Intercalation based synthesis of Chevrel phase superconductors

Abstract

This work focuses on the superconducting and magnetic properties of Chevrel phase (CP) compounds (Cu₂Mo₆S₈ and Ni₂Mo₆S₈) processed via a novel self-propagating high-temperature synthesis (SHS) method involving ternary cation intercalation in MoS₂ high-temperature polymorphs. Green pellets of the precursor materials sealed under vacuum and exposed to 1050 °C or 1200 °C rapidly transformed from a powder mixture into submicron CPs within minutes. X-ray diffraction analysis confirmed the presence of Cu₂Mo₆S₈ CP (67%), and Ni₂Mo₆S₈ (93%). This group has illustrated the first time CP were formed at low energy cost and full stoichiometry. Magnetization versus temperature data obtained for the samples under different field conditions confirmed the superconducting behavior of the Cu–CP at a critical temperature of 8.3 K. The Ni–CP behaved as a superparamagnetic at low temperature (4.2 K). The results of this work suggest that the novel SHS process offers a viable scalable synthesis route for the Chevrel family of compounds and that it is worth exploring the functional applications of the new compounds of this material system.

Introduction

Chevrel phase (CP) compounds were discovered in 1973 when Dr. Marcel Sergent and Roger Chevrel synthesized compounds via solid state synthesis at elevated temperatures inside of a fused silica tube using MoS₂ and a third element (M) [1]. The formula of the CP family of compounds is illustrated as M_xMo₆Z₈, where M is a ternary cation, and Z is either sulfur, selenium, or tellurium; x may either fall between zero and two for bivalent M or between zero and four for monovalent M. The unique structure of CP (discussed in detail below) allows for the insertion of over 40 different elements as ternary cations, which can be used to tune the specific properties of the compound for different applications leading to a multifunctional material. Due to this flexibility, CP compounds have been investigated for applications as thermoelectrics [2, 3], battery electrodes [4, 5], electrocatalysts [6, 7], superconductors [8, 9], and others. Although superconductivity in CP has been observed before, superconducting compounds had not been synthesized via self-propagating high-temperature synthesis (SHS), which has the potential for industrial scalability over other methods of CP synthesis. Therefore, it is critical to test SHS viability if CP compounds are to be pursued for applications in the superconductivity industry.

Self-propagating high-temperature synthesis refers to a reaction in which precursor materials are ignited and spontaneously transformed into the product due to the reaction’s highly exothermic nature [10]. Specifically, a combustion front propagates through the reactants with a certain velocity to transform the reactants into the product [11]. A novel SHS process was illustrated in Gouma’s previous work, which involved a multi-step reaction mechanism for the rapid and sustainable processing of 3D and 2D ternary chalcogenides [12]. This reaction mechanism is attributed to a synergetic action of polymorphic, intercalation, and exfoliation processes taking place via the following stages: (1) MoS₂ phase transformation; (2) Ternary cation intercalation; (3) Exfoliation; (4) CP formation [13].

SHS was employed for this experiment to demonstrate a low energy method of synthesizing CP compounds. Other methods of synthesis include solid state [1, 14, and 15], molten salt [16], solution precipitation [17], high energy mechanical milling [18], and microwave [19]. While, there are many options available for synthesizing CP, these methods include drawbacks that would prove difficult in scalability. For example, solid state, molten salt, and solution precipitation syntheses all require lengthened furnace times of up to 60 h [16, 17, 20, 21], which can cost large amounts of energy. Furthermore, while high energy mechanical milling and microwave synthesis do not require these extended heating times, they do suffer from consistent incomplete phase transitions [18] and the need for more complex and/or expensive equipment (that has its own difficulties in upscaling) [19] that reduce their viability for scaling to industrial levels. Due to these factors, SHS is the most promising method for scalability as it uses a simple mechanism of mixing commercially available dry powders, pressing, vacuuming, and heat treating; it requires reduced heat treatment times leading to lower energy inputs; and it has shown viability in creation of fully transformed samples [12].

For this specific SHS route, MoS₂ was chosen as a precursor over just elemental Mo and S to allow for an intercalation-based synthesis method. Bulk MoS₂ has three major polytypes with 1T (trigonal), 2H (hexagonal), and 3R (rhombohedral) symmetry. The 2H phase is thermodynamically stable at room temperature and the metastable phases arise from highly non-equilibrium thermodynamic conditions. Transition metals can easily intercalate through the Van der Waals gaps in MoS₂. This process usually involves electron transfer from the guest intercalate to the higher energy unoccupied d band of molybdenum. The negatively charged S layer is considered to be a driving force for the coulombic attraction between the intercalates and Mo. The steps involved in the intercalation process include adsorption of ternary cations (intercalate) on the surface of the MoS₂ crystal, formation of an activated complex, followed by rapid diffusion around the edge, with subsequent entry into the Van der Waals gaps of MoS₂. For example, intercalation of Fe in MoS₂ forms the layered compound FeMo₂S₄. In Gouma’s earlier work, formation of FeMo₃S₄, hypothesized through intercalation of Fe through MoS₂, was observed within just 1 min of residence time in the SHS chamber [13]. This intercalation-based SHS process allows for the rapid synthesis of CP compounds from a low energy method that could be highly desirable for commercial applications.

Experimental method

Material synthesis via SHS

Powders of Cu (99.5% 0.8 µm, US Research Nanomaterials), Ni (99.9% 1–1.5 μm, SkySpring Materials), Mo (99.9% 0.5–0.8 µm, SkySpring Materials), and MoS₂ (Super Fine Grade, Climax Molybdenum) were used in synthesizing the CPs. Powders with a ratio of 4Cu:2Mo:4MoS₂ to a total weight of 10 g were vigorously mixed using a Janke and Kunkel laboratory grinder. 2 g of the mixture was then measured out and mechanically pressed into a cylindrical pellet of 0.5-inch diameter with 15000lbs of force. A novel reaction chamber using a quartz tube and a vacuum adapter was employed to encapsulate the pellet in vacuum in an effort to remove oxygen and moisture from the reaction environment (see Fig. 1 below). Additionally, alumina wool was added into the reaction chamber to assist with thermal isolation.

Figure 1

Novel reaction chamber with encapsulated pellet sample (circled). Built-in thermocouple, added alumina wool, and connection for vacuum attachment have been annotated.

The encapsulated 4Cu:2Mo:4MoS₂ sample in the reaction chamber was introduced into a general–purpose tube furnace (Thermo Scientific Lindberg) at 1050 °C for 20 min before the reaction chamber was removed and allowed to cool to room temperature. Once cooled, the vacuum was released under a fume hood to avoid vapor exposure. The same process of grinding, pressing, and encapsulating was used for a ratio of 2Ni:2Mo:4MoS₂. The Ni sample was heat-treated at 1200 °C for 10 min before removal.

XRD

Portions of the heat-treated samples were ground into a fine powder using a mortar and pestle for powder X-ray diffraction (XRD). XRD was carried out at the center for electron microscopy and analysis (CEMAS) using a Rigaku SmartLab X-ray diffractometer (Bragg–Brentano unit). The tube voltage and current were set to 40 kV and 15 mA, respectively. An angle range of 10°–60° with a step size of two per min was used for 2θ values. Resulting spectra were analyzed using MDI JADE and the ICDD database.

SEM

Powder samples were also observed via scanning electron microscopy (SEM) to analyze morphology and crystallite size. Processed samples were imaged using the Thermo Scientific Apreo II SEM and the Thermo Scientific Quattro environmental scanning electron microscope (ESEM) at CEMAS. Image analysis was completed using ImageJ.

SQUID magnetometer

In order to fit within the ~ 5 mm homogeneous region of the superconducting quantum interference device (SQUID) magnetometer, each pellet was fractured, and a piece of suitable size was attached to the inside of a plastic drinking straw using GE 7031 varnish. The straw was cut so the sample was approximately 66 mm from the holder end, then attached to a sample holder rod and lowered into a Quantum Design MPMS 3 SQUID magnetometer. The temperature was set to 2 K with zero applied field (zero-field-cooled condition, or ZFC), after which a 5 m T field was applied and the sample was centered by sensing the moment as a function of vertical position. Measurements of magnetic moment were performed in Vibrating-Sample magnetometer (VSM) mode with an amplitude of 5 mm and frequency of 11.7 Hz. The magnetic moment was measured versus temperature, sweeping from 1.8 K to 10 or 20 K at a rate of 3 K/min. The sample was cooled under applied field (field-cooled or FC) and then moment versus increasing temperature was measured again. The field was then reset to zero. This sequence was repeated for applied magnetic fields of 1 T for the Cu sample and 1, 3, 5, and 7 T for the Ni sample. A magnetization versus field (M vs. H) measurement was performed, where the field was swept from zero, to a positive value, then to a negative value, and then to positive (five quadrants) at a rate of 10 m T/s. Moment versus temperature was then measured up to 313 K at a faster sweep rate. An M versus H measurement was also conducted at 300 K. The susceptibility was obtained by dividing the measured moment by the sample volume.

Results

XRD

Figure 2 below illustrates the XRD patterns for the 4Cu:2Mo:4MoS₂ sample (Fig. 2a) and 2Ni:2Mo:4MoS₂ sample (Fig. 2b). As can be seen, the 4Cu:2Mo:4MoS₂ sample was mostly transformed into the Cu₂Mo₆S₈ CP with a presence of MoS₂. Additionally, in this sample was a presence of Cu₂O due to a side reaction during the heat treatment process (discussed later). For the 2Ni:2Mo:4MoS₂ sample (Fig. 2b), Ni₂Mo₆S₈ CP is the prominent phase with some presence of MoS₂. No additional Ni peak was present in the scan as during heat treatment, some stray Ni powder adhered to the quartz tube enclosure.

Figure 2

XRD spectrum of samples. a 4Cu:2Mo:4MoS₂ (blue) sample was transformed into Cu₂Mo₆S₈ (magenta) with a presence of MoS₂ (red) and Cu₂O (cyan). b 2Ni:2Mo:4MoS₂ (blue) sample was transformed into primary phase Ni₂Mo₆S₈ (magenta) with a presence of MoS₂ (red).

Figure 3 displays the wt% of each phase as calculated by JADE via peak intensities for each sample. As shown, the CP for the Ni sample is much more prominent than the respective CP of the Cu sample. Further investigation is ongoing to determine if cation particle size and/or time is a main contributor to this difference.

Figure 3

Relative weight fraction of phases as calculated by JADE for a 4Cu:2Mo:4MoS₂ and b 2Ni:2Mo:4MoS₂.

SEM

ImageJ was used to analyze the crystallite size of the CPs as well as the MoS₂ platelets in the 2Ni:2Mo:4MoS₂ sample. Figure 4 below exhibits the morphology of the 4Cu:2Mo:4MoS₂ sample. The morphology is uniform with faceted crystals of a size ranging between 700 nm and 2.2 μm.

Figure 4

SEM images of 4Cu:2Mo:4MoS₂ sample.

Figure 5 illustrates the morphology of the 2Ni:2Mo:4MoS₂ sample. There are large Ni CP crystals in the range of 900 nm–2.2 μm along with MoS₂ platelets in the range of 1.2 um–3.6 μm. Also, observed is a large amount of sintering and coarsening, suggesting that the heat treatment time of 10 min is rather long for this particular reaction.

Figure 5

SEM images of 2Ni:2Mo:4MoS₂ sample.

SQUID magnetometer

The M versus T data indicates superconductivity in both samples. At low temperature, the sample cooled in zero-field conditions (ZFC) exhibited shielding currents once field was applied due to the Meissner effect which opposed the applied magnetic field, giving rise to a magnetization. However, in the field cooled condition, the Meissner effect would be expected to cause flux expulsion below T_c; the fact that the magnetization remains near zero is indicative of significant flux pinning.

The critical temperature T_c was chosen as the temperature above which the gap between the zero-field-cooled and field-cooled curves went to zero, in the case of the Cu CP, or stopped rapidly converging, in the case of the Ni CP. The modified definition was necessary in the case of the Ni sample because above T_c the sample is paramagnetic and there remains a gap between the ZFC and FC curves. See, Table 1 below for T_c values of the two samples.

Table 1 Superconducting transition temperatures T_c [K]

The M versus T data for the 4Cu:2Mo:4MoS₂ sample (Fig. 6) shows that the material is superconducting up to 8.3 K in 5 m T applied field and 7.0 K in 1 T field. The shape of the M versus H loop (Fig. 7) is also indicative of a type II superconductor with flux pinning. The magnetization increases with decreasing magnitude of the applied field, but then goes rapidly to near zero in the range ± 0.2 T, a behavior consistent with flux jumps and likely due in part to a large size of the superconducting regions (the tendency to flux jump is proportional to the critical current density times the sample size). While, a single flux jump was recorded in each pass through the low field in this case, measurements on similar Cu CP samples showed multiple flux jumps.

Figure 6

Susceptibility of 4Cu:2Mo:4MoS₂ versus temperature over a temperature range of a 2–20 K and b 2–313 K. The curves measured under field-cooled and zero-field cooled conditions are indicated in a. The inset in a shows a positive bump in susceptibility surrounding T_c. The cause of this is unknown.

Figure 7

Magnetic moment normalized to sample volume of 4Cu:2Mo:4MoS₂ versus applied magnetic field at 4.2 K with a field sweep rate of 10 m T/s.

Figure 8 below illustrates the susceptibility versus T measurement for the 2Ni:2Mo:4MoS₂ sample and shows a slight difference between the ZFC and FC curves at 5 m T below ~ 3.4 K, consistent with a superconducting transition, though this ΔM is orders of magnitude smaller than in the Cu sample. In Fig. 9 the M versus H measurement shows behavior which might indicate a super paramagnet or spin glass at 4.2 K. At 300 K, the sample is paramagnetic with a ferromagnetic component that saturates around 0.1 T, which may be present in the 4.2 K data as well. The susceptibility versus T data above T_c, at 5 m T, is well within the range that the ferromagnetic component would dominate.

Figure 8

Susceptibility of 2Ni:2Mo:4MoS₂ versus temperature for various applied magnetic fields over a temperature range of a 2–10 K and b 2–313 K.

Figure 9

Magnetic moment normalized to sample mass of 2Ni:2Mo:4MoS₂ versus applied magnetic field at 4.2 K with a field sweep rate of 10 m T/s.

Figure 10 below illustrates the M versus H data for both samples at 300 K.

Figure 10

M versus H for both samples at 300 K. Both samples show paramagnetic behavior. The Ni sample appears also to have a ferromagnetic component.

Discussion

Chevrel structure

The multifunctionality of the CP family stems from the unique building blocks of Mo₆Z₈, in which Mo atoms form an octahedron with each atom placed slightly outside of a surrounding distorted cube of Z atoms (see Fig. 11a). Each corner of each Mo₆Z₈ cube lies opposite of the face center of the adjacent cubes, leading to close contacts of the six Mo atoms of one unit and the Z atoms of the surrounding six units. In the full structure, each Mo atom in a unit is located close to five Z atoms, creating a square base pyramid. The four Z atom base of such a pyramid constitutes the face of a single Mo₆Z₈ unit, and the fifth apex Z atom is a part of an opposite Mo₆Z₈ unit. Therefore, the 6f Z atoms in a building block belong to bases of the pyramids and are simultaneously the apexes of six pyramids, while the remaining 2c Z atoms belong only to the square faces of the pyramids [6].

Figure 11

a Mo₆Z₈ single unit. b Full Mo₆Z₈ structure with labeled cavities.

This layout provides different cavities in the chalcogenide network. The largest (Site 1 in Fig. 11b) is at the origin of the rhombohedral unit cell and creates a distorted cube with eight atoms from eight different Mo₆Z₈ units. The second and third cavities are much smaller; the second cavity (Site 2) is located between two of Site 1, and the third cavity (Site 3) is located between two consecutively misaligned Mo₆Z₈ units [6]. Ternary cations will fill specific cavities depending on their size and create channels of ternary cations due to the interconnected network. This network allows for the insertion of over 40 elements as ternary cations. Figure 12 illustrates a model of the CP structure of the formula M₄Mo₆S₈.

Figure 12

Structure of M₄Mo₆S₈.

Naturally, having different sites in the Mo₆S₈ can result in different CP structures. Singstock et al. [22] used machine learning to predict the placement of over 50 elements in the Chevrel structures for Mo₆S₈, Mo₆Se₈, and Mo₆Te₈. According to Fig. 13a below, for Mo₆S₈ the Cu atoms have a near even affinity towards Site 1 center and Site 1 off-center (Site 2 in Fig. 11) with minimal chance of occupying Site 2 (Site 3 in Fig. 11). Conversely, Ni shows a greater affinity towards Site 1 off-center and a less chance to occupy Site 1 center or Site 2. From this analysis larger cations prefer Site 1, while smaller cations prefer Site 1 off-center. In Mo₆S₈, Site 2 is predicted to be reserved mainly for 4D and 5D transition metal intercalants. In addition to changes in stability for an element in varying sites, Singstock also predicted that changes in the electronic structure may occur. For example, some CPs may transition from metallic to semiconducting or from nonmagnetic to antiferromagnetic as the ternary cation site changes.

Figure 13

Cation prediction sites for different Chevrel structures. a Mo₆S₈. b Mo₆Se₈. c Mo₆Te₈. Reprinted with permission from [22] copyright 2024 American chemical society.

Within this study, the unique electronic and magnetic properties of Cu and Ni in the cluster compounds was shown, but further work is needed to establish the specific property effects of Cu and Ni on the clusters.

SHS process

After the synthesis, the CP present for the 4Cu:2Mo:4MoS₂ sample was found to be Cu₂Mo₆S₈. This phase (from [23]) was confirmed to most closely match the experimental XRD peaks, which was also compared to reference scans for Cu₂Mo₃S₄ [24], Cu_1.73Mo₆S_7.9, Cu_0.67Mo₂S_2.53 [25], CuMo₆S₈ [26], as well as a few other copper–deficient stoichiometries. These other scans displayed peak variations from the 4Cu:2Mo:4MoS₂ experimental sample especially around 2θ = 45°. E.g., at this degree there are two prominent peaks shown in the experimental sample and the Cu₂Mo₆S₈ reference (at 45.70° and 46.20°), but for Cu₂Mo₃S₄ this is one large peak (45.70°) and for Cu_1.73Mo₆S_7.9 (at 45.20°, 45.60°, and 46.40°), this is three smaller peaks. Cu_0.67Mo₂S_2.53 contains the closest peak position to the experimental, however the relative peak intensities are off compared to the closer matched Cu₂Mo₆S₈ phase.

Similarly, the 2Ni:2Mo:4MoS₂ sample was compared to reference XRD scans for Ni_0.7Mo₃S₄ [27], NiMo₃S₄ [28], Ni_2.5Mo₆S_6.7 [29], and Ni₂Mo₆S₈ [30]. Ni_2.5Mo₆S_6.7 shows two very small peaks around 2θ = 33.30° and 32.60° that do not appear in the experimental sample, and Ni_0.7Mo₃S₄ showed a small peak shift that was not seen in the sample. Although NiMo₃S₄ and Ni₂Mo₆S₈ are the same stoichiometrically, there exists a very small difference in a peak at 2θ = 45.70°, where in NiMo₃S₄ a second peak begins to emerge out of the side at 45.90°. This second peak is not seen in Ni₂Mo₆S₈ or in the experimental sample leading to the conclusion that Ni₂Mo₆S₈ is the CP present.

While, the samples were not transformed into 100% CP, the SHS process was shown to be viable for CP formation. One noted result was the presence of Cu₂O in the 4Cu:2Mo:4MoS₂ sample. This phase was found to be due to a rather prominent amount of Cu₂O contaminant within the Cu powder precursor (See Fig. 14 below). The XRD scan below shows the presence of about 31.1% Cu₂O within the powder along with 17.7% Na₂SO₄. The Cu₂O is believed to have caused the low Chevrel yield as the Cu₂O could not participate in the intercalation mechanism. The peaks for Na₂SO₄ are expected to have been minor enough in the final product that they fell within the background of the spectrum. Nb., this copper powder was re-sourced for ongoing experiments and the purity has been confirmed.

Figure 14

a XRD of copper powder precursor. b Weight fraction of the different phases present as calculated by JADE.

Superconducting properties

Copper-based CP (Cu_xMo₆S₈) is stable for a range of copper content [10]. Two low temperature phases are reported: LT1 with a critical temperature T_c which increases from 10.0 K at x = 1.75 to 10.9 K at x = 1.85, and LT2, with a T_c which decreases from 6.3 K at x = 3.1 to 4.0 K at x = 3.3 [11]. See below in Table 2 for a list of selected common superconducting materials for comparison. The T_c of 8.3 K in our sample lies between the T_cs of the two LT phases, and no composition is shown in [11] to have that T_c. The reason for the difference in our T_c is unknown at this time, but T_c is known in other materials to be affected by composition and lattice strain. The magnetization versus field loop for the Cu CP shows large hysteresis consistent with flux pinning behavior. The loop widths decrease to near zero but have long tails before finally closing at the irreversibility field B_irr, which is > 4 T.

Table 2 Critical temperatures and fields of selected superconducting materials

In addition to the superconducting behavior, there also appears to be a paramagnetic component. This background makes determination of the upper critical field, B_C2, difficult, since in magnetometry measurements it is commonly determined by the point, where the slope of the M–H loop goes from positive to zero.

While, the low T_c and B_C2 rule out these samples for use in applications such as high-field magnets, a T_c above 4.2 K enables use in liquid helium cooled environments without the complexity involved in obtaining lower temperatures and could be useful in electronics, sensors, or other novel applications.

Our susceptibility measurement for the 2Ni:2Mo:4MoS₂ sample and showed a slight difference between the ZFC and FC curves at 5 m T below ~ 3.4 K, consistent with a superconducting transition, though this ΔM is orders of magnitude smaller than in the Cu sample. This is interesting, because the Ni-containing compound is one of several CP materials (M = Cr, Mn, Fe, Co, Ni) which is not known to be superconducting because even small amounts of magnetic impurity are enough to destroy superconductivity in most such materials [12]. We also observe superparamgnetic signatures, and an apparent additional soft ferromagnetic behavior. Paramagnetism has been reported previously in Ni containing CP materials using tellurium instead of sulfur [20].

Conclusions

CPs of Cu₂Mo₆S₈ and Ni₂Mo₆S₈ were processed via intercalation-assisted SHS process and were tested for their magnetic and superconducting properties. The 4Cu:2Mo:4MoS₂ sample was shown to be superconducting up to 8.3 K in the 5 m T field and decreased to 7.0 K in 1 T applied field. The Cu CP performed approximately within the T_c range of Cu CP with different compositions, synthesized via other methods. Even though the T_c is still relatively low compared to more conventional superconducting materials, it was worth assessing the novel microstructures. The 2Ni:2Mo:4MoS₂ sample showed some evidence for a superconducting response below ~ 3.4 K. Also, we saw superparamagnetic behavior at low temperature, which is consistent with other (ternary) Ni chalcogenides. Additionally, the samples were found to have ferromagnetic or paramagnetic properties. This background made the determination of the upper critical field via magnetometry difficult. A T_c above 4.2 K enables use in liquid helium cooled environments without the complexity involved in obtaining lower temperatures. The energy efficient and novel rapid synthesis route demonstrated here to produce compounds of the CP family may open the pathway for the novel configurations and properties of these interesting cluster compounds.