CuCrP2S6 (CCPS): Multiferroic Photonic Material

• CuCrP2S6 (CCPS) is a layered van der Waals quaternary thiophosphate characterized by an ordered honeycomb-derived lattice and coupled dipolar and magnetic ordering.
• It undergoes distinct temperature-dependent phase transitions—from paraelectric to ferroelectric and antiferroelectric regimes—with evidence from Raman spectroscopy and neutron diffraction.
• Its excellent crystal growth and exfoliation enable integration in photonic, magneto-optic, and ion-driven neuromorphic devices, highlighting its versatile functionality.

CuCrP$_2$S$_6$ (CCPS) is a layered van der Waals quaternary thiophosphate in which Cu$^+$ and Cr$^{3+}$ occupy an ordered honeycomb-derived lattice within the broader metal thiophosphate family. It is studied as a multiferroic material because its Cu sublattice supports dipolar ordering, its Cr sublattice supports magnetic ordering, and both are coupled to lattice, optical, and ionic degrees of freedom. In bulk crystals, room-temperature and low-temperature phases differ substantially, with paraelectric, ferroelectric, quasi-antiferroelectric, antiferroelectric, and antiferromagnetic regimes reported as a function of temperature, while few-layer and electrically poled specimens show switchable out-of-plane ferroelectricity. CCPS has also become a platform for integrated photonics, magneto-optics, Raman studies of coupled phase transitions, and ion-driven resistive switching (Selter et al., 2021, Lai et al., 2018, Dushaq et al., 2 Sep 2025).

1. Crystal chemistry, growth, and exfoliation

Bulk CCPS single crystals have been grown by chemical vapor transport using iodine as transport agent, with stoichiometric Cu:Cr:P:S starting materials in a 1:1:2:6 ratio. The reported optimized growth profile used a hot zone at $750^\circ\mathrm{C}$ and a cold zone at $700^\circ\mathrm{C}$ for 100 hours, following ampule preparation under Ar atmosphere and evacuation to $10^{-8}$ bar. Compositional analysis gave a mean composition Cu$_{1.05(3)}$Cr$_{1.04(2)}$P$_{2.03(1)}$S$_6$0, and no detectable iodine contamination was found in bulk or at the surface (Selter et al., 2021).

At room temperature, CCPS crystallizes in the monoclinic space group $_6$1 with refined lattice parameters $_6$2, $_6$3, $_6$4, and $_6$5. The Cu atoms are strongly displaced from the plane and away from octahedral centers, adopting near trigonal planar coordination and generating local electric dipoles, whereas Cr atoms remain near octahedral centers. This cation differentiation distinguishes CCPS from the random site occupation common in the $_6$6P$_6$7S$_6$8 family and underlies its antipolar and multiferroic behavior (Selter et al., 2021).

The material forms shiny, plate-like layered crystals with hexagonal habitus and typical dimensions up to $_6$9. Mechanical exfoliation by the Scotch tape method yields flakes down to 2.8 nm, corresponding to 4 layers, with a monolayer thickness of about 0.7 nm. Flakes from 50 nm down to a few layers were reproducibly obtained, and thin flakes were reported to remain stable in air for several weeks. This combination of crystallographic order, air stability, and exfoliability is central to the later literature on photonic, spintronic, and ferroic-device integration (Selter et al., 2021).

2. Ferroic phases and multiferroicity across thickness and temperature

Bulk CCPS shows a temperature-dependent phase sequence in which structural and dipolar order evolve before magnetic order appears at lower temperature. Raman-based phase analysis reported the following sequence (Tang et al., 15 Apr 2025):

Temperature range	Reported phase	Reported order
$^+$0 K	Paraelectric	none
$^+$1 K	Ferroelectric	FE
$^+$2 K	Quasi-antiferroelectric	partially antipolar
$^+$3 K	Antiferroelectric	AFE
$^+$4 K	AFE + AFM regime	multiferroic below $^+$5 K

Specific-heat measurements on bulk crystals independently detected broad anomalies between 140 and 220 K, associated with structural changes accompanying antiferroelectric ordering, including a paraelectric-to-intermediate transition around 190 K and ordering into an antipolar state below 140–145 K. These measurements also yielded a structural entropy anomaly of $^+$6 for 120–220 K integration (Selter et al., 2021).

In reduced thickness, the reported phenomenology is different. First-principles calculations for monolayer CCPS found that the antiferroelectric phase is lower in energy than the ferroelectric phase by $^+$7, while the switching barrier from FE to AFE is about $^+$8 and a larger field-driven barrier of about $^+$9 was discussed for nanoscale switching. The same calculations predicted intrinsic ferromagnetism in both FE and AFE monolayers, with the ferromagnetic state favored over the antiferromagnetic state by about 13 meV/f.u., and a total moment of about $^{3+}$0 (Lai et al., 2018).

Few-layer experiments then reported room-temperature out-of-plane ferroelectricity in exfoliated CCPS nanosheets. Piezo force microscopy on approximately 13.3 nm thick flakes showed square phase-voltage hysteresis, amplitude-voltage butterfly loops, and polarization writing with non-destructive reading. Magnetic hysteresis at 10 K was observed in massively stacked nanosheets, consistent with ferromagnetic behavior in the few-layer limit. A later spin-dynamics study also stated that room-temperature ferroelectricity, including switchable domains, was confirmed down to 48 nm thickness (Lai et al., 2018, Basnet et al., 23 Feb 2026).

The theoretical work further proposed a coupled spin–electric-dipole–valley structure in the ferroelectric state, expressed as

$^{3+}$1

with a valley Zeeman splitting of about 20 meV in the FE phase with SOC. This places CCPS in a category of multiferroics where dipole, spin, and valley physics may be intertwined, although direct experimental valley readout was not reported in the supplied studies (Lai et al., 2018).

3. Magnetic ground state, anisotropy, and field-driven dynamics

Magnetically, CCPS is a quasi-2D A-type antiferromagnet: Cr$^{3+}$2 moments are aligned ferromagnetically within the layers and antiferromagnetically between layers. Bulk magnetization and specific heat place the Néel temperature at about $^{3+}$3 K or $^{3+}$4 K, while Curie–Weiss fits give positive $^{3+}$5 values of 33–35 K, indicating dominant ferromagnetic intralayer coupling and weaker antiferromagnetic interlayer exchange. Effective moments of $^{3+}$6 for $^{3+}$7 and $^{3+}$8 for $^{3+}$9 are close to the spin-only value expected for $750^\circ\mathrm{C}$0 Cr$750^\circ\mathrm{C}$1, and saturation magnetization is about $750^\circ\mathrm{C}$2 per formula unit (Selter et al., 2021).

Single-crystal neutron diffraction resolved a long-standing ambiguity about the ordered moment direction. The ordered state was identified as A-type AFM with moments aligned along the crystallographic $750^\circ\mathrm{C}$3 axis and magnetic space group $750^\circ\mathrm{C}$4. The refined ordered moment was $750^\circ\mathrm{C}$5 per Cr$750^\circ\mathrm{C}$6, again consistent with spin-only Cr$750^\circ\mathrm{C}$7 (Guo et al., 26 Mar 2026).

Broadband and sub-THz magnetic resonance studies explain this behavior in terms of Cr$750^\circ\mathrm{C}$8 ions under an octahedral crystal field. Because the $750^\circ\mathrm{C}$9 configuration yields a half-filled $700^\circ\mathrm{C}$0 manifold, the orbital moment is quenched, SOC is weak, and single-ion anisotropy is very small. The resulting dynamics are governed mainly by isotropic Cr–Cr exchange, with weak anisotropy and a low-field spin-flop transition. The standard relation used in the literature is

$700^\circ\mathrm{C}$1

and a spin-flop field of about 0.44 T was reported for in-plane field at low temperature. Exchange fields are much larger than anisotropy fields; for example, $700^\circ\mathrm{C}$2 T at 1.2 K was extracted from AFMR fits in one study, whereas neutron-based analysis gave $700^\circ\mathrm{C}$3 T and $700^\circ\mathrm{C}$4 T at 2 K from the static field response model (Basnet et al., 23 Feb 2026, Guo et al., 26 Mar 2026).

The field evolution is anisotropic within the plane. For field along $700^\circ\mathrm{C}$5, the system undergoes a spin-flop transition and then approaches a fully polarized state. For field along $700^\circ\mathrm{C}$6, the moments rotate smoothly toward the field, and the AFM Bragg peak disappears above a critical field, giving way to a field-polarized ferromagnetic state. Thermodynamic work reported suppression of the AFM phase at about 6 T for field parallel to the $700^\circ\mathrm{C}$7 plane and about 8 T for field perpendicular to it, while sub-THz resonance measurements detected a field-induced FM-like polarization above about 6.4 T (Selter et al., 2021, Basnet et al., 23 Feb 2026).

Two additional aspects are notable. First, AFMR persists far above $700^\circ\mathrm{C}$8, with short-range AFM correlations extending to about $700^\circ\mathrm{C}$9, emphasizing the quasi-2D character of the magnetism. Second, the Gilbert damping parameter was reported as $10^{-8}$0–$10^{-8}$1, attributed to weak SOC, orbital quenching, and insulating character (Basnet et al., 23 Feb 2026).

4. Raman fingerprints, soft modes, and magnetoelastic coupling

Raman spectroscopy has been used to track both ordinary lattice anharmonicity and the ferroic phase sequence. High-frequency modes assigned to the $10^{-8}$2 framework occur near 380.6 cm$10^{-8}$3 and 599.7 cm$10^{-8}$4 in one assignment scheme, while Cu/Cr-involving modes appear near 202.6 cm$10^{-8}$5 and 266.1 cm$10^{-8}$6. A separate temperature-dependent Raman study extrapolated the P–P and P–S stretching modes to 384.5 cm$10^{-8}$7 and 610.1 cm$10^{-8}$8 at 0 K, respectively; the small numerical differences reflect different analysis protocols and labeling conventions in the supplied literature (Tang et al., 15 Apr 2025, Rao et al., 2021).

For the P–P and P–S stretching modes, first-order temperature coefficients between 200 and 400 K were reported as $10^{-8}$9 cm$_{1.05(3)}$0K$_{1.05(3)}$1 and $_{1.05(3)}$2 cm$_{1.05(3)}$3K$_{1.05(3)}$4, respectively. The non-linear evolution of both mode frequency and linewidth was modeled with three-phonon and four-phonon anharmonic terms, and the extracted coefficients satisfied $_{1.05(3)}$5, indicating that three-phonon decay dominates in CCPS. Room-temperature linewidths then imply phonon lifetimes of about 0.8 ps for the P–P mode and about 0.25 ps for the P–S mode (Rao et al., 2021).

Low-frequency Raman modes are more directly tied to ferroic transitions. Two soft modes at 36.1 cm$_{1.05(3)}$6 and 70.5 cm$_{1.05(3)}$7 appear only below 145 K and disappear at the antiferroelectric-to-quasi-antiferroelectric transition, identifying Cu-dominated lattice instabilities associated with antipolar ordering. Their temperature dependence was fitted by $_{1.05(3)}$8 with $_{1.05(3)}$9–0.46, consistent with soft-mode critical behavior near a second-order transition in the supplied analysis (Tang et al., 15 Apr 2025).

Below about 65 K, several Raman modes harden when temperature increases from 4 to 65 K, contrary to the usual high-temperature softening trend. This was interpreted as evidence for negative thermal expansion and strong magnetoelastic coupling in the low-temperature regime. Independent neutron diffraction later directly identified a magnetoelastic coupling between interlayer spacing and magnetic order, with the interlayer distance contracting below $_{1.04(2)}$0 and expanding under field-induced canting; the Raman and neutron results are therefore mutually consistent at the level of observed spin-lattice coupling (Tang et al., 15 Apr 2025, Guo et al., 26 Mar 2026).

A later Raman study added a kinetic perspective. It reported pronounced thermal hysteresis in phonon self-energy parameters and in the dynamic Raman susceptibility, confirming a first-order antipolar transition near $_{1.04(2)}$1 K and a second-order transition near $_{1.04(2)}$2 K. Low-frequency mode $_{1.04(2)}$3 near 37 cm$_{1.04(2)}$4 showed non-monotonic temperature dependence and intensity enhancement near 60 K, while quasielastic central-peak response coexisted with incomplete softening of the low-frequency modes. The supplied interpretation was a crossover from displacive to order-disorder transition mechanisms, together with spin-phonon coupling below $_{1.04(2)}$5 K (Auti et al., 24 Dec 2025).

5. Optical, electro-optic, and magneto-optic functionality

CCPS is transparent in the near- and short-wave infrared and has a relatively high refractive index, with one photonics study describing anisotropic indices exceeding 3.1. Combined with exfoliable flakes and dry-transfer compatibility, this has enabled CCPS integration with both silicon and silicon nitride photonic platforms (Dushaq et al., 2023, Dushaq et al., 2024, Dushaq et al., 2 Sep 2025).

Hybrid platform	Function	Reported metrics
Si microring resonator + CCPS	Electro-refractive tuning	$_{1.04(2)}$6 up to $_{1.04(2)}$7 RIU; 2.7 dB/cm loss; $_{1.04(2)}$8 V·cm
Si microring resonator + CCPS + magnetic field	Non-reciprocal response	28 dB isolation; 0.15 dB insertion loss; 0.4 nm splitting; 50 GHz bandwidth
SiN microring/straight waveguide + CCPS	TM-pass filtering and rotation	PER $_{1.04(2)}$9 dB; 0.2–0.4 dB insertion loss; azimuth rotation up to $_{2.03(1)}$0

In silicon microring resonators, electrically driven Cu-ion migration in CCPS was used to tune the effective refractive index without measurable broadening of resonance linewidth or degradation of extinction ratio. The reported change in effective index reached $_{2.03(1)}$1 RIU, optical loss was 2.7 dB/cm, and the modulation efficiency was $_{2.03(1)}$2 V·cm. The resonance shift was consistently blue, and the response was strongly polarization dependent: TE modes showed a pronounced shift, whereas TM modes showed negligible tuning. The extracted phase response was based on the standard resonance-shift relation

$_{2.03(1)}$3

with the supplied interpretation that Cu-ion redistribution changes the real part of the dielectric function while leaving absorption largely unaffected (Dushaq et al., 2023).

Under magnetic field, silicon micro-ring resonators loaded with multilayer CCPS exhibited non-reciprocal TE-mode response at 1550 nm. Reported performance included up to 0.4 nm resonance wavelength splitting between clockwise and counterclockwise propagation, a 28 dB isolation ratio, 0.15 dB insertion loss, and a 50 GHz optical bandwidth. Interaction lengths were 22–55 $_{2.03(1)}$4m for flake thicknesses of 39–62 nm, and the response was flat across a 100 nm C-band window. The supplied mechanism was a magnetic-field-induced divergence in effective refractive index for the two propagation directions in the CCPS-loaded ring (Dushaq et al., 2024).

In silicon nitride photonics, CCPS has been used for polarization-selective filtering and rotation. Microring resonators incorporating 30–90 nm CCPS flakes on 400 nm-thick SiN waveguides showed TM-pass filtering with polarization extinction ratio exceeding 25 dB and insertion loss of about 0.2–0.4 dB across 1500–1600 nm. Straight SiN waveguides loaded with 120–300 $_{2.03(1)}$5m CCPS flakes produced azimuth angle shifts up to $_{2.03(1)}$6 for TM input and $_{2.03(1)}$7 for TE input, with ellipticity largely preserved. The reported rotation model used the accumulated phase difference

$_{2.03(1)}$8

and the experimental conclusion was that polarization-dependent optical mode overlap, set mainly by the refractive-index profile of the CCPS/SiN hybrid and by waveguide geometry, is the primary mechanism; the anisotropy of CCPS enhances the effect but is secondary rather than principal (Dushaq et al., 2 Sep 2025).

6. Ionic transport, redox switching, and neuromorphic relevance

Beyond ferroic and photonic behavior, CCPS is a mobile-ion solid in which Cu$_{2.03(1)}$9 transport can dominate electrical response. A 2026 study isolated resistive switching in paraelectric CCPS and concluded that, under the studied conditions, the switching is purely ion-driven rather than ferroelectric in origin. Exhaustive PFM and switching-spectroscopy PFM showed no measurable ferroelectric domains or phase hysteresis loops, while conductance modulation was reproducible and strongly dependent on voltage amplitude and sweep rate (Lancaster et al., 24 Jun 2026).

The central mechanism is electric-field-driven migration of native Cu$_6$00 ions, coupled to interfacial solid-state redox activity described by

$_6$01

Threshold fields around 0.045–0.06 V/nm were reported for significant ion migration. Slower sweeps or longer pulses produced larger and less volatile conductance changes, whereas faster sweeps produced more volatile states. The decay of the programmed current was fitted by a stretched exponential,

$_6$02

with $_6$03 around 0.5–0.6, indicating a broad distribution of relaxation times and disorder-governed ionic relaxation (Lancaster et al., 24 Jun 2026).

Cyclic-voltammetry-like measurements showed $_6$04, consistent with the Randles–Ševčík law and diffusion-limited redox currents. From this analysis, the Cu$_6$05 diffusion coefficient was extracted as

$_6$06

close to an independent estimate of about $_6$07 from ionic conductivity. Both inert and reactive electrodes supported the switching, implying that native Cu$_6$08 ions, rather than electrode-derived species, drive the process. In a soft-breakdown regime, volatile filamentary conduction attributed to metallic Cu clusters could be formed and erased, whereas excessive cycling caused irreversible structural change, including S$_6$09 formation detected by Raman spectroscopy (Lancaster et al., 24 Jun 2026).

This electrical literature materially clarifies a common ambiguity in CCPS research: mobile Cu ions are not only relevant to ferroelectric or antiferroelectric ordering, but can also dominate transport even in the paraelectric state. A plausible implication is that the same Cu-ion degree of freedom underlies several otherwise separate CCPS research directions, including electro-refractive photonic tuning, resistive switching, and thermally hysteretic structural dynamics (Lancaster et al., 24 Jun 2026, Dushaq et al., 2023).