WorksPhysics and Astronomy
Anomalous Hall effect from inter-superlattice scattering in a noncollinear antiferromagnet
Abstract
Superlattice formation dictates the physical properties of many materials, including the nature of the ground state in magnetic materials. Chemical composition is commonly considered to be the primary determinant of superlattice identity, especially in intercalation compounds. Here, we find that, contrary to this conventional wisdom, kinetic control of superlattice growth leads to the coexistence of disparate domains within a compositionally "perfect" single crystal. We demonstrate that CrTaS is a bulk noncollinear antiferromagnet in which scattering between bulk and minority superlattice domains engenders complex magnetotransport below the Néel temperature, including an anomalous Hall effect. We characterize the magnetic phases in different domains, image their nanoscale morphology, and propose a mechanism for nucleation and growth. These results provide a blueprint for the deliberate engineering of macroscopic transport responses via microscopic patterning of magnetic exchange interactions in superlattice domains.
Introduction
Superlattices impart emergent properties to many materials, including metal alloys,[1] semiconductors,[2] and ceramics.[3] The periodicity and symmetry of the long-range order determine the underlying many-body physics, with especially profound implications for quantum materials. Moiré superlattices, for example, host a plethora of emergent electronic phenomena, such as unconventional superconductivity,[4] correlated insulating states,5,6 and the quantum anomalous Hall effect.7,8 In magnetic materials, superlattices of the spin-bearing centers can dictate the distances between them, strongly influencing the microscopic magnetic exchange interactions and hence the ground state.9–11
The link between superlattice identity and magnetic ordering is especially apparent in transition metal dichalcogenides (TMDs) intercalated with first-row transition metals.12–14 In this family of materials, many types of magnetism can be accessed by varying the host lattice, intercalant, and stoichiometry, including hard ferromagnetism,15–17 noncollinear antiferromagnetism,18–20 and spin glass phases.21,22 Although off-stoichiometry and structural disorder are known to modify their magnetotransport properties,23–25 composition is generally treated as a proxy for the periodicity of the intercalant superlattice. Assuming a formula of , where is a first-row transition metal, is Nb or Ta, and is S or Se, it is typically assumed that results in a superlattice, results in a superlattice, and intermediate compositions result in defective, disordered, or mixed superlattices.24,26–30 Nevertheless, since the magnetism of these materials is highly sensitive to the intercalant superlattice,26,31,32 precise control over atomic-scale ordering is necessary to realize transport signatures associated with specific magnetic phases, including noncollinear textures and altermagnetism.32–36
In this work, we synthesize the new material and find that it is a bulk noncollinear antiferromagnet with a well-ordered superlattice of Cr. Below the Néel temperature () of 145 K, it exhibits an anomalous Hall response, which is inconsistent with the symmetry of the magnetic structure. Detailed characterization of the nanoscale atomic structure reveals minority domains with different superlattice ordering, even in high-quality, stoichiometric single crystals. We propose a mechanism for kinetically arrested growth of disparate superlattice domains and show that the complex magnetotransport originates from scattering between domains. The results show how variation of local ordering can be leveraged to tune magnetotransport in an antiferromagnet, with implications for designing transport responses via deliberate superlattice engineering without changes in chemical composition.
Results
We initially targeted the compound with the hypothesis that () is a promising intercalant for targeting a noncollinear ground state: it has a largely quenched orbital moment37 and strong easy-plane anisotropy,38 which can lead to predictable interplay between second-order perturbations and Heisenberg exchange. For example, in and , the Dzyaloshinskii–Moriya interaction arising from broken inversion symmetry competes with ferromagnetic coupling, resulting in chiral helimagnetism.39–41 We thus surmised that putting on a geometrically frustrated triangular lattice with antiferromagnetic (AFM) coupling would favor a noncollinear ground state to relieve the frustration (Figure 1a), potentially yielding an anomalous Hall response.18–20,42–44
Figure 1. is a noncollinear antiferromagnet. (a) Local -orbital splitting for , and design principle for noncollinear antiferromagnetism. (b) Structure of from single-crystal X-ray diffraction. The unit cell ( relative to -) is indicated in black and the triangular lattice is emphasized in purple. (c) Heat capacity () vs. normalized to the formula . (d) Single-crystal neutron diffraction data: Integrated intensities of the peak vs. , and the peak at different temperatures (solid lines indicate Gaussian fits), associated with the propagation vector . (e) antiferromagnetic structure as obtained from neutron diffraction ( representation). The magnetic unit cell (relative to the Cr superlattice) is shown in light blue, and the nuclear unit cell is shown in black. The magnetic structure is invariant with respect to the three mirror planes perpendicular to the magnetic moments (dotted black lines).
Single crystals of were grown from the constituent elements using chemical vapor transport with iodine as a transport agent. We found that Cr crystallographic disorder was minimized in a two-zone furnace with the hot zone at 1100 °C and the cold (growth) zone at 1000 °C, and a cooling rate of 20 °C/hour. The structure, as determined by single-crystal X-ray diffraction (SCXRD), consists of - layers with 1/4 of the pseudo-octahedral sites in the van der Waals gap occupied by Cr, forming a superlattice (Figure 1b; more details of the refinement in Table S1 and Table S2). Within the sensitivity limits of a laboratory diffractometer, we detect no evidence of Cr deficiency on the site, electron density on the site, or additional reflections corresponding to the superlattice (Figure S1). Unless otherwise indicated, all experiments were carried out on crystals from two batches with “perfect” (within experimental error) Cr stoichiometries of as determined by energy dispersive X-ray spectroscopy (EDS, Figure S2), and a fully occupied superlattice as determined by SCXRD (see the Supplementary Information for more details on crystal growth and compositional characterization).
| Empirical formula | |
| Formula weight (g/mol) | 1032.27 |
| Temperature (K) | 293(2) |
| Wavelength (Å) | 0.71073 |
| Crystal system | Hexagonal |
| Space group | |
| (Å) | 6.5959(3) |
| (Å) | 12.0391(7) |
| Volume (Å) | 453.60(5) |
| 2 | |
| Density (calculated) (g/cm) | 7.558 |
| Absorption coefficient (mm) | 50.985 |
| (000) | 888 |
| Crystal size (mm) | 0.031 × 0.023 × 0.015 |
| (°) | 3.384 to 29.622 |
| Index ranges | |
| Reflections collected | 18880 |
| Independent reflections | 272 |
| Completeness to | 0.975 |
| Absorption correction | Semi-empirical from equivalents |
| Refinement method | Full-matrix least-squares on |
| Data / restraints / parameters | 272 / 0 / 18 |
| Goodness-of-fit on | 1.331 |
| Final indices [] | , |
| indices (all data) | , |
| Largest diff. peak and hole (e Å) | 4.46 and −4.24 |
Table S1. Crystal data and refinement details for from single-crystal X-ray diffraction.
| Atom Label | Site | ||||
| Cr1 | 0 | 0 | 2a | 0.0121(13) | |
| Ta2 | 0 | 0 | 2b | 0.0064(5) | |
| Ta3 | 0.49435(7) | 0.50565(7) | 6h | 0.0063(4) | |
| S4 | 0.1667(3) | 0.3334(5) | 0.3809(3) | 12k | 0.0074(8) |
| S5 | 0.1188(5) | 4f | 0.0076(11) |
Table S2. Atomic coordinates, Wyckoff positions, and equivalent isotropic displacement parameters for from single-crystal X-ray diffraction.

Figure S1. Reconstructed , , and scattering planes from single-crystal X-ray diffraction of . Bragg peaks associated with the - host lattice and Cr superlattice are circled in light blue and navy, respectively.
Figure S2. Representative energy dispersive X-ray spectroscopy data for a single crystal of CrTaS, with peaks corresponding to Cr, Ta, S, and C labeled. The atomic ratio determined by fitting the Cr K, Ta M, and S K peaks yields a formula of CrTaS for this crystal.
Heat capacity and neutron diffraction measurements establish that is a bulk antiferromagnet with K. A single anomaly in the heat capacity () is observed at 145 K (Figure 1c and inset), indicating a bulk phase transition at this temperature. Using single-crystal neutron diffraction in the scattering plane, we observe satellite peaks below 150 K consistent with the magnetic propagation vector , corresponding to the onset of AFM ordering (Figure 1d). (Due to crystal size requirements, we measured a slightly Cr-deficient crystal with .) Representation analysis of the observed using SARAh45 indicates that the magnetic representation can be decomposed into six irreducible representations (), of which and are consistent with the easy-plane anisotropy. The only stable refinement of data collected at 1.5 K is obtained with , which corresponds to an in-plane cycloidal structure with parallel orientation of moments along . When an equal moment constraint is applied, the value refines to , with the spins oriented along the directions, i.e. a 120° AFM structure (Figure 1e). The lower-than-expected moments (compared to the theoretical ) could be a consequence of itinerant magnetism, which has been observed in related materials.35,36,46 Additional details of the refinement are available in the Supplementary Information (Table S3–Table S5). Altogether, the heat capacity and neutron diffraction data unambiguously point to a single bulk AFM transition in at 145 K.
Table S3: Refinement details for the nuclear structure of from single-crystal neutron diffraction.
| Parameter | Value |
| Crystal mass (mg) | 16.0 |
| Temperature (K) | 1.5 |
| Wavelength (Å) | 1.486 |
| Space group | |
| (Å) | 6.619(5) |
| (Å) | 11.961(9) |
| Volume (Å) | 453.8(6) |
| 4.64 | |
| 6.36 | |
| 5.05 | |
| 5.87 | |
| reflections | 58 [] |
| Scale factor | 693(12) |
Table S3.
| Parameter | Value |
| Crystal mass (mg) | 16.0 |
| Temperature (K) | 1.5 |
| Wavelength (Å) | 1.486 |
| Representation | |
| 19.8 | |
| 30.8 | |
| 10.2 | |
| 3.22 | |
| reflections | 19 |
| Cr1 moment () | 2.07(8) |
Table S5. Refinement details for the magnetic structure of from single-crystal neutron diffraction.
Electrical transport measurements on bulk single crystals of reveal metallic behavior from the in-plane longitudinal resistivity () (Figure 2a). A sharp drop at K is consistent with reduced carrier scattering upon bulk AFM ordering. The residual resistivity ratio, , is about 10, indicating good crystal quality. The in-plane Hall resistivity () is dominated by the ordinary Hall effect (Figure 2b). An anomalous Hall contribution, evidenced by jumps in centered at zero field, is present below (Figure 2c,Figure 2d). We fit for fields smaller than 5 T to a single-band model comprising an ordinary Hall component and a small component from an anomalous Hall effect (AHE), . From the ordinary Hall component, we extract carrier concentrations () for the dominant hole carriers on the order of , and carrier mobilities () up to at 2.5 K (Figure 2e), reasonable values for an electron-doped intercalation compound of 2H-TaS.16 The faster decrease in below 150 K may correspond to Fermi surface reconstruction as a result of AFM ordering.47

Figure 2. Magnetotransport properties of CrTaS. (a) In-plane longitudinal resistivity () and vs. . (b) In-plane Hall resistivity () vs. , and schematic of the measurement configuration. Colors indicate temperatures as labeled in (C). (c) Anomalous Hall resistivity () vs. at different temperatures. Translucent lines are the raw data; opaque lines are denoised. (d) The saturation values of vs. . (e) Charge carrier concentration () and carrier mobility () vs. , as derived from the ordinary Hall component of . (f) Magnetoresistance (MR) vs. at different temperatures. Colors indicate temperatures as labeled in (C). (g) MR at 12 T vs. .
Several interesting features are observed in the temperature dependence of (Figure 2c,Figure 2d). Below 40 K, is negligible. Between 40 to 100 K, is positive in sign and exhibits a small zero-field hysteresis of less than 1 T up to 80 K. The saturation magnitude of increases with increasing in this regime. At 110 K, the circulation of changes sign while retaining a similar magnitude. then approaches zero with increasing temperature and vanishes above . The anomalous Hall conductivity, , has a maximum magnitude of about 9 at 50 K.
The magnetoresistance (MR), defined as , also exhibits a complex temperature dependence (Figure 2f,Figure 2g). At 10 K and below, the MR is positive and approximately quadratic with respect to field. Above 10 K, the MR decreases quickly with increasing . The high-field MR becomes negative at 60 K and increases gradually in magnitude until 100 K. There is a significant decrease in the magnitude of the negative MR between 100 and 110 K. At intermediate temperatures in the positive and negative MR regimes, cusps are observed around zero field (Figure 2f inset). At 150 K and above, i.e. above , the MR becomes uniformly quadratic in field and remains negative.
The fact that a non-zero and complex temperature-dependent MR are observed only below indicates that these magnetotransport phenomena are tied to the bulk AFM order.
The intrinsic anomalous Hall conductivity (AHC) can be evaluated from the integral of Berry curvature over the Brillouin zone within linear response theory. However, symmetry analysis of the magnetic structure shows that it is invariant with respect to the mirror planes perpendicular to the magnetic moments, as indicated by the dashed lines in Figure 1e. The Berry curvature is odd under these mirror symmetry transformations. Therefore, the intrinsic AHC sums to zero across the Brillouin zone.48 Hence, we sought to investigate mechanisms for an extrinsic, scattering-mediated AHE that could also explain the rich MR behavior.
Considering the nature of as an intercalation compound, we used more sensitive probes of local symmetry to study possible intercalant disorder as a source of magnetic scattering. In confocal Raman microscopy (with a laser spot size of approximately ), we observe a sharp, intense phonon mode at , which we attribute to the Cr superlattice (Figure 3a,Figure 3b).49,50 A small but sharp feature at is also present, which matches the superlattice mode observed in the compound .51 This suggests that a small amount of ordering with sub-micron in-plane domain sizes is present in bulk crystals of , and that these -containing domains coexist with the dominant superlattice.

Figure 3. Evidence for domains in . (a) Structures of -, , and , with , , and unit cells. (b) Raman spectra of , , and -, with Cr superlattice phonon modes highlighted. (c) Selected area electron diffraction of an exfoliated flake of , with primitive and superlattice reflections indicated. (d) Virtual dark-field images from four-dimensional scanning transmission electron microscopy (4D-STEM) reconstructed using the intensities of and Bragg disks. (e) 4D-STEM diffraction patterns from the regions indicated in (d). (f) Field-cooled (FC) and zero-field-cooled (ZFC) DC magnetic susceptibility () vs. , measured with a 100 Oe field parallel and perpendicular to . (g) Real part of the AC susceptibility () vs. , measured with an in-plane AC field of 10 Oe. (h) Isothermal magnetization () vs. . (i) Arrott plot with linear fit to the 98 K trace.
Electron diffraction studies on unambiguously confirm the existence of ordered -containing domains, and furthermore demonstrate nanoscale domains of different superlattices. Selected area electron diffraction (SAED) of a mechanically exfoliated flake in a circular region with a radius of shows superlattice reflections, as well as significantly weaker but sharp reflections (Figure 3c). To investigate the spatial distribution of superlattice domains, we used four-dimensional scanning transmission electron microscopy (4D-STEM). In 4D-STEM, a converged electron beam ( under our experimental conditions) is scanned across a two-dimensional (2D) area of the sample, and 2D diffraction data is collected at each probe position.52 Virtual dark-field images reconstructed using the intensities of and Bragg disks reveal in-plane spatial separation of superlattices (Figure 3d). The superlattice is dominant over the entire area except for two small regions that resemble pinched triangles. These -deficient regions precisely
correspond to the brightest regions in the map. Individual electron diffraction patterns shown in Figure 3e illustrate the evolution of superlattice order over a 100 nm length scale, showing mixed superlattices with maximal order, weaker order, and exclusive order in moving from a triangular -containing region to a region.
The presence of order offers a clue for interpreting bulk magnetometry data on , which are inconsistent in several respects with the fully compensated AFM structure associated with the Cr superlattice. A sharp rise in the field-cooled DC magnetic susceptibility () below 100 K, along with marked bifurcation between zero-field-cooled and field-cooled traces, is suggestive of a minority ferromagnetic (FM) transition (Figure 3f). The larger with the magnetic field applied in-plane is consistent with the expected easy-plane behavior. Fitting to the Curie–Weiss law, , yields emu K (mol Cr) and K (Figure S3). We then obtain , close to the theoretical spin-only value of for (). The value of is positive but considerably smaller than the temperature of the upturn in , suggesting the coexistence of AFM and FM coupling.53
Figure S3. Fitting of inverse DC susceptibility data vs. to the Curie–Weiss law, , for a 1 T magnetic field perpendicular to .
The real part of the AC magnetic susceptibility () shows a cusp below 100 K, as well as a broader and more prominent feature with a maximum at about 40 K with a modest frequency dependence (Figure 3g and Figure S4). The peak at 40 K is especially pronounced in the imaginary part of the AC susceptibility (Figure S5). These observations suggest magnetic irreversibility at lower temperatures. We tentatively propose a reentrant spin glass (SG) state below 40 K that is driven by spin disorder arising from competition between AFM and FM exchange interactions.22,23 This interpretation is corroborated by the slow relaxation dynamics observed in thermoremanent magnetization measurements (Figure S6 and Table S6).54

Figure S4. Real part of the AC susceptibility (), measured with an in-plane AC driving field of 10 Oe, fitted to a univariate spline in the vicinity of the spin glass transition temperature.

Figure S5. Imaginary part of the AC susceptibility (), measured with an in-plane AC driving field of 10 Oe.
Figure S6. Thermoremanent magnetization curves obtained by field-cooling in a field of 1 T, removing the field, then measuring the magnetization as a function of time. Open circles represent the data, while solid lines represent fits to stretched exponential functions for 2 K, 15 K, 50 K, and 80 K.
| (K) | (/Cr) | (/Cr) | (s) | |
| 2 | 0.28768(2) | 0.003641(6) | 1224.4(4) | 0.6541(7) |
| 15 | 0.25815(2) | 0.007156(6) | 1231.3(2) | 0.6274(4) |
| 50 | 0.18978(5) | 0.009451(9) | 2315.6(5) | 0.6586(4) |
| 80 | 0.07196(3) | 0.01006(7) | 1425.8(2) | 0.6399(3) |
Table S6. Parameters for fitting thermoremanent magnetization data to a stretched exponential function, , where is the total magnetization, is the time, is the constant magnetization, is the glassy component of the magnetization, is the characteristic relaxation time, and is a stretching exponent.
Partial FM ordering in a minority of the sample is also supported by the isothermal magnetization () vs. field traces (Figure 3h). At 4 K, with shows a small zero-field hysteresis of 400 Oe (Figure 3h inset). A detectable hysteresis persists up to 94 K (Figure S7), above which a slight “S”-shape remains below 150 K. The magnetization is non-saturating up to fields of 12 T, reaching a maximum value of 0.55 /Cr at 4 K, which is reasonable for compensated AFM ordering in the majority of the sample. An Arrott plot indicates a first-order FM transition with = 98 K (Figure 3i), 55 in agreement with the AC and DC susceptibility results.
Figure S7. Isothermal magnetization curves with , showing zero-field hysteresis at 94 K and below.
It is known that the fully occupied Cr superlattice in exhibits local FM coupling both in- and out-of-plane, with chiral helimagnetic order developing below 140 K because of competition between FM Heisenberg exchange and the Dzyaloshinskii–Moriya (DM) interaction. 40,41 The introduction of vacancies on the Cr superlattice in the closely related material lowers markedly and suppresses the DM interaction, resulting in bulk FM behavior. 26,28 Hence, we attribute the partial FM ordering observed in to the minority -containing structural domains. The lower of 98 K observed in our samples compared to the reported of is consistent with a defective Cr superlattice in the minority FM domains of , which is expected from the average stoichiometry of our samples.
To investigate the effects of Cr intercalation and superlattice ordering on the electronic structure of , we first consider the symmetry of the AFM and the dominant structural superlattice relative to the host lattice of -. The 2D Brillouin zone (BZ) for the AFM BZ is scaled down by and rotated by relative to the structural BZ, which is itself scaled down by relative to the host lattice BZ (Figure 4a). As a result, extensive zone folding is expected in-plane. For the three-dimensional (3D) BZ, the out-of-plane reciprocal lattice vector remains unchanged.
Figure 4. Electronic structure of CrTaS. (a) Two-dimensional Brillouin zone (BZ) for 120° AFM, superlattice, and host lattice, and three-dimensional BZ for 120° AFM, with high-symmetry points labeled. (b) Experimental Fermi surfaces from angle-resolved photoemission spectroscopy measured with linear horizontal (LH) and linear vertical (LV) polarization. Dotted lines indicate the host lattice BZ. (c) Energy vs. dispersion along –––. Overlay: density functional theory (DFT) band structure of 2H-TaS shifted down by 0.2 eV. Purple arrows indicate replica bands from folding that cross . (d) DFT band structure for CrTaS in the 120° AFM phase. (e) Energy vs. dispersion at from the photon energy dependence. (f) Energy vs. dispersions from constant cuts centered at A and . Overlays: DFT band structures along the –– direction with and for A and , respectively. All ARPES data were measured at 10 K with 80 eV, LH-polarized photons unless otherwise indicated.
Using angle-resolved photoemission spectroscopy (ARPES), we find that the experimental Fermi surfaces show clear evidence of reconstruction (Figure 4b). With linear horizontal (LH) polarized photons (80 eV), we see replica hole pockets at that are folded from . In linear vertical (LV) polarization, we also observe duplication of the hole pockets from at the corners of the BZ. By fitting the main hole pocket around from the LH data, we obtain Å along – (Figure S8), which is intermediate between 2H-TaS and CrTaS.51,56
Figure S8. Momentum distribution curve for eV and Å, fitted to multiple Lorentzian peaks.
The ARPES energy vs. momentum dispersion shows further evidence for superlattice-induced electronic reconstruction. As expected, the most prominent bands are consistent with the overlaid density functional theory (DFT) band structure for 2H-TaS with the Fermi level () shifted down by 0.2 eV (Figure 4c).57 However, clear replica bands are present throughout the BZ that do not correspond to any bands from the unfolded host lattice (indicated by purple arrows in Figure 4c and Figure S9). The DFT band structure calculated for the 120° AFM structure of indicates that the magnetic and structural folding results in several bands crossing (Figure 4d).
Figure S9. Increasing cell size and the folding of the bandstructure (yellow, shifted down by 0.2 eV) of -TaS in the reciprocal space in comparison with the ARPES experiment. (a) , (b) , (c) unit cells of -TaS. The calculated bandstructure of (d) , (e) , (f) unit cells of -TaS.
We observe a marked dispersion of bands near , as shown in Figure 4e, consistent with the –A dispersion from DFT (Figure 4d). Constant cuts from the high-symmetry points qualitatively match the DFT band structures for and , respectively (Figure 4f and Figure S10). This constitutes additional evidence that the Cr superlattice and 120° AFM structure strongly modulate the electronic structure beyond the rigid band picture in a manner consistent with other intercalated TMDs.51,58–61 More broadly, the ARPES experiments further confirm that the superlattice is dominant, as we do not observe evidence of electronic reconstruction associated with minority -containing structural domains.
Figure S10. (a) Band structures of with U=0 (red) and U=4 (blue) eV. (b) Band structures of with (red), (blue), and (green).
Discussion
From the magnetometry data and geometry of domains observed in 4D-STEM, we propose the following simplified scheme for Cr ordering during crystal growth and the subsequent cooling process (Figure 5a). At high temperatures, a defective superlattice, which has higher configurational entropy, is more stable. At intermediate temperatures, enthalpically favored domains nucleate and grow. At lower temperatures, Cr intercalants become immobile due to insufficient thermal energy, thus freezing the superlattice domain configurations at room temperature and below. This implies that if domains are not given enough time to grow during cooling, defective -containing domains will persist between them. The spatial separation of superlattices revealed by 4D-STEM suggests that despite cooling at the relatively slow rate of 20 °C/hour and being within experimental error of the perfect stoichiometry, growth of pure domains appears to be kinetically limited in our best samples. This picture explains how order can be present in samples with compositions that should favor a superlattice from purely thermodynamic considerations,32 consistent with computational predictions.62

Figure 5. In-plane superlattice domain formation and magnetic phases in . (a) Schematic illustration of growth and ordering of Cr intercalant domains at elevated temperatures. Gray dots indicate possible interstitial sites, blue dots indicate Cr, light green shaded regions indicate local order, and light blue shaded regions indicate order. (b) Proposed magnetic phases in and structural domains at 145 K and below, comprising antiferromagnetic (AFM), ferromagnetic (FM), paramagnetic (PM), and spin glass (SG) phases. (c) Schematic illustration of inelastic scattering of spin-up and spin-down carriers between AFM domains and minority domains or domain walls.
Accordingly, we propose the following coexistence of magnetic phases in based on the transition temperatures for the AFM, FM, and SG phases established previously (Figure 5b). Below 145 K, the majority domains exhibit AFM order, while the minority -containing domains remain paramagnetic (PM). Below 98 K and above 40 K, the -containing domains exhibit FM order, and below 40 K, they enter an SG state. We note that our proposed scheme for superlattice domain formation considers only in-plane ordering. From the 4D-STEM data, the -containing domains also show reflections (Figure 3e), implying that some mixing of superlattices occurs in the out-of-plane direction. It is possible that the structural proximity of superlattices with FM vs. AFM coupling both in- and out-of-plane is responsible for the observed SG behavior.
The magnetic transition temperatures correlate well with changes observed in the AHE. A nonzero AHE emerges below 145 K, coinciding with the onset of AFM ordering. We attribute the negative sign between 100 and 145 K to an extrinsic AHE from inelastic scattering, specifically from spin defects at domain walls between AFM domains and PM minority superlattice domains (Figure 5c).16,63 The AHE is positive between 40 and 100 K, which corresponds with FM ordering in the minority domains. The complex scaling between the anomalous Hall conductivity () and the longitudinal conductivity () suggests that different mechanisms are responsible in different temperature regimes (Figure S11). No clear signature of AHE is observed below 40 K, reminiscent of the disappearance of some components of the AHE in the cluster glass phase of the kagome AFM MnSn.42
Figure S11. Plot of the longitudinal conductivity () vs. the anomalous Hall conductivity ().
These temperature regimes are further supported by the MR behavior, which is generally consistent with disparate magnetic phases coexisting on the nanoscale. The crossover between positive and negative MR and low-field cusps at intermediate temperatures are reminiscent of dilute magnetic semiconductors and magnetically doped topological insulators with clustering of the magnetic species.64,65 We note also that CrNbS likewise exhibits complex field dependence of the MR and a sign change in the AHE, which have been attributed to field evolution of noncoplanar spin textures.66,67 In CrTaS, although the minority -containing domains are defective, local DM interactions or domain wall effects may still lead to spin canting or chiral textures that influence the transport behavior.
To test our proposed mechanisms, we carried out an analogous set of magnetometry and transport experiments on Cr-deficient crystals with the composition CrTaS. The Cr-deficient material clearly shows an AFM transition at 145 K, with no second transition at 98 K. It exhibits a negative and a small, linear vs. , indicating the absence of FM/SG domains (Figure S12). The Raman spectrum does not show a well-defined peak, and the superlattice mode is slightly broadened and red-shifted, consistent with a defective superlattice (Figure S13). Critically, we observe zero detectable AHE at 200 K and below (Figure S14). The Cr-deficient material has a value of 1.6, almost an order of magnitude lower than CrTaS, which we attribute to increased scattering off of defects in the Cr superlattice. We also observe higher carrier concentration, lower mobility, and lower MR, which are additionally consistent with lower crystal quality and the absence of FM/SG domains. These findings indicate that the combined presence of a well-ordered and fully occupied Cr superlattice and minority -containing domains is required for the rich magnetotransport behavior observed in .
Figure S12. Magnetometry of CrTaS. (a) Zero-field-cooled (ZFC) DC magnetic susceptibility () vs. , measured with a 1 T magnetic field perpendicular to . (b) Fitting of inverse DC susceptibility data vs. to the Curie–Weiss law, . (c) Isothermal magnetization () vs. .
Figure S13. Raman spectra of and , with and superlattice peak regions highlighted for .
Figure S14. Magnetotransport properties of . (a) In-plane longitudinal resistivity () vs. . (b) In-plane Hall resistivity () vs. . (c) Charge carrier concentration () and carrier mobility () vs. , as derived from the ordinary Hall component of . (d) Anomalous Hall resistivity () vs. at different temperatures, obtained by subtracting the ordinary Hall component of . (e) Magnetoresistance (MR) vs. at different temperatures. (f) MR at 12 T vs. .
Conclusions
is a metallic antiferromagnet with a noncollinear AFM ground state. High-quality crystals with a well-ordered Cr superlattice nevertheless also contain minority domains with local order. These samples host complex magnetotransport phenomena below the AFM , including temperature-dependent sign changes in AHE and MR. Interactions between the majority AFM phase and minority PM/FM/SG domains engender the observed transport responses through scattering between the disparate superlattices. Altogether, this work shows that growth conditions (in addition to composition) must be taken into account to determine superlattice identity and microstructure. Our results corroborate a growing body of literature indicating that subtle differences in crystallographic order in intercalated TMDs can be harnessed towards realizing a wide array of magnetic and electronic properties.24,25,28,31,68–70
The possibility of engineering disparate magnetic phases by controlling the geometry of superlattice domains, as well as the topology of domain walls, is an intriguing prospect and worthy of further exploration.71,72 In addition, intercalated TMDs might be promising platforms to study analogues of the three- and four-state Potts universality classes (corresponding to 2D and ordering, respectively) in three dimensions.73 More generally, our work points to the importance of using sensitive and local probes to investigate the possibility of disorder or inhomogeneity in intercalation compounds with more than one stable superlattice. Further tailoring of scattering interactions via optimization of superlattice nucleation and growth may lead to the development of new materials exhibiting electrical transport responses relevant to spintronic devices.
Materials and Methods
Single crystals of were grown from Cr (powder, 99.97%, Alfa Aesar), Ta (powder, 99.98%, Alfa Aesar, and S (powder, 99.999%, Acros Organics) using (99.999%, Spectrum Chemicals) as a transport agent. The constituent elements in a 0.33:1:2.05 molar ratio were sealed in an evacuated fused quartz ampoule (14 mm inner diameter, 1 mm wall thickness, 25 cm long) along with 2 mg/cm of . The ampoule was placed in a two-zone furnace with the hot zone maintained at 1100 °C and the cold (growth) zone maintained at 1000 °C for 200 hours, after which both zones were cooled down to room temperature over the course of 50 hours (about 20 °C/hour). Hexagonal plate-shaped crystals were obtained, with lateral dimensions of several mm and thicknesses of 1–2 mm. Single crystals of were grown in an analogous fashion using a 0.30:1:2.02 molar ratio, 2.3 mg/cm of , and a 1100 °C hot zone and 950 °C cold zone.
Single-crystal X-ray diffraction data were collected on a Rigaku XtaLAB P200 with Mo K radiation at 295 K. Data reduction and scaling and empirical absorption correction were performed in CrysAlis Pro. Structures were solved by direct methods using SHELXT74 and refined against on all data by full-matrix least squares with SHELXL75 using the ShelXle graphical user interface.76 Reconstructed scattering planes were generated from the frames in CrysAlis Pro. Energy dispersive X-ray spectroscopy data were acquired on a FEI Quanta 3D FEG scanning electron microscope with an accelerating voltage of 20 kV.
Single-crystal neutron diffraction measurements were conducted on WAND at the High Flux Isotope Reactor (Oak Ridge National Laboratory) with an incident wavelength of 1.486 Å. A single crystal was mounted on an aluminum rod with GE Varnish in the scattering geometry. The data were integrated using Mantid Workbench.77 Representational analysis was carried out using SARAh,45 and structural refinement was performed using FullProf.78
Heat capacity, electrical transport, and magnetometry measurements were carried out in a Quantum Design Physical Property Measurement System Dynacool equipped with a 12 T magnet. For heat capacity measurements, single crystals were affixed to the stage using Apiezon N grease. For electrical transport measurements, cleaved single crystals with thicknesses of 30 or less were affixed using GE Varnish and contacted using silver paint and gold wire. Measurements were conducted using Stanford Research Systems SR830 lock-in amplifiers by applying a 5 mA AC current (17.777 Hz) and measuring the transverse and longitudinal voltages in typical four-probe or Hall configurations. and data were symmetrized and antisymmetrized, respectively. and MR data were denoised with a Savitzky–Golay filter; a notch filter was applied to MR data to remove low-frequency noise on the order of 0.001 Hz. Magnetometry measurements were performed using the Vibrating Sample Magnetometer option for DC measurements, and the AC Measurement System II option for AC measurements, with crystals affixed to quartz paddles or brass holders using GE Varnish.
Confocal Raman microscopy data were collected on a Horiba LabRAM HR Evolution with an ultra-low frequency filter using 633 nm laser excitation and powers between 1 and 8 mW.
Plan-view transmission electron microscopy (TEM) imaging of a nm-thick flake was performed using an FEI TitanX operating at 80 keV. Selected area electron diffraction (SAED) and four-dimensional scanning transmission electron microscopy (4D-STEM) were acquired in the same sample region. Samples were prepared via a dry transfer method utilizing a poly(bisphenol A carbonate)/polydimethylsiloxane polymer stamp.17 In this process, flakes were mechanically exfoliated using Kapton tape onto /Si, and were then transferred onto a 200 nm silicon nitride holey TEM grid (Norcada) that had been treated with plasma for 5 minutes immediately prior to stacking. For the SAED experiments, a 40 diameter aperture was used, defining a selected diameter of nm. On the other hand, 4D-STEM was acquired with a 0.55 indicated convergence semi-angle yielding a nm converged electron probe size. The acquired 4D-STEM data were analyzed using the py4DSTEM Python package.79 Peak detection algorithms identified the – Bragg peaks, which were used to calculate the reciprocal lattice vectors of the – structure. Based on their symmetry relations to these vectors, the and Cr superlattice reciprocal vectors were determined. Virtual apertures were then constructed separately. for each superlattice by masking all regions of the diffraction patterns except for the corresponding Cr superlattice peak areas. These apertures were applied to the 4D-STEM data to extract the integrated intensities of the two distinct superlattices at each probe position.
Angle-resolved photoemission (ARPES) measurements were carried out on beamline 4.0.3 (MERLIN) of the Advanced Light Source (Lawrence Berkeley National Laboratory) equipped with a Scienta Omicron R8000 hemispherical electron analyzer. Crystals were cleaved in situ under high vacuum (base pressures of Torr or less) by carefully knocking off alumina posts affixed to the top surface using silver epoxy. Photon energy-dependent measurements were conducted between 30 and 124 eV, and momentum conversion was carried out using an inner potential () of 8 eV. The primary datasets were collected at eV, close to as determined by the photon energy dependence. Data analysis was carried out using the PyARPES software package.80
First-principles calculations based on density functional theory (DFT) were performed using the Vienna Ab Initio Simulation Package (VASP)81,82 and the open source plane-wave code QuantumEspresso (QE).83 The projector-augmented wave (PAW) methods84 implemented in VASP was used with a kinetic energy cutoff of 400 eV. The optimized norm-conserving Vanderbilt (ONCV) pseudopotentials from the PseudoDojo project85,86 were applied for calculations using QE with a kinetic energy cutoff of 86 Ry. The lattice constants ( Å, Å) of CrTaS were taken from a single-crystal X-ray diffraction measurement at 100 K. A -center k-mesh was used to sample the Brillouin zone for CrTaS, and k-mesh was used for TaS. The exchange-correlation interaction was described by the Perdew-Burke-Ernzerhof (PBE) functional.87 An effective on-site Coulomb interaction of 4 eV on Cr atom was employed within DFT+ of the Dudarev scheme.88 The magnetic ordering in CrTaS was taken as the 120° noncollinear antiferromagnetic structure using a magnetic unit cell shown in Figure 1e. VASP was used for the calculations of band structure and magnetic properties in CrTaS, and QE was used for the calculation of band structure in TaS.
Acknowledgments
We thank Sae Hee Ryu for assistance with ARPES measurements. A portion of this research used resources at the High Flux Isotope Reactor, a DOE Office of Science User Facility operated by the Oak Ridge National Laboratory. The beam time was allocated to WAND on proposal number IPTS-30492.1. This research used resources of the Advanced Light Source, which is a DOE Office of Science User Facility under contract no. DE-AC02-05CH11231. Work at the Molecular Foundry, LBNL, was supported by the Office of Science, Office of Basic Energy Sciences, the U.S. Department of Energy under Contract no. DE-AC02-05CH11231. Computational work in this paper used the Lux Supercomputer at UC Santa Cruz, funded by NSF MRI Grant No. AST 1828315. L.S.X. acknowledges support from the Arnold and Mabel Beckman Foundation through an Arnold O. Beckman Postdoctoral Fellowship (award no. 51532). B.H.G. was supported by the University of California Presidential Postdoctoral Fellowship Program (UC PPFP), Schmidt Science Fellows in partnership with the Rhodes Trust, and the Max Planck Society. O.G. acknowledges support from a NSF Graduate Research Fellowship Grant DGE 1752814. S.H. acknowledges support from the Blavatnik Innovation Fellowship.
Author Contributions
L.S.X. and D.K.B. conceived of the study. L.S.X. synthesized samples and carried out transport and EDS experiments. L.S.X., O.G., and S.S.F. collected and analyzed SCXRD data. M.D.F., S.S.F., and L.S.X. collected and analyzed neutron diffraction data. W.F. and K.L. carried out symmetry analysis and first-principles DFT calculations. L.S.X., M.P.E., and C.M. collected and analyzed Raman data. S.H. collected and analyzed electron diffraction data; B.H.G. and I.M.C. provided analysis and interpretation. C.M., L.S.X., and S.S.F. collected and analyzed magnetometry and heat capacity data. L.S.X., J.P.D., S.S.F., and M.P.E. collected and analyzed ARPES data. D.K.B. and Y.P. supervised the project. L.S.X. wrote the manuscript with input from all authors.
Competing Interests
The authors declare no competing interests.
Single-Crystal X-ray Diffraction
Energy-Dispersive X-ray Spectroscopy
Single-Crystal Neutron Diffraction
Table S4: Atomic coordinates, Wyckoff positions, isotropic displacement parameters, and occupancies for the nuclear structure of from single-crystal neutron diffraction.
| Atom Label | Site | Occupancy | ||||
| Cr1 | 0 | 1 | 1/2 | 2a | 1.73(7) | 0.848(26) |
| Ta2 | 0 | 1 | 1/4 | 2b | 1.73(7) | 1 |
| Ta3 | 0.495(1) | 0.989(2) | 1/4 | 6h | 1.73(7) | 1 |
| S4 | 0.1670(9) | 0.8330(9) | 0.3802(5) | 12k | 1.73(7) | 1 |
| S5 | 1/3 | 0.118(1) | 4f | 1.73(7) | 1 |
Table S4.
Magnetometry
To extract the maximum of the real part of , corresponding to the spin glass freezing temperature () at each frequency, we fit the data to a univariate spline and take the root of the derivative. We then quantify the frequency dependence of by calculating the Mydosh parameter (), according to the formula at different frequencies (). We obtain , which is somewhat larger than typical values for canonical spin glasses (e.g. 0.0045 for AuMn),[1] and in line with expected values for cluster spin glasses.[2, 3, 4]
The obtained values for and are consistent with those expected for a spin glass.[2] The persistence of thermoremanent magnetization above K and below K suggests that the FM state also exhibits some glassy characteristics.
Angle-Resolved Photoemission Spectroscopy
Density Functional Theory Calculations
In Figure S9a–c, we show different unit cells of -TaS from , , to . One effect of increasing the unit cell size is the folding of the band structure in the reciprocal space, as seen in Figure S9d–f. In spite of the band folding, the calculated band structure of -TaS always contains the key features that are measured from the ARPES experiment. The position of the key features in DFT calculation differs by 0.2 eV with respect to the experiment. The comparison indicates that the main contribution to ARPES measurement is from -TaS, and that the intercalated Cr shifts the Fermi level by introducing extra electrons.
In Figure S10a, we compare the band structures of CrTaS with and eV. Including Hubbard interaction moves the bands upward, which brings the agreement between DFT band structures and ARPES measurements closer. It also increases the magnetic moment on Cr from to , which makes it close to the theoretical spin-only value of /Cr for Cr (). In Figure S10b, we compare the band structures of CrTaS with different . Increasing makes many bands degenerate. Overall, including Hubbard with a gives the best agreement between DFT band structures and the main ARPES dataset collected with eV.
Magnetotransport of CrTaS
Characterization of CrTaS
References
- [1]Mulder, C. A. M., van Duynneveldt, A. J. & Mydosh, J. A. Frequency and field dependence of the ac susceptibility of the AuMn spin-glass. Phys. Rev. B 25, 515–518 (1982).
- [2]Mydosh, J. A. Spin Glasses: An Experimental Introduction (CRC Press, 1993).
- [3]Anand, V. K., Adroja, D. T. & Hillier, A. D. Ferromagnetic cluster spin-glass behavior in PrRhSn₃. Phys. Rev. B 85, 014418 (2012).
- [4]Bag, P., Baral, P. R. & Nath, R. Cluster spin-glass behavior and memory effect in Cr₀.₅Fe₀.₅Ga. Phys. Rev. B 98, 144436 (2018).



