Introduction

Propylene oxide (PO), a pivotal chemical intermediate, serves as an essential precursor for the synthesis of propylene glycol (PG), polyurethane foams, propylene carbonate, and other industrial chemicals1,2,3. The existing industrial PO production predominantly relies on propylene oxidation through processes such as the hydrogen peroxide propene oxide (HPPO) method3,4,5,6,7,8. Recent advancements in electrochemical synthesis have highlighted the potential utilization of the electrolytic cell for chemical production9,10,11,12,13,14,15, positioning the direct electrochemical oxidation of propylene as a promising alternative route for PO synthesis. This route is environment-friendly and can be operated under mild reaction conditions, but suffers from low reaction selectivity and yield. Therefore, further exploration of electrocatalysts is essential to regulate the reaction pathway, improve the reaction kinetics, and eventually enhance the reaction efficiency.

Noble-metal-based catalysts, including Pd16,17,18, Pt18,19,20,21, and Ag22,23,24 have been studied for propylene electrooxidation. However, challenges persist due to ambiguous reaction mechanisms and complex products, which contain PO, PG, acrylic acid, and other derivatives. For example, in the mechanistic study of the Ag3PO4 catalyst, two distinct reaction pathways (OH-mediated and O-mediated pathways) for electrochemical propylene oxidation were discussed, selectively producing propylene oxide and acrylic acid, respectively23. These observations emphasize the critical need for the development of efficient epoxidation electrocatalysts with well-defined active sites and well-controlled reaction pathways.

Single-atom catalysts (SACs), with atomically dispersed active centers, offer distinct advantages in regulating reaction networks through precise electronic and geometric modulation25,26,27,28,29. For instance, decorating Rh single atoms onto nanoporous mesh-type NiFe layered double hydroxides enhanced the adsorption of 5-hydroxymethylfurfural and promoted its subsequent oxidation to generate 2,5-furandicarboxylic acid30. Given the inherent complexity of propylene oxidation products and the proposed different reaction intermediates of *O, *OH, and *OOH, the synergistic catalysis between spatially isolated catalytic sites and support in SACs may offer a promising approach to steer reaction pathways toward selective PO formation. This approach necessitates an atomic-level identification of active sites and mechanistic understanding to guide the design of dual-site catalysts. To understand the reaction mechanism of electrochemical propylene oxidation, it is essential to elucidate the dynamic evolution of catalytic structures and reactive intermediates31,32,33. In-situ characterizations were applied to capture the reactive intermediates and transition of surface status in the electrochemical propylene oxidation reaction34,35,36,37. For example, the formation of μ-C = CHCH3 and evolution of propylene intermediates on PdO/C electrocatalyst at high anodic potentials were detected by in-situ ATR-FTIR, uncovering a distinct reaction pathway different from thermal catalysis16.

In this work, a series of noble-metal decorated FeOOH electrocatalysts is designed and prepared for electrochemical propylene oxidation reaction, among which, the single-Ag-atom decorated FeOOH electrocatalyst (Ag1-FeOOH) displays the optimized performance, achieving a PO Faradaic efficiency (FEPO) of 32.0% at an anodic potential of 2.4 V vs. RHE. In-situ ATR-SEIRAS and in-situ 57Fe Mössbauer spectroscopy measurements reveal propylene adsorption at FeOOH and reactive oxygen species generation at single-Ag-atom sites. DFT calculations indicate that the single-Ag-atom sites on FeOOH can decrease the energy barrier for generating *O intermediate, facilitating *O and *C3H6 coupling to form PO.

Results

Synthesis and structural characterization

The Ag-decorated FeOOH catalysts were synthesized via a conventional hydrothermal method, followed by metal loading through an impregnation approach (Fig. S1). The concentration of Ag+ in the impregnation solution was adjusted to synthesize Ag-FeOOH with different Ag loadings, yielding the following four distinct Ag-FeOOH catalysts: Ag1-FeOOH-1 and Ag1-FeOOH-2, which only contain atomically dispersed Ag species, and Ag-FeOOH-2.5 as well as Ag-FeOOH-3, which contains Ag nanoparticles. Inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis gave the actual Ag loadings of Ag1-FeOOH-1 (0.3 wt.%), Ag1-FeOOH-2 (1.7 wt.%), Ag-FeOOH-2.5 (4.6 wt.%), and Ag-FeOOH-3 (9.2 wt.%) (Table S1).

Transmission electron microscopy (TEM) measurements showed that the as-synthesised FeOOH consisted of well-defined nanorods with an average length of ~100 nm (Fig. S2a). High-resolution TEM (HRTEM) analysis identified a characteristic lattice spacing of 0.53 nm, corresponding to the 200 plane of β-FeOOH (PDF #34-1266) (Fig. S2b). The scanning electron microscopy (SEM) images showed that FeOOH exhibited a nanorod morphology (Fig. S3), consistent with TEM imaging. After loading Ag, the Ag-FeOOH-3 sample maintained the original nanorod morphology. Specifically, the TEM image revealed the presence of Ag nanoparticles with an average size of 5.4 nm, which were uniformly decorated on the FeOOH nanorod surface (Figs. 1a and S4). HRTEM image of these nanoparticles revealed a distinct lattice spacing of 0.23 nm attributable to the 111 plane of metallic Ag (PDF #04-0783, Fig. 1b). As the Ag loading amount decreased, Ag-FeOOH-2.5 exhibits similar morphological features to those of Ag-FeOOH-3 (Fig. S5). The Ag1-FeOOH-2 sample preserved the nanorod morphology without observable Ag nanoparticles (Fig. 1c). HRTEM image of Ag1-FeOOH-2 revealed only a lattice spacing of 0.53 nm, corresponding to the 200 planes of β-FeOOH (Fig. S6).

Fig. 1: Structural characterization of the Ag-FeOOH catalysts.
Fig. 1: Structural characterization of the Ag-FeOOH catalysts.The alternative text for this image may have been generated using AI.
Full size image

a TEM image and b HRTEM image of Ag-FeOOH-3. c TEM image of Ag1-FeOOH-2. d Elemental mapping and e HAADF-STEM image of Ag1-FeOOH-2; the yellow circles correspond to the single-Ag-atom. f XRD patterns of FeOOH and Ag-FeOOH with different Ag loadings. g Normalized Ag K-edge XANES and h EXAFS spectra for Ag1-FeOOH-2, Ag2O, and Ag foil. i Room temperature 57Fe Mössbauer spectrum of Ag1-FeOOH-2. Source data was provided as a Source data file.

Energy-dispersive X-ray spectroscopy (EDS) elemental mappings showed a uniform distribution of Ag, Fe, and O elements across the entire sample (Fig. 1d), and the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) (Fig. 1e) image of Ag1-FeOOH-2 clearly showed atomically dispersed Ag, suggesting that the Ag species were likely atomically dispersed on the FeOOH nanorod surface. With a further reduced Ag loading amount, the Ag1-FeOOH-1 sample retained the identical morphology as Ag1-FeOOH-2 (Fig. S7). X-ray diffraction (XRD) patterns provided complementary structural information (Fig. 1f). All samples exhibited dominant diffraction peaks corresponding to β-FeOOH. Notably, different from Ag1-FeOOH-1 and Ag1-FeOOH-2, the Ag-FeOOH-2.5 (Fig. S8) and Ag-FeOOH-3 samples displayed additional diffraction peaks at 38.1°, 44.3°, and 64.4°, which could be assigned to the 111, 200, and 220 planes of face-centered cubic Ag (PDF #04-0783), corroborating TEM observations of Ag nanoparticles in Ag-FeOOH-2.5 and Ag-FeOOH-3.

Next, the electronic structures of the as-prepared samples were investigated by X-ray photoelectron spectroscopy (XPS). The Ag 3d XPS spectra of Ag1-FeOOH-1 and Ag1-FeOOH-2 exhibited similar features (Fig. S9a), with the Ag 3d3/2 and 3d5/2 peaks located at approximately 374.4 eV and 368.4 eV, respectively. This indicates that the Ag species in both materials are in an oxidized state. In contrast, for Ag-FeOOH-3, the binding energies of the Ag 3d3/2 and 3d5/2 peaks shifted to higher values, suggesting the presence of Ag clusters, which is consistent with the HRTEM results38,39,40,41,42. The Fe 2p XPS spectra displayed similar features for all Ag-FeOOH samples (Fig. S9b), with Fe 2p3/2 and 2p1/2 peaks at 711.2 eV and 724.9 eV, respectively, accompanied by the satellite peaks at 719.1 eV and 733.4 eV, indicating that Fe mainly existed as Fe(Ⅲ)43,44. The O 1s XPS spectra could be deconvoluted into lattice oxygen in Fe-O-Fe, hydroxyl group in Fe-O-H, and adsorbed H2O (Fig. S9c)44,45. To further study the chemical state and coordination environment of Ag in Ag1-FeOOH-2, X-ray absorption spectroscopy (XAS) measurements were conducted. As shown in the Ag K-edge X-ray absorption near-edge structure (XANES) spectrum (Fig. 1g), the white line of Ag in Ag1-FeOOH-2 aligned closely with that in Ag2O, and both were located on the right side of that in Ag foil, indicating that Ag species in Ag1-FeOOH-2 were mainly present as Ag(Ⅰ), consistent with the XPS results. The corresponding X-ray absorption fine structure (EXAFS) spectrum of Ag1-FeOOH-2 also displayed a prominent peak, which could be assigned to the Ag-O bonds (Figs. 1h, S10 and Table S2). The detailed coordination environment of the single-Ag-atom site was further confirmed by EXAFS fitting. The obtained parameters, listed in Table S2, further confirm the coordination environment of single-Ag-atom sites in Ag1-FeOOH-2.

To study the influence of Ag loading on the coordination environment and structural characteristics of Fe in FeOOH57, Fe Mössbauer spectroscopy measurements were conducted. Two distinct Fe(Ⅲ) components were identified via fitting, with quadrupole splitting (QS) values of around 1.0 mm s−1 (Fe(Ⅲ)-a) and 0.6 mm s−1 (Fe(Ⅲ)-b), respectively. The larger QS value indicates the poorer symmetry of Fe (Figs. 1i and S11). The fitted parameters for the Mössbauer spectra of FeOOH and Ag-FeOOH with different Ag loadings are listed in Table S3. Along with increasing Ag loading amount, the relative content of Fe(Ⅲ) increased from 57.5% for FeOOH to 65.4% for Ag1-FeOOH-2, indicating that Ag decoration induced a local structural distortion at the adjacent Fe sites. Notably, the Ag-FeOOH-3 sample exhibited a reduced Fe(Ⅲ) content (62.4%), indicating weakened interfacial interaction of Ag and FeOOH once Ag species aggregated together to form Ag nanoparticles. Furthermore, electron spin resonance (ESR) spectroscopy was employed to probe the presence of oxygen vacancies in Ag1-FeOOH (Fig. S12). The spectrum of Ag1-FeOOH-2 reveals a distinct signal at g = 2.003, which is characteristic of oxygen vacancies, confirming their existence in the material.

Electrochemical propylene epoxidation performance

The catalytic performance of the as-prepared samples for electrochemical propylene epoxidation was assessed in a standard H-type electrochemical cell in a three-electrode configuration. The reaction system employed a 0.1 M phosphate buffer solution (PBS) as the electrolyte. Following 1-hour constant potential electrolysis, the liquid-phase products were quantitatively determined by proton nuclear magnetic resonance (1H NMR) spectroscopy (Fig. S13–S15). Additionally, gas-phase products were collected throughout the reaction and characterized by GC-MS for qualitative analysis (Fig. S16).

Firstly, in order to screen the metal-dependent catalytic activities, a series of M-FeOOH (M = Ag, Ir, Ru, and Pd) samples with identical metal loading were synthesized via the same procedure except using different metal precursors. Their electrochemical propylene oxidation performances were comparatively studied across varying applied potentials (Fig. 2a). Notably, Ir-FeOOH and Ru-FeOOH exhibited negligible PO Faradaic efficiencies (FEPO < 4.0%) across all tested potentials in comparison with FeOOH. Pd-FeOOH slightly increased the maximum FEPO to 5.8% at 2.4 V vs. RHE. In contrast, Ag1-FeOOH-1 markedly enhanced the electrochemical propylene epoxidation activity, reaching a maximum FEPO of 21.1% at 2.4 V vs. RHE, highlighting a super catalytic improvement of Ag for electrochemical propylene epoxidation.

Fig. 2: Electrochemical propylene oxidation performance (Electrolyte: 0.1 M PBS, pH: 7.0, solution resistance: 4.5 Ω, electrode surface area: 1.0 cm2).
Fig. 2: Electrochemical propylene oxidation performance (Electrolyte: 0.1 M PBS, pH: 7.0, solution resistance: 4.5 Ω, electrode surface area: 1.0 cm2).The alternative text for this image may have been generated using AI.
Full size image

a FEPO on M-FeOOH (M = Ag, Ir, Ru, or Pd). b FEPO on Ag1-FeOOH-X (X = 1 or 2) and Ag-FeOOH-3. c jPO on FeOOH, Ag1-FeOOH-X (X = 1 or 2) and Ag-FeOOH-3, d FE for acetone (yellow), PO (red) and acetic acid (blue) on different catalysts at 2.4 V vs. RHE. All error bars represent the standard deviation of three independent measurements. All voltage was recorded without iR correction. Source data was provided as a Source Data file.

Then, we compared the electrochemical propylene oxidation performance of various Ag-FeOOH samples with different Ag loading amounts (Figs. 2b and S17). Among which, Ag1-FeOOH-2 demonstrated the optimal performance, reaching a highest FEPO of 32.0% at 2.4 V vs. RHE. In addition, the variety of gas products produced over Ag1-FeOOH-2 at 2.4 V vs. RHE was quantified by GC-MS, and the results revealed two main peaks in the chromatogram (Fig. S16a): Peak 1 (0.065–0.750 min) and Peak 2 (1.460–2.140 min). The results from mass spectra indicate that Peak 1 is attributed to C3H6, N2, CO, O2, and CO2, whereas Peak 2 includes formaldehyde and propylene glycol (Fig. S16b). The detected gas-phase propylene electrooxidation products are thus CO, CO2, formaldehyde, and propylene glycol, together with O2 generated from competing OER. Additionally, other liquid products such as acetone and acetic acid were observed (Fig. S18–20).

To assess the intrinsic catalytic activity, the electrochemically active surface areas (ECSAs) of the prepared samples were characterized (Fig. S21, S22 and Table S4). The PO partial current density (jPO) increased with increasing Ag loading amount on Ag-FeOOH (Fig. 2c), while the formation of Ag nanoparticles on FeOOH greatly promoted competitive oxygen evolution reaction (OER) and thus significantly reduced the FEPO over Ag-FeOOH-3 (Fig. S23).

To further investigate the effect of FeOOH on the electrochemical propylene oxidation reaction, FeOOH was replaced by carbon (C) or NiOOH substrate decorated with the equivalent amount of Ag (Fig. 2d). Ag1-FeOOH-2 exhibited higher FEPO (32.0%) relative to Ag-C (10.0%) and Ag-NiOOH (20.0%) at 2.4 V vs. RHE, underscoring the role of FeOOH in enhancing the electrochemical propylene epoxidation selectivity. The stability of the Ag1-FeOOH-2 catalyst was evaluated at a fixed potential of −2.4 V vs. RHE for 6 hours (Fig. S24). Throughout the test, the FE of propylene oxide remained relatively high, exceeding 30%, while the current density showed only a slight decrease. Post-reaction characterizations confirmed the structural stability of the catalyst. HAADF-STEM imaging confirmed the atomic dispersion of Ag sites after the stability test, demonstrating the structural stability of Ag1-FeOOH-2 under reaction conditions (Fig. S25a). HRTEM further confirmed that the facets of FeOOH were kept after the stability test (Fig. S25b). The catalytic performance of the Ag1-FeOOH-2 catalyst is well-positioned among the reported candidates, deserving a further exploration of the underlying catalytic mechanism for propylene epoxidation (Table S5).

In-situ identification of catalytic sites

To probe the key intermediates and reaction pathways of electrochemical propylene epoxidation to PO, the in-situ ATR-SEIRAS measurements were conducted over FeOOH and Ag1-FeOOH-2 catalysts in 1.0 M PBS electrolyte under He or propylene atmosphere. Figure 3a revealed that in the He-saturated electrolyte over FeOOH, a positive peak at 1270 cm−1 emerged progressively with increasing anodic potential, resulting from the bending vibration of adsorbed *OOH, a typical OER intermediate46,47, verifying OER activity of FeOOH in an inert environment. When switching to a propylene-saturated environment (Figs. 3b and S26a), distinct adsorption features appeared at 2823–3090 cm−1, 1449 cm−1, and 1646 cm−1 at open circuit voltage (OCV), indicating propylene adsorption on FeOOH. These peaks initially increased in intensity with increasing anodic potential, followed by transitioning to inverted peaks at anodic potentials ≥ 1.1 V vs. RHE, indicating the potential-dependent adsorption and consumption of propylene. Notably, a developing peak at 1428 cm−1 emerged at higher anodic potentials, which could be attributed to the C = C vibration in *O-*C3H648, confirming the progression of propylene oxidation.

Fig. 3: In-situ ATR-SEIRAS spectra recorded in a He- or propylene-saturated 1.0 M PBS (pH: 7.0, solution resistance: 4.5 Ω).
Fig. 3: In-situ ATR-SEIRAS spectra recorded in a He- or propylene-saturated 1.0 M PBS (pH: 7.0, solution resistance: 4.5 Ω).The alternative text for this image may have been generated using AI.
Full size image

Spectra on (a, b) FeOOH and (c, d) Ag1-FeOOH-2 over an anodic potential range of 0.7 to 2.7 V vs. RHE. The insets in b, d show the adsorption structures (Fe: yellow, Ag: light blue, O: red, C: brown, and H: white). All voltage was recorded without iR correction. Source data was provided as a Source data file.

Compared to FeOOH, the *OOH formation over Ag1-FeOOH-2 in He atmosphere occurred at a much lower anodic potential (Fig. 3c), indicating greatly enhanced oxygenated intermediates adsorption after decorating Ag single atoms on FeOOH. Under the condition of saturated propylene (Figs. 3d; S26b and Table S6), while maintaining a similar propylene adsorption trend, the *OOH exhibited a delayed emergence, having reduced intensity over Ag1-FeOOH-2 (as compared to FeOOH) along with the increase in anodic potential, suggesting the quick consumption of oxygenated species due to electrochemical propylene oxidation. Crucially, the appearance of a peak at 1620 cm−1 corresponding to PO indicated a completed propylene epoxidation via *O-*C3H6 intermediate on Ag1-FeOOH-249,50. These in-situ ATR-SEIRAS results identified FeOOH as the propylene adsorption sites and Ag single atoms as active sites for generating oxygenated species, providing mechanistic insights into the electrochemical propylene epoxidation process. Besides using FeOOH as the support, the adsorption of propylene on other Fe-based supports, such as Fe2O3 and metallic Fe, was also investigated (Fig. S27). The infrared vibration peaks of propylene (near 2800–3200 cm−1) were observed (Figs. S28 and S29), among which, the FeOOH surface showed the highest intensity. These results indicate that propylene can be adsorbed on different Fe-based supports, and FeOOH shows a higher propylene adsorption capacity compared with Fe2O3 and metallic Fe.

To further clarify the phase changes of the materials and the adsorbed intermediate species during the reaction, in situ Raman spectroscopy measurements were supplemented. As shown in Fig. S30a, the support exhibited three obvious peaks at 312, 409, and 730 cm−1, which corresponded to the β-phase FeOOH. Under the Ar atmosphere, the peaks did not show significant changes, indicating that the phase of the support was stable in an inert atmosphere. As shown in Fig. S31b, under a propylene atmosphere, an obvious peak appeared at 550 cm−1, and an obvious broad peak appeared between 1300 and 1500 cm−1. The peak height became more obvious with increasing applied potential. These two peaks could be attributed to the vibration peaks of Fe-C and C = C bonds on the material surface, respectively. This indicated that FeOOH had an adsorption effect on propylene, and the adsorption occurred at the Fe sites. To further probe the changes in the coordination environments of Ag and Fe during the reaction, quasi in-situ XPS measurements were conducted on Ag1-FeOOH-2 (Fig. S31). These measurements revealed that the single-Ag-atom could serve as the generation site for oxidized species, exhibited a higher valence state during the reaction, in contrast, Fe remained relatively stable with no significant changes (Fig. S32).

To further quantitatively monitor the evolution of Fe and validate propylene adsorption at Fe sites during the electrochemical propylene oxidation process, in-situ 57Fe Mössbauer spectroscopy measurement was performed on Ag1-FeOOH-2 in propylene-saturated 0.1 M PBS (Fig. S33). Figure 4a–e present the potential-dependent ⁵⁷Fe Mössbauer spectra with the corresponding current-time profiles shown in Fig. 4f. Spectral fitting revealed two Fe(Ⅲ) states (Fe(Ⅲ)-a and Fe(Ⅲ)-b), similar to the ex-situ measurements. Quantitative analysis (Fig. 4g and Table S7) showed that the relative content of Fe(Ⅲ) a increased from 59.6% at OCV to the maximum 72.2% at 2.4 V vs. RHE, and then decreased to 67.5% at 2.8 V vs. RHE. Similar to what had been discussed in the ex-situ ⁵⁷Fe Mössbauer spectra, the QS values for Fe(Ⅲ)-a and Fe(Ⅲ)-b were also approximately 1.0 mm s−1 and 0.6 mm s−1, respectively. The results from in-situ 57Fe Mössbauer measurements indicated that the structure and oxidation state of FeOOH were well maintained. Furthermore, the relative content of Fe(Ⅲ)-a was observed to increase with the relative content of Fe(Ⅲ)-b decreased, indicating that a poorer symmetry of electronic structure around Fe could be induced by adsorption of reaction intermediates on the Fe(Ⅲ)-a sites under reaction conditions. AFT 57Fe Mössbauer spectrum demonstrated an approximate restoration of Fe(Ⅲ)-a to the initial level after electrochemical oxidation reaction, suggesting reversible adsorption and desorption of the adsorbed reactant and reaction intermediates over Ag1-FeOOH-2. These in-situ ⁵⁷Fe Mössbauer findings further corroborated the in-situ ATR-SEIRAS observations, verifying FeOOH as the propylene adsorption sites. Figure 4h schematically illustrated how single-Ag-atom decoration modified the Fe coordination in FeOOH (the incorporated Ag single atoms bonded to the adjacent Fe sites and converted them into Fe(Ⅲ)-a). Figure 4i showed the effect of propylene adsorption on FeOOH (Fe atoms formed bonds with propylene molecules, thereby increasing the relative content of Fe(Ⅲ)-a).

Fig. 4: In-situ 57Fe Mössbauer spectroscopy.
Fig. 4: In-situ 57Fe Mössbauer spectroscopy.The alternative text for this image may have been generated using AI.
Full size image

In-situ 57Fe Mössbauer spectra of Ag1-FeOOH-2 recorded in propylene-saturated 0.1 M PBS (pH: 7.0, solution resistance: 4.5 Ω) at (a) OCV, b 2.0 V vs. RHE, c 2.4 V vs. RHE, d 2.8 V vs. RHE, and e after the test (AFT). f Current-time curve of Ag1-FeOOH-2 recorded during in-situ ⁵⁷Fe Mössbauer measurements. g Relative signal intensities of different Fe species obtained from the fitting results of in-situ ⁵⁷Fe Mössbauer spectra. h The schematic diagram of Fe (Ⅲ)-a (green and blue) and Fe(Ⅲ)-b (yellow) in Ag1-FeOOH. i The schematic diagram of Fe(Ⅲ)-a (pink and blue) and Fe(Ⅲ)-b (yellow) in FeOOH after propylene adsorption. All voltage was recorded without iR correction. Source data was provided as a Source data file.

Theoretical insights into the catalytic mechanism

Density functional theory (DFT) calculations were performed to further investigate the propylene oxidation mechanism over Ag1-FeOOH. Based on the nanorod morphology observed and the dominant 200 lattice, the thermodynamically stable and low-energy β-FeOOH 200 surface was selected to model the catalyst support. The FeOOH 200 surface was constructed to simulate the FeOOH support (Fig. S34a, b). In addition, a surface Fe atom was replaced with an Ag atom to model Ag1-FeOOH (Fig. S34c, d). Considering that the potential oxidation of Ag to Ag2O under the reaction condition, additionally, the Ag2O (Fig. S34e, f) model was built to simulate Ag nanoparticles and Ag clusters on FeOOH. Furthermore, the ab initio molecular dynamics (AIMD) simulation shows a stable total energy and preserved atomic configuration (Fig. S35), demonstrating the stability of the optimized structure of Ag1-FeOOH.

The electronic structures of the Fe and Ag sites were firstly analyzed to evaluate their catalytic performance. For the surface Fe site on Ag1-FeOOH (Fig. 5a), the spin-up and spin-down electrons exhibited an asymmetric distribution. The d-band center is located at −3.72 eV for spin-down electrons and −0.18 eV for spin-up electrons. In contrast, the spin-up and spin-down electrons of the single-Ag-atom display symmetric distributions and are situated farther from the Fermi level, with d-band centers at −2.56 eV and −2.53 eV, respectively. According to the d-band center theory, the single-Ag-atom site would exhibit weaker adsorption properties toward reaction intermediates such as C3H6 compared to the Fe site. Furthermore, relative to pure FeOOH, which has d-band centers of −3.76 eV (spin-up) and −0.27 eV (spin-down), Ag doping may weaken the adsorption of intermediates on the Fe site.

Fig. 5: Theoretical calculations.
Fig. 5: Theoretical calculations.The alternative text for this image may have been generated using AI.
Full size image

a Density of states plots of Ag1-FeOOH and FeOOH. Charge density differences of propylene adsorption on (b) FeOOH and c Ag1-FeOOH. d Scheme of the reaction pathways of propylene epoxidation on Ag1-FeOOH. (Fe: yellow, Ag: light blue, O: red, C: brown, H: white). e Free energy diagram of propylene epoxidation on Ag1-FeOOH. O*-path (red) and OH*-path (yellow). f Free energy diagram of OER on Ag1-FeOOH (red), FeOOH (yellow), and Ag2O (black).

From the in-situ ATR-SEIRAS results, propylene could be initially adsorbed on the Fe site of the FeOOH support. To identify the preferred adsorption site, the adsorption energies of propylene on FeOOH and the single-Ag-atom site of Ag1-FeOOH were calculated. As shown in Fig. 5b, the adsorption energy of propylene on the Fe site of Ag1-FeOOH is −0.36 eV, stronger than that on the single-Ag-atom site (−0.30 eV, Fig. 5c). This suggests that the Fe site is more favorable for propylene adsorption, particularly at high anodic potential, where electron-deficient catalytic sites tend to exhibit stronger adsorption. Additionally, the adsorption energy of propylene on the Fe site of pure FeOOH was calculated to be -0.45 eV, which is stronger than that on the Ag site of Ag1-FeOOH, further supporting the preference for C3H6 adsorption on Fe sites.

Furthermore, to explore the role of oxygen vacancies in Ag1-FeOOH, we constructed an Ag1-FeOOH model featuring a surface oxygen vacancy near the single-Ag-atom site (named as Ag1-FeOOH-V, Fig. S36) and computed the H2O adsorption energy to probe the Ag site’s oxidation activity. The calculation results show that the Ovac significantly enhances H2O adsorption, strengthening the adsorption energy from −0.28 eV to −1.15 eV and shortening the Ag-O bond from 2.44 Å to 2.43 Å. These findings indicate that oxygen vacancies are beneficial for H2O activation. By promoting Ag-O* formation and oxygen activation, the Ag-induced Ovac in Ag1-FeOOH-2 created an electron-deficient surface at the expense of adsorbing electron-rich propylene, aligning well with our proposed reaction mechanism.

The free energy profile for the propylene epoxidation reaction was then calculated on the basis of the proposed catalytic mechanism (Fig. 5d). This mechanism is well consistent with the in-situ ATR-SEIRAS results, which indicate that propylene is firstly adsorbed on the active site, followed by the formation of an oxidized intermediate, ultimately leading to the epoxidation reaction. Both *O and *OH involved epoxidation reaction pathways were considered. Initially, propylene was adsorbed on the Fe site (deduced from in-situ ATR-SEIRAS measurements) and subsequently a *OH was formed on the adjacent single-Ag-atom site. The *OH could react with propylene to form a C3H6OH* or transfer to O*. Then, *O coupled with adsorbed C3H6* on the Fe site to form C3H6O. It should be noted that *O could also be further oxidized to *OOH, leading to the formation of O2 as a byproduct. From the free energy profile of Ag1-FeOOH (Figs. 5e and S37, S38), it is evident that the C3H6 was firstly adsorbed on the Fe site (-0.36 eV on the Fe site). Subsequently, OH* was formed on the single-Ag-atom site with an energy barrier of 1.32 eV. The next step of OH* coupling with C3H6* displayed an energy barrier of 0.82 eV, higher than OH* oxidation to O* (0.74 eV). Note that the energy barrier of the O* path could be overcome by applying anodic potential. For example, at 1.23 V (vs. RHE), the energy barrier of OH* formation is only 0.09 eV, while the OH* epoxidation path is still limited by the OH*-C3H6* coupling step (0.82 eV). In other words, the propylene epoxidation reaction follows the O*-C3H6* coupling mechanism. Compared with Ag2O (Fig. S39, S40), with a high energy barrier for OH* generation, and FeOOH (Fig. S41, S42) with a high energy barrier in C3H6*-O* coupling, Ag1-FeOOH shows a lower free energy barrier for propylene epoxidation (Fig. S43).

Additionally, the free energy profile for the side OER over Ag1-FeOOH and FeOOH was also calculated (Figs. 5f and S37–S42). The rate-limiting step for OER is the conversion of *O to *OOH. Compared to FeOOH (0.91 eV), the energy barrier for *O to *OOH conversion on Ag1-FeOOH is obviously higher (2.08 eV), indicating that the introduction of the single-Ag-atom promotes propylene epoxidation, while not facilitating OER on the single-Ag-atom site. Note that the OER performance of the adjacent Fe site may be boosted because of the interaction between Ag and Fe sites from the results of experimental results (Fig. S23). During the propylene epoxidation process, the Fe site is occupied by C3H6. The side product O2 is thus also inhibited. Besides, the OER energy barrier of Ag2O is 0.53 eV, in line with our experiment results that Ag-FeOOH-3 shows a higher current density of OER and a lower FE of PO.

Furthermore, the propylene epoxidation free energy profiles of Ag2-FeOOH (Fig. S44a, b) and Ag3-FeOOH (Fig. S44c, d) were also considered. The O* intermediate involved in the epoxidation process was considered for all three catalysts (Fig. S45, S46). In the case of Ag2-FeOOH, the rate-limiting step is the second reaction step, namely the formation of the OH* intermediate, while for Ag3-FeOOH, the rate-limiting step is the coupling between C3H6* and O*. Both Ag2-FeOOH and Ag3-FeOOH exhibit distinct catalytic characteristics compared to Ag1-FeOOH, demonstrating that the local coordination environment of the active site is directly correlated with catalytic performance (Fig. S47).

Discussion

In summary, in-situ techniques were employed to elucidate the reaction mechanism of electrochemical propylene oxidation to PO over the single-Ag-atom decorated FeOOH catalyst. Results from in-situ ATR-SEIRAS and in-situ 57Fe Mössbauer spectroscopy measurements identified a dual-site synergistic catalytic mechanism, where single-Ag-atom sites activated water to generate reactive oxygen species, while the adjacent Fe sites served as adsorption sites to activate propylene. DFT calculations further demonstrated that propylene could be preferentially adsorbed on the asymmetrically coordinated Fe sites, and incorporation of Ag single atoms on FeOOH facilitated the generation of *O intermediates, which significantly reduced the energy barrier of *O and *C3H6 coupling and facilitated the formation of PO. This research elucidates the dual-site synergistic catalytic mechanism governing electrochemical propylene epoxidation and establishes a framework for rationally designing single-atom catalysts for electrochemical organic synthesis.

Methods

Chemicals and materials

Iron (Ⅲ) nitrate nonahydrate (Fe(NO3)3·9H2O; Analytical Reagent (AR), CAS No.7782-61-8, Aladdin), ammonium fluoride (NH4F, 98%, CAS No.12125-01-8, Aladdin), urea (CH4N2O, 99.5%, CAS No.57-13-6, Aladdin), silver nitrate (AgNO3, AR, CAS No.7761-88-8, Sinopharm Chemical Reagent Co., Ltd (China)), propylene oxide (PO) (C3H6O, 99.5% (GC), CAS No.75-56-9, Aladdin), 1,2-propanediol (C3H6O2, AR, CAS No. 57-55-6, GHTECH), acetone (C3H6O, AR, CAS No.67-641, KESHI), acetic acid (C2H4O, 99.5%, Cas No.64-19-7, Aladdin), Nafion solution (5 wt.%, Dupont), dimethyl sulfoxide (C2H6OS, AR, Cas No.67-68-5, Sinopharm Chemical Reagent Co., Ltd (China)), deuterium oxide (D2O, 99.9% (D), Cas No.7789-20-0, CIL), potassium phosphate dibasic (K2HPO4, 99%, Cas No.7758-11-4, Aladdin), potassium phosphate monobasic (KH2PO4, 99.5%, Cas No.7778-77-0, Aladdin). Deionized water (18.2 mΩ) was used in all experiments.

Preparation of β-FeOOH

The β-FeOOH was prepared by a modified method that was reported in prior literature51. It was in 16 mL of deionized (DI) water that 0.88 mmol Fe(NO3)3·9H2O, 1.6 mmol NH4F, and 4 mmol urea were dissolved under vigorous stirring for 30 min in a typical experiment. Then the above solution, which was transferred into a 50 mL Teflon-lined autoclave, was maintained at 120 °C for 6 h. The final product, after the system had cooled down to room temperature, was washed with DI water for 3 times and ethanol for 3 times. Then the final product was dried in a vacuum overnight for further synthesis and characterization.

Synthesis of Ag-decorated FeOOH

First, disperse 100 mg FeOOH in 20 mL of a solution containing 0.01 M NaOH and apply ultrasonic vibration. Next, dropwise add 4 mL of an AgNO3 solution (0.005 mmol mL−1) while stirring at room temperature52. After continuous stirring for 12 h, centrifuge the suspension and then wash it three times with DI water and ethanol. Subsequently, dry the obtained solid in a vacuum oven and calcine it in an argon (Ar) atmosphere at 200 °C for 2 h to obtain Ag1-FeOOH-1.

The same method was used for the synthesis of Ag1-FeOOH-2, except that the concentration of AgNO3 solution was 0.03 mmol mL−1.

The same method was used for the synthesis of Ag-FeOOH-3, except that the concentration of AgNO3 solution was 0.2 mmol mL−1.

Preparation of catalyst ink and electrodes

The catalyst ink was prepared by dissolving 5 mg of the catalyst in 1 mL of a mixed solution composed of Nafion solution, DI water, and isopropyl alcohol (a volume ratio of 4:48:48). Subsequently, the solution was subjected to sonication for a sufficient duration to ensure a uniform dispersion, forming a homogeneous catalyst ink. The 200 μL well-dispersed catalyst ink was uniformly coated onto a 1 × 1 cm2 carbon paper substrate. This coated carbon paper served as the electrode for the subsequent electrolysis experiments.

Electrochemical measurements

The electrooxidation of propylene was performed at ambient temperature (~298 K) in an H-type cell. A standard three-electrode configuration, controlled by a CHI660e electrochemical workstation. The electrolyte selected for these experiments was an aqueous 0.1 mol L−1 phosphate buffer solution (PBS, pH ~7.0). The cathode and anode are separated by Nafion 117 (183 μm, Dupont). Prior to use, the Nafion 117 membrane was pretreated by immersing it in a 5 wt.% H2O2 solution at 80 °C for 1 h, rinsing with deionized water, then treating it with a 0.5 mol L−1 HNO3 solution at 80 °C for 0.5 h, and finally rinsing again with deionized water. For the electrode configuration, the prepared electrode served as the working electrode, a saturated calomel electrode (SCE) functioned as the reference electrode, and a platinum plate with dimensions of 1 × 1 cm2 was utilized as the counter electrode. All measurements were performed without iR correction, given a solution resistance of ~4.5 Ω. The reference electrode was calibrated by measuring the HOR and HER on Pt electrodes in H2-saturated 0.1 M HClO4 and 0.1 M KOH. The CV curves were scanned at 10 mV s−1 until stable, and the RHE potential was determined from the open-circuit potential in each electrolyte. The RHE conversion was based on the Nernst equation, which is presented as follows:

$${{\rm{E}}}\, ({{\rm{vs}}}.{{\rm{RHE}}})={{\rm{E}}}\, ({{\rm{vs}}}.{{\rm{SCE}}})+0.2415{{\rm{V}}}+0.0591\times {{\rm{pH}}}$$
(1)

A gas mass flow controller at the anode inlet was used to maintain a constant flow at 20 mL min−1 of the gas mixture, which consisted of 99.5% propylene balanced with argon. To ensure complete saturation of the electrolyte with propylene, the system was continuously purged with the gas mixture prior to electrochemical testing. The quantification of liquid-phase products was accomplished through proton nuclear magnetic resonance (1H-NMR) spectroscopy using a standard curve.

The Faradaic efficiency of PO (FEPO) at any potential was calculated as follows:

$${{\rm{FE}}}=c\times V\times N\times F/Q$$
(2)

where c represents the concentration of PO, V is the volume of the electrolyte, N is the number of electrons transferred per product molecule (for PO, N = 2), F represents the Faraday constant, and Q is the charge quantity calculated from the i–t curve.

A volume of 10 μL of the catalyst ink, formulated with the identical composition as previously described, was dropped onto an L-shaped glassy carbon electrode to form a well-defined catalyst layer. Subsequently, under an argon atmosphere, cyclic voltammetry (CV) measurements were conducted on different samples at a series of scan rates (10, 20, 40, 60, 80, and 100 mV s−1) to determine the double-layer capacitance (Cdl). To obtain the Cdl, a linear fitting approach was employed by plotting Δj (ja–jc) against the scan rate. Herein, within the defined scan range, ja and jc denote the anodic and cathodic current densities, respectively. It is worth noting that the slope of the fitted straight line corresponds to twice the value of Cdl. The electrochemically active surface area (ECSA) was calculated using the following equation:

$${{\rm{ECSAs}}}={R}_{f}\times S$$
(3)

where Rf represents the roughness factor of the FeOOH surface, and S is the surface area of the carbon paper electrode, which is 1 cm2 in this experimental setup. The Rf was calculated according to the relationship:

$${R}_{f}={C}_{{dl}}/40$$
(4)

which is based on the Cdl value of the smooth oxide surface (specifically, 40 μF cm−2 for the FeOOH surface53).

Characterization

Transmission electron microscopy (TEM), energy-dispersive X-ray spectroscopy (EDS) mapping, and high-resolution TEM (HRTEM) were performed using a JEOL JEMF200 STEM/TEM microscope (Japan). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) characterization was performed on a probe-corrected JEOL JEM-ARM200F STEM/TEM microscope. X-ray diffraction (XRD) measurements were carried out on a PAN Analytical X’pert Pro diffractometer with Cu Kα radiation (λ = 1.5418 Å). The GC-combustion-MS analysis (GC-MS) was performed on a Shimadzu GC-MS QP-2020 NX instrument. Fe Mössbauer spectra were acquired at room temperature in constant acceleration mode with a WSS-10 spectrometer controlled by Wissoft 2003 software. A 57Co (Rh) source was used for irradiation. Spectral analysis was performed using the MossWinn 4.0 software, with all spectra fitted to Lorentzian profiles to derive the relevant parameters. The minimum experimental line width obtained was 0.27 mm s−1. Isomer shifts are reported relative to that of α-iron at 298 K. Electron spin resonance (ESR) spectra were acquired on a Bruker A200 spectrometer. Inductively coupled plasma optical emission spectrometry (ICP-OES) was employed to quantify the silver concentration in Ag-FeOOH. X-ray photoelectron spectroscopy (XPS) measurements were carried out on a Thermo Fisher Scientific ESCALAB 250 spectrometer (USA), which was furnished with an Al Kα X-ray source. The binding energy values were calibrated by referencing the C 1s peak set at 284.8 eV. X-ray absorption fine structure (XAFS) spectra at the Ag K-edge were measured at Shanghai Synchrotron Radiation Facility (SSRF).

In-situ ATR-SEIRAS measurements

The in-situ attenuated total reflectance surface enhanced infrared absorption spectroscopy (ATR-FTIR) was carried out to analyze the reaction process and identify intermediates. The FTIR instrument was Nicolet iS20 with a MCT detector cooled by liquid nitrogen and a PIKE VeeMax III variable-angle ATR sampling accessory. The working electrode was constructed by coating catalyst ink (0.5 mg cm−2) on a nano-Au covered monocrystalline silicon, the saturated calomel electrode was used as reference. The relevant data are available from the corresponding electrode; the graphite rod electrode was counter electrode. The electrolyte was 0.1 M PBS solution, identical to that used in constant current test. Before the test, Ar or propylene was continuously introduced into the electrolyte for 30 min. The test was performed from OCV to reaction potential range (from 0.7 to 2.7 V vs. RHE with an interval of 0.1 V vs. RHE). At each potential, the voltage was applied for 60 s, and the spectrum was subsequently acquired for 30 s.

In-situ 57Fe Mössbauer measurements

In-situ 57Fe Mössbauer spectroscopy measurements were carried out in a propylene-saturated 0.1 M PBS electrolyte. In a specially designed H-type electrolytic cell, two pieces of carbon paper with a catalyst loading of 10 mg cm−2 were used, and the cross-sectional area irradiated by the γ-ray source was 1 cm2. The reference electrode and counter electrode used were of the same type as those in the electrochemical tests. Spectra were collected at the open-circuit voltage (OCV), as well as at applied potentials of 2.0 V, 2.4 V, and 2.8 V vs. RHE. After the reaction, spectra were also collected after the potential was removed (AFT).

Density functional theory calculations

The density functional theory (DFT) calculations were performed in the Vienna ab initio Simulation Package (VASP) code, version 6.154,55. The projector augmented wave (PAW) method was used to describe the interactions between ion cores and valence electrons56,57. The generalized gradient approximation (GGA) with Perdew-Burke-Ernzerhof (PBE) functional was used to calculate the exchanged correlation interactions58. The valence electronic states were expanded in the plane wave basis sets. The cutoff energy was set as 450 eV. The FeOOH (200) slab was built by a cleavage cell of 100 surface with a supercell of 2 * 2. The vacuum thickness was larger than 1.5 nm. The Ag1-FeOOH was built by replacing a surface Fe atom with Ag. The geometric optimizations converged when the energy difference was smaller than 10−6 eV, and the forces were less than 0.03 eV Å−1. The Brillouin zone of the calculations was sampled with a 2 × 2 × 1 Monkhorst-Pack mesh. During the structure optimization, only the surface two layers are relaxed. The free energy of adsorbed intermediates and the propylene oxidation free energy profile were calculated as:

$$G={E}_{{dft}}+{{\rm{ZPE}}}+{H}_{{vi}}b-T{S}_{{vib}}$$
(5)

In the Eq. (5), Edft is the DFT-calculated total energy (E0), ZPE is the zero-point energy obtained from frequency calculation, T is the temperature (298 K), and Hvib and Svib are the enthalpy and entropy parts of non-imaginary vibrations derived from the harmonic approximation to the potential at the temperature T. The VASPKIT code was used for postprocessing of the VASP calculated data, including free energy correction and charge density difference analysis59

The structural model involved in DFT is in Supplementary Data 1

The spin-polarized Ab initio Molecular Dynamics simulation on Ag1-FeOOH was performed in the NVT ensemble at 300 K using the Nosé-Hoover thermostat, with a time step of 1 fs for a total simulation duration of 20 ps.