Synthesis and Characterization of Ultrathin FeTe2 Nanocrystals.

Ultrathin crystals including monolayers have been reported for various transition-metal dichalcogenides (TMDCs) with van der Waals bonds in the crystal structure. Herein, we report a detailed synthesis procedure and characterization of ultrathin iron ditelluride crystals. This material crystallizes in an orthorhombic marcasite Pnnm crystal structure whose bonding is dominantly covalent and without loosely connected van der Waals (vdW) bonds, making monolayer FeTe2 synthesis less straightforward than other TMDC monolayer syntheses. The chemical vapor deposition synthesis process described is simple, effective, and results in a range of crystal thicknesses from approximately 400 nm down to the FeTe2 monolayer.

Introduction

Iron ditelluride (FeTe2) is one of the several transition-metal dichalcogenide crystals first synthesized and observed by Steffan Tengner in 1938.[1] FeTe2 is known to form in either marcasite-type or pyrite-type crystals, depending on the synthesis conditions. The orthorhombic marcasite structure is the more common crystal phase formed under ambient conditions, while the cubic pyrite phase is formed through high-temperature and high-pressure synthesis techniques.[2]

Marcasite FeTe2 adopts either a Pnnm(1,3) or Pnn2-type[4] orthorhombic unit cell, where Fe is octahedrally coordinated by Te. Density functional theory found no significant energy difference between the two arrangements.[5] Early studies show that FeTe2 crystals have an anisotropic antiferromagnetic ground state below about 80 K and a possible ferromagnetic (FM) state below approximately 30 K.[6−8] The electronic structure features a reported band gap of 0.35–0.67 eV, as deduced from electrical transport measurements.[7,9] FeTe2 thin films have been found to feature a low thermal conductivity value of 1.20 W m–1 K–1.[10] Electron-doped marcasite FeTe2 crystals are a good candidate for thermoelectric applications due to the sharp increase in density of states around the band edges, which favors a large increase in the thermoelectric power factor S2σ, where S is the thermopower and σ is the electrical conductivity.[5]

FeTe2 nanoparticles have previously been synthesized using solution-based methods, where all reactions take place between 120 and 300 °C. Low-temperature hydrothermal synthesis produced 25 nm diameter spherical nanoparticles[11] and nanorods[12] that can be indexed to orthorhombic FeTe2. FM FeTe2 nanocrystallites[13] were also prepared using a hydrothermal method. Hot-injection solvothermal methods under atmospheres of inert gasses like nitrogen and argon have produced FeTe2 nanorods,[14] a colloidal solution of FeTe2 nanocrystals,[15] and magnetic spherical FeTe2 crystallites that aggregate into treelike architectures.[16] Irregular FeTe2 nanocrystallites[13,16] with dimensions on the order of several hundred nanometers were found to be ferromagnetic at room temperature. These solutions utilize relatively mild temperatures and deliver crystals of variable morphologies. However, they are also relatively time-consuming, require complicated precursor solution syntheses prior to the FeTe2 formation reaction, and necessarily involve filtration, washing, and drying/resuspension steps. These steps not only add extensive hands-on involvement but can also be a source of contamination.

The development of a more facile synthesis method for FeTe2 nanocrystals is of high interest, particularly methods that can produce relatively clean materials for future studies. It has been known for some time that nanocrystals with vdW bonds in the crystal structure, such as TMDCs, can be produced using chemical vapor deposition methods.[17−21] The synthesis of ultrathin TMDC nanocrystals has enabled many important discoveries.[22−24]

Since neither marcasite nor pyrite crystals are vdW solids, monolayer FeTe2 was formerly considered nonexistent.[25] This conclusion was reasonable, as it can be seen in Figure that neither the 1T nor 2H atomic layer structures common in TMDCs are directly derived from the Pnnm space group crystal unit cell. However, multiple first-principles studies, where monolayer FeTe2 was studied,[5,26−29] concluded that it is possible for this material to be energetically stable in monolayer form in ambient conditions based on calculated phonon-dispersion frequencies. When both the octahedral (T) and trigonal-prismatic (H) coordinations of Fe were compared, it was found that the H phase had lower energy than the T phase,[2] making it the more thermodynamically favorable of the two. As a result, most first-principles calculations of monolayer FeTe2 consider the H phase.

Figure 1

Comparison of marcasite Pnnm space group with the 1T (P31m, CdI2-type) and 2H (63/, NbS2-type) crystal structures of transition-metal dichalcogenides. Note the absence of van der Waals bonds in the marcasite crystal structure.

The magnetic state of monolayer FeTe2 is predicted to be either FM or ferromagnetic with calculated moment values in the range from 1.08 to 2 μB.[5,28,29] Magnetocrystalline anisotropy (MCA) that favors the in-plane direction[28] and the MCA dependence on an applied electric field for multiple MTe2 compound monolayers[27] were predicted. The FeTe2 easy axis flips from in-plane to out-of-plane as the applied electric field increases, giving monolayer FeTe2 excellent potential for applications in magnetic memory devices. Magnetic anisotropy has also been noted in experiments on bulk FeTe2 crystals.[7]

There is some debate about the electronic character of the FeTe2 monolayer. Two studies found it to exhibit semiconducting behavior, with band gaps of and 0.87[2] and 0.35 eV,[5] which are on the order of those observed in TMDCs. However, other studies found FeTe2 monolayer to exhibit metallic behavior[25,26,28] and half-metallicity,[27,29] making it a potential candidate for applications in spintronics as spin valves. Gudelli et al.[5] also found in their first-principle theoretical study of the thermoelectric properties of monolayer FeSe2 and FeTe2 that marcasite FeTe2 is a good thermoelectric material that favors electron doping. Its hole and electron concentrations were calculated to be 9.2 × 1019 and 1.4 × 1021 cm–3 for the thermopower values of 96 and −74 mV K–1 at 300 and 600 K, respectively. FeTe2 also had calculated relaxation times of 2.3 × 10–14 and 3.1 × 10–14 s at those respective temperatures.

It is clear that FeTe2 is a material with promising qualities for future applications that remains relatively unstudied experimentally. With the lack of efficient large-scale syntheses of this material being an obstruction of progress in this field, it is vital to further develop its nanomaterial production methods. In this work, we present a chemical vapor deposition process to synthesize FeTe2 nanocrystals of various sizes down to monolayer thickness. This more facile process allows for nanocrystal synthesis directly on the substrate from simple powder precursors in a relatively short interval of time. Crystal structure and unit cell parameters of deposited nanostructures were investigated through X-ray diffraction. A scanning electron microscope (SEM) was used to obtain images confirming nanoparticle deposition, lateral sizes, and morphologies. The energy-dispersive X-ray spectroscopy (EDS) capability of the SEM was used to determine the chemical composition of nanostructures within the sample. Raman spectroscopy also confirmed that FeTe2 phase crystals formed in this experiment. Atomic force microscopy (AFM) data was used to calculate and compare the lateral and vertical size dimensions of deposited nanostructures.

Results

X-ray Diffraction and Energy-Dispersive Spectroscopy

Figure shows the X-ray diffraction scan of the substrate after deposition. All diffraction peaks can be indexed in the orthorhombic Pnnm (58) space group, with the calculated lattice parameters a = 5.262(2) Å, b = 6.2608(2) Å, and c = 3.871(2) Å. These values correspond well to the reported standard values[10,13] with no impurity present.

Figure 2

X-ray diffraction pattern for deposited FeTe2 crystals. The inset shows the FeTe2 crystal structure.

Diffraction peaks can be seen at 28.49° (020) and 34.01° (200), and weak peaks at 22.05° (110), 27.11° (011), 44.00° (211), 44.98° (220), and 58.92° (040). The strongest peak is at 28.49° (020), and the peak for its parallel plane (040) is visible as well. Due to the orthorhombic lattice structure of this crystal, the (200) plane is also symmetrical to the (020) and (040) planes. The symmetry of the two largest peaks indicates the preferential orientation of crystals on the substrate with their {0k0} faces aligned to the substrate direction, or faces of the type {200} or {100}. The presence of peaks for the (110) and (220) planes, which are symmetrical, as well as the (211) and (011) planes, indicate that although there is preferential orientation along the (010) plane, the deposits have a certain range of preferred crystallization.

Scanning Electron Microscopy

Figure a–e consists of several medium- and high-magnification SEM images of the crystallized deposition products on the substrate. As can be seen in the medium-magnification image (a), the chemical vapor deposition process delivers results with a high dispersibility and nanostructures that range in size and morphology. The nanoparticles produced in this reaction can be separated into three classes based on the order of their largest lateral dimension: those with dimensions greater than 1 μm, those with dimensions between 100 nm and 1 μm, and those with all lateral dimensions less than 100 nm.

Figure 3

(a–e) Scanning electron microscope images at various magnifications of the as-synthesized FeTe2 nanocrystals. (f, g) Elemental mapping of target nanocrystals for tellurium (f) and iron (g) concentrations.

Crystal morphology was not found to be size-dependent, with multiple configurations present throughout the sample. The most common structure, seen in Figure a,b, consists of oblong plates with rounded edges. These elongated oval or rectangular structures are seen across all size ranges, with their length being their largest dimension. A less common morphology, visible in the bottom right corner of Figure a,c, is that of the elongated hexagonal particle, which is often darker in color than the oblong structures.

More common than the hexagonal particles are structures like those seen in Figure c, where deposits form from the clusters of smaller oblong and spherical particles. The spherical crystallites that make up the deposit in Figure c can also be seen individually in Figure d. Their lateral dimensions are all less than 100 nm, but their relative brightness indicates a significant thickness. The morphology that makes up the majority of the deposited particles in Figure d consists of oblong nanoplates with irregular edges. These also have lateral dimensions in the tens and hundreds of nanometers, but their dimness indicates a relatively lower vertical dimension. Most deposited structures have one elongated dimension. Elemental mapping (Figure f–g) done through energy-dispersive X-ray spectroscopy (EDS) confirms that deposited nanostructures are composed of both iron and tellurium.

The ratio of iron (Fe) to tellurium (Te) in structures of different sizes and morphologies as observed through EDS can be seen in Figure . The oblong structures (Figure b) have a Fe/Te ratio of 1:2, whereas deposits in (c) and (d) show the width of formation with Fe/Te ratio of 1:1.81 and 1:2.16, respectively. This indicates that different morphologies may result from minor fluctuations in molar ratio. Based on these observations and those from the diffractogram in Figure , synthesized nanocrystals exhibit the orthorhombic marcasite FeTe2 crystal structure. The deviation from ideal stoichiometry could arise from the interference with the silicon substrate, which could generate a stronger signal than the relatively thin deposited nanoparticles. The relatively high percentages of oxygen seen only in Figure d could be an evidence of oxidation and may indicate size-based air instability.

Figure 4

Energy-dispersive X-ray spectroscopy spectra taken on a substrate before deposition (a) and on selected crystals (b–d).

Atomic Force Microscopy

Figure shows topographic atomic force microscopy images of the substrate surface with varying scan sizes. Each scan captures information about the vertical dimensions of the different size classes of deposited nanostructures. Images in Figure a,b, with scan sizes of 20 μm, confirm the high dispersibility of the nanostructures as seen through SEM. Their line scans show the vertical dimensions of features crossed by the matching line across the image. Nanocrystals with a dimension greater than 1 μm were found to have vertical dimensions ranging from 100 to 250 nm, and even up to 400 nm (not shown). Nanoparticles in the class with maximum lateral dimensions equal to or less than 1 μm have thicknesses ranging from 10 to 150 nm.

Figure 5

Atomic force microscope images at two different areas (a, b) with scan sizes of 20 μm with line scans corresponding to the horizontal red lines across each image.

The nanocrystals measured in Figure are significantly smaller, with lateral dimensions under 250 nm. Those measured in Figure a can be compared to the bright, round particles seen clustered together in Figure c. These particles have maximum lateral dimensions between 88 and 214 nm, and thicknesses in the range of 12–50 nm. For all particles discussed thus far, their height ranged between 10 and 33% of their largest lateral dimension. However, the particles in Figure b were found to have a significantly reduced height to length ratio.

Figure 6

Atomic force microscope images with scan sizes of (a) 2.0 μm and (b, c) 500 nm and their accompanying line scans. (c) Boxes to designate the borders of each scanned nanoplate and an inset that shows a proportionate scanning electron microscope image of nanoplates for comparison.

The particles in Figure b appear to be the flat, irregularly shaped FeTe2 nanoplates seen in Figure d. These crystals are in the lowest size range of those observed for both the SEM and AFM, and these images are directly compared in Figure c. These particles have smaller lateral dimensions than those in Figure a, with lengths measured between 43 and 62 nm. Their vertical dimensions are significantly lower, with thicknesses between approximately 400 pm and 2 nm, or between 0.5 and 3% of their lateral dimensions. These measurements are on the order of the lattice parameters found when analyzing the sample diffraction pattern, with the smallest lattice parameter measuring 386 pm. This indicates that the nanocrystals in Figure b,c are single to quadruple molecular layers of FeTe2 unit cells.

Raman Spectroscopy

Raman peaks for bulk FeTe2 crystals were previously reported at 119 and 137 cm–1 at room temperature.[7] A 154 cm–1 peak was also observed at low temperature,[7] and at room temperature in a different study.[30] The first two peaks are attributed to the B1g mode associated with antiphase Te–Te vibration, whereas the 154 cm–1 mode is associated with A1g in-phase mixed Te–Te stretching related to mutual Fe–Te motion within FeTe6 octahedron and is sensitive to changes in bond lengths.[7,30] Raman spectra in this work (Figures and 8) support the X-ray diffraction and energy-dispersive X-ray conclusion that FeTe2 phase crystals were formed in this experiment.

Figure 7

Raman peaks of FeTe2 nanoplates with lateral dimensions above 1 μm of oblong (a, b) and two different rounded shapes (c–f). (g) Two-dimensional mapping at 520.7 cm–1 Si crystalline Raman peak. Since FeTe2 nanoplates block the Si Raman signal, the shape and dimension of the dark feature in the mapping are used to identify the nanocrystal. (h–j) Raman intensity peak mapping for the crystal (e). Position dependence of Raman peaks of FeTe2 (k–m); scale identical to (g).

Figure 8

(a) Two-dimensional mapping at 520 cm–1 of Si crystalline Raman peak, analogous to Figure g. (b) Raman peaks, (c, d) Raman intensity, and (e–j) position peak mapping of oblong FeTe2 nanoplates with lateral dimensions below 1 μm.

We note that the elongated, oblonglike nanocrystals of several micron size show somewhat lower A1g frequency and somewhat higher B1g frequency when compared to bulk crystals (Figure a,b). In contrast, both frequencies are slightly increased in the rounded crystals (Figure c–f) of about 1 μm size. Interestingly, peak intensity mapping of one such round-shape nanoplate (Figure g–j) reveals that the broad B1g mode (Figure f) splits into weak intensity bulk crystal like 137 cm–1 and strong intensity ∼141 cm–1 vibration, whereas the weak A1g mode shows a small shift to higher frequencies compared to bulk crystal. The selected spot Raman spectroscopy suggests that the A1g mode tends to diminish near the crystal edge (Figure k–m).

Both B1g (136 cm–1) and A1g (155 cm–1) modes were observed in nanoplate crystals with lateral dimensions below 1 μm (Figure a,b), confirming the synthesis of submicron FeTe2 nanoplates. Both modes are more intense near submicron crystal edges (Figure c,d), even though position mapping indicates the presence of both peaks along the crystallite length (Figure e–j). This could suggest that the FeTe2 nanoplate growth mechanism involves the reaction of Fe vapor from sublimed FeCl2 (Scheme ) with Te powder particle of several hundred micron sizes, where FeTe2 growth is facilitated along the lateral edges of Te powder. As nanoplates grow to micron (Figure e–m) and oblong rodlike shapes of several micron lateral lengths (Figure a–d), small Raman peak shifts imply changes in Fe–Te and Te–Te bond lengths in mesoscale FeTe2.

Scheme 1

Schematic Diagram for FeTe2 Chemical Vapor Deposition Synthesis

The bond lengths and coordination numbers of surface atoms and bulk atoms are usually different; therefore, a shift in Raman frequencies could arise due to competing influences of surface atom undercoordination and bond-length changes. The undercoordination of surface atoms has a great impact on their vibrational frequency.[31,32] The above argument can be applied to the subtle changes of Raman peak shift seen due to the variety in size and shape among the nanoplates and between nanoplates and bulk. In particular, the atoms in crystal corners are usually undercoordinated when compared to the atoms in bulk and surface. Hence, the change in coordination number in the corners of nanoplates may influence the Fe–Te motion mode in Figure i,j. The peaks for the Te–Te mode are still observed at the corners, suggesting that the corners of the FeTe2 plates are mainly terminated by Te atoms. The FeTe2 plates in Figure show a higher peak intensity mode of A1g Fe–Te motion compared to the B1g Te–Te mode.

The corner or edge effects are more obvious in smaller plates or elongated, oblonglike nanocrystals, so the peak intensity of the Te–Te mode is relatively higher than that of the Fe–Te mode, as shown in Figure a,b. For both single-selected spot Raman spectroscopy and 2D mapping over the surface areas of FeTe2 plates, both Fe–Te motion (∼151 cm–1) and Te–Te mode (∼137–140 cm–1) peaks are observed and comparable, suggesting elementary stoichiometric structure in the FeTe2 nanoplates.

Discussion

FeTe2 nanocrystals are of high interest for application in the new generation of hybrid energy storage devices such as lithium–ion capacitors and sodium–ion batteries that feature high power and energy density and a good cycle retention rate.[33,34] Further experiments on in situ chemical vapor deposition of FeTe2 ultrathin crystals in porous carbon network are therefore of high interest. Moreover, since FeTe2 monolayers are predicted to preserve a ferromagnetic state due to magnetocrystalline anisotropy arising from spin-orbit coupling,[28] further magneto-optic experiments are of interest to test theoretical predictions. Two-dimensional materials with ferromagnetic ground states attract great interest as they are possible candidates for spintronic and memory devices but can also be valuable in fundamental science studies.[35−37] Finally, since reproducible switching behavior without crystal structure changes has recently been reported in FeTe2 nanostructures,[38] it would be of interest to investigate the size dependence of switching properties in future devices made from FeTe2 ultrathin crystals.

Conclusions

In this experiment, FeTe2 nanocrystallites of varying sizes and thicknesses down to the monolayer were grown using a facile chemical vapor deposition process. The deposited nanoparticles showed a high dispersibility and lateral dimensions ranging from less than 100 nm to above a micrometer. The most common crystal morphology is that of an oblong nanoplate with rounded edges. The observed nanocrystallites have stoichiometries corresponding to FeTe2.0(2). The X-ray diffractogram of this sample can be indexed to that of orthorhombic marcasite FeTe2 with a Pnnm space group and no crystalline impurity phase present. The preferential orientation of deposited nanocrystals along the (010) plane is indicated by differences in relative peak intensities. Synthesized nanocrystallites have thicknesses ranging from picometers to 200 nm or between 0.5 and 33% of their largest lateral dimension. This indicates that the chemical vapor deposition process in this work is effective at synthesizing nanocrystalline FeTe2 with disparate thicknesses down to the monolayer level on one substrate. Comparison of shape- and size-dependent Raman spectra suggests that FeTe2 crystallites grow from the reaction of sublimed Fe on Te submicron particles on lateral edges. Future work of interest includes studies of chemical defects that could influence magnetic properties, chemical stability, and protection of dangling covalent bonds as well as size-dependent studies of magnetism in reduced dimensions using optical methods.

Experimental Section

Materials and Equipment

Anhydrous iron(II) chloride (FeCl2) powder (99.5% Alfa Aesar) and Tellurium (Te) powder (99.99% Alfa Aesar) were used as chemical precursors for the synthesis of FeTe2. Metal chloride powder precursors are preferable to their pure metal counterparts due to significantly increased vapor pressures at lower temperatures. For example, FeCl2 has a vapor pressure of 0.1 mmHg at 490 °C and 1 mmHg at 570 °C,[39] whereas pure iron requires a temperature of 1412.5 °C to reach a vapor pressure of 0.1 mmHg.[40] Tellurium, however, has a vapor pressure of 1 mmHg at the relatively low temperature of 520 °C[39] and therefore can be used in pure form. The deposition substrates used were silicon thermal oxide wafers with a 285 nm silicon oxide layer (MTI Corporation), with lateral dimensions of 5 mm and a thickness of 0.5 mm. The precursors and substrate were placed in rectangular alumina crucibles of varying dimensions. The synthesis was performed inside a horizontal single-zone tubular furnace with a heating chamber 1 in. in diameter and 12 in. in length. A quartz process tube was used to contain the process. Experiments were done under a constant gas flow of 3% H2 in Ar.

Synthetic Procedures

There are multiple crystalline phases of iron telluride that form based on synthesis temperature and precursor stoichiometry.[3,42] The marcasite structured ε-phase has been reported to have a homogeneity range from 66.1 to 67.4% tellurium and an accepted peritectic formation temperature of 649 °C.[42−44] The ε-FeTe2 also forms from a eutectoid reaction at temperatures ranging from 514 to 517 °C.[43]

Scheme presents the experimental setup for the chemical vapor deposition of FeTe2. FeCl2 and Te powders were obtained and used without further purification. They were placed in a single alumina crucible with a molar ratio of 1 mole FeCl2/3 moles Te. The powder precursor masses were separated from each other by 3 cm to control their respective temperatures during the reaction. The crucible was placed in the furnace with the Te directly above the heating element and the FeCl2 powder 3 cm upstream. Placing the FeCl2 at a greater distance from the deposition substrate was done to lower the ratio of iron that reached the substrate compared to tellurium and to induce a 1:2 deposition stoichiometry on the substrate.

The substrate was removed from a sealed plastic sleeve and placed vertically at an angle inside a smaller crucible, with its polished face directed upstream. This crucible was placed, so the substrate was 8.5 cm downstream from the heating element. Crucible placement was designed to induce temperatures of 545, 550, and 520 °C on the FeCl2, Te, and substrate, respectively, based on the measured temperature gradient using an external thermocouple since the result of synthesis is sensitive to the temperature of precursors and substrate. These temperatures were chosen based on precursor vaporization rates and the formation temperature for FeTe2. The ends of the furnace’s quartz tube were then closed, and the chamber flushed with 3% hydrogen balance argon carrier gas for 10 min prior to the start of the chemical vapor deposition reaction.

The chemical vapor deposition was performed under a constant carrier gas flow rate of 50 cubic centimeters per minute. The furnace was heated to a peak temperature of 550 °C from room temperature over 15 min at a rate of about 35 °C min–1. This temperature was maintained for 30 min to induce precursor vaporization and deposition on the substrate, as well as the nucleation and growth of FeTe2 crystals. After 30 min, the furnace was shut off and the system cooled under a constant gas flow. The furnace cooled naturally from 550 to 170 °C, at which point the furnace casing was opened for more rapid cooling. The substrates were removed from the furnace once it reached room temperature and placed in a glovebox to prevent oxidation. This entire process was completed over the course of approximately 4 h.

Characterization Methods

X-ray diffraction data was obtained using a Rigaku Miniflex with Cu Kα (λ = 0.15418 nm) radiation. The obtained diffraction pattern was analyzed using Rietica software. Scanning electron microscopy and energy-dispersive X-ray spectroscopy data were obtained using a JEOL JSM-7600F scanning electron microscope.

An Asylum Research MFP 3D operated using noncontact AC Air topography mode was used to take atomic force microscopy images and measure the surface topology of the sample. No sample treatment was necessary prior to the measurement for this experiment.

Raman experiments for single-selected point Raman spectrum acquisition and 2D Raman spectral imaging were performed using a WITec confocal Raman microscope alpha 300 equipped with a solid-state laser (λ = 532 nm), an electron multiplying CCD detector, and a 100×/0.9NA objective lens. Unpolarized Raman scattered light was focused onto a multimode fiber and monochromator with an 1800 line/mm grating. Instrument calibration was verified by checking the position of Si at 520.7 cm–1.