1T-TaS2: CDW, Mott, and Hidden Phases

• 1T-TaS2 is a layered transition-metal dichalcogenide characterized by a Star-of-David commensurate CDW that orchestrates its insulating, metallic, and metastable states.
• Its electronic behavior arises from the interplay of lattice reconstruction, Mott splitting, band folding, and interlayer stacking, leading to a variety of correlated phases.
• Extensive studies show that dimensionality, defect doping, and external perturbations unlock hidden states and tunable transport properties in this complex material.

1T-TaS2 is a layered transition-metal dichalcogenide in the 1T polytype, built from van der Waals–coupled S–Ta–S trilayers in which Ta is octahedrally coordinated by S. Its defining low-temperature state is a commensurate charge-density wave (C-CDW) with a $\sqrt{13}\times\sqrt{13}\,R13.9^\circ$ “Star-of-David” superlattice, within which lattice reconstruction, band folding, on-site Coulomb repulsion, and interlayer stacking jointly determine whether the material is insulating, metallic, or metastably transformed. Across bulk, few-layer, and monolayer limits, 1T-TaS2 has become a central system for studying the coexistence of CDW order and Mottness, the dimensional crossover of correlated states, hidden and metastable phases, and possible quantum-spin-liquid-like behavior (Cho et al., 2015, Chen et al., 3 Mar 2026).

1. Crystal structure and charge-density-wave hierarchy

Bulk 1T-TaS2 exhibits a sequence of temperature-driven CDW phases. The high-temperature state is metallic; below approximately $550\ \mathrm{K}$ an incommensurate CDW appears, below about $350\ \mathrm{K}$ a nearly commensurate CDW (NC-CDW or NCCDW) develops, and below about $180\ \mathrm{K}$ the system enters the commensurate C-CDW phase. In heating protocols, the C phase can transform to a triclinic phase near $223\ \mathrm{K}$ and then to the NC phase near $283\ \mathrm{K}$, reflecting pronounced hysteresis and the first-order character of the low-temperature transition (Chen et al., 3 Mar 2026, Mahajan et al., 2018).

The C-CDW consists of 13-Ta clusters arranged as “Star-of-David” units. Each cluster contains one central Ta, a first ring of six Ta displaced inward, and a second ring of six Ta displaced less strongly. In real space the superlattice is observed as a $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$ pattern rotated by $13.9^\circ$ relative to the $1\times1$ lattice, with periodicity $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$ or $550\ \mathrm{K}$0 depending on experimental description. In diffraction- or Fourier-space language, the commensurate modulation corresponds to well-defined superlattice peaks and higher harmonics characteristic of a coherent periodic lattice distortion (Cho et al., 2015, Hovden et al., 2016).

The NC phase is structurally distinct. It consists of commensurate domains separated by discommensurations or domain walls, and in thin exfoliated flakes its periodic lattice distortions can coexist with stacking boundaries and increased disorder. Atomic-resolution STEM has shown that NC modulations persist both inside stacking domains and at their boundaries, while the low-temperature commensurate superstructure can remain visible in projection over regions as thick as roughly 65 layers, implying at least partial out-of-plane CDW order (Hovden et al., 2016).

Two further structural aspects have become important in recent work. First, the in-plane Star-of-David lattice can form chiral triangular superlattices with two degenerate handed domains and a ferro-rotational axial-vector order parameter. Second, interlayer registry is not ancillary: stacking order directly modulates out-of-plane hybridization and therefore the electronic bandwidth and gap. This makes 1T-TaS2 unusual among layered CDW compounds in that intralayer order and interlayer order must be treated together rather than hierarchically (Chen et al., 3 Mar 2026, Liu et al., 2024).

2. Electronic structure: band folding, CDW gaps, and the Mott state

The low-temperature insulating state of 1T-TaS2 is set by a two-stage reconstruction. The C-CDW potential and accompanying lattice distortion narrow the original metallic Ta $550\ \mathrm{K}$1 band into a manifold of subbands, leaving a single narrow half-filled cluster band at the Fermi level. Strong on-site Coulomb repulsion then splits this half-filled band into lower and upper Hubbard states, opening a Mott gap. A minimal description is therefore a cluster Hubbard model,

$550\ \mathrm{K}$2

with the crucial caveat that both $550\ \mathrm{K}$3 and the low-energy basis are themselves generated by the CDW reconstruction (Cho et al., 2015, Chen et al., 3 Mar 2026).

Spatially resolved STS at $550\ \mathrm{K}$4 resolved a “multiply formed” gap structure in pristine C-CDW regions. The central Mott gap is $550\ \mathrm{K}$5, bounded by a lower Hubbard state at $550\ \mathrm{K}$6 and an upper Hubbard state at $550\ \mathrm{K}$7. Additional CDW-derived gaps segment the occupied and unoccupied spectrum into several subbands, with a characteristic CDW gap of $550\ \mathrm{K}$8. Real-space phase analysis distinguishes the two origins: crossing a CDW gap produces a phase flip of the LDOS modulation, whereas crossing the Mott gap does not. This directly separates band gaps opened by periodic lattice distortion from the correlation gap at $550\ \mathrm{K}$9 (Cho et al., 2015).

Other spectroscopies report somewhat different gap scales. Bulk surfaces and monolayers commonly show a “large-gap” spectrum of about $350\ \mathrm{K}$0–$350\ \mathrm{K}$1 with sharp Hubbard peaks on Star-of-David centers, while monolayer ARPES detects the lower Hubbard band near $350\ \mathrm{K}$2 and a gap of about $350\ \mathrm{K}$3 to $350\ \mathrm{K}$4 at low temperature. Earlier ARPES and inverse-ARPES values of $350\ \mathrm{K}$5–$350\ \mathrm{K}$6 have been attributed to resolution limits, broadening, or incomplete access to both Hubbard edges. The literature therefore does not support a single universal number independent of probe, stacking, and dimensionality; rather, it supports a correlation gap whose apparent magnitude depends strongly on how Hubbard edges, broadened subbands, and CDW-derived structures are operationally defined (Chen et al., 3 Mar 2026, Cho et al., 2015).

This spectroscopic complexity underlies a long-running controversy over whether the low-temperature insulator should be described primarily as a Mott insulator or, in some stackings, as a band insulator. DFT without $350\ \mathrm{K}$7 can produce a $350\ \mathrm{K}$8-point gap or a full gap for specific bilayer registries such as A or AL stacking, due to interlayer hybridization of $350\ \mathrm{K}$9-derived states. Against a purely stacking-driven account, monolayer 1T-TaS2 remains insulating and shows Hubbard peaks even when interlayer effects are absent, and alkali-doping experiments distinguish non-rigid Mott-like spectral-weight transfer from rigid band shifts characteristic of paired bilayer surfaces. The most defensible synthesis is that both Mottness and stacking matter, but their relative weights are configuration dependent (Chen et al., 3 Mar 2026, Zhu et al., 2019).

3. Dimensionality, orbital texture, and interlayer stacking

Reduced dimensionality amplifies the electronic consequences of the Star-of-David reconstruction. In DFT for distorted bulk 1T-TaS2, the reconstructed half-filled band has bandwidth $180\ \mathrm{K}$0, weak in-plane dispersion, and strong out-of-plane dispersion, producing a quasi-one-dimensional metallic state in the paramagnetic limit. In the monolayer, by contrast, removal of $180\ \mathrm{K}$1–A dispersion leaves an extremely narrow half-filled band with $180\ \mathrm{K}$2 in GGA, and with $180\ \mathrm{K}$3 this band splits into Hubbard bands opening a gap $180\ \mathrm{K}$4 (Yu et al., 2014).

The same calculations identify an orbital-density-wave (ODW) structure locked to the CDW. The low-energy manifold is dominated by Ta $180\ \mathrm{K}$5 character, strongest on the central Ta of each Star-of-David and decreasing radially outward. Quantitatively, the central Ta carries $180\ \mathrm{K}$6 weight in $180\ \mathrm{K}$7 versus $180\ \mathrm{K}$8 in other $180\ \mathrm{K}$9 components; the first ring carries $223\ \mathrm{K}$0 versus $223\ \mathrm{K}$1; the second ring $223\ \mathrm{K}$2 versus $223\ \mathrm{K}$3. In this description the Star-of-David distortion acts as a Jahn–Teller-like electron–phonon effect that polarizes orbital occupancy and predisposes the narrow cluster band to Mott splitting (Yu et al., 2014).

Interlayer stacking adds a second control parameter. Optical and ARPES studies distinguish an insulating AL stacking associated with layer dimerization from a nearby metallic L stacking, and recent work identifies the ultrafast hidden state as a real-space mosaic of AL and L domains. STEM on exfoliated crystals independently showed A..A.. and A..B.. registries, as well as stacking boundaries formed by a one-bond-length in-plane translation of about $223\ \mathrm{K}$4 over roughly $223\ \mathrm{K}$5. Such a shift induces a fractional interlayer CDW phase offset $223\ \mathrm{K}$6 of order $223\ \mathrm{K}$7 radian, providing a direct structural route for tuning interplane hybridization and the dimensional crossover between more three-dimensional metallic and more quasi-two-dimensional insulating behavior (Hovden et al., 2016, Liu et al., 2024).

Intermediate-thickness flakes reveal how sensitive this balance is. Micro-ARPES on exfoliated $223\ \mathrm{K}$8 and $223\ \mathrm{K}$9 flakes found an equilibrium hidden-phase-like metallic state with a shallow $283\ \mathrm{K}$0-centered band crossing $283\ \mathrm{K}$1 while CDW-related hybridization gaps remain visible. This metallic band persists from $283\ \mathrm{K}$2 to room temperature, with CDW-related suppression at $283\ \mathrm{K}$3–$283\ \mathrm{K}$4 weakening above $283\ \mathrm{K}$5 and collapsing near $283\ \mathrm{K}$6–$283\ \mathrm{K}$7. By contrast, an $283\ \mathrm{K}$8 flake shows only a broadened weaker metallic band, and a $283\ \mathrm{K}$9 flake is fully gapped and incoherent. A plausible implication is that exfoliation-induced modifications of interlayer coherence can stabilize an electronic configuration spectroscopically close to the ultrafast hidden phase without any optical pump (Yilmaz et al., 21 May 2026).

Out-of-plane transport in vertical graphene-contacted heterostructures reaches a related conclusion from a different direction. Few-layer 1T-TaS2 develops a robust Fermi-level gap in c-axis transport, with Arrhenius activation energies of about $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$0 at higher temperature and $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$1 at low temperature. Self-consistent DFT+$p(\sqrt{13}\times\sqrt{13})R13.9^\circ$2 finds that the gap increases as thickness is reduced, from about $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$3 for four layers to about $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$4 in the monolayer, and that in the few-layer limit this gap does not require spin-paired bilayers. This sharpens the distinction between bulk stacking physics and genuine low-dimensional confinement (Boix-Constant et al., 2020).

4. Magnetism, low-energy excitations, and the quantum-spin-liquid question

The C-CDW leaves one nominally unpaired electron per Star-of-David cluster, generating an effective triangular lattice of $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$5 moments. That geometry naturally motivates frustration-based scenarios, yet conventional long-range magnetic order has remained elusive. $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$6SR measurements found no static magnetic order or spin freezing down to $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$7, with zero-field and transverse-field relaxation rates remaining temperature independent and a longitudinal field of $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$8 fully decoupling the signal. Magnetic susceptibility shows a Curie–Weiss fit with $p(\sqrt{13}\times\sqrt{13})R13.9^\circ$9, $13.9^\circ$0, and $13.9^\circ$1, while heat capacity reveals a field-dependent linear term $13.9^\circ$2 that is suppressed by about a factor of three at $13.9^\circ$3. These observations established the absence of magnetic ordering together with intrinsic gapless low-energy excitations in an electrically insulating state (Ribak et al., 2017).

Subsequent thermodynamic and $13.9^\circ$4SR analysis argued that the C-CDW phase contains not one but several competing QSL-like regimes. In that interpretation, the temperature range below the C-CDW transition decomposes into three regimes separated near $13.9^\circ$5 and $13.9^\circ$6, consistent with arrays of undimerized QSL layers interleaved with spin-paired bilayers. Region II was assigned to a nodal $13.9^\circ$7 QSL with linear spinon density of states, while the lowest-temperature regime was assigned to a quantum-critical $13.9^\circ$8 QSL with constant DOS. The proposed stacking-resolved picture is technically specific and attempts to reconcile bulk measurements with the known multiplicity of interlayer CDW registries (Mañas-Valero et al., 2020).

An alternative line of work emphasizes disorder and broad distributions of exchange couplings rather than a clean triangular-lattice QSL. Magnetization studies found thermomagnetic irreversibility below about $13.9^\circ$9, metastability under thermal cycling, and logarithmic magnetic relaxation above roughly $1\times1$0, changing to rapid-rise-and-saturate dynamics below that temperature. Those results were interpreted as evidence for a glass-like random singlet state distinct from a canonical spin glass. Low-field susceptibility measurements likewise showed that $1\times1$1 rises already in the metallic nearly commensurate phase and, below $1\times1$2, follows a power law $1\times1$3 with $1\times1$4–$1\times1$5 depending on field and protocol, again favoring random-singlet physics over a simple Curie tail (Pal et al., 2019, Pal et al., 2022).

In the strictly two-dimensional limit the debate broadens further. A 2026 review summarizes monolayer ARPES reporting an anomalously narrow lower Hubbard band of about $1\times1$6 width, an opening of about $1\times1$7, and a broad momentum-space continuum interpreted in terms of fractionalized spinon and chargon excitations. The same review also discusses impurity-induced Kondo-like responses, in which magnetic adatoms reduce lower-Hubbard-band weight and fill intensity into the Mott gap, whereas non-magnetic alkali dopants mainly shift the chemical potential. These observations are suggestive but not definitive. The present state of the field is therefore plural: 1T-TaS2 is well established as magnetically unconventional, but whether its low-energy spin sector is best described by a clean QSL, a random-singlet manifold, or sample-dependent coexistence remains unresolved (Chen et al., 3 Mar 2026).

5. Hidden states, ultrafast dynamics, and nonequilibrium metallicity

Ultrafast excitation accesses metallic states that are not equivalent to the thermally driven high-temperature metal. Broadband pump–probe spectroscopy at $1\times1$8 showed prompt collapse and recovery of the $1\times1$9 Mott gap, but the transient state exhibited suppressed low-frequency conductivity, strong Fano phonon asymmetry, and a new mid-infrared absorption band centered at $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$0. These features were interpreted as polaronic transport in a state where photodoping melts the correlation gap while the low-temperature CDW lattice symmetry remains largely intact. In this regime the photoinduced metal is not Drude-like; it is a correlated, strongly electron–phonon-coupled conductor (Dean et al., 2010).

The now-standard “hidden state” has been clarified as a mixed-stacking phase rather than a simple domain-wall metal. In situ ARPES during write–erase cycles showed that a single $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$1 pulse at $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$2 and fluence near $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$3 transforms the insulating C-CDW into a nonvolatile state whose spectrum is reproduced by a superposition of insulating AL and metallic L interlayer stackings. The write efficiency oscillates at about $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$4 in pulse-pair experiments, identifying the Star-of-David amplitude mode as the coherent trigger. Erasure proceeds instead through progressive domain coarsening under weak pulse trains or thermal cycling, and the hidden state remains stable below about $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$5 (Liu et al., 2024).

Raman work on exfoliated flakes adds a distinct equilibrium perspective. In $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$6–$\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$7 samples, temperature-dependent Raman spectra revealed a low-temperature transition near $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$8 marked by additional weak phonon modes, including S7, S10, S14, and S24, together with anomalies in phonon self-energy and in the low-frequency electronic Raman slope. The same study tracked a low-frequency Raman response consistent with a metallic-to–Mott-insulating crossover below roughly $\sqrt{13}\,a_0 \approx 12.1\ \text{\AA}$9–$550\ \mathrm{K}$00. This suggests that “hidden” terminology in 1T-TaS2 now spans at least three contexts: ultrafast metastable mixed stacking, equilibrium hidden-phase-like metallicity in intermediate-thickness flakes, and an equilibrium hidden quantum CDW state inferred from low-temperature Raman anomalies (Kumar et al., 5 May 2025, Yilmaz et al., 21 May 2026).

More broadly, nonequilibrium work has become a means of separating electron–phonon and electron–electron dynamics. The characteristic amplitude-mode frequencies, the persistence or collapse of layer dimerization, and the distinct recovery channels of metallic spectral weight and CDW coherence have made 1T-TaS2 a reference material for time-domain studies of correlated CDW insulators. What is unusual is not merely the existence of hidden states, but the fact that several inequivalent hidden-like states can be reached depending on whether the control parameter is light, thickness, cooling history, or interlayer registry (Chen et al., 3 Mar 2026).

6. Defects, doping, and heterointerface proximity effects

Defects in 1T-TaS2 are not electronically generic; they couple to the CDW and the Mott state in defect-specific ways. Spatially resolved STS found that a bright point defect at a CDW maximum can reduce the local Mott gap from about $550\ \mathrm{K}$01 to about $550\ \mathrm{K}$02 while leaving CDW order and CDW gaps largely intact. This behavior was interpreted using a Hubbard–Holstein framework,

$550\ \mathrm{K}$03

where local phonon softening or enhanced electron–phonon coupling lowers the effective on-site repulsion. Optical conductivity places relevant optical phonons near $550\ \mathrm{K}$04, making defect-induced local renormalization of $550\ \mathrm{K}$05 a plausible microscopic route to quasi-metallicity in the C-CDW phase (Cho et al., 2015).

A broader STM/STS, ARPES, and DFT survey at $550\ \mathrm{K}$06 resolved five defect classes, D1–D5. D1 is an acceptor-like dopant that lowers a CDW maximum in topography, narrows the Mott gap to about $550\ \mathrm{K}$07, and produces band bending of about $550\ \mathrm{K}$08 over $550\ \mathrm{K}$09–$550\ \mathrm{K}$10. D2 is a local metallic defect with a peak at $550\ \mathrm{K}$11 and was identified with a sulfur vacancy. D3 introduces an in-gap shoulder near $550\ \mathrm{K}$12; D4 is donor-like and slightly increases the gap to about $550\ \mathrm{K}$13; D5 generates a three-site LDOS depletion motif. In DFT, oxygen substitution closes the gap and suppresses magnetization from $550\ \mathrm{K}$14 per supercell in the pristine monolayer to $550\ \mathrm{K}$15, implying that oxidation is especially consequential for magnetic interpretations of thin 1T-TaS2 (Lutsyk et al., 2023).

Surface alkali doping provides a distinct route to metallization. Potassium adsorption up to about $550\ \mathrm{K}$16 fills the Mott gap with in-gap states while preserving long-range CDW order, and at about $550\ \mathrm{K}$17, $550\ \mathrm{K}$18 out of $550\ \mathrm{K}$19 counted K adatoms occupy Star-of-David centers. The resulting metallic state is unusual because the new in-gap spectral weight appears near the top of the lower Hubbard band rather than beneath the upper Hubbard band, which is the opposite of the simplest electron-doped Mott-insulator expectation. Numerical calculations attribute this to local reduction of $550\ \mathrm{K}$20 on K-dosed stars rather than to CDW disorder (Zhu et al., 2019).

Heterointerfaces show that 1T-TaS2 can both induce and receive proximity effects. In graphene/1T-TaS2, STM observed a proximity-induced CDW in graphene with the same $550\ \mathrm{K}$21 periodicity as the TaS2 substrate, while STS on the covered region found that the TaS2 Mott gap shrinks from $550\ \mathrm{K}$22 to $550\ \mathrm{K}$23, a $550\ \mathrm{K}$24 reduction. DFT modeled this by reducing the effective Hubbard parameter from $550\ \mathrm{K}$25 for bare TaS2 to a phenomenological $550\ \mathrm{K}$26 under graphene, consistent with screening by Dirac carriers (Altvater et al., 2022).

7. External tuning, transport control, and device-oriented phenomena

External tuning can collapse or reshape the CDW–Mott state in several ways. A 2026 review synthesizes evidence that hydrostatic pressure, ionic gating, and chemical substitution or intercalation suppress the C-CDW/Mott phase and can induce superconductivity, typically yielding a superconducting dome adjacent to the collapsed CDW/Mott region. In thin flakes and monolayers, gate-tunable transitions between C-CDW/Mott, metallic, and superconducting regimes are discussed, although that review does not specify quantitative $550\ \mathrm{K}$27, pressure, or carrier-density values (Chen et al., 3 Mar 2026).

Transport heterostructures have converted these phase changes into measurable barrier engineering. In back-gated 1T-TaS2/2H-MoS2 heterojunctions, the commensurate-to-triclinic transition increases the TaS2/MoS2 Schottky barrier, and comparison of phase-dependent barrier heights yields an electrical estimate of the commensurate-phase Mott gap of about $550\ \mathrm{K}$28. Across the nearly commensurate to incommensurate transition, the barrier is suppressed and the heterojunction exhibits gate-controlled resistance switching up to $550\ \mathrm{K}$29, about $550\ \mathrm{K}$30 times larger than standalone TaS2. The result is not merely electrical amplification of an intrinsic transition; it is a phase-sensitive interface effect mediated by Fermi-level pinning and barrier modulation (Mahajan et al., 2018).

Electrical switching in ultrathin flakes also reveals that intermediate resistance states are intrinsic metastable CDW configurations rather than just micron-scale mixtures of NC and C domains. In vertical 1T-TaS2/WSe2/graphite structures, temperature-driven imaging shows micron-sized domains during the NC–C transition, but electrically driven changes at fixed temperature are spatially uniform at the $550\ \mathrm{K}$31 scale. Bidirectional switching between multiple nonvolatile states occurs with an effective lateral reset field $550\ \mathrm{K}$32, and the results were interpreted with a free-energy landscape for the discommensuration density rather than a simple Joule-heating picture (Patel et al., 2020).

Strain is another effective control parameter because the room-temperature NC–IC transition is Joule-heating driven. In flexible thin-flake devices, tensile strain increases the NC-phase resistance and shifts the switching threshold upward with slope about $550\ \mathrm{K}$33, while compressive strain lowers the threshold with slope about $550\ \mathrm{K}$34 when strain magnitude is plotted positively. The gauge factor is about $550\ \mathrm{K}$35 in tension and $550\ \mathrm{K}$36 in compression, and the threshold obeys the quadratic relation $550\ \mathrm{K}$37, directly linking piezoresistance to phase-transition control. The same devices function as threshold-like strain and displacement sensors with large current jumps and programmable detection windows (Merino et al., 29 Oct 2025).

Taken together, these results show that 1T-TaS2 is not defined by a single low-temperature state. It is better described as a correlated CDW material whose electronic phase space is navigated by lattice reconstruction, Mottness, dimensionality, stacking, disorder, and external drive. The material’s continuing significance lies precisely in that multiplicity: the same Star-of-David background supports Mott gaps, band-hybridization gaps, hidden and hidden-like metals, defect-driven quasi-metallicity, possible spin-liquid-like excitations, and tunable transport responses across optical, electrical, and mechanical control channels.