Advancing Ionic Liquid‐Based Membrane Fabrication: Spray‐Coating as a Pathway for Mixed‐Matrix Membranes

Abstract

This work investigates two fabrication strategies, solvent casting and spray‐coating, for producing advanced IL‐based mixed‐matrix membranes (MMMs) incorporating functionalized silica IL (SiIL) microparticles, [Si][C₃C₁im]Cl. Membranes with 20 wt% of poly(diallyldimethylammonium) bis(trifluoromethylsulfonyl)imide, poly[pyr₁₁][NTf₂], PIL and 80 wt% of 1‐ethyl‐3‐methylimidazolium bis(trifluoromethylsulfonyl)imide, ([C₂mim][NTf₂]), IL with varying filler loadings were evaluated for CO₂ separation performance, including gas permeance, permeability, diffusivity, solubility, and CO₂/N₂ selectivity. Spray‐coating enabled the formation of ultrathin, selective layers, resulting in CO₂ permeances up to 27 GPU, approximately twice than without silica and up to ten times higher than solvent‐cast samples, while maintaining selectivity values, between 29 and 40. In contrast, solvent‐cast membranes exhibited lower CO₂ permeances (2–3 GPU) but benefited from uniform filler dispersion, achieving a 63% increase in CO₂ permeability at optimal SiIL loadings. Mixed‐gas experiments (15% CO₂ and 85% N₂) demonstrated that spray‐coated membranes retained most of their performance under competitive conditions, with CO₂ permeance losses of only 5%–9%, whereas solvent‐cast membranes showed more stable selectivity but lower overall flux. These results highlight that membrane fabrication methodology, rather than filler content alone, critically determines the membrane architecture and thus gas transport behavior. Overall, spray‐coating emerges as a highly effective approach to designing scalable, high‐performance MMMs for industrial CO₂ separation applications.

Spray‐coating is demonstrated as a highly effective strategy for fabricating ionic liquid‐based mixed‐matrix membranes with exceptional CO₂ separation performance. The formation of ultrathin selective layers increases CO₂ permeance by up to tenfold compared with solvent‐cast membranes and doubles permeance relative to particle‐free systems, while maintaining high CO₂/N₂ selectivity under both single‐ and mixed‐gas conditions, highlighting its potential for scalable industrial carbon capture applications.

1. Introduction

Addressing anthropogenic carbon dioxide (CO₂) emissions remains a critical challenge at the heart of global climate solutions [1]. Among the diverse strategies in carbon capture utilization and storage (CCUS) and essential CO₂ preconcentration from flue gas, natural gas, and biogas streams, the ability to selectively and efficiently concentrate CO₂ is paramount. The creation of CO₂‐philic materials is thus central to advancing these technologies. Ionic liquids (ILs) stand out due to their highly tunable chemistry and virtually negligible vapor pressure [2], allowing precise modulation of both physical and chemical interactions with CO₂ by integrating functional groups such as amines, fluorinated chains, or other polar moieties [3], alongside the advantage of low volatility [4].

IL‐based membranes have generated significant research momentum, as they combine the inherent advantages of membrane separations with the designer properties of ILs. Demonstrated across multiple studies, these materials show promise for large‐scale applications, including CO₂ extraction from flue gas [5], natural gas purification [6], and biohydrogen upgrading [7]. Polymeric ILs (PILs) further expand this platform, providing a processable solid‐state matrix whose properties can be tailored for specialized separations. High‐efficiency standards drive the membrane field, and to meet the escalating benchmarks required for CO₂ capture, property enhancements are continually pursued.

A leading strategy in the membrane field is the incorporation of fillers, giving rise to mixed‐matrix membranes (MMMs) [8]. MMMs merge the molecular selectivity of polymers with the superior permeability of inorganic additives. For optimal performance, fillers should possess high selectivity for CO₂ over competing gases, disperse uniformly, and yield membrane morphologies favorable to industrial deployment. Micro‐ and nanostructured materials, carbon nanotubes, graphene, silica particles, zeolites, and metal–organic frameworks (MOF), have all been leveraged to augment both mechanical and separation characteristics of membranes. These fillers create new gas transport pathways via physical interactions, increased free volume, molecular sieving, and changes to polymer rigidity, and their inclusion at the nanoscale can vastly improve compatibility and CO₂ separation efficiency [9].

Breakthroughs in high‐performing MMMs have stemmed from innovative fabrication methods. Micro‐ and nanomaterials have been instrumental in propelling the development of MMMs, significantly improving both mechanical robustness and gas separation capabilities. Advanced fillers such as carbon nanotubes, graphene, zeolites, and MOFs have been widely investigated for their remarkable ability to increase membrane performance [10]. These additives boost gas permeability through several synergistic mechanisms, including enhanced interactions with gas molecules, increased free volume, precise molecular sieving effects, and strategic modifications to polymer rigidity. Cutting‐edge research has revealed that the integration of single or combined nanoscale fillers within polymer matrices not only strengthens polymer‐filler compatibility but also delivers notable advances in CO₂ separation efficiency [11]. For example, Kim et al. [12] developed a thin‐film composite (TFC) MMM integrating IL and ZIF‐8 into a PIL matrix, combining polymer adhesion and improved CO₂ solubility imparted by the IL, and molecular sieving from ZIF‐8 for outstanding separation results. Zhao et al. [13] utilized solution mixing to embed 13 wt% ZIF‐8 in PVAm, supported on polysulfone, creating defect‐free, highly selective MMMs with 325% improved CO₂ permeability and a 65% increase in CO₂/N₂ selectivity versus neat membranes. Similarly, Wong et al. [14] showed that trace loadings of β‐cyclodextrin‐functionalized carbon nanotubes within PEG/PES blends deliver striking increases in both CO₂ permeance and selectivity by optimizing filler dispersion and interfacial compatibility. Their work also demonstrates how excessive loading can diminish performance, underscoring the necessity of careful filler optimization. Modified nanofillers not only improve transport but also bolster resistance to fouling and recyclability by enhancing interfacial adhesion and membrane robustness [15]. In the field of PIL‐based membranes, Atlaskina et al. [16] reported imidazolium‐based PIL homopolymers and block copolymers for combined CO₂ capture and catalytic conversion. Their study showed that membrane performance strongly depended on the nature of the counter‐anion, with chloride‐containing systems exhibiting CO₂/N₂ selectivity of 18–22 while iodide‐based membranes reached selectivity up to 30. However, this improvement in selectivity for iodide‐based membranes was accompanied by substantial reduction in CO₂ permeability (∼7.5 Barrer), highlighting the classic permeability‐selectivity tradeoff and the critical role of anion chemistry in regulating the free volume and gas transport pathway. Martin et al. [17] investigated the effect of the addition of IL functionalized silico‐metallic particles to Pebax1657‐based MMMs, showing that filler addition significantly enhances CO₂ transport, increasing from CO₂ permeability from 84.21 Barrer for neat Pebax to 192.17 Barrer at 20 wt% filler loading, while maintaining high selectivity. Notably, the highest CO₂/N₂ selectivity reached 91.96 at low filler loading (0.5 wt%), and all MMMs surpassed the 2008 Robeson upper bound. While separation performance is crucial, the practical value of any CO₂ capture technology also depends on its environmental footprint. Compared with absorption, which is most industrially used carbon capture technology, membrane‐based separations are inherently more energy‐efficient because they avoid phase changes and do not rely on volatile chemical solvents. Moreover, IL‐based membranes combine the attractive properties of membranes with those of ILs, namely the high CO₂ affinity and the negligible vapor pressure. The introduction of silica fillers in IL‐based membranes further enhance the CO₂ separation performance due to the complex membrane structure and increased free volume [18]. Silica is abundant, chemically stable, and can be functionalized using relatively mild conditions [19]. In addition, achieving CO₂ separation high performance at low filler loadings reduces material use and supports longer membrane lifetimes. Together, these features make PIL–IL MMMs using silica a practical platform for CO₂ separation that aligns with the goals of sustainable and green chemical processing without sacrificing separation efficiency.

Against this background, scalability and process efficiency become just as important as materials design. Traditional MMM fabrication methods, including solution casting, phase inversion, and interfacial polymerization, often involve high solvent use, prolonged processing, and uneven particle distribution. In contrast, spray‐coating overcomes these barriers by minimizing solvent consumption, accelerating film formation, and enabling uniform, reproducible selective layers that are essential for reliable gas separation performance [9]. Furthermore, spray‐coating also offers exceptional scalability for industrial use, precisely controls film deposition and filler dispersion [20], and dramatically cuts processing times thanks to rapid evaporation and film formation [21]. These advantages position spray‐coating as a highly attractive and sustainable membrane fabrication strategy, enabling membrane architectures that deliver separation performances comparable to or exceeding those of traditionally made membranes, while improving process efficiency and scalability [22]. Such findings are increasingly being reported in the recent literature on mixed‐matrix and IL‐based membrane systems. For example, Wang et al. [23] developed a layer‐by‐layer spray‐coated sandwich membrane composed of confined CuBDC MOF nanosheets and [BMIM][BF₄] IL within interlayers for hydrogen purification. Their design successfully sealed nonselective defects and created CO₂‐philic transport domains. As a result, the membrane achieved an H₂ permeance of 1904 GPU and H₂/CO₂ selectivity of 12.6. Similarly, Neves et al. [24] reported Pebax1657 MMMs combining ZIF‐8 MOFs with [EMIM][Tf₂N] IL, which were further developed into TFC membranes using spray‐coating, without the need for an intermediate gutter layer. The optimized membrane, containing Pebax1657, 30 wt% ZIF‐8, and 30 wt% [EMIM][Tf₂N], showed CO₂ permeability of 148.9 barrer and CO₂/N₂ selectivity of 73.8, when tested with a mixture of 15% CO₂ + 85% N₂ at 3 bar and 30 °C at 0%relative humidity. This study demonstrates that spray‐coating enables scalable fabrication of high‐performance MMM architectures.

In this work, spray‐coating was used to fabricate polymerized IL–IL MMMs (PIL–IL MMMs) incorporating IL–functionalized silica fillers. This approach aims to produce thin, mechanically robust membranes with uniform silica particle dispersion with enhanced CO₂ permeability and selectivity, while reducing material waste.

The membrane matrix consists of the polymerized IL poly(diallyldimethylammonium) bis(trifluoromethylsulfonyl)imide (poly[pyr₁₁][NTf₂]) and the IL 1‐ethyl‐3‐methylimidazolium bis(trifluoromethylsulfonyl)imide ([C₂mim][NTf₂]). These compounds were selected due to the extensive experience of our group with this system [3, 25, 27], its well‐documented stability, and its consistently reported high CO₂ permeability combined with excellent CO₂/N₂ selectivity [3, 26, 28, 29]. Silica particles were chosen as fillers because of their low cost and widespread availability. To improve compatibility with the PIL–IL matrix, the silica surface was functionalized with ILs, [Si][C₃C₁im]Cl, promoting strong interfacial interactions, homogeneous dispersion, and improved membrane uniformity. The results here shown demonstrate that spray‐coating significantly outperforms conventional solvent casting, achieving an increase in CO₂ permeance of approximately one order of magnitude. Additionally, the separation performance under mixed‐gas conditions was evaluated, providing a more realistic assessment of membrane behavior. Emphasis was placed on the use of ultra‐low filler loadings to prevent particle aggregation and mechanical degradation, enabling the development of membranes that simultaneously achieve high separation performance, mechanical integrity, sustainability, and cost‐effectiveness.

2. Experimental Section

2.1. Materials

Poly(diallyldimethylammonium) bis(trifluoromethylsulfonyl)imide (poly[pyr₁₁][NTf₂]) was synthesized via an anion exchange reaction using commercially available precursors: poly(diallyldimethylammonium chloride) (average molecular weight 400,000–500,000; 20 wt% aqueous solution) and lithium bis(trifluoromethylsulfonyl)imide (LiNTf₂, 99%). These chemicals were obtained from Sigma‐Aldrich and Iolitec, respectively, and used as received. Additionally, 1‐ethyl‐3‐methylimidazolium bis(trifluoromethylsulfonyl)imide ([C₂mim][NTf₂], 99%) was purchased from Iolitec.

Gases used for permeation tests—CO₂, nitrogen (N₂), and methane (CH₄)—were supplied by Air Liquide with a minimum purity of 99.99%, and were used without further purification. The ILswere supported on porous, hydrophobic poly(vinylidene fluoride) (PVDF) Durapore membranes, characterized by a pore size of 0.22 μm and an average thickness of 125 μm, provided by Merck Millipore Corporation.

Reagents used for IL‐functionalized silica preparation included silica gel 60 (0.2–0.5 mm, Merck), 3‐chloropropyltrimethoxysilane (≥98.0%, Acros Organics), N‐methylimidazole (99.0%, Acros Organics) (ethanol (≥99.8%, Fluka), hydrochloric acid (37.0% w/w, Honeywell), methanol (HPLC grade, Fisher Scientific), and toluene (99.8%, Fisher Scientific).

2.2. Synthesis of Poly[pyr₁₁][NTf₂]

The polymer containing pyrrolidinium pendant groups was synthesized following a well‐established anion exchange protocol [30, 31]. In this process, poly(diallyldimethylammonium) chloride was reacted with lithium bis(trifluoromethanesulfonyl)imide (LiNTf₂) in aqueous solution, resulting in the incorporation of the bis(trifluoromethanesulfonyl)imide ([NTf₂]⁻) anion. This exchange caused the new polymer to precipitate, after which it was thoroughly washed, filtered, and dried to a constant weight.

2.3. Synthesis of Functionalized Silica Particles

The IL‐functionalized silica particles, [Si][C₃C₁im]Cl, were prepared through a two‐step reaction process. First, activated silica particles were reacted with the silane coupling agent 3‐chloropropyltrimethoxysilane. Subsequently, the resulting chloropropyl‐functionalized silica was reacted with N‐methylimidazole as the cation source. This synthesis procedure was adapted from Qiu et al. [32, 33] and was more recently optimized and validated by our group [19]. Detailed characterization of the resulting materials can be found in the literature [34, 35].

2.4. Fourier‐Transform Infrared (FTIR) Spectroscopy

FTIR‐ATR spectra of the neat PILs and ILs and their respective composites were acquired using a Thermo Electron Corporation Nicolet 5700 spectrometer with a DTGS TEC detector. All spectra were collected using 28 scans from 500 to 4000 cm⁻¹.

2.5. Membrane Preparation

For consistency, membrane compositions are designated as XPIL–YIL–ZSiIL, where PIL refers to poly[pyr₁₁][NTf₂], IL to [C₂mim][NTf₂], and SiIL to the functionalized silica particles [Si][C₃mim]Cl. Thus, a membrane containing 20 wt% PIL, 80 wt% IL, and 1 wt% SiIL is abbreviated as 20PIL–80IL–1SiIL throughout the manuscript.

2.5.1. Mixed‐Matrix Membranes by Solvent Casting

The MMMs were prepared by solvent casting via slow evaporation at room temperature of 12% (w/v) acetone solution of a PIL/IL mixture with the filler particles dispersed. Initially, the PIL (poly[pyr₁₁][NTf₂]) and IL ([C₂mim][NTf₂]) were mixed in appropriate mass proportions, then dissolved in half the final volume of acetone. In parallel, the IL functionalized silica particles were dispersed in the remaining acetone, in a separate vial. Both mixtures were stirred separately for 8 h, followed by 45 min of sonication to ensure homogeneity.

To further improve dispersion, a stepwise protocol was employed to mix the two solutions: first, 10% of the mass of the polymer–IL solution was added to the filler dispersion, stirred for 4 h and sonicated for 5 min. After sonication, 45% of the mass of the PIL–IL solution was introduced and stirred for another 4 h. Finally, the remaining polymer–IL solution was added, and the final blend was stirred for 24 h at room temperature (∼20 C). The prepared homogeneous solution was cast onto the PVDF support placed in a petri dish and left covered at room temperature, to allow for slow acetone evaporation. After 48 h, the resulting membrane was carefully detached with a spatula.

2.5.2. Thin Film Mixed‐Matrix Membranes by Spray‐Coating

The membrane solution for spray‐coating was prepared using the same methodology as that used for the preparation of MMMs, though only 5% (w/v) acetone was used. The solution is loaded into the spray‐coating setup rather than cast on a petri dish. This thin film MMM were fabricated using a spray‐coating setup with the substrate thermostated at 45 °C. The membrane support was rotated continuously during deposition to ensure uniform coverage. The spray nozzle was kept 10 cm from the surface. Ten spray cycles were performed, each consisting of 5 s of spraying followed by a 15‐s pause to allow partial solvent evaporation and the stabilization of each layer. This pulsed deposition built up a uniform and defect‐free membrane. After the tenth cycle, the membrane remained rotating for an additional 3 min at the same temperature to ensure complete drying and solvent removal. This method produced a dense, homogeneous selective layer designed for high‐performance separation.

2.6. Membrane Characterization

Microscopy analyses of the integral asymmetric PIL–IL membranes were performed using a JEOL 7001F FEG–scanning electron microscopy (SEM) equipped with Oxford light elements EDS and EBSD detectors. Samples were fractured, mounted on stubs, and sputter‐coated with gold before imaging. Both surface and cross‐sectional SEM images were obtained.

FTIR spectra were recorded using a PerkinElmer Spotlight 200i FTIR Microscope System. Spectra were collected over the range of 400–4000 cm⁻¹, with a resolution of 4 cm⁻¹, using eight scans per measurement.

2.7. Gas Permeation Experiments

Permeation measurements for CO₂ and N₂ were conducted at 293.15 K using the time‐lag apparatus, a standard method for evaluating membrane permeability, diffusivity, and solubility from dynamic single‐gas experiments [36]. Prior to analysis, each membrane was degassed under vacuum within the permeation setup for 24 h to eliminate residual solvents or trapped gases. For each membrane, three separate measurements were performed with both gases. Between runs, samples were degassed until the downstream pressure dropped below 0.1 kPa. Membrane thickness was averaged from six different locations, both before and after testing.

Transport across the membranes was interpreted using the solution–diffusion model [37]. Upon obtaining permeability (P) and diffusivity (D) values, solubility (S) was calculated as their ratio, according to Equation (1).

$$P=D\times S$$ (1)

Additionally, the ideal gas permselectivity (α) was defined as the ratio of the permeabilities for gases A and B, reflecting the membrane’s discrimination ability between the two species [31]. Permselectivity could also be expressed as the product of diffusivity selectivity and solubility selectivity, as outlined in the subsequent Equation (2)

$${\alpha }_{A/B}=\frac{P_A}{P_B}=\left(\frac{D_A}{D_B}\right)\times \left(\frac{S_A}{S_B}\right)$$ (2)

Furthermore, two membranes corresponding to the highest permeabilities within each fabrication method, namely solvent casting and spray‐coating, were selected for mixed‐gas permeation studies. In this case, gas transport properties under mixed‐gas conditions were evaluated using a steady‐state membrane permeation system operating in continuous‐flow mode [25]. The membrane cell separated the incoming gas mixture into permeate and nonpermeate streams, with helium continuously introduced on the downstream side to remove permeating species. Independent electronic mass flow controllers regulated all inlet and outlet flow rates, while pressure sensors positioned downstream of the membrane provided real‐time pressure measurements. The membrane cell was maintained at a constant temperature of 20 °C inside a controlled temperature chamber. The exposed membrane surface area was 7.1 cm².

For all experiments, the feed gas flow rate was fixed at 25 cm³ (STP) min⁻¹. The permeate side of the cell was purged with helium at a flow rate of 7.25 cm³ (STP) min⁻¹ to ensure effective sweeping of studied gases. Under the studied operating conditions, the permeate flow rate was approximately 0.03 mL min⁻¹. The chemical composition of the permeate and retentate streams was continuously monitored using gas chromatography (Thermo Scientific Trade 1300) coupled with a thermal conductivity detector. Measurements were recorded at predefined time intervals until stable permeation behavior was observed, typically after 1–2 h. Final permeability values were obtained by averaging three consecutive steady‐state measurements, with variability consistently below 5%.

Component permeabilities were determined from the measured permeate flow rates and gas compositions according to Equation (3)

$$P_i=\frac{Q_{\text{perm}\text{,}\mathit{\text{T}}}\cdot x_{\text{perm}\text{,}\mathit{\text{i}}}\cdot l}{A\cdot \mathrm{\Delta }P}=\frac{Q_{\text{perm}\text{,}\mathit{\text{T}}}\cdot x_{\text{perm}\text{,}\mathit{\text{i}}}\cdot l}{A\cdot (p_{\mathrm{feed}}\cdot x_{\text{feed}\text{,}\mathit{\text{i}}}-p_{\mathrm{perm}}\cdot x_{\text{perm}\text{,}\mathit{\text{i}}})}$$ (3)

In this expression, Q _perm,T denotes the total volumetric permeate flow, x _perm,i and x _feed,i are the mole fractions of species i in the permeate and feed streams, respectively, l is the membrane thickness, and A is the effective membrane area. The parameters p _feed and p _perm represent the total pressures on the upstream and downstream sides of the membrane.

Each mixed‐gas experiment was repeated at least three times to ensure statistical reliability. Selectivity under mixed‐gas conditions was calculated by dividing the permeability of one gas component by that of the other, consistent with the definition given in Equation (2).

3. Results

This section presents the results obtained for the development and optimization of PIL‐based MMMs for CO₂ separation. The study begins by evaluating solvent casting membranes and the IL effect on gas permeation properties, followed by the influence of silica particle loading to identify the most promising membrane composition. The role of silica particles in the gas transport mechanism is further investigated through CO₂ solubility and diffusivity analyses. Subsequently, spray‐coating is explored as an alternative fabrication method and its impact on membrane performance is assessed. The membranes prepared by solvent casting and spray‐coating are then characterized and compared to evaluate the influence of the fabrication technique. Finally, mixed‐gas permeation experiments are performed using CO₂/N₂ mixtures to assess membrane performance under more realistic conditions, and the obtained results are benchmarked against representative data from the literature.

3.1. Effect of IL Content on Gas Permeation Properties of MMM by Solvent Casting

To determine whether increasing the IL fraction in the XPIL–XIL membrane matrix enhances particle incorporation and gas permeation, membranes with two IL contents (60PIL–40IL and 20PIL–80IL) were fabricated by solvent casting. Each formulation was combined with four silica filler loadings (0, 0.01, 0.1, and 5 wt%), cast into membranes, and evaluated using the time‐lag permeation method. At 293.15 K and a differential pressure of 1 bar, permeability (P), diffusivity (D), and solubility (S) coefficients for CO₂ and N₂ were measured. The results (see Figure 1, Tables S1 and S2) reveal the influence of IL content and silica loading on gas transport performance.

FIGURE 1.

CO₂ permeability (bars, right axis) and CO₂/N₂ ideal permselectivity (circle, left axis) of 60PIL‐40IL‐XSiIL (dotted bars and triangles) and 20PIL‐80IL‐XSiIL (full bars and circles) MMMs, prepared by solvent casting, with various amounts of functionalized silica particles.

For all contents of SiIL filler, membranes with higher IL content (20PIL–80IL) displayed significantly greater CO₂ permeability than those with lower IL content (60PIL–40IL), underscoring the dominant influence of the IL fraction over filler loading in boosting gas transport. This effect was especially pronounced in 20PIL–80IL membranes, where enhanced chain mobility and improved particle dispersion in the more fluid matrix favored gas diffusion.

At equal filler loadings, increasing IL content reliably raised CO₂ permeability, while CO₂/N₂ selectivity remained largely unchanged. Within the 20PIL–80IL series, adding more silica led to a marked increase in CO₂ permeability—a relationship absent in the 60PIL–40IL group.

At the highest filler loading (5 wt% SiIL), the selectivity of the two systems diverged: 20PIL–80IL membranes maintained high CO₂/N₂ selectivity (∼33) across all loadings, indicating robust and defect‐free structures, whereas selectivity in 60PIL–40IL membranes dropped sharply (∼10), suggesting poor dispersion and increased nonselective pathways.

Overall, higher silica content integrates more effectively into IL‐rich matrices, yielding membranes that are both stable and selective. The notable improvement in CO₂ permeability for high‐IL content membranes is mainly due to enhanced diffusivity; solubility remained essentially constant across all groups.

It should be noted that despite the mechanical stabilization offered by the polymer, the 20PIL–80IL MMMs are gelatinous and fragile, requiring PVDF support. Nevertheless, their superior CO₂ permeation performance makes this composition especially promising for further development, striking an excellent balance between gas separation efficiency and processability when supported.

To further assess the operational stability of the prepared membranes, downstream pressure evolution plots obtained during the time‐lag measurements are included in the Supporting Information (Figures S1 and S2). These results show stable and reproducible pressure profiles throughout the experiments, with no observable drift either within individual runs or between consecutive measurements. The absence of significant pressure fluctuations indicates that the membranes maintained their structural integrity under the applied differential pressure conditions. Furthermore, the reproducibility of the CO₂ and N₂ permeation measurements, including repeated CO₂ tests performed after N₂ permeation experiments, confirms the stability of the membrane performance over the duration of the measurements. It should also be noted that, unlike supported IL membranes, where the IL is physically confined within the pores of a support and may be susceptible to IL displacement or loss under the operational conditions, the spray‐coated membranes developed in this work are composite PIL–IL membranes, in which the IL is immobilized within a polymerized IL network. This configuration provides enhanced mechanical and operational stability during gas permeation experiments.

3.2. Effect of Silica Particles on Gas Permeation Properties of MMM by Solvent Casting

The effect of functionalized silica particles ([C₃mim][Cl], SiIL) on gas transport was systematically investigated in 20PIL–80IL MMMs by solvent casting. SiIL was incorporated at concentrations ranging from 0 to 10  wt% (0, 0.007, 0.01, 0.1, 1, 5, and 10  wt%) to enhance CO₂ permeability by fostering efficient gas transport pathways while preserving membrane stability. The SiIL particles were uniformly dispersed during solvent casting, and the resulting changes in CO₂ permeability and CO₂/N₂ permselectivity were recorded (see Figure 2 and Table S3). Increasing SiIL content from 0 to 10 wt% led to a substantial rise in CO₂ permeability, from 498 to 811 Barrer. However, membranes with 20 and 30  wt% SiIL proved mechanically brittle and unsuitable for practical use. Representative photographs of membranes containing different SiIL loadings are provided in the Supporting Information (Figure S3). A comparison between a stable membrane (10 wt% SiIL) and a failed membrane (30 wt% SiIL) clearly illustrates the detrimental effect of excessive filler incorporation on membrane integrity. While lower filler loadings produced homogeneous and flexible membranes suitable for handling and testing, higher SiIL content resulted in brittle materials that were prone to cracking and fracture, thereby limiting their practical applicability.

FIGURE 2.

CO₂ permeability (bars, green, right axis), N₂ permeability (bars, blue, right axis), and CO₂/N₂ ideal permselectivity (circle, left axis) of 20PIL‐80IL‐XSiIL membranes, prepared by solvent casting, with various amounts of functionalized silica particles.

The unfilled (0  wt% SiIL) membrane displayed high intrinsic CO₂ permeability (498 Barrer), reflecting strong baseline diffusivity and solubility. Introducing a small amount of filler (0.007 wt% SiIL) led to a slight increase in CO₂ permeability (511 Barrer) but also caused a notable drop in selectivity (from 38 to 29), likely due to nonselective pathways created by increased free volume around dispersed particles.

Further increasing the SiIL loading to 1 wt% simultaneously improved both permeability (601 Barrer) and CO₂/N₂ permselectivity (40). With even higher SiIL contents, CO₂ permeability continued rising, and at the highest tested level (10  wt%), reaching a peak at permeability and selectivity of about 850 Barrer and 35, respectively. This unexpected synergistic gain suggests that high filler content, combined with slow solvent evaporation, promoted homogeneous particle distribution and strong polymer‐filler interactions, yielding highly selective gas transport pathways resulting in a 70% increase in permeability versus the neat membrane.

Attempts to use filler contents above 10 wt% (such as 20 and 30 wt%) resulted in membranes that were mechanically unstable, indicating an upper limit for usable SiIL content.

Unlike CO₂ transport, N₂ permeability was relatively constant across all SiIL concentrations—rising slightly from 13 to 23.1 Barrer up to 10 wt%. This stability, together with increasing CO₂ permeation, led directly to maintain the high selectivity.

3.3. Effect of Silica Particles on CO₂ Solubility and Diffusivity Properties

The impact of SiIL filler incorporation on CO₂ solubility and diffusivity in 20PIL–80IL MMMs was assessed (see Figures 3, 4, and Table S4). Adding a small amount of SiIL reduced CO₂ solubility from 12 to 8 m³(STP) m⁻³ Pa⁻¹ × 10⁶ (for 0 to 0.007 wt%), with this lower value remaining nearly constant at higher filler loadings, aside from a slight uptick at 10 wt%. This decline is attributed to dilution of the CO₂‐philic sites within the IL as more inorganic filler is added. N₂ solubility followed a similar trend. Notably, solubility selectivity closely mirrored the observed overall permselectivity, indicating that solubility differences are likely the dominant factor controlling membrane selectivity.

FIGURE 3.

CO₂ solubility (bars, green, right axis), N₂ solubility (bars, blue, right axis), and CO₂/N₂ solubility selectivity (circle, left axis) of 20PIL‐80IL‐XSiIL membranes, prepared by solvent casting, with various amounts of functionalized silica particles.

FIGURE 4.

CO₂ diffusivity (bars, green, right axis), N₂ diffusivity (bars, blue, right axis), and CO₂/N₂ diffusivity (circle, left axis) of 20PIL‐80IL‐XSiIL membranes, prepared by solvent casting, with various amounts of functionalized silica particles.

In contrast to solubility, CO₂ diffusivity increased sharply with the addition of SiIL fillers, especially at ultra‐low concentrations (0.007 and 0.01 wt%), rising from 292 to 990 and 758 × 10⁻¹² m² s⁻¹ for 0, 0.007, and 0.01 wt%, respectively. At these loadings, enhanced surface area, stronger interfacial interactions, tailored pore structures, and increased free volume from SiIL created more efficient CO₂ diffusion pathways. Even at higher filler contents (1–10 wt%), CO₂ diffusivity remained elevated compared to the neat membrane. These results confirm that morphological changes from filler incorporation—by boosting CO₂ diffusivity—drive the observed permeability increases, even with lower solubility. N₂ diffusivity showed similar values to CO₂, yielding a diffusivity selectivity of 1–2. Thus, solubility remains the primary factor determining overall membrane selectivity, despite potential pathway formation.

3.4. Impact of Spray‐Coating on the Gas Transport Properties of 20PIL–80IL

Moving beyond the general impact of filler incorporation, we examined the spray‐coating technique, which provides advantages over traditional solvent casting by enabling more precise particle deposition and formation of uniform superficial layers. This method can optimize gas transport pathways, as highlighted in Figure 5, which presents CO₂ permeance and CO₂/N₂ selectivity for spray‐coated membranes across a range of SiIL loadings. The PVDF support exhibits a highly porous structure that provides mechanical stability while offering negligible resistance to gas transport. Gas permeation measurements were not performed for the neat PVDF support, as its absence of a dense selective layer prevents meaningful gas separation and does not allow a representative comparison with the prepared PIL–IL membranes.

FIGURE 5.

CO₂ permeance (bars, orange, right axis), N₂ permeance (bars, yellow, right axis), and CO₂/N₂ ideal permselectivity (circle, left axis) of 20PIL‐80IL‐XSiIL membranes, prepared by spray‐coating, with various amounts of functionalized silica particles.

Remarkably, adding just 0.01 wt% SiIL via spray‐coating nearly doubled the CO₂ permeance (from ∼13.5 to ∼24.5 GPU), likely due to increased surface area, additional voids, and efficient gas pathways. While a 1 wt% SiIL loading yielded the highest absolute CO₂ permeance (∼27 GPU), the most significant impact occurs at the 0.01 wt% level, where permeance jumps by ∼11 GPU; beyond this point, increasing the filler content produces only marginal improvements, with permeance values remaining relatively stable. This suggests either a morphological saturation point or highly uniform dispersion that minimizes agglomeration.

CO₂/N₂ ideal permselectivity also remained steady (around 20–36) across all SiIL loadings studied. Unlike solvent casting, which can allow filler sedimentation and aggregation during slow drying, spray‐coating, thanks to controlled deposition and intermittent drying, facilitates uniform, superficial particle distribution and maintains consistent separation performance even at higher filler contents.

3.5. Characterization of Solvent Casting and Spray‐Coating 20PIL‐80 IL MMMs

FTIR spectra were collected for [Si][C₃C₁im]Cl particles, various solvent‐cast MMMs (60PIL–40IL–0.01SiIL, 20PIL–80IL–1SiIL, 20PIL–80IL–10SiIL), and spray‐coated MMMs (20PIL–80IL–0.01SiIL, 20PIL–80IL–1SiIL) to verify successful silica functionalization and membrane composition, presented in Figures S4–S7 and Tables S5–S7. In [Si][C₃C₁im]Cl, the Si–OH bending mode at 960 cm⁻¹ disappeared, confirming the functionalization. FTIR spectra for PVDF supports displayed characteristic α and β phase bands, with the β phase at 1431, 1404, 1271, 1162, and 838 cm⁻¹, and the α phase at 1073, 876, and 760 cm⁻¹; overlapping peaks from the support were also apparent in the membrane spectra.

MMM spectra showed all [NTf₂]⁻ anion vibrational bands, confirming the incorporation of both PIL and IL. Key features included asymmetric/symmetric SO₂ stretches at 1347 and 1130 cm⁻¹, SO₂ bending at 612/508 cm⁻¹, S–N–S stretching at 841 cm⁻¹, S–N–S bending at 650 cm⁻¹, C–S stretching at 1050/790 cm⁻¹, and CF₃ signals from 1167–1227 cm⁻¹ and 573 cm⁻¹, alongside CH₃ and C–H vibrations for the IL cation. Spectra for the spray‐coated membranes were similar, showing matching bands with variations in intensity but no major wavelength shifts; additional PVDF vibrations appeared at 1401 cm⁻¹ (β) and 879 cm⁻¹ (α).

No new or shifted bands were observed for PIL or IL after filler addition, indicating that the particles did not preferentially interact with either component.

SEM was used to evaluate the surface and cross‐sectional morphologies of the membranes and is presented in Figure 6. For all solvent‐cast membranes with a 20PIL–80IL composition and varying silica loadings, the membrane surfaces appeared uneven and wrinkled, which can be attributed to the gel‐like behavior of the casting solutions. However, cross‐sectional images revealed dense and uniform membrane interiors, indicating good incorporation of the silica particles within the polymer matrix. The fine silica particles (approximately 5–20 µm) were not distinctly visible on the surface of the 20PIL–80IL–0.01 wt% and 20PIL–80IL–1 wt% membranes due to the low particle content, whereas at higher loading (20PIL–80IL–10 wt%), the presence of silica particles became clearly visible. Regarding spray‐coated membranes, to facilitate the identification of the membrane architecture, the cross‐sectional SEM images were placed in Supporting Information, Figure S8, to distinguish the porous PVDF support layer from the spray‐coated selective membrane layer.

FIGURE 6.

SEM images of 20PIL–80 IL‐XSiIL MMMs with different amounts of functionalized silica particles, fabricated by solvent casting and spray‐coating.

In contrast to the solvent‐casted membranes, spray‐coated membranes exhibited distinct morphological changes as a function of silica loading. At low SiIL loadings (0.01 and 1 wt%), the membrane surfaces appeared relatively rough, although still homogeneous, indicating uniform film formation. As the SiIL loading increased to 10 wt%, the membrane surface became noticeably smoother. This smoothing effect can be probably associated with possible particle agglomeration and sedimentation phenomena during the rapid evaporation of acetone inherent to the spray‐coating process. Similar to the solvent‐cast membranes, silica particles were not visible in the spray‐coated membranes at low filler loadings (0.01 and 1 wt%), while multiple particles could be clearly observed at the highest loading (10 wt%). Cross‐sectional SEM images of the spray‐coated membranes consistently revealed a bilayer architecture, comprising an approximately 20 µm thick dense selective layer deposited onto a porous PVDF support, regardless of SiIL loading.

3.6. Comparative Analysis of Fabrication Methods

To deepen the analysis, a direct comparison was made between membranes prepared by solvent casting and spray‐coating with the composition 20PIL–80IL–XSiIL. For consistency and comparability, all permeability values from the solvent‐cast MMMs were converted to permeance (GPU), facilitating a side‐by‐side performance assessment. The resulting data are summarized in Figure 7.

FIGURE 7.

Comparison of CO₂ permeance (bars) and CO₂/N₂ ideal permselectivity (symbols) for 20PIL–80IL MMM, with several SiIL loadings, produced using either solvent‐casting method (green and blue) or spray‐coating method (orange and yellow).

Figure 7 illustrates the comparative performance of 20PIL–80IL–XSiIL membranes produced by solvent casting and spray‐coating across various silica loadings (see Table S5). Two clear and distinct trends emerge when comparing the two fabrication methods. Solvent‐cast membranes exhibit consistently low gas permeances, with CO₂ permeance remaining in the narrow range of 2–3 GPU across all silica concentrations and N₂ permeance remaining negligible. In contrast, spray‐coated membranes display markedly higher permeances, with CO₂ permeance increasing substantially to values between 13 and 27 GPU. The highest CO₂ permeance (∼27 GPU) is observed at 1 wt% silica and remains stably elevated up to 10 wt%, indicating that silica loading has a pronounced effect only when combined with the spray‐coating process.

Despite these large differences in permeance, both fabrication methods maintain stable and comparable CO₂/N₂ selectivity. N₂ permeance stays near zero in all cases, confirming that the enhanced CO₂ transport achieved through spray‐coating does not compromise separation efficiency. CO₂/N₂ selectivity values span a relatively narrow range from 29 to 40 for both membrane types, with spray‐coated membranes exhibiting selectivities similar to or slightly higher than those of solvent‐cast membranes. Overall, these results highlight the critical role of fabrication method: while silica addition has minimal impact on solvent‐cast membranes, it significantly enhances gas transport in spray‐coated membranes, particularly at low filler contents, enabling an order‐of‐magnitude improvement in CO₂ permeance while preserving strong and stable selectivity.

3.7. Mixed‐Gas Permeation

To assess separation performance under more realistic operating conditions, mixed‐gas permeation experiments were carried out on membranes with 0.01 and 10 wt% SiIL for both fabrication methods. This choice was due to 0.01 wt% membranes showing the largest initial increase in CO₂ permeance compared to the 0 wt% baseline, highlighting the most significant effect of filler addition, while the 10 wt% membranes were included to evaluate whether very high filler loading influences performance under mixed‐gas conditions, where competitive sorption and diffusion could play a role. This selection thus allows us to capture both the maximum incremental improvement and the potential impact of high particle content.

The selected membranes were evaluated using a 15% CO₂/85% N₂ gas mixture at a transmembrane pressure difference (Δp) of 1 bar and a temperature of 20 °C. As expected, CO₂ permeance decreased under mixed‐gas conditions due to the lower CO₂ partial pressure and competitive sorption with N₂, which reduces both CO₂ solubility and diffusivity within the membrane. Similarly, CO₂/N₂ selectivity declined relative to single‐gas values, as the presence of N₂ diminishes the effective separation efficiency. This loss in selectivity is a commonly reported phenomenon in mixed‐gas permeation experiments and was anticipated based on previous studies conducted by our group under comparable conditions [25].

From the graphical results in Figure 8, and Table S8—green bars representing solvent‐cast membranes and orange bars representing spray‐coated membranes, with full bars for single‐gas data and patterned bars for mixed‐gas data—it is apparent that the spray‐coated membranes maintain the most consistent CO₂ permeances under mixed‐gas conditions. Specifically, CO₂ permeance losses were minimal for spray‐coated membranes, only 9% and 5% for 0.01 and 10 wt% SiIL, respectively, indicating that the uniform particle dispersion and optimized gas pathways created by spray‐coating preserve transport even under competitive conditions. In contrast, CO₂/N₂ selectivity remained more stable for solvent‐cast membranes, with losses of 22% and 43% for 0.01 and 10 wt% SiIL, respectively, suggesting that while permeance is low, the slower transport in solvent‐cast membranes makes the selectivity less sensitive to competition between gases.

FIGURE 8.

Comparison of CO₂ permeance for single gas measurements (full bars), CO₂ permeance for mixed‐gas measurements (patterned bars), and CO₂/N₂ ideal permselectivity or CO₂/N₂ selectivity (symbols) for 20PIL‐80IL MMM, with 0.01% and 10% SiIL loadings, produced using either solvent‐casting method (green) or spray‐coating method (orange).

Overall, these observations can be explained by the differences in morphology and gas transport mechanisms between the two fabrication methods. Spray‐coated membranes feature well‐dispersed superficial filler particles and controlled thin layers, which maintain high CO₂ flux even in the presence of N₂. Solvent‐cast membranes, in contrast, are denser and may allow slight filler aggregation, leading to inherently lower permeances but relatively more stable selectivity under mixed‐gas conditions. Thus, the choice of membranes for mixed‐gas testing effectively highlights both the maximum achievable permeance enhancement (0.01 wt%) and the behavior at high filler content (10 wt%), providing a comprehensive view of performance under realistic operating conditions.

While the present study focuses on membrane fabrication and performance evaluation under standard conditions, future work will extend this assessment to temperatures representative of industrial flue gas streams.

3.8. Gas Permeation and Selectivity Comparison

The membrane performance was evaluated using a Robeson plot based on CO₂ permeability and CO₂/N₂ selectivity, as permeability expressed in Barrer remains the most widely accepted parameter for benchmarking membrane materials and comparing intrinsic transport properties across different studies. Although membrane performance is frequently reported in terms of CO₂ perm‐eance (GPU), particularly for thin‐film and composite membranes, the use of permeability allows a more direct comparison with the Robeson upper bound and with a broader range of literature data. For completeness, literature data originally reported in permeance units were converted to permeability whenever membrane thickness information was available, while additional performance data are provided in Table S9.

The separation performance of the prepared MMMs was evaluated against the Robeson upper bound [38] for CO₂/N₂ selectivity and CO₂ permeability. As can be seen in Figure 9, both solvent‐cast and spray‐coating membranes approach the Robeson upper bound, with the membranes containing 1% and 10% of filler loading exhibiting the closest performance among the solvent‐cast and spray‐coating fabrication methods, respectively. In both cases, this behavior is due to the significant increase in CO₂/N₂ selectivity achieved at those filler loadings. The CO₂/N₂ separation performance of the prepared MMMs can also be benchmarked against other membranes reported in the literature [16, 17, 24, 39], whose values are plotted in Figure 9 and listed in Table S9 of the Supporting Information. Overall, the MMMs developed in this work outperform most literature results, especially when comparing the CO₂ permeability of spray‐coating membranes. Although some Pebax‐based MMMs containing ILs achieve higher permeability–selectivity combinations and, in certain cases, surpass the Robeson upper bound, the results obtained here demonstrate that spray‐coating is an effective strategy for enhancing the performance of IL‐based MMMs through the fabrication of thinner and more homogeneous selective layers. These findings further highlight the potential of combining tailored filler incorporation with scalable thin‐film processing techniques to develop high‐performance membranes for CO₂ separation.

FIGURE 9.

Performance of MMMs and spray‐coating membranes relative the Robeson upper bound [38], as well as literature values for Pebax MMMs [17], spray‐coating composite membranes [24], Bifunctional PIL membranes [16] and GRTIL membranes [39].

4. Conclusions

Among the two fabrication methods investigated, spray‐coating clearly stands out as the more advanced approach for the development of IL‐based MMMs incorporating functionalized SiIL microparticles. The results demonstrate that the membrane fabrication method has a decisive influence on gas transport properties and separation efficiency. Although the CO₂ permeance remains below that of some commercial membrane systems, the spray‐coating approach enables the formation of thinner, more homogeneous selective layers while preserving the intrinsic advantages of IL membranes. By enabling the formation of ultrathin, highly selective layers, spray‐coating achieves up to an order of magnitude higher CO₂ permeance than conventional solvent‐cast membranes, while maintaining high CO₂/N₂ selectivity. Notably, even minimal filler incorporation leads to significant performance enhancements. In particular, the 20PIL–80IL–0.01SiIL spray‐coated membrane exhibited an 81% increase in CO₂ permeance and a 20% improvement in CO₂/N₂ selectivity compared to the neat spray‐coated membrane, highlighting the strong synergy between precise processing control and targeted material design.

Mixed‐gas permeation experiments further confirmed the robustness of the spray‐coated membranes under realistic operating conditions. CO₂ permeance losses were minimal, amounting to only 9% and 5% for membranes containing 0.01 and 10 wt% SiIL, respectively. As commonly observed under mixed‐gas conditions, a reduction in CO₂/N₂ selectivity was also detected due to competitive sorption effects; however, the membranes maintained effective separation performance.

Overall, these findings establish spray‐coating as a fast, energy‐efficient, and highly tunable fabrication route for IL‐based membranes. Given its widespread use in the fabrication of thin functional films for photovoltaics, batteries, fuel cells, and sensors, spray‐coating also offers a scalable and industrially relevant pathway for the future development of high‐performance CO₂ separation membranes.

Funding

This work was supported by Fundação para a Ciência e a Tecnologia (2021.05450.BD, UID/00100/2025, UID/PRR/100/2025, LA/P/0056/2020, UID/50011, and LA/P/0006/2020).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material

Data Availability Statement

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