INTRODUCTION

The development of exfoliatable 2D van der Waals (vdW) materials with magnetic properties has attracted considerable interest ever since CrI 3 was shown to maintain ferromagnetism (FM) down to the monolayer limit. 1 These 2D magnetic materials have offered the possibility to explore many intriguing phenomena, such as proximity effect induced magnetism, 2 spin dynamics, 3,4 and new moirémagnetic states, 5 and have many potential applications in spintronics and as components in bulk heterostructures. [6][7][8] However, many 2D van der Waals magnetic systems, such as the transition metal trihalides, [9][10][11][12][13][14] suffer from extreme air and water sensitivity and low transition temperatures especially when exfoliated into the few-layer regime. 1,15 A rigorous characterization of the atomic structure of monolayers whose properties are directly under investigation is rarely performed.

Together these difficulties can often result in a divergence of property measurements for the same materials. For instance, many chromium chalcogenide materials have received a lot of debate over their magnetic properties. One of the most recently studied 2D materials is metallic 1T-CrTe 2 . A number of synthesis methods have been used to make the material including the deintercalation of K from KCrTe 2 , 16 chemical vapor deposition (CVD), 17 and molecular beam epitaxy (MBE). 18 All three methods report ferromagnetism for 1T-CrTe 2 below 300 K. However, the easy axis of the material has been reported in both the in-plane and cross-plane directions, while the stability of the material is poor, decomposing above >300 K and requiring a Te capping layer to protect it from oxidation when used in a heterostructure. 1T-CrSe 2 has been shown to be antiferromagnetic with a Neél temperature of ∼165-180 K and is a metastable phase prepared via I 2 -based deintercalation of KCrSe 2 . 19 However, CrSe 2 monolayers grown on WSe 2 monolayers have been reported to be ferromagnetic, with a Curie temperature (T C ) of 110 K, as a result of an increase in lattice constants and interlayer charge transfer. 20 Additionally, CrGeTe 3 , or Cr 2 Ge 2 Te 6 , is another well explored 2D vdW system with soft ferromagnetism, a bulk T C of 68 K, and a cross-plane easy axis. This T C decreases to 30 K in the bilayer, which has been shown to be stable in air for 1.5 h before degradation begins. 6 Another family of 2D magnets that has attracted a lot of interest is the magnetically doped TMDs in which a magnetic element (V, Cr, Mn, Fe, Co, Ni) is substituted onto the TMD lattice. However, the maximum amount of magnetic element doping before phase separation is almost always ≤10%, which often puts them in the regime of dilute magnetic semiconductors. For instance, monolayers of MoS 2 can be doped with ∼2% Fe. 21 Additionally, ∼3% Mn-doped MoS 2 shows ferromagnetism with a T C of 350 K with a magnetic moment of 1.06 μB/Mn. 22 The effects of magnetic doping of Co, Ni, Fe, and Mn (10%) on MoS 2 nanocrystals, Mn-doped (10%) MoSe 2 , and V-doped (12%) WS 2 or (9%) WSe 2 have also been explored. [23][24][25][26] Furthermore, while a handful of compounds such as MoS 2 and WS 2 are air-stable in both the bulk and monolayer form, most early transition metal tellurides have a strong propensity to oxidize, especially upon exfoliation, due to the thermodynamic driving force of metal oxide formation. It is a well-regarded principle in molecular inorganic chemistry that early transition metals tend to be much more oxophilic and prone toward oxidation than late transition metals. 27 Indeed, the late transition metal Pd and Pt TMDs have been shown to be air stable both in the bulk and in monolayer form. [28][29][30][31][32][33] Moreover, PtTe 2 has been claimed to be a type II Dirac semimetal, but these crossing points are energetically far removed from the Fermi level and thus do not cause any relevant exotic transport effects. 34,35 Herein, we show the preparation of an air-and water-stable series that merges the stability of a late transition metal dichalcogenide (PtTe 2 ) with an early transition metal magnetic element. After an initial screening of transition metal alloys of V, Cr, Mn, and Ni with PtTe 2 , we found that Cr substitutes onto the Pt site, resulting in a van der Waals Cr x Pt 1-x Te 2 series with Cr concentrations reaching up to 45% before phase separation. These materials are ferromagnetic with Cr 0.45 Pt 0.55 Te 2 , having a T C of 220 K and an easy axis along the c-axis. The bulk material is metallic with appreciable magnetoresistance and can be exfoliated to few layer thicknesses.

RESULTS AND DISCUSSION

The Cr x Pt 1-x Te 2 series (0 ≤ x ≤ 0.45) was synthesized from stoichiometric mixtures of the elements via traditional quartz tube reactions for the powders and from slow cooling of the melt for crystals. Figure 1 shows representative crystal structures of Cr 0.21 Pt 0.79 Te 2 and Cr 0.40 Pt 0.60 Te 2 as determined from single-crystal X-ray diffraction, as well as representative crystals of PtTe 2 and Cr 0.40 Pt 0.60 Te 2 . PtTe 2 crystallizes into a one-layer trigonal unit cell with a P3̅ m1 space group. Singlecrystal XRD analysis indicates that this 1T P3̅ m1 structure is maintained for x = 0.21 and x = 0.31, with the Cr atoms substituting onto the Pt position (Figure 1a). However, singlecrystal analysis of Cr 0.40 Pt 0.60 Te 2 indicated that it crystallized into a two-layer trigonal (2T) cell having a P3̅ m1 space group. In this case, one layer of the new cell was found to be Cr rich (∼60%) and the other was Cr poor (∼20%) (Figure 1b). Although it is very common for metals to intercalate into van der Waals space of 2D transition metal dichalcogenides, 36 no Cr was found to occupy the octahedral or tetrahedral holes in the van der Waals space. This confirms that Cr is alloying onto the Pt site and not intercalating. In addition, large, well-faceted crystals having dimensions of >5 mm × 5 mm and thicknesses of ∼1 mm can be readily prepared from the melt across the entire series, as shown for PtTe 2 and Cr 0.40 Pt 0.60 Te 2 (Figure 1c andd). Single-crystalline domains can be extracted from the solidified ingot, through mechanical exfoliation. The Cr:Pt stoichiometries in these extracted single crystals of Cr x Pt 1-x Te 2 can vary by x = 0.05 in a particular batch, based on singlecrystal XRD.

Powder diffraction in combination with Rietveld analysis was used to establish the changes in the structure that occur with increasing Cr alloying (Tables S1-S12). A uniform change in the lattice parameters is observed in the XRD patterns across the series (Figure 2a andb). In addition, the transformation from the 1T polytype (x ≤ 0.35) to the 2T polytype (x = 0.45) is apparent based on the emergence of a new 001 peak at 8°2θ in Cr 0.45 Pt 0.55 Te 2 . With increasing Cr content, the interlayer spacing increases from 5.224 Å to 5.490 Å, whereas the inplane a lattice parameter uniformly decreases from 4.025 Å to 3.965 Å (Figure 2c). Overall, the unit cell volume per layer increases from 73.294 Å 3 to 74.732 Å 3 with increasing Cr content (Figure 2d). The uniform changes in lattice constants and volume indicate a homogeneous alloy up to 45% Cr substitution. Synthetic attempts with Cr substitution above 45% resulted in the emergence of a Cr 2 Te 3 impurity phase. Next, Rietveld analysis of the Cr x Pt 1-x Te 2 powders produced a much better fit when Cr was alloyed onto the Pt site rather than intercalated onto the octahedral or tetrahedral hole in the van der Waals space. This matches the single-crystal diffraction analysis and further confirms the absence of Cr intercalation. In all powder patterns with x < 0.45, no reflections of other Cr x Te y phases were observed, including Cr x Te, 37 CrTe 3 , 38 Cr 2 Te 3 , 39 and Cr 5 Te 8 . [40][41][42] Occasionally very weak reflections corresponding to excess Te were observed. Finally, all samples can be stored in air and light for at least eight months without any changes in the observed diffraction pattern.

The magnetic properties of the Cr x Pt 1-x Te 2 compounds were investigated via SQUID magnetometry. The 500 Oe field cooled (FC) magnetic susceptibility of powders from 2 to 300 K are shown in Figures 3a andS1. PtTe 2 was previously reported to have an overall tiny and slightly negative diamagnetic mass susceptibility (-5 × 10 -8 cm 3 g -1 at 90 K), which we confirm here (-17 × 10 -8 cm 3 g -1 at 90 K). Alloying even a small amount of Cr into the PtTe 2 phase leads to a significant increase in magnetic susceptibility. Furthermore, incorporating greater amounts of Cr leads to an increase in both the molar susceptibility and the magnetic transition temperature of the material. Herein, we define T C as the inflection point based on the second derivative of the magnetic It has been well established that in dilute magnetic metals the magnetic coupling can oscillate based on the separation distance of magnetic elements (such as in a Ruderman-Kittel-Kasuya-Yosida mechanism), which we suspect occurs here as well, but will be the subject of future inquiry.

We then investigated the magnetic properties of single crystals of Cr 0.45 Pt 0.55 Te 2 in detail. The 500 Oe FC and zerofield cooled (ZFC) magnetic susceptibility measured with the 500 Oe measurement field applied along the in-plane and cross-plane directions are shown in Figure 3c. When the field is applied along the in-plane direction, the ZFC and FC curves are nearly superimposable. However, when the field is applied along the cross-plane direction, a significant deviation between the ZFC and FC curves is apparent. The cross-plane and inplane orientations of the single crystal both had a T C of ∼191 K, which is relatively close to the Weiss constant (Θ) of ∼211 K obtained from the Curie-Weiss fitting (Figure S3). The inplane and cross-plane M vs H plot is shown in Figure 3d. When the field is applied along the cross-plane direction, there is an appreciable hysteresis with a coercive field of 0.17 T, and the magnetization saturates to ∼2.37 μ B /Cr atom at a much lower applied field. Thus, the cross-plane direction is the easy axis and the direction of spin alignment. The unusual difference between the FC and ZFC susceptibility along the cross-plane direction in Figure 3c arises from the fact that the measurement field is smaller than the coercive field. Finally, the 2.25-2.37 μ B /Cr is relatively close to the expected spinonly moment of a Cr 4+ d 2 spin system, which would be 2.8 μ B / Cr. The assignment of a formal oxidation state to Cr is complicated by the metallic nature of this compound, for which the delocalization of electrons often leads to a reduced magnetic moment. Also, further confirmation of the chromium oxidation state using X-ray photoelectron spectroscopy is confounded from the spectral overlap between the Cr 2p and Te 3d peaks. In order to shed additional light on this question, the magnetic moment per Cr atom was examined with density functional theory (DFT) for an ideal defect-free sample at zero temperature. DFT finds a magnetic moment of 2.72 μB/Cr, giving strong support to the hypothesis of a Cr 4+ d 2 spin system where the expectation is 2.8 μB/Cr. The origin of the magnetization from the Cr d-states can be clearly seen in the spin-resolved density of states in Figure 4. crystallographic phases, respectively, only the P3̅ m′1 magnetic group can lead to ferromagnetic order. This magnetic group requires the Cr spins to be oriented along the c-axis. Rietveld analysis was performed on each data set, and a ferromagnetic structure was required for the 100 and 10 K data sets to achieve a good fit. At 100 and 10 K, the Cr atoms in the Crrich layer had a magnetic moment of 1.0(5) and 1.6(3) μ B /Cr atom, respectively, whereas the Cr-poor layers had 0.8(7) and 1.3(5) μ B /Cr atom, respectively. The 10 K magnetic structure is shown in Figure 5, inset. It is important to point out that a trace (∼3% phase fraction based on neutron Rietveld analysis) Cr 2 O 3 impurity formed in this large-scale synthesis of powder, as evidenced by both in-house lab X-ray and neutron diffraction. The low intensity and minimal overlap of the Cr 2 O 3 reflections did not affect the fitting of the Cr 0.45 Pt 0.55 Te 2 phase in the neutron diffraction, especially since Cr 2 O 3 is an antiferromagnet with a Neél temperature of 308 K. 43 A Rhodes-Wohlfarth ratio (RWR) analysis was used to characterize the amount of delocalization of the magnetic moment within the material. 44 RWR = P c /P s where P c is the magnetic moment calculated from P c (P c + 2) = P 2 eff . P 2 eff is taken from the susceptibility of the paramagnetic phase via a high-temperature Curie-Weiss fitting. P s is the saturation magnetic moment collected from the M vs H plots below the transition temperature of the material. 45 When RWR = 1, the P c and P s values are equal, indicating localized magnetic moments. However, when RWR > 1, this indicates that the magnetic moments are from electrons that are itinerant. 46,47 The P c value was determined to be 4.18 μB/Cr from the Cr 0.45 Pt 0.55 Te 2 cross-plane FC Curie-Weiss plot, while the P s value of 2.37 μB/Cr was taken from the Cr 0.45 Pt 0.55 Te 2 crossplane magnetization curve at 2 K. The calculated RWR of 1.76 suggests that the magnetic moments in Cr 0.45 Pt 0.55 Te 2 are from itinerant electrons. 48 We then investigated the changes in Raman spectra as Cr is alloyed onto the lattice, since Raman spectroscopy has been established as one of the most powerful characterization techniques for 2D materials. In all compounds two major Raman modes are observed, one at 110-125 cm -1 and a second at 150-160 cm -1 (Figure 6a). Generally, as more Cr is alloyed onto the PtTe 2 lattice, the decrease in reduced mass causes these Raman modes to shift to higher frequencies. To further confirm the symmetry of these modes, polarized Raman spectroscopy was performed looking down the c-axis in backscattering geometry. Both the 1T and 2T phases belong to the P3̅ m1 space group, in which the Te atoms occupy the 2d Wyckoff positions that have local C 3v symmetry. Consequently, the 1T phases will have two Raman-active modes: one A 1g and one E g mode. The 2T phases will have four Raman-active modes: two A 1g and two E g modes. In the 2T phase, the Crrich and Cr-poor layers have very similar local structures. Thus, the two A 1g modes will likely occur at similar energies, as will the two E g modes. The A 1g and E g modes are both expected to appear when the polarized filter is parallel to the polarization of the light source, i.e., -Z(XX)Z configuration in Porto's notation. When the polarized filter is perpendicular to the polarization of the light source, or -Z(XY)Z configuration, only the E g modes are expected to be visible. In the polarized Raman spectra of Cr 0.45 Pt 0.55 Te 2 (Figure 6b), the intensity of the Raman mode at ∼155 cm -1 significantly decreases in crosspolarization, confirming that it has A 1g symmetry. It is important to note the absence of additional Raman peaks in the range of 100-200 cm -1 , which would be indicative of TeO 2 , further showing the resilience of these systems against oxidation. 49 As previous studies have found 1T-PtTe 2 to be metallic, four-probe in-plane resistivity measurements were collected on Cr 0.45 Pt 0.55 Te 2 from 300 to 80 K, to determine the influence of Cr alloying on the electronic properties. The resistivities are very low and increase from 7 μΩ m at 80 K to 11 μΩ m at 300 K, indicative of metallic behavior (Figure 7a) as expected from the DOS (Figure 4). The measured thermopowers are also quite small, as would be expected for metallic materials, and vary from 0.9 to 2.4 μV K -1 from 80 to 300 K (Figure 7b). It is well established that the resistance of ferromagnetic metals decreases when a field is applied due to the suppression of spin fluctuations throughout the material. Indeed, below T C , a negative magnetoresistance of 2.5% was observed when a magnetic field was applied along the easy axis (Figure 7c). The magnetoresistance has a quadratic dependence with respect to the applied cross-plane magnetic field.

Finally, to demonstrate the possibility of accessing few to single layers, bulk single crystals of Cr 0.45 Pt 0.55 Te 2 were mechanically exfoliated onto 285 nm SiO 2 /Si substrates using Kapton tape. From this technique, regions with thicknesses of ∼10 nm are readily prepared and visible via optical microscopy (Figure 8a inset). Furthermore, regions down to 1.1 nm were observed near thicker flakes, which corresponds to a thickness of two van der Waals layers (Figure 8a andb). In contrast to many other magnetic phases that exhibit extreme air-sensitivity such as CrI 3 , 15 there are no apparent changes in optical microscopy or AFM measurements indicative of oxidation, such as hole formation, or increases in surface roughness, despite exposure to air for eight months (Figure S4). Future studies of the properties of exfoliated monolayer and bilayers should provide further verification of the resistance against oxidation.

CONCLUSIONS

We have developed a family of exfoliatable, 2D van der Waals metallic ferromagnets, Cr x Pt 1-x Te 2 , that marries the stability of the late transition metal Pt with magnetism from Cr. These compounds are readily synthesized in single-crystal form, simply from cooling the melt, further underlining its thermodynamic stability. It is possible to substitute up to 45% of the Pt in PtTe 2 with Cr, allowing for a robust high transition temperature (T C = 220 K) ferromagnet that, despite eight months of air exposure, shows minimal degradation or changes in properties. This stable 2D ferromagnet is an exciting building block for future explorations in spintronic and magnetoelectronic properties and devices.

EXPERIMENTAL METHODS

Synthesis Methods. Polycrystalline powders and crystals of Cr x Pt 1-x Te 2 (x ≤ 0.45) were synthesized from the elements. Due to the air reactivity of Cr at higher temperatures, stoichiometric amounts of the elemental powders of Cr (99.99%, Alfa Aesar) and Pt (99.98%, Alfa Aesar) with Te pieces (99.999%, Aldrich) were added to a quartz crucible and sealed in a larger quartz tube under a ≤70 mTorr Ar atmosphere. To produce powdered Cr x Pt 1-x Te 2 , the elements were heated to 800 °C over 3 h, held at temperature for a week, and allowed to cool to room temperature over 60 h. To produce single crystals of Cr x Pt 1-x Te 2 from the melt, the elements were heated to 1100 °C over 3 h, held at temperature for 24 h, and let cool to 700 °C at 2 °C/h and air cool below 700 °C to room temperature. Single crystals can be extracted from the ingot via mechanical exfoliation. Single crystals of PtTe 2 were grown via Te flux with a 1:17 Pt to Te ratio. Pt powder and Te pieces were placed in a quartz crucible inside a large quartz tube with quartz wool placed above the crucible and sealed under a ≤60 mTorr Ar atmosphere. The material was heated to 1000 °C over 4 h, held at temperature for 8 h, and then cooled at 2 °C/h to 500 °C. At 500 °C, the tube was removed from the furnace, inverted, and centrifuged at 1000 rpm for 10 min to remove excess Te.

Characterization Methods. After synthesis, all samples were handled and stored in air. A Bruker D8 Advance powder X-ray diffractometer, with Cu Kα radiation, was used to collected X-ray diffraction (XRD) patterns for the Cr x Pt 1-x Te 2 powder series.

The single-crystal X-ray diffraction studies were carried out on a Nonius Kappa diffractometer equipped with a Bruker APEX-II CCD and Mo Kα radiation (λ = 0.710 73 Å). A 0.071 × 0.055 × 0.052 mm piece of a metallic silver block was mounted on a cryoloop with clear enamel. Data were collected at ambient conditions using ϕ and ϖ scans. The crystal-to-detector distance was 40 mm and the exposure time was 5 s per frame using a scan width of 2.0°. Data collection was 100% complete to 25.00°in θ. A total of 3749 reflections were collected covering the indices -5 < = h ≤ 5, -5 ≤ k ≤ 5, -7 ≤ l ≤ 7. A total of 109 reflections were found to be symmetry independent, with an R int of 0.0491. Indexing and unit cell refinement indicated a primitive, trigonal lattice. The space group was found to be P3̅ m1. The data were integrated using the Bruker SAINT software program and scaled using the SADABS software program. Solution by direct methods (SHELXT) produced a complete phasing model for refinement. All atoms were refined anisotropically by full-matrix least-squares (SHELXL-2014).

A Quantum Design MPMS3 superconducting quantum interference device (SQUID) magnetometer was used to measure the molar magnetic susceptibility and magnetization of the powder and singlecrystal Cr x Pt 1-x Te 2 series. The powder series (∼50-100 mg) was placed in a gelatin capsule, sealed with Kapton tape, and inserted into a straw for SQUID measurements. Single crystals were enclosed between two pieces of Kapton tape and attached inside the straw via a cutout with Kapton tape for both in-plane and cross-plane orientation measurements. For M vs H plots, the magnetization data were collected at 2 K by cooling the material to temperature and then sweeping the applied field from 0 T up to 4.5 T, down to -4.5 T, and back to 4.5 T. To account for the demagnetization factor when measuring along the cross-plane direction, a correction factor was applied to the external field using H i = H e -NM where H i is the intrinsic field, H e is the applied external field, N is the shape factor (which is equal to 4π for a sheet that is perpendicular to H e ), and M is the measured magnetization.

DFT calculations were performed on a 2 × 5 × 1 60-atom Cr 0.45 Pt 0.55 Te 2 2T supercell generated by the mcsqs code of the Alloy Theoretic Automated Toolkit (ATAT) 50 as a special quasi-random structure (SQS) 51 with different Cr concentrations in the two layers as measured. The structure was relaxed, and electron density of states (DOS) and magnetization were calculated with the Vienna ab initio Simulation Package (VASP) 52 using the plane wave projector augmented-wave (PAW) method 53 and the PBE 54 functional, together with van der Waals corrections. 55,56 The Brillouin zone was sampled on a 4 × 4 × 4 k-point grid, and a plane-wave energy cutoff of 400 eV was used.

Time-of-flight neutron diffraction on powders was collected using the POWGEN beamline at Oak Ridge National Laboratory. Frames 1 and 3 were collected and utilized to characterize the magnetic structure of Cr 0.45 Pt 0.55 Te 2 at 250, 100, and 10 K.

Raman scattering spectra were collected using a Renishaw Raman IR microprobe with an inVia confocal Raman microscope, with a 633 nm laser light source and a charge-coupled detector. For polarized Raman measurements, a Pike Technologies KRS-5 polarizer was placed before the charge-coupled detector.

Four-probe temperature-dependent in-plane resistivity and thermopower measurements were performed simultaneously on a single crystal of Cr x Pt 1-x Te 2 using a home-built setup in a liquid N 2 cryostat. A small brass sheet was attached to the top of the crystal underneath a resistance heater, and a second brass sheet was attached to the bottom of the crystal and connected to an alumina heat sink. Copper-constantan thermocouples were attached at two points along the crystal with Ag epoxy to measure the temperature gradient. The copper legs of the thermocouples were also used to measure the thermopower voltage. For resistivity measurements, current was applied through the top and bottom brass sheets, while the copper legs were used to measure the voltage. The same setup was also used for magnetoresistance measurements, but with an applied magnetic field ranging from ±1.4 T in the cross-plane direction.

Atomic force microscopy measurements were performed on exfoliated flakes of Cr 0.45 Pt 0.55 Te 2 that were prepared via Kapton tape exfoliation onto 285 nm SiO 2 /Si substrates. A Bruker AXS Dimension Icon atomic/magnetic force microscope with ScanAsyst was used for thickness and few-layer stability measurements.