betaFeedback

WorksPhysics and Astronomy

Magnetic frustration in a van der Waals metal CeSiI

Abstract

The realization of magnetic frustration in a metallic van der Waals (vdW) coupled material has been sought as a promising platform to explore novel phenomena both in bulk matter and in exfoliated devices. However, a suitable material platform has been lacking so far. Here, we demonstrate that CeSiI hosts itinerant electrons coexisting with exotic magnetism. In CeSiI, the magnetic cerium atoms form a triangular bilayer structure sandwiched by van der Waals stacked iodine layers. From resistivity and magnetometry measurements, we confirm the coexistence of itinerant electrons with magnetism with dominant antiferromagnetic exchange between the strongly Ising-like Ce moments below 7 K. Neutron diffraction directly confirms magnetic order with an incommensurate propagation vector k ~ (0.28, 0, 0.19) at 1.6 K, which points to the importance of further neighbor magnetic interactions in this system. The presence of a two-step magnetic-field-induced phase transition along c axis further suggests magnetic frustration in the ground state. Our findings provide a novel material platform hosting a coexistence of itinerant electron and frustrated magnetism in a vdW system, where exotic phenomena arising from rich interplay between spin, charge and lattice in low dimension can be explored.

The exploration of novel quantum phenomena in van der Waals (vdW) coupled materials is a promising and rapidly growing field motivated by both fundamental physics interest and, thanks to recent advances in exfoliation technologies, device applications[1, 2, 3, 4]. An essential component to accelerate the development of the field is to increase the number of materials that host states whose properties are described by multiple coupled degrees of freedom, such as charge, spin, orbital, and lattice[5]. Among these, a particularly interesting playground for discovering such emergent phenomena has been identified in systems which manifest an interplay between itinerant electrons and magnetism, and intensive studies have been pursued and reported[6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. However, a promising area of this playground which simultaneously remains relatively unexplored is that of vdW materials, where itinerant electrons coexist with low-dimensional and potentially frustrated magnetism. Magnetic frustration can result in non-collinear spin textures and field-induced transitions to exotic quantum phases[18, 19]. This includes emergent topologically non-trivial spin textures which lead to exotic couplings between the magnetism and the motion of itinerant electrons[20, 21]. Furthermore, non-collinear spin textures can couple to lattice degrees of freedom to produce phenomena like multiferroicity[22].

Part of the reason why magnetic metallic vdW materials are so scarce is that most vdW materials studied so far have been chalcogenides, which tend to be nonmagnetic. Another important class of materials with vdW structures is the halides. The large ionic radius and small number of chemical bonds of halides leads to low dimensional vdW bonded structures[23]. While typical transition metal halides[24, 25, 26, 27] are low-dimensional Mott-insulators, this is not the case for reduced rare-earth halides LnX2LnX_2. In rare-earth elements with a strong tendency to stabilize the trivalent state, an electron produced by reduction occupies the outer 5d5d orbital of lanthanide rather than inner 4f4f orbital and metallicity results[28, 29, 30]. Moreover, the finding of giant magnetoresistance and ferromagnetic order above room temperature in GdI2_2 motivates us to expect that 4f4f magnetism strongly couples to conduction electrons in rare earth halide materials[31, 32].

The ternary system LnAILnAI (Ln=La,Ce,Pr,GdLn = \mathrm{La, Ce, Pr, Gd}. A=Al,Si,Ga,GeA = \mathrm{Al, Si, Ga, Ge}) is a class of materials closely related to the reduced rare earth halides[33, 34, 35]. CeSiI appears as a cleavable material in 2DMatpedia[36], and its crystal structure is depicted in Figure 1a. Ce atoms form a triangular bilayer in the abab plane sandwiching a honeycomb net of Si. These blocks are terminated by a vdW-coupled layer of iodine. While its synthesis and crystal structure have been reported previously, its physical properties, especially the effect of geometrical frustration and the influence of the Fermi surface on magnetism, have never been investigated. In order to demonstrate CeSiI as a promising host of itinerant electrons with frustrated magnetism, we have synthesized crystalline CeSiI by a high-temperature-solid-state reaction[33] in a bulk form (Figure 1b) and performed resistivity, magnetometry, heat capacity, and neutron diffraction measurements.

Image not extracted

Figure 1. a, Crystal structure of CeSiI. Red, blue, and purple spheres indicate Ce, silicon, and iodine atoms, respectively. Red, green, and blue arrows indicate the aa, bb, and cc axes, respectively. b, Optical microscope image of a single crystal of CeSiI. The image is taken in a glove box filled with pure Ar gas. c, Temperature dependence of magnetic susceptibilities from a single crystal of CeSiI and zero-field resistivity of a polycrystalline sample of CeSiI. The magnetization measurements were carried out upon cooling in a magnetic field of 1 T applied along the aa and cc axes.

The coexistence of metallicity and antiferromagnetic interactions in a CeSiI is confirmed. Figure 1c shows the temperature-dependence of the susceptibility χ(T)\chi(T) along the aa and cc axes (left axis), together with the temperature dependence of the resistivity (right axis). χ(T)\chi(T) exhibits strongly anisotropic behavior below 100 K due to the trigonal crystalline electric field splitting of the J=5/2J = 5/2 manifold of the Ce3+^{3+} 4f4f electrons. The lowest Kramers doublet dominates the magnetism below ∼\sim10 K because the other J=5/2J = 5/2 levels are typically located above 10210^{2}–10310^{3} K in the trigonal crystalline electric field[37]. Below 7 K, the susceptibility component along the cc axis drops, while that along aa axis slightly increases, which indicates long-range antiferromagnetic ordering at TN∼7T_{\mathrm{N}} \sim7 K. The presence of a magnetic phase transition is also confirmed by the drop of resistivity below 7 K due to suppression of electron-spin scattering.

instrument at ILL. Whereas single crystal neutron diffraction would be ideal for pinning down the magnetic structure, the required sample volume is difficult to obtain. Nevertheless, as we describe hereafter, powder diffraction still provides essential information to understand this relatively new compound. In Figure 2a, we plot the temperature dependence of the low-angle magnetic scattering of CeSiI obtained by subtracting a high-temperature intensity from each 2θ2\theta. Six magnetic peaks, which are labelled by θ1∼θ6\theta_1 \sim\theta_6 in Figure 2a, are visible below 40° (see Supplementary Figure S1 for the nuclear refinement). As shown in Figure 2b, the positions of these peaks are temperature-independent, while their intensity gradually increases below TNT_{\mathrm{N}}, which is consistent with the presence of a second order phase transition. The quality of the data did not allow for a determination of the critical exponent, however. The increase in the intensity of θ5\theta_5 above TNT_{\mathrm{N}} is affected by a ferromagnetic impurity CeSi1.7_{1.7} with Tc∼11KT_{\mathrm{c}} \sim11\mathrm{K}[38]. We note that the change of background below TNT_{\mathrm{N}} is due to suppression of diffuse scattering by paramagnetic Ce3+^{3+}. We employed a long scan of the 1.6K data subtracted by that of the 7.7 K data for the magnetic structure analysis in order to disentangle the effect of diffuse scattering and magnetic Bragg peak.

Image not extracted

Figure 2. Neutron diffraction study of CeSiI. (a) Temperature dependence of magnetic diffraction in the temperature range from 16 to 3.8 K. Nuclear reflections were removed from the data by subtracting data obtained at 17 K. The positions of the six observed magnetic Bragg peaks are labeled θi\theta_i (i=1,…,6i = 1, \ldots, 6). The color scales of the neutron intensity are shown on the right side. (b) The evolution of the integrated intensity around the magnetic Bragg peaks derived from (a). The temperature is scaled by TN=7T_{\mathrm{N}} = 7 K. The integration intervals for θi\theta_i (i=1,…,6i = 1, \ldots, 6) are (9.96, 10.59), (13, 14.8), (27.31, 28.15), (28.81, 30.00), (34.13, 35.42), and (35.76, 36.7), respectively. A linear background was assumed in integration. In order to elucidate the magnetic structure, powder neutron diffraction was performed below 20 K on the D20

Image not extracted

Figure S1. Powder neutron profile of CeSiI at 19.4 K fitted by the Rietveld method, showing observed (black circle), calculated (red line), and difference (blue line). The green ticks represent the positions of the Bragg reflections of CeSiI and impurity phases: CeI3_3, CeOI, and CeSi1.7_{1.7}.

Indexing the magnetic peaks produced a unique solution with an incommensurate wavevector of k∼(0.28,0,0.19)\mathbf{k} \sim(0.28, 0, 0.19). In the space group P-3m1P\mathrm{\text{-}}3m1, the identity and mirror perpendicular to bb axis render the propagation vector invariant, and these make up the little group GkG_{\mathbf{k}}. The magnetic representation of a crystallographic site at 2c2c site (0,0,z)(0, 0, z) can be decomposed into the irreducible representation Γ1+2Γ2\Gamma_1 + 2\Gamma_2, for which the projected basis vectors are listed in Table 1. While the Ce site is separated into two orbits (0,0,z)(0, 0, z) and (0,0,−z)(0, 0, -z) in GkG_{\mathbf{k}}, the symmetry operation that maps k\mathbf{k} to −k-\mathbf{k} in the full group relates these two; the real and imaginary coefficients of the basis vectors have opposite and same values between the two orbits, respectively. Inversion and two-fold rotation about the bb axis are added to GkG_{\mathbf{k}} in the full group that also includes time reversal (Supplementary Note 2).

IRBVmxm_xmym_ymzm_z
Γ1\Gamma_1ψ1\psi_1010
Γ2\Gamma_2ψ2\psi_2100
ψ3\psi_3001

Table 1. Irreducible representations (IRs) and basis vectors (BVs) for the space group P-3m1P\mathrm{\text{-}}3m1 with k=(0.28,0,0.19)\mathbf{k} = (0.28, 0, 0.19). The decomposition of the magnetic structure representation for the 2c2c site (0,0,z)(0, 0, z) is Γmag=Γ1+2Γ2\Gamma_{\mathrm{mag}} = \Gamma_1 + 2\Gamma_2. The BV components of one orbit along a∗a^*, bb, and cc are shown by mxm_x, mym_y, and mzm_z, respectively. The other orbit (0,0,−z)(0, 0, -z) has same BVs.

To determine the magnetic structure below TNT_{\mathrm{N}}, we refined the neutron diffraction pattern using all possible magnetic structures described by a single irreducible representation: SDW along the bb axis (SDWb_b), which belongs to Γ1\Gamma_1 (in BasIreps notation), and SDW (SDWa∗c_{a^*c}) and counter-rotating cycloid (Cycloid1_1) in the a∗ca^*c plane, which belong to Γ2\Gamma_2. Cycloid1_1 has opposite chiralities between the two orbits to maintain the inversion symmetry. We also consider a structure that breaks sublattice symmetry but is an element of the little group and is more common than the counter-rotating cycloid; a co-rotating cycloid (Cycloid2_2) in the a∗ca^*c plane. In all cases, impurity phase CeSi1.7_{1.7} is included in all the refinement and we confirmed that the existence of the secondary phase does not have any impact on the main conclusions of this study. Further details of the magnetic structures and fits are described in Supplementary Note 3. We plot the results of the Rietveld refinement in Figure 3a-Figure 3d and Table 2. As shown in Figure 3a, SDWb_b is clearly excluded from the candidate magnetic structure because of the poor agreement around the (010)−(010)-, (011)−(011)-, (011)+(011)+, and (011ˉ)+(01\bar{1})+ reflections. By contrast, SDWa∗c_{a^*c}, Cycloid1_1, and Cycloid2_2 reproduce the overall behavior of the observed diffraction pattern as indicated by Figure 3b-Figure 3d. Thus, based on diffraction experiments, we confirm that a common feature of these three magnetic structures, all of which belong to Γ2\Gamma_2, are a dominant out-of-plane component of the order of 1.01.0-1.1 μB1.1\,\mu_{\mathrm{B}}. Two of the magnetic structures that yielded the smallest RR factors are schematically shown in Figure 4a and Figure 4b as representatives of possible magnetic structures with and without relatively large in-plane magnetic moment.

ModelIRmx,1m_{x,1}my,1m_{y,1}mz,1m_{z,1}mx,2m_{x,2}my,2m_{y,2}mz,2m_{z,2}RpR_{\mathrm{p}} (%)RwpR_{\mathrm{wp}} (%)χ2\chi^2
SDWb_bΓ1\Gamma_101.00(1)001.00(1)039.228.86.15
SDWa∗c_{a^*c}Γ2\Gamma_20.14(1)01.08(1)0.14(1)01.08(1)26.918.22.59
Cycloid1_1Γ2\Gamma_20.07(2)ii01.11(1)0.07(2)ii0−1.11(1)-1.11(1)26.718.62.76
Cycloid2_2Γ2\Gamma_20.71(2)ii01.024(9)0.71(2)ii01.024(9)21.815.92.26

Table 2. Results of Rietveld refinement of neutron diffraction pattern. The magnetic structure models are described in the main text. The mi,jm_{i,j} represents the component of jj site along direction (i=x,y,zi = x, y, z and j=1,2j = 1, 2). RpR_{\mathrm{p}}, RwpR_{\mathrm{wp}}, and χ2\chi^2 represent profile factor, weighted profile factor, and reduced chi square, respectively.

Image not extracted

Figure 3. Powder neutron diffraction pattern of CeSiI and a Rietveld fitting to the magnetic spin structures Γ1\Gamma_1 SDWb_b (a), Γ2\Gamma_2 SDWa∗c_{a^*c} (b), Γ2\Gamma_2 Cycloid1_1 (c), and Γ2\Gamma_2 Cycloid2_2 (d). The red circles represent data taken at 1.6 K after the subtraction of the 7.7 K data as a reference of nuclear contributions. The black, blue, and green line represent a Rietveld fit, a residual of fitting, and background, respectively. The red thin and thick green bar represents position of magnetic Bragg peaks of CeSiI and CeSi1.7_{1.7}, respectively. The arrows indicate the indices of magnetic Bragg peaks, where (hkl)±(hkl)_{\pm} represents a Miller index of (hkl)±(0.28,0,0.19)(hkl) \pm(0.28, 0, 0.19) reflection.

Image not extracted

Figure 4. Magnetic frustration in CeSiI (a,b) Magnetic structures are schematically drawn for the case of Cycloid2 and SDWa∗c_{a*c}. Red spheres and magenta arrows indicate the magnetic Ce ions and their magnetic moments, respectively. Red, green, and blue arrows indicate the aa, bb, and cc axes, respectively. (c) Magnetization process of CeSiI. The red, green, and blue circles and red and blue triangles represent magnetization along cc axis at 2, 4, and 7 K and along abab plane at 2 and 7 K, respectively. The red dotted lines are derivative of magnetization at 2K, where the peaks indicate the two-step metamagnetic transition.

Finally, we point out that magnetization process is consistent with the presence of magnetic frustration. Figure 4c shows magnetization process along out-of-plane (//c//c) and in-plane (//ab//ab) direction. Above 4 T, the magnetization along the cc axis reaches ∼1.1 μB\sim1.1\,\mu_{\mathrm{B}}, which is close to the expected saturated magnetization of SDWa∗c_{a^*c} and Cycloid2_2 model and suggests localized nature of the Ce 4f moment. The magnetization along the aa axis does not saturate below 7 T because this is the hard axis. The key feature is seen at 2 K for out-of-plane (//c) case. Between magnetic fields of 2 T and 4 T, two metamagnetic transitions are clearly observed. These anomalies are visible only below TNT_{\mathrm{N}} and are therefore thought to be associated with the magnetic order. It is important to note that the anomalous behavior should be distinct from a simple spin flop transition in a collinear antiferromagnet, which is usually a single-step process. A similar magnetization process with multiple steps is seen in layered triangular lattice antiferromagnets in insulating systems[39, 40, 41, 42, 43], which include MnI2_2, NiI2_2, and CoI2_2. In these compounds, magnetic frustration is supposed to occur due to competing nearest neighbor and further neighbor magnetic coupling, leading to a helical order with in-plane periodicity of (a,0)(a, 0). The magnetic structure in these insulating iodides is analogous to Cycloid2 case (see table 2) and competing further neighbor couplings between localized spins are naturally present in metallic systems. As far as we recognize, the observation of multiple magnetization process in a vdW metal is the first report in CeSiI.

Our experimental findings collectively point towards CeSiI being a novel material platform hosting magnetic frustration coexisting with itinerant electrons within a vdW material family. In the case of the out-of-plane SDW (Figure 4b), the realization of a quantum disordered state by frustration and reduced dimensionality will be an interesting question to be explored with the help of the exfoliation technique. Further exotic phases can be expected in the case of the co-rotating cycloid case (Figure 4a). For example, the point group of this structure is m1’, which allows a finite electric polarization perpendicular to bb axis. Even if polarization is allowed by symmetry, the conduction electrons are expected to screen polarization in metallic compounds. However, in the van der Waals metal WTe2_2, which features a polar-nonpolar structural transition, an out-of-plane electric field can switch the displacement of W when atomically-thin flakes are used[44].

In summary, we investigated CeSiI using resistivity, magnetometry measurements and neutron diffraction experiments on both powder and single crystal samples. Through these, we demonstrated that CeSiI is a novel material platform to search exotic phenomena arisen from rich interplay between spin, charge and lattice degree of freedom, for example, by using exfoliation-based device fabrication techniques.

Acknowledgement

We thank Takeshi Yajima, and Hajime Ishikawa for advice on the synthesis of CeSiI, Dmitry Khalyavin for help with the determination of the magnetic structure, and Rieko Ishii, Daichi Ueta, and Wonjong Lee for helping the experiments. This work was carried out by the joint research in the Institute for Solid State Physics, the University of Tokyo.

Method

Single crystal growth

Single crystals of CeSiI were synthesized by a high-temperature-solid-state reaction following ref. 33. The handling of the elements and compounds was performed in an argon-filled glovebox. CeI3_3 was made by reaction of stoichiometric amounts of each element: Ce and I2_2 were sealed in an evacuated quartz tube in vacuo and reacted at 300°C for 24 hours followed by subsequent heating at 900°C for 24 hours to complete the reaction. The quartz tube was then placed vertically with one end outside the furnace and heated again at 900°C in order to facilitate the sublimation of polycrystalline CeI3\mathrm{CeI_3}. Yellow needle-like crystals of CeI3\mathrm{CeI_3} were formed at the colder end of the quartz tube. Stoichiometric amounts of CeI3\mathrm{CeI_3}, Ce, and Si in a total weight of 1 g were sealed in a Nb tube with a diameter of 1 cm and length of 15 cm by arc welding. The sealed Nb tube was sealed in a quartz tube to prevent oxidation of Nb at elevated temperatures. The sealed quartz tube was then placed horizontally in a two-zone furnace. The cold and hot ends were kept 800°C and 1000°C, respectively. After a week, shiny bronze plate-like crystals with a maximum dimension of 2×2×0.05 mm32 \times2 \times0.05\ \mathrm{mm^3} formed at the hot end. The synthesis of powder CeSiI\mathrm{CeSiI} was performed at 1000°C for four days.

X-ray diffraction, magnetometry, heat capacity, and resistivity measurement

Single crystals of CeSiI\mathrm{CeSiI} could be indexed by a trigonal space group P3ˉm1P\bar{3}m1 with lattice constants of a=4.1872(1)a = 4.1872(1) Å and c=11.7036(4)c = 11.7036(4) Å at 300 K using a single crystal X-ray diffractometer Bruker Venture D8 (Mo Kα, λ=0.71069\lambda= 0.71069 Å). DC SQUID Magnetometry measurements were performed using MPMS-3 (Quantum Design). A single crystal was placed on a diamagnetic sapphire substrate and coated with paraffin. Resistivity was measured by conventional four-probe method using the resistivity option of PPMS (Quantum Design). An as-grown polycrystalline chunk of CeSiI\mathrm{CeSiI} was attached to indium electrodes and gold wires.

Powder Neutron Diffraction experiment and data analysis.

Powder neutron diffraction of was performed on the high-flux D20 thermal diffractometer at Institut Laue-Langevin at temperatures from 20 K to 1.6 K using a constant wavelength λ1∼2.4\lambda_1 \sim2.4 Å[45]. The powder sample was sealed in a vanadium can with an indium seal to prevent oxidation. The magnetic peaks were indexed by the kk-search software in the FULLPROF suite[46]. The analysis of the irreducible representations is performed using SARAh suite software (little group)[47] and ISODISTORT software (full group)[48, 49]. Structural and magnetic refinement of powder diffraction data was performed by Fullprof[46].

Data availability

The data files are also available from the corresponding author upon request.

Supporting information for “Magnetic frustration in a van der Waals metal CeSiI”

Ryutaro Okuma; Clemens Ritter; Gøran J. Nilsen; Yoshinori Okada

Nuclear structure refinement of powder neutron diffraction data {#sec:Supplementary Note 1}

We show here that neutron analysis for drawing main conclusion is not altered due to impurity phases in our samples. Figure S1 shows the result of nuclear Rietveld refinement in Fullprof[1] using data at 19.4 K considering four phases according to reported structural data. It successfully converged to Rwp=8.76%R_{\mathrm{wp}} = 8.76\% and Rp=7.45%R_{\mathrm{p}} = 7.45\%. The refined value of molar ratio of the sample was CeSiI : CeI3_3 : CeSi1.7_{1.7} : CeOI = 56.4(6) : 25.3(4) : 14.4(3) : 3.9(1). As CeI3_3 and CeOI are paramagnetic above 2 K and CeSi1.7_{1.7} is ferromagnetic below ∼11\sim11 K[2], magnetic scattering away from nuclear scattering should originate purely from magnetic moments on CeSiI.

FormulaCeSiI
Space groupPP-3m13m1
aa / Å4.1697(1)
cc / Å11.6581(4)
VV / Å3^3175.537(9)
ZZ2
AtomWyck.xxyyzzOcc.
Ce12c2c000.1751(6)1
I12d2d1/32/30.3548(4)1
Si12d2d1/32/30.9924(5)1

Table 6.

FormulaCeI3_3
Space groupCmcmCmcm
aa / Å4.3636(3)
bb / Å13.917(1)
cc / Å9.9333(7)
VV / Å3^3603.23(8)
ZZ4
AtomWyck.xxyyzzOcc.
Ce14c4c03/41/41
I14c4c00.0803(7)1/41
I28f8f00.3541(5)0.0700(7)1

Table 7.

FormulaCeSi1.7_{1.7}
Space groupImmaImma
aa / Å4.1195(4)
bb / Å4.1835(4)
cc / Å13.883(1)
VV / Å3^{3}239.26(4)
ZZ4
AtomWyck.xxyyzzOcc.
Ce14e01/40.124(1)1
Si14e01/40.533(1)0.85
Si24e01/40.702(1)0.85
FormulaCeOI
Space groupP4/nmmP4/nmm
aa / Å4.1051(5)
cc / Å9.100(2)
VV / Å3^{3}153.35(5)
ZZ2
AtomWyck.xxyyzzOcc.
Ce12c1/41/40.1311
O12a3/41/401
I12c1/41/40.6721

Supplementary Table 1. Structural parameters of CeSiI, CeSi1.7_{1.7}, CeI3_3, CeOI, and obtained from the Rietveld refinements. Isotropic displacement parameter BisoB_{\mathrm{iso}} is fixed to 0.5/Å2^{2} for all atoms.

Symmetry analysis of the magnetic structure

Details of the symmetry analysis of the magnetic structure is given in this section. CeSiI crystallizes in a trigonal space group P3‾m1P\overline{3}m1 with lattice constants a=4.1697(1)a = 4.1697(1) Å and c=11.6581(4)c = 11.6581(4) Å at 20 K and Ce ions occupy single Wyckoff position 2c as described in the Supplementary Note 1. To comply with the symmetry of the single-kk magnetic structure described by k=0.28a∗+0.2c∗k = 0.28a^{*} + 0.2c^{*}, we use (a∗ b c)(a^{*}\ b\ c) as principal axes in the following discussion. The magnetic moment at the position r=(x,y,z)\mathbf{r} = (x, y, z) can be expressed as

S(r)=eik⋅r(mx,my,mz)+e−ik⋅r(mx∗,my∗,mz∗).\mathbf{S}(\mathbf{r}) = e^{i\mathbf{k}\cdot\mathbf{r}}(m_x, m_y, m_z) + e^{-i\mathbf{k}\cdot\mathbf{r}}(m_x^{*}, m_y^{*}, m_z^{*}).

Here mxm_x, mym_y, mzm_z, and * correspond to the component along a∗a^{*}, bb, and cc and a complex conjugate operator, respectively. Symmetry of the magnetic structure can be characterized by the symmetry operations in P3‾m11′P\overline{3}m1 1^{\prime} that transform k⋅r\mathbf{k}\cdot\mathbf{r} to ±k⋅r\pm\mathbf{k}\cdot\mathbf{r}, namely,

1′:(x,y,z,mx,my,mz)→(x,y,z,−mx,−my,−mz),mb:(x,y,z,mx,my,mz)→(x,−y,z,−mx,my,−mz),2b:(x,y,z,mx,my,mz)→(−x,y,−z,−mx∗,my∗,−mz∗).\begin{aligned} 1^{\prime}: &(x,y,z,m_x,m_y,m_z) \to(x,y,z,-m_x,-m_y,-m_z),\\ m_b: &(x,y,z,m_x,m_y,m_z) \to(x,-y,z,-m_x,m_y,-m_z),\\ 2_b: &(x,y,z,m_x,m_y,m_z) \to(-x,y,-z,-m_x^{*},m_y^{*},-m_z^{*}). \end{aligned}

where 1′1^{\prime}, mbm_b, and 2b2_b describe time-reversal operation, mirror operation perpendicular to the bb axis, and two-fold rotation about the bb axis, respectively. Superspace groups of the candidate magnetic structures are obtained by ISODISTORT[3, 4] as shown in Supplementary Table 2 and Figure 2.

IRSuperspace groupMagnetic point groupS+S_{+}S−S_{-}Types of magnetic structures
Γ1\Gamma_{1}C2/m1′(α,0,β)00sC2/m1^{\prime}(\alpha,0,\beta)00s2/m1′2/m1^{\prime}(0,m,0)(0,m,0)(0,m∗,0)(0,m^{*},0)SDW // bb axis
Cm1′(α,0,β)0sCm1^{\prime}(\alpha,0,\beta)0sm1′m1^{\prime}(0,m,0)(0,m,0)(0,u,0)(0,u,0)SDW // bb axis
Γ2\Gamma_{2}C2/m1′(α,0,β)0ssC2/m1^{\prime}(\alpha,0,\beta)0ss2/m1′2/m1^{\prime}(m,0,u)(m,0,u)−(m∗,0,u∗)-(m^{*},0,u^{*})SDW and counter-rotating cycloid // a∗ca^{*}c
Cm1′(α,0,β)ssCm1^{\prime}(\alpha,0,\beta)ssm1′m1^{\prime}(m,0,u)(m,0,u)(v,0,w)(v,0,w)SDW and co-rotating cycloid // a∗ca^{*}c

Supplementary Table 2. Superspace groups of single-kk magnetic structures with the incommensurate propagation vector k=0.28a∗+0.2c∗k = 0.28a^{*} + 0.2c^{*}. S±S_{\pm} denotes the allowed (mx,my,mz)(m_x,m_y,m_z) at (0,0,±z)(0,0,\pm z) in the given superspace group and mm, uu, vv, and ww are complex numbers. The ss in superspace groups denotes the shift of the internal coordinate by a factor of π\pi.

In C2/m1′(α,0,β)00sC2/m1^{\prime}(\alpha,0,\beta)00s and C2/m1′(α,0,β)0ssC2/m1^{\prime}(\alpha,0,\beta)0ss, two-fold rotation related the two orbits inside a unit cell and inversion symmetry is retained whereas in Cm1′(α,0,β)0sCm1^{\prime}(\alpha,0,\beta)0s and Cm1′(α,0,β)ssCm1^{\prime}(\alpha,0,\beta)ss, two orbits are not related by symmetry and only mirror symmetry is present as shown in Supplementary Figure 2. In terms of types of magnetic structure, C2/m1′(α,0,β)00sC2/m1^{\prime}(\alpha,0,\beta)00s allows only a collinear spin density wave order parallel to bb axis and both orbits have same amplitude. Cm1′(α,0,β)0sCm1^{\prime}(\alpha,0,\beta)0s also allows only a collinear SDW but the amplitude between the sites can be different. C2/m1′(α,0,β)0ssC2/m1^{\prime}(\alpha,0,\beta)0ss in general describes elliptic counter rotating cycloids in the sense that the chirality defined as b⋅Sr×Sr+a\mathbf{b}\cdot\mathbf{S}_{r}\times\mathbf{S}_{r+a} takes the opposite sign and same amplitude between the two orbits, which includes SDW parallel to the bb axis. Cm1′(α,0,β)ssCm1^{\prime}(\alpha,0,\beta)ss can describe more general elliptic cycloids including co-rotating cycloids.

Image not extracted

Supplementary Figure 3. Typical magnetic structure allowed in the superspace group given in Supplementary Table 2. The red sphere and purple and green arrows represent Ce ions and magnetic moments with different sites, respectively. Sky-blue bond and shaded plane indicate two-fold rotation axis and mirror plane, respectively. The sign indicates phase shift of the internal coordinate; it is multiplied to the magnetic moment after of the symmetry operation is performed to keep the magnetic structure unchanged. SDW, counter-rotating cycloid, and co-rotating cycloids are sketched as examples for C2/m1′(α,0,β)00sC2/m1'(\alpha, 0, \beta)00s and Cm1′(α,0,β)0sCm1'(\alpha, 0, \beta)0s, C2/m1′(α,0,β)00sC2/m1'(\alpha, 0, \beta)00s, and Cm1′(α,0,β)ssCm1'(\alpha, 0, \beta)ss, respectively.

Magnetic structure refinement

To construct physical models of the magnetic structure in CeSiI, we considered typical magnetic structures allowed in the superspace groups of Supplementary Table 2, which is discussed in the main text. Each parameter used to describe the magnetic structures is shown in Supplementary Table 3.

ModelSuperspace groupS+S_{+}S−S_{-}
SDWb_bC2/m1′(α,0,β)00sC2/m1'(\alpha, 0, \beta)00se−iφ/2(0,My,0)e^{-i\varphi/2}(0, M_y, 0)eiφ/2(0,My,0)e^{i\varphi/2}(0, M_y, 0)
SDWa∗c_{a*c}C2/m1′(α,0,β)0ssC2/m1'(\alpha, 0, \beta)0sse−iφ/2(Mx,0,Mz)e^{-i\varphi/2}(M_x, 0, M_z)eiφ/2(Mx,0,Mz)e^{i\varphi/2}(M_x, 0, M_z)
Cycloid1_1C2/m1′(α,0,β)0ssC2/m1'(\alpha, 0, \beta)0sse−iφ/2(iMx,0,Mz)e^{-i\varphi/2}(iM_x, 0, M_z)eiφ/2(iMx,0,−Mz)e^{i\varphi/2}(iM_x, 0, -M_z)
Cycloid2_2Cm1′(α,0,β)ssCm1'(\alpha, 0, \beta)sse−iφ/2(iMx,0,Mz)e^{-i\varphi/2}(iM_x, 0, M_z)eiφ/2(iMx,0,Mz)e^{i\varphi/2}(iM_x, 0, M_z)

Supplementary Table 3. Magnetic structure models used in the Rietveld refinement. S±S_{\pm} denotes the allowed (mx,my,mz)(m_x, m_y, m_z) at (0,0,±z)(0, 0, \pm z) in the given superspace group and MxM_x, MyM_y, MzM_z, φ\varphi, and θ\theta are real numbers.

We note that there are three features that require careful treatment in the magnetic Rietveld refinement of the powder diffraction data while they do not affect the main conclusion that the magnetic structure has an incommensurate dominant easy-axis component: ferromagnetic impurity phase CeSi1.7_{1.7}, asymmetric peak broadening and diffuse scattering at low angles. First, the ferromagnetic reflections of CeSi1.7_{1.7} were included in all the analysis assuming that the Ce moment points in the <100><100> direction[1]. Second, asymmetric peak broadening, which manifests itself at (000)±(000)_{\pm} and (001)−(001)_{-} peaks, indicates finite correlation length along the out-of-plane direction due to two-dimensional character of the magnetism. This effect is simulated by plate-like magnetic domains via Size-model function in Fullprof[1]. Finally, presence of diffuse scattering is inferred from the residual intensity around 15°, between (000)±(000)_{\pm} and (001)−(001)_{-} reflections, which cannot totally be reproduced by plate-like domains. For the background, we initially used a linear interpolation of fixed points in the regions without magnetic Bragg peaks. Then to reproduce the actual background containing broad peaks of diffuse scattering, we also refined the background intensity of intermediate positions, which include magnetic Bragg peaks. The results of the refinement with the fixed background points and the refined background points are presented in Supplementary Table 4 and Supplementary Figure 4 and Supplementary Table 5 and Supplementary Figure 5, respectively.

ModelMxM_x or MyM_yMzM_zφ\varphiξ\xika∗k_a^*kc∗k_c^*MCeSi1.7M_{\mathrm{CeSi1.7}}RpR_pRwpR_{wp}RexpR_{exp}χ2\chi^2
SDWb_b1.157(7)−0.351(3)15.7(5)0.2784(1)0.2029(8)0.65(2)36.531.79.1811.9
SDWa∗c_{a^*c}0.25(1)1.262(6)0.363(2)14.0(3)0.27807(5)0.2032(4)0.43(1)24.718.49.174.0
Cycloid1_10.22(3)1.288(7)0.187(1)13.6(3)0.27824(6)0.2059(5)0.46(1)26.620.89.175.1
Cycloid2_20.40(4)1.27(1)0.179(2)12.5(3)0.27800(6)0.2047(5)0.42(2)27.921.69.175.5

Supplementary Table 4. Magnetic structure refinement with fixed background. MM, θ\theta, and φ\varphi are parameters that define the magnetic structure in Supplementary Table 3. ξ\xi is anisotropic Lorentzian contribution of particle size for a plate-like coherent domain[3]. ka∗k_a^*, kc∗k_c^* are a∗a^* and c∗c^* component of the propagation vector. MCeSi1.7M_{\mathrm{CeSi1.7}} is length of ferromagnetic moment of CeSi1.7_{1.7}. RpR_p, RwpR_{wp}, and RexpR_{exp}, χ2\chi^2 represent profile factor, weighted profile factor, expected weighted profile factor, and reduced chi square, respectively.

ModelMxM_x or MyM_yMzM_zφ\varphiξ\xika∗k_a^*kc∗k_c^*MCeSi1.7M_{\mathrm{CeSi1.7}}RpR_pRwpR_{wp}RexpR_{exp}χ2\chi^2
SDWb_b1.00(1)−0.344(5)10.4(6)0.27833(8)0.1936(6)0.37(3)39.228.811.66.2
SDWa∗c_{a^*c}0.14(1)1.08(1)0.355(3)7.5(3)0.27809(5)0.1935(4)0.52(1)26.918.211.32.6
Cycloid1_10.07(1)1.11(1)0.175(2)6.5(3)0.27808(5)0.1938(4)0.53(1)26.718.611.22.8
Cycloid2_20.71(2)1.024(9)0.161(2)8.1(3)0.27806(4)0.1936(3)0.55(1)21.815.910.62.3

Supplementary Table 5. Magnetic structure refinement with refined background. The refinable background consists of 36 points between 9 and 99° in equal spacing of 2.5°.

Image not extracted

(a)

Image not extracted

(b)

Image not extracted

(c)

Image not extracted

(d)

Supplementary Figure 5. Powder neutron diffraction pattern of CeSiI and a Rietveld fitting to the magnetic spin structures with refined background parameters.

In the fitting without refinable background, the obtained MzM_z exceeds the saturation field of 1.1μB1.1\mu_{\mathrm{B}} and χ2\chi^2 is much larger than 1 due to poor agreement near (000)±(000)_{\pm} and (001)−(001)_{-} peak. By contrast, in the fitting with refinable background, MzM_z is closed to the observed value in Γ2\Gamma_2 models and χ2\chi^2 is greatly improved. The refined CeSi1.7_{1.7} moment also agrees well with the reported value of 0.5μB0.5\mu_{\mathrm{B}} regardless of the background refinement. Therefore, fittings with the refined background are more likely to capture the feature of the magnetic structure and Cycloid2_2 is the primary candidate of the magnetic structure. However, small difference in the agreement factor between the collinear and noncollinear orders certainly calls for polarized neutron diffraction experiments using single crystals in order to measure the chiral term iMxMq∗iM_{x}M_{q}^{*}, which is proportional to spin chirality. Although the limitation of the crystal size hinders such a measurement at the current stage, a large single crystal will uncover the magnetic structures even under magnetic fields and thus shed light on the origin of the frustration in CeSiI.

Image not extracted

Figure S4. Powder neutron diffraction pattern of CeSiI and a Rietveld fitting to the magnetic spin structures Γ1\Gamma_{1} SDWb_{b} (a), Γ2\Gamma_{2} SDWa∗c_{a^{*}c} (b), Γ2\Gamma_{2} Cycloid1_{1} (c), and Γ2\Gamma_{2} Cycloid1_{1} (d) with fixed background parameters. Fitting regions are between 8∘8^{\circ} and 100∘100^{\circ}. The red circle represents data taken at 1.6 K after the subtraction of the 7.7 K data as a reference of nuclear contributions. The black, blue, and green line represent a Rietveld fit, a residual of fitting, and background, respectively. The red thin and thick green bar represents position of magnetic Bragg peaks of CeSiI and CeSi1.7_{1.7}, respectively. The arrows indicate the indices of magnetic Bragg peaks, where (hkl)±(hkl)_{\pm} represents a Miller index of (hkl)±(0.28,0,0.20)(hkl)_{\pm}(0.28, 0, 0.20) reflection.

Supplementary References

References

  1. [1]Geim, A. K., & Grigorieva, I. V. Van der Waals heterostructures. Nature 499, 419 (2013).DOI
  2. [2]Burch, K. S., Mandrus, D., & Park, J.-G. Magnetism in two-dimensional van der Waals materials. Nature 563, 47–52 (2018).DOI
  3. [3]Gong, C. & Zhang, X. Two-dimensional magnetic crystals and emergent heterostructure devices. Science 363, eaav4450 (2019).DOI
  4. [4]Li, H., Ruan, S., & Zeng, Y. J. Intrinsic van der Waals magnetic materials from bulk to the 2D limit: New frontiers of spintronics. Adv. Mater. 31, 1900065 (2019).DOI
  5. [5]Imada, M., Fujimori, A., & Tokura, Y. (1998). Metal-insulator transitions. Rev. Mod. Phys., 70, 1039.
  6. [6]Keimer, B., & al., . et . (2015). From quantum matter to high-temperature superconductivity in copper oxides. Nature, 518, 179–186.
  7. [7]Stewart, G. R. (2011). Superconductivity in Iron Compounds. Rev. Mod. Phys., 83, 1589–1652.
  8. [8]Stewart, G. R. (1984). Heavy-fermion systems. Rev. Mod. Phys., 56, 755.
  9. [9]Tokura, Y., Yasuda, K., & Tsukazaki, A. (2019). Magnetic topological insulators. Nat. Rev. Phys., 1, 126–143.
  10. [10]Zhang, Y., & al., . et . (2018). Emergence of Kondo lattice behavior in a van der Waals itinerant ferromagnet, $\mathrm{Fe_3GeTe_2}$. Sci. Adv., 4, eaao6791.
  11. [11]Fei, Z., & al., . et . (2018). Two-dimensional itinerant ferromagnetism in atomically thin $\mathrm{Fe_3GeTe_2}$. Nat. Mater., 17, 778–782.
  12. [12]Kim, K. et al. Large anomalous Hall current induced by topological nodal lines in a ferromagnetic van der Waals semimetal. Nat. Mater. 17, 794 (2018).DOI
  13. [13]Deng, Y., & al., . et . (2018). Gate-tunable room-temperature ferromagnetism in two-dimensional $\mathrm{Fe_3GeTe_2}$. Nature, 563, 94.
  14. [14]Li, J. et al. Intrinsic magnetic topological insulators in van der Waals layered MnBi₂Te₄-family materials. Sci. Adv. 5, eaaw5685 (2019).DOI
  15. [15]Deng, Y., & al., . et . (2020). Quantum anomalous Hall effect in intrinsic magnetic topological insulator $\mathrm{MnBi_2Te_4}$. Science, 367, 895–900.
  16. [16]Lei, S. et . al . (2020). High mobility in a van der Waals layered antiferromagnetic metal. Sci. Adv., 6, eaay6407.
  17. [17]Okuma, R. et al. "Fermionic order by disorder in a van der Waals antiferromagnet." *Sci. Rep.* **10**, 1-8 (2020).DOI
  18. [18]Nakatsuji, S., Kiyohara, N. & Higo, T. Large anomalous Hall effect in a non-collinear antiferromagnet at room temperature. *Nature* **527**, 212 (2015).DOI
  19. [19]Okuma, R. et al. A series of magnon crystals appearing under ultrahigh magnetic fields in a kagomé antiferromagnet. *Nat. Commun.* **10**, 1-7 (2019).DOI
  20. [20]Nagaosa, N. & Tokura, Y. Topological properties and dynamics of magnetic skyrmions. *Nature Nanotech.* **8**, 899–911 (2013).DOI
  21. [21]Kurumaji, T. et al. Skyrmion lattice with a giant topological Hall effect in a frustrated triangular-lattice magnet. *Science* **365**, 914–918 (2019).DOI
  22. [22]Katsura, H., Nagaosa, N., & Balatsky, A. V. (2005). Spin current and magnetoelectric effect in noncollinear magnets. Phys. Rev. Lett., 95, 057205.
  23. [23]Martin, J. D. et al. Metal halide analogues of chalcogenides: A building block approach to the rational synthesis of solid-state materials. *Chem. Mater.* **10**, 2699-2713 (1998).DOI
  24. [24]Jiang, S. et al. Controlling magnetism in 2D CrI₃ by electrostatic doping. *Nat. Nanotech.* **13**, 549-553 (2018).DOI
  25. [25]Huang, B. et al. Electrical control of 2D magnetism in bilayer CrI₃. *Nat. Nanotech.* **13**, 544-548 (2018).DOI
  26. [26]Klein, D. R. et al. Probing magnetism in 2D van der Waals crystalline insulators via electron tunneling. *Science* **360**, 1218-1222 (2018).DOI
  27. [27]Banerjee, A. et al. Neutron scattering in the proximate quantum spin liquid α-RuCl₃. *Science* **356**, 1055-1059 (2017).DOI
  28. [28]Corbett, J. D., Sallach, R. A., & Lokken, D. A. Physical Characterization of the Metallic LaI₂ and CeI₂ and of the Phase LaI₂. 42. *Advances in Chemistry* **71**, 56-66 (1967).DOI
  29. [29]Burrow, J. H. et . al . (1987). The electronic conditions in the 5d¹ layer-metal LaI₂ making comparison with the iso-electronic tantalum dichalcogenides, with the other RE di-iodides, and with the RE monochalcogenides. J. Phys. C: Solid State, 20, 4115.
  30. [30]Meyer, G., Gerlitzki, N., & Hammerich, S. (2004). Rare-earth diiodides and derivatives. J. Alloy. Comp., 380, 71–78.
  31. [31]Kasten, A., Müller, P. H., & Schienle, M. Magnetic ordering in GdI₂. *Solid state Commun.* **51**, 919-921 (1984).DOI
  32. [32]Felser, C. et . al . (1999). Giant negative magnetoresistance in GdI₂: Prediction and realization. J. Solid State Chem., 147, 19–25.
  33. [33]Mattausch, H. & Simon, A. Si₆, Si₁₄, and Si₂₂ Rings in Iodide Silicides of Rare Earth Metals. *Angew. Chem. Int. Ed.* **37**, 499-502 (1998).pubmed.ncbi.nlm.nih.gov/29711166
  34. [34]Mattausch, H. et . al . (2005). Reduced Lanthanum Halides with Ge as Interstitials: La₂I₂Ge, La₂I₂Ge₂, La₃Cl₂Ge₃, La₃Br₂Ge₃, La₃I₃Ge, La₆I₅Ge₃ and La₇+xI₁₂Ge. Z. Anorg. Allg. Chem., 631, 302–308.
  35. [35]Zheng, C. et al. RE₂+xI₂M₂+y (RE = Ce, Gd, Y; M= Al, Ga): Reduced Rare Earth Halides with a Hexagonal Metal Atom Network. *Z. Naturforsch. B* **62**, 633-641 (2007).DOI
  36. [36]Zhou, J. et al. 2DMatPedia, an open computational database of two-dimensional materials from top-down and bottom-up approaches. *Sci. Data* **6**, 86 (2019).DOI
  37. [37]Banda, J. et al. Crystalline electric field of Ce in trigonal symmetry: CeIr₃Ge₇ as a model case. *Phys. Rev. B* **98**, 195120 (2018).DOI
  38. [38]Yashima, H. & Satoh, T. Nonmagnetic-magnetic transition in Ce-Si system. *Solid State Commun.* **41**, 723-727 (1982).DOI
  39. [39]Cable, J. W. et . al . (1962). Neutron Diffraction Investigation of the Magnetic Order in MnI₂. Phys. Rev., 125, 1860.
  40. [40]Kurumaji, T. et . al . (2011). Magnetic-field induced competition of two multiferroic orders in a triangular-lattice helimagnet MnI₂. Phys. Rev. Lett., 106, 167206.
  41. [41]Kuindersma, S. R., Sanchez, J. P., & Haas, C. Magnetic and structural investigations on NiI₂ and CoI₂. *Physica B+C* **111**, 231-248 (1981).DOI
  42. [42]Kurumaji, T. et al. Magnetoelectric responses induced by domain rearrangement and spin structural change in triangular-lattice helimagnets NiI₂ and CoI₂. *Phys. Rev. B* **87**, 014429 (2013).DOI
  43. [43]Ju, H. et al. Possible Persistence of Multiferroic Order down to Bilayer Limit of van der Waals Material NiI2. Nano Lett. 21, 5126–51 (2021).DOI
  44. [44]Fei, Z. et . al . (2018). Ferroelectric switching of a two-dimensional metal. Nature, 560, 336–339.
  45. [45]Hansen, T. C. et al. The D20 instrument at the ILL: a versatile high-intensity two-axis neutron diffractometer. Meas. Sci. Technol. 19, 034001 (2008).DOI
  46. [46]Rodríguez-Carvajal, J. (1993). Recent advances in magnetic structure determination by neutron powder diffraction. Physica B, 192, 55–69.
  47. [47]Wills, A. S. (2000). A new protocol for the determination of magnetic structures using simulated annealing and representational analysis (SARAh). Physica B: Condens. Matter, 276, 680–681.
  48. [48]Stokes, H. T., Hatch, D. M., & Campbell, B. J. (n.d.). ISODISTORT, ISOTROPY Software Suite. Retrieved iso.byu.edu
  49. [49]Campbell, B. J. et . al . (2006). ISODISPLACE: An Internet Tool for Exploring Structural Distortions. J. Appl. Cryst., 39, 607–614.

Paper details

vlnc:W_npp4xdbq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents