Interplay Between Ionic Liquids, Kolbe Chemistry, and 2D Photocatalyst Supports in Aqueous CO₂ Photoreduction over Pd/TiO₂ and Pd/g-C₃N₄

Abstract

The photocatalytic reduction of CO₂ in aqueous media offers a sustainable route for solar-to-fuel conversion, yet remains challenged by CO₂’s thermodynamic stability and kinetic inertness, low solubility, and competitive hydrogen evolution. Here, we investigate the interplay between ionic liquids (ILs), photocatalyst supports, and additive composition in directing product selectivity among CO, CH₄, and H₂. Using imidazolium acetate as a benchmark, we demonstrate that ILs not only pre-activate CO₂ but can also undergo decomposition pathways under illumination, notably Kolbe-type reactions leading to methane formation from acetate rather than from CO₂. Comparative studies of Pd-decorated TiO₂ and g-C₃N₄ nanosheets reveal distinct behaviors driven by their interfacial interactions with the imidazolim-based ionic liquid: weak interaction with TiO₂ strongly promotes hydrogen evolution, whereas strong coupling with g-C₃N₄ synergizes with C₁C₄ImOAc to trigger acetate-derived Kolbe reactivity. The systematic evaluation of alternative salts confirms the determinant role of anion basicity and medium-pH-basic anions facilitate CO₂ activation, whereas weakly basic or non-coordinating anions favor water splitting. Overall, these results clarify the dual role of ionic liquids as both CO₂ activators and sacrificial agents, and highlight design principles for improving product selectivity and efficiency in aqueous CO₂ photoreduction systems.

1. Introduction

The ever-increasing levels of atmospheric CO₂ and the global demand for sustainable energy solutions have driven intense research into the direct conversion of CO₂ into value-added fuels using solar energy [1]. Among the available strategies—photovoltaic systems, photoelectrochemical cells, and photocatalysis—the photocatalytic reduction of CO₂ stands out as most promising, as it does not rely on carbon-based external energy sources [2,3,4,5]. However, its efficiency remains limited due to three fundamental challenges: the thermodynamic stability and kinetic inertness of CO₂, and its low solubility in reaction (aqueous) media. The first one-electron reduction of CO₂ to CO₂•⁻ is both sluggish and energetically demanding (E° = −1.90 V vs. NHE at pH 7), often giving way to the competitive hydrogen evolution reaction (HER).

To overcome these limitations, imidazolium-based ionic liquids (Im-ILs) have emerged as multifunctional media that promote CO₂ activation and selectivity [6,7]. Their ability to absorb CO₂, lower reduction overpotentials, suppress HER, and stabilize reactive intermediates makes them highly valuable for photocatalysis [8,9]. Notably, certain basic anions in Im-ILs can facilitate carbene formation via deprotonation at the C2 position, enabling the generation of CO₂-activating zwitterionic species [10,11,12,13,14,15,16].

Our previous work has shown that among various Im-ILs, only C₁C₄ImOAc is capable of pre-activating CO₂ in bare or diluted media and under ambient pressure, a prerequisite for selective CO formation under our reaction conditions. Moreover, bare Im-ILs enhance charge separation in photocatalysts by inhibiting electron–hole recombination, thereby prolonging the lifetime of photoinduced carriers. Im-ILs can even function as organocatalysts and photosensitizers, contributing to electron transfer directly, as evidenced by correlations between CO production and both UV–vis absorption properties and Kamlet–Taft β parameters [14]. In highly diluted systems (xH₂O > 0.99), the ionic nature of the Im-ILs evolves from contact ion pairs (CIP) to solvent-separated ion pairs (SIP) [17,18,19], modifying their reactivity from CO₂ activation to a solvent one [14]. These solvent–anion interactions—quantified by Kamlet–Taft β values—are crucial for modulating H⁺ availability, thereby influencing CO₂ reduction pathways, particularly via hydrogen carbonate intermediates.

In this context, we report herein the role of Im-ILs in diluted aqueous media and assess their effect in the presence of heterogeneous photocatalysts, which, to our knowledge, has never been reported. Specifically, we explore the photocatalytic CO₂ reduction using nanosheet (NS) architectures, such as NS-TiO₂/Pd and NS-g-C₃N₄/Pd, selected for their high surface area, exposed active sites, short charge migration distances [13,20,21,22], and possible surface doping [23]. By maintaining constant morphology and cocatalyst (Pd), we focus solely on the influence of the support material and the Im-IL environment.

2. Results

2.1. Synthesis and Characterization of Photocatalyst Supports

2.1.1. NS-TiO₂ and NS-TiO₂/Pd

Rectangular TiO₂ nanosheets (NS-TiO₂) were synthesized via a hydrothermal route using hydrofluoric acid, following procedures reported in the literature [22]. The obtained material exhibited high crystallinity as anatase, as confirmed by X-ray diffraction (XRD) analysis (Figure 1a), with diffraction peaks corresponding to the anatase phase (JCPDS 21-1272). Transmission electron microscopy (TEM) revealed the rectangular nanoplatelet morphology (Figure 1b), and the Brunauer–Emmett–Teller (BET) surface area was measured at 91 m²·g⁻¹, significantly larger than that of commercial P25, indicating the formation of high-surface-area anatase nanosheets.

In situ deposition of palladium nanoparticles was then carried out on NS-TiO₂ in aqueous medium. While the low Pd loading (~1 wt%) precluded the observation of Pd reflections in the XRD pattern (Figure S1), HRTEM imaging demonstrated uniform dispersion of Pd nanoparticles across the nanosheet surfaces (Figure 2). The mean particle size was in the 4–5 nm range (Figure 2), and the lattice spacing of 0.47 nm, observed along the top and bottom facets, was consistent with the (002) planes of anatase TiO₂. These results confirm the successful preparation of Pd-decorated TiO₂ nanosheets (NS-TiO₂/Pd), combining high surface area with well-dispersed metallic cocatalyst nanoparticles.

2.1.2. NS-g-C₃N₄ and NS-g-C₃N₄/Pd

Graphitic carbon nitride (g-C₃N₄), an n-type polymeric semiconductor with appealing optical and structural properties [24,25,26], was prepared in nanosheet form (NS-g-C₃N₄) using melamine as a precursor and ammonium chloride as a bubble-forming agent, as reported in our previous work [15]. The high-temperature condensation of melamine yielded exfoliated nanosheets with a porous structure, as confirmed by TEM (Figure S2b). XRD analysis revealed the characteristic reflections at 13.1° and 27.4°, corresponding, respectively, to the in-plane (100) structure and the interlayer (002) stacking (JCPDS 87-1526) (Figure S2a). The weakening of the (100) peak indicated a partial disruption of interlayer interactions, consistent with the exfoliated morphology observed by microscopy.

The synthesis yielded NS-g-C₃N₄ as a yellow powder with 13.2% efficiency, and the Brunauer–Emmett–Teller (BET) surface area was measured at 36 m²·g⁻¹. The adsorption isotherm displays a Type IV(a) profile, according to the IUPAC classification, which is characteristic of mesoporous materials (Figure 3). At intermediate pressures (approximately 0.4 < P/P₀ < 0.8), a noticeable hysteresis loop appears between the adsorption and desorption branches, suggesting that the material possesses mesopores with a slit-like or cylindrical geometry, depending on the specific shape of the loop (commonly H3 or H4 types). At high relative pressures (P/P₀ > 0.8), the sharp increase in the adsorbed volume reflects the filling of larger mesopores and interparticle voids. Overall, the isotherm confirms that the sample exhibits a mesoporous structure with a significant surface area and well-developed pore network, consistent with materials such as graphitic carbon nitride (g-C₃N₄).

The deposition of Pd nanoparticles onto the g-C₃N₄ nanosheets was achieved in situ (NS-g-C₃N₄/Pd). XRD patterns of the composite showed only the characteristic g-C₃N₄ reflections, with no distinct Pd peaks, reflecting the low Pd content (Figure S3a). However, STEM and TEM images combined with EDX mapping confirmed the presence of Pd nanoparticles homogeneously dispersed on the nanosheets (Figure S3b–d). The Pd particle size distribution was narrow, with an average diameter of ~2 nm. Elemental analysis indicated a Pd content of 2.74 wt%, in agreement with the target loading. These features underline the successful preparation of NS-g-C₃N₄/Pd, where the two-dimensional semiconductor sheets served as a high-surface-area support for finely dispersed Pd nanoparticles.

2.2. Photocatalytic Reduction of CO₂

The dataset presented in

Table 1. Photocatalytic data in the presence of ionic liquid or salts diluted in H₂O *.

Entry	Photocatalyst + Additive	CO (μmol·h−1)	CH4 (μmol·h−1)	H2 (μmol·gPd−1·h−1)	pH
1	C1C4ImOAc	0.44	0	0	5.53
2	NS-g-C3N4/Pd + C1C4ImOAc	0	0.38	0	5.53
3	NS-g-C3N4/Pd +C1C4ImOAc (Argon)	0	0.12	0	5.53
4	NS-g-C3N4/Pd + NaOAc (Argon)	0	0.03	0	8.15
5	NS-g-C3N4/Pd + HOAc (Argon)	0	5.41	0	2.68
6	NS-TiO2/Pd	0.13	0	301	5.34
7	NS-TiO2/Pd + C1C4ImOAc	0.13	0.28	298	5.53
8	NS-TiO2/Pd + C1C4ImOAc (Argon)	0.25	0.50	416	5.53
9	NS-TiO2/Pd + NaOAc	0	1.89	269	8.15
10	NS-TiO2/Pd + C1C4ImCF3CO2	0	0	449	3.80
11	NS-TiO2/Pd + KHCO3	0	0	0	8.34
12	NS-TiO2/Pd + Na2CO3	0	0	0	11.70
13	NS-TiO2/Pd + CholineCl	0	0	1394	8.12
14	NS-TiO2/Pd + CholineNTf2	0	0	2867	8.12
15	NS-TiO2/Pd + LiNTf2	0	0	3515	8.11

* Reaction conditions: 0.4 mmol IL + 3 mL H₂O + 6 mg photocatalyst + 4 h UV-vis + CO₂, unless noted as Argon.

provides valuable insights into product selectivity during photocatalytic experiments in the presence of both photocatalysts (NS-g-C₃N₄/Pd & NS-TiO₂/Pd) and C1C4ImOAc or salts diluted in H₂O.

Three major trends can be extracted.

• The central role of imidazolium acetate (C1C4ImOAc): In the absence of a heterogeneous photocatalyst, C1C4ImOAc alone produces CO selectively (0.44 μmol·h⁻¹), consistent with its demonstrated ability to act as a homogeneous photocatalyst [10,11,12,13,14,15,16]. This confirms that the imidazolium cation functions as a photosensitizer while the acetate anion activates the solvent, leading to alkylcarbonate or bicarbonate intermediates [14]. When combined with NS-g-C₃N₄/Pd, the selectivity shifts from CO toward CH₄ (0.38 μmol·h⁻¹), underlining that the coexistence of heterogeneous and homogeneous pathways can alter the reduction mechanism (entry 2). In this hybrid system, the pre-activated C1C4Im–CO₂ species may be reduced further than CO, thereby promoting C–H bond formation. The results obtained under argon confirm that the two phenomena take part in CH₄ production, since its production decreases drastically but does not become zero (entry 3). Furthermore, it appears that the protic character of the cation or the presence of a proton due to acid conditions strongly enhances CH₄ production (entries 4–5).
• The decisive influence of photocatalyst composition: A striking contrast emerges between NS-g-C₃N₄/Pd and NS-TiO₂/Pd. The former, when combined with C1C4ImOAc, supports CO₂ reduction but with altered selectivity (favoring CH₄ over CO) (entry 2). The latter, however, overwhelmingly drives hydrogen evolution (entries 6 and 7), with H₂ yields exceeding several thousand μmol·g_Pd⁻¹·h⁻¹ in the presence of salts such as CholineNTf₂ or LiNTf₂ (entries 14–15). This divergence illustrates that TiO₂ strongly promotes proton reduction, while g-C₃N₄, in synergy with IL-mediated pre-activation, can direct electrons toward C–C bond chemistry and deeper CO₂ reduction pathways. Surprisingly, under argon and in the presence of C1C4ImOAc, NS-TiO₂/Pd affords higher amounts of CO, CH₄, and H_2, while H₂ remains the dominant product (entry 8). No CO, CH₄, and H₂ are produced in the absence of palladium (Table S1, entries S5–S7).
• The impact of pH and additive nature: The correlation between solution pH, anion basicity, and selectivity is particularly revealing. At neutral to slightly acidic pH (≈5–6), imidazolium acetate affords CO or CH₄, whereas alkaline conditions (>8) are detrimental: both KHCO₃ and Na₂CO₃ completely suppress CO₂ reduction and H₂ production (entries 11 and 12), in line with the accumulation of inactive carbonate species. Salts such as NaOAc, operating around pH 8.1, bias the reaction toward methane (1.89 μmol·h⁻¹) but with substantial co-production of H₂ (entry 9), suggesting that basic acetate stabilizes pathways leading to C–H bond formation but destabilizes selective CO formation. Non-coordinating or weakly basic anions (NTf₂⁻ in CholineNTf₂ or LiNTf₂) promote massive hydrogen evolution, confirming that only sufficiently basic anions can activate CO₂. These observations are in full agreement with the mechanistic picture outlined in [14], where the Kamlet–Taft β parameter (anion basicity) was shown to positively correlate with CO₂ reduction efficiency. No CO, CH_4, and H₂ are produced in the absence of a photocatalyst (Table S1, entries S1–S4).

3. Discussion

Both supports—NS-TiO₂ and NS-g-C₃N₄—were successfully synthesized with controlled morphology and surface properties. NS-TiO₂ exhibits high crystallinity, large specific surface area, and well-defined anatase nanoplatelets, while NS-g-C₃N₄ displays a porous, exfoliated layered structure with weakened interlayer interactions. The deposition of Pd nanoparticles was effective in both systems, yielding nanoscale particles uniformly distributed on the support surfaces. However, the Pd content and average particle size differed: ~1 wt% Pd with slightly larger nanoparticles (4–5 nm) on NS-TiO₂ versus 2.74 wt% Pd with narrowly distributed 2 nm nanoparticles on NS-g-C₃N₄. These structural differences are expected to significantly influence photocatalytic performance, with NS-TiO₂/Pd favoring H₂ evolution due to its strong affinity for proton reduction, and NS-g-C₃N₄/Pd promoting selective CO₂ activation pathways in synergy with ionic liquids.

The photocatalytic behavior of diluted imidazolium-based ionic liquids (ILs) in aqueous solution was first evaluated in the absence of heterogeneous catalysts. In line with previous findings, selective CO production was observed with C1C4ImOAc (0.44 μmol·h⁻¹), significantly higher than with other imidazolium salts such as C1C4ImO₂CF₃, C1C4ImBF₄, or C1C4ImCl. This trend directly correlates with the Kamlet–Taft β parameter of the anion: the acetate anion, with its high hydrogen-bond basicity, strongly interacts with water molecules and facilitates CO₂ activation, leading to the highest CO yields. Such results are consistent with the conclusions that both anion basicity and solution pH are critical determinants of CO₂ reduction efficiency. When Pd-loaded nanosheet photocatalysts were introduced, the product distribution shifted significantly. With NS-g-C₃N₄/Pd in the presence of C1C4ImOAc, methane was selectively generated instead of CO. However, isotope-labeling experiments with ¹³CO₂ revealed that the detected CH₄ did not originate from CO₂ reduction but possibly from the photodecomposition of acetate species (Figure 4). This conclusion is supported by the absence of the m/z = 17 ion peak corresponding to ¹³CH₄ in the GC–MS spectrum, confirming that no methane derived from labeled ¹³CO₂ was formed under these conditions. Control reactions under argon confirmed the same outcome, indicating that acetic acid (formed from OAc-based imidazolium) was the main source of CH₄. The mechanism can be rationalized by the Kolbe-type decarboxylation reaction: oxidation of acetate anions (h⁺ + CH₃COO⁻ → CH₃• + CO₂), followed by reduction of methyl radicals (CH₃• + H⁺ + e⁻ → CH₄) [27,28]. This pathway highlights the key role of acetate photodecomposition in methane production, rather than direct CO₂ photoreduction [29,30].

In contrast, NS-TiO₂/Pd exhibited dual reactivity: both CO₂ reduction and H₂O splitting occurred concomitantly. With C1C4ImOAc, CH₄ was again observed alongside CO and H₂, under both CO₂ and argon atmospheres. Under a CO₂ atmosphere and in the presence of NaOAc, CO production becomes zero (entry 9), showing that the imidazolium cation in the ionic liquid plays a crucial role in CO₂ photoreduction, and that methane production does indeed stem from the presence of acetate ligands. This is confirmed by the use of the trifluoroacetate anion CF₃COO⁻, which also stops methane production under the same conditions, yielding only H₂ (entry 10). DRIFT spectroscopy of NS-TiO₂/Pd powders before (Figure 5C) and after reaction with diluted aqueous solution of C1C4ImOAc (Figure 5A,B) confirmed the strong surface coordination of acetate anions, as evidenced by characteristic stretching vibration ν(COO) between 1400 and 1700 cm⁻¹, supporting their direct participation in photocatalytic processes. The DRIFT spectrum of NS-TiO₂/Pd powder in the presence of the dilute C1C4ImOAc solution also shows ν(C-H) stretching vibration bands around 3000 cm⁻¹, characteristic of the imidazolium cation, which therefore might also interact with the photocatalyst surface. To confirm the chemisorption of the imidazolium cation, these bands are totally absent in the presence of acetic acid (Figure 5D) but are weakly present using C1C4ImCl (Figure 5E), demonstrating that imidazolium can interact with the photocatalytic support but only in synergy with the anion. It has already been shown that an Im-IL coating of TiO₂/Metal nanoparticles can drastically increase hydrogen production because of band gap change, enhancement of charge transfer between the photocatalyst and the electrolyte, and prevention of the recombination of photogenerated electron–hole pairs [8,31,32]. Although in our case, the quantities of ionic liquid used are not sufficient to achieve a uniform coating of the photocatalyst surface, the performances in H₂ production are exceptional (between 3000 and 3500 μmol-g_Pd⁻¹-h⁻¹) compared with literature values (around 100 and 500 μmol-g_Pd⁻¹-h⁻¹ for TiO₂/Pt [32] and TiO₂/Au [31] nanomaterials, respectively).

Kinetic monitoring over 20 h of NS-TiO₂/Pd in diluted C1C4ImOAc showed that CH₄ and CO formation were favored under argon (Figure A1a,b), while H₂ evolution dominated under CO₂ (Figure A1c), suggesting that the photo-Kolbe reaction (that produces CH₄ and CO₂) and CO₂ photoreduction (that produces CO) pathways compete under these conditions.

The nature of the support was then found to play a decisive role in governing reactivity. For NS-g-C₃N₄, strong π–π interactions between the nanosheet and the imidazolium cation generate a hydrophobic, positively charged surface layer that suppresses proton reduction (HER) while enhancing CO₂ solubility and acetate adsorption. The latter can undergo oxidation to yield CO₂ and CH₃• radicals, which are subsequently reduced to CH₄—thus diverting the pathway away from CO₂ reduction (entry 2). The positive surface potential further repels protons, inhibiting H₂ evolution. Moreover, as the small fraction of Im-ILs remains confined near the surface, their limited availability likely prevents effective CO₂ pre-activation, thereby hindering CO formation (Figure 6a). In contrast, TiO₂ exhibits weaker interaction with the imidazolium cation but a stronger affinity for the acetate anion through acetic acid coordination. This promotes CH₄ formation via the photo-Kolbe reaction. Because the TiO₂ surface is not positively charged, proton reduction becomes feasible, while the presence of imidazolium/carbon species in solution enables parallel CO₂ photoreduction. Consequently, Kolbe decomposition of acetate, direct CO₂ activation, and HER can occur simultaneously—accounting for the coexistence of CH₄, CO, and H₂ in the product mixture. Additional experiments with alternative anions further confirmed their decisive influence on photocatalytic behavior. Carbonate and bicarbonate salts (KHCO₃, Na₂CO₃) completely suppressed activity, consistent with the known inhibitory effect of alkaline conditions on CO₂ reduction. In contrast, the use of trifluoroacetate anion selectively inhibited CO and CH₄ formation, emphasizing the essential role of the protic cationic component in driving H₂ evolution (Figure 6b). Choline-based salts markedly enhanced hydrogen production, particularly in the presence of the NTf₂⁻ anion (entries 13 and 14), demonstrating that low-basicity anions favor water reduction pathways over CO₂ activation. Remarkably, for TiO₂ combined with NTf₂⁻ salts, an exceptionally high yield of H₂ was observed, suggesting a specific surface interaction between TiO₂, Pd, and NTf₂⁻. This strong synergetic effect points to a distinct interfacial mechanism that warrants further investigation to unravel its exact contribution to hydrogen generation.

In conclusion, the overall photocatalytic behavior arises from a decisive interplay between the nature of the support, the ionic liquid composition, and their interfacial interactions. The electronic structure and surface charge of the support dictate charge-transfer dynamics and product selectivity, while the choice of anion and cation modulates proton availability and reaction pathways. Specifically, π–π interactions on NS-g-C₃N₄/Pd favor CH₄ formation and suppress HER, whereas TiO₂ surfaces enable concurrent CO₂ reduction, acetate oxidation, and H₂ evolution. Moreover, anion basicity governs the competition between water reduction and CO₂ activation, with NTf₂⁻ salts promoting exceptional H₂ production through synergistic TiO₂–Pd–anion coupling.

4. Materials and Methods

4.1. Materials

The ionic liquids 1-butyl-3-methyl-imidazolium acetate (C1C4ImOAc, 95%), 1-butyl-3-methyl-imidazolium trifluroacetate (C₁C₄ImCF₃CO₂, 95%), and Choline bis(trifluoromethylsulfonyl)imide (CholineNTf₂, 99%) were supplied by Iolitec Company (Heilbronn, Baden-Württemberg, Germany), dried overnight under high vacuum (10⁻⁵ M bar) at room temperature, and then stored in an Ar-filled MBRAUN glovebox (the concentration of both H₂O and O₂ lower than 0.5 ppm) (MBRAUN France, Merignac, France). The potassium hydrogen carbonate (KHCO₃, 98%), sodium hydroxide (NaOH, 98%), sodium carbonate (Na₂CO₃, 99.5%), ethylene glycol (EG, 75%), sodium borohydride (NaBH₄, 99.6%), melamine (99%), Ti(OBu)₄ (TBOT, 97%), hydrofluoric acid in water solution (HF, 40%), acetic acid (CH₃COOH, 99.7%), Choline Chloride (CholineCl, 99%), Bis(trifluoromethane)sulfonimide lithium salt (LiNTf₂, 99.99%), and palladium chloride (PdCl₂, 99%) were purchased from Merck–Sigma Aldrich (Saint Louis and Burlington, MA, USA). CO₂ (99.995%) gas was supplied from Air Liquid (Paris, France). All of them were used without further purification.

Nanosheet 2D-TiO₂: The rectangular TiO₂ nanosheets were synthesized by a hydrothermal route in the presence of hydrofluoric acid solution [22]. Hydrofluoric acid solution (3 mL of 40 wt% in H₂O, 0.07 mol HF) was dropped into Ti(OBu)₄ (TBOT) (25 mL, 0.073 mol) under stirring for 2 h until the mixture became a gel. Afterwards, the gel was transferred into a dried Teflon autoclave with a capacity of 50 mL and kept at 180 °C for 36 h. After cooling down to room temperature, the powder was separated by high-speed centrifugation, then washed with ethanol and distilled water (15 mL) several times and dried in the oven at 50 °C for 10 h. Afterwards, the powder was dispersed in aqueous NaOH solution (0.1 M) and further stirred for 8 h at room temperature. After being recovered by high-speed centrifugation, the solid was washed with distilled water and ethanol several times (15 mL), and then dried at 80 °C for 6 h. In the end, 3.4 g of white powder (yield = 59%) were isolated and noted as NS-TiO₂.

Nanosheet 2D-TiO₂/Pd: The Pd-NPs loading of NS-TiO₂ (40 mg, 0.5 mol) was dispersed in ethylene glycol (EG) (20 mL) under ultrasonic, and a solution of PdCl₂ (0.04 mL, 0.09 M) was added dropwise. Then, a solution of NaOH/EG (0.5 M) was rapidly added to adjust the pH of the mixed solution to around 10–10.5, followed by the slow addition of an aqueous solution of NaBH₄ (25 mL, 2.0 mg·mL⁻¹). The total mixture was kept under stirring at room temperature for 8 h. Finally, the NS-TiO₂/Pd nanosheets were centrifuged and washed repeatedly with ethanol and ultra-pure water (15 mL). The resulting product was dried at 60 °C for 10 h. The theoretical loading of Pd onto NS-TiO₂ is around 1 wt%. The product was denoted as NS-TiO₂/Pd.

The preparation of the nanosheets NS-g-C₃N₄ and NS-g-C₃N₄/Pd by in situ loading of Pd onto g-C₃N₄ synthesized in H₂O was previously reported [15]. The elemental analysis of this sample yielded a Pd mass concentration of 2.74%, with an average size of the Pd nanoparticles around 2 nm.

4.2. Characterizations

Solution NMR spectra were recorded on Bruker AVANCE III 300 MHz with BBFO probe (Z gradient) using Topspin software 4.5.0 (Bruker BioSpin, Ettlingen, Germany). The ¹H and ¹³C NMR data were collected at room temperature on a Bruker AC 300 MHz spectrometer with the resonance frequency at 300.13 MHz for the ¹H nucleus and 75.47 MHz for the ¹³C nucleus. The spectra were obtained at 298 K unless otherwise specified. Chemical shifts are reported in parts per million (ppm) referenced to CD₂Cl₂ or (CD₂)₃SO as external reference capillary. (CD₂Cl₂, δ 5.32 ppm for ¹H, 53.84 for ¹³C; (CD₂)₃SO, δ 2.5 ppm for ¹H, 39.52 for ¹³C).

Solid-state NMR spectra were acquired using a Bruker Avance III spectrometer, equipped with an 11.74 T wide-bore magnet operating at Larmor frequencies of 500.13 MHz for ¹H and 125.72 MHz for ¹³C. The polypropylene Z-1500 sample was packed into a 4 mm ZrO₂ rotor and rotated at magic angle spinning (MAS) rates of 12 KHz, using a conventional 4 mm HX probe. The PP Z-1500 sample was ground into smaller pieces in order to achieve stable spinning of the NMR rotor. For CP MAS spectrum, transverse magnetization was obtained from ¹H using contact pulse duration of 2 ms, recycle delay of 5 s, and 2028 scans. SPINAL 64 ¹H decoupling was applied during acquisition. The DD MAS experiment was recorded with a recycle delay of 5 s and 3072 scans. The chemical shifts were referenced to the TMS using adamantine as an external standard. All results shown have been recorded at room temperature at 12 kHz MAS frequency.

The mass spectrometry analyses were performed using an ion trap (AmaZon SL, Bruker) equipped with an electrospray ion source (ESI) operated in positive or negative ion mode. Samples were prepared in 1 mg/mL in CH₂Cl₂ and diluted by 100 in a solvent mix of 46.1% methanol, 38.4% dichloromethane, 15.4% ultra-pure Elga water, and 0.1% formic acid, before being injected at 10 µL/min into the mass spectrometer.

The infrared spectra were obtained by diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy on a Thermo Scientific Nicolet TM iS TM 50 IR-TF spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The spectra were recorded from 4000 to 400 cm⁻¹ at room temperature.

The elemental analysis of carbon, nitrogen, and hydrogen was recorded on a Thermo Scientific FlashSmart^TM Elemental Analyzer. The elemental analysis of Pd was carried out by ICP-AES (Icap 6500 DUO-Thermo Scientific) by the Crealins Company (6NAPSE group, Lyon, France) at Villeurbanne.

X-ray diffraction (XRD) was performed on a PANalyticalX’Pert Pro diffractometer equipped with an X’Celerator Scientific detector and a Cu anticathode (Kα1/Kα2) (Malvern Panalytical, Limeil-Brévannes, France). Intensities were collected at 150 K by means of the CrysAlis^Pro software v43.22a. Reflection indexing, unit-cell parameters refinement, Lorentz-polarization correction, peak integration, and background determination were carried out with the CrysalisPro software.

Electron microscopy analyses were carried out at the Centre Technologies of Microstructures (CTµ), a platform of the Université Claude Bernard Lyon1, Villeurbanne, France. A drop of a water dispersion was deposited on a copper grid (200 mesh), and the solvent was allowed to evaporate. Samples were observed by transmission electron microscopy (TEM) at room temperature with a JEOL (Peabody, MA, USA) 2100F high-resolution microscope operating at an accelerating voltage of 200 Kv (JEOL Europe SAS, Croissy-sur-seine, France). The specimen holder was tilted at 15° (with respect to the energy-dispersive detector). HAADF-STEM was carried out at the Centre Technologies of Microstructures (CTµ), a platform of the Université Claude Bernard Lyon 1, Villeurbanne, France. A drop of a water dispersion was deposited on a carbon grid (200 mesh), and the solvent was allowed to evaporate. Samples were observed by HAADF-STEM at room temperature with a JEOL 2100F high-resolution microscope.

Textural properties were investigated by determining the specific surface areas of materials using the Micrometrics ASAP 2020 system, which is based on the BET method (Micromeritics France SA, Merignac, France). Prior to measurement, solids were desorbed at 423 K (150 °C) for 3 hrs. Nitrogen (N₂) gas was used as adsorptive gas and relative pressure (P/P₀) was varied from 0.019 to 0.986.

4.3. Photocatalytic Reduction of CO₂ in Diluted Ionic Liquid

All reactions were carried out under CO₂ atmosphere (CO₂ bubbling) and UV-vis irradiation (MAX-303 Xenon light source 300 W, 300–600 nm using UV-VIS mirror module, Asahi Spectra Co., Ltd., Tokyo, Japan), at room temperature. Photocatalytic reduction of CO₂ was performed in a 25 mL Pyrex glass tube, and a glass flask was connected by a tee joint for vacuum, CO_2, or Argon purging and gas sampling in the system. Typically, 6 mg of NS-TiO₂/Pd or of NS-g-C₃N₄/Pd, 3 mL H₂O and 0.4 mmol of selected ILs or salts were added in the 25 mL glass Pyrex tube, under ultrasonic for 10 min to mix all of them well, then, with CO₂ or Argon bubbling for 45 min without any light, later, the reactor was vacuumed and purged 3 times with CO₂ or Argon at 1 bar pressure. The stirred medium was exposed to UV-vis irradiation for 4 or 20 h, after 2 gas syringes (2 × 500 µL) containing a Hamilton sample lock valve were removed for multinuclear NMR, TCD, and FID of GC analyses, respectively.

The gas analyses were performed with a gas chromatograph (HP 6890) equipped with an HP-PLOT Mole sieve column (TCD) and a GS-carbon PLOT column (JetanizerTM/FID) to detect the production of CO, H_2, and other possible reduction products. A standard calibration curve was made to quantify the CO yield. Calibration curve for CO and CH₄ was determined following the equation Y(mol) = 3.23 × 10^11X − 2.68 × 10⁻¹⁰ and Y(mol) = 3.77 × 10^−11X for H₂. The GCMS analysis was performed in an Agilent (Santa Clara, CA, USA) 6850 (GC)/5975C (MS) under EI (electron impact) mode. The GCMS was equipped with a PoraBond Q column; the oven was kept at 60 °C during the analysis.

5. Conclusions

This study highlights the central role of ionic liquids, and in particular, C1C4ImOAc, in governing photocatalytic reactivity under aqueous UV–vis irradiation. In the absence of a heterogeneous catalyst, C1C4ImOAc enables selective CO formation via CO₂ pre-activation, confirming its dual role as both photosensitizer and activator, in agreement with previous reports. When coupled with nanosheet photocatalysts, however, new reactivity emerges: in g-C₃N₄ systems, methane is predominantly formed through the photo-Kolbe decomposition of acetate, while in TiO₂ systems, multiple pathways coexist, combining CO₂ reduction, H₂O splitting, and acetate photodecomposition. From a mechanistic standpoint, these findings clarify the origin of methane in such systems, ruling out CO₂ as the primary carbon source under certain conditions. The results also underline the sensitivity of CO₂ photoreduction to the interplay between photocatalyst surface properties, ionic liquid structure, and reaction medium. Anion basicity and pH emerge as decisive factors: alkaline environments or low-basicity anions suppress carbon product formation and enhance H₂ evolution.

Looking ahead, several directions arise. First, suppressing the competing photo-Kolbe reaction will be essential to distinguish genuine CO₂ reduction from acetate photodecomposition. Strategies may include tailoring OAc–based IL structures to avoid strong interaction with g-C₃N₄ surface (i.e., steric hindrance) or to enhance its affinity with water (i.e., introduction of –OH group) in order to increase its availability in solution for the photoreduction of CO₂. For metal-based support, a less basic anion than OAc must be used in order to avoid strong surface interaction. Second, engineering photocatalyst surfaces to modulate IL–support interactions could also provide selective control over electron transfer pathways. For TiO₂, introducing cocatalysts or defect engineering may mitigate its strong bias toward hydrogen evolution. These findings highlight that mastering the support–electrolyte (ionic liquid) interface is the key to directing multi-electron photocatalytic processes—offering a powerful strategy to rationally design next-generation hybrid systems for selective CO₂ photoconversion and solar fuel, as well as hydrogen, production.