Evolution of reduction process from tungsten oxide to ultrafine tungsten powder via hydrogen

3.1 Reduction progress

The reduction extent was calculated by oxygen index of the powder; therefore, the reduction content of 0 and 100% represents WO₃ and W, respectively. XRD was used to confirm the oxygen index of powder comprising W and WO₂. As shown in Figure 1, samples in condition 1 were not completely reduced because of relatively low temperature. In condition 2, samples were completely reduced in 150 min. Three stages are shown in Figure 1 for both conditions. Stage 1: in the preheating zone of the furnace, the reaction was not significant as the temperature was still low. Stage 2: a rapid reaction stage continued for about 20 min. Change in color can be easily observed from blue to violet and brown, suggesting that the reduction sequence was WO_2.9 → WO_2.72 → WO₂. The reaction velocity was relatively fast [11,14], and by the end of this stage, the reduction extent was 33%. XRD results in Figure 1(b) show that the content of WO₂ reached upto 98%. Stage 3: in the third stage, WO₂ was reduced to W directly. The reduction curves were parabolic, which are typical of a chemical reaction at the moving interface [15]. The increase of reduction velocity from 100 to 130 min in condition 2 (red line) was because of the increase in temperature. The whole sequence was WO_2.9 → WO_2.72 → WO₂ → W. In both reduction conditions, no β–W was found in XRD results.

Figure 1

(a) Reduction extent in two reduction conditions; (b) XRD curve at the end of stage 2.

Figure 2 shows the evolution of BET values in two reduction conditions. In the process from WO_2.9 to WO₂, the BET values underwent increase and subsequent decrease to about 3.5 m²/g. The increase of BET values results from the formation of WO₂. However, the size of WO₂ is not the key factor in final size of tungsten powder. In the stage 3, temperature affected the evolution of BET values. In condition 1, BET values increased from 3.5 to 5.5 m²/g in the first 70 min of stage 3 and then raised to 9 m²/g rapidly in 30 min. In condition 2, with the rise in temperature, the BET values decreased to ∼2 m²/g.

Figure 2

BET measurements in two reduction conditions.

3.2 Morphology evolution

The reduction sequence has been investigated to be WO_2.9 → WO_2.72 → WO₂ → W. The morphology shows the corresponding evolution in each stage. In the preheating zone of furnace, microstructure of powder seemed unchanged, as shown in Figure 3. The color of powder became dark blue because of the tiny differences in chemical composition and roughness of the surface. The time shown in all figures is from the boat pushed into the furnace and not the effective reduction time.

Figure 3

Morphology of samples. (a) Initial state; (b) in preheating zone.

As shown in Figure 4, rod-like forest spread to the whole surface when the boat was pushed into the heating zone. The color of the powder became violet, indicating that WO_2.72 was formed. Subsequently, WO_2.72 was rapidly transformed into WO₂. Figure 5 shows the transformation process from WO_2.72 to WO₂. Small rods aggregated together and disappeared gradually, instead of the granules (∼2 μm) split by star-shaped cracking structure [12] at the end of stage 2. In industrial reduction conditions, WO₂ with star-shaped cracking structure is inevitable. As W was reduced from these granules, final morphology of tungsten powder was decided by the process from WO₂ to W. In dry conditions, the granular morphology would be retained after WO₂ reduced into tungsten.

Figure 4

Morphology of rod-like forest at (a) low magnification and (b) high magnification.

Figure 5

Transformation process from WO_2.72 to WO₂. (a) Aggregation at the top of WO_2.72 forest; (b) WO₂ with star-shaped cracks.

In the long reduction process from WO₂ to W, the granules that were split by the star-shaped cracks dispersed gradually. Figure 6a shows the evolution process of the granules. Firstly, trigonal granules changed into a circular floret structure. The floret structure probably evolved from the star-shaped cracks morphology because residual star shape cracks can be observed in Figure 6(a). Then, the floret granules changed into irregularly fibrous, leading to the emergence of a covered microstructure.

Figure 6

Morphology evolution of WO₂ on the surface. (a) Floret structure; (b) irregularly fibrous.

Finally, as shown in Figure 7(b), spherical tungsten particles were formed in condition 2. Because of low temperature and dry atmosphere in producing ultrafine powder, a part of the granules was reduced in situ, resulting in a considerable number of agglomerates in condition 1 (Figure 7(a)).

Figure 7

Morphology of tungsten powder. (a) With agglomerates; (b) spherical particles.

4.1 Mechanism of morphology evolution

The formation of WO₂(OH)₂ is one of the major reasons for the coarsening of tungsten powder during reduction [16,17]. The morphology evolution in Figures 4–7 indicates that this kind of chemical vapor transport works in all the reduction procedures. The vapor intermediary is probably WO _x (OH) _y because of the variation of valency of W in different reduction steps. Figure 8 presents the forepart of WO_2.72 and WO₂ formation. Rod-like WO_2.72 forest grew up from the rough surface by the deposition of reduced hydrated WO_2.72 vapor on the head of each rod, leaving small holes on the original surface of WO_2.9. Similarly, prior WO₂ formed on the head of WO_2.72 rod and then captured hydrated WO₂ vapor rising from the bottom of WO_2.72 forest, resulting in aggregated WO₂ granules.

Figure 8

Morphology change by chemical vapor transport. (a) WO_2.9 → WO_2.72; (b) WO_2.72 → WO₂.

4.2 Heterogeneity in reduction process

To produce homogeneous tungsten powder, it might be expected that raw oxide has a single component. However, the reduction process on each reduction stage was heterogeneous. As shown in Figure 9, rod-like WO_2.72 could be observed on the surface but the fracture surface was still flat. Only a small amount of WO_2.72 was detected by XRD. The fracture surface in Figure 9(a) was formed after the samples were removed from the furnace. It suggests that the reduction path inside the particle was different from that on the surface; as intermediate WO_2.72 would not have formed inside. As predicted by Donoso et al. [14], only 20% of WO_2.72 form in the whole reduction process at 750–900°C.

Figure 9

Fracture surface of the samples reduced for 35 min in condition 1. (a) Low magnification; (b) high magnification.

As presented in Figure 10, in the process of aggregation of rods, some regions reacted preferentially, leaving rod-like residues. Star-shaped cracks observed in Figure 10 are supposed to result from the evaporation of residual WO_2.72 or the internal stress originating from the shrinkage of aggregated WO₂. When the star-shaped cracking structure was spread to the whole surface, a different internal structure with nanocrystalline WO₂ particles could be observed (see Figure 11). In addition to the observed inhomogeneity in one oxide particle, the difference in reduction content between oxide particles was great, leading to a wide distribution of particle size.

Figure 10

Morphology of residual rod-like WO_2.72. (a) Low magnification; (b) high magnification.

Figure 11

Morphology of nanocrystalline particles on the back of WO₂ granules. (a) Low magnification; (b) high magnification.

4.3 Agglomeration in ultrafine tungsten powder

In industrial conditions, undesirable agglomerates are difficult to avoid in order to produce submicron powder. As presented in Figure 12, these agglomerates cause the powders to maintain the original morphology of raw material. The coarse agglomerates may lead to formation of inhomogeneous mixture with carbon, even to coarse WC particles, which induces abnormal grain growth in production of cemented carbide.

Figure 12

Ultrafine tungsten powder with coarse agglomerates.

As discussed above, the agglomeration is due to lack of chemical vapor transport in the step 3. To avoid agglomeration, the key factor is the promotion of aggregate transformation into spherical particles. The paradox is that the factors promoting chemical vapor transport also lead to coarsening of the powder. One possible method to address the issue is the design of a suitable boat structure and hydrogen flux to suppress the coarsen mechanism of repeated oxidization and reduction of powder by rapidly discharging the water vapor, enhancing the contribution of chemical vapor transport mechanism.