*2.4. Recycling Experiments*

Catalyst recycling experiments were performed with 5Fe-ACs catalyst at 170 ◦C with a 3 h reaction time. After the reaction, the catalyst was filtered, washed with methanol and water, dried and then used again in the same conditions for three cycles. As a result, the furfural yield decreased from 48% to 30% after the first run and then remained constant (Figure 6). Conversion and reaction selectivity behaved similarly; conversion decreased from 76% to 53% and reaction selectivity from 66% to 59% after the first run and then they remained constant (Figure 6). The decrease in furfural yield, conversion and selectivity may be explained by iron leaching. Leaching was observed by measuring the iron content of the water phase after every reaction. After the first run, 66% of the initial iron amount was leached out but after second and third runs leaching was only 3% and 1%, respectively. It has been reported that hot acidic water promotes the solubility of some metal oxides and in particular the solubility of iron oxides is strongly influenced by the solution pH [48,49]. In our experiments, the pH changed from 6 to 2 during the reaction with 5Fe-ACs, so the reaction media was clearly acidic. Chang et al. were able to decrease iron leaching by post-treating the catalyst after impregnation, first with NaOH and then with HCl [43]. Post treatment was designated to convert iron to ferric hydroxide, which is poorly soluble in water. This treatment also decreased iron leaching at a low pH (pH = 2) to 21%, while the loss of iron during the post-treatment process was very minor (from 5.2 to 4.5 wt%). In future work, this must be taken into consideration.

**Figure 6.** Furfural yield, xylose conversion and reaction selectivity in catalyst recycling experiments after 1–3 cycles.

The stability of the catalyst was also observed by monitoring changes at the catalyst surface with SEM and STEM. The images revealed that the iron content was decreased and leaching was more connected to larger iron particles (approx. 15–40 nm) than to small ones (approx. 5 nm) (see Figure S6). Small iron particles were present throughout the recycled catalysts, while large ones occurred only randomly. However, some large particles were also left on the catalyst surface after the third cycle, indicating that they were not all leached out. Leaching of large particles (or agglomerates of small particles) is reasonable since they are more vulnerable to leaching because they have poorer interaction with the support surface than smaller particles. In future work, it would be important to optimize the catalyst preparation to obtain less-agglomerated metal particles.

Based on catalyst weighing, the mass of the catalyst was increased from 5 to 6.7 mg in each cycle. This indicates that the catalyst was adsorbing some reaction products, such as humins. Humins

are black carbonaceous side-products generated either by the cross-polymerization of furfural by itself, between the just-formed furfural and free xylose present in solution or between furfural and intermediate products [50–52]. Solid humins can plug the pores of the catalyst surface, block the access of xylose to metal surface sites or totally encapsulate a metal particle [53]. It is also possible that this blocking of the catalytic sites on the porous structure of the catalyst caused the decrease in furfural yields. Indeed, pore blocking caused by carbon deposition is a common problem in heterogeneous porous catalysts; for example, with MCM-41-SO3H, the surface area and pore volume decreased by 50%–60% after the first run and the furfural yield decreased from 54% to 37% [54]. BET analysis of the used catalysts could not be performed because a very small amount of the catalysts was used in the reactions. In the SEM and STEM images, there was not any clear coating visible on the surface of the carbon, as in Reference [55] but some slight changes were observed in the SEM images (Figure S7). However, based on the figures, no differences were found between catalysts that were used one, two or three times.

#### **3. Materials and Methods**
