Assessing the CO₂ Capture and Electro-Reduction in Imidazolium-Based Ionic Liquids: Role of the Ion Exchange Membrane

Abstract

The electrochemical CO₂ reduction (eCO₂RR) to valuable chemicals offers a promising method to combat global warming by recycling carbon. Among the possible products, syngas—a CO and H₂ mixture—is especially valuable for industrial reactions. The use of Room Temperature Ionic Liquids (RTILs) electrolytes presents a promising pathway for eCO₂RR because of the lower overpotential required and the increased CO₂ solubility with respect to the aqueous ones. Ensuring a constant CO/H₂ production is essential, and it relies on both the catalyst and reactor design. This study explores eCO₂RR in RTIL mixtures of 1-butyl-3-methyl imidazolium trifluoromethanesulfonate (good for CO₂ conversion) and 1-butyl-3-methyl imidazolium acetate (good for CO₂ capture), with various amounts of water as a proton source. We evaluated syngas production stability across different electrochemical cells and ion exchange membranes after determining the appropriate electrolyte mixture for a suitable CO/H₂ ratio near 1:1. The two-chamber cell configuration outperformed single-cell designs by reducing oxidative RTILs degradation and by-products formation. Using a bipolar membrane (BPM) in forward mode led to catholyte acidification, causing an increase of HER relative to eCO₂RR over time, confirmed by Multiphysics modeling. Conversely, an anionic exchange membrane (AEM) maintained constant syngas production over extended periods. This work offers guidelines for syngas generation in RTIL-based systems from waste-CO₂ reduction, which can be useful for other green chemical synthesis processes.

1. Introduction

Increasing CO₂ atmospheric concentration (from 280 ppm during the pre-industrial era to 426 ppm reached in 2024) is a crucial threat to humankind, leading to global warming. To face this challenge, many efforts have been dedicated to valorizing waste CO₂ to obtain value-added products. Electrochemistry plays a pivotal role since CO₂ can be reduced to other chemical species, such as CO [1,2,3], HCOOH [4,5,6], CH₄ [7,8,9], C₂H₄ [10,11,12,13], and CH₃OH [14,15,16] via the electrochemical CO₂ reduction reaction (eCO₂RR). In particular, the simultaneous production of CO and H₂, defined syngas, is a crucial reagent for industrially relevant reactions [17,18,19,20,21,22,23] When investigating the conversion of CO₂ dissolved in the liquid electrolyte, aqueous-based electrolytes suffer from low CO₂ solubility (30 mM) [17], leading to sluggish reaction kinetics and competing side reactions, like the H₂ evolution reaction (HER). Therefore, organic solvents such as acetonitrile [18] (CH₃CN) are used, enabling a higher CO₂ solubility (0.02 mol%) [19], but usually entailing low conductivity and lack of stability under electrochemical conditions because of electrode passivation or solvent degradation. However, overcoming these challenges requires the development of electrolytes that provide improved CO₂ transport, stability, and tailored interactions with reaction intermediates. In this scenario, Room Temperature Ionic Liquids (RTIL) have been studied as effective electrolytes for eCO₂RR applications [20,21,22,23,24]. RTILs are molten salts at ambient conditions. Their ionic nature makes RTILs very conductive. However, due to the strong ionic interactions between the cations and the anions, they usually suffer from high viscosity. Thus, dissolving RTILs in an organic solvent (e.g., CH₃CN or 3-methoxypropionitrile, 3-MPN) is a good strategy for obtaining highly conductive electrolytes. RTILs offer superior CO₂ solubility and a wider electrochemical stability window compared to other non-aqueous electrolytes, making them the ideal choice for CO₂ capture and conversion applications; they act as co-catalysts by interacting with CO₂ and stabilizing the intermediate CO₂^•− radical anion (generally considered to be the rate-determining step for CO₂ activation), lowering the eCO₂RR overpotential [25,26]. Imidazolium-based RTILs have shown excellent CO₂ capture properties [19], and high conversion efficiencies [27,28]. The capture properties are related to the ability of the imidazolium cation to release a proton, forming a highly reactive carbene species; this carbene can interact with CO₂, forming a Bmim-CO₂ complex. The ability to form the carbene species is controllable by choosing the right anion, as highlighted by the work of Fortunati et al. [29,30].

Not only the electrolyte or the electrocatalyst design [31,32], but the reactor setup also plays an important role [33,34]. The electrochemical configurations (e.g., the type of electrochemical cell employed) strongly influence the outcome of the process. A single-compartment cell is a simple, easy-to-operate setup commonly used in lab experiments exploiting RTILs. It requires a few amount of electrolyte and bypasses the anolyte optimization process. However, coexistence in the same cathodic and anodic processes environment can lead to undesired oxidation/reduction of the cathodic/anodic products. Besides, the electrolyte should be optimized to be simultaneously suitable for eCO₂RR at the cathode and Oxygen Evolution reaction (OER) at the anode [35,36]. Instead, the membrane is a decisive parameter when using two-compartment cells (e.g., H-type cells) [37]. For aqueous-based eCO₂RR applications, a bipolar membrane (BPM) is generally preferred to enhance the CO₂ single-pass conversion (SPC) with respect to an anionic exchange membrane (AEM). CO₂ forms (bi)carbonate anions in water, which migrate from the catholyte to the anolyte through the AEM and re-oxidizes at the anode, causing CO₂ loss [38].

The present work studies mixtures of imidazolium-based RTILs, with and without adding water as a proton source, as electrolytes for eCO₂RR to syngas on Ag by using laboratory-scale electrochemical cells. Based on a recent work of some of us [30], we selected 1-butyl-3-methyl imidazolium trifluoromethane sulfonate ([Bmim] [CF₃SO₃]), which promotes highly selective CO production, and 1-butyl-3-methyl imidazolium acetate ([Bmim] [CH₃CO₂]), which is a good CO₂ capture RTIL and promotes HER. Hence, the synergistic effect of the two anions was exploited here to achieve a tunable and constant CO/H₂ ratio during syngas production [39,40,41,42]. The system’s selectivity and stability towards syngas evolution depended on configuration parameters such as electrolyte composition and electrochemical cell employed. In the case of the two-compartment H-cell configuration, the role of the membrane employed in the reaction outcome was studied experimentally and with COMSOL 6.1 (COMSOL Srl, Brescia, Italy) Multiphysics modeling.

This work provides guidelines to control the CO/H₂ ratio and stability in RTIL-based electrolytes by exploiting their CO₂ capture and cocatalyst features. Moreover, we (i) highlight the necessity to take into account the eventual degradation processes of the RTILs, avoidable by separating the anode and cathode chambers; (ii) most importantly, we warn the reader that BPM is not always the right choice when talking about eCO₂RR. This work paves the way for developing sustainable integrated RTILs-based technologies for synthesizing fine chemicals (for example, via carbonylation [43,44], hydroformylation or thermocatalytic processes for methanol production), requiring different CO/H₂ ratios (from pure CO to CO/H₂ = 2 to 0.5), starting from syngas electrochemically generated from CO₂ [45,46].

2. Results

2.1. Electrolyte Preparation

Electrolytes were prepared by mixing two different ILs, i.e., [Bmim] [CH₃CO₂ and [Bmim] [CF₃SO₃] (The chemical structures of the RTILs are depicted in Figure 1), in 3-MPN. Different amounts of H₂O were added to the mixture to act as a proton source for eCO₂RR and HER. In principle, controlling the CO/H₂ ratio should be possible by varying the water content. Mixtures of 1.6 M of [Bmim] [CF₃SO₃] and 2.4 M of [Bmim] [CH₃CO₂] in 3-MPN were used to maximize the conductivity of each IL component (see Figure S1 in the Supplementary Materials). Then, different volumetric percentages of H₂O were added to the original solution, namely 20%v/v, 10%v/v, 5%v/v, and 1%v/v, until a total volume of 50 mL (see molar compositions in

Table 1. Molar composition of the electrolytes.

Name	H2O (%v/v)	H2O (mL)	H2O (mmol)	Added H2O (M)	[SO₃CF₃] (mmol)	[SO₃CF₃] (M)	[CO2CH₃] (mmol)	[CO2CH₃] (M)	[Anion] (mmol)	Conductivity (mS/cm)
MIX20%	20	10	550.0	11.0	48	0.96	24	0.48	72	21.7
MIX10%	10	5	275.0	5.5	54	1.08	27	0.54	81	18.8
MIX5%	5	2.5	137.5	2.75	57	1.14	28	0.57	85	16.1
MIX1%	1	0.5	27.5	0.55	59	1.19	29	0.59	88	12.7
MIX0%	~0.2	0	~3	0	59	1.20	30	0.60	89	11.6

). The different mixes were named based on the water addition (%v/v). MIX0% stands for the electrolyte without any further addition of water (besides water traces < 110 ppm are present, see Section 4.1), while MIX20% contains 20%v/v of H₂O. Logically, the molarity of the ILs decreases with the increase in water content. The ratio between triflate’s and acetate’s molarities was kept constant at 2.

2.2. RTIL-Based Electrolyte Electrochemical Characterization

The activity towards eCO₂RR of the electrolyte mixtures was studied in a single compartment cell (Figure S2A, Supplementary Materials) via Linear Sweep Voltammetry (LSV) between 0 and −2.5 V vs. Ag/AgCl with a scan rate of 10 mV/s. In N₂-saturated electrolytes, the onset potential shifts toward more positive values for higher water contents (Figure 2a), because of higher HER activity due to the increased concentration of protons. With water traces (MIX0%), a more negative onset potential was observed in the CO₂-saturated rather than the N₂-saturated, indicating that the eCO₂RR is favorably promoted in the RTIL-based electrolyte. This phenomenon was observed for any mixes with water content until 5%. Conversely, analogous onset potentials were reported in the N₂- and CO₂-saturated MIX20%, indicating that the HER is a non-negligible process at 20% water concentration in the electrolyte (see Figure 2b and inset). The positive shift of the onset potential in the presence of CO₂ evidences the reactivity of the electrolytes towards eCO₂RR. In N₂-saturated electrolyte, every mixture is electrochemically stable until an applied potential of −2 V vs. Ag/AgCl, up to 5%v/v, where the water content becomes high enough to shift the onset potential due to the HER. For more negative potential, the electrochemical reduction of the ILs in the electrolyte began to be the primary Faradaic process.

2.3. CO₂ Electrolysis Tests

Electrolysis tests were conducted in a single-compartment cell and an H-type (Figure S2B, Supplementary Materials) cell to assess the cell configuration’s influence on selectivity and the CO and H₂ production stability in the RTILs-based electrolytes with different water contents. Figure 3 reports the Faradaic Efficiency (FE) of the principal gaseous products during 90 min CP experiments in the single cell at −20 mA/cm². These results suggest that water is the leading proton donor for HER and eCO₂RR. Indeed, by lowering the H₂O content in the electrolyte, the FE shifts towards mainly CO production. However, even with traces of water in the electrolyte (no added water for MIX0% and 1.6 M Bmim OTf), the eCO₂RR occurs, while the HER is significantly suppressed. Fortunati et al. recently showed that a stronger Lewis base anion (as acetate) turns the C2 proton of the imidazolium cation more acidic. Thus, if released, this proton can serve as a co-proton source for the reaction together with the low amount of water present in the electrolyte.

Moreover, the total Faradaic efficiency decreases following a decrease in water content and applied current (Figure S3, Supplementary Materials). These results may be attributed to the products’ partial reoxidation and the possible presence of other Faradaic processes, such as the degradation of ILs. When a polycrystalline Ag plate is employed as a working electrode, the three detectable reduction products are CO and HCOO− (from eCO₂RR) and H₂ (from HER). However, as pointed out by the pioneering work of Hori [40], in the potential range explored, in an aqueous system, the selectivity towards CO production on Ag is one order of magnitude higher than HCOO⁻. Moreover, previous work from our group using similar electrolytes and catalysts [30] highlighted how the Ionic Liquid here employed (e.g., Bmim OTf and Bmim OAc) are selective towards producing CO and H₂, respectively. No other liquid products (such as HCOO− or alcohols) were detected significantly either by HPLC or GC-MSD. Kroon et al. [47] employed theoretical calculations to predict that cation radicals may form at cathodic potential, causing imidazolium dimerization through C2 hydrogen. Cation dimerization during the reaction is supported by the evident color change in the solution, from pale yellow to dark brown, especially at low water content in the electrolyte (Figure S4, Supplementary Materials). Besides, CH₄ and C₂H₄ were detected when potentials over −2 V were applied in a single-compartment electrochemical cell, and the concentration of these products increased with the applied potential and with the ILs molarity in the mixture (Figure S5, Supplementary Materials), thus suggesting both to be IL’s degradation products. Hence, the electron count is probably closed by a reductive mechanism of Ionic Liquids degradation, as reported by Kroon et al. However, since these RTIL cathodic degradation mechanisms are radical-involved mechanisms, it is difficult to make an exact count to obtain an efficiency value, and this is beyond the scope of the present paper. To confirm this hypothesis, a CP ramp ranging from −5 mA/cm² to −90 mA/cm² was conducted in an inert N₂ atmosphere (Figure 4), continuously analyzing the outlet gases to confirm whether the C-based products are generated even without CO₂. CH₄ and C₂H₄ were detected, while CO was absent, confirming that methane and ethylene are formed as IL degradation products. CO started forming when the electrolyte was saturated with CO₂, and a sudden decrease in the cathodic potential, E(we), happened. Since no variation was detected on the anode potential, E(ce), after CO₂ saturation of the electrolyte, we hypothesized that the RTILs degradation occurred via oxidation at the Pt anode (placed close to the cathode in the single-compartment cell). Moreover, Ag is known to be selective towards CO production, and only a few works report CH₄ and C₂H₄ production under specific mass transport conditions, further excluding any CO₂ reduction origin of those products under the tested conditions [48,49]. As a final note, LSVs carried out before and after electrolysis for the MIX0% showed significant difference (while minor changes occurred for MIX20%, see Figure S6A,B, Supplementary Materials), indicating that the electrolyte degradation strongly depends on IL concentration, being higher at lower water contents.

Previous experimental evidence showed that the separation between the catholyte and the anolyte is of utmost importance, even at the expense of higher cell potential. In this way, the electrolyte degradation at the anode can be prevented, consequently preserving the RTIL-based electrolyte’s stability and the syngas’ purity out-stream and suppressing the evolution of unwanted side products such as CH₄ and C₂H₄. By changing the experimental setup from a single-cell to an H-cell, the membrane employed to separate the cathodic and anodic compartments becomes the primary concern.

From the data obtained in the single cell, MIX20% was selected as the most suitable electrolyte for syngas production, allowing a total CO:H₂ ratio close to 1:1. In the single cell, despite the initial high CO selectivity, HER generally increased during operation becoming the primary faradaic process at the end of the test (Figure S7, Supplementary Materials), probably due to the RTIL-electrolyte partial oxidation. To overcome this limitation, the co-electrolysis experiments were performed in an H-type cell (see the Section 4.3) with MIX20%, and the obtained FE values are summarized in Figure S8 (Supplementary Materials). In this configuration, the membrane dramatically influences selectivity over time. When using a bipolar membrane, the CO production was rapidly suppressed, arriving at zero after a few minutes of operation (see Figure 5a). The system’s selectivity becomes stable only when an anionic membrane is employed (Figure 5b). This phenomenon can be ascribed to the proton production at the interface layer of the BPM, which makes the system more selective towards HER. As detailed in the Section 2.4, a descriptive COMSOL model has been built to confirm this hypothesis. Moreover, adding 20% of water to the ILs-based electrolyte also reduced the possibility of the imidazolium cation to act as a proton source and ensured a higher electrolyte stability.

On the other hand, catholyte and anolyte cross-over is a severe issue when a two-compartment cell (e.g., H-cell or Flow cell) is employed, and organic electrolytes can exacerbate such phenomena by destabilizing the polymeric matrix of the membrane. To investigate such phenomena, two samples of the anolyte (KOH aqueous solution) and two samples of the catholyte, were collected before and after several hours of chronopotentiometry with the AEM, and were analyzed by IR spectroscopy. As shown in the Figure S9 (Supplementary Materials), the two anolyte spectra are similar, which confirms the absence of ILs crossover from the catholyte to the anode compartment. Indeed, no peaks related to the RTILs nor the 3-MPN organic solvent are present in the anolyte after 3 h of test: only the vibrational modes of water (between 3000 and 3000 cm⁻¹) related to O-H bonds stretching, and the sharper peak (around 1600 cm⁻¹) related to the H-O-H scissoring vibration are visible. Similarly, Figure S10 (Supplementary Materials) compares the two catholyte spectra before and after the test, demonstrating no appreciable aqueous KOH cross-over during the reaction. The broad O-H band in the region between 3000 and 3500 cm⁻¹ is not visible in the catholyte’s IR spectra at the end of the chronopotentiometry, suggesting that the anolyte crossover is not a severe issue in the proposed setup.

2.4. Computational Modeling

To explain the experimental results shown in Figure 5, we developed a simplified 1-dimensional COMSOL Multiphysics model. In particular, the model aims to validate the following hypothesis: (i) Water is the main proton source for HER; (ii) Protons in the C2 position of the imidazolium cations of the ILs can act as additional proton sources for CO₂ reduction when the water concentration is low; (iii) the use of a Bipolar Membrane determines unstable syngas production due to increased proton concentration at the cathode during operation. The structural model included the Ag surface (cathode) as a point electrode, a diffusion layer of 100 µm, and the bulk electrolyte (as a point concentration reference). Since the density of [Bmim][CH₃SO₃] is equal to 1.2135 kg/L at 292.95 K and CO₂ molality in [Bmim][CH₃SO₃] is 0.4927 mol/kg at 293.2 K [50], then we assumed a CO₂ solubility in [Bmim] of 0.60 mol/L, equivalent to 600 mM. This CO₂ concentration was then set at the bulk electrolyte, while [H⁺] and [OH⁻] were generically assumed to be 0 mM, and [CH₃SO₃⁻], [CO₂CH₃⁻], and [H₂O] bulk concentrations were taken from Table 1. For simplicity, no further species were included in the model. Anode and, consequently, anodic reactions were not considered since the focus was on the catholyte chamber of the H-cell experimentally employed.

As heterogeneous reactions at the cathode, CO and H₂ formation were considered, with kinetic parameters, charge transfer coefficients, and equilibrium potentials retrieved from Ref. [51] (see

Table 2. Kinetic parameters included in the model.

	j0 (mA/cm2)	α	E0/V vs. RHE	Reference
H2	6.36 10−2	0.14	0	[51]
CO	2.01 10−3	0.35	−0.11	[51]

). The CO exchange current density from Ref [51] was multiplied by a correction factor of 5 to match experimental results. Instead, the kinetic parameters evaluated from our own experimental study were not used due to the significant mass transfer limitations occurring in our system (due to varying H₂O content). The proton source for CO₂ reduction and Hydrogen Evolution was either [Bmim⁺] protons (at low water content in the electrolyte, see Equations (1) and (2)) or H₂O itself (Equations (3) and (4)). CO and H₂ current densities were then modeled via Butler-Vomer equations, including a specific dependence on CO₂ (for eCO₂RR) and H⁺/H₂O surface concentrations (for HER, depending on the proton source). This mass transfer factor was normalized by a reference concentration of 1 M. The actual conductivity of the bulk electrolyte (data in Table 1) was used in the model.

$${\mathrm{C}\mathrm{O}}_2+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }\mathrm{C}\mathrm{O}+{\mathrm{H}}_2\mathrm{O}$$

(1)

$${2\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }{\mathrm{H}}_2$$

(2)

$${\mathrm{C}\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }\mathrm{C}\mathrm{O}+2{\mathrm{O}\mathrm{H}}^{-}$$

(3)

$$2{\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\stackrel{\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}\mathrm{s}}{\to }{\mathrm{H}}_2+2{\mathrm{O}\mathrm{H}}^{-}$$

(4)

Regarding chemical equilibria reactions at the bulk electrolyte, water self-ionization (Equation (5)), and release of C2 proton (Equations (6) and (7)) were included, assuming the reaction rates reported in

Table 3. Homogeneous reaction rates.

Constant	Value	References
kw1 (mol/m3 s)	2.4·10−2	[17]
kw2 (m3/mol s)	2.4·106	[17]
ΔG([Bmim]:)SO3CF3 (eV)	1.03	[29]
ΔG([Bmim]:)CO2CH3 (eV)	2.24	[29]

, where values on ionic liquids are taken from our recent density functional theory study [29]. In Equation (7), we set a frequency factor of 1 fs⁻¹. Given the previous considerations, H⁺ and OH⁻ evolution within the electrolytes followed Equations (8) and (9).

$${\mathrm{H}}_2\mathrm{O}\stackrel{\mathrm{k}\mathrm{w}1,\mathrm{k}\mathrm{w}2}{\to }{\mathrm{H}}^{+}+{\mathrm{O}\mathrm{H}}^{-}$$

(5)

$$\left[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}\right]\left[\mathrm{a}\mathrm{n}\mathrm{i}\mathrm{o}\mathrm{n}\right]\stackrel{{\mathrm{k}}_{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{m}}\left(\right[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}]:)}{\to }\left[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}\right]:+{\mathrm{H}}^{+}+{\mathrm{a}\mathrm{n}\mathrm{i}\mathrm{o}\mathrm{n}}^{-}$$

(6)

$$\left[{\mathrm{H}}^{+}\right]=A\mathrm{e}\mathrm{x}\mathrm{p}\left({\displaystyle \frac{-{∆G\left(\right[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}]:)}_{{\mathrm{S}\mathrm{O}}_3{\mathrm{C}\mathrm{F}}_3}}{{\mathrm{k}}_{\mathrm{B}}\mathrm{T}}}\right)\left[{\mathrm{S}\mathrm{O}}_3{\mathrm{C}\mathrm{F}}_3\right]+A\mathrm{e}\mathrm{x}\mathrm{p}\left({\displaystyle \frac{-{∆G\left(\right[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}]:)}_{{\mathrm{C}\mathrm{O}}_2{\mathrm{C}\mathrm{H}}_3}}{{\mathrm{k}}_{\mathrm{B}}\mathrm{T}}}\right)\left[{\mathrm{C}\mathrm{O}}_2{\mathrm{C}\mathrm{H}}_3\right]$$

(7)

$$R_{{\mathrm{H}}^{+}}={\mathrm{k}\mathrm{w}}_1\left[{\mathrm{H}}_2\mathrm{O}\right]-{\mathrm{k}\mathrm{w}}_2\left[{\mathrm{H}}^{+}\right]\left[{\mathrm{O}\mathrm{H}}^{-}\right]+{\mathrm{k}}_{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{m}-{\mathrm{S}\mathrm{O}}_3{\mathrm{C}\mathrm{F}}_3}\left[{\mathrm{S}\mathrm{O}}_3{\mathrm{C}\mathrm{F}}_3\right]+{\mathrm{k}}_{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{m}-{\mathrm{C}\mathrm{O}}_2{\mathrm{C}\mathrm{H}}_3}\left[{\mathrm{C}\mathrm{O}}_2{\mathrm{C}\mathrm{H}}_3\right]$$

(8)

$$R_{{\mathrm{O}\mathrm{H}}^{-}}={\mathrm{k}\mathrm{w}}_1\left[{\mathrm{H}}_2\mathrm{O}\right]-{\mathrm{k}\mathrm{w}}_2\left[{\mathrm{H}}^{+}\right]\left[{\mathrm{O}\mathrm{H}}^{-}\right]$$

(9)

The diffusion coefficients and charges reported in

Table 4. Diffusion coefficients assumed in the model.

Species	Diffusion Coefficient (cm2/s)	Charge (|e|−)	Solvent	Reference
H+	9.311·10−5	+1	Water	[17]
OH−	5.273·10−5	−1	Water	[17]
H2O	2.57·10−5	0	Water	[52]
CO2	1.91·10−5	0	Water	[17]
CO2	2.9·10−6	0	[Bmim][PF6]	[53]

were used to solve the Nernst- Planck equation. Regarding diffusion coefficients, values estimated in water for H⁺, OH⁻, and H₂O were multiplied by a correction factor obtained as ${\displaystyle \frac{D_{{\mathrm{C}\mathrm{O}}_2}\left(\right[\mathrm{B}\mathrm{m}\mathrm{i}\mathrm{m}\left]\right[{\mathrm{P}\mathrm{F}}_6\left]\right)}{D_{{\mathrm{C}\mathrm{O}}_2}\left(\right[{\mathrm{H}}_2\mathrm{O}\left]\right)}}$. Diffusion coefficients for [CH₃SO₃⁻] and [CO₂CH₃⁻] were assumed to be 0.1·10⁻⁵ cm²/s, while their charges were set to −1 |e|⁻.

By running a stationary simulation assuming the molar compositions in Table 1 and an applied current density of −20 mA/cm², the model returns values of Faradaic efficiency toward H₂ and CO well aligned with the experimental results in Figure 3. As shown in Figure 6, H₂ selectivity is close to zero at low water content, suggesting the need for H₂O to sustain hydrogen production. Instead, a high selectivity toward CO is observed with very low water content, strengthening the hypothesis of [Bmim] C2-H as an additional proton source for the eCO₂RR.

To qualitatively investigate the role of the bipolar membrane in tuning the H₂/CO ratio, we included in the model an artificial proton source at the electrode (H⁺ flux of 0.5 mol/(m²s)) to resemble the water splitting occurring at the BPM. As evident from the time-dependent study shown in Figure 7, after an initial CO₂ reduction activity, the diffusion of H⁺ from the membrane to the cathode leads to high H₂ and negligible eCO₂RR activity (Figure 7a). Indeed, a very high H⁺/CO₂ concentration ratio is observed at the surface. Instead, by employing the anion exchange membrane (i.e., without any proton flux, Figure 7b), H⁺ concentration does not increase, as confirmed by the H⁺/CO₂ ratio around 0.

3. Discussion

Our study suggests that several phenomena are responsible for the instability in syngas evolution in RTIL-based electrolytes. The anodic degradation of the RTILs during electrolysis in a single-compartment cell is a severe issue that determines the generation of side products like CH₄ and C₂H₄ and, thus, the unreliable determination of eCO₂RR products in single cells. Electrolysis experiments have evidenced acetate anion as the main cause of degradation by-products. Acetic acid electrolysis on Pt at oxidative potential follows a mechanism known as Kolbe reaction. This mechanism leads to the cleavage of the C-C bond of acetic acid with the subsequent coupling of two CH₃^• radicals to form ethane. Ethylene is reported to be a non-Kolbe product—coming from a side mechanism of the Kolbe reaction [54]. However, the absence of ethane as a gaseous byproduct indicates the existence of another degradation mechanism that has led to the evolution of CH₄ and C₂H₄. The reactivity of acetic acid towards electrolysis on Pt anode leading to a similar Kolbe mechanism supports what is reported in Figure 3, where MIX0% displayed a total FE lower with respect to 1.6 M Bmim OTf. This phenomenon can be explained by the presence in MIX0% of the acetate anion coming from Bmim Acetate. The hypothesis is that when the Pt anode is immersed in the electrolyte within a single cell, the presence of acetate anions increases the probability of radical species formation, thereby raising their concentration. This makes the electrolyte chemically less stable, even at the cathodic interface. When only Bmim OTf is employed as electrolyte, the anion degradation doesn’t occur, and the total FE, which is close to 100%, accounts for only the eCO₂RR to CO and HER. The remaining percentage of FE can then be explained by the previously cited cation degradation reported and calculated by Kroon et al. [47] However, the mechanism of anodic RTILs degradation requires additional studies, which will be discussed in a follow-up manuscript. Regarding the identity of the proton source, the selectivity towards HER rises dramatically by increasing the amount of water in the electrolyte. Besides, eCO₂RR still happens even in electrolytes containing very low amounts of H₂O, such as MIX0% and 1.6 M [Bmim] [CF₃SO₃], indicating that C2 of the imidazolium cation can serve as an additional proton source. This hypothesis is supported by COMSOL modeling, which shows good agreement with the experimental data (Figure 6). Nearly 100% of Faradaic efficiency toward CO is predicted even when the model assumes the absence of water and the equilibrium reaction in Equation (6), confirming the reliability of the RTIL C2-H as an alternative proton source.

Experimentally, we note that in the case of 1.6 M [Bmim] [CF₃SO₃], the total FE is over 90%, confirming the hypothesis that the IL with the [CF₃SO₃]⁻ anion is less prone to degradation than with the [CH₃CO₂]⁻ one. CO production remains stable when no [CH₃CO₂]⁻ anion is present in the electrolyte. Under this condition, no other Faradaic process can occur at the Ag surface other than eCO₂RR, and the protons coming from the imidazolium cation are selectively employed for CO evolution. When only [CF₃SO₃]⁻ anions are present, the electrode-electrolyte interface is more hydrophobic due to the presence of the -CF₃ moiety; thus, less water is present at the surface. Given the high CO selectivity reported, the hypothesis that H from C2 can act as a proton source for eCO₂RR is further confirmed. Cation degradation to release C2-H may be an issue regarding the system’s long-term stability since this eventually led to a change in electrolyte composition. Moreover, imidazolium cations at the electrode interface, when losing the proton in C2 position, can form a carbene species that poisons the Ag surface [30], hindering eCO₂RR. These processes can be avoided by adding H₂O to the electrolyte. Proton evolution from H₂O is easier and more favorable than the IL cation disruption.

Regarding the electrochemical setup, changing from a single cell to an H-type cell is crucial to avoid IL oxidative degradation. However, the membrane employed to separate the anodic and cathodic chambers also plays a crucial role in the outcome of the process. When MIX20% was tested in the H-type cell equipped with BPM, syngas was produced, but the CO/H₂ ratio was not stable due to the rapid suppression of eCO₂RR at the Ag interface. After 1 h of −20 mA/cm² chronopotentiometry, the system’s selectivity switched towards HER. Conversely, in the same electrolytic condition but equipping the H-cell with an AEM, the CO/H₂ ratio reached a steady state after 1.5 h, and the syngas productivity was maintained for up to 3.5 h, after which the system was still fully functioning.

Moreover, as pointed out in Figure 5b, once the stability of the gas evolution is obtained, the syngas composition can be easily tuned by varying potentiometric parameters (Figure S8, Supplementary Materials). This gives a potent tool for online syngas production with on-demand CO/H₂ that can be exploited to perform coupled organic reactions that require different syngas compositions. Carbonylation reaction usually requires only the presence of CO as reactant; by contrast, high-pressure and high-temperature hydroformylation require syngas with CO:H₂ equal to 1:1.

To explain the cause of syngas instability caused by BPM, Figure 8 depicts a scheme of the process hypothesized to happen at the membrane. BPM comprises a cation exchange layer (CEL) and an anion exchange layer (AEL) pressed together [37,55]. When a bias is applied to the system, water splitting occurs at the interface, and H⁺ and OH⁻ are produced in the catholyte and the anolyte, respectively (if the membrane is operated in forward bias mode, with the CEL facing the cathode). This phenomenon causes progressive acidification of the catholyte with a consequent progressive switching of the selectivity towards HER. The diffusion coefficient of the two electroactive species can also be considered in this scenario. H⁺ displays a diffusivity almost ten times more than CO₂; this would further explain the reason for the total suppression of eCO₂RR with time (Table 4). This hypothesis is strongly supported by COMSOL modeling. In fact, from Table 4 it is evident how the diffusion of H⁺ from the bipolar membrane completely takes over CO₂ adsorption at the surface, leading to almost 100% selectivity to H₂ during a −20 mA/cm² chronopotentiometry. Instead, when the AEM is in place, j_CO and j_H2 become stable after a short time due to an effective depletion of H⁺ from the catalytic surface.

4. Materials and Methods

4.1. Chemicals

The Ionic liquids (ILs) employed, i.e., 1-Butyl-3-methylimidazolium acetate ([Bmim] [CH₃CO₂]) and 1-Butyl-3-methylimidazolium trifluoromethansulfonate ([Bmim] [CF₃SO₃]) were synthesized by IoLiTec Gmbh (Heilbronn, Germany) (high purity grade) and were used without any further purification ([Bmim] cation purity > 99.4%, Triflate anion purity > 99.9%, water content in [Bmim] [CF₃SO₃] < 110 ppm, Acetate anion purity > 98%, water content in [Bmim] [CH₃CO₂] < 0.6%). The organic solvent employed, 3-methoxyporpionitrile (3-MPN), was purchased from (Merck KGaA, Darmstadt, Germany). Double-distilled Milli-Q water (R > 18.2 MΩ, 25 °C) was obtained using a Millipore Direct-Q 3 UV system (Merck KGaA, Darmstadt, Germany) and employed as the protic solvent.

4.2. Characterizations

Conductivity measurements were made by using a VWR pHenomenal MU 6100 H Multi-Parameter instrument (VWR International Srl, Milano, Italy), and the results are shown in Figure S1 in the Supplementary Materials. The ATR-IR spectra of the anolytes and catholytes (in shown in Figures S9 and S10) were collected in the range between 400 and 4000 cm⁻¹ with 4 cm⁻¹ resolution, using a Bruker TENSOR 27 spectrometer (Bruker Italia Srl, Milano, Italy).

4.3. Electrochemical Tests

The electrochemical characterization was carried out using a single-compartment electrochemical cell. Linear Sweep Voltammetry (LSV) curves were collected in a single cell with a scan rate of 10 mV/s and are reported hereafter applying the iR drop correction; the uncompensated resistance (R_u/Ω) was measured for each electrolyte via single-frequency (100 mHz) impedance measurement (ZIR). Chronopotentiometries (CPs) were conducted in both a single and an H-type cell. An Ag foil, Ag/AgCl (KCl sat.) and Pt mesh were employed as working electrode (WE), reference electrode (RE) and counter electrode (CE), respectively. Gas inlet/outlet in the cells allowed the outgassing and the saturation of the electrolytes with N₂ and CO₂. During the CPs, the gas was continuously fed into the cell at a constant rate of 20 NmL/min using a mass flow controller (Bronkhorst EL-FLOW, Bronkhorst, Ruurlo, The Netherlands), and magnetic stirring was maintained to favor the convection of electroactive species towards the electrode-electrolyte interface. In the H-type cell, either a Fumasep BPM membrane or an FAA 3 (thickness 130) µm peek-reinforced AEM were used. An aqueous solution of 0.5 M KOH was employed in the H-type cell as anolyte. All electrochemical measurements were conducted using a Biologic SP-300 potentiostat (supplied from BioLogic, Seyssinet-Pariset, France) and EC-Lab software version 11.50 (released 4 May 2023) for data collection and analysis. The online detection of gaseous products during eCO₂RR was done through a Varian 490 micro-gas chromatograph. Further details are provided in the text. All potentials (E) are referred to the Ag/AgCl reference electrode.

5. Conclusions

In this work, we investigated different conditions and setups to achieve a tunable syngas production with a constant CO/H₂ ratio. Starting with the electrolyte, we showed how an RTIL-based electrolyte enhances eCO₂RR activity. By selecting the proper mix of RTILs, it is possible to balance the two competing reactions (eCO₂RR and HER) to produce syngas with a CO/H₂ ratio constant over time. This is thanks to the specific choice of the RTILs employed: one (Bmim OAc) is more indicated for the capture and consequently HER, while the other one (Bmim OTf) is selective towards eCO₂RR to CO. Water is the preferred proton source for HER; however, eCO₂RR can occur even in low-water-concentration electrolytes (few ppm), suggesting that the acidic proton at C2 position of the imidazolium ring can act as well as a co-proton source for CO₂ reduction. These data have shown good agreement with modeling results through COMSOL Multiphysics.

When choosing the most suitable cell for the reaction, a two-compartment cell (e.g., H-cells or flow cells) is the best solution. When a single cell was used, it was impossible to maintain the two processes equally stable, with the HER overcoming the eCO₂RR in a short time. Although in previous works it was possible to produce CO stably in a single cell with FE_CO > 90% with Bmim OTf in 3-MPN as the electrolyte, herein, the addition of Bmim OAc led to oxidative phenomena of the RTIL-based electrolyte at the anode, with a negative effect in the eCO₂RR. Indeed, while controlling the potential at the cathode, the anodic potential can increase to values that may degrade the RTILs, producing by-products such as ethylene and methane, which can be wrongly attributed to the eCO₂RR and reduce the purity of the produced syngas.

The usual average higher cell potential of H-cells due to high ohmic resistances can be avoided using flow cells or better-designed H-cells with a low distance between the electrodes, even if a few works have implemented such a setup to work with RTILs-based electrolytes. Separating the anodic chamber from the cathodic chamber would prevent issues related to the degradation of products and enable the choice of the anodic reaction to lower the overall cell potential further. Besides, we have demonstrated the importance of choosing the right membrane. When a BPM was employed, we observed constant acidification of the catholyte due to water splitting at the CEL, which makes HER favored with time, in agreement with our modeling data supporting such a mechanistic hypothesis. An anionic membrane is preferred in this IL-based system to obtain stable syngas production for hours without losing catalytic activity [45].

A trade-off between the different phenomena in a specific electrochemical cell should be found to optimize the production of the target product. We demonstrated that a stable syngas production could be obtained while exploiting the advantage of the mixture of Bmim OAc and Bmim OTf as electrolytes, i.e., high CO₂ solubility and co-catalytic effect, by using a two-chamber electrochemical cell with an AEM. Depending on the application, the syngas CO/H₂ ratio can be controlled by changing the electrolyte’s water/ionic liquid ratio. The known issue of the loss of CO₂ from the catholyte by bi/carbonates crossover through the anionic membrane should be mitigated differently. For instance, the bi/bicarbonates (being oxidized to CO₂ in the anode) can be recovered by caustic washing with KOH of the anodic outgas, producing a bicarbonate solution that can be recirculated and reused as catholyte or by a pressure swing absorption process to separate the CO₂ to be recirculated in the gaseous form. The best option should be selected based on techno-economic considerations beyond this work’s scope. Herein, we offer guidelines for syngas generation in RTIL-based systems from waste-CO₂ reduction, which can be helpful for the development of novel green chemical synthesis processes.