1D Magnetic MX3 Single‐Chains (M = Cr, V and X = Cl, Br, I)

Magnetic materials in reduced dimensions are not only excellent platforms for fundamental studies of magnetism, but they play crucial roles in technological advances. The discovery of intrinsic magnetism in monolayer 2D van der Waals systems has sparked enormous interest, but the single‐chain limit of 1D magnetic van der Waals materials has been largely unexplored. This paper reports on a family of 1D magnetic van der Waals materials with composition MX3 (M = Cr, V, and X = Cl, Br, I), prepared in fully‐isolated fashion within the protective cores of carbon nanotubes. Atomic‐resolution scanning transmission electron microscopy identifies unique structures that differ from the well‐known 2D honeycomb lattice MX3 structure. Density functional theory calculations reveal charge‐driven reversible magnetic phase transitions.


Introduction
Magnetism at surfaces, interfaces, and lowdimensional materials generally has long been of central interest for both fundamental research and technological advances. [1]he reduced dimensionality often leads to strong spin fluctuations that affect magnetic ordering and give rise to a variety of novel magnetic phenomena [2] including enhanced magnetic anisotropy.1a,g] Recent discoveries of intrinsic magnetism in 2D van der Waals materials have intensified the search for new atomically thin magnetic materials, and single 2D layers of CrI 3, [1e] Cr 2 Ge 2 Te 6, [1c] FePS 3 , [3] VSe 2 , [4] and MnSe 2 [5] are prime examples.However, there has been limited exploration of other dimensional polymorphs, particularly 1D chain structures.Moreover, although some quasi-1D magnetic materials have received modest attention, the synthesis of fully isolated, truly 1D single-chain magnets has proven difficult.In a promising recent advancement, [6] some single CrCl 3 chains were observed amongst collections of multi-chain CrCl 3 bundles grown on NbSe 2 substrates.Synthesis and further experimental studies of such structures are exceedingly challenging because of the sensitivity of halide materials to light, oxygen, and moisture. [7]ne promising method for fabricating and enabling further characterization of sensitive 1D structures is to employ hollowcore nanotubes as nano-reaction vessels.Carbon nanotubes (CNTs) and boron nitride nanotubes (BNNTs) have facilitated synthesis of various materials, including monoelements, [8] metal halides, [9] chalcogenides, [10] and perovskites. [11]9b,10g] Furthermore, especially in the case of CNT encapsulation, possible charge transfer between the CNT and the core chain provides an additional control parameter for altering material properties.
Here, we report the creation and study of a family of 1D magnetic van der Waals single-chains of MX 3 (M = Cr, V and X = Cl, Br, I), fully isolated within the cores of CNTs.Nanotube encapsulation enables stabilization of the structure of the 1D MX 3 single-chains, and simultaneously prevents environmental degradation, facilitating detailed structural characterization.Atomic-resolution scanning transmission electron microscopy (STEM) imaging and simulation clearly identify the 1D MX 3 chain configurations.The 1D MX 3 chains consist of face-sharing MX 6 octahedra, in distinct contrast to the edge-sharing octahedral found in the complementary 2D honeycomb lattice MX 3 structure.Density functional theory (DFT) calculations reveal that a significant fraction of electronic charges is transferred from the CNTs to the chain.The electron transfer stabilizes the face-sharing single-chain structure against the edge-sharing layered structure, and strongly affects the magnetic energies of the chains.Notably, CrX 3 chains tend to undergo antiferromagnetic (AFM) to ferromagnetic (FM) transitions as a function of electron doping.

Results and Discussion
CrI 3 chains are directly synthesized by vacuum annealing CrI 3 precursors in the presence of open-ended nanotubes at high temperatures (see Experimental Section for the details).The samples are primarily screened using transmission electron microscopy (TEM) to confirm that the target material has successfully filled inside the nanotubes.The atomic structure of the filled material is further investigated using annular dark field (ADF) aberrationcorrected STEM.
Figure 1a shows an atomic-resolution ADF-STEM image of a 1D CrI 3 single-chain encapsulated within a nanotube with an inner diameter of 1.1 nm.The CrI 3 chain encapsulated nanotubes are filled to around 90% with chain length of several hundred nanometers (Figure S1, Supporting Information).The composition ratio of the encapsulated CrI 3 is determined by energy-dispersive spectroscopy (EDS), which confirms a composition 25.4 ± 2.8 atomic percent (atom%) Cr and 74.6 ± 3.6 atom%, yielding an atomic ratio X/M = 2.9 (expected ratio = 3).The ADF-STEM image contrast strongly depends on the atomic number (Z) which allows direct distinguishing of Cr (Z Cr = 24) from I (Z I = 53).Our DFT binding energy calculations (Figure S2, Supporting Information) show that CrI 3 single-chains are most stable when encapsulated within a nanotube of inner diameter 1.1 nm for both metallic armchair and semiconducting zigzag CNTs; this critical diameter is corroborated by our experimental observations for metallic CNTs.
Based on the observed STEM images, we find that the 1D CrI 3 chain structure consists of face-sharing CrI 6 octahedra along the chain axis (Figure 1b and Figure S3, Supporting Information).Figure 1c and d are magnified high-resolution (HR) ADF-STEM images at the different viewing directions (0°and 90°, respectively), which clearly show the atomic structure of 1D CrI 3 .Figure 1e,f shows intensity line profiles from the regions marked in Figure 1c,d, respectively.In the case of a 0°rotated image, the intensity difference of I 2 and I 1 along the chain direction is clearly distinguishable.The uniform intensities of I 1 are obtained from the 90°-rotated image.The lattice constant along the chain directions is 6.8 Å from the observed images.The simulated STEM images are generated using the identified atomic structure, which match well with the experimentally observed results, as shown in Figure 1g,h.
The observed face-sharing in 1D CrI 6 octahedrons is different from that observed in the 2D CrI 3 counterpart.While the 2D CrI 3 also contains the same octahedrons, they are connected to each other in an edge-sharing mode to form a 2D honeycomb lattice (Figure S4, Supporting Information).Similar 1D face-sharing octahedral MX 3 chain structures are observed in other quasi-1D crystals such as -TiCl 3 [12] and -RuCl 3 . [13]We emphasize, however, that our observed 1D CrI 3 chain structure, and the isolation of this chain in a single-chain form, have not been previously reported.2D CrI 3 often requires graphene or h-BN encapsulation for protection from ambient exposure or light-induced damages.We find that the synthesized CrI 3 chains inside nanotubes are stable under ambient exposure.9b,10g,11] Extending the above findings for CrI 3 , we study the 1D MX 3 single-chain with different metal and halogen combinations.In our study we use in-house synthesized crystals of CrBr 3 and VI 3 , as well as commercially available powders of CrCl 3 , VBr 3 , and VCl 3 for MX 3 encapsulated nanotube fabrication (see Experimental Section for details).Figure 2 shows the experimental and simulated atomic-resolution STEM images of 0°-rotated MX 3 singlechains within the nanotube with different halogens and metals.All synthesized MX 3 single-chains inside the nanotube consist of face-sharing octahedral, identical to the CrI 3 case.The experimentally observed STEM images (left side) of CrBr 3 , VI 3 , and VBr 3 show similar image contrast to those of CrI 3 (Figure 2a,c,d) as here again the atomic numbers of the metals (Z V = 23 and Z cr = 24) are lower than those of the corresponding halogens (Z I = 53 and Z Br = 35).In the case of CrCl 3 and VCl 3 , the atomic number of Cl (Z cl = 17) is lower than that of the metals, result-ing in lower intensity compared to the metals (Figure 2b,e).The simulated STEM images (right side) of the MX 3 chain also show good agreement with the experimental results.The 90°-rotated simulated STEM images are shown in Figure S5, Supporting Information.
The measured lattice constants of the MX 3 chains along the chain direction are 6.3 Å for CrBr 3 , 6.0 Å for CrCl 3 , 6.8 Å for VI 3 , 6.3 Å for VBr 3 , and 6.0 Å for VCl 3 , respectively (Figure S6, Supporting Information).The lattice constant of the MX 3 materials is primarily influenced by the halogen element rather than the transition metal.The size of the halogen ion affects the distance between neighboring atoms, which in turn affects the lattice constant.Transition metals, on the other hand, have a minor influence on the lattice constant of these materials.Our results strongly suggest that the face-sharing 1D MX 3 chain structures inside nanotubes can be stabilized universally in MX 3 compounds.
We now theoretically investigate the atomic, electronic, and magnetic properties of MX 3 single-chains by using firstprinciples DFT calculations.We first focus on isolated MX 3 single-chains with no surrounding nanotube.The atomic positions of isolated MX 3 single-chains are optimized by minimizing the total energy.The optimized lattice parameters are in excellent agreement with the experimental lattice parameters as shown in Table 1.For all MX 3 compounds, the MX 6 octahedra are elongated along the chain axis compared to the ideal octahedral structure.Although the elongation brings a small splitting of t 2g orbitals into a doubly-degenerate e  g and a nondegenerate a 1g orbital, [14] three (two) t 2g electrons in Cr 3+ (V 3+ ) magnetic ions remain fully spin-polarized to form S = 3/2 (S = 1) AFM (FM) chains (Figure 3a,b).CrCl 3 , CrBr 3 , and VI 3 have the easy axis perpendicular to the chain axis, and CrI 3 , VCl 3 , and VBr 3 parallel to the chain axis.For both Cr and V chains, the magnetic anisotropy is the strongest for chains with iodine.We note that the dependence of the magnetic anisotropy on the ligand atom is also pronounced in the monolayer CrX 3 , where the spin-orbit coupling of the ligand atom plays the dominant role in determining the magnetic anisotropy energy. [15]e now determine the magnetic exchange parameters of the 1D anisotropic Heisenberg model where S || i and S ⊥ i are spin operators parallel and perpendicular to the easy axis, respectively, and J || ij and J ⊥ ij are corresponding exchange interaction parameters.First, we obtain the density matrix of isolated MX 3 single-chains in vacuum including the spinorbit interaction, and then the exchange parameters are calculated based on the magnetic force theorem. [16]Figure S7, Supporting Information shows the calculated exchange interaction parameters as a function of distance between spins.We find that the magnetism of CrX 3 and VX 3 chains are of localized and itinerant character, respectively.For CrX 3 , the magnetic interactions are short-ranged and XXZ-type (J || ij ≠ 0).The dominant exchange parameters for CrX 3 are summarized in Table 1.On the other hand, VX 3 has oscillating and slowly decaying exchange interactions, which indicates itinerant magnetic interactions.
Figure S8, Supporting Information shows the calculated electronic structures of isolated nanotube-free MX 3 single-chains.Cr-based chains are insulating when in either an FM or AFM state.Interestingly, the conduction bands of CrX 3 chains in the AFM phase are extremely flat and energetically well isolated.The narrow bandwidth can be understood as arising from kinetic frustration, [17] where the direct hopping between Cr d orbitals destructively interferes with indirect hopping mediated by p orbitals in the halide atoms (see Figure S9, Supporting Information showing the conduction band wave functions).2b] On the other hand, FM VX 3 chains are half-metals where majority spin states are metallic and minority spin states are insulating with a large band gap over 2 eV.
To illustrate the role of nanotube encapsulation, we examine theoretically MX 3 single-chains encapsulated within CNTs. Figure 3c,d shows that, because of the work function mismatch, Table 1.Experimental (a exp ) and calculated lattice constants (a calc ) of MX 3 chains and DFT results of formation energies (E f ), easy axis, magnetic anisotropy energy (E MAE ), and magnetic exchange interaction parameters parallel (J || 1,2 ) and perpendicular (J ⊥ 1,2 ) to the easy axis up to the second nearest neighbors.an electronic charge transfers from the CNT.The amount of the electron transfer is around 0.25 e per formula unit (f.u.) and does not sensitively depend on the composition.The additional electrons from the CNT are distributed among both transition metal and halide atoms in the chain.10b,h,18] The charge transfer makes the encapsulating CNTs hole-doped, which can have experimental signatures such as raising of the Raman G band frequency in CNTs. [19]o understand the effects of electron doping on MX 3 chains, we calculate the electron doping dependence of MX 3 singlechains in vacuum.Figure 3e compares the total energies of MX 3 in the single-chain phase against the monolayer phase consisting of a honeycomb structure of edge-sharing octahedra.While the monolayer phase is more stable without electron doping, the single-chain phase quickly becomes more stable with electron doping.When the doping amount is 0.25 e/f.u., which represents the amount of the electron transfer from the CNT, all MX 3 chains are more stable than their monolayer counterparts.This shows that electron transfer from the CNT plays a crucial role in further stabilizing the single-chain phases.For single-chain CrCl 3 synthesized on NbSe 2 substrates, charge transfer may similarly aid stabilization of the chain structure. [6]In addition, our calculations show that the magnetic states of MX 3 single-chains are strongly affected by electron doping.Figure 3f shows the magnetic energy as a function of electron doping.CrX 3 chains are AFM at the neutral phase but they become FM as electrons are added.Based on the results presented in Figure 3f, the critical doping level for the transition is about 0.35, 0.25, and 0.12 e/f.u. for Cl, Br, and I, respectively.
Therefore, within a CNT with intrinsic charge transfer, the magnetic state of CrI 3 is nominally in the FM state.This doping dependent switching of magnetism for CrX 3 appears in the single-chain phase, but not in its monolayer 2D phase, where the FM ground state is not changed by doping (Figure S10, Supporting Information).On the other hand, the magnetic energy of VX 3 chains also tends to decrease in magnitude with increased doping, but the system remains FM for all doping ranges considered.Our calculations suggest that electrically tunable magnetism in CrX 3 single-chains can be realized by electrostatic gating of CrX 3filled CNTs.
Figure 4 shows the electronic structures of MX 3 single-chains encapsulated in CNTs.The Dirac point energy of CNT is pushed upward because of electron transfer from the CNT to the chains.For CrX 3 , electrons transferred from the CNT populate the conduction band of the chain.Notably, with CNT encapsulation, electron doping brings the flat bands in the AFM CrX 3 to the Fermi level, and such states are energetically well separated from the other chain-derived states.Those flat bands can provide an ideal system to study correlated electron physics in 1D.For the FM CrI 3 chain, the additional electron goes into spin-polarized conduction bands, so CrI 3 becomes a half metal.Since the isolated FM VX 3 chains are already half-metallic, charge transfer from the CNT provides additional spin-polarized carriers to the chain.

Conclusions
In conclusion, we present 1D CrX 3 and VX 3 magnetic singlechain structures via nanotube encapsulation.We demonstrate the 1D face-sharing octahedron MX 3 structure can be universally synthesized as a single-chain limit inside a nanotube.The nanotube vessel stabilizes and protects the materials, allowing ac-cess to 1D single-chain limit, specifically CrI 3 , CrBr 3 , CrCl 3 , VI 3 , VBr 3 , and VCl 3 .Our DFT calculations suggest that charge transfer from CNT to MX 3 single-chain plays a critical role in stabilizing the chain structures.Notably, AFM CrX 3 chains host isolated flat bands to which electron doping can be achievable via charge transfer from the CNT.We find that magnetic states of CrX 3 chains can be switched by electron doping.These findings can pave the way for further investigations into low-dimensional magnetic systems and confinement-stabilized materials in nanotubes, which offer exciting opportunities for future research and applications across various fields.

Experimental Section
Materials: CNTs were purchased from Sigma Aldrich (single-walled: 704113) and CheapTubes (90% SW-DW CNTs) and were annealed in air at 510 °C for 15 min prior to filling to open the end caps.The precursors utilized for filling included commercially available CrI 3 (purchased from Ossila), CrCl 3 (99.99%Sigma Aldrich and 99.9% Alfa-Aesar), VBr 3 (99.5% Alfa-Aesar), and VCl 3 (99% Alfa-Aesar) powders, as well as synthesized CrBr 3 and VI 3 crystals.The chemical vapor transport (CVT) method was used to grow CrBr 3 and VI 3 precursors.For CrBr 3 , a 1:0.8 molar ratio of Chromium (99% Alfa-Aesar) and TeBr 4 (99.9%Alfa-Aesar) was mixed with a total mass of 1 g in a quartz ampule.For VI 3 , a 1:3 molar ratio of Vanadium (99.7% Alfa-Aesar) and Iodine (99.99% Alfa-Aesar) was mixed with a total mass of 1 g in a quartz ampule.The ampule (10 mm diameter and 15 cm long) was then sealed under high vacuum (≈10 −6 torr) and placed in a horizontal one-zone furnace with the hot end at 750 °C for 5 days before being cooled to room temperature.The synthesized small crystals were extracted inside an Ar-filled glove box to minimize oxidation.
Growing MX 3 @ Nanotube: The CNTs (≈3 mg) were mixed with 30 mg of precursor materials and sealed in a 6 mm inner diameter and 15 cm long quartz ampule under high vacuum (≈10 −6 torr).The sealed ampule was then heated to 650 °C in a single-zone box furnace and kept there for 3 days before being cooled down to room temperature over 1 day.The synthesized materials (MX 3 @nanotube) were dispersed in isopropanol using a bath sonicator for 15 min, and drop-cast onto lacey carbon TEM grids for TEM/STEM characterization.
TEM/STEM Imaging and Simulations: Initial sample screening was conducted using a JEOL 2010 microscope at 80 kV.Atomic-resolution ADF-STEM images were acquired using the double spherical (Cs) aberration-corrected JEOL ARM-200F and TEAM 0.5 at the National Center for Electron Microscopy.The JEOL ARM-200F microscopy was set at 80 kV with a 23 mrad convergence angle and collection semiangles ranging from 40 to 160 mrad, while the TEAM 0.5 instrument was operated at 80 kV with a convergence angle of 30 mrad and collection semiangles from 37 to 187 mrad.The electron dose for atomic-resolution STEM imaging was estimated to be ≈ 1 × 10 8 e -nm -2 .The STEM images were calibrated using 3.4 Å inter-wall spacing of CNTs.
HR STEM image simulations were simulated using MacTempas software based on multislice calculations.The simulation parameters were similar to the experimental parameters (e.g., a probe semiangle of 23 or 30 mrad, 0.05 Å pixel −1 sampling, and 20 frozen phonon calculations) for each simulation.Image analysis and processing were performed using Im-ageJ software.Poisson noise was added for the simulated STEM images to match the experimental results.
Calculations: First-principles DFT calculations were performed as implemented in SIESTA. [20]The Perdew-Burke-Ernzerhof functional, [21] fully relativistic optimized norm-conserving pseudopotentials, [22] and a localized pseudoatomic orbital basis were used.Van der Waals interactions were included within the Grimme-D2 scheme. [23]A real-space mesh cutoff of 800 Ry was used.A 40-thick cell was used along the transverse vacuum direction.The primitive Brillouin zone of isolated MX 3 chains was sampled by 32 k points, and the number of k points was proportionally reduced in supercell calculations.The atomic positions of MX 3 chains with and without a CNT were optimized with a force threshold of 0.01 eV Å −1 while fixing the position of carbon atoms in CNT.Electron doping to the isolated chain was simulated by adjusting the total number of electrons and adding positive compensating background charges.The sisl package was used to process and plot the real-space electron density. [24]10k]

Figure 1 .
Figure 1.Structure characterization of 1D CrI 3 single-chain inside a nanotube.a) Aberration-corrected ADF-STEM image of a CrI 3 single-chain encapsulated within a double-walled CNT.Scale bar: 1 nm.b) Atomic model of face-sharing 1D MX 3 single-chain within nanotube.The face-sharing MX 6 octahedra are shadowed.The transition metal atoms are displayed in blue and the halogen atoms are displayed in red.c,d) Experimentally observed atomic-resolution STEM images of c) 0°and d) 90°rotated CrI 3 single chains encapsulated inside nanotubes.Cr, I 2 , and I 1 are marked by blue, yellow, and red circles, respectively.Scale bar: 0.5 nm.e,f) Intensity line profile along the colored line in panels (c) and (d), respectively.g,h) Simulated STEM images of g) 0°and h) 90°rotated CrI 3 single chains encapsulated inside nanotubes.Cr, I 2 , and I 1 are marked by blue, yellow, and red circles, respectively.Scale bar: 0.5 nm.

Figure 2 .
Figure 2. Universal stabilization of 1D MX 3 within a nanotube.a-e) Experimental (left) and simulated (right) atomic-resolution STEM images of a) CrBr 3 , b) CrCl 3 , c) VCr 3 , d) VBr 3 , and e) VCl 3 , respectively.Scale bar: 0.5 nm.The positions of transition metal atoms, halogen double atoms, and halogen single atoms are marked by blue, yellow, and red circles, respectively.

Figure 3 .
Figure 3. Calculated magnetic state, charge transfer from CNT, and electron doping effects.Schematics of a) antiferromagnetic and b) ferromagnetic state of CrI 3 single-chains.The easy axis is along the chain direction.c,d) Electron density transferred from the CNT to the chain.e) Relative stability of 1D chain phase against 2D monolayer phase and f) magnetic energy of single-chains as a function of electron doping.

Figure 4 .
Figure 4. Calculated electronic structures of single-chain MX 3 encapsulated in CNTs.a) Antiferromagnetic CrCl 3 , b) antiferromagnetic CrBr 3 , c) ferromagnetic CrI 3 , and d-f) ferromagnetic VCl 3 , VBr 3 and VI 3 .Supercell band structures are projected onto and unfolded with respect to the primitive Brillouin zone (PBZ) of the chain and CNT, respectively, and Z chain and Z CNT denote the PBZ boundaries.Black lines indicate the CNT states, and red and blue lines are the majority and minority spin states of the chain, respectively.