Fe-Loaded Montmorillonite/TiO₂ Composite as a Promising Photocatalyst for Selective Conversion of Glucose to Formic Acid under Visible-Light Irradiation

Abstract

The development of efficient and inexpensive photocatalysts for the production of high-value chemicals from the photoreforming of biomass is a highly attractive strategy to establish the production of chemicals from sustainable resources. In this work, Fe-loaded montmorillonite/TiO₂ composite (Fe-Mt/TiO₂), pure TiO₂, Mt/TiO₂ and Mt/Fe-TiO₂ were fabricated and further utilized as photocatalysts for the production of formic acid from glucose under visible-light irradiation. Among the as-prepared composites, the Fe-Mt/TiO₂ exhibited the highest glucose conversion (83%), formic acid production (44%) and formic acid selectivity (53%). The effective heterojunction between Fe-Mt and TiO₂ is proposed to describe the superior photocatalytic activity of Fe-Mt/TiO_2, which effectively suppressed the recombination of the photogenerated electrons and holes during the reaction. Mechanism investigations suggested that the selective photocatalytic oxidation of glucose into formic acid by Fe-Mt/TiO₂ mainly occurred through an α-scission reaction pathway, driven by the main active species as ^•O₂⁻ and ¹O₂. The research findings in this work suggested that the Fe-Mt/TiO₂ composite can be applied as a low-cost, easy-to-prepare, reusable and selective photocatalyst for sustainable synthesis of high-value chemicals from biomass-derived substrates.

1. Introduction

Global energy demand is rapidly rising due to the growth of the population and economy, with a projected 50% increase by 2050. Currently, fossil-based energies have been dominantly utilized and contributed over 85% of total energy demand [1]. However, fossil resources are nonrenewable, and the widespread use of fossil fuel has led to significant environmental concerns. Consequently, there is an urgent need to explore sustainable alternative energy resources. Lignocellulosic biomass, abundant and renewable carbon resources [2], has been considered as a potential alternative to fossil resources for overcoming the energy and environmental concerns. Moreover, transitioning to the production of chemicals and fuels from lignocellulosic biomass demonstrates a promising step towards achieving carbon neutrality. Previous works showed that various fuels and chemicals including alkanes [3], organic acids [4,5], furans [6,7], and ethanol [8] can be synthesized from lignocellulosic biomass as feedstocks. It is worth mentioning that glucose, a simple model compound of biomass, has been considered as a potential alternative to lignocellulose, which has a complex structure, in the investigation of photoreforming mechanisms [9].

Formic acid is a versatile organic building block chemical and can be used in several applications such as food additives, bactericide and in various industries including dyes, textile, rubber and leather industries [10]. The global formic acid production reached one million metric tons (Mt) in 2019, and the market is anticipated to expand with an annual rate of approximately 3% [11,12]. Moreover, the global formic acid consumption in 2021 was around 870.03 kilotons, with a projected increase of around 4% over the next six years [13]. In general, two main processes have been intensively developed for the production of formic acid from lignocellulosic biomass, which are wet oxidation and catalytic wet oxidation. Typically, the addition of oxidant (H₂O₂) and alkaline salt (NaOH or KOH) is conducted in the wet oxidation process to activate the oxidation reaction and prevent the undesired decomposition of formic acid [14]. For catalytic wet oxidation, the lignocellulosic biomass was selectively oxidized into formic acid promoted by the catalyst with the gaseous O₂ supply [15]. More importantly, the wet oxidation and catalytic wet oxidation are energy consumption processes and mostly operated at high temperature (>150 °C) and pressure (>3 MPa of O₂) to achieve high yield [1]. In contrast to these energy-intensive processes, the photocatalytic conversion of lignocellulosic biomass has been considered as a promising route because it can possibly be driven by solar light at the evaluated temperature (<100 °C) and atmospheric pressure [5]. In the photocatalytic processes, the photon energy is utilized instead of thermal energy, demonstrating several advantages over the conventional methods such as mild reaction operations, high efficiency, cost-effectiveness and great cleanness [16].

TiO₂ is the most extensively employed photocatalyst due to its robust oxidizing capabilities, non-toxicity, low cost and chemical stability [17]. Thus, many research works have developed various methods for fabricating highly efficient TiO₂-based photocatalysts with tailored properties to enhance their photocatalytic performance such as TiO₂ nanoparticles [18], doped-TiO₂ with other elements (Fe-TiO₂ [19]) and combined TiO₂ with other semiconductor materials [20]. Because the photocatalytic reaction happens on the surface of the photocatalysts, the immobilization of TiO₂ particles on porous materials such as natural clays can also improve the photoactivity performance. Previous works demonstrated that several kinds of clays have been used to immobilize TiO₂ particles including sepiolite, bentonite, zeolite and montmorillonite [17,21].

Recently, montmorillonite (Mt)-based composites with TiO₂ have gained attention as promising photocatalysts. This is attributed to the properties of Mt such as abundant availability, substantial two-dimensional surface area, cost-effectiveness, low toxicity, remarkable chemical stability and environmentally friendly characteristics [21,22]. Furthermore, our previous works demonstrated that the incorporation of Fe³⁺ into Mt can enable the photocatalytic activity of Fe-Mt materials. This phenomenon was possibly caused by the creation of a mid-gap energy level through the d orbital of the doped iron (suggested by the molecule simulation) [22,23], which subsequently enhances the photocatalytic efficiency. Interestingly, a previous work also reported that the presence of Fe³⁺ or Fe²⁺ (sensitizing ions) in the photocatalytic materials can enhance the photocatalytic conversion of biomass into oxygenated hydrocarbon. Also, using mixture solvents of aqueous media and/or acetonitrile in the photocatalytic process can improve the selectivity oxidation of biomass [5]. Thus, the immobilization of TiO₂ particles on the Fe-Mt (acting as a photocatalyst support) can yield highly efficient Fe-Mt/TiO₂ photocatalysts due to the formation of Fe-Mt/TiO₂ heterostructure. This innovative material showed significant potential for diverse photocatalytic applications in addressing global problem challenges.

Herein, the Fe-Mt/TiO₂ composite was prepared and further applied as a photocatalyst for transformation of glucose to formic acid under visible-light illumination. Photocatalytic activity was determined in a mixture solvent of aqueous/acetonitrile solution. Additionally, the photocatalytic performance results were also compared to those achieved using pure TiO₂, Mt/TiO₂ and Mt/Fe-TiO₂. The improved light absorption capability and the heterostructure between Fe-Mt and TiO₂ contributed to enhance the photocatalytic performance. Furthermore, optimal parameters for the production of formic acid and possible mechanism pathways were also explored and discussed.

2. Materials and Methods

2.1. Materials and Chemicals

Iron (III) nitrate nonahydrate (Fe(NO₃)₃·9H₂O, 99%), hydrochloric acid (HCl, 35.0–37.0%), nitric acid (HNO₃, 70.0–71.0%) and sodium hydroxide (NaOH, 97%) were supplied by FUJIFILM Wako Pure Chemical Corp., Japan. Titanium (IV) chloride (TiCl₄, 99%) was bought from KISHIDA Chemical Co., Ltd., Japan. The Mt material ((Na_0.97Ca_0.08)^+1.13 (Si_7.68Al_0.32)(Al_2.94Fe^III_0.25Fe^II_0.03Mg_0.78)O₂₀(OH)₄^−1.13·nH₂O, Kunipia-F) with the cation exchange capacity (CEC) of 1.114 mmolg⁻¹ was supplied from Kunimine Industries Co., Ltd., Tokyo, Japan. All chemicals were utilized as received without further purification. All solutions in this work were prepared using deionized (DI) water with resistivity > 18.5 MΩcm⁻¹ (Synergy UV Water Purification System, Millipore, MA, USA) as a solvent.

2.2. Preparation of Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ Composites

The Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ composites were systematically prepared following the previous report with some modifications [23]. Typically, an acidified Mt (H-Mt) was synthesized by dispersing Mt powder (2 g) in 200 mL of DI water. Then, the suspension was adjusted to a pH of 2 using 0.1 M of HNO₃ or 0.01 M of NaOH and stirred at 500 rpm at room temperature for one hour. Then, the solid H-Mt was separated from an aqueous media by centrifugation, followed by washing with DI water and freeze-drying. Similarly, the Fe³⁺-doped Mt (Fe-Mt) sample was fabricated in a similar procedure, but the Fe(NO₃)₃·9H₂O with the concentration equivalent to 0.114 mmol Fe/gMt was added into the Mt suspension. All samples were ground and sieved (<107 µm) prior to further use.

To prepare the Ti(OH)₄/TiO₂ sol–gel precursor, the TiCl₄ was dropwise added into 4.5 mL of HCl (1 M) aqueous solution to obtain the final concentration of 0.83 M. Then, the solution was stirred at 500 rpm for one hour and then aged for 6 h at room temperature under ambient air to obtain a transparent solution. After that, the pH of the Ti(OH)₄/TiO₂ sol–gel solution was adjusted to a pH of 2 using NaOH aqueous solution. Also, the Fe-Ti(OH)₄/TiO₂ sol–gel was prepared following the identical procedure, but the Fe(NO₃)₃·9H₂O with stochiometric equivalence of 0.114 mmol Fe/g(Mt/TiO₂) was introduced into the Ti(OH)₄/TiO₂ solution before stirring and aging.

In order to synthesize Mt/TiO₂ by heterocoagulation, the suspension of Mt aqueous solutions (1% w/v) was firstly prepared, which was subsequently added with the Ti(OH)₄/TiO₂ sol–gel. Afterwards, the above suspension was stirred at 500 rpm for 30 min to obtain a homogenous dispersion and then thermally heated at 70 °C for 15 h by left standing in the oven. Later, the solid product was extracted from the liquid media by centrifugation, washed with DI water and freeze-dried at −40 °C and 20 Pa (Freeze-dryer, FDU-1200, EYELA, Bohemia, NY, USA). The preparation of Fe-Mt/TiO₂ and Mt/Fe-TiO₂ composites were also conducted using similar routes, but the Fe-Mt aqueous suspension (1% w/v) and the Fe-Ti(OH)₄/TiO₂ sol–gel were employed as starting precursors for Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples, respectively. Theoretically, the TiO₂ content in the Mt/TiO₂-based composites was approximately 30 wt%, consistent with the results from XRF analysis (Table S1). The procedure for preparation of pure TiO₂ was followed using the same procedure, using Ti(OH)₄/TiO₂ sol–gel as precursors without adding the Mt aqueous suspension. All as-synthesized samples were ground and sieved (<107 µm) prior to further use. The concentration of Fe³⁺ in the supernatant after preparation of all Fe³⁺-loaded samples was determined by inductive plasma–optical emission spectroscopy (ICP–OES, Perkin-Elmer 8500, Waltham, MA, USA). The Fe³⁺ concentration in the supernatant was lower than the detection limit of the ICP–OES, suggesting that stoichiometric equivalence of Fe³⁺ in all Fe³⁺-loaded samples is entirely incorporated in the solid matrixes.

2.3. Sample Characterization

The composition of Fe-Mt/TiO₂ and Mt/Fe-TiO₂ composites was examined by X-ray fluorescence spectroscopy (XRF) using an XRF spectrometer (Shimadzu-EDX800, Kyoto, Japan). The crystal phases of the samples were determined by X-ray diffraction (XRD) analysis using an Ultima IV X-ray diffractometer (Rigaku, Tokyo, Japan) with Cu Kα radiation operating at 40 kV and 40 mA and a scanning rate of 2°/min. The morphology and the elemental dispersion of the samples were examined by scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM–EDX, FlexSEM1000II and AZteclive lite FX, Hatachi, Tokyo, Japan). The optical absorption properties of the samples were investigated by a UV–vis diffuse reflectance spectroscopy (UV–vis DRS, UV-2540, Shimadzu, Japan), and BaSO₄ was used as a reference. Then, the obtained UV–vis DRS spectrum was employed to calculate the energy band gap (Eg) values of the samples following Tauc’s plot method [24]. A solid-state photoluminescence (PL) measurement was performed on a spectrofluorometer (JASCO F-6600, Tokyo, Japan) at room temperature with an excitation wavelength of 320 nm. Transient photocurrent and electrochemical impedance spectroscopy (EIS) measurements were conducted on a Solartron 1280C electrochemical test system (Solartron analytical, Leicester, UK) in 0.1 M of NaCl electrolyte solution (pH = 2 adjusted by using 0.1 M of HCl) to inspect photoelectrochemical properties of the samples. The as-prepared samples were coated onto a Pt sheet electrode (1 cm²) as the working electrode, whereas the Pt wire and Ag/AgCl (in 0.1 M of NaCl) are applied as the counter and the reference electrodes, respectively.

2.4. Photocatalytic Experiments

All experiments of the photocatalytic reaction were conducted in a sealed quartz flask (50 mL) under visible-light irradiation (500 W Xe lamp, 6258, USHIO, Tokyo, Japan) with the UV-cutoff filter (λ < 380 nm, ASAHI SPECTRA, Super Cold Filter 750) under magnetic stirring at 300 rpm. In a typical run, 25 g of glucose and 25 mg of the sample were put in the reactor with 25 mL of the mixture solvent (acetonitrile/water, 90% v/v). Prior to performing the reaction at a desired temperature, the reactor was sealed with the rubber septum. Then, the reaction was stirred at 300 rpm under dark condition for 30 min to allow the equilibrium of adsorption–desorption of glucose on the sample surface. At the specific reaction time, the solution in the reactor was taken and filtered (0.22 µm filter PTFE membrane). The oxidized products and remaining glucose concentrations were quantified by using a high-performance liquid chromatography system (HPLC, CO-2065, JASCO, Tokyo, Japan) equipped with an Aminex HPX-87H (BIO-RAD, Hercules, CA, USA) column and a refractive index detector (RI-2031 Plus, Tokyo, Japan) for analysis of sugar together with a UV–vis detector (UV-2075 Plus, JASCO, Tokyo, Japan) at 210 nm for analysis of possible organic acid products. The HPLC system was performed at 50 °C and a flow rate of mobile phase (5 mM of H₂SO₄ aqueous solution) of 0.6 mL·min⁻¹ in the isocratic mode. The glucose conversion and formic acid yield were calculated by the following formulas [25,26].

$$\mathrm{Glucose} \mathrm{conversion}\left(\%\right)=(1 - (\frac{ \mathrm{Concentration} \mathrm{of} \mathrm{glucose} \mathrm{after} \mathrm{reaction} \left(\mathrm{M}\right)}{\mathrm{Initial} \mathrm{concentration} \mathrm{of} \mathrm{glucose} \left(\mathrm{M}\right)}\left)\right)\times 100$$

(1)

$$\mathrm{Formic} \mathrm{acid} \mathrm{yield}\left(\%\right)=(\frac{ 1 \times \mathrm{Concentration} \mathrm{of} \mathrm{formic} \mathrm{acid} \mathrm{after} \mathrm{reaction} \left(\mathrm{M}\right)}{6 \times \mathrm{Initial} \mathrm{concentration} \mathrm{of} \mathrm{glucose} \left(\mathrm{M}\right) })\times 100$$

(2)

3. Results and Discussion

3.1. Sample Characterization

The XRD patterns of pure TiO₂, Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples are shown in Figure 1. The pure TiO₂ demonstrated the typical diffraction peaks at 25.4° and 39.7°, which are corresponded to the 101 and 004 planar planes of anatase TiO₂ (JCPDS 01-071-1168) [27], respectively. Notably, a broad peak of brookite TiO₂ (121) at 2θ of 30.3° was also observed in the XRD pattern of pure TiO₂ (JCPDS 01-075-1582) [27] and no rutile peaks. By using the Rietveld refinement method, the contents of anatase and brookite TiO₂ in pure TiO₂ sample were found to be 93% and 7%, respectively. The XRD patterns of Mt-based samples show a strong peak around 2θ = 6.4–6.5° due to the d(001) basal spacing reflection of the montmorillonite. Additionally, diffraction peaks at 19.8° and 35.2° attributed to the 003 and 101 planar planes of montmorillonite (JCPDS 13-0259) [27], respectively, were also found in the XRD patterns. It is important to note that a single layer of Mt has an interlayer space of 0.96 nm (from the d value of the 001 planar plane). Consequently, following the exclusion of a single Mt layer (Scheme S1), the interlayer spaces of Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ were determined to be 0.41 nm, 0.43 nm and 0.44 nm, respectively. This phenomenon could be attributed to the smaller size of TiO₂ particles formed on the surface of each Mt layer in the Fe-Mt/TiO₂ compared to that of Mt/Fe-TiO₂. No diffraction peaks corresponding to iron oxide compounds (Fe₂O₃ or Fe₃O₄) were found in the XRD patterns of Fe-loaded samples, suggested that the Fe ion was incorporated in the Mt layer or TiO₂ crystal structure. Interestingly, the diffraction peak of 110 rutile TiO₂ (JCPDS 01-076-0324) [27] were found at 28.2°–28.4° together with a peak of anatase TiO₂ (101) in XRD patterns of Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples (as seen in Figure S1). The contents of anatase and rutile TiO₂ were estimated to be 52% and 48% for Mt/TiO₂, 64% and 36% for Fe-Mt/TiO₂ and 84% and 16% for Mt/Fe-TiO₂, respectively. This result suggested that the presence of Fe³⁺ in both Fe-Mt and Fe-TiO₂ has a strong impact on the formation of anatase phase, possibly due to the repulsion between the positively charged Fe³⁺- and Ti⁴⁺-containing groups linked to Mt layers by interacting with their negative charge [23].

The morphology of Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples are displayed in Figure 2a–d. As seen in Figure 2a, the morphology of pure TiO₂ is nearly spherical-shaped, with a size ranging from 5 µm to 20 µm, which is similar morphology to the TiO₂ spherical shape prepared by hydrolysis of TiCl₄ solution with HCl [28]. Interestingly, the size of TiO₂ particles was reduced when incorporated into the Mt layer of Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples. As seen in TEM images of Fe-Mt/TiO₂ in Figure S2, the particle size of TiO₂ particles on the Mt layer were in the range of an estimated 7–19 nm. This result suggests that Mt can serve as the supporting material for TiO₂ particle dispersion to mitigate the particle agglomeration and further enhance the photocatalytic activity. SEM–EDX elemental mapping analysis of Fe-Mt/TiO₂ (Figure S3) and Mt/Fe-TiO₂ (Figure S4) samples demonstrated that the Ti, Fe, Mg, Al, Si and O existed with good dispersion, which can further suggest that the TiO₂ nanoparticles intensively covered the surface of the Mt supporting materials. Moreover, the uniform dispersion of Fe signals in the Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples implied the good distribution of Fe throughout the Mt phase, TiO₂ phase and the Mt-TiO₂ interface.

The UV–vis DRS was utilized to examine the optical properties of the pure TiO₂, Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples. As seen in Figure 3a, the pure TiO₂ and Mt/TiO₂ samples exhibited UV-region adsorption edges, whereas Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples showed a slight shift towards the visible region, potentially enhancing the photocatalytic activity. To estimate the band gap energy (Eg) of all samples, the data were processed by Tauc’s equation (Equation (3)) [24,29].

αhv = A (hv − Eg)^n/2

(3)

where α, h, v, A and Eg are the absorption coefficient, Plank constant, the frequency of light, a proportionality constant and the band gap energy, respectively. The “n” value is 1 for the direct transition semiconductor and 4 for indirect transition of the semiconductor. Figure 3b shows the Tauc plot of all samples, and the Eg values were about 3.20, 3.31, 3.07 and 3.01 for the pure TiO₂, Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples, respectively. Additionally, reversed double-beam photoacoustic spectroscopy (RDB-PAS) [30] was also utilized to estimate the Eg of Fe-Mt/TiO₂ and Mt/Fe-TiO₂, and the obtained results (Figure S5) were comparable to those estimated by Tauc’s equation (3.06 eV for Fe-Mt/TiO₂ and 3.01 eV for Mt/Fe-TiO₂). The higher Eg of Mt/TiO₂ than that of pure TiO₂ may be due to the quantum confinement effect, consistent with previous works where the Eg values of TiO₂/montmorillonite composites were greater than those of TiO₂ prepared in their works [31,32]. Interestingly, both Fe-Mt/TiO₂ and Mt/Fe-TiO₂ samples exhibited smaller Eg values than that of pure TiO₂ because the presence of Fe³⁺ can generate a new Fermi level between the energy band of Mt and TiO₂ [33,34].

Photoluminescence (PL) analysis was conducted to study the emission properties and the migration and recombination of the photogenerated electron–hole pairs. The PL intensity of Mt/TiO₂ was significantly decreased compared to that of the pure TiO₂ (Figure 3c), which was possibly due to the improved suppression of photogenerated charge recombination by the Mt supporting material. Furthermore, PL intensities of Fe-Mt/TiO₂ and Mt/Fe-TiO₂ were lower than those of the pure TiO₂, Mt/TiO₂, suggesting that the presence of Fe³⁺ in Mt and TiO₂ in both samples can mitigate recombination of the photogenerated electron–hole pairs. Notably, the Fe-Mt/TiO₂ exhibited lower PL intensity than that of Mt/Fe-TiO₂, indicating that the Fe-Mt component played an important role to migrate and suppress the photogenerated charge carriers. A previous work reported that incorporation of Fe³⁺ in clay is able to demonstrate the photocatalytic activity, suggesting that Fe-Mt can serve as a photocatalytic supporting material [23,34]. Thus, the heterojunction between Fe-Mt and TiO₂ could be constructed in Fe-Mt/TiO₂ and further facilitate the migration and suppression of the photogenerated charge carriers and improve the photocatalytic performance. Conversely, there should be no formation of a Mt/Fe- heterojunction in the Mt/Fe-TiO₂, as unmodified Mt was employed as a non-photoactive supporting material for the dispersion of Fe-TiO₂ particles.

The transient photocurrent responses under chopped irradiation and EIS were further investigated to study the migration and generation of electron in all samples. As seen in Figure 4a, the photocurrent density exhibited a gradual increase in the order of Fe-Mt/TiO₂ > Mt/Fe-TiO₂ > Mt/TiO₂ > pure TiO₂. This indicates improved charge migration in the composite materials compared to pure TiO₂ and enhanced charge transfer in Fe-Mt/TiO₂ compared to Mt/Fe-TiO₂. In addition, in EIS Nyquist diagrams (Figure 4b), the Mt-based composites displayed a smaller arc radius than pure TiO₂, which suggested a lower charge transfer resistance (Table S2) than that of pure TiO₂. Similarly, the Fe-Mt/TiO₂ has a smaller arc radius than that of Mt/Fe-TiO₂, indicating that the Fe-Mt/TiO₂ has weaker resistance for charge transportation than Mt/Fe-TiO₂ [26].

3.2. Photocatalytic Performance on Formic Aicd Production from Glucose

The photocatalytic activities of the pure TiO₂, Mt/TiO₂, Fe-Mt/TiO₂ and Mt/Fe-TiO₂ under visible-light irradiation were evaluated for the conversion of glucose to formic acid in mixture solvent (acetonitrile/water, 90 vol%). The glucose conversion, formic acid yield and formic acid selectivity after a 2 h reaction at 50 °C are reported in Figure 5a. It is worth mentioning that pristine Mt demonstrated no photocatalytic oxidation of glucose. Clearly, the incorporation of TiO₂ particles into the Mt layer in Mt/TiO₂ led to an enhanced photocatalytic performance. The Mt/TiO₂ exhibited 60% glucose conversion and 13% formic acid yield, surpassing the performance of pristine TiO₂, which achieved 58% of glucose conversion and 7% of formic acid production. This enhancement can be attributed to the smaller size of TiO₂ particles and the better photogenerated charge transportation within the Mt/TiO₂ composite compared to pure TiO₂. Interestingly, the Fe-Mt/TiO₂ composite demonstrated superior photocatalytic performance, yielding 83% glucose conversion and 44% formic acid yield compared to Mt/Fe-TiO₂, which produced 78% glucose conversion and 38% formic acid production. In addition, the formic acid selectivity was increased following the order Fe-Mt/TiO₂ (53%) > Mt/Fe-TiO₂ (49%) >Mt/TiO₂ (21%) > TiO₂ (12%), which implied that the incorporation of Fe³⁺ in Mt and TiO₂ components can also boost the photocatalytic performance and the formic acid selectivity, which is consistent with the UV–vis DRS, PL, photocurrent and EIS results. The better photocatalytic performance of Fe-Mt/TiO₂, in comparison to Mt/Fe-TiO₂, is mainly attributed to the presence of Fe³⁺ in the Mt layer, enabling the formation of an Fe-Mt/TiO₂ heterojunction in the Fe-Mt/TiO₂ composite. Conversely, in Mt/Fe-TiO₂, the Mt component serves merely as a support without photocatalytic contribution. The possible intermedia products during the photocatalytic glucose conversion by Fe-Mt/TiO₂ were investigated using HPLC analysis based on previous research. As seen in Figure 5b,c, the HPLC chromatograms obtained from the refractive index (RI) detector revealed the presence of glucose and arabinose, while the HPLC chromatograms from the UV detector detected formic acid after a 30 min reaction. As the photocatalytic reaction proceeded, the peak intensity of glucose tended to decrease, whereas the peak intensities of arabinose and formic acid continuously increased along with the reaction time. Based on the HPLC results, arabinose was concluded to be the main product intermediated in the photocatalytic glucose conversion to formic acid by the Fe-Mt/TiO₂ composite, which is similar to the previous work [35].

According to the results mentioned earlier, it is clear that the Fe-Mt/TiO₂ composite produced the highest amount of formic acid. As a result, additional experiments were conducted using Fe-Mt/TiO₂ in the photocatalytic conversion of glucose to investigate the optimal conditions for formic acid production. Figure 6a demonstrates the influence of catalyst loading on the photocatalytic glucose conversion to formic acid. The glucose conversion and formic acid yield were significantly increased from 62% to 74% and 18% to 25%, respectively, when the catalyst loading was increased from 0.25 to 0.5 gL⁻¹. The maximum glucose conversion (83%) and formic acid yield (44%) were obtained using the catalyst loading of 2.0 gL⁻¹. A further increase in catalyst loading for 2.0 to 4.0 gL⁻¹ led to less improvement of glucose conversion (85%) and a slight decrease in the formic acid production (38%), which could be caused by the decomposition of formic acid with the excess amount of catalysts [1].

Next, the effect of the reaction temperature on the photocatalytic conversion of glucose to formic acid by Fe-Mt/TiO₂ was investigated at temperatures ranging from 30 to 70 °C. As seen in Figure 6b, the glucose conversion was much improved, from 22% at 30 °C to 92% at 60 °C. After the reaction temperature was further increased to 70 °C, the glucose conversion slightly increased to almost 100%. Interestingly, the formic acid yield was increased from 32% at 30 °C to reach the maximum value of 44% at 50 °C. Additionally, the formic acid yield had a negligible improvement when the reaction temperature was further increased to 60 °C and 70 °C, resulting in a reduction in formic acid production by 41% and 40%, respectively. As seen in Figure S6, gluconic acid and an unknown product were found in the HPLC chromatogram of the solution from the reaction at 70 °C. It is possible that the increase in reaction temperature can enhance the kinetic of the reaction for the glucose conversion and promote the side reactions to produce gluconic acid and other organic acid compounds. The decrease in formic acid yield at higher temperature (>50 °C) may be attributed to the exothermic nature of the oxidation of glucose to formic acid. Thermodynamically, this reaction is more favorable at lower temperatures, leading to reduced formic acid formation as the temperature increases [36].

The effect of solvent composition on the photocatalytic glucose conversion to formic acid by Fe-Mt/TiO₂ was also investigated in three conditions, altering the ratio of water (W) to acetonitrile (ACN) as 100:0, 50:50 and 10:90 (%vol). As illustrated in Figure S7, increasing the acetonitrile content in the mixture solvent resulted in an increase in glucose conversion and formic acid production. In the 10: 90 (W: ACN vol%), glucose conversion and formic acid formation reached 83% and 44%, respectively, compared to 13% glucose conversion and 4% formic acid production in pure water. The better formic acid selectivity after increasing the ACN content in the solvent could be attributed to the low water content, leading to a lower concentration of highly unselective ^•OH radicals, which is similar to the previous works [5,26,37].

The recycling test of the Fe-Mt/TiO₂ on photocatalytic glucose conversion to formic acid was evaluated and is shown in Figure 6c. About 53% of formic acid yield and 83% glucose conversion were obtained in the first cycle. After water-washing and -drying of the spent Fe-Mt/TiO₂ composites, the formic acid yield and glucose conversion slightly decreased to 40% and 80%, respectively. The minor reduction in formic acid production and glucose conversion could be attributed to the loss of some photocatalyst powder during the reusability testing procedure.

3.3. Possible Reaction Pathways

To further investigate the possible active species involved the photocatalytic glucose conversion by Fe-Mt/TiO₂, scavenging experiments [36] were conducted employing scavengers such as tryptophan (TRP,) para-benzoquinone (BQ), methanol and isopropanol (IPA) to quench ¹O₂, ^•O₂⁻, h⁺ and ^•OH, respectively. Figure 6d displays the effect of each scavenger on the photocatalytic conversion of glucose to formic acid. The photocatalytic reaction with methanol and IPA exhibited comparable glucose conversion (~80%) and formic acid production (~40%) to those of reaction without scavenging reagents, suggesting that only a few h⁺ and ^•OH existed in the reaction. However, the reaction in TRP and BQ delivered lower glucose conversion (41% in TRP and 37% in BQ) and formic acid yield (18% in TRP and 8% in BQ), indicating that ¹O₂ and ^•O₂⁻ were produced in the reaction under visible-light illumination and played the most crucial roles in the selective glucose conversion to formic acid. Previous works reported that ¹O₂ and ^•O₂⁻ were milder oxidants and resulted in better selective oxidation of organic compounds than ^•OH. Compared to the scavenging experiments of pure TiO₂ (Figure S8), the reaction in BQ and IPA gave lower glucose conversion (22% in IPA and 38% in BQ) and formic acid yield (2% in IPA and 4 % in BQ) than the reaction without. Thus, ^•OH and^•O₂⁻ radicals were produced in the reaction when TiO₂ was used as a photocatalyst and played important roles in the reaction. Thus, the ¹O₂ and ^•O₂⁻ in our photocatalytic reaction catalyzed by Fe-Mt/TiO₂ can effectively oxidize glucose while avoiding the overoxidation, resulting in the high yield of formic acid production [38]. Additionally, the high photocatalytic conversion of glucose with good selectivity may be attributed to the presence of ACN in the mixture solvent, which could potentially prolong the lifetime of ¹O₂ in the reaction [39].

The analysis of the electronic band position of Fe-Mt/TiO₂ is important to understand the mechanism of glucose conversion. The Mott–Schottky measurements (Figure S9) were conducted to estimate flat band potentials of TiO₂ and Fe-Mt, which are about −0.69 and −0.46 V vs. the normal hydrogen electrode (NHE), respectively [40]. Additionally, the negative slope in the Mott–Schottky plots in both TiO₂ and Fe-Mt suggested that they are n-type semiconductors. Generally, the conduction band potential value of n-type semiconductors is more negative, about 0.1 V, than the flat band potentials. Thus, estimated conduction band (CB) potentials of TiO₂ and Fe-Mt were to be −0.79 and −0.56 V, respectively. In addition, the valence band (VB) was calculated using the CB (Mott–Schottky measurements) and Eg obtained from Tauc’s plot (Figure 3b). According to the formula of E_VB = E_CB + Eg, the potential of the VB of TiO₂ and Fe-Mt were determined to be 2.41 and 2.51 V, respectively. Based on the above results and previous work [22,23], the possible electronic band structure of Fe-Mt/TiO₂ was proposed (Figure 7), revealing that the VB of TiO₂ was more negative than that of Fe-Mt (2.51 V), whereas the CB of TiO₂ was more positive than that of Fe-Mt (−0.56 V). Notably, the potential of Fe-mid-gap states in Fe-Mt at 1.09 eV (estimated from the previous work) was included in the proposed electronic band structure [22,23]. Thus, the Z-scheme Fe-Mt/TiO₂ heterojunction could be constructed and promote the separation of the photogenerated charge carries in the material. Under light irradiation, the Fe-Mt and TiO₂ can absorb photons, and the photogenerated hole and photogenerated electron were subsequently created in the CB and VB of both materials. Simultaneously, the electrons from the VB of Fe-Mt were excited to the Fe-mid-gap states and subsequently migrated to the VB of TiO₂ because the VB of TiO₂ was relatively more positive than that of Fe-Mt. Then, the recombination of photogenerated electrons from the Fe-mid-gap states in Fe-Mt and photogenerated hole in VB of TiO₂ may prolong the lifetime of photogenerated electrons in the CB of TiO₂. Thus, the photoexcited electrons accumulating on the CB of TiO₂ can transform O₂ to ^•O₂⁻ because the E_CB of TiO₂ was more negative than −0.33 eV (O₂/^•O₂⁻ vs. NHE). Also, the ¹O₂ was generated through the reaction between h⁺ and ^•O₂⁻ in the reaction. Subsequently, the generated ^•O₂⁻ and ¹O₂ radicals oxidized glucose into arabinose and further into formic acid, exhibiting relatively high selectivity in the process. Interestingly, the VB of Fe-Mt was more positive than that of the redox potential of ^•OH production (H₂O/^•OH vs. NHE, 2.27 eV), resulting in a possible generation of ^•OH in the reaction. The proposed electronic band structure and the possible formation of ¹O₂ and^•O₂⁻ as the main reactive species align with the scavenging test results. These findings can support that the ¹O₂ and ^•O₂⁻ are major reactive radicals for the conversion of glucose, whereas ^•OH exhibited less effect on the reaction.

Possible photocatalytic oxidation pathways of glucose to formic acid catalyzed by the Fe-Mt/TiO₂ composite are proposed in Figure 7. Because arabinose was found to be the primary intermediate product, the oxidation of glucose could mainly occur through the C1-C2 (α-scission) bond cleavage, driven by the activity of ^•O₂⁻ and ¹O₂ radicals. In this process, the initial breakage of the C1-C2 (α-scission) bond in glucose yielded one equivalent of arabinose and one formic acid. This cycle was then repeated until all aldose was completely converted into formic acid [35].

4. Conclusions

In this work, Fe-loaded montmorillonite/TiO₂ composite (Fe-Mt/TiO₂), pure TiO₂, Mt/TiO₂ and Mt/Fe-TiO₂ were systematically prepared and subsequently utilized as photocatalysts to produce formic acid via glucose conversion under visible-light irradiation. The Fe-Mt/TiO₂ exhibited the highest photocatalytic conversion of glucose (83%) and formic acid yield (44%) after 2 h at 50 °C. The better photocatalytic performance of Fe-Mt/TiO₂ was possibly due to the suppression of the photogenerated charge recombination, suggested by the PL, photocurrent and EIS results. Additionally, the Fe-Mt component within the Fe-Mt/TiO₂ composite functioned as a photocatalyst support, facilitating the possible formation of a Z-scheme Fe-Mt/TiO₂ heterojunction. Mechanism studies showed that the Fe-Mt/TiO₂ material promoted the selective photocatalytic conversion of glucose to formic acid via an α-scission reaction route with arabinose as an intermediate, driven by the ^•O₂⁻ and ¹O₂ as major reactive species. This work demonstrates the possible potential of clay-based materials for inexpensive and visible-light-driven photocatalytic conversion of biomass-derived substances into value-added chemicals. Additional investigations are required to find the optimal contents of Fe and TiO₂ in Fe-Mt/TiO₂, identify suitable modification methods, and explore milder reaction conditions for formic acid production and more environmentally friendly solvents than acetonitrile.