Magnetism out of antisite disorder in the $J=0$ compound ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$

We systematically investigate the magnetic properties and local structure of ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ to demonstrate that Y and Ir lattice defects in the form of antiphase boundary or clusters of antisite disorder affect the magnetism observed in this $5d^4$ compound. The experimental investigation involved comparison of the magnetic properties and atomic imaging of (1) a slow-cooled crystal, (2) a crystal quenched from $900{}^{\circ }\mathrm{C}$ after growth, and (3) a crystal grown using a faster cooling rate during growth than the slow-cooled one. Atomic-scale imaging by scanning transmission electron microscopy (STEM) shows that quenching from $900{}^{\circ }\mathrm{C}$ introduces Ir-rich antiphase boundaries in the crystals, and a faster cooling rate during crystal growth leads to clusters of Y and Ir antisite disorder. Compared to the slow-cooled crystals, ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ crystals with clusters of antisite defects have a larger effective moment and a larger saturation moment, while quenched crystals with Ir-rich antiphase boundary show a slightly suppressed moment. Our DFT and model magnetic Hamiltonian calculations suggest magnetic condensation is unlikely, as the energy to be gained from superexchange is small compared to the spin-orbit gap. However, once Y is replaced by Ir in the antisite disordered region, the picture of local nonmagnetic singlets breaks down and magnetism can be induced. This is because of (a) enhanced interactions due to increased orbital overlap and (b) increased number of orbitals mediating the interactions. Our work highlights the importance of lattice defects in understanding the experimentally observed magnetism in ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ and other $J=0$ systems.

Figure 1

(a) Temperature profiles for the growth of ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ crystals. BYIO-4d: slow-cooled crystal. BYIO-4d-Q: quenched into iced water from $900{}^{\circ }\mathrm{C}$. BYIO-16d: a faster cooling rate ($16{}^{\circ }\mathrm{C}/\mathrm{h}$) during crystal growth. (b) A summary of defects observed by STEM, lattice parameter, effective moment, and saturation moment at 2 K for all three samples.

Figure 2

HADDF-STEM images in the [110] projection of ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ crystals. (a)–(c) show the low magnification HAADF image for BYIO-4d, BYIO-16d, and BYIO-4d-Q, respectively. (d)–(f) show the high-resolution HAADF image for BYIO-4d, BYIO-16d, and BYIO-4d-Q, respectively. Inset of (d) highlights the ordering in BYIO-4d (yellow: Ba, green: Y, blue: Ir, and red: O). Inset of (e) shows the details of a disordered region of BYIO-16d. Inset of (f) highlights a narrow defected region through the ordered crystal. (g), (h), and (i) show the intensity profile of atomic columns along [001] as marked by the (yellow) lines in (d)–(f) for BYIO-4d, BYIO-16d, and BYIO-4d-Q, respectively.

Figure 3

Temperature dependence of magnetic susceptibility $\chi$ and $1/\chi$ for BYIO-4d single crystals in the temperature range $2\mathrm{K}\le T\le 750\mathrm{K}$ measured in an applied magnetic field of 20 kOe. The solid (blue) line shows the linear fitting of 1/$\chi$ above 200 K. The dashed curve shows the fitting with a Curie-Weiss law as described in the text.

Figure 4

Temperature dependence of magnetic susceptibility of ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ single crystals in the temperature range $2\mathrm{K}\le T\le 300\mathrm{K}$ measured in an applied magnetic field of 50 kOe. The dashed curves show the Curie-Weiss fitting in the temperature range $15\mathrm{K}\le T\le 300\mathrm{K}$. The fitting parameters are listed in the inset.

Figure 5

Field dependence of magnetization of ${\mathrm{Ba}}_2{\mathrm{YIrO}}_6$ single crystals at 2 K.

Figure 6

The GGA band structure without spin-orbit coupling is shown. A tight-binding model with $t_{2g}$ Wannier orbitals is fit to the three bands pictured.

Figure 7

The triplet excitation spectrum is plotted using a spin-orbit gap of 350 meV. Superexchange reduces the cost to make a triplet excitation at the edge of the Brillouin zone. The lowest energy excitations appear along $X-W$ about 20 meV lower than the original 350 meV spin-orbit gap.