Redox-Switchable Single-Atom Catalyst Enables Efficient Aqueous Hydroxymethylfurfural Oxidation

1Redox-Switchable Single-Atom Catalyst Enables Efficient Aqueous 2Hydroxymethylfurfural Oxidation 3Jacky H. Advani, David Panácek, Petr Langer, Daniela Plachá, En Zhao, Shibo Xi, Zupeng Chen, 4Rajenahally V. Jagadeesh, Paolo Fornasiero, Giorgio Zoppellaro,*Aristides Bakandritsos,* 5and Radek Zboril* Cite This: https://doi.org/10.1021/acscatal.5c06280 Read Online ACCESS Metrics & More Article Recommendations * sı Supporting Information 6ABSTRACT: The selective aerobic oxidation of biomass-derived 57hydroxymethylfurfural (HMF) to 2,5-diformylfuran (DFF) is a 8pivotal step toward biobased polymers, pharmaceuticals, and fuels. 9Yet, most high-performance catalysts require noble metals and 10 organic solvents and lose activity in water. Here, we report a robust 11 and recyclable heterogeneous catalyst comprising mixed-valence 12 single-atom iron dimers anchored on nitrogen-doped graphene acid 13 (Fe−NGA), which mimics the powerful oxidation center in 14 nonheme diiron oxidases. Spectroscopic and theoretical studies 15 reveal a redox-flexible Fe2+/Fe3+ manifold that, under basic aqueous 16 conditions, evolves into a Fe3+−Fe4+ ferryl species capable of highly 17 selective proton-coupled two-electron oxidations. Fe−NGA achieves 18 97% HMF conversion with 95% DFF selectivity, a turnover 19 frequency of 17.3 h−1, and a specific productivity of 12.5 mmolDFF gcat −1h−1in pure water, surpassing state-of-the-art homogeneous 20 and heterogeneous catalysts. The catalyst is stable with very low performance loss for at least six reactions. By merging such 21 functionalities within a stable and reusable heterogeneous framework, Fe-NGA provides a benchmark earth-abundant catalyst for the 22 effective oxidation of renewable feedstocks. 23 KEYWORDS: iron single-atom catalyst, dimers, green oxidation, biomass valorization, 2,5-diformylfuran 1. INTRODUCTION 24 The development of highly active and selective oxidation 25 catalysts is essential for many important organic trans26 formations, including green chemicals at the energy−environ27 ment interface. 1−5 Among these, the oxidation of biomass28 derived 5-hydroxymethylfurfural (HMF; one of the top 12 29 sustainable platform chemicals 6 ) to 2,5-diformylfuran (DFF) is 30 a flagship reaction in sustainable catalysis. 1,7 DFF is a versatile 31 C6building block used in polymer manufacture, pharmaceut32 icals, antifungals, and fine chemicals, and is a key intermediate 33 toward biobased aromatic replacements. 8−10 Achieving this 34 transformation effectively in water, without sacrificial oxidants 35 or noble metals, is a central challenge for the integration of 36 biomass upgrading into environmentally benign, scalable 37 processes. 11 38 Despite its apparent simplicity, the selective two-electron 39 oxidation of HMF’s primary alcohol to DFF is challenging. 40 The catalyst must oxidize the hydroxymethyl group without 41 overoxidizing the aldehyde to the carboxylic acid or promoting 42 side reactions at the α,β-unsaturated aldehyde moiety. 12−14 In 43 fact, the aldehyde group can get oxidized even in base alone, 44 without a catalyst. 6,15 Current state-of-the-art HMF-to-DFF 45 catalysts still face major limitations related to sustainability 46 (i.e., use of noble metals 13,16−18 ), activity, selectivity, and 47 recyclability. Homogeneous catalysts offer high activity, but are 48 hard to recover and reuse, whereas heterogeneous analogues 49 rarely reach comparable rates. 19,20 For example, the benchmark 50 oxovanadium complex reaches a mean turnover frequency 51 (TOF) of 9.6 h−1under full conversion conditions. 21 Yet, 52 heterogenizing such vanadium catalysts on carbon, reduces the 53 activity to 5.7 h−1, 22 retaining a specific productivity (SP) of ca. 54 6 mmolDFF gcat −1h−1. 16 Efforts have also focused on 55 photocatalysts, but, so far, have led to low activities (e.g., 56 TOF < 1.5 h−1or SP < 1.7 mmolDFF gcat −1h−1for a single atom 57 catalyst of Cu on carbon nitride, 23 and a ZnIn2S4plasmonic 58photocatalyst 12 ). 59 Another bottleneck is the narrow solvent window, because 60most reported catalysts require predominantly aprotic organic Received: September 8, 2025 Revised: November 27, 2025 Accepted: December 2, 2025 Research Articlepubs.acs.org/acscatalysis © XXXX American Chemical Society A https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX *Unknown * | ACSJCA |JCA11.2.5208/W Library-x64 |manuscript.3f (R5.2.i3:5013 |2.1) 2022/08/03 13:05:00 |PROD-WS-397 |rq_4235882 |12/04/2025 05:32:58 |12 |JCA-DEFAULT 61 media (Table S6). Bulk water generally promotes over62 oxidation to carboxylic acids and suppresses activity by 63 displacing or blocking surface oxygen species with OH−/ 64 H2O. 1,18,24,25 Interestingly, trace water has been reported to 65 create a narrow optimum by tuning surface-bound H2O2and 66 reactive oxygen species (ROS), where for a CdZnS photo67 catalyst, adding ∼0.16 mL H2O to 10 mL acetonitrile raised 68 DFF yield to 66%. 26 However, yields declined sharply at higher 69 water content, highlighting the persistent difficulty of 70 maintaining both activity and selectivity in aqueous-rich 71 media. Importantly, this challenge is not limited to photo72 catalysts that rely on free, nonadsorbed ROS, which are readily 73 quenched in water. Even catalysts that activate O2through 74 lattice oxygen migration (Mars−van Krevelen pathways) or 75 surface-bound oxygen intermediates are inhibited in high water 76 concentrations because hydroxyl adsorption blocks the oxygen 77 activation on the surface. 27 Moreover, the mononuclear sites in 78 SACs restrict access to cooperative and controlled oxygen 79 activation pathways. In contrast, nature’s nonheme diiron 80 oxidases provide an instructive blueprint: they exploit water 81 and hydroxides as bridges in binuclear Fe centers, forming 82 high-valent Fe4+�O (ferryl) intermediates that are both highly 83 oxidizing and intrinsically selective in water. 28 In these systems, 84 water and hydroxyls are not poisons but essential participants. 85 Translating such redox-cooperative motifs into robust, 86 recyclable heterogeneous catalysts for operation in water 87 represents an exciting but largely unexplored frontier in 88 catalyst design. 89Here we report an oxidation catalyst featuring preorganized 90single-atom iron dimers anchored on nitrogen-doped graphene 91acid (Fe−NGA), which stabilizes a redox-flexible Fe2+/Fe3+ 92manifold. Under the basic aqueous conditions, this catalyst 93forms an oxo-bridged Fe3+−Fe4+ ferryl species analogous to the 94reactive cores found in nonheme diiron enzymes. 28 This rare 95active site structure enables selective, proton-coupled two96electron oxidation of HMF to DFF in pure water, achieving 97complete HMF conversion within 3 h, with a mean TOF of 9817.3 h−1and an SP of 12.5 mmolDFF gcat −1h−1, while 99maintaining DFF selectivity up to 95%. These performance 100metrics surpass even state-of-the-art homogeneous systems. 101Additionally, the catalyst demonstrates stability and recycla102bility for at least six reactions, with very small loss in activity. 103Fe−NGA thus achieves previously inaccessible high oxidizing 104power and aqueous-phase activity, offering a promising 105blueprint for sustainable oxidation processes and chemicals 106production in water. 2. RESULTS AND DISCUSSION 1072.1. Synthesis and Characterization of the Fe-NGA 108Catalyst. The NGA support was synthesized by first 109exfoliating commercially available fluorographite via sonication, 110followed by reaction with sodium azide. 29 The nucleophilic 111azide ions attack the electrophilic centers associated with 112fluorine vacancies on fluorographene (FG). 30 Subsequently, 113the azides disproportionate, leading to nitrogen atoms 114incorporated into the graphene plane, affording a 16 at.% N115 f1doped graphene (NG, Figure 1a), as previously demonstraFigure 1. (a) Synthesis of Fe-NGA: iron atoms are coordinated with nitrogen and carboxyl/carboxylate groups. (b) HR-TEM and SEM (in the inset) micrographs of Fe-NGA, (c) HAADF image, and (d−g) EDS mapping of Fe-NGA. (h) HR-XPS spectra of the NGA support and Fe-NGA for the O 1s region. (i) XANES spectra and (j) R-space FT-EXAFS of Fe-NGA and Fe foil. ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX B 116 ted. 31 According to our earlier study, the resulting NG is a 117 highly nitrogen-doped graphene containing numerous and 118 large vacancies. HR-TEM and Raman spectroscopy revealed an 119 intense D-band, indicative of sp3-type defect carbon sites that 120 persist even after high-temperature treatment, confirming that 121 these defects originate from the edges of vacancies and the 122 periphery of the NG sheets. 31 These structural features make 123 NG particularly susceptible to oxidative treatment. Accord124 ingly, the NG was oxidized using nitric acid to introduce a high 125 density of carboxylic (COOH) groups, as validated by XPS 126 results. 29 The resulting NGA, rich in COOH functionalities, 127 was then mixed with an aqueous solution of Fe(NO3)3to 128 immobilize Fe cations toward the preparation of the Fe-NGA 129 catalyst (Figure 1a). 130 X-ray photoelectron spectroscopy (XPS) indicated extensive 131 defluorination and nitrogen doping upon the transformation of 132 FG to NG (as previously shown in detail 32 ). Upon further 133 transformation of NG to NGA, a substantial increase in 134 oxygen-containing groups (as carboxyls) and a decrease in 135 nitrogen content took place (as previously studied; 32 XPS of 136 the new batch of NGA is also provided here for direct 137comparison with the Fe-NGA, Figures S1 and S2). Fourier138transform infrared spectroscopy (FT-IR) of NGA showed the 139characteristic stretching band of the carboxylic groups at 1720 140 cm−1(Figure S3). The spectrum also demonstrated two 141intense bands at ca. 1550 cm−1and 1230 cm−1, both 142corresponding to skeletal vibrations of the sp2aromatic carbon 143 rings of graphene. 33,34 For the Fe-NGA, the intensity of the 144carboxylic groups decreased due to the interactions with the Fe 145cations, leading to partial ionization to carboxylates. 35 As a 146 result, the vibrations at ∼1600 cm−1and ∼1420/1350 cm−1, 147corresponding to the asymmetric and symmetric stretching of 148the −CO2 −groups, respectively, 36 increased markedly (Figure 149S3). 150 Transmission electron microscopy (TEM) revealed a flaky 151NGA/Fe-NGA morphology with lateral dimensions of ∼150 152nm, a result of oxidative cleavage during nitric acid treatment 153 (Figure 1b). High-angle annular dark-field scanning TEM 154(HAADF-STEM) showed no nanoparticles or clusters, only 155bright atomic-scale spots corresponding to individual Fe atoms 156 (Figure 1c). Elemental mapping revealed the uniform 157dispersion of iron, with no observable local Fe aggregation Figure 2. X-band EPR spectra recorded at T= 90 K and in ambient oxygen of (a, h) the neat NGA in water after 5 min sonication at 80 °C; (e, l) the neat NGA in the presence of K2CO3(100 mM); (b, (i) the Fe-NGA catalyst in neat water after sonication at 80 °C (5 min), and (c, j) after addition of H2O2(30% w/w in water; 10 μL added to the 150 μL water suspension of NGA) and 5 min sonication at 80 °C; (d, k) the Fe-NGA catalyst in the presence of K2CO3(100 mM); (f, m) the Fe-NGA catalyst in water and HMF (50 mM); (g, n) the Fe-NGA catalyst with K2CO3 (100 mM), water and HMF (50 mM). Experimental parameters for (a, h) 9.084 GHz and 0.6 mW; (b, (i) 9.079 GHz and 0.9 mW; (c, j) 9.077 GHz and 0.9 mW; (d, k) 9.081 GHz and 0.2 mW; (e, l) 9.079 GHz and 0.9 mW; (f, m) 9.074 GHz and 0.6 mW; (g, n) 9.082 GHz and 0.9 mW. 0.6 mT modulation width, 30 ms time constant, 12 min acquisition time. The EPR intensities have been divided by the square root of the applied power. (o) Molecular electrostatic potential map surface of the NGA model (Table S7) obtained from RHF/PM3 calculation (C63H33N5O24, neutral, Heat of formation = −2349.23 kJ/mol) (p) Molecular electrostatic potential map of the Fe-NGA model (Table S5) obtained from UHF/ PM3tm calculation (C63H30N5O24 ×Fe2+Fe3+ ×H2O, dication, Heat of formation = −4115.21 kJ/mol). ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX C 158 (Figure 1d−g). The content of iron was 4 wt %, as determined 159 by ICP analysis. 160 High-resolution XPS (HR-XPS) provided insight into 161 chemical environments and bonding. In NGA and Fe-NGA, 162 C 1s spectra contained sp2/sp3carbon (284.7 eV), C−N 163 (286.2 eV), and deprotonated/protonated carboxyls (287.2 164 and 288.6 eV; Figure S2a;Table S1). The O 1s spectra 165 displayed two components associated with carboxylic groups: 166 one corresponding to C�O (531.4 eV), and the other to C− 167 OH (533.1 eV; Figure 1h; Table S3). The noticeable decrease 168 in the C−OH area upon Fe ion incorporation confirms the 169 interaction between the Fe ions and the carboxylic groups on 170 the NGA support. This observation is in agreement with the 171 deprotonation and higher charge density of carboxylate 172 oxygens during carboxylic-carboxylate conversion and with 173 the FT-IR results. For the Fe-NGA catalyst, XPS analysis 174 showed 1.1 at.% in Fe (Figure S1b). The deconvolution of N 175 1s spectra for NGA and Fe-NGA showed three components, 176 reflecting −N�(sp2), −NH−(sp3), graphitic, and oxidized 177 nitrogen configurations (Figure S2b;Table S2). Fe 2p spectra 178 revealed Fe2+ (710.4 eV) and Fe3+ (712.8 eV) species with 179 characteristic satellites (Figure S2c), indicating partial Fe3+→ 180 Fe2+ reduction during immobilization. Such metal−graphene 181 charge transfer and reduction is typically observed in such 182 systems, as previously reported for Cu and Au cations using a 183 nitrile functionalized graphene. 5,37 184 The Fe K-edge XANES and EXAFS spectra clarify the 185 oxidation state and local geometry of Fe in Fe-NGA. The 186 absorption edge of Fe-NGA lies above that of Fe foil, 187 confirming the oxidized state of Fe (Figure 1i). The pre-edge 188 peak, arising from 1s →3d/4p transitions, is weak in Fe(0) 189 due to high symmetry, but pronounced in Fe-NGA, which 190 indicates a low symmetry coordination environment, con191 sistent with a distorted octahedral geometry (Figure 1i). 38,39 192 Additionally, the postedge region exhibits dampened oscil193 lations compared to Fe foil (Figure 1i), indicating that Fe is 194 atomically dispersed rather than embedded in a crystalline, 195 ordered local environment, as further confirmed by k-space 196 EXAFS (Figure S4a). 40 The FT-EXAFS is dominated by a first197 shell Fe−N/O contribution at ∼2.02 Å with a coordination 198 number of ∼6 (Figure 1j, Figure S4b, and Table S4). In the 199 phase-uncorrected R-space plots there are no intense features 200 beyond ∼2.5 Å, excluding long-range order and ruling out Fe 201 nanoparticles or extended Fe−Fe networks (Figure 1j; see also 202 k-space data in Figure S4a). 41 A small second-shell 203 contribution is captured by including a weak Fe···Fe scattering 204 path at ∼2.86 Å in the fit with a coordination number of ∼1.2 205 (Table S4), implying spatially proximate single-atom Fe sites 206 and not metallic Fe−Fe bonds or clustering. This improves the 207 fit to a subtle shoulder near ∼2.4−2.5 Å in phase-uncorrected 208 R. 209 A theoretical structural model (UHF/PM3tm) for the Fe210 NGA system (Figure 1a and Figure S5) converged in excellent 211 agreement with synchrotron and XPS results. In the model, the 212 NGA plane contains a carbon divacancy, where two Fe cations 213 are positioned in close proximity, with Fe−N (1.80−1.88 Å), 214 Fe−O (1.93−2.16 Å), and Fe−Fe (2.51 Å) distances. The 215 apical sixth coordination in the model is occupied by a water 216 molecule. The double vacancy is selected as the most stable 217 configuration in nitrogen-doped derivatives originating from 218 fluorographene. 34,42 The model is also in full agreement with 219 the XPS analysis, containing 1 part −N�(sp2), 3 parts 220 −NH−(sp3), very low graphitic nitrogen, and one-part 221 oxidized nitrogen configurations (Figure S2b), as well as 222 matching with the total carbon, oxygen, and nitrogen atoms. It 223 is noted that some iron cations could also interact with 224 carboxyl groups only, further above the plane of the NGA, as 225previously observed for other types of d-block metal cations. 29 226 2.2. The Electronic Configuration of the Catalyst. To 227 unveil the electronic spin configuration and gain further 228 insights into the structure and properties of Fe-NGA, X-band 229 EPR was performed. The EPR spectrum of neat NGA support 230 f2 in water (Figure 2a,h) showed one strong and isotropic 231resonance signal at geff = 1.998. The signal was unchanged, in 232 both intensity and signal line shape, in a basic environment 233 (0.1 M, K2CO3, pH = 11, Figure 2e,l). This EPR signal arises 234 from the spin-containing defects located on a carbon center, 235 which belongs to the NGA framework. This clearly indicates 236 that the NGA support remained stable with respect to the spin 237 defects concentration, under the reaction conditions used later 238for the catalytic reaction. 239 The EPR fingerprints of the Fe-NGA catalyst were 240 substantially different from those witnessed in neat NGA. 241 When Fe-NGA was suspended in neat water, the spectrum 242 (Figure 2b) displayed two distinct resonance features. In place 243 of the strong and narrow isotropic resonance signal at geff = 244 1.998 (ΔBpp = 0.7 mT) in neat NGA, a much broader and 245 weaker in intensity resonance feature (ΔBpp = 1.7 mT) 246emerged at geff = 1.997 in Fe-NGA. The signal change of the 247 spin-containing defects in the NGA framework is affected by 248 the presence of the paramagnetic Fe cations, causing a signal 249 broadening due to the dipolar magnetic interactions acting on 250 spin centers of different natures (Fe, C-radicals). The strength 251 of these interactions, and the broadening effects are inversely 252 proportional to the distance (r) of the two different spin 253 centers, ΔBbr ∝g2β2SFe.Srad/r3. 43 The second resonance signal 254 that appears in the low magnetic field region (geff > 4.00) 255corresponds to the middle-Kramer doublet (ms±3/2) of high 256spin (S= 5/2) Fe3+ cations coordinated to the NGA backbone. 257 The observed signal anisotropy indicates that the Fe3+ cations 258 are entrapped in different coordination environments; a 259 fraction of Fe3+ cations experience strong rhombic distortion, 260 as given by the E/D ratio of the zero-field-splitting components 261 (E/D ∼0.3, geff = 4.25), while the resonance shoulder around 262 geff = 5.24 indicates the presence of bound Fe3+ cations with 263 significantly weaker rhombic field (E/D ∼0.2). The notations 264 D and E correspond to axial and rhombic zero-field-splitting 265terms of high-spin Fe3+, respectively. Moreover, the EPR 266 spectra performed on the Fe-NGA sample in water with or 267 without a sonication step at 80 °C revealed no net changes in 268 the signal shape and intensities associated with the Fe3+ and 269 radical centers, validating the stability of the Fe-NGA catalytic 270 system, from the EPR signatures perspective, in the temper271 ature range close to that used during HMF oxidation. Upon 272 direct addition of H2O2inside the EPR tube containing the 273 water dispersion of Fe-NGA, the EPR spectrum showed an 274increase in the signal intensities of both the radical centers 275located on NGA, as well as the signal associated with the Fe3+ 276 cations (Figure 2c,j). Comparison of the double integrated 277 signal intensity of the overall Fe3+ resonances (B-field range of 278 100−200 mT) before and after the addition of hydrogen 279 peroxide indicated that ∼30% of the Fe cations were present in 280 NGA in the reduced form, as Fe2+. The higher Fe3+/Fe2+ ratio 281 observed in EPR in comparison to that in XPS is ascribed to 282 the different measurement conditions: EPR is performed in 283water, where oxidation of Fe2+ is commonly observed, whereas ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX D 284 in XPS, the sample is measured in a dry state and under inert 285 conditions. It should be noted that these Fe2+ centers can 286 adopt either the high-spin configuration (S= 2, integer spin 287 system), with large zero-field-splitting components (D, E > 0.3 288 cm−1), or can be present in the low-spin configuration (S= 0); 289 both spin configurations, at X-band frequency, result into EPR 290 silent species. 291 When Fe-NGA was dispersed in water with a base (10 μL of 292 K2CO3solution to reach a pH ∼11), and the solution heated 293 at 80 °C for 5 min, followed by fast freeze-quench, a very 294 strong, narrow (ΔBpp = 1.0 mT) and anisotropic resonance 295 signal is recorded (T= 90 K), expressing gtensor parameters at 296 g1= 2.000, g2= 1.997, g3= 1.993 (geff = 1.997) (Figure 2d,k). 297 Moreover, no resonance signal attributable to Fe3+ cations was 298 detected. The strong resonance fingerprint at geff = 1.997 299 differs substantially from the S = 1/2 weak signal associated 300 with the carbon-based spin-containing defects seen in the neat 301 NGA framework, both in the presence (Figure 2e,l) and 302 absence (Figure 2a,h) of base. In addition, the appearance of 303 such a signal (geff = 1.997, Figure 2d,k) was independent of the 304 specific base used. Substituting K2CO3with 0.05 M NaOH 305 produced an identical resonance feature. This signal dis306 appeared upon neutralizing the basic environment to pH 7, 307 indicating that it is linked to the Fe spin-active species involved 308 in a pH-dependent equilibrium. We suggest that such a strong 309 S= 1/2 resonance signal, with small g-anisotropy, originates 310 from the formation of a mixed-valence Fe3+-Fe2+ dimer 311 structure, where the two iron centers are bridged by a μ312 hydroxo group [Fe3+−OH-Fe2+]. In such a scenario, the S= 5/ 313 2 (Fe3+) and S= 2 (Fe2+) states interact antiferromagnetically, 314 giving a radical-like S= 1/2 signature. Comparable signals in 315 the g= 2.00 region, with radical-like feature and S= 1/2 spin 316 configuration, have been observed in several dinuclear Fe 317 complexes (μ-hydroxo) in which the mixed valence states 318 (Fe2+/Fe3+) were conveniently produced by radiolytic 319 reduction of the ferric dimers 44 or by chemical conversion of 320 μ-oxo into the μ-hydroxo systems, as those seen in diiron 321 dipyrrin Pacman complexes. 45 To study the catalytic potency 322 of these Fe dimers in Fe-NGA to perform in oxidation 323 f3 reactions, we used HMF as a model substrate (Figure 3a) and 324 monitored the changes in the EPR signals of Fe-NGA during 325the oxidation process. 326 The EPR spectrum recorded after adding HMF to the 327aqueous Fe-NGA suspension and sonicated at 80 °C for 5 min Figure 3. (a) HMF oxidation to DFF, (b) Catalytic performance benchmarking with control experiments, (c) Recyclability of the Fe−NGA catalyst in the oxidation of HMF to DFF. The GC yields of DFF for each cycle are indicated above the corresponding bars, and (d) comparison of the performance (TOF and SP) of previously reported catalysts with the Fe-NGA for HMF oxidation to DFF. Reaction conditions: 0.125 mmol HMF, 3.1 mg Fe-NGA (1.7 mol % Fe; HMF/metal molar ratio 56:1), 0.5 MPa O2, 0.25 mmol K2CO3, and 2 mL water. In the case of Fe-salts and other Fe-materials, the amount of catalyst corresponds to 1.7 mol % of Fe. Conversion and yields were determined by HPLC. ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX E 328 (without base; Figure 2f,m) did not show the signal associated 329 with the formation of the mixed-valent state Fe3+-μ−OH-Fe2+ 330 system, confirming that the basicity is key for the Fe3+-μ−OH331 Fe2+ generation. Furthermore, the resonance associated with 332 the presence of high-spin Fe3+ centers became much weaker, as 333 seen from the intensity of the signals evolving at geff = 4.25 and 334 geff = 5.24. The weakened intensity, compared to the Fe3+ 335 signal intensity seen in Figure 2b,i (without HMF, without 336 base), indicates that HMF interacts with the Fe3+ metal 337 centers, resulting in their reduction to Fe2+, which is EPR silent 338 at X-band. However, only a small fraction of the total Fe in 339 NGA appears to be active in the absence of base, as 340 experimentally observed (vide infra). When the base is 341 added, the electronic properties of the Fe-NGA system 342 dramatically changed (Figure 2g,n), showing again a strong S 343 = 1/2 resonance at geff = 1.997, signaling the Fe3+μOH-Fe2+ 344 formation, complemented by a well-defined and sharp Fe3+ 345 component (geff = 4.25). Taking all the information gained 346 from EPR measurements and analysis, we suggest that in the 347 Fe-NGA catalyst, the Fe-centers act cooperatively as Fe pairs 348 during HMF oxidation. 349 The organization and electronic features of Fe-NGA mimic 350 the behavior known in binuclear nonheme iron enzymes that 351 activate oxygen in biocatalytic processes. 28 In these natural 352 systems, a strong radical signal that expresses small g353 anisotropy is seen at g∼2.00, displaying nearly identical 354 fingerprints to those shown in Figure 2d,k,g,n. This signal is 355 known to originate from a [Fe3+-μ-oxo/μ-hydroxo-Fe4+] or 356 [Fe3+-bis μ-oxo-Fe4+] species formed upon O2activation at the 357 diferrous site. Such a signal, in the nonheme di-iron system 358 ribonucleotide reductase R2 (RNR R2), has been termed 359 intermediate X. 46−49 A nearly identical radical signal, observed 360 here for FeNGA in K2CO3and K2CO3+ HMF around geff = 361 2.00, has also been reported for the tris(3,5-dimethyl-4362 methoxylpyridyl-2-methyl)amine ligand chelating a di-iron 363 center. In this case, a valence-localized [HO−Fe3+−O− 364 Fe4+�O] open core is formed upon addition of OH−. 50 365 Therefore, we suggest that Fe-NGA is capable of exhibiting 366 high redox flexibility, forming the initial Fe dimer core (Fe3+- 367 μ−OH-Fe2+ in neat base), and the Fe3+−(μ−OH)-O2−Fe4+ 368 ferryl-containing species upon addition of the base and 369 substrate in the presence of O2. The formation of the high370 valent Fe(IV) species during HMF catalysis should signifi371 cantly enhance the oxidation process. 372 The formation in the Fe-NGA of a pocket prone to favor 373 oxidation is also reflected in the computed molecular 374 electrostatic potential maps from geometry-optimized models 375 for neat NGA (RHF/PM3) and Fe-NGA (UHF/PM3tm) 376 (Figure 2o,p). Significant changes in the electron density 377 distribution upon Fe binding are observed. The initially 378 negative (electron-rich, red color) areas around the C 379 divacancy in the NGA plane shift to positive (electron-poor, 380 blue color) regions, indicating that these sites are prone to 381 interact with electron donors, for instance, with a substrate 382 undergoing oxidation reaction. 383 2.3. Catalytic Performance. To exploit the unique 384 oxidizing features of the Fe-NGA catalyst (cooperative oxygen 385 activation in water and ferryl species formation), we evaluated 386 the activity for the oxidation of HMF in more detail (Figure 387 3a). Reaction parameters (temperature, catalyst amount, base, 388 and O2pressure) were optimized (Table S5). The base 389 (K2CO3) plays a critical role in the catalytic system. Its primary 390 function is to facilitate the formation of the Fe3+-μ−OH-Fe2+ 391 dimer species by providing hydroxide ions that bridge the Fe 392 centers, as supported by spectroscopic and mechanistic 393 analyses. To evaluate the effect of base, we also tested 394 NaOH and K3PO4. NaOH promoted catalytic activity but also 395increased the formation of byproducts. K3PO4resulted in 396 lower HMF conversion. These observations indicated K2CO3 397 as the most appropriate base for the reaction. Under optimum 398 conditions (100 °C, 0.25 mmol K2CO3, 0.5 MPa of O2), Fe399 NGA showed a remarkable performance for the selective 400 oxidation of HMF to DFF in pure water. The Fe-NGA catalyst 401 with 4 wt % Fe loading achieved almost complete HMF 402 conversion (97%) with 95% DFF selectivity, a TOF of 17.3 403 h−1, and a specific productivity of 12.5 mmolDFF g−1h−1 404(Figure 3b,d). To probe the role of adjacent Fe centers for 405 effective oxidation, a Fe-NGA catalyst with low Fe loading (2 406 wt %) was evaluated under identical conditions. In this case, 407 the possibility of having adjacent Fe cations is significantly 408 reduced due to the 2-fold lower Fe-loading. The total catalyst 409 amount in the reaction was adjusted to maintain the same 410 overall Fe content, as in the case of the 4 wt % Fe-NGA 411 catalyst. Despite this adjustment, the catalytic activity 412 significantly decreased, with a conversion of only 48% and a 413TOF of 8.3 h−1. This finding supports the EPR results, 414 highlighting the crucial role of the Fe metal-ion synergy for 415achieving the highest performance. 416 To confirm that the activity and selectivity of the catalyst are 417not exclusively attributed to any of its individual components, 418control experiments were conducted (Figure 3b). A negligible 419 catalytic activity was observed for NGA alone. Iron(III) nitrate, 420 in the absence of NGA, led to 23% HMF conversion, with 20% 421 DFF selectivity and a TOF of 0.9 h−1. When ferrous chloride 422 was used as the sole catalyst, the DFF selectivity was only 19%, 423 giving a TOF of 2.4 h−1. Additionally, a mixture of iron(II) and 424 iron(III) salts was tested, which resulted in 30% HMF 425 conversion, 62% DFF selectivity, and TOF of 3.5 h−1. These 426outcomes show that freely diffusing Fe ions do not reproduce 427 the activity-selectivity balance of Fe-NGA and underscore the 428 need for preorganized Fe sites on NGA to enable controlled 429 O2activation. When the selectivity toward DFF was lower than 430 in the optimized conditions, product analysis revealed the 431 presence of intermediate oxidation products, such as HMFCA 432 and trace amounts of FFCA. Several additional unidentified 433 peaks also appeared in the chromatograms, especially under 434 harsher reaction conditions. These signals likely correspond to 435products formed via well-known secondary pathways, including 436 degradation, condensation of HMF or its intermediates, and 437 polymerization (humins). Their formation is consistent with 438 the complex reaction network typically associated with HMF 439 oxidation. Another control experiment highlighted the 440 importance of the immobilization of Fe3+ cations on the 441 NGA support and of the stable binding of Fe under turnover. 442 This experiment involved the addition of the Fe(NO3)3salt 443 together with NGA. Unlike the case of adding only Fe(NO3)3, 444the presence of NGA restored performance to levels 445 comparable with Fe-NGA, highlighting the role of support in 446 anchoring Fe under turnover and creating the active environ447 ment. The catalyst exhibited good recyclability; however, a 448 decrease in HMF conversion was observed mainly in the first 449 cycles (1st cycle 12.5%; second cycle 5%, third cycle 2%, fifth 450 cycle 1%, Figure 3c). The activity loss is ascribed to small 451 catalyst losses during the recovery steps, due to its high 452hydrophilicity and dispersibility in water (the solvent of the 453reaction). XPS analysis of the spent catalyst (Figure S8, ESI) ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX F 454 revealed no significant structural changes or loss of active metal 455 sites. Both the atomic percentages and the high-resolution Fe 456 2p spectra remained unchanged compared to the fresh catalyst. 457 2.4. Evaluation of Results in the Context of the State458 of-the-Art. To correlate the performance of the Fe-NGA 459 catalyst in context to the state-of-the-art, we compared the 460 TOF and SP values (Figure 3d and Table S6). Homogeneous 461 catalysts, such as oxovanadium complexes or VOSO4/Cu462 (NO3)2demonstrated some of the highest TOF for HMF 463 oxidation to DFF. 21,51 The oxovanadium complex achieved a 464 TOF of 9.6 h−1(Table S6, entry 3), 21 while VOSO4/ 465 Cu(NO3)2displayed a TOF of 4.9 h−1(Table S6, entry 14). 51 466 However, the inherent challenges of catalyst separation and 467 nonrecyclability associated with homogeneous systems have 468 led to the development of heterogeneous catalysts. In this 469 regard, vanadium complexes heterogenized on carbon supports 470 showed promising results, achieving a production rate of 6.1 471 mmolDFF g−1h−1; however, with a quite low TOF of 5.7 h−1 472 (Table S6, entry 4). 22 Several Ru-based heterogeneous systems 473 have also been explored for this transformation. 17,18,25,52 For 474 example, Ru supported on γ-Al2O3exhibited poor DFF 475 selectivity (21.2%), despite a decent HMF conversion of 476 91.2%, likely due to overoxidation and other side reactions 477 (Table S6, entry 10). 17 On the other hand, Ru supported on 478 covalent triazine frameworks (Ru@CTF) delivered a remark479 able performance with a TOF of 9.4 h−1and an SP of 3.5 480 mmolDFF g−1h−1, highlighting the advantage of tailored porous 481 supports in achieving improved selectivity (Table S6, entry 482 12). 18 Fe2O3@HAP-Ru reached full conversion with a TOF of 483 5.9 h−1and an SP of 1.2 mmolDFF g−1h−1, underscoring the 484 synergistic effect between iron oxide and ruthenium on 485 hydroxyapatite (Table S6, entry 13). 25 Ru/MnCo2O4also 486 delivered 98.3% conversion and complete selectivity for DFF, 487 although with a low TOF of 0.5 h−1(Table S6, entry 16). 52 488 Gold-based catalysts, such as Au NPs@sPSB, showed 489 moderate activity with a TOF of 1.0 h−1and SP of 0.1 490 mmolDFF g−1h−1. However, the lower DFF selectivity (80%), 491 possibly due to competing oxidation pathways, remains a 492 drawback (Table S6, entry 11). 24 PdNi supported on MnO2 493 achieved full conversion and excellent selectivity (99%) within 494 1 h, with a SP of 6.3 mmolDFF g−1h−1, albeit a modest TOF of 495 0.8 h−1(Table S6, entry 8). 16 Among non-noble catalysts, 496 Mn5O8and Mn6Fe1Ox achieved TOF values of 0.9 h−1and 497 ∼0.2 h−1, respectively, suggesting that while selectivity remains 498 high (>94%), the activity is substantially lower than 499 homogeneous catalysts (Table S6, entries 7 and 15). 53,54 500 FeCo/C attained full conversion with excellent selectivity 501 (99%), but its TOF and SP remained comparatively low (0.8 502 h−1and 0.4 mmolDFF g−1h−1, respectively) (Table S6, entry 503 9), 55 further illustrating the limitations in the activity of earth504abundant transition metal-based catalysts. Figure 4. (a) Proposed reaction mechanism (main pathway) for the oxidation of HMF by Fe-NGA catalyst in a basic environment. (b) The computational models (UHF/PM6) show the Fe-NGA intermediates involved in the HMF catalysis, highlighting the variation of Fe−Fe distances (Å) within the catalytic cycle. Hfindicates Heat of Formation energy (kJ/mol) and < S2> the expectation value. Spin density isosurfaces (0.002 IsoVal) are shown in the bottom structures. ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX G 505 While heterogeneous and homogeneous catalysts have 506 shown varying degrees of success, photocatalysts have gained 507 significant attention for their potential in renewable energy and 508 environmental applications. However, with respect to HMF-to509 DFF oxidation, their performance remains suboptimal. For 510 instance, S-scheme heterojunctions between ZnIn2S4with 511 sulfur vacancies and CeO2have been explored for HMF 512 oxidation to DFF, achieving a production rate (SP) of 0.9 513 mmolDFF g−1h−1(Table S6, entry 6). 56 Similarly, oxygen514 doped ZnIn2S4nanosheets with atomic-scale edge steps and 515 lattice defects showed improved performance, achieving an SP 516 of 1.6 mmolDFF g−1h−1(Table S6, entry 5). 12 A Ru complex 517 supported on CdS quantum dots catalyzed HMF photo518 oxidation to DFF, showing very low activity with a TOF of 519 0.03 h−1and an SP of 0.05 mmolDFF g−1h−1(Table S6, entry 520 2). 13 Additionally, a photocatalyst comprising Cu−N4and C− 521 S−C as dual active sites supported on carbon nitride produced 522 DFF with an SP of 0.3 mmolDFF g−1h−1and a TOF of 1.2 h−1 523 (Table S6, entry 1). 23 524 Despite these advances, three constraints recur across the 525 literature: (i) overoxidation (suppressing DFF selectivity), (ii) 526 low turnover frequencies under practical conditions, and (iii) 527 the reliance on organic solvents. Particularly critical is the third 528 constraint, as crude HMF streams contain water from fructose 529 dehydration, and, in most cases, even trace amounts of H2O 530 often degrade DFF selectivity and/or catalyst activity 18,24,25,52 531 (see also Figure 3d for comparative activities in organic 532 solvents and in water). In contrast, the Fe-NGA catalyst 533 demonstrates unprecedented performance in pure water with 534 97−100% HMF conversion, 93−95% selectivity to DFF, a 535 TOF of 17.3−17.4 h−1, and an SP of ∼12.5 mmolDFF g−1h−1. 536 These figures not only surpass most noble-metal and 537 homogeneous benchmarks in activity per active metal site 538 but also demonstrate robust compatibility in aqueous-phase, 539 effectively addressing a key bottleneck in sustainable HMF 540 upgrading. We attribute this performance to the NGA 541 framework’s ability to stabilize redox-flexible Fe−Fe dimers, 542 which form the highly active catalytic center (ferryl-based iron 543 dimer) in a basic aqueous environment. This, in turn, activates 544 O2cooperatively with high selectivity and rates, even in water. 545 Moreover, to assess the scalability of the Fe-NGA system, the 546 oxidation of HMF was conducted at a 10-fold higher substrate 547 concentration under unoptimized conditions. Remarkably, 548 complete HMF conversion and 96% DFF selectivity were 549 achieved after 12 h, demonstrating that the catalyst retains 550 excellent activity and selectivity even at higher substrate 551 loadings. This result further highlights the robustness and 552 potential industrial applicability of Fe−NGA for aqueous553 phase aerobic oxidation. Collectively, these results position 554 Fe−NGA as a practical blueprint for high-rate, selective 555 aerobic oxidations directly in water. 556 2.5. Reaction Mechanism. Considering the obtained 557 results, the presence of adjacent Fe cations (Fe2+ and Fe3+) in a 558 bridged configuration [Fe3+-μ−OH-Fe2+], and the emergence 559 of activated oxygen species under turnover [Fe3+-(O2)(μ− 560 OH)-Fe4+], the possible mechanism was explored. The two Fe f4 561 sites are labeled as Site (1) and Site (2), as shown in Figure 4 562 (with additional information on intermediates in Figure S9). In 563 the NGA framework, two closely interacting Fe centers can be 564 present as a statistical combination of Fe3+-Fe2+, Fe3+-Fe3+, and 565 Fe2+-Fe2+ active pairs in the resting state. Note that the Fe3+- 566 Fe3+ and Fe2+-Fe2+ combinations, in the presence of base, 567 cannot produce the strong radical-like signal observed in 568 Figure 2d,k, because they are known to give EPR silent 569 spectra. 46 Therefore, two pathways are possible. The one 570 shown in Figure 4a involves the formation of the spin-active 571 dimers [Fe3+-μ−OH-Fe2+] and [Fe3+-(O2)(μ−OH)-Fe4+]. 572 The other pathway, starting from Fe3+-Fe3+ and Fe2+-Fe2+ 573 couples expressing activate intermediates [Fe3+-(μ−OH)2574 Fe3+] and [Fe4+-(O2)-Fe4+], is shown in Figure S9. In the 575 aerobic oxidation of HMF to DFF, oxygen species such as 576 superoxide radical (O2 •−), peroxides (O2 −), singlet oxygen 577 (1O2), or hydroxyl radical (•OH) are thought to play key roles 578 in the oxidation process. Generally, O2acts as the terminal 579 oxidant; the alcohol is initially dehydrogenated on the metal 580 catalyst, which undergoes reduction (−CH2OH of HMF being 581 oxidized to −CHO in DFF), and then O2reoxidizes the 582 catalyst. However, in Fe-NGA the activated oxygen specie 583 (peroxide) bound to the Fe-centers directly acts in the HMF 584 oxidation process. From the Fe3+-Fe2+ pair in the resting state, 585 the addition of base rapidly forms intermediate I (Fe3+-(μ− 586 OH)-Fe2+,S= 1/2) (Figure 4a), which activates an oxygen 587 molecule and transforms into the intermediate II (Fe3+-(μ− 588 OH)-(O2)-Fe4+), the hydroxo-peroxo intermediate (S= 1/2). 589 These intermediates (I) and (II) give identical EPR signals as 590 experimentally observed in Figure 2d,k. The HMF substrate 591 then interacts with the active Fe3+-Fe4+ site (see also Figure 592 S10, Monte Carlo simulations), delivering two electrons and 593 two protons to the Fe bound peroxo-specie, triggering the 594 release of water molecule and formation of a ferryl 595 intermediate (Fe4+�O, intermediate III). This provides the 596 first evidence under turnover of an uncoupled Fe−Fe system 597 containing Fe3+−OH and Fe4+�O centers. The ferric center 598 (Fe3+−OH) gives sharp and isotropic EPR signatures seen in 599 Figure 2g, with g= 4.25, while the Fe4+�O site is EPR silent. 600 A second HMF molecule then reacts with the high-valent 601 Fe4+�O, in another proton-coupled two-electron transfer, 602 releasing a second water molecule, generating intermediate 603 (IV), Fe3+−OH, and Fe2+. These sites reform the bridged 604 Fe3+/Fe2+ dimeric unit through the excess in solution of 605 OH−anions, restarting the catalytic cycle back to intermediate 606 I. Note that the reaction step I →II, which involves the oxygen 607 binding and activation, is endothermic (PM6 method), and 608 requires supply of thermal energy (T) (Figure 4b). This is 609 consistent with our experimental evidence that the oxidation 610 reaction to proceed needs both high temperature (100 °C) and 611 O2pressure, so to express effective substrate oxidation. 612 Moreover, in the Fe3+-Fe2+ mixed valent spin configuration, 613 the Fe3+ site is expected to provide the center interacting with 614 the −CHO moiety of HMF (Figure S10), directing the HMF 615 alcoholic residue (−CH2−OH) toward the Fe2+ site, involved 616 in O2binding and activation. Thus, the dimeric Fe-centers act 617 synergistically to enhance the catalytic activity and selectivity 618of the Fe-NGA. 619 The mechanism is compared to nonheme diiron enzymes to 620 illustrate the formation of high-spin species; however, there is 621 currently no direct evidence that HMF is cooperatively 622 activated by both iron centers in the Fe3+-μ−OH-Fe2+ dimer. 623 The proposed mechanism shows sequential oxidation steps 624 mediated by the dimer, with both iron atoms necessary for 625 catalytic activity, but it does not imply simultaneous 626 cooperative activation of the substrate. Determining whether 627 cooperative activation occurs will require further mechanistic 628investigation. ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX H 3. CONCLUSION 629 We have developed a robust oxidation catalyst comprising 630 redox-flexible single-atom iron dimers stabilized on nitrogen631 doped graphene acid. The Fe-NGA catalyst integrates enzyme632 like active center features with the stability of heterogeneous 633 catalysts, enabling selective aerobic oxidation of an important 634 biomass-derived platform chemical under mild conditions in 635 pure water. Comprehensive spectroscopic analyses revealed 636 that under catalytic turnover, Fe-NGA mimics nonheme diiron 637 oxidases by generating a high-valent oxo-bridged Fe3+−Fe4+ 638 intermediate. This enables efficient, proton-coupled two639 electron oxidations with high rates and remarkable selectivity, 640 even in pure water, surpassing state-of-the-art systems. Thus, 641 Fe-NGA catalyzed the complete HMF transformation to DFF 642 with 93% selectivity, and a turnover frequency of 17.4 h−1, 643 when the reaction was performed at 1 MPa of O2(entry 4, 644 Table S5). Mechanistic studies reveal that the cooperative 645 action of adjacent Fe centers is essential for O2activation and 646 substrate oxidation, with the NGA support providing the 647 precise coordination environment required to stabilize redox648 switchable Fe2+/Fe3+/Fe4+ states. Importantly, this study 649 introduces a previously unreported μ-hydroxo-bridged Fe2+− 650 Fe3+ active-site architecture on a graphene-based support, 651 formed in the presence of base, which serves as the true 652 catalytically competent structure. This structurally defined and 653 mechanistically validated diiron motif represents a new class of 654 heterogeneous active centers that emulate enzymatic redox 655 cooperation while operating in fully aqueous conditions. Such 656 a discovery bridges biological and synthetic oxidation 657 chemistry, offering a conceptual and structural advance in 658 the design of earth-abundant, sustainable oxidation catalysts. 659 This cooperative, redox-flexible manifold opens previously 660 inaccessible mechanistic pathways in heterogeneous oxidation 661 catalysis, offering a remarkable paradigm in aerobic oxidations 662 with heterogeneous synthetic catalysts. This work offers a 663 blueprint for developing next-generation heterogeneous 664 systems for the green valorization of renewable feedstocks. 4. EXPERIMENTAL DETAILS 665 4.1. Synthesis of Fe-NGA Catalyst. NGA was synthe666 sized as previously described. 29 0.5 g of fluorographite was 667 dispersed in 30 mL of DMF, followed by 24 h of sonication to 668 ensure uniform dispersion. Then, 3 g of NaN3was introduced, 669 and the mixture was transferred to a round-bottom flask with a 670 condenser and stirred at 130 °C for 3 days. The resulting 671 product was subjected to purification by sequential washes 672 with DMF, acetone, ethanol, water, and hot water, with each 673 wash followed by centrifugation at 14000 rcf. A portion of this 674 N-doped graphene was then oxidized by treatment with 45% 675 HNO3at 100 °C for 24 h in a glass flask with a condenser. The 676 oxidized material was further purified by repeated washes with 677 hot water and then dialyzed using a cellulose membrane (14 678 kDa cutoff). Finally, the oxidized nitrogen-doped graphene 679 acid was mixed with an aqueous Fe(NO3)3solution and stirred 680 at room temperature for 24 h. The resulting material was then 681 washed with water and freeze-dried to obtain the Fe-NGA 682 catalyst. 683 4.2. Catalytic Tests. The catalytic oxidation of the HMF 684 was carried out in a 25 mL pressure reactor. The reactor was 685 charged with 0.125 mmol HMF, 3.1 mg Fe-NGA catalyst (1.7 686 mol % Fe, HMF/metal molar ratio was 56:1), 0.25 mmol 687 K2CO3, and 2 mL water, and the reaction mixture was 688sonicated for 5 min. The reaction was performed under a 689constant O2pressure (typically 0.5 or 1 MPa) and maintained 690at the desired reaction temperature (100 °C) for a specific time 691(typically 3 h) with a stirring rate of 700 rpm. The reaction was 692quickly terminated by cooling the reactor to room temperature 693in an ice bath, and aliquots were taken from the mixture for 694product analysis by HPLC, as described in the Supporting 695Information. Larger scale testing was performed by increasing 696all reagents by 10-fold, except of the amount of water due to 697reactor limitations. 698Additional experimental details are available in the 699Supporting Information file (Chemicals, Characterization, 700Product analysis from the catalytic reaction, Supplementary 701results figures and tables). 702 ■ASSOCIATED CONTENT 703Data Availability Statement 704The data that support the findings of this work are openly 705available in Zenodo under the same title. 706* sı Supporting Information 707The Supporting Information is available free of charge at 708https://pubs.acs.org/doi/10.1021/acscatal.5c06280. 709Materials and methods; Figure S1: the XPS survey 710spectra of NGA support and Fe-NGA catalyst; Figure 711S2: the HR-XPS of the NGA support and Fe-NGA 712catalyst for the spectral regions of C 1s, N 1s, and Fe 2p; 713Figure S3: FT-IR spectra of NGA and Fe-NGA; Figure 714S4 reports the XAS analysis; Figure S5: the geometry 715optimized structures of NGA support and the Fe-NGA 716catalyst; Figure S6: calibration curves for HMF and for 717DFF; Figure S7: a representative chromatograph of 718product analysis for the HMF oxidation; Figure S8: the 719XPS analysis of the reused Fe-NGA catalyst; Figure S9: 720the other possible reaction mechanism for HMF 721oxidation by the Fe-NGA catalyst; Figure S10: the 722interaction of HMF with Fe-NGA in the presence of 723intermediate [II]; Table S1: comparative deconvoluted 724areas for C 1s region; Table S2: comparative 725deconvoluted areas for N 1s region; Table S3: 726comparative deconvoluted areas for O 1s region; Table 727S4: the structural parameters for Fe-NGA catalyst; Table 728S5: reaction optimization studies; Table S6: the 729comparative overview of the Fe-NGA with previously 730reported catalysts for HMF oxidation; Table S7: the 731Cartesian coordinates for the NGA system; Tables S8− 732S13: the Cartesian coordinates for the Fe-NGA system 733in different models (PDF) 734 ■AUTHOR INFORMATION 735Corresponding Authors 736Giorgio Zoppellaro −Nanotechnology Centre, Centre for 737Energy and Environmental Technologies, VSB−Technical 738University of Ostrava, Ostrava 708 00, Czech Republic; 739Regional Centre of Advanced Technologies and Materials, 740Czech Advanced Technology and Research Institute 741(CATRIN), PalackyUniversity Olomouc, Olomouc 783 71, 742Czech Republic; Email: [email protected] 743Aristides Bakandritsos −Nanotechnology Centre, Centre for 744Energy and Environmental Technologies, VSB−Technical 745University of Ostrava, Ostrava 708 00, Czech Republic; 746Regional Centre of Advanced Technologies and Materials, 747Czech Advanced Technology and Research Institute ACS Catalysis pubs.acs.org/acscatalysis Research Article https://doi.org/10.1021/acscatal.5c06280 ACS Catal. XXXX, XXX, XXX−XXX I