Chromium Thiophosphate (CrPS4): Low-Symmetry Magnet

• Chromium thiophosphate (CrPS4) is a layered van der Waals magnetic semiconductor known for its distorted CrS6 octahedra and anisotropic A-type antiferromagnetic order.
• Its unique crystal structure with a rectangular Cr network drives quasi-one-dimensional in-plane ferromagnetic exchange and field-tunable spin reorientation phenomena.
• CrPS4 exhibits diverse optical and transport responses, including ligand-field excitations, linear dichroism, and electrically modulated tunneling magnetoresistance.

Chromium thiophosphate, $\mathrm{CrPS_4}$, is a layered van der Waals magnetic semiconductor in which $\mathrm{Cr^{3+}}$ ions in distorted sulfur octahedra form a low-symmetry rectangular network, producing strongly anisotropic magnetism, field-tunable antiferromagnetic order, localized $d$-orbital optical excitations, and multiple spin-dependent transport phenomena (Calder et al., 2020, Sternemann et al., 21 Nov 2025). Across bulk, few-layer, and heterostructure realizations, CrPS$_4$ has become a reference system for studying A-type antiferromagnetism, quasi-one-dimensional intralayer correlations, room-temperature linear dichroism, magnon transport, and electrically tunable tunneling anisotropic magnetoresistance in an air-stable 2D material (Huang et al., 2023, Cordero-Silis et al., 7 Apr 2026, Fu et al., 2024).

1. Crystal chemistry and structural symmetry

CrPS$_4$ is a ternary chromium thiophosphate composed of Cr, P, and S, and its crystal architecture is layered: the layers are strongly bonded in plane and stacked out of plane with weak van der Waals coupling, enabling mechanical exfoliation to few-layer and monolayer thicknesses (Lee et al., 2020). Structurally, the Cr ions reside in slightly distorted $\mathrm{CrS_6}$ octahedra, while P occupies tetrahedral coordination in $\mathrm{PS_4}$ units; this low-symmetry coordination is central to both the magnetic anisotropy and the optical anisotropy of the material (Kim et al., 2020, Cordero-Silis et al., 7 Apr 2026).

The crystallographic symmetry has been discussed in two closely related forms. Earlier structural work placed CrPS$_4$ in monoclinic $C2/m$, with representative room-temperature lattice parameters $a = 10.871\,\text{\AA}$, $\mathrm{Cr^{3+}}$0, $\mathrm{Cr^{3+}}$1, and $\mathrm{Cr^{3+}}$2 (Lee et al., 2020). Later neutron powder diffraction on the ordered low-temperature phase identified a non-centrosymmetric monoclinic $\mathrm{Cr^{3+}}$3 structure with $\mathrm{Cr^{3+}}$4, $\mathrm{Cr^{3+}}$5, $\mathrm{Cr^{3+}}$6, and $\mathrm{Cr^{3+}}$7 at 4 K (Calder et al., 2020). The literature therefore treats the $\mathrm{Cr^{3+}}$8 to $\mathrm{Cr^{3+}}$9 distinction as a genuine structural issue rather than a purely notational one; low-symmetry optical work explicitly emphasizes that earlier studies placed the compound in $d$0, whereas more recent x-ray studies suggest the lower $d$1 symmetry that naturally explains the pronounced in-plane anisotropy (Cordero-Silis et al., 7 Apr 2026).

Within each layer, CrPS$d$2 differs from the honeycomb geometry typical of many $d$3 magnets. The Cr ions form a rectangular 2D lattice, and the dominant magnetic couplings run along a single crystallographic direction, producing quasi-one-dimensional chains inside an otherwise quasi-two-dimensional layer (Calder et al., 2020). High-resolution structural studies further showed unequal Cr–Cr spacings along the basal-plane axes and preferential cleavage along diagonal Cr rows, providing direct crystallographic signatures of this anisotropy in exfoliated flakes (Lee et al., 2020).

2. Magnetic ground state and exchange hierarchy

The magnetic ion in CrPS$d$4 is nominally $d$5 with $d$6 and $d$7, and the low-temperature ordered moment is close to the spin-only expectation: neutron powder refinement gives $d$8 per Cr at 3–4 K (Calder et al., 2020). Neutron diffraction established that the bulk ground state is a layered antiferromagnet with propagation vector

$d$9

so the magnetic unit cell is doubled along $_4$0; spins are ferromagnetically aligned within each layer, antiferromagnetically stacked between adjacent layers, and lie in the $_4$1–$_4$2 plane with a dominant $_4$3-axis component at zero field (Calder et al., 2020). In the later few-layer literature this is described as A-type antiferromagnetism: each monolayer is ferromagnetic, while neighboring monolayers are oppositely magnetized (Huang et al., 2023).

The exchange network is unusually anisotropic. In a local-moment Hamiltonian,

$_4$4

the fitted inelastic-neutron-scattering parameters are $_4$5, $_4$6, $_4$7, $_4$8, and $_4$9, using the sign convention in which negative $_4$0 denotes ferromagnetic coupling (Calder et al., 2020). This immediately defines the hierarchy $_4$1: CrPS$_4$2 is quasi-two-dimensional because interlayer coupling is an order of magnitude weaker than the strongest in-plane couplings, and it is simultaneously quasi-one-dimensional because the strongest ferromagnetic exchange runs along chain-like bonds inside each layer (Calder et al., 2020, Baral et al., 15 May 2026). The single-ion anisotropy is correspondingly small, giving a fitted spin gap $_4$3 and nearly Heisenberg behavior with weak easy-axis anisotropy (Calder et al., 2020).

The modern neutron result superseded an earlier bulk interpretation based on magnetization, specific heat, and ESR that proposed a C-type antiferromagnetic ground state with in-plane antiferromagnetic coupling and ferromagnetic interlayer coupling (Pei et al., 2016). That earlier model was useful as a first phenomenology, but later neutron diffraction and spin-wave analysis fixed the bulk structure as ferromagnetic layers stacked antiferromagnetically and resolved the microscopic exchange constants directly (Calder et al., 2020).

3. Field response, spin reorientation, and reduced dimensionality

CrPS$_4$4 does not exhibit a single simple magnetic transition. In bulk neutron diffraction, additional magnetic Bragg peaks appear below about 40 K, but the intensities of different reflections evolve differently; the ordered state develops through a spin reorientation between 36 and 40 K without changing $_4$5, producing what was described as a “non-trivial magnetic transition” (Calder et al., 2020). The origin of that two-stage evolution remains open: the neutron study discussed either a dimensional crossover from 2D in-plane correlations to full 3D order or a weak structural effect that changes local anisotropy, while noting that no strong structural anomaly was resolved in the lattice constants (Calder et al., 2020).

The low anisotropy also makes the spin structure highly field-tunable. With field along the easy axis, a spin-flop occurs near $_4$6: the zero-field structure with spins largely along $_4$7 reorients into a state with spins predominantly along $_4$8, while the propagation vector remains $_4$9 (Calder et al., 2020). Few-layer and nanoscale probes observe the same basic metamagnetic sequence. In an 8 nm flake, dynamic cantilever magnetometry resolved three regimes for $\mathrm{CrS_6}$0: low-field A-AFM, an intermediate canted AFM region beyond a spin-flop near $\mathrm{CrS_6}$1 T at 5 K, and a field-polarized state above $\mathrm{CrS_6}$2 T (Li et al., 2024). For $\mathrm{CrS_6}$3, the same study found an M-shaped $\mathrm{CrS_6}$4 at low temperature, a hump developing near $\mathrm{CrS_6}$5 K, and a negative W-shaped response by $\mathrm{CrS_6}$6–$\mathrm{CrS_6}$7 K, which it interpreted as a continuous temperature-driven reorientation from out-of-plane toward in-plane anisotropy (Li et al., 2024).

Real-space neutron total-scattering has now sharpened that picture. Magnetic PDF analysis resolved the local spin direction and showed that the polar angle of the local moment rotates from $\mathrm{CrS_6}$8 away from $\mathrm{CrS_6}$9 at 2 K to $\mathrm{PS_4}$0 at 200 K, while ferromagnetic intrachain correlations persist far above $\mathrm{PS_4}$1 (Baral et al., 15 May 2026). Combining those correlations with a DFT-derived spin Hamiltonian showed that single-ion and exchange-anisotropy channels are renormalized differently as temperature increases: the single-ion terms remain local, whereas the exchange-anisotropy contribution scales with the decay of intersite correlations, rotating the effective easy axis (Baral et al., 15 May 2026). Above $\mathrm{PS_4}$2, the continued experimental rotation exceeds the dominant-chain model prediction, explicitly identifying the limits of that minimal description rather than invalidating the correlation-driven mechanism itself (Baral et al., 15 May 2026).

In the atomically thin limit, CrPS$\mathrm{PS_4}$3 preserves its layered antiferromagnetic character but acquires a pronounced even–odd effect. Wide-field NV magnetometry directly imaged strong stray field in trilayers and negligible stray field in bilayers, consistent with uncompensated net magnetization in odd-layer A-type stacks and compensation in even-layer stacks; the reconstructed trilayer magnetization at 5 K is spatially averaged at roughly $\mathrm{PS_4}$4, and the trilayer ordering temperature is reduced to $\mathrm{PS_4}$5 K (Huang et al., 2023). Because all intralayer couplings in bulk are ferromagnetic while the interlayer coupling is small and antiferromagnetic, this suggests that removing $\mathrm{PS_4}$6 in the monolayer limit favors ferromagnetic order within a single CrPS$\mathrm{PS_4}$7 layer (Calder et al., 2020).

4. Electronic structure and optical spectroscopy

CrPS$\mathrm{PS_4}$8 is a semiconductor, but the relevant optical energy scales depend strongly on the probe. Earlier few-layer optical work reported an optical bandgap near $\mathrm{PS_4}$9 eV and a dominant photoluminescence peak at $_4$0 eV, both associated with localized $_4$1 $_4$2–$_4$3 transitions (Kim et al., 2020). Differential-reflectance analysis in another study gave $_4$4 eV and identified two weak sub-gap absorption features at 1.54 and 1.65 eV (Gu et al., 2019). More recent optical anisotropy work therefore framed the bandgap as being on the order of $_4$5–2 eV while emphasizing that the 1.37–2.48 eV spectral window is dominated by intra-ionic $_4$6 transitions rather than simple band-edge absorption (Cordero-Silis et al., 7 Apr 2026). A plausible implication is that different experiments are emphasizing different parts of the same correlated ligand-field and charge-transfer landscape.

Momentum-resolved photoemission and DFT+$_4$7 have now supplied the missing band-structure framework. The valence band is dominated by Cr $_4$8 and S $_4$9 states, and the gap is of ligand-to-metal charge-transfer type: the valence-band maximum has strong S $C2/m$0 weight with significant Cr $C2/m$1 admixture, whereas the lowest unoccupied states are predominantly Cr $C2/m$2 (Sternemann et al., 21 Nov 2025). The $C2/m$3 manifold remains comparatively weakly hybridized and carries the local moments responsible for magnetic order, while the $C2/m$4 manifold hybridizes strongly with ligand states and forms bonding/antibonding structures that relax dipole selection rules (Sternemann et al., 21 Nov 2025). Soft-x-ray XMCD and RIXS reach a closely related conclusion from a different direction: CrPS$C2/m$5 is more covalent than a purely ionic $C2/m$6 picture, exhibits hybridization between crystal-field and charge-transfer excitations, and shows measurable spin polarization on both S and P ligands, consistent with extended superexchange paths of the form Cr–S–P–S–Cr (Buccoliero et al., 17 Feb 2026).

These orbital ingredients organize the optical spectra. In the 1.6–1.9 eV range, polarized reflectivity, photocurrent, and PLE identify two principal $C2/m$7 crystal-field transitions, $C2/m$8 near $C2/m$9–$a = 10.871\,\text{\AA}$0 eV and $a = 10.871\,\text{\AA}$1 near $a = 10.871\,\text{\AA}$2–$a = 10.871\,\text{\AA}$3 eV (Cordero-Silis et al., 7 Apr 2026). CrPS$a = 10.871\,\text{\AA}$4 also exhibits strong NIR photoluminescence. One line of work assigned room-temperature peaks at 1.32 and 1.43 eV to spin-allowed intra-$a = 10.871\,\text{\AA}$5-shell transitions of $a = 10.871\,\text{\AA}$6 in an octahedral-like ligand field (Gu et al., 2019). Another showed that the PL pathway bifurcates between fluorescence from $a = 10.871\,\text{\AA}$7 at about $a = 10.871\,\text{\AA}$8 eV and phosphorescence from $a = 10.871\,\text{\AA}$9 at about $\mathrm{Cr^{3+}}$00 eV, with the branching controlled by thickness, temperature, and defect density through activated reverse intersystem crossing (Kim et al., 2022). In that three-level Cr$\mathrm{Cr^{3+}}$01 model, thinner and more defective flakes shorten both biexponential decay times and increase the fast-component weight, consistent with defects lowering the barrier for reverse intersystem crossing (Kim et al., 2022).

Two additional spectroscopic phenomena are distinctive. First, below $\mathrm{Cr^{3+}}$02 K, photoluminescence develops a Fano-type asymmetric line shape caused by interference between a broad continuum of $\mathrm{Cr^{3+}}$03 intra-$\mathrm{Cr^{3+}}$04 transitions and a discrete impurity-related level attributed to excess phosphorus; the discrete channel becomes optically active only in the antiferromagnetic phase and is further enhanced by magnetic field (Gu et al., 2019). Second, CrPS$\mathrm{Cr^{3+}}$05 displays pronounced room-temperature in-plane optical anisotropy. In reflection, the linear dichroism reaches about $\mathrm{Cr^{3+}}$06 near $\mathrm{Cr^{3+}}$07 eV and about $\mathrm{Cr^{3+}}$08 near $\mathrm{Cr^{3+}}$09 eV, implying a sign reversal within a $\mathrm{Cr^{3+}}$10 meV window; in photocurrent, the linear dichroism reaches $\mathrm{Cr^{3+}}$11 near 1.63 eV at zero bias, and scanning photocurrent mapping shows a 3-fold enhancement along the $\mathrm{Cr^{3+}}$12-axis compared with the $\mathrm{Cr^{3+}}$13-axis (Cordero-Silis et al., 7 Apr 2026). These responses are explicitly tied to the low-symmetry monoclinic lattice and polarization-dependent coupling to the $\mathrm{Cr^{3+}}$14 and $\mathrm{Cr^{3+}}$15 $\mathrm{Cr^{3+}}$16 transitions (Cordero-Silis et al., 7 Apr 2026).

5. Transport, magnonics, and tunneling phenomena

Bulk transport establishes CrPS$\mathrm{Cr^{3+}}$17 as a semiconductor with thermally activated resistivity. The in-plane resistivity follows $\mathrm{Cr^{3+}}$18, with $\mathrm{Cr^{3+}}$19 and activation energy $\mathrm{Cr^{3+}}$20 eV in early single-crystal measurements (Pei et al., 2016). More recent device studies shifted attention from charge transport itself to spin and magnon transport through this insulating antiferromagnet.

A central development is non-local magnon transport. In unconventional Pt/CrPS$\mathrm{Cr^{3+}}$21 devices operating in the collinear state at in-plane fields $\mathrm{Cr^{3+}}$22 T and $\mathrm{Cr^{3+}}$23 K, the same contact can act as injector and gate or the gate can be placed outside the nominal transport channel, yet still modulate diffusive magnon transport of incoherent magnons (Wal et al., 2024). In this platform, electrically generated magnon signals are modulated by up to $\mathrm{Cr^{3+}}$24/mA and thermally generated signals by up to $\mathrm{Cr^{3+}}$25/mA, with the gate current altering the magnon chemical potential through the spin Hall effect and the magnon temperature through Joule heating (Wal et al., 2024). CrPS$\mathrm{Cr^{3+}}$26 therefore supplies a vdW antiferromagnetic realization of gate-tunable magnon transistors rather than merely a passive magnon medium.

Vertical tunneling devices reveal a complementary regime. In tetralayer graphite/CrPS$\mathrm{Cr^{3+}}$27/graphite spin-filter junctions, conductance is low in the low-field A-type antiferromagnetic state, jumps at the spin-flop near $\mathrm{Cr^{3+}}$28 kG, and saturates at higher field as the layers align ferromagnetically, yielding tunneling magnetoresistance of about $\mathrm{Cr^{3+}}$29 at 10 K and $\mathrm{Cr^{3+}}$30 mV (Huang et al., 2023). Few-layer CrPS$\mathrm{Cr^{3+}}$31 tunnel junctions go further: out-of-plane and in-plane field sweeps reveal layer-dependent and anomalous tunneling-magnetoresistance oscillations, especially in 4-layer devices, together with bias- and gate-controllable tunneling anisotropic magnetoresistance whose sign can reverse as bias or gate voltage is tuned (Fu et al., 2024). First-principles transport calculations in that work attribute the TAMR sign reversal primarily to the energy dependence of the difference between tunneling currents for out-of-plane and in-plane spin orientations rather than to a wholesale change of the magnetic easy axis (Fu et al., 2024).

CrPS$\mathrm{Cr^{3+}}$32 also functions as an interfacial antiferromagnet in exchange-bias heterostructures. In pristine CrPS$\mathrm{Cr^{3+}}$33/Fe$\mathrm{Cr^{3+}}$34GeTe$\mathrm{Cr^{3+}}$35 devices, anomalous Hall loops show exchange bias of about $\mathrm{Cr^{3+}}$36 mT at 5 K, vanishing above about 20 K and even reversing sign around 15 K; in oxidized CrPS$\mathrm{Cr^{3+}}$37/(O-FGT)/FGT heterostructures, the exchange bias becomes non-monotonic with temperature and persists to about 140 K because CrPS$\mathrm{Cr^{3+}}$38 interplays with ferrimagnetic Fe$\mathrm{Cr^{3+}}$39O$\mathrm{Cr^{3+}}$40 and FeO-like oxide layers (Balan et al., 2023). The same “fully uncompensated” character of a single CrPS$\mathrm{Cr^{3+}}$41 monolayer that makes it attractive for exchange bias is precisely the property that underlies odd-layer uncompensated magnetization in the few-layer NV measurements (Huang et al., 2023, Balan et al., 2023).

6. Defects, proximity effects, and spin-optical control

Because CrPS$\mathrm{Cr^{3+}}$42 combines a layered antiferromagnetic host with optically and electronically addressable excitations, it has become a useful platform for engineered proximity and impurity coupling. In bulk-CrPS$\mathrm{Cr^{3+}}$43/monolayer-WSe$\mathrm{Cr^{3+}}$44 heterostructures, type-II band alignment and spin-polarized charge transfer allow WSe$\mathrm{Cr^{3+}}$45 defect luminescence to act as an optical transducer of the CrPS$\mathrm{Cr^{3+}}$46 surface magnetic order; below $\mathrm{Cr^{3+}}$47 K, the heterostructure exhibits zero-field splitting and finite polarization contrast in the WSe$\mathrm{Cr^{3+}}$48 defect PL, enabling optical readout of the surface layer and, by extension, the Néel vector orientation (Shaikh et al., 2024). The same study reported that in the A-type AFM regime the circularly polarized CrPS$\mathrm{Cr^{3+}}$49 transition remains nearly constant with small field variations, whereas the WSe$\mathrm{Cr^{3+}}$50 response changes because the spin-polarized transfer channel is field-sensitive (Shaikh et al., 2024).

Rare-earth doping introduces an even sharper spin-optical probe. In Yb$\mathrm{Cr^{3+}}$51-doped CrPS$\mathrm{Cr^{3+}}$52, exchange coupling between isolated Yb$\mathrm{Cr^{3+}}$53 dopants and the Cr spin lattice produces large exchange splittings in narrow $\mathrm{Cr^{3+}}$54–$\mathrm{Cr^{3+}}$55 photoluminescence lines below $\mathrm{Cr^{3+}}$56, and those splittings track the field-driven spin-flop with exceptional sensitivity (Baillie et al., 15 Jul 2025). In that system the spin-flop at $\mathrm{Cr^{3+}}$57–1.0 T modulates the Yb$\mathrm{Cr^{3+}}$58 luminescence energies and exchange splittings strongly enough that an additional 405 nm optical beam can drive a reversible, optically induced spin-flop through photothermal heating of the CrPS$\mathrm{Cr^{3+}}$59 lattice and spin system (Baillie et al., 15 Jul 2025). This places CrPS$\mathrm{Cr^{3+}}$60 among the few 2D magnets where a metamagnetic phase boundary can be both read out and traversed optically.

Across these diverse experiments, two unresolved issues recur. One is crystallographic: the relationship between the $\mathrm{Cr^{3+}}$61 and $\mathrm{Cr^{3+}}$62 descriptions remains an important part of the low-symmetry phenomenology rather than a closed historical footnote (Calder et al., 2020, Cordero-Silis et al., 7 Apr 2026). The other is magnetic: the spin reorientation between 36 and 40 K, and more generally the temperature dependence of the easy axis above and below $\mathrm{Cr^{3+}}$63, is now microscopically constrained but not exhausted by minimal dominant-chain models (Calder et al., 2020, Baral et al., 15 May 2026). Even with those open questions, the accumulated evidence defines CrPS$\mathrm{Cr^{3+}}$64 rather precisely: it is a low-symmetry, quasi-one-dimensional-in-a-layer, A-type antiferromagnetic semiconductor whose ligand-field excitations, exchange hierarchy, and weak anisotropy make its spin structure unusually susceptible to dimensional reduction, field, gating, defects, and optical control.