Induced 2H-Phase Formation and Low Thermal Conductivity by Reactive Spark Plasma Sintering of 1T-Phase Pristine and Co-Doped MoS2 Nanosheets.

Pristine and Co-doped MoS2 nanosheets, containing a dominant 1T phase, have been densified by spark plasma sintering (SPS) to produce a nanostructured arrangement. The structural analysis by X-ray powder diffraction revealed that the reactive sintering process transforms the 1T-MoS2 nanosheets into their stable 2H form despite a significantly reduced sintering temperature and time testifying to the fast kinetics of phase change. Together with the phase conversion, the SPS process promoted a strong texturing of the nanosheets, which drives additional scattering processes and alters the electronic and thermal transport properties. In the pristine sample, it produced one of the lowest thermal conductivities ever reported on MoS2 with a minimal value of 0.66 W/m·K at room temperature. The effect of Co substitution in the final sintered samples is not significant, compared to the pristine MoS2 sample, except for a non-negligible improvement of the electrical conductivity by a factor of 100 in the high-Co content (6% by mass) sample.

Introduction

The layered transition-metal dichalcogenides (TMDC) are two-dimensional (2D) materials with the formula MX2 (where M = group IVb, Vb, or VIb transition metal and X = S, Se, or Te). This material family has been intensively screened due to the substantial variety of transport (electronic and thermal) and structural properties.[1−4] Their layered structures are built as an assembly of two-dimensional covalently bonded X-M-X layers separated by a van der Waals gap leading to their crystallization in various polytypes such as 1T, 2H, 3R, 4H, and 6R phases (where the numeral quantifies the number of X-M-X layers per unit cell along the c-axis, while T, H, and R indicate trigonal, hexagonal, and rhombohedral symmetry, respectively).

Among this broad family of compounds, MoS2 is considered as a versatile material due to its various physicochemical and structural properties as it changes from bulk to nanoscales.[5] Its major properties comprise high carrier mobility at room temperature and a layer-dependent bandgap varying from an indirect bandgap of 1.2 eV to a direct bandgap of 1.9 eV, which make it suitable for possible electronic and optoelectronic device applications. Its common polytypes are 2H-MoS2 (space group: P63/mmc; a ≈ 3.16 Å; c ≈ 12.30 Å),[5] considered as the most stable form, the metastable 1T-MoS2 (space group: P3̅m1; a ≈ 3.19 Å; c ≈ 5.95 Å)[6] and its intermediate polymorphs 1T′, 1T″ 1T‴ and 3H-MoS2 (space group: R3m; a ≈ 3.17 Å; c ≈ 18.38 Å).[5] Nowadays, the synthesis of a single/multi-layer MoS2 is easy, which opens a large gateway for the development of new materials based on two-dimensional (2D) MoS2. Indeed, 2D nanomaterials and thin films have recently drawn an increasing interest of the scientific community due to the technological development for realizing and using nanostructures in various fields such as sensor, catalysis, and thermoelectricity.[7−14] In the thermoelectric (TE) field, the material TE efficiency is quantified by the dimensionless figure of merit zT = PF × Tk–1 = S2σTk–1 (PF, power factor; T, absolute temperature; S, Seebeck coefficient; σ, electrical conductivity; k, thermal conductivity). The nanostructuring approach presents a major interest due to the possibility to enhance the figure of merit zT by simultaneously reducing the thermal conductivity with selective phonon scattering effects and, possibly, increasing the Seebeck coefficient by the quantum confinement effect on carriers.[15−18] In particular, the thermal conductivity of layered materials have been of interest due to its anisotropic and interfacial effects.[19−25] The native bulk MoS2 is reported to be an intrinsic p-type semiconductor with a large Seebeck coefficient (S ≈ 500–700 μV/K@RT), but its poor electrical conductivity limits the overall power factor.[26,27] Combined with its relatively large thermal conductivity, the figure of merit zT is still reported to be negligible in comparison to other sulfide TE materials with values rarely exceeding zT = 0.15 even in the high-temperature range (T ≈ 1000 K).[28] However, theoretical studies have estimated a high potential at room temperature of the MoS2 nanoribbons with a predicted zT of about 3. This extraordinary value is obtained from mainly a high electrical conductivity induced by the reduction of the gap resulting in a strong edge reconstruction and gives an insight of the MoS2’s potential for TE applications by nanostructuring and/or enhancing electrical transport properties.[29] In a recent report, the calculated electronic structure revealed that a MoS2 monolayer doped with Co becomes half-metallic due to the impurity band formation near the Fermi level. Considering the favorable substitution of Co in the Mo site by forming an S vacancy, it is also expected that the charge balance conservation will lead to hole doping, enabling the possibility to increase the carrier concentration of the MoS2 and improve its electrical conductivity.[30,31] Moreover, the Co doping in MoS2 has been reported, theoretically and experimentally, to induce dilute magnetism in the natively non-magnetic MoS2.[30−32] Similar magnetism has also been reported to constructively promote a magnetic enhancement of the Seebeck coefficient, leading to a better TE performance.[33,34]

In a previous study, the development of Co-doped MoS2 nanosheets composed of a dominant metastable 1T phase has been achieved with promising catalytic properties.[35] In the present study, we have investigated the densification of these 1T-MoS2 nanosheets by reactive spark plasma sintering, which constitutes a nanostructure-engineering strategy for the development of the TMDC in contrast to the classical film approach usually employed.[36−39]

Results and Discussion

Structural Characterization

The XRD pattern of the as-prepared undoped MoS2 nanosheets before densification suggests that the sample is poorly ordered and exhibits increased basal spacing due to NH3/NH4+ intercalation (Figure S1A). The as-prepared Co(6%)-doped MoS2 nanosheets (Figure S1A) exhibits a broad (002) reflection at 11.0 Å, indicating the presence of a guest species in the interlayer. The guest entity could, possibly, be NH3/NH4+ ions released as byproducts of hydrazine used as a reductant in the hydrothermal reaction. The asymmetric 2D reflections at 2θ = 33 and 57° reveal the presence of stacking faults within the few-layered Co-doped MoS2. The SEM image (Figure S1B) of the as-synthesized Co(6%)-doped MoS2 indicates that the sample consists of clusters of sheets.

Figure displays the room temperature powder XRD patterns of the MoS2 samples after reactive SPS sintering. The main diffraction peaks indicate that all the samples agree with the trigonal prismatic polymorph 2H–MoS2 structure type (space group: P63/mmc; a ≈ 3.16 Å; c ≈ 12.30 Å) rather than the 1T–MoS2 (Figure a) as highlighted by their respective simulated pattern in Figure b.[6] Interestingly, the dominant 1T phase of the initial powder of MoS2 nanosheets (Supporting Information, Figure S1) has been fully converted into the 2H phase after a short SPS process (only 5 min of dwell time), indicating that the phase change kinetics between the trigonal antiprismatic and prismatic geometry is very fast.[35] Indeed, the 2H polytype is the thermodynamically stable form of TMDC-MoS2 and it has been already reported that temperatures above 70–100 °C induces the 1T polytype to turn into 2H form, which is correlated experimentally with a strong endothermal signal.[5,40−42] A comparable signal has been confirmed in our current native nanopowder (Figure S2). The dynamical process of the transition between 1T/2H phases involves intra- and interlayer atomic plane gliding, which is caused by atom displacement due to extra thermal energy.[43] The 2H form has a hexagonal lattice with a threefold symmetry and an atomic stacking sequence (S–Mo–S′) of ABA. Each Mo atom in the 2H phase lies in a center prismatically coordinated by six surrounding S atoms, with the S atoms in the upper layer lying directly above those of the lower layer (Figure a). In contrast, the Mo atom in the 1T phase is octahedrally coordinated to six neighboring S atoms, with an atomic stacking sequence of ABC, where the bottom S′ plane occupies the hollow center of the top S lattice. For the 1T-to-2H transformation, the basic process involves a sulfur atomic plane gliding to hollow center sites due to thermal energy. The activation energy of this phase transition mechanism is estimated to be 400 ± 60 meV (38 ± 6 kJ/mol).[44] The diffraction peaks corresponding to the (00l) indexation of the 2H-MoS2 structure appear sharper and stronger than other peaks (as highlighted in Figure b), especially in the pristine sample, which are consistent with the preferential orientation due to the favorable ordering along the stacking direction. The other indexations (hkl ≠ 00l) are characterized by a large full width at half maximum (FWHM), suggesting a reduced crystallite size or a possible strain in the structure. These observations have been confirmed and discussed later with the analysis by scanning electron microscopy (SEM) of a fractured cross section along the SPS pressure axis (Figure ). With the increase in Co doping, the main diffraction peaks of the 2H-MoS2 phase see their intensities significantly reduce and simultaneously, new diffraction peaks emerged from the background (* in Figure b). It clearly indicates that Co contained in the native nanosheets promotes the formation of byproduct phases during the reactive SPS step. This secondary phase seems to correspond to the CoS structure (space group: P63/mmc; a ≈ 3.37 Å; c ≈ 5.17 Å).[45] The formation of the secondary phase is consistent with the expected high reactivity of the raw powder composed of metastable nanosheets. Indeed, during the densification step, it has been confirmed that the sintering temperature has to be significantly reduced at 1223 and 1073 K for the pristine and the Co-doped samples, respectively, to reach the final consolidation step and to avoid the decomposition of the samples (Figure S3). However, below the sintering temperature, all the samples present sintering curves characteristic of a high reactivity process with an intermediate step (blue highlight in Figure S3). This intermediate step occurs below the pyrometer control, which makes it difficult to prevent the byproduct formation.

Figure 1

(a) Representation of the 1T and 2H crystal structure of MoS2 and (b) X-ray powder diffraction patterns of MoS2 samples after SPS.

Figure 3

SEM micrograph of a fractured cross-sectional surface normal to the SPS pressure axis of (a) MoS2, (b) Co(3%)-doped MoS2, and (c) Co(6%)-doped MoS2 samples after SPS.

The energy-dispersive X-ray spectroscopy (EDS) elemental mapping (Figure and Figures S4–S6) sustains the increasing amount of Co consistently with the expected doping level (Figure a) but reveals a disruption in the sample homogeneity in the Co-doped samples, attesting the by-product formation. The EDS point analysis indicates a representative composition close to CoS stoichiometry (Figure S7) in line with the XRD observation. Unfortunately, the limitation of the analysis accuracy did not allow us to obtain reliable composition values due to the fact that the characteristic energy peaks of Mo (Lα, 2.292 eV) and S (Kα, 2.309 eV) are too close and produced an uncertainty of the Mo/S quantification. Nevertheless, we can observe a (00l) peak shifting of the 2H-MoS2 phase to the lower angle with the increasing content of Co, indicative of the c lattice expansion (Figure b). It suggests that the Co2+ dopant is still present in the MoS2 lattice/interlayer. In previous studies, it has been demonstrated that Co2+ occupy the S vacancies in basal planes as well as the unsaturated S-edges produced by the hydrothermal synthesis of the MoS2 nanosheets and confirmed experimentally by XPS analysis.[35,39,46,47] The Le Bail XRD refinement[48] of the (002) peak allows obtaining of a reasonable c parameter estimation of 2H-MoS2 (space group: P63/mmc; a ≈ 3.16 Å; c ≈ 12.30 Å), which reveals that the average spacing between MoS2 layers expands from 12.327(2) to 12.509(1) Å, respectively, for the pristine and the Co(6%)-doped MoS2 samples. This is consistent with the increasing presence of Co in the interlayer planes of the MoS2 phase. The EDS confirmed the increasing presence of Co in the MoS2 matrix of the doped samples (Figure a) with the characteristic energy peaks (Lα, 0.776 eV and Kα, 6.924 eV) rising in the Co(3%)- and Co(6%)-doped samples compared to the pristine MoS2 sample (Figure b).

Figure 2

(a) Elemental analysis composition and (b) polished surface SEM image with the corresponding EDS spectrum of the MoS2 samples after SPS.

The microstructure analysis by SEM imaging of the sample’s cross section (Figure ) confirms the oriented microstructure suggested by the XRD analysis. A strong texturing normal to the SPS pressure axis can be observed in the pristine sample (Figure a) and Co(3%)-doped sample (Figure b). In addition, it shows that the lateral dimensions of the sheets have significantly increased after SPS. The initial 1T nanosheets were a few-layer thick and a few hundred nanometers in the lateral dimension.[35] However, after SPS, the sheets have grown in the lateral dimensions into the micrometer range. It indicates that the initial nanosheets have, likely, merged for favoring the formation of a larger sheet along the in-plane axis. The anisotropic microstructure observed is consistent with the 2D-layered structure of 2H-MoS2, wherein the weak van der Waals interaction between the layers will allow the layers to propitiously slide over one another due to the uniaxial pressure. The texturing is visibly reduced with the increasing Co content due to the formation of the secondary phase. Some inclusion of equiaxial grains, undoubtedly the secondary phase, appears in the Co-doped sample and becomes a non-negligible part of the microstructure in the Co(6%)-doped MoS2 as displayed in Figure c. In this latter case, it seems that the MoS2 layers are still present but, between layers, the equiaxial grains form a large agglomerate and break partially the global texturing.

Electrical Transport Properties

To observe the influence of the texturing and the phase change from 1T to 2H on the electrical properties, the temperature dependence of the in-plane electrical transport properties has been probed and the results are displayed in Figure . All the samples revealed a noticeable n–p-type transition behavior with the increasing temperature, indicated by the change of the sign of the Seebeck coefficient. This behavior seems to be in good agreement with the previous report on pristine bulk-MoS2 composition.[28,49] At room temperature, all the samples exhibited a negative Seebeck coefficient (Figure a), indicating that electrons (n-type) are the major carriers and turn to positive values at higher temperatures (T > 500 K) indicative of the holes (p-type) dominating the transport properties. The band structure of the bulk 2H-MoS2 is reported with an indirect bandgap of 1.29 eV and a Fermi level lying in the top of the valance band representative of the material’s native p-type character.[50,51] To explain the observed n-type conduction at room temperature, we hypothesize that the reactive sintering process causes partial sulfur volatilization, leading to an excess of Mo remaining in the van der Waals gap as already reported in other TMDC.[52,53] Self-intercalated Mo will therefore provide additional electrons to the system, leading to an extrinsic n-type character at room temperature. Then, the n–p-type transition implies that the temperature rising activates hole carriers and turns the character to the native p-type semiconductor at a medium temperature in agreement with the intrinsic MoS2 band structure. The chaotic trend, visible between 300 and 400 K in the pristine and Co(3%), is attributed to the expected competition between the native p-type behavior of the 2H-MoS2 phase and the extrinsic n-type behavior produced by self-intercalation. The magnitude of the Seebeck coefficient is higher in the pristine sample, with values varying from S = – 126.14@300 K to S = + 339.68@775 K, compared to the Co-doped samples, with values from S = – 2.59@300 K to S = + 39.02@775 K for Co(6%)-doped MoS2, for example. The variations of the Seebeck coefficient magnitude are associated with an apparent increase in the electrical conductivity in the Co-doped samples, especially for the larger content of 6% of Co (Figure b). According to the Mott equation prediction, the electrical conductivity increases with the carrier concentration and/or mobility and, reversely, the Seebeck coefficient decreases with the increase in carrier concentration.[54] It suggests that the current Co doping may increase the carrier concentration/mobility in our samples, which therefore can explain the drastic enhancement of the electrical conductivity together with the Seebeck coefficient decreasing. This experimental trend has been predicted by the electronic structure calculation performed on monolayer Co-doped MoS2, showing the formation of impurity bands related to the Co substitution, which turns the behavior from the semiconductor to half-metallic.[31] In addition, so far, there is no report on CoS transport properties but some Co–S binary phases have been reported with n-type metallic transport properties.[55] It is, therefore, not excluded that the secondary phase affects the electrical transport properties by affecting the carrier concentration/mobility. To confirm the electrical behavior trend in a better way, Hall measurements were attempted to determine the charge carrier concentration in the pristine and doped samples. Unfortunately, the successive measurements remained unsuccessful or yielded inconsistent values attributed to several issues: the large sample resistance (pristine sample), the presence of the secondary phase (Co-doped samples), and the n–p-type transition attesting to bipolar conduction near room temperature. All the samples show a positive temperature dependence of σ, agreeing with the semiconducting state of the phase. The pristine sample is characterized by a higher σ = 11.7 S/m@300 K compared to the literature value (σ ≈ 0.1–1 S/m@300–325 K) consistent with probable superior carrier mobility promoted by the texturing (Figure a). However, the absolute values remains too low for providing a large thermoelectric PF (Figure S8a). Despite the large σ enhancement, the PF of the Co(6%)-doped MoS2 sample reduces significantly at high temperatures, compared to the pristine sample, due to the decrease in the Seebeck coefficient magnitude (Figure a). The maximum PF is obtained for the pristine sample with the highest values of 0.18 μW/cm·K2@775 K.

Figure 4

Temperature dependence of (a) Seebeck coefficient S and (b) electrical conductivity σ of the MoS2, Co(3%)-doped MoS2, and Co(6%)-doped MoS2 samples after SPS.

Thermal Transport Properties

The measured out-of-plane thermal conductivity as a function of temperature is depicted in Figure a. The pristine sample exhibits an intrinsic low k varying between 0.66 and 0.71 W/m·K in the entire temperature range. These values are far lower than the native thermal conductivity of the MoS2 single crystal (85–110 W/m·K in the basal plane)[56] and are one of the lowest ever reported for polycrystalline bulk samples (Table ). Indeed, the thermal conductivity at room temperature of MoS2 is reported to range from 34.5 W/m·K for monolayer MoS2 to 52 W/m·K for 11-layer MoS2.[57,58] Some recent papers report thermal conductivities below 1 W/m·K in polycrystalline thin films with a record lowest value of 0.27 W/m·K reported on polycrystalline films with perfect controlled grain orientation.[38,59,60] Experimentally, the out-of-plane thermal conductivity of bulk MoS2 at 300 K is reported with values in a common range of 2.2–6.1 W/m·K and the lowest value of 1.05 W/m·K has been reported in restacked compounds with a similar 2H phase in the room temperature range.[28,49,61] This current state of the art ranks our samples in the lowest range of MoS2 thermal conductivity ever reported. The significantly reduced thermal conductivity is consistent with the high texturization degree of the native nanosheets, which will consequently promote an enhanced phonon scattering. The increase in the phonon scattering is likely to be larger with the increased density of grain boundaries in the direction perpendicular to the pressure axis. The mean free path of phonon is then reduced due to the increase in the intergranular thermal resistance (Figure b). The rising grain boundary density, coupled with the possible presence of planar defects induced by the high kinetic conversion of the 1T phase into the 2H phase, like stacking faults, is possibly relevant additional contributors to the increase in phonon scattering.

Figure 5

(a) Temperature dependence of the thermal conductivity of MoS2, Co(3%)-doped MoS2, and Co(6%)-doped MoS2 after SPS and (b) schematic representation of the out-of-plane enhanced phonon scattering induced by the texturing.

Table 1

Representative Literature Comparison of the MoS2 Thermal Conductivities

formatting	MoS2 sample type	direction of measurement	κ (W/m·K)	temperature (K)	reference
single crystal	single crystal	basal	85–100	300	(56)
		c-axis	2–2.5	300	(56)
nanolayer	monolayer	basal	34.5	300	(57)
	multi-layer	basal	44–52	300	(58)
thin film	polycrystalline nanomembrane	in plane	0.75	300	(38)
	nanoflake film	in plane	1.5	300	(60)
	film with controlled grain orientation	out of plane	0.27–2	300	(59)
bulk (* correspond to the pristine MoS2)	MoS2 with VMoS4 nanoinclusion	in plane	16–40*	300	(28)
		out of plane	4.2 - 6.7*	300	(28)
	MoS2 with MoO2 nanoinclusion	in plane	20–40*	327	(49)
		out of plane	3.1–5.5*	327	(49)
	exfoliated and restacked MoS2	out of plane	1.05*	300	(61)
	this work	out of plane	0.71*–1.24	300

The Co-doped samples are characterized by a slightly larger thermal conductivity than the pristine sample, which monotonously increases with the Co content. The origin of this phenomenon is likely governed by two interconnected contributions. First, the secondary phase formation (CoS) induced by the Co doping breaks the texturing of the sample and therefore induces a rise of the thermal transport along the out-of-plane axis (Figure c). Second, it is not excluded that this byproduct phase might be intrinsically characterized by a larger thermal conductivity than MoS2.

Finally, the dimensionless figure of merit dependence on temperature, zT, has been quantified and revealed a direct dependency of the zT with the respective PF of each sample (Figure S8). Up to 600 K, the zT remains low in all series before rising up to the highest value of 0.02@775 K in the pristine sample.

Conclusions

We have achieved densification and texturing of 1T-MoS2 nanosheets by reactive spark plasma sintering. This approach enables the possibility to significantly reduce the sintering temperature (<1000 °C) and time (∼5 min dwell time) as well as promoted a strong texturing of the MoS2 nanosized layers as observed through XRD and SEM analysis. We highlight that the conversion of the metastable 1T to the stable 2H form is extremely fast and fully achieved by reactive densification. We report the lowest values of the out-of-plane thermal conductivity of the MoS2 bulk material of 0.66–0.71 W/m·K in the pristine sample due to the nanosheet texturing. The Co-doped MoS2 has been found to favor the formation of composite materials enhancing the electrical conductivity of the system by 100 times in the case of a large Co content. However, it induced a negative effect in the thermal conductivity and the magnitude of the Seebeck coefficient, which restricted the final zT.

Experimental Section

Preparation of Co-Doped MoS2 Nanosheets

The cobalt-doped MoS2 nanosheets were prepared as described previously.[35] An aqueous solution (45 mL) of a mixture of cobalt acetate (0.107 g) and ammonium tetrathiomolybdate (0.442 g) was stirred for 30 min. The second precursor acts as a source of ammonium, sulfur, and molybdenum in the reaction. At the end of 15 min, hydrazine hydrate (5 mL) was added to the solution and the stirring was continued. The black-brown solution was hydrothermally reacted in a Teflon-lined stainless-steel autoclave at 180 °C for 24 h and cooled to room temperature under ambient conditions. The black precipitate that formed was washed with distilled water till the pH of the washings was ∼7 followed by washing with acetone. The product was dried in air at ambient temperature. The preparation was repeated using 0.054 g of cobalt acetate to vary the cobalt content in the product. Cobalt contents were found to be ∼6 and 3% by mass in the samples prepared using 0.107 and 0.054 g of cobalt acetate, respectively,[35] and these samples would be called hereafter Co(6%)-doped MoS2 and Co(3%)-doped MoS2.

As a control experiment, the synthesis was repeated in the absence of cobalt acetate, which results in ammoniated MoS2 nanosheets.

Densification by Spark Plasma Sintering

The ground powders were densified by spark plasma sintering (Dr. Sinter, SPS-322Lx) in a Ø10 mm graphite dye at various temperatures for 5 min of dwelling time under a uniaxial pressure of 50 MPa (heating and cooling rate of 100 K min–1). The temperatures steps were 1223 and 1073 K, respectively, for the pristine MoS2 and the two Co-doped MoS2 samples (Figure S2). The densities measured by Archimedes’ method were 4.52, 4.58, and 4.47, respectively, for the pristine, Co(3%)-, and Co(6%)-doped samples. It corresponds to ≥90% relative densities if normalized to the standard MoS2 density. The sintered pellets were then cut and polished to the required shapes and dimensions for various measurements.

Characterization

The crystal structures of the sintered pellets were examined using X-ray powder diffraction (Rigaku Smart Lab 3 diffractometer) with Cu Kα radiation. Data were collected over a 2θ range of 10–120° with a step size of 0.02° and a step time of 2°/min. Le Bail fittings were performed using the FullProf program included in the WinPLOTR software.[48,62,63] The shape of the diffraction peaks was modeled using a pseudo-Voigt profile function. Zero-point shifts, asymmetry parameters, and lattice parameters were systematically refined, and the background contribution was manually estimated. Observations of microstructural aspects of the sintered samples were performed on the fractured cross section and polished surface using a Hitachi SU-4800 scanning electron microscope (SEM) and a mini-SEM (TM3000, Hitachi) both equipped with an energy-dispersive spectrometer (EDS).

The thermal diffusivity α and heat capacity Cp were measured using LFA-467 Hyperflash (Netzsch) under a flowing argon atmosphere (50 mL/min). The thermal conductivity κ was derived as a product of the sample’s density (measured by Archimedes’ method), thermal diffusivity, and heat capacity Cp. The measurements of electrical resistivity ρ and Seebeck coefficient S were performed simultaneously using a commercial instrument Ulvac ZEM-2 under partial helium pressure.