Magnetic switching of exciton lifetime in CrSBr

TL;DR

This study demonstrates a magnetic-field–driven switch in CrSBr exciton lifetime, from $\tau_X \approx 11\ \mathrm{ps}$ in the AFM phase to $\approx 7\ \mathrm{ps}$ in the FM phase, linked to localization effects that enhance the exciton oscillator strength in the FM state. Time-resolved measurements show that the lifetime follows magnetic order and increases with temperature, suggesting disorder-induced localization as the controlling factor rather than nonradiative channels. Ab initio BSE/DFT+$U$+$J$ calculations reveal a redshift of the lowest exciton and a larger oscillator strength in the AFM phase, implying that localization broadens the radiative channel in the FM phase; this discrepancy is reconciled by considering localization volumes and thermal population of dark states. The combined experimental and theoretical framework highlights disorder-induced localization as a key mechanism shaping exciton dynamics in CrSBr and points to localization engineering as a route to control light–matter interactions in 2D magnetic semiconductors.

Abstract

Exciton dynamics in layered magnetic semiconductors provide a sensitive probe of the interplay between spin order and light-matter interaction. Here, we study thin CrSBr layers using time-resolved photoluminescence spectroscopy in an external magnetic field, revealing a step-like reduction in the exciton lifetime from 11 to 7 ps, during the magnetization flip from the antiferromagnetic to the ferromagnetic phase. The reduction of the exciton lifetime in the ferromagnetic phase persists below the Néel temperature, as evidenced by its strong magnetic-field dependence that disappears in the paramagnetic phase. Ab initio calculations reveal a one-dimensional nature of free excitons accompanied by a pronounced change in the oscillator strength across the magnetic phase transition predicting a shorter radiative lifetime of free excitons in the antiferromagnetic phase of CrSBr contradicting the experimental observations. This discrepancy is explained by strong localization of excitons at low tempature. We show both experimentally and theoretically that the observed magnetic switching of the exciton lifetime is attributed to a larger exciton localization volume leading to a larger oscillator strength in the ferromagnetic phase. The results show that disorder-induced localization effects play a key role in exciton dynamics in CrSBr.

Figures (8)

• Figure 1: (a) PL and reflectivity spectra in the AFM phase of a characteristic flake in zero magnetic field. Temperature $T = 2$ K. PL is excited with photon energy $E_{\rm exc} = 1.771$ eV. PL spectrum is normalized to the exciton peak (X). (b) PL spectra in the AFM phase from flakes of different thicknesses (as shown in c-e). The arrow marks the spectral position of 1s exciton. $T = 2$ K, $E_{\rm exc} = 1.771$ eV. PL intensity is normalized by the X peak intensity. (c), (d) and (e) Optical microscope images of flakes with 8, $\sim$36 and $\sim$70 nm thicknesses, correspondingly, which spectra are shown in (b). Thicknesses are measured using atomic force microscopy, see Supplementary Information Fig. S1. The thicknesses $\sim$36 and $\sim$70 nm are derived as mean values across the flakes. Directions of CrSBr $a$, $b$ and $c$ axes are shown by arrows in (e), the same directions are applicable to (c) and (d).
• Figure 2: 2D plots of PL signal as a function of magnetic field and photon energy measured at $T = 2$ K on a characteristic flake (thickness of $\sim$70 nm), $E_{\rm exc} = 1.771$ eV. (a), (b) Voigt geometry with $\textbf{B} \parallel a$ and $\textbf{B} \parallel b$, respectively. (c) Faraday geometry with $\textbf{B} \parallel c$. (d) Schematic representation of the Faraday experimental geometry.
• Figure 3: (a) TRPL transients in the AFM ($B =0$, blue) and FM ($B =0.5$ T, red) phases at low temperature ($T = 2$ K). The instrument response function (IRF) is shown by green. Symbols stand for experimental data and solid lines for fit with $I(t)=A\exp{(-\frac{4 \tau_X t - \sigma^2}{4\tau_X^2})}(1+{\rm erf}(\frac{2 \tau_X t-\sigma^2}{2 \tau_X \sigma}))$, representing rise and decay of the PL intensity, where $\tau_X$ is decay time, $A$ is a measure of amplitude of the signal, $\sigma = \rm{HWHM}/\ln 2 \approx 4.3$ ps is the measure of half-width half-maximum (HWHM) of the apparatus function, and ${\rm erf}(t)$ is the error function. (b) Magnetic field dependence of the exciton lifetime $\tau_{\rm X}$ in the AFM (green circles) and FM (red diamonds) phases. The switch of $\tau_{\rm X}$ from $11.1\pm0.3$ to $7.1\pm0.3$ ps takes place during the transition between the AFM and FM phase at $B=0.35$ T. (c) Magnetic field dependence of PL spectrum. Average flake thickness is $\sim$36 nm.
• Figure 4: Magnetic field dependence of PL spectra and the exciton lifetime $\tau_{\rm X}$ (green circles) for (a) low temperature $T=2$ K, (b) intermediate temperature $T<T_{\rm N}$ below Néel temperature, $T=96$ K, and (c) high temperature $T>T_{\rm N}$, $T=140$ K. The data are taken for the magnetic field directed along $c$-axis, in the Faraday geometry. The average flake thickness is $\sim$36 nm. Dashed white line shows transition to the FM phase. (d) Temperature dependence of the exciton lifetime. Green circles show the data at $B=0$ corresponding to the AFM phase. Red diamonds show the data measured in the range of $2-3$ T (larger than the saturation field $B_S$) corresponding to the FM phase. Solid lines are fits with Eq. \ref{['eq:tau_fit']} with parameters given in the text. Decay time in the limit of zero temperature is $\tau_{\rm X}^{\rm AFM}=11.7$ ps in AFM phase and $\tau_{\rm X}^{\rm FM}=7.5$ ps in FM phase. Black dashed line shows Néel temperature $T_{\rm N}=132$ K, so that $\tau_{\rm X}$ value above $T_{\rm N}$ (red and green triangles) correspond to paramagnetic phase (PM) disregarding the applied magnetic field. Blue dashed line shows the dependence for the free exciton lifetime $\tau_{\rm FX}$ in 1D case, described in the Discussion section.
• Figure 5: Ab initio simulations of excitons in CrSBr. (a) Absorption spectra for light linearly polarized along the $b$-axis, obtained from the Bethe--Salpeter equation (BSE) and independent particle approximation (IPA), are shown for the AFM and FM phases. (b) Momentum-space exciton composition of the lowest energy exciton eigenstate projected on the IPA band structures. The amplitude of the single-particle states (see SI3, Eq. (S2) for details) is shown by the black shadows under corresponding lines. (c) Dispersion curves for the lowest exciton, illustrating significant anisotropy in the transferred electron-hole momenta. The black line is meant to be a guide for the eye. Insets show the parabolas (solid lines) in the $\Gamma\rightarrow Y$-direction used to estimate the curvature of the exciton dispersion and mass. (d) Real-space electron probability density of the lowest excitonic state with the hole positioned near one of the sulfur atoms. Left panels display the in-plane profiles, where Cr–S chains extend along the $b$-axis. Right panels show the out-of-plane profiles.