A highly stable MOF-engineered FeS2/C nanocatalyst for heterogeneous electro-Fenton treatment: Validation in wastewater at mild pH.

Herein, the novel application of FeS2/C nanocomposite as a highly active, stable and recyclable catalyst for heterogeneous electro-Fenton (EF) treatment of organic water pollutants is discussed. The simultaneous carbonization and sulfidation of an iron-based metal organic framework (MOF) yielded well-dispersed pyrite FeS2 nanoparticles of ~100 nm diameter linked to porous carbon. XPS analysis revealed the presence of doping N atoms. EF treatment with an IrO2/air-diffusion cell ensured the complete removal of the antidepressant fluoxetine spiked into urban wastewater at near-neutral pH after 60 min at 50 mA with 0.4 g L-1 catalyst as optimum dose. The clear enhancement of catalytic activity and stability of the material as compared to natural pyrite was evidenced, as deduced from its characterization before and after use. The final solutions contained < 1.5 mg L-1 of dissolved iron and became progressively acidified. Fluorescence excitation-emission spectroscopy with PARAFAC analysis demonstrated the large mineralization of all wastewater components at 6 h, which was accompanied by a substantial decrease of toxicity. A mechanism with •OH as dominant oxidant was proposed: FeS2 core-shell nanoparticles served as Fe2+ shuttles for homogeneous Fenton's reaction and provided active sites for heterogeneous Fenton process, whereas nanoporous carbon allowed minimizing the mass transport limitations.

Undoubtedly, Fenton process is currently one of the most attractive technologies to tackle 40 the global water contamination by toxic, recalcitrant, non-biodegradable organic 41 pollutants, owing to its great effectiveness combined with simplicity and low cost. 1

42
Aiming to overcome some inherent shortcomings, 2 gradual optimization led to the 43 development of electro-Fenton (EF) process, which has become the most successful 44 method among the so-called electrochemical advanced oxidation processes (EAOPs). 3,4 45 The scientific fundamentals of EF are now quite well understood, but the lack of 46 robustness and reliability of some of the materials involved still hampers its final 47 implementation at industrial scale. 2 On the one hand, much progress has been made on 48 cathode development to enhance the H2O2 electrogeneration from the 2-electron O2 49 reduction reaction (1). 5 The greatest H2O2 accumulation can be achieved using air-50 diffusion cathodes equipped with a gas chamber, 6-10 although high efficiencies for H2O2 51 production are also feasible with modified three-dimensional carbonaceous cathodes. 11-15 52 Substantial advances have also been made in the selection of electrocatalytic anodes (M) 53 that promote the simultaneous generation of adsorbed M( • OH) from water oxidation. [16][17][18] 54 O2(g) + 2H + + 2e − → H2O2 (1) 55 Conversely, crucial concerns arise when the third cornerstone, i.e., the catalyst, is 56 considered. Conventional EF treatment based on homogeneous catalytic decomposition 57 of H2O2 in the presence of soluble Fe 2+ , according to Fenton's reaction (2) at optimum 58 pH ~ 3.0, is still the sole well-established application. 2

59
Fe 2+ + H2O2 → Fe 3+ + • OH + OH − (2) 60 Lately, some approaches have been proposed to work at less acidic pH, thus trying 61 to broaden the potential market of EF, which could embrace the treatment of urban 62 5 wastewater at near-neutral pH. For example, homogeneous EF is viable at high pH upon 63 use of iron complexed with organic chelators. 19 Nonetheless, heterogeneous catalysts 64 seem a smarter choice, since they can facilitate the post-treatment clean-up and minimize 65 the dissolved iron content that eventually causes sludge production. 20 These catalysts 66 include several types of synthetic iron-loaded structures, such as resins or zeolites, 21,22 as 67 well as zero-valent ion, 2 iron-rich clays 23 , layered double hydroxides (LDHs) 24 and 68 minerals like iron oxides [25][26][27] or pyrite (FeS2). 26 In particular, mineral pyrite has been 69 confirmed as a very good candidate for Fenton 28,29 or EF 30-32 treatments, since it is an 70 excellent electron donor whose S2 2conversion to sulfate via reaction (3) and (4)  carried out in urban wastewater (Text S1). All CAS numbers and purities are given in 116 Table S1.

117
Synthesis of the catalyst. The synthetic route followed to obtain the FeS2/C catalyst, 118 adapted from Pham et al., 55 is schematized in Figure S1 (Supplementary Information). The performance of the synthesized catalyst was compared with that of commercial 128 pyrite, which was milled and washed with ethanol and nitric acid to obtain the dark shiny 129 powder shown in Figure S1. in the SEM images, whose aggregation gave rise to larger secondary nanostructures.

221
Based on this finding, the as-synthesized powder will be properly dispersed by means of 222 ultrasounds prior to its use as catalyst in EF treatments described in next subsections.

223
Worth noting, the size of MOF-derived FeS2 particles synthesized by sulfidation using 224 other protocols was also around 100 nm. 58 It is expected that nanometric size will have a 225 very positive contribution to the catalytic activity, which will also benefit from a 226 relatively large BET surface area of 25.96 m 2 g -1 , a value much higher than that reported 227 upon hydrothermal synthesis of FeS2 (i.e., 1.17 m 2 g -1 ). 33 As expected, the raw iron-MOF 228 precursor had a greater BET surface area of 516.2 m 2 g -1 , thanks to its inherent 3D porous 229 structure.

230
The XRD pattern depicted in Figure 1b Figure 1e), but the carbon shell was not so easy to identify. Worth noticing, the signal for 257 oxygen was strong enough, as a result of residual Fe2O3, whereas that from nitrogen was 258 very weak. In order to confirm the formation of the core-shell structure, TEM-EELS 259 analysis was carried out. Figure S3d shows the STEM image of an aggregate, along with 260 13 the EELS spectra recorded from two different regions: region i, whose composition agrees 261 with that of a core since it reveals the presence of Fe (major edge at 708 eV (L3) and a 262 smaller one at 721 eV (L2)) and S (major edge at 165 eV (L2,3)), apart from carbon with 263 a major K-edge at 284 eV; and region ii, which clearly matches with a carbon shell.

264
Peaks associated to the carbon shell could not be identified in the XRD pattern due 265 to its amorphous structure. Nonetheless, the Raman spectrum of the FeS2/C catalyst 266 depicted in Figure S3e evidences the presence of two main bands related to carbon, 267 namely D and G located at 1309 and 1541 cm -1 . 59 The smaller peaks at 339 cm -1 (Eg), 375 268 cm -1 (Ag) and 464 cm-1 (Tg) can be attributed to pyrite. 59

269
The surface composition was further analyzed by XPS. The general spectrum for the 270 as-synthesized FeS2/C nanocatalyst, depicted in Figure S4, reveals the energy range of 271 the five elements identified above, and the three most important were evaluated in detail.

272
In the high resolution Fe 2p core level XPS spectrum, shown in Figure  and (iii) the presence of carbon enhanced both, the mass transport due to its porosity, and 336 the catalytic activity, as also found for Fe3O4/C catalyst during octane degradation. 62 The 337 reactivity was also favored by doping with N.

338
It is very remarkable that iron release with the new catalyst was much lower than that which kept the pH constant (~ 6.0, Figure S7b). At such high pH, the dissolved iron was 360 almost negligible (< 0.25 mg L -1 , Figure S7c) and hence, the contribution of homogeneous 361 Fenton's reaction to the final 46% drug removal could be presumed as insignificant. Since 362 in EO-H2O2 in phosphate buffer the degradation at 60 min was 20% ( Figure S7a), it could 363 be inferred that the FeS2/C catalyst is able to yield 26% fluoxetine degradation via pure 364 heterogeneous catalysis. Now, going back to Figure 3a, as a first approach one could 365 conclude that the 91% drug removal was caused by a combined mechanism involving 366 heterogeneous Fenton (~26%, as just calculated form Figure S7a) and EO-H2O2 (~41%, 367 Figure 3a), but being also remarkable the role of homogeneous Fenton (~24%). From and homogeneous Fenton's reaction thanks to more dissolved iron (see Figure S9).

391
However, further increase to 0.5 g L -1 FeS2/C did not improve the performance, probably 392 because of the parasitic reaction between • OH and the excess of Fe(III) or Fe(II). Figure   393 4c evidences the positive contribution of current increase, being more significant from 15 394 to 30 mA, which resulted from a gradually greater iron release ( Figure S10a) and H2O2 395 electrogeneration ( Figure S10b). The latter species had a prevailing role due to its higher Mineralization and proposed mechanism. Longer trials were performed to assess 437 the mineralization ability of the heterogeneous EF treatment, using the optimum FeS2/C 438 content shown in Figure 4b (i.e., 0.4 g L -1 ). A BDD anode and a current of 100 mA were 439 employed, looking for a more powerful system thanks to the production of physisorbed 440 • OH. It was a right choice since, as can be seen in Figure S14, an impressive 90% TOC 441 removal was achieved at 6 h. This outperforms even the conventional EF process at pH organic compounds had disappeared. Therefore, the residual TOC in Figure S14 470 corresponded to aliphatic products.

471
Taking into account the main homogeneous and heterogeneous reactions and species 472 mentioned throughout the manuscript, a thorough mechanism is proposed in Figure 6 for 473 the FeS2/C-catalyzed EF treatment of fluoxetine, as model organic pollutant, at mild pH.

474
In conclusion, the pyrite-like nanocomposite made of a nanosized FeS2 core embedded in 475 a carbon shell has been confirmed as an outstanding candidate for heterogeneous EF