Nature of “Disorder” in the Ordered Double Perovskite ${\mathrm{Sr}}_2{\mathrm{FeMoO}}_6$

The degree of $B/B^{\prime }$ alternate cation order is known to heavily influence the magnetic properties of $A_{2}B{B}^{\prime }{\mathrm{O}}_6$ double perovskites although the nature of such disorder has never been critically studied. Our detailed x-ray absorption fine structure studies in conjunction with synchrotron radiation x-ray diffraction experiments on polycrystalline ${\mathrm{Sr}}_2{\mathrm{FeMoO}}_6$ samples with various degrees of disorder reveal that a very high degree of short range order is preserved even in samples with highly reduced long range chemical order. Based on these experimental results and with the help of detailed structural simulations, we are able to model the nature of the disorder in this important class of materials and discuss the consequent implications on its physical properties.

Figure 1

Slices of 3D models of $\mathrm{Fe}/\mathrm{Mo}$ cubic lattices (light and dark grey sites representing the Fe and Mo ordered positions, red (blue) sites being Fe (Mo) on the Mo (Fe) sublattice) built (see text) with the same degree of long rang order, $S=0.51$, but different kinds of local disorder are shown in panels (a) to (c). (a) random AS distribution: $\xi ={\xi }_{\mathrm{AS}}=0.63$, (b) small antiphase domains with $\xi =0.85$ and (c) large antiphase regions with $\xi =0.99$. (d) (pictorial view): an applied ${\overrightarrow{B}}_{\mathrm{ext}}$ aligns neighboring domains while AFM coupling persists in the vicinity of APB defects.

Figure 2

XRD patterns for all the investigated samples. In the insets: (a) the region corresponding to the (101) superlattice peak for samples $A$, $B$ and $E$. (b) the width (FWHM) of lattice (open diamonds) and superlattice (filled diamonds) reflections.

Figure 3

XAFS results. (a),(b) experimental (dots) and best fit (full lines) for all samples/edges investigated. (c) modulus and imaginary part of FT from Fe $K$-edge XAFS data of the sample $D$ (dots), the best fit obtained refining the local order parameter $\xi$ (upper curves) and fixing it to the value given by XRD (lower curves) are shown for sake of comparison. (d) $|FT|$ of $k$-weighted experimental (dots) and best fit (full lines) XAFS data at the Mo $K$ edge.

Figure 4

Left: the calculated SL peak intensity for the two cases with the same LRO $S=0.51$, but different kinds of defects [see Figs. 1a, 1b]. The superlattice peak calculated for the $S=1$ case is also shown for sake of comparison. Right: the same data plotted in semilog scale, the weak effect due to different kinds of local order can be put in evidence.