1. Introduction
Tungsten (W) is considered to be an important structural material for high-temperature applications. Tungsten possesses a melting point of 3422 °C [
1,
2], exceeding that of any other pure metal, allowing it to withstand extreme thermal environments. Its high tensile strength of 1725 MPa allows it to resist deformation and breakage under immense loads [
3]. Furthermore, tungsten exhibits outstanding corrosion resistance, making it a valuable material for applications involving harsh chemical environments [
4]. Coupled with a low thermal expansion coefficient (4.32 × 10
−6 K
−1 at 300 K), it can maintain its dimensions with minimal change under significant temperature changes. This makes it ideal for applications requiring dimensional stability in extreme environments, such as jet engine components and nuclear reactors [
5,
6,
7]. Moreover, ultrafine tungsten powder is the essential raw material used for the production of ultrafine cemented carbides, widely used in the printed circuit board industry as a component of micro-drilling tools [
8,
9].
Obtaining high-quality tungsten powder is a critical step in the fabrication of advanced tungsten-based materials [
10,
11]. The main challenge is focused on densifying the material, often requiring a sintering temperature exceeding 2000 °C. Researchers have explored various methods to synthesize nanoscale tungsten powder, such as high-energy ball milling [
12,
13], molten salt synthesis [
14], and spray conversion processing [
15]. Nanocrystalline tungsten powder has been fully densified at temperatures as low as 1400 °C via spark plasma sintering, resulting in an ultrafine-grained microstructure and excellent mechanical performance [
16]. However, the most widely-used method is the traditional reduction method [
17,
18], where tungsten powder is obtained from the reduction of tungsten oxide or ammonium paratungstate (APT) under the flow of hydrogen (H
2). Unfortunately, water vapor produced by the reduction leads to an increase in particle size due to the formation of volatile WO
2(OH)
2, which transports tungsten through the gas phase and promotes particle growth [
19]. Therefore, reducing the particle size of the starting powder is a key strategy to lower the densification temperature and achieve a controllable microstructure. This necessitates the development of novel synthesis methods that circumvent the limitations of the traditional hydrogen reduction approach.
As previously reported, a nitridation–carburization method was used to synthesize ultrafine tungsten carbide (WC) powder, in which nitridation of WO
3 using NH
3 and carburization using CH
4-H
2 were conducted sequentially [
20]. This two-step process involved the sequential nitridation of WO
3 using NH
3 followed by carburization using CH
4-H
2. Interestingly, the nitridation step resulted in the formation of W
2N, which subsequently transformed into WC powder through carburization at 800 °C. A WC powder was formed with an impressively fine particle size of approximately 20–30 nm. The hypothesis suggested that the initial formation of metallic tungsten occurred via the rapid decomposition of W
2N nanoparticles, a process significantly faster than WO
3 reduction. Rather than the growth and coarsening attributed to the existence of WO
2(OH)
2, as mentioned previously, the particle size of WC was be maintained at the nanoscale due to the absence of H
2O in the reaction system. Inspired by this work, an in situ synthesis process was proposed to produce metallic tungsten powder, using a one-step nitridation–decomposition method. This method involved the initial preparation of W
2N intermediate through the nitridation of WO
3 in NH
3, followed by in situ decomposition under an inert atmosphere. The phase and morphological evolution throughout the process were studied. As an alternative method, carbothermal reduction offers the advantages of faster reaction times due to the absence of solid–gas reactions and generally low costs. In the present work, we also attempted to produce tungsten powder via carbothermal reduction of WO
3, and the morphology of the final product and reaction mechanism are compared.
2. Experimental Procedures
A sol–gel route was employed to synthesize WO
3 powder, starting from ammonium metatungstate (Zhuzhou Cemented Carbide Group Corp. Ltd., Zhuzhou, China) and anhydrous citric acid (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). After gelation, drying, and calcination, nanosized WO
3 precursor with an average particle size of approximately 40 nm was obtained for the following process. More details regarding the procedures can be found in a previous study [
20].
The prepared WO3 powder was placed on a graphite boat and heat-treated in a horizontal tube furnace under a constant flow of NH3 (600 mL/min) at 800 °C for 3 h to facilitate complete nitridation. Following nitridation, the NH3 flow was immediately switched to Ar (1000 mL/min) and the temperature was elevated to 900 °C and 1000 °C for in situ decomposition of W2N to metallic W. The intermediate W2N after ammonolysis of WO3 at 800 °C was also collected and characterized.
For carbothermal reduction, the molar ratio of WO3 to carbon black (purity: 99.9%, Beijing Enoch technology Ltd., Beijing, China) was controlled at 1:2.9. The starting mixtures were mixed for 24 h in a polyethene jar using Si3N4 balls and ethyl alcohol as a carrier fluid. The slurry was dried in a rotary evaporator and then moved to a graphite element resistance furnace; the temperature was increased to 1300 °C at 10 °C/min followed by a dwell time of 3 h under vacuum.
The phase assemblage of the as-received powders was investigated using X-ray diffraction with Cu Kα radiation at 40 kV and 40 mA (XRD; Rigaku D/MAX-2 550V, Tokyo, Japan). The mass fraction of W in the mixture of W
2N/W was determined based on the relative intensity of the strongest diffraction peak using the K value method [
21]. The reference intensity ratio of W
2N (18.34) and W (18.00) was adopted from JCPDS Card nos. 25-1257 and 89-2767, respectively. Transmission electron microscopy (TEM; JEOL 2100F, JEOL Co. Ltd., Tokyo, Japan) and scanning electron microscopy (SEM; LYRA 3 XMU, Tescan, USA) were used to observe the powder morphology, where the powdered samples from both method were dispersed onto a Cu grid. The specific surface area was examined using the BET method (ASAP 2010, Micromeritics, Norcross, USA).
3. Results and Discussion
XRD analysis (
Figure 1a) confirmed the formation of pure tungsten product through carbothermal reduction of WO
3 at 1300 °C under vacuum, as evidenced by the presence of only tungsten reflections matching JCPDS Card no. 89-2767. This indicated the absence of residual oxides or unreacted starting materials. However, compared to the initial WO
3 powder with an average particle size of ~40 nm [
20], TEM images (
Figure 1b) revealed the coarsening of W particles after the reaction. The estimated particle size based on TEM analysis was around 0.8 μm, which was further supported by the measured specific surface area of 0.7 m
2/g calculated using the following equation:
where S denotes the surface area, D represents the particle diameter, and
is the density. This corresponded to a theoretical particle diameter of 4.4 μm. Notably, the average particle size observed in SEM images (seen in
Figure 1c,d) was determined to be within the range of 1–2 μm. Importantly, both XRD and SEM results confirmed the absence of any graphite phase, indicating a complete reaction and no residual carbon black contamination. The TEM observation (
Figure 1b) reveals a finer particle structure compared to the SEM images, primarily attributed to the dispersion of the powdered sample onto a Cu grid during TEM sample preparation.
In a previous study, flowing NH
3 was utilized at 600 °C to produce W
2N powder through nitridation of WO
3 [
20]. To ensure complete nitridation of WO
3 and prevent oxygen impurities, the nitridation temperature was increased to 800 °C.
Figure 2 presented XRD patterns of the nitrided products, which illustrated phase evolution during nitridation–decomposition of WO
3 at various stages. The dominant phase of the product after ammonolysis of WO
3 at 800 °C was W
2N, as evidenced in
Figure 2a by characteristic reflections at (111), (200), and (220). However, the presence of W at 2θ = 40.0° as a minor phase in the product suggests simultaneous decomposition of the nitride intermediate due to the relatively high temperature employed. The W mass fraction was calculated to be 0.9 wt. %. In situ decomposition was subsequently performed by switching the flowing gas from NH
3 to Ar and further increasing the heat-treatment temperature to 900 °C and 1000 °C. As shown in
Figure 2b,c, the intensity of W peaks gradually increased with increasing temperature, while those of W
2N decreased. Complete decomposition of W
2N to W was achieved at 1000 °C, as confirmed by the disappearance of W
2N peaks in
Figure 2e. Importantly, no intermediate phases were observed during the decomposition process (
Figure 2a–e).
The morphology of the W
2N intermediate, obtained from the nitridation of WO
3 at 800 °C for 3 h, is presented in
Figure 3a. The particles possessed a near-spherical shape with an average size of approximately 30 nm and exhibited slight agglomeration. Notably, mesopores with a size of approximately 5 nm were observed within the particle structure, as indicated by red arrows in
Figure 3b, in contrast to the starting WO
3. The unique formation of mesoporous W
2N was mainly attributed to the nitridation reaction.
Tungsten powder derived from the in situ decomposition of W
2N at 1000 °C was analyzed, and the resulting morphology and particle size were investigated. As shown in the TEM observation in
Figure 4, the synthesized W powder exhibited a uniform particle size of ~15 nm with narrow distribution. Furthermore, some elongated W particles were observed, possibly originating from the mesoporous W
2N particles. The specific surface area of the W powder synthesized at 1000 °C was measured to be 6.52 m
2/g, indicating a relatively high surface area. In addition,
Figure 4b also exhibits that the as-obtained nanosized tungsten powder are bonded together to form some agglomeration. Notably, the particle size of the W product obtained through nitridation–decomposition was smaller than both the raw WO
3 powder and the W product derived from the carbothermal reduction (seen in
Figure 1b). This highlighted the potential of this method for synthesizing high-quality, nanosized tungsten powder.
Figure 5 illustrates a proposed mechanism for the morphology changes observed during the nitridation of WO
3. The initial stage involved the formation of a rigid surface layer composed of interconnected W
2N crystallites on the WO
3 particle surface. Due to the great density difference between W
2N (18.1 g/cm
3) and WO
3 (7.28 g/cm
3), a significant volume reduction of up to 67% occurred during the conversion of WO
3 to W
2N. This newly formed nitride layer likely restricted shrinking as the nitridation process continued, consequently, leading to the formation of nanosized pores or cracks within the particles. The W
2N particles formed were comprised of nanocrystalline aggregates, where the mesoporous structure formed within or between these aggregates, which can be observed in
Figure 3b. This hypothesis was supported by the variation in the specific surface area. Despite small changes in overall particle size, the W
2N powder exhibits a significantly higher specific surface area (23.63 m
2/g) compared to the starting WO
3 powder (18.10 m
2/g). This increase suggests the development of internal porosity within the W
2N particles, consistent with the proposed mechanism. Due to the crack form within the particle itself, the newly-formed tungsten featured a freshly exposed surface, leading to the formation of soft agglomerates (seen in
Figure 4). These agglomerates are expected to be easily broken apart, due to the absence of the water vapor at high temperatures, which would result in the agglomerates of tungsten product derived using the traditional hydrogen reduction method.
Although the overall mechanism of the carbothermal reduction reaction of tungsten oxide can be expressed as WO
3 + 3C = W + 3CO, the carbothermal reduction of WO
3 to W proceeds through a multi-step reaction mechanism, as previously reported [
22,
23], namely WO
3 → WO
2.
72 → WO
2 → W. In the initial stage, oxygen removal from WO
3 forms intermediate oxides such as WO
2.
72, creating lattice defects that facilitate further reduction. The subsequent reaction between WO
2 and CO produces tungsten and CO
2:
However, it was confirmed that a competing side reaction involving the spontaneous disintegration of WO
2 occurs simultaneously [
24]:
WO
2 was heat-treated under vacuum and WO
3 was identified as the gas phase according to the following reaction [
25,
26]:
During this process, the highly mobile gaseous WO3 phase underwent sublimation and was deposited onto the surface of WO2 or carbon black particles as the carbothermal reduction was occurring. The resulting gaseous (WO3)x species contributed to gas-phase diffusion, potentially promoting tungsten particle coarsening via surface energy reduction. Notably, achieving minimal carbon impurities often required non-stoichiometric ratios like WO3:C = 1:2.9 (used in this study), which presented challenges in precise control due to activity variations in the carbon source.
Unlike carbothermal reduction, the decomposition of W
2N offered a direct route to metallic W with a significantly smaller particle size. This faster process avoided the growth of W particles through chemical vapor transport of tungsten-containing species, as confirmed by the absence of intermediate phases during decomposition. Notably, the direct conversion from nanosized W
2N particles led to inherently smaller W particles compared to the coarsening observed in carbothermal reduction.
Table 1 lists the average particle size of W powder obtained through both the nitridation–decomposition approach and carbothermal reduction. The comparison clearly exhibits the particle refinement and advantage for the proposed approach in the present work.
Synthesizing high-quality, fine-grained tungsten posed a significant challenge due to the limitations of existing methods. This work directly addressed this need by presenting a novel route for W
2N decomposition, which overcame the drawbacks of particle coarsening originating from intermediate species [
26]. Our findings not only offer potential improvements in the densification of tungsten but also open doors for applications demanding a high surface area and precise control over tungsten nanostructures, for instance, in the fields of catalysis [
27], nanocomposites [
28], and microelectronics [
29]. Importantly, further exploration of this method and optimizing reaction parameters and precursor properties hold immense promise for developing an even more efficient and scalable synthesis of fine-grained tungsten. This paves the way for broader adoption of this valuable material in advanced technological applications across various fields.