NiPS3: Layered 2D Antiferromagnetic Semiconductor

• NiPS3 is a layered van der Waals antiferromagnetic semiconductor with a planar honeycomb lattice of Ni ions and intrinsic zigzag spin order.
• It displays pronounced thickness-dependent behavior with strong exchange interactions, anisotropy, and tunable optical and transport properties.
• NiPS3 supports antiferromagnetic spintronics and device applications through its robust electrical readout, proximity effects, and scalable film formation.

NiPS$_3$ is a layered van der Waals antiferromagnetic semiconductor in the $M$PS$_3$ family that has become a central material for research on low-dimensional magnetism, correlated electronic structure, and spin-coupled optical excitations. Across the recent literature, it is consistently treated as a mechanically exfoliable honeycomb-lattice Ni compound with zigzag antiferromagnetic order in bulk, strong sensitivity to thickness, pressure, carrier density, and interlayer registry, and an unusual set of optical and transport signatures that remain active subjects of microscopic interpretation (Wildes et al., 2022, Tan et al., 2023, Cheon et al., 17 Apr 2026).

1. Crystal structure and magnetic ground state

At ambient conditions, NiPS$_3$ is reported as a monoclinic $C2/m$ layered compound in which Ni$^{2+}$ ions form a planar honeycomb lattice within the $ab$ planes, each Ni being coordinated by sulfur in an octahedral environment, while phosphorus dimers occupy the centers of the hexagons. The layers stack along the $c$ axis and are held together by van der Waals forces, which is the structural basis for exfoliation to few-layer thicknesses (Wildes et al., 2022).

Bulk NiPS$_3$ orders magnetically below a Néel temperature reported near $155$–$M$0 K, with a zigzag antiferromagnetic structure. Multiple studies describe the ordered moments as lying predominantly in the $M$1 plane and aligned close to the crystallographic $M$2-axis, which functions as the magnetic easy axis. Neutron-scattering results cited in optical-domain work describe a small out-of-plane canting of about $M$3, while an XMLD-PEEM study models the spins as canted by about $M$4 toward the $M$5-axis; these values are therefore best read as analysis-dependent reports rather than a single universally fixed number (Tan et al., 2023, Lee et al., 2024).

The magnetic pattern is not a nearest-neighbor Néel state. Instead, the zigzag order consists of ferromagnetic zigzag chains coupled antiferromagnetically to adjacent chains, and the interlayer coupling is reported as ferromagnetic in both neutron-based and optical-domain descriptions (Wildes et al., 2022, Tan et al., 2023). This combination of layered crystallography, in-plane easy-axis order, and survival of antiferromagnetism into the few-layer limit underlies the material’s importance for van der Waals antiferromagnetism and for antiferromagnetic spintronics (Cheon et al., 17 Apr 2026).

2. Exchange hierarchy, anisotropy, and spin dynamics

Single-crystal neutron spectroscopy establishes that NiPS$M$6 is governed by an exchange hierarchy extending to third-neighbor coupling within the plane plus a non-negligible interlayer term. The fitted parameters are

$M$7

with negative $M$8 denoting ferromagnetic exchange and positive $M$9 antiferromagnetic exchange in that convention. The dominant interaction is therefore the third-neighbor in-plane antiferromagnetic exchange $_3$0, while both $_3$1 and $_3$2 are ferromagnetic. On this basis, NiPS$_3$3 has been described as less two-dimensional than its sister compounds, because the interlayer exchange is large enough to help rationalize its comparatively high ordering temperature and the persistence of magnetic order down to bilayer thickness (Wildes et al., 2022).

The anisotropy is described as largely XY-like with a small uniaxial component. In the neutron analysis, this is encoded by

$_3$4

where the positive $_3$5 favors easy-plane behavior and the small negative $_3$6 weakly selects the in-plane easy axis. The same framework explains the two low-energy spin-wave modes at the magnetic zone center, since distinct in-plane and out-of-plane fluctuations are no longer degenerate (Wildes et al., 2022).

Optical magneto-spectroscopy further resolves the fundamental magnon gap into two components, measured at low temperature as

$_3$7

This splitting is used to argue for biaxial rather than purely uniaxial antiferromagnetism, with estimated parameters $_3$8, $_3$9, and $_3$0 (Jana et al., 2023). Ultralow-frequency Raman spectroscopy subsequently identified low-energy modes M1, M2, and M3 in bulk, with M1 at about $_3$1 and M3 at about $_3$2, and concluded that M1 and M3 are magnetic while M2 likely is not a single magnon. The same Raman work reports direct observation of the $_3$3 meV magnon scale down to the bilayer limit and uses polarization dependence to infer that M1 and M3 are dominated by different exchange-scattering paths (Na et al., 2024).

The resulting picture is that NiPS$_3$4 is a frustrated zigzag antiferromagnet with strong third-neighbor exchange, appreciable interlayer coupling, easy-plane anisotropy, and a weak in-plane pinning field. This suggests that neither an Ising description nor a strictly decoupled-layer limit is adequate for the full spin dynamics.

3. Thin-limit magnetism, domains, and vestigial order

Layer-dependent optical measurements show that antiferromagnetic order remains detectable from bulk down to bilayer NiPS$_3$5, but becomes strongly fluctuation-dominated as thickness decreases. Magneto-optical linear dichroism measurements report robust LD signals in $_3$6L–$_3$7L flakes, no detectable LD in monolayer NiPS$_3$8, and a systematic decrease of the transition temperature from $_3$9 K in $C2/m$0L to $C2/m$1 K in $C2/m$2L. In the same analysis, the critical exponent $C2/m$3 increases from $C2/m$4 in thick and $C2/m$5L samples to $C2/m$6 in $C2/m$7L, which is interpreted as evidence for stronger spin fluctuations in thinner flakes (Tan et al., 2023).

The same multimodal study identifies three distinct antiferromagnetic domains in atomically thin NiPS$C2/m$8, with Néel-vector orientations at $C2/m$9, $^{2+}$0 $^{2+}$1, and $^{2+}$2. Angular photoluminescence measurements agree with these assignments by showing PL polarization maxima shifted by $^{2+}$3 relative to the Néel vector. Thermally activated domain evolution is directly observed in a $^{2+}$4L–$^{2+}$5L sample between $^{2+}$6 K and $^{2+}$7 K, including domain flips such as a change from domain$^{2+}$8 to domain$^{2+}$9 or domain$ab$0, which the authors attribute to competition between local strain and thermal fluctuation (Tan et al., 2023).

A separate thickness-driven study argues that the few-layer regime is not merely a weakened form of the bulk zigzag antiferromagnet, but a distinct $ab$1 vestigial Potts-nematic phase. There, bulk NiPS$ab$2 is described as breaking both translational and rotational symmetry, whereas few-layer NiPS$ab$3 retains broken rotational symmetry while translational symmetry is restored. Experimentally, the zone-folded phonon near $ab$4, identified as a signature of broken translational symmetry, disappears below a critical thickness of $ab$5 nm, while the phonon splitting near $ab$6, identified with broken rotational symmetry, persists down to bilayer thickness. NV spin relaxometry and Raman quasi-elastic scattering in the same work show enhanced fluctuations as thickness is reduced, and Monte Carlo simulations for bilayer NiPS$ab$7 reproduce a $ab$8 Potts-nematic phase without long-range zigzag spin order (Sun et al., 2023).

Bulk-domain imaging adds a complementary perspective. XMLD-PEEM at the Ni $ab$9 edge resolves stripe-like antiferromagnetic domains about $c$0 nm wide with spatial resolution on the order of $c$1 nm, and follows their evolution from $c$2 K to $c$3 K. That study reports local Néel vectors parallel to the $c$4-axis in one domain class and rotated by about $c$5 or $c$6 in others, with clear thermal reorganization on cycling through the ordered phase. The authors attribute this behavior to exceptionally weak in-plane anisotropy arising from a small Ni orbital-moment anisotropy,

$c$7

corresponding to an in-plane magnetocrystalline anisotropy of about $c$8 meV/Ni (Lee et al., 2024).

Taken together, these results indicate that NiPS$c$9 supports several domain- and fluctuation-dominated regimes whose character depends strongly on thickness. A plausible implication is that “the” antiferromagnetic state of NiPS$_3$0 is better regarded as a family of closely related phases rather than a single rigid order parameter extending unchanged from bulk to monolayer.

4. Antiferromagnetic transport and electrical readout

NiPS$_3$1 is one of the clearest examples of electrical Néel-vector readout in an ultrathin van der Waals antiferromagnetic semiconductor. In transistor and tunnel-junction geometries, anisotropic magnetoresistance has been observed down to $_3$2 nm, corresponding to two layers, which the transport study emphasizes as the thinnest antiferromagnetic channel reported to show AMR. In the lateral FET geometry, the AMR magnitude is described as essentially thickness-independent down to the bilayer limit, and the same gate-dependent sign reversal appears in the $_3$3 nm device as in thicker flakes (Cheon et al., 17 Apr 2026).

The Néel vector is manipulated through a spin-flop transition of uniaxial antiferromagnetic NiPS$_3$4. For an in-plane magnetic field applied along the easy axis, the spins remain in the layer plane but rotate perpendicular to the easy axis above a spin-flop field of about $_3$5 T in the AMR work. The corresponding mean-field description is written as

$_3$6

where $_3$7 is the in-plane angle of the Néel vector relative to the magnetic field and $_3$8 is the initial angle between field and easy axis. By placing the field a few degrees away from the easy axis, nearly $_3$9 clockwise or counterclockwise rotation of the Néel vector can be forced over a sweep to $155$0 T, which allows distinct AMR contributions to be separated experimentally (Cheon et al., 17 Apr 2026).

The transport analysis resolves two AMR terms. At high carrier density, the dominant contribution is noncrystalline, depending on the angle between the current and the Néel vector; in a $155$1 nm FET, extracted amplitudes are reported as roughly $155$2 and $155$3 for one field alignment, and $155$4, $155$5 for the opposite alignment. At low charge density near threshold, the noncrystalline term fades, the magnetoresistance sign at $155$6 reverses, and the curves for opposite field orientations overlap, indicating dominance of the crystalline AMR term. The crossover occurs near carrier densities $155$7 for the positive noncrystalline regime, while low density near threshold favors positive crystalline AMR. Vertical tunnel-junction devices with $155$8–$155$9 nm NiPS$M$00 barriers independently isolate the crystalline term because the current–Néel-vector angle does not change during spin-flop in that geometry (Cheon et al., 17 Apr 2026).

This transport phenomenology gives NiPS$M$01 an unusual combination of attributes within antiferromagnetic spintronics: AMR that survives to the bilayer limit, gate-controlled switching between distinct microscopic AMR mechanisms, and compatibility with both lateral and vertical device architectures. The same material therefore supports both order-parameter readout and electrostatic control.

5. Electronic structure and correlated optical excitations

The electronic classification of NiPS$M$02 remains a major point of discussion. An earlier combination of spectroscopic ellipsometry, XAS, XPS, and DFT+$M$03 interpreted NiPS$M$04 as a self-doped negative charge-transfer insulator, with an optical gap of about $M$05 eV, strong optical transitions A, B, and C near about $M$06, $M$07, and $M$08 eV, weaker $M$09-$M$10 transitions $M$11 and $M$12 near $M$13 and $M$14 eV, and a marked redistribution of optical spectral weight at $M$15. In that description, the ground state is dominated by ligand-hole character, with

$M$16

and weights

$M$17

which was used to explain strong charge-spin coupling and the reduced ordered moment (Kim et al., 2017).

A later bulk-sensitive XAS and RIXS study reached the opposite conclusion, arguing that NiPS$M$18 is a Mott-Hubbard insulator rather than a charge-transfer insulator. That work attributes earlier small-$M$19 interpretations in part to surface oxidation affecting TEY XAS, reports that exfoliated bulk-sensitive spectra are best matched by charge-transfer multiplet calculations with $M$20 eV, and finds dominant $M$21 excitations at about $M$22 and $M$23 eV in RIXS, with charge-transfer features too weak to observe (Cao et al., 2024). The most cautious summary is therefore that the charge-transfer versus Mott-Hubbard classification remains historically contested, with later evidence emphasizing bulk sensitivity and surface preparation.

Momentum-resolved spectroscopy adds a further layer of complexity. $M$24-ARPES on an approximately $M$25-layer flake identifies a magnetically induced band shift of about $M$26 meV across $M$27, centered on a relatively flat band near $M$28 at $M$29 eV, and attributes it to a mixed Ni–S band related to superexchange involving Ni $M$30 orbitals. The same study also reports a weakly dispersive shoulder above the valence-band maximum around $M$31 eV that is not reproduced by DFT+$M$32, suggesting many-body interactions beyond that approximation and distinguishing NiPS$M$33 from MnPS$M$34 and FePS$M$35, where DFT+$M$36 was found more adequate (Pestka et al., 20 Jul 2025).

The most famous optical feature of NiPS$M$37 is the sharp line near $M$38–$M$39 eV, but its microscopic origin is debated. One RIXS study identifies a sharp feature near $M$40 eV as a Hund’s exciton: a triplet-to-singlet excitation formed primarily by on-site Hund’s exchange on the Ni $M$41 shell, with $M$42 eV and a measured dispersion of about $M$43 meV that tracks the double-magnon dispersion (He et al., 2024). A later optical and crystal-field study instead argues that the $M$44 eV photoluminescence is localized Ni-centered spin-flip luminescence, not a collective magnetic excitation or Zhang–Rice–type charge-transfer exciton, and models it as a singlet–triplet crystal-field transition of Ni$M$45 in an octahedral environment using Tanabe–Sugano and charge-transfer multiplet calculations (Schue et al., 16 Jun 2025). By contrast, work on solution-processed films interprets the narrow $M$46 eV emission and related transient-absorption signals as spin-entangled Zhang–Rice multiplet excitons with nanosecond-range lifetimes (Shcherbakov et al., 2023), while optical magneto-spectroscopy explicitly states that a coherent Zhang–Rice exciton assignment is doubtful because the $M$47 eV line splits in an in-plane magnetic field and follows spin-axis rotation (Jana et al., 2023).

The common ground across these otherwise divergent interpretations is narrower than sometimes assumed. All of the cited studies agree that the $M$48–$M$49 eV excitation is strongly coupled to magnetic order, is anomalously sharp, and is not well described as an ordinary band-edge exciton. The disagreement concerns whether its dominant character is Zhang–Rice-like, Hund’s-excitonic, or localized crystal-field spin-flip.

6. Tuning by pressure, doping, substitution, and molecular intercalation

Pressure tuning reveals a strong dependence on stress conditions. Under hydrostatic pressure, NiPS$M$50 undergoes a two-stage structural evolution,

$M$51

interpreted as layer-by-layer slip along the $M$52-axis. In that study, an insulator–metal transition occurs near $M$53 GPa, magnetism collapses during the pressure-induced 2D-to-3D crossover, the two-magnon Raman feature near $M$54 disappears by about $M$55 GPa, and the Néel temperature is suppressed to $M$56 K at the LP/HP-II boundary (Ma et al., 2020). A later quasi-uniaxial study, however, finds a sluggish IMT only at $M$57 GPa, with no superconductivity down to about $M$58 K, and concludes that the phase stability fields of NiPS$M$59 are highly strain dependent (Matsuoka et al., 2021). Pressure alone is therefore not a complete control parameter; the stress tensor matters.

Electron doping by organic-cation intercalation produces a nonmonotonic magnetic evolution. Electrochemical insertion of cations into the van der Waals gap changes NiPS$M$60 from AFM to ferrimagnetic at doping levels of $M$61–$M$62 electrons/cell and back to AFM at $M$63 electrons/cell, which is attributed to competition between Stoner exchange and super-exchange. THA-intercalated NiPS$M$64 shows ferrimagnetism with a Curie temperature around $M$65 K and an average net moment of about $M$66 per cell, while DFT calculations identify the Stoner criterion $M$67 as satisfied at intermediate doping (Mi et al., 2021).

Light Cr substitution offers another route to modifying the antiferromagnetic state. In $M$68 with $M$69, the Néel temperature drops from about $M$70 K in pristine NiPS$M$71 to about $M$72 K, the spin-flop field decreases from about $M$73 T to about $M$74 T, and a field-induced polarized regime emerges near $M$75 T for both in-plane and out-of-plane fields. The reported interpretation is that Cr strongly suppresses both the effective exchange scale and, especially, the magnetic anisotropy (Basnet et al., 2024).

Molecular intercalation with pyridine modifies the magnetic order differently. That work keeps the monoclinic structure but reports systematic changes in the lattice angle $M$76, reversible shifts in transition temperatures measured for $M$77 and $M$78, orientation-dependent pyridine dipoles inferred from polarized Raman spectroscopy, and recovery of the pristine magnetic response after deintercalation at $M$79 in vacuum within about $M$80 deviation. The proposed mechanism emphasizes charge transfer, stacking distortion, and a double-exchange or “double spin-exchange” picture rather than simple ferrimagnetic conversion (Chakraborty et al., 2024).

Across these perturbations, NiPS$M$81 emerges as a material whose magnetism is unusually sensitive to interlayer spacing, layer registry, charge injection, and anisotropy engineering. This suggests that control of stacking and local coordination is at least as important as control of carrier density alone.

7. Heterostructures, scalable films, and nonequilibrium functionality

NiPS$M$82 has also become a component material in hybrid platforms. In monolayer WSe$M$83/NiPS$M$84 heterostructures, nanoscale indentations create localized WSe$M$85 quantum emitters that interact through magnetic proximity with NiPS$M$86. The NiPS$M$87 layer in that work is approximately $M$88 nm thick and acts as an antiferromagnetic insulator whose defects or domain walls provide localized uncompensated out-of-plane magnetization. The resulting quantum emitters produce circularly polarized single photons in the $M$89–$M$90 nm range at zero external magnetic field, with degree of circular polarization up to $M$91 and single-photon purity up to $M$92. The proposed mechanism is exchange-mediated magnetic proximity rather than a simple dipolar stray field, because the estimated dipolar field of about $M$93–$M$94 mT is far too small to account for the observed chirality (Li et al., 2022).

At the level of scalable materials processing, liquid-phase exfoliation and modified Langmuir–Schaefer assembly produce centimeter-scale NiPS$M$95 thin films that retain key low-temperature optical signatures. In those films, the main photoluminescence feature appears at $M$96 eV with subpeaks at $M$97, $M$98, and $M$99 eV, each with FWHM $_3$00 meV, while transient absorption identifies related signals at $_3$01 and $_3$02 eV and relaxation extending into the nanosecond regime (Shcherbakov et al., 2023). Whatever the preferred microscopic assignment of these excitations, the processing result is notable because the narrow spin-coupled optical response survives a disordered, solution-processed morphology.

Ultrafast spectroscopy extends the same physics into the time domain. Pump–probe reflection measurements on NiPS$_3$03 report acoustic phonon oscillations near $_3$04 GHz, a fast decay component assigned to spin-orbit entangled exciton coherence with timescale $_3$05–$_3$06 ps, and a slower $_3$07–$_3$08 ns component assigned to spin reordering. The coherence time shortens from about $_3$09–$_3$10 ps below the exciton dissociation temperature $_3$11 K to about $_3$12 ps above $_3$13, while the slow component shows critical slowing down near $_3$14 K (Sahu et al., 5 Sep 2025). This places NiPS$_3$15 among the few van der Waals antiferromagnets in which excitonic coherence, spin fluctuations, and lattice motion have been resolved on distinct ultrafast timescales.

The broader technological implication is not a single device concept but a convergent materials profile: NiPS$_3$16 supports atomically thin AMR readout, thickness-dependent and strain-sensitive antiferromagnetic domains, scalable film formation, proximity-enabled quantum-light generation, and nonequilibrium optical access to spin-coupled excitations. A plausible implication is that its long-term value will lie less in any one interpretation of the $_3$17 eV excitation than in the fact that structure, transport, spin order, and optical response remain strongly co-tunable within one exfoliable antiferromagnetic platform.