Optical Readout of Reconfigurable Layered Magnetic Domain Structure in CrSBr

TL;DR

This work demonstrates optical readout of a reconfigurable layered magnetic domain structure in the van der Waals magnet CrSBr. A magnetic field along the easy axis drives a cascade of metastable intermediate magnetic states (iMS) whose richness scales with sample thickness, enabling layer-by-layer magnetic reconfiguration. The optical response, captured by magneto-reflectance and magneto-PL, maps the magnetic configurations and is explained by a transfer-matrix multilayer model with AFM/FM dielectric functions and a thickness-dependent FM exciton energy, validating CrSBr as a light-guiding, reconfigurable optical metamaterial. These findings position CrSBr as an intelligent-matter platform for neuromorphic, light-driven information processing where information can be encoded, processed, and stored in reconfigurable magnetic layers.

Abstract

The emergence of intelligent matter has sparked significant interest in next generation technologies. We report on the discovery of a reconfigurable magnetic multilayer domain structure in the van der Waals magnet CrSBr, exhibiting a unique combination of magnetic and optical properties. Applying an external magnetic field along the easy axis drives the hysteretic antiferromagnetic-to-ferromagnetic transition that is not universally binary, but instead develops through a cascade of intermediate magnetic configurations whose multiplicity and stability scale systematically with thickness. This material can be considered as a prototypical intelligent matter, capable of encoding, processing, and storing information through its tunable magnetic structure. The directly linked optical properties of CrSBr, modulated by the magnetic structure, provide a readout mechanism for the stored information compatible with modern information distribution using light. With its adaptive properties, CrSBr is an attractive candidate for neuromorphic circuitries, enabling the design of brain-inspired computing architectures that can learn and evolve in response to changing environments.

Figures (8)

• Figure 1: Thickness dependence of field-driven optical switching in CrSBr. Magneto-reflectance spectra $R/R_0$ with $B$ applied along the easy ($b$) axis. The reflectance ratio $R/R_0$ is color-coded. Shown are magnetic field $B_\mathrm{up}$ and $B_\mathrm{down}$ sweeps for (a) 20 nm thick CrSBr and (b) for 150 nm thick CrSBr. Vertical dashed lines separate AFM, iMS, and FM phases and indicate the critical switching fields $B_\mathrm{c}$ between AFM and iMS and $B'_\mathrm{c}$, between iMS and FM. (c) Integrated $R/R_0$ reflectance ratio (between 1364.1 and 1365.3 meV --- range for the exciton in AFM state) for 20 nm CrSBr for up $B_\mathrm{up}$ and down $B_\mathrm{down}$ sweep with iMS windows marked by the shaded regions. (d) Number of steps in the iMS region as a function of layer thickness for up- and down-sweep in semi-logarithmic representation. (e) Magnetic field range $\Delta B$ of the iMS phase for up- and down-sweep as a function of layer thickness in semi-logarithmic representation. Full symbols stem from reflectance and open from PL spectra. (f) Schematic dielectric profiles for alternating $\varepsilon_{\mathrm{AFM}}$ (AFM), mixed AFM/FM stack (iMS), and uniform $\varepsilon_{\mathrm{FM}}$ (FM).
• Figure 2: Comparison of experimental and simulated spectra for 25-layer CrSBr.(a) Experimental magneto-reflectance spectra $R/R_0$ for selected magnetic field values for the field applied along the CrSBr easy axis, where 0 T and 0.5 T correspond to AFM and FM order and all values in between to an iMS state; traces are offset for clarity. (b) Simulated magneto-reflectance spectra using the TMM based on layered magnetic domains for AFM, FM and representative iMS, with increasing number of FM layers (given in %). Simulations are purely based on $\varepsilon_\mathrm{AFM}(\omega)$, $\varepsilon_\mathrm{FM}(\omega)$ and layer thickness and not fitted to experimental data.
• Figure 3: Magneto-reflectance investigation for 7-layer CrSBr as the lower limit for layered-magnetic domain formation. Magneto-reflectance $R/R_0$ spectra for a 7L-CrSBr flake; (a) up-sweep, $B_\mathrm{up}$, (b) down-sweep, $B_\mathrm{down}$. Vertical dashed lines indicate critical switching fields. (c) Selected reflectance spectra for AFM, iMS and FM state for $B_\mathrm{up}$. The asterisk (*) marks an additional resonance that appears only within the intermediate-state window for $B_\mathrm{up}$. (d) Simulated magneto-reflectance spectra using the TMM for pure AFM, pure FM and iMS (4L FM, 1L AFM, 2L FM) configuration depicted at the right.
• Figure S1: Comparison between three different approaches to the construction of the dielectric function used in the calculations, along with the subsequent result of the model. In each column: the top row shows the dispersion of the refractive index $n(E)$; the middle row shows the energy of the excitonic resonance depending on the number of layers, all other than energy parameters of the excitonic resonance used to the construction of the dielectric function are the same. The bottom row shows the normalized reflectance $R/R_0$ for exemplary configurations of AFM/FM stacks with different FM content (given in %). First approach (a) is differing only the monolayer exciton energy (AFM state) and FM state (the number of layers $\geqslant 2$). Even with this simple model we can observe characteristics of the intermediate AFM/FM stack --- occurrence of more features in the spectrum and blueshifts. (b) The second approach with additional intermediate energy for thin layers (2--3L); The approach presented in (c) was used in the main paper.
• Figure S2: Left: calculated reflectance spectra for flakes of 7 ML, 25 ML, and 300 ML thickness. Spectra are shown for the pure AFM (top), FM (bottom) and intermediate states between. For 7 ML, the sequence corresponds to one AFM layer sandwiched between two ferromagnetic layers with thickness of 4 (top) and 2 (bottom). For 25 ML and 300 nm, representative mixed configurations are presented. The content of the FM layers is increasing from top (AFM) to bottom (FM) of the graphs.