Nonvolatile nematic order manipulated by strain and magnetic field in a layered antiferromagnet Authors: Zili Feng, Weihang Lu, Tao Lu, Fangyan Liu, Joseph R. Sheeran, Mengxing Ye, Jing Xia, Takashi Kurumaji, Linda Ye Published: 2026 ABSTRACT Electronic nematicity refers to a state whose electronic properties spontaneously break rotation symmetries of the host crystalline lattice, leading to anisotropic electronic properties. In this work, we demonstrate that the layered antiferromagnet CoTa3S6 exhibits a switchable nematic order, evidenced by the emergence of both resistivity anisotropy and optical birefringence. This nematic state sets in at a temperature T* distinct from that of the antiferromagnetic transitions in the system, indicating a separate symmetry-breaking mechanism. The nematic order can be manipulated either by an in-plane rotation symmetry-breaking strain or in-plane magnetic field, with the latter exhibiting a pronounced non-volatile memory effect. Remarkably, we find that the broken three-fold rotation symmetry in electronic transport is restored with a moderate out-of-plane field. We hypothesize that the nematicity is of electronic origin and emerges from instabilities associated with van Hove singularities. The resulting phase diagram points to an intertwined interplay between the electronic nematicity and the proposed underlying collinear and non-coplanar spin orders. Our findings establish CoTa3S6 as a versatile antiferromagnetic platform with highly tunable functionalities arising from the breaking of rotational, time-reversal, and inversion symmetries. Introduction Antiferromagnets, characterized by magnetic orders with zero or nearly zero net moments, have emerged as promising candidates for next-generation data storage devices due to minimal stray fields, robustness against magnetic field fluctuations, and fast switching dynamics1–3. However, the same set of attributes also poses significant challenges in manipulating antiferromagnetic domains, in stark contrast to ferromagnets, where magnetic fields—serving as the conjugate variable to the order parameter—enable straightforward writing of magnetic bits. The realization of functional memory devices based on antiferromagnets thus hinges on identifying robust, non-volatile orders that can be manipulated by external stimuli and reliably read by external probes. Recent advances include electrical readout of antiferromagnetic Néel vectors4,5 and their current-switching in non-centrosymmetric antiferromagnetic systems5–8. From a band structure perspective, antiferromagnets also allow a versatile interplay between various symmetries with topology and geometry of the underlying electronic states3,9,10. As a distinct class of quantum phases, electronic nematicity refers to states where the electronic responses spontaneously break rotational symmetries of the underlying crystalline lattice, and can be viewed as a quantum analog of the orientational orders in liquid crystals11,12. Initially identified in two-dimensional electron gas at the quantum Hall regime13, nematicity has been reported in various strongly correlated oxides and pnictides12,14–18, and most recently in moiré superlattices19,20 and kagome superconductors21–23. A nematic state selects a special direction in space, often indicated by a headless vector termed nematic director, while the particular rotation symmetry that’s broken determines the number of domains which would in principle serve as memory bits. Examples include Ising nematicity with two distinct states11,17 and Potts nematicity with three24–26. Analogous to liquid crystals, the anisotropy induced by the nematic order, and its response with respect to external parameters are key to its functionalities. Here we present an unexpected discovery of a unique three-state nematic state that coexists and interacts with underlying antiferromagnetic orders in a layered compound CoTa3S6. The electronic anisotropy of the system is found to be closely intertwined with the antiferromagnetic orders. CoTa3S6 belongs to a family of magnetic element-intercalated transition metal dichalcogenides (TX2, T=Nb, Ta, X=S, Se) that were first studied in the 1970s driven by the rich magnetic states that can be realized depending on the intercalant species and concentration27–32. CoTa3S6 can be viewed as an ordered structure for fractional (1/3) intercalation of Co into the van der Waals gap between TaS2 layers forming a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sqrt{3}\times \sqrt{3}$$\end{document}3×3 superlattice. The crystal structure is shown in Fig. 1a, while in reality a small off-stoichiometry may be allowed33. The cobalt atoms between the TaS2 layers form triangular lattices (Fig. 1b) and the overall crystal structure lacks inversion symmetry and is chiral, belonging to the space group P6322 34.Fig. 1Emergence of resistivity anisotropy and optical birefringence in CoTa3S6.a Crystal structure of CoTa3S6. Co atoms (blue spheres) are sandwiched by TaS2 layers composed of TaS6 triangular prisms. Ta (S) atoms are shown in red (yellow). b The Co-triangular lattice in the ab-plane; throughout the work we adopt a Cartesian coordinate system (x, y, z) and the definition for the axes with respect to the principal crystallographic axes is also shown. c Schematic of three-state nematicity on the Co triangular lattice; each of the inequivalent nematic domains is labeled as A, B and C, respectively. The long axes of the red shaded ellipses correspond to the direction along which Δρ > 0. d–f Temperature (T) dependence of magnetization (d), anomalous Hall resistivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{yx}^{A}$$\end{document}ρyxA (e), and normalized resistivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}={\rho }_{ii}/{\rho }_{{{\rm{ZFC}}}}$$\end{document}ρ~ii=ρii/ρZFC for i = x, y (f). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tilde{\rho }$$\end{document}ρ~ is defined as the ratio between field-trained and zero field cooled states (see text and Methods). Inset of (d) shows a magnified view of M(T) near TN1. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{yx}^{A}$$\end{document}ρyxA shown in (e) is the spontaneous Hall resistivity at each T with scanning H∥z. T* in (f) indicates the onset temperature of resistivity anisotropy. g Polar plots of optical polarization rotation θT as a function of incident light polarization (α) for three different nematic domains. Plus/minus are the signs of the polarization rotation by birefringence. Solid curves are fits to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\theta }_{T}={\theta }_{P}\sin [2(\alpha -{\alpha }_{0})]$$\end{document}θT=θPsin[2(α−α0)] with θP and α0 as fitting parameters; the two-way arrows indicate the slow axis, and the diameter of the gray circles corresponds to 600 microrad. h Fitted birefringence amplitude θP from measurements at various temperatures (raw data see Supplementary Information (SI) Sec. II). The blue solid line is a guideline to the eye. Emergence of resistivity anisotropy and optical birefringence in CoTa3S6. a Crystal structure of CoTa3S6. Co atoms (blue spheres) are sandwiched by TaS2 layers composed of TaS6 triangular prisms. Ta (S) atoms are shown in red (yellow). b The Co-triangular lattice in the ab-plane; throughout the work we adopt a Cartesian coordinate system (x, y, z) and the definition for the axes with respect to the principal crystallographic axes is also shown. c Schematic of three-state nematicity on the Co triangular lattice; each of the inequivalent nematic domains is labeled as A, B and C, respectively. The long axes of the red shaded ellipses correspond to the direction along which Δρ > 0. d–f Temperature (T) dependence of magnetization (d), anomalous Hall resistivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{yx}^{A}$$\end{document}ρyxA (e), and normalized resistivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}={\rho }_{ii}/{\rho }_{{{\rm{ZFC}}}}$$\end{document}ρ~ii=ρii/ρZFC for i = x, y (f). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tilde{\rho }$$\end{document}ρ~ is defined as the ratio between field-trained and zero field cooled states (see text and Methods). Inset of (d) shows a magnified view of M(T) near TN1. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{yx}^{A}$$\end{document}ρyxA shown in (e) is the spontaneous Hall resistivity at each T with scanning H∥z. T* in (f) indicates the onset temperature of resistivity anisotropy. g Polar plots of optical polarization rotation θT as a function of incident light polarization (α) for three different nematic domains. Plus/minus are the signs of the polarization rotation by birefringence. Solid curves are fits to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\theta }_{T}={\theta }_{P}\sin [2(\alpha -{\alpha }_{0})]$$\end{document}θT=θPsin[2(α−α0)] with θP and α0 as fitting parameters; the two-way arrows indicate the slow axis, and the diameter of the gray circles corresponds to 600 microrad. h Fitted birefringence amplitude θP from measurements at various temperatures (raw data see Supplementary Information (SI) Sec. II). The blue solid line is a guideline to the eye. A recent surge of attention on CoTa3S6 and isostructural CoNb3S6 was stimulated by reports of large anomalous Hall effects in the xy-plane arising in both compounds within their antiferromagnetic states35–39. For CoTa3S6 in particular, two successive magnetic transitions are reported at TN1 ≈ 37 K and TN2 ≈ 26 K, while the anomalous Hall effect is exclusively observed below TN227,28,36. Neutron scattering studies have revealed magnetic modulation vectors {1/2, 0, 0} below both TN1 and TN2, and a single-q collinear antiferromagnetic state is proposed between TN1 and TN2 while a four-sublattice non-coplanar spin structure is proposed below TN240–42. The latter as a triple-q state is supported by the observation of higher harmonics {1/2, 1/2, 0} and symmetry analysis of magnetic point group to account for the observed anomalous Hall effect40,41,43–45. That anomalous Hall effect is allowed here suggests that the system is a promising candidate for multi-functional, antiferromagnetic spintronic applications. Results Identification of nematic order in CoTa3S6 We highlight our central observation of anisotropic electronic behaviors in CoTa3S6–which allows identification of a three-state nematic state (Fig. 1c) –in close comparison with the magnetic transitions in the system in Fig. 1d–h. Magnetization (M) along z was measured with H∥z at μ0H = 0.1 T (Fig. 1d); a weak kink can be observed at TN1 = 37 K (see Fig. 1d inset) while an onset of a net moment can be observed below TN2 = 26.5 K. Importantly, we note that the net moment is two orders of magnitude smaller than that expected for the effective magnetic moment of Co2+ ( ~ 3.8μB). The onset of the net moment below TN2 is also accompanied by the appearance of a spontaneous anomalous Hall effect \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{yx}^{A}$$\end{document}ρyxA as summarized in Fig. 1e, comparable with that reported in previous studies40,41 (see (SI) Sec. I for characterization of our crystals from the anomalous Hall effect). The electronic anisotropy is sensitively captured through in-plane transport as exemplified by the temperature dependence of the normalized resistivities \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}$$\end{document}ρ~xx and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{yy}$$\end{document}ρ~yy in Fig. 1f. The x axis is along the bonds of the Co triangular lattice as shown in Fig. 1b. Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}(T)$$\end{document}ρ~ii(T) is defined by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}={\rho }_{ii}(T,{{\rm{FC}}})/{\rho }_{ii}(T,{{\rm{ZFC}}})$$\end{document}ρ~ii=ρii(T,FC)/ρii(T,ZFC); ρii(T, FC) is measured at zero field after the field-trained process at 9 T of H∥x (we will also get back to the training procedure below), while ρii(T, ZFC) is measured after zero-field cooling. ρxx and ρyy are obtained through simultaneous measurements using a modified Montgometry method (see Methods and SI Sec. IV). Note that for a hexagonal crystalline lattice, no resistivity anisotropy is expected, i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}={\tilde{\rho }}_{yy}=1$$\end{document}ρ~xx=ρ~yy=1. We have found that the in-plane resistivity anisotropy turns on below TN1 at T * ≃ 34 K, which persists to low temperature. Recalling that the measurements are performed at zero field, it indicates that the six-fold rotation symmetry of the crystal is spontaneously broken below the characteristic temperature scale T *. Furthermore, the antisymmetric evolution of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}$$\end{document}ρ~xx and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{yy}$$\end{document}ρ~yy with temperature implies that the transport tensor indeed develops a nematic-like anisotropy in the xy plane with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rho }_{xx}=\bar{\rho }+\Delta \rho,{\rho }_{yy}=\bar{\rho }-\Delta \rho$$\end{document}ρxx=ρ¯+Δρ,ρyy=ρ¯−Δρ, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\bar{\rho }$$\end{document}ρ¯ is the average in-plane resistivity and Δρ reflects the difference in ρxx and ρyy relative to the corresponding isotropic state \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({\rho }_{xx}={\rho }_{yy}=\bar{\rho })$$\end{document}(ρxx=ρyy=ρ¯). We take the convention that the long axes of the nematic directors (Fig. 1c) has Δρ > 0. We have verified that an isotropic, multi-domain state is realized in ρ(ZFC) (see SI Sec. IV), and expect \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}-{\tilde{\rho }}_{yy}=2\Delta \rho /\bar{\rho }$$\end{document}ρ~xx−ρ~yy=2Δρ/ρ¯ to play the role of a nematic order parameter. We note that a very weak kink in resistivity anisotropy may be identified near the onset of net moment/anomalous Hall effect at TN2. The emergence of electronic anisotropy in CoTa3S6 is further evidenced by the observation of optical birefringence at low temperature (see Methods and SI Sec. II). In Fig. 1g we show the polar plot of the birefringence response at 10 K of three domains we have identified. Their optical principal axes are close to that expected for domains A, B, and C illustrated in Fig. 1c, respectively. We find that the slow axis tends to align with the a or equivalent crystallographic axes. The T-evolution of birefringence is summarized in Fig. 1h, which closely follows the T-evolution of resistivity anisotropy (Fig. 1f, raw data see Fig. S5); both highlighting a growing electronic anisotropy with lowering T. That electronic anisotropy emerges at a temperature between TN1 and TN2 and persists in the triple-q ground state is rather unexpected, as the superposition of three equivalent q vectors is expected to retain the three-fold rotational symmetry40,41. These all point to the fact that the nematicity in CoTa3S6 likely is a distinct phase from the antiferromagnetic orders; we will return to this below. Controlling nematic domains with in-plane strain With a finite coupling between an anisotropic, nematic state and the lattice, one may expect using in-plane strain that breaks rotation symmetry in the same manner of the nematic order parameters to couple to the former and control the nematic domains16,25,26,46. In Fig. 2a we illustrate a possible scenario of domain selection (the validity of the scenario is corroborated by experiments discussed below) as a function of in-plane antisymmetric strain \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{2}\left({\epsilon }_{xx}-{\epsilon }_{yy}\right)$$\end{document}12ϵxx−ϵyy: with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{2}\left({\epsilon }_{xx}-{\epsilon }_{yy}\right) > 0$$\end{document}12ϵxx−ϵyy>0 one out of the three Potts nematic domains is selected while for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{2}\left({\epsilon }_{xx}-{\epsilon }_{yy}\right) < 0$$\end{document}12ϵxx−ϵyy<0 two out of the three nematic domains are selected. In our experiment, we apply a uniaxial stress along x which generates a finite component in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{2}\left({\epsilon }_{xx}-{\epsilon }_{yy}\right)$$\end{document}12ϵxx−ϵyy. The longitudinal resistivity ρxx at different bias strains ϵxx ≡ ΔLx/Lx are summarized in Fig. 2b, and here the magnetic field is applied along the z-axis.Fig. 2Resistivity anisotropy and nematic domains manipulated by in-plane strain.a Schematic strain effect on the Co triangular lattice and expected nematic domain selection by in-plane strain. b Out-of-plane magnetic field (H∥z) dependence of ρxx measured at 2 K under various magnitudes of strain along the x axis (ϵxx). c Strain dependence of the magnetoresistance MR = (ρxx(0 T) - ρxx(9 T))/ρxx(9 T) (blue circle). Dark blue curve is a guideline to the eye. The expected activated nematic domains at the compressive and tensile ends of the strain range are also shown. Inset is the strain dependence of ρxx(9 T) fitted with a linear function of ϵxx. Resistivity anisotropy and nematic domains manipulated by in-plane strain. a Schematic strain effect on the Co triangular lattice and expected nematic domain selection by in-plane strain. b Out-of-plane magnetic field (H∥z) dependence of ρxx measured at 2 K under various magnitudes of strain along the x axis (ϵxx). c Strain dependence of the magnetoresistance MR = (ρxx(0 T) - ρxx(9 T))/ρxx(9 T) (blue circle). Dark blue curve is a guideline to the eye. The expected activated nematic domains at the compressive and tensile ends of the strain range are also shown. Inset is the strain dependence of ρxx(9 T) fitted with a linear function of ϵxx. Qualitatively, as a function of ϵxx, the magnetoresistance in Fig. 2b appears to abruptly switch sign from negative on the most tensile side (blue curves) to positive on the most compressive side (red curves). This can also be seen in the ϵxx-dependence of both the zero-field and high-field ρxx summarized in Fig. 2c. At 9 T, ρxx appears to evolve linearly with ϵxx (Fig. 2c inset), and similar linear elastoresistivity can arise from a combination of the strain gauge effect and modification of the band structure under strain47,48. To eliminate such contributions in the strain-evolution of ρxx(0 T), we compare the percentage magnetoresistance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left[{\rho }_{xx}(0{{\rm{\,T}}})-{\rho }_{xx}(9{{\rm{\,T}}})\right]/{\rho }_{xx}(9{{\rm{\,T}}})$$\end{document}ρxx(0 T)−ρxx(9 T)/ρxx(9 T) against ϵxx in Fig. 2c. This quantity can be found to be highly nonlinear with ϵxx, and it changes sign near ϵxx = 0% and tends to saturate towards both the compressive and tensile ends of ϵxx. We note that owing to the mismatch of thermal expansion coefficients of the strain cell, sample plate and the sample itself, it is difficult to determine the real zero strain. Here we define the strain at which the magnetoresistance is nearly zero as the nominal zero strain (see SI Sec. V for similar results at 10 K). As demonstrated above in Fig. 1f, ρxx and ρyy evolves in an antisymmetric manner within the nematic state; this allows us to use either ρxx and ρyy on their own as an indication of the nematic director. We therefore attribute the different ρxx values at the compressive/tensile sides to a resistive detection of the strain-induced selection of nematic directors (see schematics in Fig. 2c). The dark blue curve in Fig. 2c is a guideline to the eye. We note that in the ideal case, in-plane strain tuning of resistivity anisotropy is expected to exhibit a linear interpolation between values at the two ends, as ϵxx − ϵyy is a thermodynamic variable expected to be proportional to imbalance of the nematic domains49. In-plane magnetic field manipulation of nematic domains Having demonstrated that in-plane strain can control the nematic domains, next we turn to the manipulation of nematic domains by in-plane magnetic fields14,17,50. Similar with in-plane strains, the presence of an in-plane magnetic field breaks the six-fold rotation symmetry of the lattice and is expected to couple to and can act to select amongst three nematic domains, as schematically illustrated in Fig. 3a.Fig. 3Manipulation of resistivity anisotropy by in-plane magnetic fields.a Schematic of the relationship between in-plane magnetic field direction and the nematic director. θ is the angle of the magnetic field with respect to the x axis. b ρxx and ρyy as a function of in-plane magnetic field along y-axis (θ = 90∘). The initial state is prepared by ZFC from T = 50 K and 2 K. The magnetic field is applied in the sequence of μ0H = 0 T → + 14 T → − 14 T → + 14 T. c–e Time-dependence of ρxx and ρyy (c) with subsequent application of Hx and Hy training fields with field history as a function of time shown in (d, e). f Schematic of domain selection with in-plane magnetic field alternating between Hx and Hy. Ellipses with darker shade indicate the favored domain(s) at respective field configurations. Manipulation of resistivity anisotropy by in-plane magnetic fields. a Schematic of the relationship between in-plane magnetic field direction and the nematic director. θ is the angle of the magnetic field with respect to the x axis. b ρxx and ρyy as a function of in-plane magnetic field along y-axis (θ = 90∘). The initial state is prepared by ZFC from T = 50 K and 2 K. The magnetic field is applied in the sequence of μ0H = 0 T → + 14 T → − 14 T → + 14 T. c–e Time-dependence of ρxx and ρyy (c) with subsequent application of Hx and Hy training fields with field history as a function of time shown in (d, e). f Schematic of domain selection with in-plane magnetic field alternating between Hx and Hy. Ellipses with darker shade indicate the favored domain(s) at respective field configurations. Figure 3 b shows a typical in-plane field-evolution of resistivity anisotropy at T = 2 K starting from a zero field cooled, multi-domain state. Here the in-plane field is applied along y(θ = 90∘). The ZFC state is isotropic in ρxx and ρyy, while as field increases, a finite resistivity anisotropy starts to emerge from 7 T with ρxx and ρyy again bifurcating in opposite, antisymmetric manners. The weak parabolic magnetoresistance in both ρxx and ρyy can be attributed to the orbital Lorentz force. We note that as field reaches 14 T, the resistivity anisotropy saturates, and when the magnetic field is swept back and forth between ± 14 T, the system maintains the resistivity anisotropy initialized at high magnetic fields. This suggests that (1) in-plane field along y acts to select B and C domains, both of which exhibit a resistivity anisotropy ρxx − ρyy < 0 and (2) the memory of resistivity anisotropy and nematic domains is retained when the magnetic field is removed. The non-volatile nature of the nematic domains allows one to utilize them as medium for a field-controlled nematic memory. We therefore devise a writing procedure using pulse-like in-plane magnetic fields to train the domains and subsequently using resistivity anisotropy as a readout mechanism (Fig. 3c–e). At 5 K the switching between states with ρxx − ρyy > 0 and ρxx − ρyy < 0 are triggered by x and y magnetic fields, respectively, and the operation can be repeated multiple cycles without decay in amplitude. Figure 3f illustrates the corresponding activated nematic domains in the in-plane field training sequence. Such effect is found to persist up to 25 K (see SI Sec. VIII). We note that similar non-volatile domain selection is expected for the application and removal of in-plane uniaxial stress, which is not accessed in the present set of experiments (see Methods). In SI Sec. IX, we further verify that the nematic director is pinned along three directions parallel to the principal a and equivalent axes of the crystal and a dominant domain can be selected in a discrete manner by the rotation of the in-plane training field. Temperature-field phase diagram of CoTa3S6 In order to gain additional insights on the origin of nematicity in relation to the underlying magnetic orders in CoTa3S6, below we evaluate the effect of out-of-plane field to the observed transport anisotropy. Figure 4a shows \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}$$\end{document}ρ~ii as a function of out-of-plane magnetic field (H∥z). Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}(H)$$\end{document}ρ~ii(H) is the normalized resistivity defined by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}(H)={\rho }_{ii}$$\end{document}ρ~ii(H)=ρii(H,FC)/ρii(H,ZFC); ρii(H,FC) is measured after field cooling with in-plane magnetic field from T = 50 K to 2 K, while ρii(H,ZFC) is measured after zero-field cooling (see Methods). As we have demonstrated above, following the removal of in-plane training field, the system retains the memory of the nematic domains at low temperature. Starting from an anisotropic state thus prepared, as an out-of-plane field is gradually turned on, as shown in Fig. 4a, the resistivity anisotropy of the trained state suddenly collapses at a critical magnetic field Hc ≃ 3 T, which corresponds well with the sudden drop in ρyx (inset of Fig. 4a) and sudden jump in Mz (see SI Sec. I) both observed in our own crystals and reported in literature36,40,41. This sudden collapse of resistivity anisotropy suggests that Hc is a characteristic field scale where the nematic order is undermined by a moderate out-of-plane field; in other words, at Hc the system ascends to a rotation symmetric state.Fig. 4Phase diagram of CoTa3S6 for magnetic field along the z-axis.a Magnetic field dependence of normalized in-plane resistivities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}$$\end{document}ρ~ii) with H∥z at T = 2 K. Prior to the measurement, the sample is field-cooled with in-plane magnetic field along the x axis at 9 T (see text and Methods). The inset shows the magnetic field dependence of ρyx at T = 2 K. Red arrows indicate the critical field (Hc). b Color map of the resistivity anisotropy (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}-{\tilde{\rho }}_{yy}$$\end{document}ρ~xx−ρ~yy) in the H∥z − T plane. The transition temperatures TN1 (gray circles) and TN2 (black circles) are extracted from the temperature dependence of magnetization. Blue circles (triangles) are defined by singularities in ρyy(H) (ρyy(T)) scans (see SI Secs. IV and XI), and green stars are defined by kinks in ρyx(H). c Temperature evolution of nematic, collinear and non-coplanar antiferromagnetic order parameters in a Landau free energy picture (see Methods and text). d, e Resistivity anisotropy ρxx − ρyy (d) and dρxx/dT (e) as a function of temperature of Sample B01. ρxx and ρyy were measured in a FC state by the Montgomery method under zero magnetic field. Here TN1 and TN2 are determined from kinks in dρxx/dT and T* from the onset of ρxx − ρyy (see text). Phase diagram of CoTa3S6 for magnetic field along the z-axis. a Magnetic field dependence of normalized in-plane resistivities (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{ii}$$\end{document}ρ~ii) with H∥z at T = 2 K. Prior to the measurement, the sample is field-cooled with in-plane magnetic field along the x axis at 9 T (see text and Methods). The inset shows the magnetic field dependence of ρyx at T = 2 K. Red arrows indicate the critical field (Hc). b Color map of the resistivity anisotropy (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{\rho }}_{xx}-{\tilde{\rho }}_{yy}$$\end{document}ρ~xx−ρ~yy) in the H∥z − T plane. The transition temperatures TN1 (gray circles) and TN2 (black circles) are extracted from the temperature dependence of magnetization. Blue circles (triangles) are defined by singularities in ρyy(H) (ρyy(T)) scans (see SI Secs. IV and XI), and green stars are defined by kinks in ρyx(H). c Temperature evolution of nematic, collinear and non-coplanar antiferromagnetic order parameters in a Landau free energy picture (see Methods and text). d, e Resistivity anisotropy ρxx − ρyy (d) and dρxx/dT (e) as a function of temperature of Sample B01. ρxx and ρyy were measured in a FC state by the Montgomery method under zero magnetic field. Here TN1 and TN2 are determined from kinks in dρxx/dT and T* from the onset of ρxx − ρyy (see text). We note that above the critical field Hc, ρxy is comparable with that at zero field (Fig. 4a inset), while the resistivity anisotropy vanishes thoroughly. This suggests that the nematic order and the anomalous Hall-driving antiferromagnetic phase are likely of distinct origin. Notably, in experiments with canted magnetic fields (see SI Sec. VI), we find that the control of resistivity anisotropy is independent of the sign of the anomalous Hall effect, suggesting that the former is controlled by in-plane components while the latter by z-component of the magnetic field only. These different responses again indicate that the nematicity and Hall effect arise from distinct degrees of freedom in CoTa3S6. In Fig. 4b we summarize the evolution of resistivity anisotropy in a color scale in the out-of-plane field and temperature plane, and solid marks indicate phase boundaries obtained from magnetization and transport (ρyy, ρyx) measurements. At all temperatures below TN1, the field scale where resistivity anisotropy vanishes agrees well with that defined by kinks in ρyx. Additionally, the shape of the phase diagram also implies that the nematic order below T* and Hc is a distinct phase from the collinear antiferromagnetic order which onsets at TN1 and the non-coplanar antiferromagnetic order which sets in at TN2: not only isT* at zero field distinct from both TN1 and TN2, but the T* nematic phase is much more fragile with respect to out-of-plane field; no discernible changes of TN1 and TN2 can be identified up to 9 T. Discussion Thus far, our observations have established the presence of a three-state nematic state within the antiferromagnetic orders in CoTa3S6. The state can be read out by resistivity anisotropy and optical birefringence and manipulated by in-plane strain and magnetic fields. Additionally, we identified that the rotation symmetry-breaking can be ‘turned off’ by an out-of-plane magnetic field, offering an additional means of tuning the anisotropic functionalities of the system. The fact that this suppression is also evident in both magnetization and anomalous Hall signals (see SI Fig. S2 and Fig. 4a inset) suggests that, while the nematic order constitutes a distinct degree of freedom, it is coupled to the underlying magnetic phases. Here we illustrate the interplay between the three orders in CoTa3S6 using a minimal Landau free energy picture. Here we consider (L1, L2, L3) for the magnetic order parameter (Li is a Néel vector for an antiferromagnetic configuration propagating along one of the M points in the hexagonal Brillouin zone25) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({n}_{1},{n}_{2})=n(\cos 2\theta,\sin 2\theta )$$\end{document}(n1,n2)=n(cos2θ,sin2θ), where θ is the angle between x and the nematic director for the nematic order parameter24,25. The free energy associated with the magnetic orders is (see SI Sec. III for a more complete description) 1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{rcl}{F}_{{{\rm{mag}}}}&=&a \mathop{\sum }\limits_{i=1}^{3}{{{\bf{L}}}}_{i}^{2}+{v}_{0}{\left(\mathop{\sum }\limits_{i=1}^{3}{{{\bf{L}}}}_{i}^{2}\right)}^{2}+{v}_{1}\mathop{\sum }\limits_{i < j}{{{\bf{L}}}}_{i}^{2}{{{\bf{L}}}}_{j}^{2}\\ &&+{v}_{2}\mathop{\sum }\limits_{i < j}{\left({{{\bf{L}}}}_{i}\cdot {{{\bf{L}}}}_{j}\right)}^{2}\end{array}$$\end{document}Fmag=a∑i=13Li2+v0∑i=13Li22+v1∑i 0,\tilde{a} > 0,{\tilde{v}}_{1} > 0$$\end{document}a=a~(T−TN1),v1=v~1(T−TN2),v0>0,a~>0,v~1>0. The competition between v1 and v2 terms drives the single-q to triple-q transition51. The free energy for the nematic state together with its coupling with the modulated magnetic states is 2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{rcl}&&{F}_{{{\rm{n}}}}+{F}_{{{\rm{n-mag}}}}=b{n}^{2}+c{n}^{3}\cos 6\theta+w{n}^{4}\\ &&+gn\left(\cos 2\theta \left({{{\bf{L}}}}_{1}^{2}+{{{\bf{L}}}}_{2}^{2}-2{{{\bf{L}}}}_{3}^{2}\right)+\sin 2\theta \sqrt{3}({{{\bf{L}}}}_{1}^{2}-{{{\bf{L}}}}_{2}^{2})\right)\end{array}$$\end{document}Fn+Fn-mag=bn2+cn3cos6θ+wn4+gncos2θL12+L22−2L32+sin2θ3(L12−L22) With the above formalism (see Methods for the parameter values used), we are able to produce the schematic order parameter evolution in Fig. 4c. The behaviors closely mirror the onset of the anomalous Hall effect (Fig. 1e) and transport anisotropy (Fig. 4d) of the system, as well as the weak cusp at TN2 in the latter. The free energy analysis also points to that above T*, a tail in the nematic order parameter is induced by a finite coupling with the stripe antiferromagnet order at TN1; this leads us to attribute T* to the intersection of extrapolated fits in ρxx − ρyy in Fig. 4d, which appears to occur at a temperature distinct from both TN1 and TN2 defined as kinks in dρxx/dT (Fig. 4e). We note that results in Fig. 4d and e are obtained from simultaneous measurements of ρxx and ρyy and that T* is distinct from TN1 and TN2 is reproduced in multiple samples and is independent of in-plane fields applied (see SI section X). The close comparison with our free energy model implies that the nematicity arises independently from the antiferromagnetic orders regardless of their single-q and triple-q nature. Their intertwined orders contrasts with (vestigial) nematicity induced by stripe antiferromagnetic orders in Fe1/3NbS2 and FePSe325,26,52, or rotation-symmetry-breaking charge density waves53,54. The emergence of nematicity here as a separate, non-vestigial order is rather surprising; to our knowledge, CoTa3S6 represents a quite unique case of intertwined nematic and antiferromagnetic orders with distinct origins. We hypothesize that the nematic order in CoTa3S6 is primarily driven by electronic rather than magnetic degrees of freedom. For instance, photoemission studies show hexagonal and triangular Fermi surfaces that nearly touch the M-points of the Brillouin zone40,55, indicating the presence of van Hove singularities near the Fermi level. Such van Hove singularities in hexagonal lattices have been widely discussed as a driving force of a wide variety of electronic orders, including nematicity through the Pomeranchuk instability24,56–60. This intertwined interplay of orders echoes paradigms seen in strongly correlated systems such as cuprates and iron pnictides61,62, and more recently in kagome metals with itinerant electrons on frustrated lattices with electronic singularities62–65. Elucidating the microscopic origin of the observed nematicity in CoTa3S6 represents a compelling direction for future research and may help establish design principles for electronic instabilities in metallic systems. As a final remark, we note that structurally, CoTa3S6 lacks inversion symmetry through its chiral structure, and time-reversal symmetry is known to be broken at low temperature evidenced by the anomalous Hall effect. Our experimental results further reveal that the zero magnetic field ground state of CoTa3S6 breaks the underlying rotation symmetry of the lattice. The interplay between the various broken symmetries may lead to novel electronic and optoelectronic functionalities in CoTa3S6, such as a six-state magneto-nematic memory which can be written magnetically and read electrically (via joint measurements of the anomalous Hall effect and resistivity anisotropy), and current-tunability of the nematic order and/or the anomalous Hall effect analogous to current switching of Néel vectors in Fe1/3NbS28. Moreover, the observed non-volatile memory associated with the nematic order may enable control over nonlinear electrical transport66 and spin current generation67, leveraging the simultaneous absence of inversion and time-reversal symmetries. Methods Crystal growth and characterization Single crystals of CoTa3S6 were grown by a two-step process27,28,36,41. Polycrystalline CoTa3S6 was first synthesized by solid-state reaction from stoichiometric mixture of Co (99.99%, Strem Chemicals), Ta (99.98%, Strem Chemicals) and S (99.9995%, Alfa Aesar) powders. The mixture was loaded in an alumina crucible, which was then sealed in an evacuated quartz tube. The tube was slowly heated up to 900 °C and kept for one day. Subsequently, single crystals were grown by a chemical vapor transport process using I2 (approximately 0.05 g) as the transport agent. The temperature of the source (sink) was 950 °C (850 °C). Crystals with lateral dimensions up to 2 mm were obtained after being kept at the above temperatures for about one week. The phase purity of the crystals was checked by powder X-ray diffraction, and the orientation of the single crystalline samples was checked by a Laue diffractometer. Energy Dispersive X-ray spectroscopy was used to characterize the resulting stoichiometry of the single crystals. It was measured on the ZEISS 1550VP field emission SEM with Oxford X-Max SDD X-ray Energy Dispersive Spectrometer (EDS) system. Magnetization was measured by using a Vibrating-sample magnetometer (VSM) option implemented on a commercial superconducting magnet (Quantum Design DynaCool PPMS). Transport measurement under strain Electrical transport properties were measured using both standard four-(six-) wire methods and modified Montgomery method (See dedicated section below) in commercial cryostats with a superconducting magnet. Strain was applied to single crystals of CoTa3S6 using a commercial piezoelectric uniaxial strain cell (CS-100, Razorbill Instruments). Polished bar-shape samples with typical dimension 1.5 × 0.5 × 0.1 mm3 were mounted to the strain cell with Stycast 2850FT epoxy. For data shown in Fig. 2, the center part of the sample between the mounting plates is approximately 0.7 mm – we assume this to be the length L of the strained region of the sample, and it is used to estimate the applied strain through ϵxx = ΔL/L where ΔL is obtained from the capacitive displacement sensor in the strain cell. A home-made probe was used to load the strain cell in the cryostat. The piezoelectric stacks were controlled by a combination of the DC output of Keithley 6221 current source and TEGAM 2350 high voltage amplifier. Longitudinal and Hall resistivities were measured using external lock-in amplifiers (Stanford Research Systems, SR860) with an AC current of typical amplitude 2 mA from a voltage-controlled current source (Stanford Research Systems, CS580). Modified montgomery method The modified Montgomery method is an effective technique for measuring the anisotropic resistivity in orthorgonal directions within a single sample simultaneously68–70. In this study, CoTa3S6 is polished into a thin, nearly square shape with its edges aligned along the x- and y-axes, respectively (see Fig. S8 inset for a photo of the Montgomery sample). Four electrodes were attached at the corners of the sample with relative small contact area to ensure precise measurements. The resistance Rxx (Ryy) was determined by applying current Ix (Iy) along the x-(y-) axis on one side and measuring the voltage drop Vx (Vy) on the opposite side; and Rxx = Vx/Ix (Ryy = Vy/Iy). The resistivities ρxx and ρyy were calculated based on the ratio of Rxx and Ryy (see SI Sec. IV for details). ρxx(ρyy) presented in Figs. 1,3 and 4 are obtained using this method. Polarization rotation measurement The schematic of the optical setup can be found in SI Sec. II. A linearly polarized beam is generated by passing light through a polarizer and a polarization-independent beam splitter (half mirror), which transmits half of the beam through a pinhole and discards the other half. The transmitted beam then passes through a half-wave plate (HWP), mechanically rotated such that its fast axis is at an angle α/2 to the initial polarization direction. This results in a polarization rotation of angle α. The beam passes through the optical window and reflects from the sample. Upon return, it passes through the HWP again, inducing an additional rotation by − α, leaving the polarization direction rotated by θT in total. The beam then encounters the beam splitter again, which directs part of the light to a Wollaston prism. The Wollaston prism, rotated at π/4, separates the beam into two orthogonal polarization components (s and p), which are detected by two balanced detectors (P1 and P2). For a gold mirror calibration sample, P1 and P2 are balanced, i.e., ΔP = P1 − P2 ≈ 0. The optical field amplitudes at the two detectors are: 3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{1}={E}_{0}\cos \left(\frac{\pi }{4}-{\theta }_{T}\right),\quad {E}_{2}={E}_{0}\cos \left(\frac{\pi }{4}+{\theta }_{T}\right)$$\end{document}E1=E0cosπ4−θT,E2=E0cosπ4+θT Let E0 be the total field amplitude. The total and differential intensities are: 4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {I}_{1}+{I}_{2}={E}_{1}^{2}+{E}_{2}^{2}={E}_{0}^{2} \\ {I}_{1}-{I}_{2}={E}_{1}^{2}-{E}_{2}^{2}={E}_{0}^{2} \sin (2{\theta}_{T})$$\end{document}I1+I2=E12+E22=E02I1−I2=E12−E22=E02sin(2θT) Therefore, the normalized signal becomes: 5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{I}_{1}-{I}_{2}}{{I}_{1}+{I}_{2}}=\sin (2{\theta }_{T})$$\end{document}I1−I2I1+I2=sin(2θT) Since optical power P ∝ I, we can write: 6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{rcl}{\theta }_{T}=\frac{1}{2}\arcsin \left(\frac{{I}_{1}-{I}_{2}}{{I}_{1}+{I}_{2}}\right)&=&\frac{1}{2}\arcsin \left(\frac{{P}_{1}-{P}_{2}}{{P}_{1}+{P}_{2}}\right)\\ &=&\frac{1}{2}\arcsin \left(\frac{\Delta P}{{P}_{1}+{P}_{2}}\right)\end{array}$$\end{document}θT=12arcsinI1−I2I1+I2=12arcsinP1−P2P1+P2=12arcsinΔPP1+P2 As the θT is a function of α, the polar plot can be fit with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\theta }_{T}={\theta }_{P}\sin [2(\alpha -{\alpha }_{0})]$$\end{document}θT=θPsin[2(α−α0)]. α0 defines the orientation of the slow axis, which corresponds to the direction of the nematic director. Comparing in-plane strain and magnetic fields as domain polarization forces Due to a linear coupling between in-plane strain and the nematic order, an orthorhombic distortion proportional to the nematic order parameter (e.g., ϵxx − ϵyy for domain A) is expected for each domain. As a result, in-plane strain-dependence of the nematic order parameter and resistivity anisotropy should be non-hysteretic, and a linear proportionality is expected between the two in a multi-domain state49. In contrast, when using the conjugate thermodynamic variable to strain, in-plane uniaxial stress (e.g., σxx − σyy) to tune the nematic domains, one may exhibit hysteresis analogous to M(H) curves in ferromagnets. The thermodynamic condition in our experiment is such that strain ϵxx is the control parameter; therefore assessing a potentially nonvolatile stress-tuning of the nematic domains is beyond the present experiments. Meanwhile, in-plane magnetic field, itself unconstrained by the nematic order parameters, can exhibit memory effects in the polarization process. Resistivity anisotropy under out-of-plane field The magnetic field dependence of the resistivity anisotropy for H∥z in Fig. 4a was measured in the following sequence. The sample was mounted on a probe with rotation capability and field-cooled by H∥x at 9 T from T = 50 K to 2 K, and then the field was set to zero. After this, the sample platform is slowly rotated to the H∥z configuration at fixed temperature. During this, the effect of magnetic field should be negligible. The system is then warmed up to the target temperature; after waiting for temperature stability, an out-of-plane field is applied from 0 to 9 T where ρxx and ρyy are measured simultaneously. Phenomenological modeling of phases in CoTa3S6 The phenomenological free energy of the system is given by Eq. (1) and Eq. (2). In Fig. 4c, the collinear AFM order parameter is chosen to be ∣Li∣ when the other two ∣Lj∣, j ≠ i are zero, and set to zero when at least two of the ∣Lj∣, j = 1, 2, 3 are non-zero. The non-coplanar AFM order parameter is defined as ∣(L1 × L2) ⋅ L3∣, which is proportional to the scalar spin chirality of the non-coplanar spin configuration (see SI Sec. III. A for details). For the nematic order parameter, we are plotting n. The values of the parameters are as follows: for the nematic order, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$b=\tilde{b}(T-{T}_{{{\rm{n}}}}),\tilde{b}=0.3,c=-0.05,w=5$$\end{document}b=b~(T−Tn),b~=0.3,c=−0.05,w=5, and Tn = 30; for the magnetic orders, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${v}_{0}=5,\tilde{a}=1,{\tilde{v}}_{1}=1$$\end{document}v0=5,a~=1,v~1=1, and TN1 = 37, TN2 = 25. For the coupling between the magnetic orders and nematic order, we take g = 0.05. Supplementary information Supplementary Information Transparent Peer Review file Supplementary Information Transparent Peer Review file