a propagation vector $k=(1,0,0)$. The broad peak at the (100) magnetic reflection position indicates short-range magnetic correlations at 75 K. Red line is a structural and magnetic refinement of the diffraction pattern observed at 3.5 K (gray circles). The overfitting of the (120) and (211) magnetic reflections is ascribed to inaccuracies in the ${\mathrm{Os}}^{5+}$ form factor.

Figure 6

Temperature dependence of the elastic neutron scattering intensity in ${\mathrm{Ba}}_2{\mathrm{YOsO}}_6$, for $E_i=40$ meV, integrated over the energy range $[-1,1]$ meV. A long-range order is observed below 70 K, with magnetic Bragg peaks at the $\left[100\right]$ ($\left|Q\right|=0.75 \AA {}^{-1}$) and $\left[110\right]$ ($\left|Q\right|=1.06 \AA {}^{-1}$) positions. A high-temperature background data set (150 K) has been subtracted from each data to isolate the magnetic contribution.

Figure 7

(a) Temperature dependence of the [100] intensity from the N5 data (upon warming and upon cooling), showing a discontinuity at $T_{c,2}=67.45$ K. Inset: Zoom near $T_{c,2}$ (red arrow). (b) Blue and orange lines are fits to standard power law to extract the critical behavior (see text).

Figure 8

(a)–(f) Evolution of the neutron scattering intensity $S\left(\right|Q|,E)$ across the magnetic transition at $T_{\mathrm{N}}\sim 69$ K. The background intensity of the sample cell has been subtracted off. Below 70 K, a spin gap $\Delta$ is progressively opening, finally reaching a $\sim 17$ meV value at 20 K and below. The upper intensity scale refers to (a) and (b) only, while the lower intensity scale refers to (c) to (f), highlighting the opening of the gap.

Figure 9

(a) Energy cuts of Fig. 8, integrated over $Q=[0.5,1.5] \AA {}^{-1}$, and showing the formation of the spin gap $\Delta$. The red line is a fit of the 1.8 K data to two Gaussians centered on elastic and inelastic positions, resulting in a spin gap value of $\Delta =18\left(2\right)$ meV and a spin wave bandwidth above the gap of 15(5) meV. (b) $Q$ cuts of the scattered intensity measured with an incident energy $E_i=40$ meV, integrated over $E=[6,8]$ meV, i.e., in the inelastic regime, within the gap. The low-$T$ data set (1.8 K) has been subtracted as a background, and the plot shows the magnetic spectral weight filling the gap as the temperature is increased. Tick marks show the [100] and [110] positions. (c) $T$ evolution of ${\chi }^{\prime \prime }\left(\right|Q|,E)$, extracted from the data of panels (a) and (b) (see text), normalized by their respective plateau value above $T=70$ K. Below 70 K, the intensity decreases exponentially (red line), allowing us to estimate $\Delta =16\left(2\right)$ meV from this activated temperature dependence.