Sunlight‐Driven Concurrent Synthesis of Jet Fuel Precursors and Hydrogen From Biomass‐Derived Furfuryl Alcohol Over Oxygen‐Doped Zn₂In₂S₅

ABSTRACT

Photoreforming of biomass‐derived furfuryl alcohol (FA) offers a green and economic route for the concurrent production of jet fuel precursor hydrofuroin (HF) and hydrogen. However, conventional FA photoreforming is generally hampered by low quantum yield and poor HF selectivity. Herein, an oxygen‐doping strategy to modify the Zn₂In₂S₅ (ZIS) photocatalyst is reported, which markedly enhances both quantum yield and HF selectivity. Oxygen dopants fulfill multiple critical roles: engineering shallow donor levels to facilitate photocarrier separation and extend their lifetime, accelerating the conversion of the furfural (FF) byproduct, and strengthening FA adsorption. Notably, the O‐doped Zn₂In₂S₅ (ZIS‐O) achieves 95% HF selectivity under simulated sunlight alongside high apparent quantum yields (AQY) for HF: 37.6% at 380 ± 20 nm and 12.4% at 420 ± 20 nm. Pilot‐scale tests under natural sunlight reveal a long‐term average HF selectivity of 90% with a turnover number (TON) of 1390. Theoretical investigations confirm that oxygen dopants thermodynamically and kinetically favor the FA‐to‐HF conversion. These findings provide valuable mechanistic insights for the rational design of high‐performance photocatalysts tailored for the selective conversion of biomass‐derived platform molecules into value‐added chemicals.

An oxygen‐doping strategy has been employed for the Zn₂In₂S₅ photocatalyst, which significantly enhances its conversion efficiency of biomass‐derived furfuryl alcohol to the jet fuel precursor hydrofuroin and hydrogen. The oxygen dopant plays pivotal roles in photocarrier separation, byproduct suppression, and the adsorption of the reactants, thereby boosting both the quantum yield and the selectivity toward hydrofuroin.

1. Introduction

Photoreforming of biomass‐derived compounds emerges as a green, sustainable, scalable, and cost‐effective pathway for the concurrent production of hydrogen and value‐added chemicals under mild reaction conditions [1, 2, 3, 4, 5]. Compared to conventional photocatalytic water splitting, photoreformative H₂ production offers distinct advantages in both reaction efficiency and product value [6, 7, 8]. On the one hand, the oxidative transformation of organic substrates is thermodynamically and kinetically more favorable than the oxygen evolution reaction (OER) in water splitting. Employing biomass‐derived organics as proton donors can thus drastically enhance the hydrogen production rate, which is typically constrained by the sluggish kinetics of the OER during photocatalytic water splitting [9, 10]. On the other hand, as opposed to the formation of low‐value byproduct oxygen, photoreforming enables the synthesis of high‐value chemicals under moderate, environmentally benign conditions, thereby conferring substantial techno‐economic competitiveness for large‐scale industrial implementation [11, 12]. Leveraging abundant biomass as feedstock (global annual output >120 billion tons), photoreforming holds enormous promise for reshaping our energy and chemical industries, which are severely addicted to fossil fuels.

Within the aforementioned framework, a series of lignocellulose‐derived downstream organics, including furfural [13], methylfurans [8], dimethyl ethers [14] and alcohols [15, 16, 17] have been successfully photoreformed into valuable chemicals, fuels, and fuel additives, with concurrent release of hydrogen. Notably, furfuryl alcohol (FA), a one‐step product from cellulose‐derived 5‐hydroxymethylfurfural (HMF) and hemicellulose‐derived furfural (FF) (Figure S1) [18, 19, 20], emerges as a compelling candidate for the jet fuel synthesis [21] and hydrogen production [22]. As schematically illustrated in Scheme 1, this process typically entails the photocatalytic dehydrogenative C─C coupling of FA to hydrofuroin (HF), a key jet fuel precursor that can be further hydrodeoxygenated to yield alkanes within the jet fuel carbon range (C₈‐C₁₆). Nevertheless, the dehydrogenative C─C coupling of FA necessitates the prior activation of the α─C─H bond before C─C bond formation—a significant challenge attributed to the intrinsic inertness of the sp ³ C─H bond and the coexistence of the more reactive O─H bond. Consequently, side products such as furfural (FF) and furoic acid (FAC) often dominate over the desired C─C coupling products, leading to unsatisfactory product selectivity [23, 24, 25, 26]. Currently, state‐of‐the‐art photocatalysts typically achieve a HF selectivity no more than 63% at a reasonable FA conversion efficiency (> 98%) [21, 27]. Therefore, the development of effective strategies to enhance HF selectivity is highly desirable, as it not only simplifies subsequent post‐treatment processes but also reduces overall operational costs.

SCHEME 1.

Conceptual illustration of photoreforming biomass‐derived FA to jet fuel precursor HF, the concomitant H₂ production, and their downstream industrial applications. The HF can be converted into jet fuel by hydrodeoxygenation (HDO) processes.

Apart from poor HF selectivity, photocatalytic FA photoreforming is also plagued by a low quantum yield or an inadequate utilization of photocarriers (i.e. photo‐generated electrons and holes). This deficiency is evidenced by the prevalent requirement for intense light sources (e.g., blue light‐emitting diodes (LEDs)) and prolonged illumination durations (> 20 h) to achieve viable FA conversion efficiencies. For instance, In₂S₃/Zn₅In₂S₈ hybrid photocatalysts (20 mg) need 36 h of illumination under blue LEDs (460 nm, 5 W) to achieve the conversion of 0.46 mmol FA [21, 27]. Such stringent reaction conditions further exacerbate the poor selectivity issue, as the formed products are susceptible to subsequent attack by photocarriers and the derived radical species—an observation supported by the time‐dependent degradation of HF selectivity [28]. It is generally recognized that the majority of photocarriers in photocatalytic systems undergo unproductive annihilations, attributed to their short lifetime [29]. The photocarrier lifetime of conventional photocatalysts typically ranges from 10⁻¹² s to 10⁻⁹ s, which is incompatible with the timescale of chemical reactions (10⁻⁹ s–10⁰ s) [30]. Therefore, extending the lifetime of photocarriers emerges as a critical strategy to enhance the quantum yield of HF, representing another major challenge in the field of FA photoreforming.

In this work, we demonstrate that both the selectivity and quantum yield issues of FA photoreforming can be resolved simultaneously by doping oxygen into Zn₂In₂S₅. The oxygen dopants play an important role in the critical steps of FA homo‐coupling and significantly modify the electronic structures of Zn₂In₂S₅, affording highly selective conversion of FA to HF with a high quantum yield.

2. Results and Discussion

2.1. Oxygen Doping Over Zn₂In₂S₅

Oxygen‐doped Zn₂In₂S₅ was synthesized via a hydrothermal route using polyvinylpyrrolidone (PVP) as an additive. PVP is reported to be a mild reducing agent which can create sulfur vacancies in metal sulfides for subsequent O incorporation during hydrothermal synthesis [31, 32, 33, 34]. Our Fourier transform infrared spectroscopy (FTIR) analysis revealed that PVP can be an O supplier by cleaving its C═O bonds, as evidenced by the prominent degradation of its C═O stretching vibration (1650 cm⁻¹) after hydrothermal reactions (Figure S2). For clarity, samples prepared with and without PVP are denoted as ZIS‐O and ZIS, respectively. The successful O‐doping of Zn₂In₂S₅ was verified by a series of spectroscopic characterizations.

Figure 1a and Figure S3 compare the X‐ray diffraction (XRD) patterns of ZIS‐O and ZIS, alongside the standard XRD patterns of Zn₂In₂S₅. All samples exhibit analogous X‐ray diffraction peaks that are well‐matched with the standard Zn₂In₂S₅ patterns, indicating the successful synthesis of single‐phase Zn₂In₂S₅ and the preservation of its crystal structure upon O‐doping. Nevertheless, a slight shift of diffraction peaks toward higher angles is observed for ZIS‐O relative to ZIS, as exemplified by the magnified (0024) diffraction peak in Figure 1a. This observation is attributable to the lattice contraction caused by the substitution of larger S^{2 −} ions (ionic radius: 1.70 Å) with smaller O^{2 −} ions (ionic radius: 1.26 Å).

FIGURE 1.

Structure of oxygen‐doped Zn₂In₂S₅. (a) XRD patterns of Zn₂In₂S₅ (ZIS) and O‐doped Zn₂In₂S₅ (ZIS‐O), standard patterns of ZIS (JCPDS:01‐070‐2618) are included for comparisons and (0024) reflection are enlarged on the right. Dashed line is the guide to the eye. (b) UV–vis DRS spectra of ZIS and ZIS‐O. c XPS spectra at O 1s state of ZIS and ZIS‐O, signals assignable to lattice oxygen (O²⁻), hydroxyl group (OH⁻) and water (H₂O) are labelled. (d) XAS spectra at O K‐edges of ZIS and ZIS‐O, absorption due to the transition from filled 1s orbitals to unfilled orbitals is labelled. (e) HR‐TEM image of ZIS. (f) The contrast intensity profile along the dashed yellow arrow in e. (g) O content of ZIS‐O determined by ONH analyzer as a function of polyvinylpyrrolidone (PVP) dosage used during the synthesis. (h) HR‐TEM image of ZIS‐O. (i) The contrast intensity profile along the dashed yellow arrow in h.

Since the valence band (VB) of Zn₂In₂S₅ is predominantly composed of S 3p orbitals, the replacement of S with O—an element with higher electronegativity (χ_O = 3.44, χ_S = 2.58) [35]—results in an enlarged bandgap. This is corroborated by the blue‐shifted absorption edge in the UV‐visible diffuse reflectance spectra (UV–vis DRS) of ZIS‐O (Figure 1b and Figure S4). The incorporation of O into the Zn₂In₂S₅ lattice was further confirmed by X‐ray photoelectron spectroscopy (XPS) and X‐ray absorption spectroscopy (XAS), both of which are sensitive to the chemical environment of O. The O 1s XPS spectra exhibit broad signals that can be deconvoluted into multiple peaks (Figure 1c). The peaks centered at 532.6 and 531.4 eV are assigned to surface‐adsorbed H₂O and hydroxyl groups (OH⁻), respectively—phenomena also reported for other metal sulfides [33, 36]. In contrast to ZIS, ZIS‐O exhibits an additional peak centered at 529.8 eV. This peak, which is typically characteristic of lattice oxygen (O²⁻) in metal oxides [37, 38], unequivocally confirms the incorporation of O into the Zn₂In₂S₅ lattice. Consistent results are also acquired from XPS spectra after Ar⁺ etching which detect the O²⁻ species in the subsurface of ZIS‐O (Figure S5), verifying again the successful O doping and excluding the interference of surface impurities.

Furthermore, O K‐edge XAS reveals distinct absorption features between ZIS and ZIS‐O (Figure 1d). For ZIS, the XAS spectra display relatively narrow absorption signals, which are typical of O₂ and H₂O molecules with degenerate and isolated molecular orbitals [37]. This is consistent with the XPS results, indicating that ZIS only contains extrinsic oxygen‐containing species. In contrast, ZIS‐O exhibits a broadened asymmetric absorption profile, which is attributed to the hybridized O 2p orbitals that reduce orbital degeneracy. For example, the absorption edge around 532 eV is characterized by a broad absorption line with a shoulder, which is ascribed to the electronic transition from the O 1s orbital to the hybridized O 2p/In 5s/Zn 4s orbitals [39]. A similar phenomenon is observed in the absorption profile at higher photon energies. Collectively, these XPS and XAS results confirm that O is doped into the Zn₂In₂S₅ lattice and forms strong chemical bonds with Zn/In cations.

Field‐emission scanning electron microscopy (FE‐SEM) analysis demonstrates that ZIS‐O retains the flower‐like morphology of ZIS and exhibits a homogeneous elemental distribution (including O, Figure S6). However, high‐resolution transmission electron microscopy (HR‐TEM) analysis reveals subtle changes in the local crystal structure induced by O‐doping. As shown in Figure 1e,f,h,i, ZIS‐O exhibits less distinct lattice fringes and more blurred regions compared to ZIS, which is presumably due to increased lattice strain arising from S/O substitution. This is further reflected in the contrast intensity profiles of the lattice fringes, where ZIS‐O shows a significant deviation from the periodic intensity fluctuation observed for ZIS. Brunauer‐Emmett‐Teller (BET) surface analysis (Figure S7) reveals that O doping favors a larger surface area of Zn₂In₂S₅, likely due to the fact that PVP often serves as a growth modifier to regulate particle growth, as evidenced by the SEM observation (Figure S6). Barrett‐Joyner‐Halenda (BJH) analysis further suggests that ZIS and ZIS‐O have similar pore size distributions. Water contact angle measurements (Figure S8) indicate that ZIS‐O is slightly more hydrophilic (contact angle: 67.59°) than ZIS (contact angle: 70.08°). The improved hydrophilicity originates from the enhanced surface polarity induced by oxygen doping, which facilitates the wetting of polar solvents. Such favorable surface modifications provide abundant surface active sites and strengthen the interfacial interaction with the reaction solvent, thereby benefiting the photocatalytic FA reforming reaction.

Notably, O‐doping causes a negative shift in the XPS binding energies of all elements in Zn₂In₂S₅ (Figure S9), which can be rationalized by the mild reducing capacity of PVP used for O‐doping. The O content of ZIS‐O is tunable by adjusting the PVP dosage during synthesis. As shown in Figure 1g, the O content, determined by elemental analysis (ONH analyzer), exhibits an almost linear dependence on the PVP dosage. Unless otherwise stated, ZIS‐O synthesized with 75 mg PVP was used as the representative O‐doped Zn₂In₂S₅ sample in subsequent experiments.

2.2. Photoreforming of Furfuryl Alcohol

The photoreforming of furfuryl alcohol (FA) was conducted under simulated AM1.5G illumination (100 mW cm⁻²) using ZIS‐O samples with varying O contents. Figure 2a summarizes the product yield as a function of O content in ZIS‐O. A series of critical reaction parameters, including FA conversion efficiency, product yield, and selectivity, can be modulated by tailoring the O uptake in ZIS‐O. For instance, the FA conversion efficiency is doubled when the O content is increased from 3.2 at.% to 15.9 at.%. More importantly, O‐doping enables switching of the dominant liquid product: from furfural (FF) as the sole liquid product over pristine ZIS to hydrofuroin (HF) as the major product over ZIS‐O. The hydrogen (H₂) yield is also significantly enhanced, demonstrating the positive role of O‐doping in promoting FA photoreforming. Control experiment using ZIS containing sulfur vacancies (ZIS‐V_S) indicated no such improvement in terms of both product yield and selectivity (Figure S10). Considering the fact that O doping has only mild or negligible modifications to the ZIS including surface area, morphology and hydrophilicity, such distinctive performance of ZIS‐O is attributable to O dopants. Notably, ZIS‐O with an O content of 15.9 at.% achieves a FA conversion efficiency of 98% and an HF selectivity of 92% within merely 4 h of reaction. To the best of our knowledge, this performance outperforms state‐of‐the‐art photocatalysts reported for FA‐to‐HF conversion (Table S1) [21, 27].

FIGURE 2.

Photoreforming of furfuryl alcohol (FA). (a) Yield of hydrofuroin (HF), furfural (FF), H₂ and the selectivity of HF as a function of oxygen content of ZIS‐O, reaction conditions: 10 mg photocatalyst, 0.1 mmol furfuryl alcohol (FA) in a solution containing 3 mL deionized water and 7 mL acetonitrile, 4 h under AM1.5 G, Ar atmosphere. (b) Temporal conversion of FA and product yield over ZIS. (c) Temporal conversion of FA and product yield over ZIS‐O. (d) Action spectra (AQY vs. λ) of ZIS‐O for the yield of HF. (e) Extended testing cycles for the yield and selectivity of HF over ZIS‐O under AM1.5G, each cycle has a fixed reaction time of 4 h. (f) Digital photograph of the polyethylene terephthalate (PET) bag reactor for FA photoreforming under natural sunlight. (g) Yield of HF, FF, H₂ and the selectivity of HF over the bag reactor containing 200 mg ZIS‐O photocatalyst, 60 mL FA, 132 mL H₂O and 308 mL CH₃CN under natural sunlight for 30 days. (h) Digital photograph of HF product extracted from the bag reactor after the 30‐day reaction.

Temporal experiments were performed to monitor the FA photoreforming process at different reaction time points (Figure 2b,c). For pristine ZIS, FF was the sole liquid product, and its yield was well‐matched with FA conversion at various time points. These results indicate that pristine ZIS can only catalyze the conversion of FA to FF and is incompetent in mediating the dehydrogenative C‐C coupling reaction required for HF formation. In contrast, ZIS‐O produced HF as the major liquid product throughout the reaction, with FF generated as a transient byproduct at the early stage. The accumulated FF was further consumed at the later stage of the reaction, leading to a gradual increase in HF selectivity over time. After 5 h of reaction, ZIS‐O achieved complete FA conversion (100%) with an HF selectivity as high as 95%. This result is distinct from those of ZIS and previously reported photocatalysts, where HF selectivity tends to decrease with prolonged reaction time [21]. However, the reaction pathway over ZIS‐O shows high similarity to the homo‐coupling of benzyl alcohol [40, 41], implying a plausible common reaction mechanism.

The apparent quantum yield (AQY) was determined to evaluate the photon utilization efficiency of ZIS‐O for FA‐to‐HF conversion. As shown in the action spectra (Figure 2d and Table S2), the AQY for HF production correlates well with the light absorption profile of ZIS‐O. This confirms that the photoreforming reaction is essentially photon‐driven and proceeds via bandgap excitation of ZIS‐O. Notably, the AQY reaches up to 37.6% at 380 ± 20 nm and 12.4% at 420 ± 20 nm, highlighting the superior solar energy utilization capacity of ZIS‐O. Furthermore, the stability of ZIS‐O for FA photoreforming was investigated through extended cyclic tests. As shown in Figure 2e, ZIS‐O maintains excellent performance—i.e., high FA conversion efficiency and high HF selectivity—over five consecutive cycles without noticeable degradation. Post‐reaction characterizations of the used ZIS‐O (Figures S11–S13 and Table S3) reveal that its crystal structure, microstructure, metal content, and surface elemental states remain unchanged compared to the fresh sample. Collectively, these results affirm the exceptional activity and stability of ZIS‐O for FA photoreforming, which is a crucial prerequisite for its potential industrial applications. Nevertheless, FTIR analysis indicates the presence of residual PVP in the ZIS‐O sample (Figure S2). To clarify the influence of residual PVP, we further evaluated the thoroughly washed ZIS‐O that is nearly absent in PVP residues. The thoroughly washed sample exhibits identical photocatalytic performance to that of the mildly washed one (Figure S14), thereby ruling out any contribution of residual PVP to the enhanced photocatalytic activity.

As a proof‐of‐concept for practical FA photoreforming, ZIS‐O was evaluated in a pilot‐scale polyethylene terephthalate (PET) bag reactor (20 cm × 25 cm) under natural sunlight (Figure 2f). Figure 2g presents the product yields and HF selectivity over a 30‐day testing period (08:00–18:00 local time daily). Although the product yields are highly dependent on weather conditions, ZIS‐O maintains an average HF selectivity of 90% throughout the long‐term test. At the conclusion of the test, 200 mg of ZIS‐O photoreformed 60 mL of FA into 271.1 mmol of HF, 26.9 mmol of FF, and 230.5 mmol of H₂, achieving a FA conversion efficiency of 82.1% and a turnover number (TON) of 1390 based on the moles of photocatalyst. The jet fuel precursor HF was readily separated from the reaction suspension via distillation (Figure 2h), and its chemical identity and purity were confirmed by ¹H nuclear magnetic resonance (¹H NMR) spectroscopy (Figure S15). The isolated yield of HF was approximately 67.2%, based on the initial amounts of FA used. The product purity was determined by GC to be 96.5%, with the main impurity being FF (2.8%) and FA (0.7%). These results not only validate the feasibility of the FA‐to‐HF conversion route proposed in this work but also highlight its significant potential for large‐scale practical applications.

2.3. Photocarrier Separation

Given the distinct photoreforming activities between ZIS and ZIS‐O, it is imperative to investigate the behaviors of photocarriers in the presence of oxygen dopants. First, photoluminescence (PL) spectroscopy, an effective technique for probing radiative recombination of photocarriers, was employed to compare ZIS and ZIS‐O. As shown in Figure 3a, both samples exhibit broad emission bands, which can be deconvoluted into four distinct peaks centered at 432, 466, 532, and 602 nm. The peak at 432 nm is attributed to interband emission, as its photon energy exceeds the bandgaps (E _g) of ZIS (E _g = 2.76 eV) and ZIS‐O (E _g = 2.89 eV) (Figure S16). The 466 nm peak is typically assigned to band‐edge emission, as it is close to the light absorption edge with a reasonable Stokes shift. The remaining two peaks (532 and 602 nm) correspond to sub‐bandgap emissions, which are ascribable to radiative photocarrier recombination mediated by mid‐gap states.

FIGURE 3.

Photocarrier separation properties after O doping. (a) Photoluminescence (PL) spectra of ZIS and ZIS‐O at 298 K with a 380 nm excitation. (b) Time‐resolved PL (TRPL) decay spectra of ZIS and ZIS‐O, inset shows the average PL decay lifetime (τ_average). (c) Integrated PL intensity as a function of temperature of ZIS and ZIS‐O, insets show the exciton binding energy (E_b) determined by fitting the data using Arrhenius relation. (d) EPR spectra of TEMPO electron‐scavenger experiments (TEMPO‐e⁻) over ZIS and ZIS‐O. (e) EPR spectra of TEMPO hole‐scavenger experiments (TEMPO‐h⁺) over ZIS and ZIS‐O. (f) SPV spectra of ZIS and ZIS‐O as a function of excitation wavelength. (g) Transient photovoltage (TPV) response of ZIS and ZIS‐O. (h) UPS‐SEC and UPS‐VB spectra of ZIS and ZIS‐O, work function (WF) is deduced by subtracting photon energy (40 eV) with secondary electron cutoff energy. (i) Schematic illustration of UPS‐deduced band edge positions (CB and VB) of ZIS and ZIS‐O, where photocarrier behaviors such as generation, recombination, migration and trapping by surface states are also depicted.

In comparison to ZIS, the emission intensity at 532 nm is drastically enhanced for ZIS‐O, accounting for the significantly stronger overall PL signal of ZIS‐O. It is proposed that the O dopants are intrinsically associated with the 532 nm emission. With a photon energy of 2.33 eV, this emission corresponds to radiative recombination from shallow donor or acceptor states adjacent to the band edges [42, 43]. Time‐resolved PL (TRPL) decay spectroscopy was further utilized to investigate the dynamics of PL signal decay. As presented in Figure 3b and Figure S17, ZIS‐O exhibits a much slower decay process than ZIS, indicating that O‐doping suppresses photocarrier recombination in Zn₂In₂S₅. The average decay lifetimes increase considerably when comparing pristine ZIS (69.6 ns) with ZIS‐O at different O content, i.e. 109.89 ns (8.7 at.%), 150.21 ns (15.9 at.%), and 138.70 ns (18.3 at.%). The longest average decay lifetime is found for ZIS‐O with a doping level of 15.9 at.% whose decay lifetime (150.21 ns) is 2.1 times longer than that of ZIS (69.64 ns) (inset of Figure 3b and Table S4).

Photocarriers are typically bound as excitons (i.e., e⁻‐h⁺ pairs), and their exciton binding energy (E_b) quantifies the difficulty of photocarrier dissociation. E_b can be quantitatively determined from the temperature dependence of PL intensity (Figure S18). As illustrated in Figure 3c, based on the Arrhenius equation (Table S5), the fitted E_b values are 23.35 meV for ZIS and 18.99 meV for ZIS‐O. A small E_b, particularly one much smaller than the thermal energy (23.5 meV at 273 K), indicates that photocarriers can be readily dissociated into free charge carriers under ambient conditions, as is the case for ZIS‐O. Collectively, the PL and TRPL results demonstrate that O‐doping significantly improves the photocarrier separation efficiency of Zn₂In₂S₅.

The enhanced separation efficiency of photocarriers strongly implies that reactive charge carriers are readily accessible to surface reaction sites, thereby contributing to the superior photocatalytic activity. The accessibility of photocarriers to surface redox reactions was probed via electron paramagnetic resonance (EPR) spectroscopy using 2,2,6,6‐tetramethylpiperidinyloxy (TEMPO) as a scavenger. Specifically, TEMPO undergoes redox reactions with photo‐generated electrons and holes to form EPR‐silent compounds (TEMPO‐e⁻ and TEMPO‐h⁺), which results in the attenuation of the characteristic EPR signal of TEMPO. As illustrated in Figure 3d,e and Figure S19, ZIS‐O exhibits significantly higher efficiency than ZIS in quenching the triplet EPR signals of TEMPO in both electron‐scavenging (TEMPO‐e⁻) and hole‐scavenging (TEMPO‐h⁺) experiments. These findings clearly demonstrate that oxygen doping facilitates the migration of a greater number of reactive charge carriers to the Zn₂In₂S₅ surface. The number of electrons and holes accessible to the photocatalyst's surface was then quantitatively determined based on the degradation of TEMPO EPR signals, as summarized in Table S6. It is found that the C‐C coupling rate exhibits a linear relationship with the square of hole concentration (Figure S20), indicating that the FA coupling reaction over Zn₂In₂S₅ proceeds via a two‐hole transfer mechanism.

This deduction is further corroborated by surface photovoltage (SPV) spectroscopic analysis, which enables the electrical detection of charge carriers approaching the material surface [44, 45, 46]. As presented in Figure 3f, ZIS‐O yields considerably stronger SPV responses than ZIS, in terms of both signal intensity and spectral range. This observation confirms the proposition that oxygen doping promotes the migration of photocarriers to the surface. Notably, a negative SPV signal emerges at wavelengths longer than 430 nm, which is attributed to mid‐gap states that can effectively trap photo‐generated electrons—a phenomenon also documented in other n‐type semiconductors [46, 47, 48]. Furthermore, transient photovoltage (TPV) measurements were conducted to investigate the dynamics of photocarrier separation and recombination. As shown in Figure 3g, ZIS exhibits a single, rapidly decaying transient photovoltage signal, indicating the fast recombination of photocarriers [49]. In contrast, ZIS‐O displays a distinct dual‐peak feature: the initial sharp peak (Peak 1) corresponds to the instantaneous separation and extraction of photocarriers [50], while the subsequent broad and slowly evolving peak (Peak 2) is associated with the release of photocarriers from shallow trap states followed by delayed recombination [51, 52]. This dual‐peak kinetic behavior confirms that O doping introduces shallow mid‐gap states in ZIS, which can temporarily trap electrons and thereby effectively suppress photocarrier recombination.

Surface photovoltage microscopy (SPVM), which operates based on the detection of contact potential difference (CPD) to map sample surfaces [53, 54], was employed to probe the local SPV distribution in ZIS and ZIS‐O. As illustrated in Figure 4, ZIS‐O exhibits consistently 5‐fold higher SPV signals than ZIS across the entire sampling area (5 × 5 µm). This observation demonstrates that the O‐doping‐enhanced SPV is homogeneously distributed and can effectively enrich charge carriers at local reaction sites.

FIGURE 4.

Surface photovoltage microscopy (SPVM) analysis. (a) Kelvin probe force microscopy (KPFM) image of ZIS under dark and light illumination (λ = 405 nm). (b) KPFM image of ZIS‐O under dark and light illumination (λ = 405 nm). (c) Linear scan of KPFM potential along the blue arrow of a. (d) Linear scan of KPFM potential along the blue arrow of b. The contact potential difference (CPD) before and after light illumination corresponds to the surface photovoltage (SPV).

To further verify the experimental deduction in terms of mid‐gap states, the surface density of states (DOS) of Zn₂In₂S₅ with and without O doping was studied via density functional theory (DFT) calculations. As presented in Figures S21 and S22, O doping introduces additional DOS at the bottom of the conduction band (CB), confirming the formation of shallow donor states. These computational results are well consistent with the preceding PL analysis, which revealed that O doping significantly enhances the PL emission at 532 nm—an emission corresponding to photocarrier recombination mediated by shallow mid‐gap states. The band edge positions of ZIS and ZIS‐O were then accurately determined via ultraviolet photoelectron spectroscopy (UPS) measurements, employing 40 eV ultraviolet photons from a synchrotron radiation source. As illustrated in Figure 3h and Figure S23, the work function and valence band (VB) edge positions were readily derived from the onset points in the UPS secondary electron cutoff (UPS‐SEC) and UPS valence band (UPS‐VB) spectra, respectively [55, 56]. Combining these results with the bandgap values obtained from Tauc plot analysis (Figure S23a), the band edge positions of both samples were resolved, as schematically depicted in Figure 3i and Figure S23d. O dopants exert a dual regulatory effect on the band structure of Zn₂In₂S₅. On the one hand, they induce a negative shift in the band edge positions of Zn₂In₂S₅. From an energetic perspective, this shift rationalizes the enhanced H₂ evolution activity observed for ZIS‐O. On the other hand, O dopants introduce shallow donor states below the CB, which facilitate photocarrier separation. Specifically, these states enable the trapping and detrapping of electrons, thereby extending the photocarrier lifetime and suppressing recombination.

To further elucidate the photocarrier dynamics of Zn₂In₂S₅ upon O doping, femtosecond transient absorption (fs‐TA) spectroscopy measurements were performed on ZIS and ZIS‐O, with excitation at 400 nm (pump wavelength). Figure 5a,b present the 2D contour plots of the fs‐TA spectra for ZIS and ZIS‐O, respectively. For pristine ZIS, the fs‐TA spectra are dominated by intense negative absorption signals below 520 nm. These signals are assigned to stimulated emission (SE), as evidenced by their significant overlap with the photoluminescence (PL) spectrum of ZIS. The contribution of ground‐state bleaching (GSB) to these signals can be excluded, given that ZIS exhibits an absorption edge around 450 nm. Beyond the SE‐dominated region, ZIS shows negligible fs‐TA responses in the 550–750 nm range. In sharp contrast, ZIS‐O displays strong and broad positive absorption signals in this spectral window, which are attributed to the excited‐state absorption (ESA) of photocarriers. These ESA signals are intrinsically associated with the introduction of shallow donor states by O dopants, which serve as trapping sites for photoexcited electrons from the CB of Zn₂In₂S₅. Notably, the ESA signals of ZIS‐O exhibit excellent temporal stability, as revealed by the time‐resolved evolution of the fs‐TA spectra (Figure 5c). Even after 2500 ps, the ESA signal intensity retains approximately 50% of its peak value, indicating a prolonged photocarrier lifetime in ZIS‐O. This observation is further corroborated by the kinetic analysis of the fs‐TA traces at 700 nm, which were fitted using a double first‐order exponential decay model (Figure 5d and Table S7). The fitting results yield an average decay lifetime of 3785.64 ps for ZIS‐O. As schematically illustrated in Figure 5e, ZIS‐O exhibits two distinct decay pathways: a fast decay component (τ₁ = 55.09 ps, amplitude = 24.25%) corresponding to direct band‐to‐band photocarrier recombination, and a dominant slow decay component (τ₂ = 3892.95 ps, amplitude = 75.75%) associated with photocarrier annihilation mediated by mid‐gap states (i.e., the O‐doping‐induced donor states). Collectively, these dynamic results demonstrate that O dopants significantly extend the photocarrier lifetime of Zn₂In₂S₅ by introducing mid‐gap states for charge accommodation, which is fully consistent with the preceding experimental findings. The long‐lived photocarriers (on the nanosecond timescale) are kinetically compatible with the typical time scale of chemical reactions (nanoseconds to seconds), thereby contributing to the enhanced FA conversion efficiency over ZIS‐O [30].

FIGURE 5.

Photocarrier dynamics. (a) fs‐Ta spectra of ZIS in 2D contour plot. (b) fs‐Ta spectra of ZIS‐O in 2D contour plot. (c) Temporal evolution of fs‐TA spectra of ZIS‐O, degradation of the excited‐state absorption (ESA) is indicated by the dashed arrow. (d) Kinetic traces for the decay of fs‐TA signal of ZIS‐O at 700 nm. (e) Schematic illustration of photocarrier recombination kinetics in ZIS‐O.

2.4. The Photoreforming Mechanisms

To gain deeper insights into the distinct catalytic activities between ZIS and ZIS‐O, the photoreforming mechanisms of FA were systematically investigated. First, scavenger experiments were performed to elucidate the roles of photo‐generated electrons and holes in the photoreforming process over ZIS‐O, with the results summarized in Figure 6a. In the presence of triethanolamine (TEOA), a well‐established hole scavenger, FA conversion was completely suppressed. This observation confirms that the photoreforming reaction is initiated by photo‐generated holes. In contrast, the introduction of sodium persulfate (Na₂S₂O₈), a typical electron scavenger, significantly compromised the selectivity toward hydrofuroin (HF), reducing it from 92% to 46%. This finding indicates that photo‐generated electrons play a dominant role in the homo‐coupling of FA. Second, EPR spin‐trapping experiments were conducted to identify the reaction intermediates. 2,2‐Dimethyl‐1‐oxido‐3,4‐dihydropyrrol‐1‐ium (DMPO) was employed as the spin‐trapping agent, and the EPR spectra of the trapped radical adducts are presented in Figure 6b. Both ZIS and ZIS‐O exhibited complex EPR signals, which could be deconvoluted into two sets of characteristic peaks via spectral simulation. The sextet signal was attributed to the DMPO‐C_α radical adduct [57, 58], with hyperfine coupling constants of α_H = 21.9 G and α_N = 15.1 G. The other multiplet signal was typically assigned to the DMPO‐•H radical adduct [59, 60]. These EPR results demonstrate that C_α radicals and hydrogen radicals (•H) are the key reaction intermediates in FA photoreforming, which are generated via the cleavage of α─C─H bonds in FA. Notably, ZIS‐O displayed much stronger EPR signals compared to ZIS, confirming that O doping facilitates FA photoreforming by promoting the cleavage of α─C─H bonds with higher efficiency.

FIGURE 6.

Mechanisms of FA photoreforming. (a) Yield of HF, FF and selectivity of HF in the presence of an electron/hole scavenger. Reaction conditions: 10 mg ZIS‐O, 10 mM scavenger, 7 mL acetonitrile, 3 mL deionized water, Ar atmosphere, AM 1.5G illumination for 4 h. (b) EPR spectra of DMPO spin‐trapping experiments over ZIS and ZIS‐O. (c) Photocatalytic conversion rate of FF to HF over ZIS and ZIS‐O. (d) Proposed reaction mechanisms of photoreforming FA. (e) Open‐circuit potential (OCP) response of ZIS and ZIS‐O electrode before and after FA injection. Testing conditions: 1 M KOH electrolyte, 20 mM FA injection, Ar atmosphere and in dark.

Since both catalysts generated the same types of reaction intermediates, the discrepancy in their product selectivity was attributed to the different conversion rates of furfuryl alcohol (FF). Control experiments revealed that both ZIS and ZIS‐O could catalyze the conversion of FF‐to‐HF in the presence of the hole scavenger TEOA (Figure S24). However, the FF conversion rate over ZIS‐O was approximately 11.3 times higher than that over ZIS (Figure 6c). This enhanced FF conversion efficiency accounts for the gradual decrease in FF yield during FA photoreforming (Figure 2c) and the exceptional HF selectivity of ZIS‐O. Based on the aforementioned results, a reaction mechanism for the FA photoreforming is proposed in Figure 6d. C_α radicals are identified as the critical intermediates for the C‐C coupling reaction leading to HF formation. These C_α radicals can be generated through two pathways: i) the dehydrogenative oxidation of FA, and ii) the hydrogenative reduction of FF. The latter pathway is reversible, and the forward reaction rate (i.e., FF reduction to C_α radicals) determines the HF selectivity. Specifically, a high FF consumption rate that exceeds its generation rate shifts the product distribution toward HF, whereas a low FF consumption rate results in FF accumulation and reduced HF selectivity.

Beyond the reaction mechanisms, the adsorption of reactants onto the photocatalyst surface, a prerequisite for the photoreforming reaction, was also investigated. Reactant adsorption is characterized by a shift in the photocatalyst's Fermi level, which arises from strong interactions between reactant molecules and the photocatalyst surface [61, 62]. This Fermi level shift can be probed by monitoring the variation in open‐circuit potential (OCP) of the photocatalyst electrode upon the addition of reactants. Herein, the OCP values of ZIS and ZIS‐O were recorded in the dark both before and after FA injection and the magnitude of OCP variation serves as a good indicator of their FA adsorption capacity. As illustrated in Figure 6e, ZIS‐O exhibits a significantly larger OCP change than ZIS following FA injection, demonstrating a superior FA adsorption capacity.

2.5. Theoretical Calculations

To achieve a more comprehensive understanding of the reaction mechanisms, density functional theory (DFT) calculations were performed on the FA homo‐coupling reaction to HF. (001) crystal facet of Zn₂In₂S₅ was adopted for calculation according to TEM analysis (Figure 1e,h) and oxygen doping was simulated by substituting one surface S atom with an O atom. Figure 7a illustrates the reaction geometries involving C_α radicals as the key reaction intermediates. On both the (001) surfaces of ZIS and ZIS‐O, FA adsorption and the subsequent dehydrogenative oxidation to C_α radicals are energetically downhill processes. Notably, ZIS‐O exhibits an additional free energy reduction of 0.21 and 0.17 eV compared to ZIS for these two steps, respectively. This is consistent with the previous observation that ZIS‐O enhances FA adsorption capacity and generates a higher abundance of C_α radicals than ZIS. This phenomenon is presumably attributed to the fact that C_α radicals, which interact with ZIS‐O via C─O bonds, are more strongly anchored than those bound to ZIS via C─S bonds. Crystal orbital Hamilton population (COHP) analysis reveals that the O─C bond possesses significantly higher bond strength than the S─C bond, owing to its enhanced bonding character (Figure 7b,c). The strong anchoring of C_α radicals on the photocatalyst surface increases their local concentration, which is kinetically favorable for the subsequent coupling reactions. More importantly, the coupling of C_α radicals to form HF is an energetically downhill step on ZIS‐O (ΔG = – 0.46 eV) but an energetically uphill step on ZIS (ΔG = 0.03 eV). This favorable energy profile further rationalizes the high HF yield over ZIS‐O, as the reaction is favored both kinetically and thermodynamically. Additionally, DFT calculations were extended to the hydrogenative reduction of FF to C_α radicals, a critical step governing product selectivity. As shown in Figure 7b, ZIS‐O exhibits a more significant free energy decrease (≈0.18 eV) than ZIS, which is consistent with its substantially higher FF conversion rate (Figure 6c). Collectively, these DFT results provide a comprehensive explanation for the high FA conversion efficiency of ZIS‐O and its exceptional product selectivity toward HF.

FIGURE 7.

Theoretical studies of FA photoreforming. (a) DFT‐calculated geometries of FA homo‐coupling at (001) surface of ZIS and ZIS‐O, the corresponding free energy diagram is shown at the bottom. (b) DFT‐calculated crystal orbital Hamilton populations (COHP) of S‐C bond in ZIS and O─C bond in ZIS‐O, Fermi level is denoted as the dashed line. (c) The integrated COHP up to Fermi level of S─C bond in ZIS and O─C bond in ZIS‐O, insets show the optimized atom geometries. (d) DFT‐calculated geometries of FF hydrogenative reduction to C_α radical, the corresponding free energy diagram is shown at the bottom.

3. Conclusions

An oxygen‐doping strategy was employed to modify the Zn₂In₂S₅ photocatalyst, which significantly enhanced the FA photoreforming efficiency with superior product selectivity toward HF, a high‐value jet fuel precursor. O doping was facilely achieved via the introduction of PVP into the hydrothermal synthesis process, and the doping content was tunable by regulating the PVP dosage. Notably, the O dopants exert multiple pivotal functions in the FA photoreforming reactions: 1) they introduce shallow donor levels within the bandgap of Zn₂In₂S₅, which substantially accelerates the separation of photogenerated carriers and prolongs their lifetime; 2) they remarkably promote the conversion rate of furfuryl alcohol (FF), thereby shifting the product selectivity from FF toward HF; 3) they enhance the FA adsorption capacity of the photocatalyst, which constitutes a fundamental prerequisite for efficient FA photoreforming. At an optimal doping level (15.9 at%), the O‐doped Zn₂In₂S₅ achieved complete FA conversion (100%) within 5 h under simulated sunlight irradiation, with an HF selectivity of up to 95%. The corresponding apparent quantum yield (AQY) reached 37.6% at 380 ± 20 nm and 12.4% at 420 ± 20 nm. Moreover, long‐term stability tests (30 days) under natural sunlight irradiation demonstrated that O‐doped Zn₂In₂S₅ maintained an HF selectivity of 90%, accompanied by exceptional structural and catalytic stability. Theoretical calculations revealed that O doping thermodynamically and kinetically favors the FA‐to‐HF conversion process. On the one hand, O doping contributes to an additional reduction in free energy during FA adsorption, C_α radical generation, and C_α radical homo‐coupling. On the other hand, O doping increases the local C_α radical concentration by forming strong O‐C bonds with C_α radicals. Furthermore, the enhanced FF conversion efficiency in the presence of O dopants was corroborated by a more pronounced reduction in free energy, which rationalizes the high HF selectivity of ZIS‐O. Collectively, these findings not only validate the crucial role of O doping in promoting FA‐to‐HF conversion but also deepen the mechanistic understanding of FA photoreforming reactions.

Funding

The National Natural Science Foundation of China (Grant No. 52572256, 22203046), and the Project of State Key Laboratory of Organic Electronics and Information Displays, the Tongji University Medicine‐X Interdisciplinary Research Initiative (Grant No. 2025‐0717‐YB‐01), Natural Science Basic Research Program of Shaanxi Province (Grant No. 2024JC‐DXWT‐03), CAS Pioneer Hundred Talents Program, Nanjing University of Posts and Telecommunications (Nos. GZR2023010003).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma73985‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.