Exchange anisotropy-driven noncollinear magnetism and magnetic transitions in MnTiO3 ilmenite

Abstract

Evidence for multiple magnetic transitions and unconventional spin exchange interactions in the ilmenite insulator MnTiO3 is provided via neutron scattering. On cooling, while G-type antiferromagnetic (AFM) order sets in first at 63 K with a k1 = (000) characteristic wave vector, a weaker second magnetic transition with k2 = (00 3/2 ) appears near 42 K, giving rise to a noncollinear structure. Intrinsic buckling of the honeycomb lattice along c creates bond anisotropy and a distorted crystal field that can lead to exchange paths that modulate orbital overlap and spin-orbit coupling. The inelastic spectrum is best described by magnetic exchange anisotropy that breaks the local symmetry of the honeycomb, with competing AFM Heisenberg, Dzyaloshinskii-Moriya and alternate intra-planar ferromagnetic (FM) interactions, that may yield a weakly-coupled ladder system.

Figures (4)

• Figure 1: (a) The unit cell of the R$\overline{3}$ ilmenite structure. (b) Top and side view of a honeycomb cell where red and blue indicate Mn sites that are shifted up or down from the plane. J$_1$ and J$_3 \pm \delta$ are in-plane exchange interactions. (c) The bipartite honeycomb network of magnetic Mn sites is bond anisotropic, represented by the different colored lines. (d) The magnetic susceptibility data at 0.1 T and 0.7 T fields for zero-field cooled (ZFC) and field cooled (FC) measurements and (e) its inverse for the 0.1 T measurement. The data were fit using the Curie-Weiss law from 120 to 295 K. (f) The temperature dependence of the lattice constants $a$ and (g) $c$ obtained from the diffraction data refinement. The dotted lines correspond to M1 (G-type) and M2 (A-type) transitions.
• Figure 2: (a) The diffraction data at 5 K is compared to a model calculated using the R$\overline{3}$ symmetry. The orange tick marks correspond to the nuclear Bragg peaks, the blue to the M1 phase with $k_1$ = (000), and red to the M2 phase $k_2$ = (00$\frac{3}{2}$). (b) The temperature dependence of the diffraction pattern in d-spacing, from 2.6 Å to 4.6 Å. The nuclear peaks are denoted by orange arrows (2.80 Å). Peaks associated with M1 transition are denoted by blue arrows (3.75 Å and 4.25 Å), and M2 is denoted by red arrows (4.40 Å, 3.50 Å, and 3.00 Å). The peak at 2.95 Å is a background peak. (c) The order parameters associated with the two magnetic $k$ vectors. The two transitions are linked to distinct propagation vectors, with M1 corresponding to $k_1$, and M2 to $k_2$. Error bars are smaller than data points. (d) The corresponding $k_1$ and $k_2$ spin structures, and the resultant spin order combining the two $k$ vectors.
• Figure 3: A plot of the dynamic magnetic susceptibility ($\chi^{\prime\prime}$(Q,$\omega$)) vs energy ($\hbar\omega$) as a function of temperature along (a) LQ, and (b) HQ trajectories. Bose correction and background subtraction were performed for the data at all temperatures. (c) The integrated intensity of the LQ spectra from 0 to 12 meV and from 12 to 15 meV in (d). (e) A simulation of the powder averaged spin wave dispersion, S$_\perp$(Q,$\omega$), at 0 K. The exchange constants, J, used for this calculation were, J$_1$=0.70, J$_2$=0.25, J$_3\pm\delta$=-0.32$\pm$0.11, A=-0.008, and D=0.07 meV. The $\chi^{\prime\prime}$(Q,$\omega$) spectrum at 5 K obtained from VISION (blue) along (f) LQ, and (g) HQ paths are compared to a model $\chi^{\prime\prime}$(Q,$\omega$) (green line) determined from a dispersion calculated using J values from Ref. hwang2021spin. A second model $\chi^{\prime\prime}$(Q,$\omega$) (red line) is also shown. This was calculated using J values obtained from this analysis. The calculation accounted for the energy resolution of the instrument and Q resolution set to 0.2 Å$^{-1}$.
• Figure 4: (a) The honeycomb layer with AFM J$_1$ in-plane and AFM J$_2$ out-of-plane. The layers are shifted by $\frac{1}{3}$ along the c-axis. (b) The bi-partite honeycomb cell has two different FM exchange bonds where J$_3$ exchange interactions. The split J$_3$ creates alternate stronger and weaker exchange ferromagnetic paths with $\left( J_{3} \pm \delta \right)$. Also listed are the values used in the Hamiltonian for model 2.