2.4.1. IF-WS2 Synthesis Using a New Baffled Quartz Tube

To improve the batch yield, a new baffled quartz tube was designed and adopted for the experiment (Figure S1). Based on previous optimal parameters, changes were made to reflect the significantly increased precursor from 6–18 g. For example, the gas rate of H2 and H2S, the reaction time and inclination angle *etc.* were finely adjusted in experiments B1–B3.

For the new baffled tube, a bigger inclination angle was required at the beginning of the experiment in order to drive the particles towards the hot zone, otherwise they would only move around rather than move forward. As soon as all the particles marched into the hot zone, the angle could be decreased to zero for better reaction. In a typical experiment, 18 g WO3 nanoparticles were used with an initial inclination angle of 5 degrees, and H2: H2S: Ar ratio of 10:30:160 for the whole process. After the experiment, 15 g out of the 18 g were collected from the hot zone. By reducing the H2 gas content in the flow in the later stage of the reaction (experiment B3), the amount of WS2 nanotubes has been dramatically decreased, and they were much shorter (compared to experiments B1 and B2), Figure 13a. The dominant IF-WS2 nanoparticles appeared to be uniform, fine and spherical, with diameters <100 nm (Figure 13b). Their XRD pattern is very promising, with strong WS2 peaks and very tiny WO*x* signals, which is again indicative of a good sulphidisation.

**Figure 13.** SEM images for samples collected from the hot zone of experiment B3, revealing the existence of a small amount of nanotubes (**a**), and the fine and uniform IF-WS2 (**b**).

TEM characterisation for sample B3 (Figure 14) further confirm the dominant nature of the IF-WS2, with sizes ranging from below 50 nm to up to 100 nm. Figure 14b also demonstrates that some particles coalesced from two or three nanoparticles, exhibiting a peanut shape (arrowed). Those single particles are always <50 nm. Figure 14b–d show high resolution images of well-crystallised IF-WS2 particles, which again reveal their typical hollow core and seamless shell layer features. Some particles still possess a residue WO*x* core (particle A in Figure 14d). Nevertheless, the new baffled working quartz tube was most promising, enabled a yield improvement from 5 g–15 g per batch whilst successfully maintaining the products quality. However, as the mixing efficiency drops with the filling degree [41], a too high amount of WO3 precursor input would lead to less effective mixing and thus compromise the quality of the final product. Thus, the batch yield and product quality are limited by the quantity of the precursors loaded at the beginning of each batch. Further improvement is still necessary.

**Figure 14.** TEM images for samples collected from the hot zone of experiment B3, demonstrating the overall distribution of particle (**a**) and the multi-layered characteristics of different particles (**b**–**d**).

### 2.4.2. A Continuous Feeding System

In this feeding system (Figure 1), the WO3 precursor was first stored in a pump and then introduced to the system by gravity and gas-blow. A single long feeding tube was used to act as the extended pathway for WO*x* particles directly blown into the system (as shown in Figures 1 and S2), while the reaction gases were fed in by a separated tube. The new design was tested (FB1), and the result showed a complete conversion of WO*x* into IF-WS2 for particles from the hot zone (Figure S10). The products were quite uniform, with sizes no more than 100 nm (Figure S11).

The process was further modified to simulate a real, continuous one, by immediately replacing the empty pump with a full one (FB2). All other experimental parameters made no alteration to FB1, except for the longer feeding time. In this case, around 50 g samples were collected from the hot zone. TEM images of the hot zone sample from FB2 were shown in Figure 15. Majority nanoparticles of around 50 nm in size are displayed in Figure 15a,c, some from 20 nm to almost 100 nm are exhibited in Figure 15b. A typical IF-WS2 particle with a hollow core and around 15 seamless layers is shown in Figure 15d.

To summarise, the continuous feeding system has been proven a success and has improved the yield of IF-WS2 to several tens of grams per batch, without having an obvious compromise on quality. Technically, the feeding of precursor could be simply continued by reloading; however, a longer reaction time would be required. In an industry environment, where a proper sample collection system is easily available, an automated reaction would not be a limitation. This reactor is easily adaptable as a whole continuous rotary process for the scaling up production of IF-WS2 nanoparticles, using a proper metallic working tube.

**Figure 15.** TEM images for particles collected from the hot zone from experiment FB2, showing overall uniformity of particles (**a** and **c**) and detailed closed-cage feature under high resolution (**b** and **d**).
