CrSBr: 2D Magnetic Semiconductor

• CrSBr is an air-stable van der Waals layered magnetic semiconductor characterized by pronounced anisotropy and thickness-dependent magnetic and optical properties.
• Its orthorhombic crystal structure and quasi-1D features enable unique excitonic, transport, and magneto-optical behaviors that are tunable via strain, defects, and alloying.
• Magnetic order transitions from ferromagnetic in monolayers to antiferromagnetic in bulk drive notable changes in electronic band structures and transport, offering new routes for device control.

Chromium sulfur bromide (CrSBr) is an air-stable van der Waals layered magnetic semiconductor that crystallizes in an orthorhombic structure and combines pronounced structural, optical, spin, and lattice anisotropy with layer-dependent magnetism and strong excitonic effects (Telford et al., 2020, Antoniazzi et al., 29 Aug 2025). In bulk, it exhibits A-type antiferromagnetism, with ferromagnetic order within each layer and antiferromagnetic coupling between adjacent layers, whereas the monolayer is ferromagnetic and the bilayer is an antiferromagnet (Long et al., 2023, Antoniazzi et al., 29 Aug 2025). Because its band structure, optical resonances, magnetic order, lattice modes, and transport all respond strongly to thickness, magnetic field, polarization, strain, alloying, and defects, CrSBr occupies an unusual position among two-dimensional magnetic semiconductors (Adak et al., 6 Jul 2025, Biktagirov et al., 6 Jun 2025).

1. Crystal structure, symmetry, and anisotropy

CrSBr crystallizes in the orthorhombic space group Pmmn and consists of layers bound by van der Waals forces. Each layer comprises buckled CrS planes sandwiched between Br sheets, with the layers stacking along the $c$-axis (Telford et al., 2020). The bulk lattice is rectangular in plane, and the crystallographic $a$, $b$, and $c$ axes coincide with the principal optical axes (Lee et al., 2020, Piel et al., 11 Mar 2026). This low-symmetry structure underlies the recurrent description of CrSBr as a quasi-1D or highly anisotropic material in its optical, vibrational, and magnetic responses (Mondal et al., 2024, Quintela et al., 15 Oct 2025).

The optical anisotropy is tensorial rather than perturbative. In spectroscopic imaging ellipsometry, paramagnetic CrSBr is described by a diagonal biaxial dielectric tensor,

$$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$

with $\varepsilon_c(E)$ significantly smaller than the in-plane components and $\varepsilon_a(E)\neq\varepsilon_b(E)$ across the measured spectral range (Piel et al., 11 Mar 2026). The in-plane response is dominated by excitonic resonances polarized along specific crystal directions: an A-exciton centered around 1.3 eV is strictly polarized along the $b$-axis, whereas the B-exciton around 1.7 eV appears along both in-plane axes (Piel et al., 11 Mar 2026). This directional selectivity is consistent with the chain-like Cr–S arrangement parallel to $b$, and it provides a structural basis for the pronounced linear dichroism, birefringence, Raman selection rules, and anisotropic polariton transport reported for the material (Adak et al., 6 Jul 2025, Mondal et al., 2024).

2. Magnetic order, anisotropy, and phase behavior

High-quality bulk CrSBr shows only A-type antiferromagnetic order below the Néel temperature $T_N=131$ K, with no intrinsic low-temperature transition near 40 K in DC or AC magnetization; the low-temperature phase reported elsewhere was identified as extrinsic (Long et al., 2023). In a separate bulk study, the antiferromagnetic transition was observed at $a$0 K, again with intralayer ferromagnetism and antiferromagnetic interlayer coupling (Telford et al., 2020). Above $a$1, Curie–Weiss fits of the form

$a$2

yielded positive Weiss temperatures: 212.4 K, 209.8 K, and 210.6 K along the $a$3-, $a$4-, and $a$5-axes in one study, and $a$6 K, $a$7 K, and $a$8 K in another, establishing strong ferromagnetic correlations within individual layers well above the onset of three-dimensional antiferromagnetic order (Long et al., 2023, Telford et al., 2020).

Thickness changes the magnetic ground state. Second-harmonic generation established that monolayer CrSBr is ferromagnetic below $a$9 K, while the bilayer is antiferromagnetic with $b$0 K, and a 6L sample was reported at $b$1 K; in this thickness range the ordering temperature increases as the layer number decreases (Lee et al., 2020). For mono- to trilayers, the magnetic configurations were summarized as purely ferromagnetic in 1L, a perfect antiferromagnet in 2L, and a more complex order in 3L resembling a bilayer antiferromagnetic core with an additional ferromagnetic layer (Antoniazzi et al., 29 Aug 2025).

The magnetic anisotropy is triaxial. Experiment and theory identify the $b$2-axis as easy, the $b$3-axis as intermediate, and the $b$4-axis as hard (Telford et al., 2020, Yang et al., 2021). At 4.7 K, vector-magnet optical measurements yielded saturation fields $b$5 T along $b$6, $b$7 T along $b$8, and $b$9 T / $c$0 T along $c$1, with hysteresis along the easy axis (Krelle et al., 11 Mar 2025). First-principles calculations attributed the triaxial magnetic anisotropy to the joint effects of spin-orbit coupling and magnetic dipole-dipole interaction, and Monte Carlo simulations with the extracted monolayer parameters gave $c$2 K (Yang et al., 2021).

Neutron diffraction and inelastic neutron scattering refined the microscopic spin Hamiltonian. The magnon spectrum was fitted by an in-plane Heisenberg model with significant exchanges $c$3, $c$4, $c$5, $c$6, $c$7, $c$8, and $c$9 meV, no measurable magnon dispersion along the interplanar direction, and no observable gap at $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$0 to within 0.5 meV (Scheie et al., 2022). A plausible implication is that the dominant magnetic physics is strongly two-dimensional even in the bulk crystal.

3. Electronic structure, band splittings, transport, and native doping

Electronic spectroscopy has produced several energy scales for CrSBr, reflecting differences between optical transitions, excitons, transport activation, and quasiparticle gaps rather than a single universal number. Scanning tunneling spectroscopy reported an electronic gap $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$1 eV and room-temperature photoluminescence centered at $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$2 eV (Telford et al., 2020). In transport, the zero-field conductance followed

$$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$3

with $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$4 meV, while scanning tunneling spectroscopy gave $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$5 meV and Hall measurements found $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$6 at room temperature, indicating $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$7-type electron doping in as-grown material (Telford et al., 2020).

Low-temperature photoluminescence resolved intrinsic band splittings from monolayer to bulk. A conduction-band splitting of $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$8 meV was identified between CBM1 and CBM2, while the valence-band splitting was $$\boldsymbol{\varepsilon}(E)= \begin{pmatrix} \varepsilon_a(E)&0&0\ 0&\varepsilon_b(E)&0\ 0&0&\varepsilon_c(E) \end{pmatrix},$$9 eV between VBM1 and VBM2 (Lin et al., 2024). The lowest excitonic transition $\varepsilon_c(E)$0 appeared at $\varepsilon_c(E)$1 eV at 4 K, $\varepsilon_c(E)$2 at $\varepsilon_c(E)$3 eV, and higher-energy $\varepsilon_c(E)$4 transitions near $\varepsilon_c(E)$5 eV at 300 K; these splittings were reported to be nearly thickness-independent from monolayer to bulk (Lin et al., 2024). The $\varepsilon_c(E)$6 transition, dipole-forbidden by symmetry in the single-particle picture, was interpreted as brightened below $\varepsilon_c(E)$7 by magnon–exciton coupling (Lin et al., 2024).

Interlayer magnetic order substantially reconstructs the band structure. A DFT+U and optical spectroscopy study assigned the A-AFM state a quasi-direct gap with the valence-band maximum at $\varepsilon_c(E)$8 and the conduction-band minimum at U, with $\varepsilon_c(E)$9 meV between direct and indirect gaps (Linhart et al., 2023). When the interlayer order changes from A-AFM to FM, the calculations showed a direct-to-indirect band-gap transition, significant conduction-band splitting along $\varepsilon_a(E)\neq\varepsilon_b(E)$0-Z, and a gap reduction by $\varepsilon_a(E)\neq\varepsilon_b(E)$1 meV in the FM phase (Linhart et al., 2023). In a tuned hybrid-functional treatment, the fundamental gap of bulk CrSBr was reported as $\varepsilon_a(E)\neq\varepsilon_b(E)$2–$\varepsilon_a(E)\neq\varepsilon_b(E)$3 eV, while the lowest optical excitons were reproduced at $\varepsilon_a(E)\neq\varepsilon_b(E)$4 eV and $\varepsilon_a(E)\neq\varepsilon_b(E)$5 eV (Ramasubramaniam et al., 13 Mar 2026). This suggests a large exciton binding energy in the optical spectrum.

Transport is strongly coupled to magnetic order. Below $\varepsilon_a(E)\neq\varepsilon_b(E)$6, magnetotransport revealed negative magnetoresistance up to $\varepsilon_a(E)\neq\varepsilon_b(E)$7, with maxima at 30 K of $\varepsilon_a(E)\neq\varepsilon_b(E)$8 along $\varepsilon_a(E)\neq\varepsilon_b(E)$9, $b$0 along $b$1, and $b$2 along $b$3, and the saturation field was lowest along the easy axis (Telford et al., 2020). The mechanism was attributed to suppression of interlayer tunneling in the antiferromagnetic state and restoration of tunneling in the fully polarized state under field (Telford et al., 2020).

Native defects account for the unintentional $b$4-type character. Angle-resolved photoemission spectroscopy showed conduction-band filling in as-grown bulk crystals, and first-principles calculations identified chromium interstitials $b$5 stabilized between layers as the most favorable shallow donors (Biktagirov et al., 6 Jun 2025). Bromine-on-sulfur antisites $b$6 and bromine vacancies $b$7 were also found to act as donor defects, with $b$8 about 0.27 eV below the conduction-band minimum and $b$9 about 0.89 eV below it (Biktagirov et al., 6 Jun 2025). The measured energy separation between the valence-band maximum and the onset of the additional spectral weight attributed to conduction-band occupation was $b$0 eV (Biktagirov et al., 6 Jun 2025).

4. Excitons and magneto-optical coupling

CrSBr supports a distinctive coexistence of localized and delocalized excitonic species. In mono- to trilayers, photoluminescence, reflectance contrast, and photoluminescence excitation measurements identified low-lying A and A′ excitons that are predominantly Frenkel-like, together with higher-energy B excitons, and in bilayers an additional C exciton, that are more Wannier-Mott-like (Antoniazzi et al., 29 Aug 2025). In that layer-dependent scheme, A is a sharp line with linewidth of a few meV, A′ is a broad band with linewidth of tens of meV, and first-principles calculations cited in the study suggest binding energies of $b$1 eV for A and $b$2 eV for B (Antoniazzi et al., 29 Aug 2025). Monolayer and trilayer CrSBr display both A and A′ emission, whereas the bilayer shows only a single broad A′ feature (Antoniazzi et al., 29 Aug 2025). The relative intensity of A and A′ in 1L and 3L depends strongly on excitation energy, clarifying earlier reports of apparently missing A emission in monolayer spectra (Antoniazzi et al., 29 Aug 2025).

The excitonic field response depends sharply on thickness and magnetic state. For A and A′, the field-induced redshift up to the critical field was reported as $b$3 meV and described by

$b$4

In trilayers, the B exciton exhibited a giant redshift of $b$5 meV, whereas in monolayers its shift was $b$6 meV and disappeared above $b$7 T; in bilayers the C exciton redshifted by $b$8 meV (Antoniazzi et al., 29 Aug 2025). Monolayer and trilayer samples showed similar magneto-optical evolution, while the bilayer displayed a qualitatively different response associated with its zero net in-plane moment (Antoniazzi et al., 29 Aug 2025). Field-driven transition from AFM to FM reduced the oscillator strength and intensity of sharp excitonic features, a luminescence quench attributed to reduced exciton binding energy and larger exciton radius in the FM phase (Antoniazzi et al., 29 Aug 2025).

A complementary layer-resolved optical picture was obtained from magnetic correlation spectroscopy. In multilayer CrSBr, correlated photoluminescence and differential reflectance isolated a bright exciton $b$9 at $T_N=131$0 eV and a dark exciton $T_N=131$1 at $T_N=131$2 eV, with a splitting of 38 meV (Krelle et al., 11 Mar 2025). The dark-exciton emission and its phonon replicas $T_N=131$3, redshifted from $T_N=131$4 by 4.2, 7.8, and 11.8 meV, decreased monotonically with magnetic field, while $T_N=131$5 brightened close to the saturation field (Krelle et al., 11 Mar 2025). Both $T_N=131$6 and $T_N=131$7 redshifted in the ferromagnetic phase, by $T_N=131$8 meV and 14 meV respectively, and the stepwise changes in photoluminescence tracked layer-by-layer magnetization switching (Krelle et al., 11 Mar 2025). Transfer-matrix analysis further showed that ferromagnetic and antiferromagnetic order can coexist within the same crystal near saturation (Krelle et al., 11 Mar 2025).

These results connect directly to the broader band-structure studies. Optical markers of the interlayer magnetic phase transition were found in both the excitonic sector and the electronic bands, indicating that CrSBr is a system in which excitonic selection rules, oscillator strengths, and transition energies cannot be separated from the magnetic configuration (Linhart et al., 2023, Lin et al., 2024).

5. Vibrational spectroscopy, nonlinear optics, and local electronic structure

Raman spectroscopy in CrSBr is strongly resonant and highly anisotropic. Polarization-resolved resonant Raman scattering showed that the $T_N=131$9 mode can rotate its polarization by $a$00 as the excitation energy changes: it is maximally polarized along the $a$01-axis at 2.33 eV, along the $a$02-axis at 1.96 eV, nearly isotropic at 2.06 eV, and switches back to $a$03-axis polarization at 1.58 eV (Mondal et al., 2024). By contrast, the $a$04 and $a$05 modes remain polarized along $a$06 regardless of excitation energy or thickness (Mondal et al., 2024). The required Raman tensor analysis included both real and imaginary parts, indicating resonant electron–phonon coupling to anisotropic electronic states (Mondal et al., 2024).

Temperature-dependent Raman modes also act as markers of magnetic order. A combined Raman and DFT+U study resolved the $a$07 and $a$08 in-plane modes, with a splitting of about 5 cm$a$09 at low temperature that increased up to 15 cm$a$10 at 220 K, and identified this splitting as sensitive to the interlayer magnetic phase (Linhart et al., 2023). The same work reported experimental observation of the $a$11 mode for the first time (Linhart et al., 2023).

Resonance Raman scattering also revealed pronounced nonlinear behavior. Under excitation-dependent Raman measurements, all three Raman modes showed resonant enhancement near $a$12 eV and $a$13 eV, while the photoluminescence feature at 1.72 eV showed no corresponding absorption resonance and was therefore assigned to an indirect, phonon-assisted transition rather than a direct optical transition (Sahu et al., 3 Feb 2025). The anti-Stokes to Stokes intensity ratio reached up to 0.8, varied strongly with excitation energy, laser power, and crystal orientation, and exceeded values quoted in the same study for graphene ($a$14) and MoS$a$15 ($a$16) (Sahu et al., 3 Feb 2025). Above a threshold of 300 $a$17W, both Stokes and anti-Stokes signals became superlinear in power, and the extracted Raman gain reached $a$18 cm/GW (Sahu et al., 3 Feb 2025).

X-ray spectroscopies provide a local-orbital view of the same anisotropy. At the Cr $a$19 edges, XAS and RIXS showed strong linear dichroism arising from the distorted octahedral environment of the Cr$a$20 ions (Porée et al., 8 Jan 2025). A crystal-electric-field multiplet model with $a$21 symmetry used a $a$22–$a$23 splitting of 1.57 eV, an $a$24 splitting of 0.25 eV, and a $a$25 splitting of 0.1 eV to reproduce much of the dichroic spectral structure (Porée et al., 8 Jan 2025). X-ray excited optical luminescence showed a sharp peak near 1.35 eV at low temperature, about 0.2 eV below the lowest $a$26-$a$27 excitation seen in RIXS, which was taken to indicate that the lowest bright exciton is not a purely intra-atomic $a$28-$a$29 transition and instead has substantial charge-transfer or band-edge character (Porée et al., 8 Jan 2025).

A separate spectroscopic interpretation emphasized magnetic-polaronic states. That work identified a charge-transfer band edge at 500 nm and assigned emissions at 720, 850, 920, and 990 nm to, respectively, a single-ion magnetic polaronic exciton, a field-induced dark-state pair exciton, a trimeric in-layer excitonic state, and an interlayer tetrameric state (Shen et al., 2024). Because these assignments are more specific than the broader excitonic nomenclature used elsewhere, they are best regarded as one reported microscopic interpretation of the low-temperature, high-field photophysics of CrSBr rather than a settled universal classification.

6. Collective modes, magnetic polymorphism, and routes to control

The coupled magnetic and optical degrees of freedom in CrSBr extend beyond excitons to magnons, polaritons, and layer-selective magnetic polymorphs. In a phase-resolved magnetic second-harmonic generation study, magnetic polymorphism was formulated combinatorially as

$a$30

where $a$31 is the number of layers and $a$32 the total magnetization (Sun et al., 2024). Bilayer CrSBr was shown to host two antiferromagnetic polymorphs at $a$33, and tetralayer CrSBr six such polymorphs, resolvable through the phase of the SHG signal (Sun et al., 2024). In tetralayers adjacent to a bilayer, the switching route became deterministic through a magnetic layer-sharing effect, with the bilayer acting as a control bit for the tetralayer transition pathway (Sun et al., 2024). This layer-selective control links CrSBr to broader questions of multistate magnetic memory and symmetry-resolved magneto-optics.

CrSBr also supports guided and cavity polaritons shaped by its in-plane anisotropy. In sufficiently thick flakes, low-loss guided modes propagated tens of microns along the crystal $a$34-axis while remaining strongly confined along $a$35; for photon-like branches the propagation length reached up to 9 $a$36m, whereas more exciton-like branches propagated $a$37 $a$38m (Adak et al., 6 Jul 2025). Embedding CrSBr in a microcavity produced dispersion that was continuous along $a$39 but discretized along $a$40, directly evidencing one-dimensional confinement (Adak et al., 6 Jul 2025). Because the exciton energy redshifts under magnetic field, all polariton branches shifted as well, enabling magnetic control of propagation and confinement (Adak et al., 6 Jul 2025).

Chemical substitution and strain provide additional control parameters. In CrSBr$a$41Cl$a$42 alloys with $a$43, Raman and magneto-optical experiments showed retention of direct-gap semiconducting behavior and of the coupling between electronic and magnetic properties (Badola et al., 5 Jan 2026). The exciton energy increased weakly with composition, $a$44 meV per %Cl, from 1.365 eV for $a$45 to 1.372 eV for $a$46, while the interlayer exchange $a$47 decreased from 6.7 $a$48eV to 2.1 $a$49eV and the single-ion anisotropy $a$50 from 27.5 $a$51eV to 18 $a$52eV (Badola et al., 5 Jan 2026). The saturation fields $a$53 decreased by 70% and $a$54 by 40% at $a$55, implying substantial tuning of the magnetic gap and magnon energies (Badola et al., 5 Jan 2026). Theoretical work on strained monolayers further predicted that tensile strain parallel to the probed polarization redshifts the dominant excitonic peaks, compressive strain blueshifts them, and perpendicular strain primarily reorders spectral weight and lineshape in the diagonal optical conductivity tensor (Quintela et al., 15 Oct 2025).

Taken together, these results place CrSBr in a distinctive category of layered semiconductors: an orthorhombic, air-stable crystal whose magnetic order is microscopically weak between layers yet optically conspicuous, whose excitons display both Frenkel-like and Wannier-Mott-like character, and whose transport, Raman response, nonlinear optics, magnon spectrum, SHG phase, and polariton flow are all sensitive to the same anisotropic spin-resolved electronic structure (Antoniazzi et al., 29 Aug 2025, Scheie et al., 2022). A plausible implication is that CrSBr will remain a central platform for experiments in which magnetism is not merely coexistent with optical functionality, but the primary control parameter for it.