2.2.1. WO3 and S Reaction under H2

The WO3 precursor exhibits an average particle size of around 60 nm, having a monoclinic WO3 structure (JCPDS No. 43-1035), as shown in Figures S4 and S5.

For the initial trial (experiment S1, as listed in Table 2), a composite powder of WO3 and S at a molar ratio of a 1:10 was used as the precursor, and reacted under Ar/H2 atmosphere at 800 °C, following the procedure described in Experimental Section 3.3.1. SEM images in Figure 2 show that very small nanoparticles and agglomerates were observed, with some nanoplatelets. However, the EDX spectrum shows that although WS2 has formed, WO*x* still dominates the products. This is also confirmed by XRD study as shown in Figure 3. In Figure 3, the peaks labelled with a triangle matched well with WS2 (JCPDS No. 84-1398) [26], and all the other peaks labelled with a star could be assigned to W18O49 and W20O58. The highest WS2 peak is at 2-theta 14.3°, corresponding to (002). The relatively low intensity of (002) plane for WS2 reveals that only a few WS2 layers have formed from the outside and leaving behind a WO*x* core which has been reduced from WO3 to W18O49 and

W20O58. The presence of W18O49 and W20O58 is in agreement with previous studies [26,31–33], in which partially reduced W20O58 and thoroughly reduced W18O49 were formed during the oxide-to-sulphide conversion from WO3 to IF- WS2. There is no detectable S left in the products.

**Figure 2.** SEM images (**a**–**c**) and EDX spectrum (**d**) for products from experiment S1.

**Figure 3.** XRD pattern for WS2 from experiment S1 (reaction of WO3, S and H2).

Although a molar ratio of 10:1 for S and W had been initially used, it seems that there was not enough S to react with the reduced WO*x* core, because the S could not stay long enough in the hot zone. As soon as the powders reached the high temperature zone, the S would be sublimed and blown

out of the high temperature zone, even under an optimised Ar/H2 flow rate of 80 mL/min [34]. Therefore, the oxide-to-sulphide conversion could not be completed.

There are some similar works that used the same molar ratio of S:WO3 = 10:1 [34], but only achieved a yield of less than 0.1 g per batch. It is obvious that this method is unsuitable for scaling up, as the product quality depends upon the quantity of WO3 nanoparticles used and it is difficult to increase this for traditional static furnaces. To avoid the S loss, another batch process has also been reported recently [35], using solid NaBH4 or LiAlH4 as the H2 releasing agent in a sealed ampoules. Again, these processes are not suitable for continuous production.

In this context, a continuous feeding and evaporation of S was tested, to compensate for the S loss during the reaction, as shown in Figure 1. The WO3 and S mixed powder at a molar ratio of 1:10 was fed directly into the hot zone gradually using the pump, at elevated temperature, rather than preplaced inside the quartz tube (experiment S2). The XRD analyses of the powders collected from both the inlet and hot zones have showed that only a few WS2 layers formed and the majority remained as WO*x*, and S peaks were also present (peaks labelled with circle); although samples from the hot zone exhibited a more complete oxide-to-sulphide conversion (Figure S6).

Morphologically, the blue powders collected from the inlet zone consist of both very small particles and big aggregates, with some bright S residues, as shown in Figure 4a,b. At the relatively low temperature inlet zone, S could not completely react with the reduced WO*x*, thus only a few layer WS2 formed. For the hot zone samples (Figure 4c,d), some big agglomerates are visible, and they are indeed composed of nanoparticles after ultrasonic treatment.

**Figure 4.** SEM images of products collected from the inlet zone (**a** and **b**), and hot-zone (**c** and **d**) of experiment S2.

Therefore, although this technique has the potential for scaling up based on the continuous feeding system, the quality of the products is not high enough. Furthermore, when the temperature dropped below its boiling point, the S vapour could block the outlet gas pipes, which could be a practical issue. The large amounts of S consumption, although can be recycled, makes this process not cost effective.

### 2.2.2. APT as Precursor and H2S as Reaction Gas

Since WO3 nanoparticles were fabricated by the decomposition of APT, as described in the SI (Supplementary Information 2.2.2, Figures S7 and S8), and were a very valuable precursor for IF-WS2 nanomaterials in previous two-step studies [1,25,26,36], *i.e.*, first decomposition of APT to form WO3 then via sulphidisation to create IF-WS2, it is thus interesting to combine the two steps together in our present set-up. This would be an advantage in terms of energy saving and process efficiency.

The as-received APT particles are crystals of several tens of m in size, with nano-sized particles attached to the surface of these big crystals. Under high magnification, cracks and sub m sized particles could be observed at the corners of some damaged crystals, indicating that the APT crystals might be agglomerates of small pieces. Thus, prior to experiment (experiment AHS1), the APT particles were ultrasonic treated, to break the agglomerates into small pieces. The IF-WS2 particles collected from hot zone are shown in Figure 5. On average, the agglomerates are smaller (Figure 5a), although some bigger aggregates which kept the original APT crystal shape were also observed, as shown Figure 5a,b. Higher magnification study shown in Figure 5c and d reveals the nanostructural feature within the agglomerates, and they are composed of both IF-WS2 nanoparticles and nanoplatelets, roughly at the same proportion. XRD study (Figure 6) reveals that the hot zone products have very high intensity of WS2 peaks, with only minor peaks of WO*x*. In contrary, the outlet zone products contain a high portion of WO*x*, owing to the shorter reaction time than those from the hot zone.

To form WO*x* by decomposing APT, the temperature required will be above the melting point of tungsten oxide, e.g., 1200 °C to form WO3 nanoparticles and 1350 °C for micro particles [37,38]. This allows for the tungsten oxide vapour to be brought to and deposited in the low temperature area. In the present process, the formed nanoparticles would then play the template role during subsequent sulphidisation, to form the IF-WS2. Because of the different temperature requirements for the two steps, the direct IF-WS2 synthesis using APT to react with H2S seems to be an unsuitable choice for the present furnace.

**Figure 5.** SEM images of samples from experiment AHS1, exhibiting the original shape and size of the APT (**a** and **b**), and the nanostructural feature within the agglomerates (**c** and **d**).

**Figure 6.** XRD patterns of samples collected from different areas (**a**, outlet and **b**, hot zone) in experiment AHS1.

2.2.3. WO3 and H2S Synthesis of WS2 Nanomaterials

Figure 7 shows the SEM results of experiment W1. A feature of nanoparticle domination is visible, with big agglomerates (Figure 7a). The semi-spherical IF-WS2 nanoparticles exhibit different sizes (Figure 7b): the tiny ones have diameters of <50 nm which are the same as the WO3 precursor; whilst the big ones are about 100–200 nm, possibly merging from two or more nanoparticles. The presence of nanosheets or nanoplatelets amongst nanoparticles can be seen from Figure 7c (arrowed). XRD investigation confirms that the majority of the products are IF-WS2 nanoparticles, with left-shifted (002) peak and broadened (103) and (105) peaks [3], Figure 7d. Tiny peaks at 23–25 degrees (labelled with a star) can be assigned to WO*<sup>x</sup>* core residue. XRD comparison between the produced IF-WS2 and the commercial 2H-WS2 has been presented in the Supplementary Information, as Figure S9, where the differences between XRD pattern of the resulting IF-WS2 and 2H-WS2 are discussed in detail.

**Figure 7.** SEM images (**a**–**c**) and XRD pattern (**d**) of IF-WS2 synthesised using the rotary process.

Further TEM examination shows that the sample contains both nanoparticles and nanosheets/nanoplatelets (Figure 8a). Indeed, the IFs exhibit the multi-layered, hollow core characteristics, being the dominant phase. Some particles followed the original shape of their oxide precursor, appearing in a spherical, seamless, and close-caged structure (arrows A); whilst some displayed a peanut-like structure (arrows B) or a long elliptical shape (arrow C). These unusual particles were possibly co-formed from adjacent WO3 nanoparticles that fused together during the heating. The continuous contour of WS2 layers suggests that these particles must have fused together first, then the oxide-to-sulphide conversion occurred. This observation can also explain the different particle sizes observed under SEM, as shown in Figure 8b,c. The products, regardless of their different shapes, possess a hollow core and a generally equal d(002) spacing of 0.62 nm for IF-WS2.

**Figure 8.** TEM images of the WS2 synthesised using the rotary process (**a**–**d**; b is a zoomed-in image of framed area in a).

These characterisations of the products have confirmed that IF-WS2 is the dominant phase in the product with high quality. This shows the great potential for the production of IF-WS2 at large quantities and high quality, using the present rotary process. Further investigations will focus on quality assessment and quantity improvement.
