Chitosan-based mixed matrix composite membranes for CO 2 /CH 4 mixed gas separation. Experimental characterization and performance validation

Membrane technology is acknowledged as one of the most efficient for biogas upgrading, separating simultaneously CH 4 and CO 2 in the retentate and permeate streams, respectively. The sustainability would still be improved by using renewable and environmentally friendly materials for membrane fabrication. Specifically, this study is focused on laboratory-scale mixed gas separation tests of composite membranes prepared from chitosan (CS), a hydrophilic, biodegradable and biocompatible polymer from abundant natural resources, with good adhesive and film forming properties and high affinity for CO 2 due to the primary amine and hydroxyl groups functionalities in CS. To improve mechanical resistance and CO 2 /CH 4 separation, CS was hybridized by a 5wt.% loading of non-toxic ionic [emim][acetate] liquid (IL), and variable loadings of


Introduction
The devastating effects of climate change have led governments to determine different strategies, leading to the introduction of environmental regulations to promote the use of renewable forms of energy to decarbonize the economy [1]. To address this challenge, biogas could be used for energy production in order to obtain biomethane as power source for which it is crucial the separation of CH 4 from CO 2 . Although membrane technology is acknowledged as one of the efficient tools for biogas upgrading due to advantages such as small footprint, low-cost, high-energy efficiency, easy processability and handling, the sustainability of the fabrication can still be improved by using renewable and environmentally friendly materials. The outcome of circular economy and decarbonization has thus attired the focus on the potential of biopolymers and materials from renewable sources as a means of sustainable fabrication of membranes with functionalities at least as good as those of conventional fossil-based polymers. In this regard, several reviews have appeared in the last couple of years [2][3][4][5].
In this regard, new sustainable routes for gas separation membranes are being open by the application of less toxic solvents [6][7][8] and the substitution of chemically synthesized polymers by biopolymers [2, [9][10][11]. Biopolymers are defined as polymers coming from renewable sources or biodegradable polymers after their use, thus their interest because they do not take place in landfills and may represent a worthy alternative to traditional fossil fuel-based polymers. In this aspect, biopolymers reported for membrane preparation are usually chitosan (CS) [3], cellulose acetate (CA) [4], polylactic acid (PLA) [12], polyurethane (PU) [13], poly vinyl alcohol (PVA) [14], starch [15] and in lesser way, polyhydroxybutyrate (PHBs) [8]. They all come from different sources, and they present different degrees of hydrophilicity, mechanical robustness and biodegradability, so they are very difficult to classify. Recent review papers collect the properties of most common bio-based materials [10].
CA membranes have been known for quite a while in biogas upgrading [4], among other polysaccharides. Among them, CS is of particular interest, being a non-toxic, hydrophilic, multi-functional, biodegradable and biocompatible polymer from abundant natural resources, with good adhesive and film forming properties [3,16]. Its high affinity for CO 2 is linked to the primary amine and hydroxyl functional groups, which is reinforced in humid conditions [17][18][19]. The presence of water in hydrophilic biopolymers, like PU [20], CS and starch, facilitates the transport of CO 2 through the membrane, in addition to the solution-diffusion mechanism common in polymer based materials [21]. The biopolymer capability of increasing CO 2 permeability in wet conditions, which seemed to be improved in water-swollen state, has been studied on the modification of composite membranes for CO 2 /CH 4 separation, like Pebax/PES [22]. Other polymers that are usually encountered under the 'biopolymer' classification in literature are those polymers obtained from wastes, like silk-fibroin [23]. The chemical grafting of cellulose acetate membranes originally produced for reverse osmosis and nanofiltration have shown potential on biogas upgrading [24].
CO 2 -rich residual streams contain water vapor that needs to be removed before conventional capture processes. However, CS biopolymer membranes improve their CO 2 permeance and CO 2 /CH 4 separation abilities in wet conditions [17,25,26]. The blending of other waste-based biopolymer with CS improves the CO 2 separation in humid conditions as most of CO 2 -containing effluent streams are [18]. Common properties in biopolymers are a high hydrophilicity and low to moderate mechanical strength, both parameters influencing the transport of gases, such as CO 2 , and the durability of the membrane for an acceptable lifetime [21]. Both the CO 2 affinity and mechanical properties of chitosan, as other polymers, can be improved by adding small loadings of 1-ethyl-4-methyl imidazolium acetate ionic liquid (IL), and zeolite-based three dimensional and two-dimensional particles [27][28][29][30][31] in the form of a mixed matrix membrane (MMM).
Spectroscopic techniques are relatively common in order to determine the structural-functional relationship in MMM. Jomeikian et al. used 13 C NMR spectroscopy to observe the C-C interactions between the grafting C 3 N 4 employed in order to prevent the agglomeration of ZIF-8 particles within the CS matrix and improve the CO 2 /CH 4 separation performance [32]. These techniques may as well provide insight into the fabrication of gas separation membranes from biopolymers, with ion exchange capacity and hydrophilicity of similar values of ion exchange membranes.
The relationship between membrane materials properties and the separation performance necessary to optimize the membrane operation is usually limited to the use of the Robeson upper bound and single permeability and selectivity values [33,34]. The correlation of physicochemical properties with the functional transport characterization [2,5,6] can provide insightful information for the reliability of biopolymer based membrane performance in fields such as CO 2 /CH 4 separation. In order to overcome the trade-off between transport and mechanical resistance, a common challenge in hydrophilic biopolymers, it is necessary to establish the quantitative properties-function relationship enabling prediction of new sustainable membrane materials in gas separation membranes [35].
In a previous work, we presented a cross-flow mathematical model predicting the CO 2 /CH 4 separation performance of composite ionic liquid blended chitosan membranes over porous polyether sulfone (PES) support and a commercial silicon rubber membrane based on pure gas permeation experiments [36]. This work continues that research line, by including the analysis of mixed gas separation performance of the CS-based mixed matrix composite membranes coated on compatible PES porous support, varying the feed concentration and composition of the selective layer. To form this layer, CS was hybridized with non-toxic [emim][acetate] ionic liquid (IL) and different inorganic fillers prepared with non-critical inorganic reactants and no costly organic surfactants, whose compatibility with the CS biopolymer matrix and CO 2 preferential solubility towards CH 4 [28,37] was proven in previous studies [29], before being coated on a porous PES support [38,39]. The CO 2 permeance and CO 2 /CH 4 mixture selectivity were measured at different feed compositions in order to address different processes. The structure-function relationship was evaluated by analyzing the transport and separation performance in the light of the water uptake, swelling degree, attenuated total multiple reflection Fourier transformed IR (ATR-FTIR), Electrochemical Impedance Spectroscopy (EIS) and scanning electron microscopy (SEM) observations of the selective membrane layer. The results were validated with the process simulation model previously developed in our group [36,40].

Materials
The main polymer used for the thin selective layer fabrication was chitosan biopolymer (Sigma Aldrich, Spain). Trimesoyl chloride (TMC) solution in hexane (Sigma Aldrich, Spain) was used to treat the surface of the polyether sulfone (PES) porous support (0.1 µm pore size, PALL, Micromeritics, Spain) prior to the coating of the hydrophilic biopolymer-based layer. The commercial fillers used in this work were Zeolite 4A (Fluka, Sigma Aldrich, Spain), and 1-Ethyl-3-methylimidazolium acetate [emim][Ac] ionic liquid (Sigma Aldrich, Spain).

Membrane preparation
Flat-sheet composite MMMs were prepared in our laboratory by a modification of the interfacial polymerization process. A commercial PES porous disk was used as support because of its favorable characteristics such as wide operating temperature, broad pH limit, low cost, good chemical and mechanical resistance and easy to be fabricated into different geometries, which make it an excellent option as a support [41]. Firstly, the surface of the PES commercial membrane was treated with a hydrophobic solution of 0.1 wt.% TMC/hexane at 50ºC in order to avoid the penetration and crystallization of the hydrophilic semi-crystalline chitosan in the opening of the PES pores [39]. The top selective layer consists of a solution of 1 w/v% chitosan (CS) biopolymer in 2 wt.% acetic acid/H 2 O, to which IL was added up to a 5 wt.% of the total polymer weight, as optimized previously in our research group [39]. In order to prepare the mixed matrix coated membranes, different types of Na + -containing porous inorganic fillers were used to improve the performance of the biopolymer-based coating: 3D Zeolite 4A [37] and ETS-10 titanosilicate [28] and 2D AM-4 titanosilicate [36]. Lamellar AM-4 and 3D ETS-10 titanosilicate nanoparticles were prepared by hydrothermal synthesis without organic surfactant-directing structural agents as reported elsewhere [43,44]. The fillers were dispersed in 2 mL water before being added to the IL-CS solution, as a 5 and 10 wt.% loading of the total polymer matrix. Membranes were finally neutralized by immersion in NaOH 1M for 1h and thorough rinsing with ultrapure water, to eliminate the excess of NaOH, in order to charge the membrane with OHand increase its affinity towards CO 2 before gas separation and characterization experiments.

Characterization
Membrane thickness was measured using a digital micrometer (Mitutoyo digimatic micrometer IP 65) with an accuracy of 0.001 mm. Four points of the effective area of the membrane were measured and the average thickness and standard deviation were calculated. Swelling Degree (SD) was determined using equation (1) by measuring membrane thickness before and after gas permeation experiments.
(1) (%) = -× 100 The wet weight was obtained by quickly blotting the membrane on tissue paper to remove the excess water and the dry weight was measured after the gas permeation experiments. This allows to obtain the Water Uptake (WU) as (2) WU(%) = -× 100 ATR-FTIR spectroscopy was conducted using a Perkin Elmer spectrometer to study the chemical composition by analyzing the functional groups and interactions in the membranes surface of the different mixed matrix composite membranes after the permeation experiments. The spectrum used in the study was taken over 4 scans at a wave number resolution of 4 cm −1 in the range 400-4000 cm −1 .
EIS in a 2-electrode electrochemical cell was measured for all membranes in order to correlate the resistance of the selective membrane layer with the diffusional physicalchemical properties. EIS experiments were performed using a Bio-logic system equipment equipped with an impedance module in a potentiostatic method. The alkaline activated membrane was placed between smooth stainless steel-plated electrodes with a projected geometric area of 1.13 cm 2 . The cell was subjected to a constant pressure until the Nyquist plot was repeatedly the same. The amplitude was set at 10 mV (7.0 mV rms) and the frequency range was varied between 1.0 MHz and 0.1 Hz. Measurements were performed until reproducible experiments.
Cross-sectional area and morphology of selected membranes of each composition of the selective layer were observed by scanning electron microscopy (SEM) in a JSM-IT2500 HR (JEOL) microscope. Samples were prepared by freezing in liquid nitrogen before being fractured with a sharp cut from a freshly sharpened blade and sputtering with silver prior to observation. This enabled to observe membrane thickness and compare it to measured one to determine the quality of the coating. This technique also allowed us to check the distribution as well as the homogeneity of the coating and through the surface images determine the interaction between the polymer matrix and the fillers.

Gas separation experiments
A stainless-steel module with an effective membrane area of 15.6 cm 2 was used for the characterization of the permeability of the membranes prepared in this work. It consists of two stainless steel pieces with a cavity where the membrane is placed using a 316LSS microporous disk support with a pore size of 20 μm and sealed by Viton rings. Different feed mixture concentrations of CH 4 and CO 2 were used in order to focus on different processes, such as natural gas sweetening, biogas upgrading or enhanced oil recovery, from low to high concentration of CO 2 in the feed. Feed mixtures of CO 2 :CH 4 were introduced to the system in concentration relationships of 20:80, 35:65, 50:50, and 70:30 v/v%, respectively.
The experiments were carried out with the home-made separation plant represented in Figure 1. The feed pressure in all the experiments was set at 4 bar. The total feed flow rate was set in all cases to 50 mL/min and gas mass flow controllers (KOFLOC 8500, Sequopro S.L., Madrid, Spain) were used to establish the desired quantity of each gas component. The permeate flowrate was measured with a bubble flowmeter, before deriving the stream to a BIOGAS 5000 gas analyzer (Geotech Tamarac, FL, USA) in order to quantify the composition of the permeate flowrate. Permeation experiments were performed until the system reached the steady state. After two series of experiments at a different feed concentration, the feed stream was humidified by passing 50% through a water bubbler for 1h during conditioning at the following concentration point, before sampling data measurements to ensure that the membrane was maintained at constant WU values over the whole experimental run to avoid the usual drop of flux observed when hydrophilic biopolymer-based membranes dried. The permeance of the gas (1 GPU = 10 −6 cm 3 (STP) cm −2 s −1 cmHg −1 ) is defined as the pressure-normalized flux of a gas through a membrane, where P is the intrinsic permeability of the desired component on the selective membrane layer in Barrer (1 Barrer= 10 -10 cm 3 (STP) cm cm -2 s -1 cmHg -1 ); p p and p r are the permeate and retentate pressures, in cmHg, respectively; A is the effective membrane area for permeation (cm 2 ); t is the selective layer thickness (cm); Q p is the permeate flowrate (cm 3 (STP) s -1 ) at measurement pressure and temperature conditions, and x and y are the mole fractions of component i in the feed and permeate streams, respectively.
The intrinsic selectivity is calculated as the ratio between the permeance of the fast and slow gas components in the feed flowrate, in this case, CO 2 (i) and CH 4 (j) respectively, according to (4) = The mixture separation factor of gas i over gas j, α ij , is defined as the ratio of the relative concentration of the gas components in the permeate and the retentate as [45] (5) = ( ) ( )

Description and validation of the membrane unit model
The experimental results obtained were used as input data for the simulation analysis. A membrane unit programmed in Aspen Custom Modeler was used to determine the desired membrane performance to achieve different targets related to product quality. For this purpose, a crossflow membrane model was developed where a tank-in-series model was applied where the membrane unit was divided into k uniform cells of equal size (with subscript k varying from 1 to n), with the permeate of each cell collected and mixed with the rest of the permeate streams, and the retentate of each cell being the feed of the next [46]. The scheme of the membrane module is illustrated in Figure 2. The following hypotheses were considered for the development of the mass transfer model,  The membrane compaction due to the applied pressure was negligible.  Transport of gases through the membrane was steady-state, isothermal, and onedimensional.  The concentration polarization in the active-side of the membrane module was negligible.  Permeability was dependent on feed conditions and could be estimated based on the correlation of conditions such as pressure, flow rate and composition [40,47].
Considering these hypotheses, the steady-state material balances were established for determining the changes in gas concentrations and flowrates at both sides of the membrane, as described in our previous work [36]. For the modeling of the gas transport through the membrane layer, the solution-diffusion mechanism was considered, where the driving force of the permeation is the partial pressure difference between permeate and retentate, as shown in equation (6) [48], since the water content within the membrane was kept constant during the whole experimental runs over the whole range of feed concentrations [21], thus where N c,k is the molar flowrate of each component permeating though the membrane cell k (kmol h -1 ), A k (m 2 ) is the membrane area of each k cell, P c is the permeance for each c component across the membrane in molar basis (kmol h -1 bar -1 m -2 , that was obtained by the experimental permeance in unit of m 3 (STP) h -1 bar -1 m -2 ), x c and y c are the molar fraction of each component in the mixture present in the retentate and permeate stream, and p r and p p are the pressure on the retentate and permeate sides of the membrane, respectively.
The stage-cut, θ, one of the main parameters used for comparing the membrane performance and predicting the feasibility of scaling-up of the process, is determined as the ratio between the permeate and feed flow rates, as (7) = The target variables that determine the effectiveness of the membrane in the CO 2 /CH 4 separation process are the purity and recovery of the desired components in each stream, given by equations (8) and (9), respectively, These variables allow determining the quality of the different product streams, so all the specifications must be directed to achieve the maximum purity and recovery of the most permeable component in the permeate and the slowest component in the retentate. Commonly reported targets are set up as a recovery and purity of CO 2 in the permeate side up to 90% and up to 95% for the CH 4 in the retentate side [49]. Due to the difficulty of maximizing both parameters of the process simultaneously, it is necessary to propose a multi-objective task to set both goals at the same time. Finally, another issue that had to be addressed before running the simulations was the number of cells, and its selection must consider the balance between the precision of the model and its calculation load. The cross-flow membrane model implemented in this work for the simulation of the performance of each membrane unit was based in the custom model developed in a prior work for CO 2 /CH 4 separation using membrane technology, where the best fitting results were obtained at n = 100 [40]. Considering the previous validation of the developed model with experimental data obtained in the laboratory for a commercial membrane of similar geometry [31], the model is validated further in the present work regarding the mixed gas separation performance of the CS-based composite membranes at varying CO 2 and CH 4 composition in the feed using the same value of n. The deviation between simulated and experimental values at different stage-cut values is estimated by the experimental absolute relative error (AARE) of the target variables to validate the model is estimated by (10) (%) = 100 × | -|

Membranes characterization
The membranes were first characterized regarding their water swelling degree and permeation performance considering the mixed gas permeation experiments at 50:50 (v/v%). Table 1 collects the values of the active layer thickness, water uptake, swelling degree and permeability and intrinsic selectivity data of the membranes. It is worth observing that the most hydrophilic membrane is the 10 wt.% AM-4 layered titanosilicate composite membrane offers large permeability at the expense of selectivity. Upon increasing filler loading to 10 wt.% the permeability increased for the lamellar AM-4 filler, at the expense of decreasing selectivity. This is attributed to the different morphologies, textural properties and CO 2 adsorption capacities of the particles dispersed in the ILCS matrix of the coated membrane layer, from the 2D AM-4 nanoporous sheets with 9.9 mg CO 2 /g uptake [43] to 50 mg CO 2 /g of the zeolite 4A [37] and 78 mg CO 2 /g of ETS-10 [50]. Liu et al. recently observed that the 3D GO sheets obtained by ionic assembling GO nanosheets would tune up the CO 2 uptake by increasing surface area [51]. However, 2D fillers providing barrier and selectivity effects are reported at low loadings and they do not benefit by increasing loading in the polymeric matrix, where the high aspect area/volume ratio can lead to cluster agglomerations and defects upon membrane separation performance [52,53]. The intrinsic permeability and selectivity values obtained from the 50:50 (v/v%) CO 2 :CH 4 mixed gas experiments are also shown in Table1. Here, we can appreciate that CO 2 permeability increased for the 3D ETS-10 and zeolite filled membranes, at a 5 wt.% loading, and decreased for lamellar AM-4 at a 5 wt.% loading. This can be attributed to the different hydrophilic, swelling and CO 2 uptake properties of the selective layer of the membranes as a function of each type of filler. The CO 2 /CH 4 separation properties deteriorates as mentioned earlier with increasing loading of lamellar AM-4 titanosilicate [43,54]. This was also observed for the ETS-10 and zeolite A-filled membranes, at a lower magnitude: the CO 2 permeability and CO 2 /CH 4 selectivity decreased slightly with increasing ETS-10 titanosilicate loading, despite the synergic hydrophilic and compatibility properties of the titanosilicate and the CS biopolymer reported earlier and enhanced by the IL [28], and less significantly varying with increasing zeolite A loading, which may account for the good compatibility and dispersion of this small-pore zeolite A in the polymer matrix [37].
The swelling degree, calculated by equation (1) to give evidence of the membrane mechanical integrity of the selective ILCS layer. In Table 1, it can be observed how the swelling degree of the 10 wt.% AM-4-filled membranes was higher than the 5 wt.% AM-4 filled ILCS composite membranes. The increasing loading of hydrophilic AM-4 layered titanosilicate particles caused detachment of the CS-based membrane layer from the substrate underneath leading to higher flux and lower separation factor along experimental runs [55]. This may be attributed to the different aspect ratio influencing adhesion and dispersion of fillers in the CS biopolymer matrix, observed in previous works upon self-standing flat sheet membrane characterization [28,54].
The permeability and selectivity values in Table 1 are plotted in Figure 3 against the Robeson upper bound to be compared with the state-of-the-art biopolymers reported so far. For this comparison, other biopolymer-based membranes data collected by Russo et al. [2], were included in Figure 3. We can observe that the 5 wt.%-filled ILCS/PES membranes studied in our laboratory were very close to the state-of-the-art of current research for CO 2 /CH 4 separation, even overcoming the Robeson upper bound in the case of Zeolite A/ILCS and ETS-10/ILCS composite membranes. The permeability-selectivity trade-off decreases when the inorganic filler loading was increased to 10%. From the experimental separation data in Table 2 plotted in Figure 3, we observe that the IL-CS/PES mixed matrix composite membranes prepared in this study provided CO 2 permeability and CO 2 /CH 4 separation performance in the range acceptable for CO 2 /CH 4 separation applications and other biopolymer based mixed matrix membranes recently reported. For instance, a ZIF-8-gC 3 N 4 /CS membrane with a CO 2 permeability of 153 Barrer and a CO 2 /CH 4 selectivity of 24 has been accepted before [32]. Likewise, selfstanding, un supported but 0.05 mm-thick, bio-cellulose acetate membranes were reported to have a CO 2 permeability of 293 Barrer and a CO 2 /CH 4 selectivity of 29.5 [56]. In this work, we can affirm that the direct addition of 5 wt.% compatible filler loading increased the CO 2 /CH 4 selectivity and CO 2 permeability of the unloaded ILCS/PES membranes. This is going to be analyzed upon observing the dispersion of the fillers in the coated membrane layer by EIS and SEM below. The SEM cross-sectional images shown in Figure 4 allow verifying the dispersion of the selected fillers on the selective coated top layer, more evident at the higher filler loading of 10 wt.% in ILCS. As expected from the hydrophilic and swelling properties in Table 1, when the lamellar AM-4 titanosilicate was used as inorganic filler, cluster aglomerations were formed in the membrane matrix, as well as the detachment observed at 10 wt.% loading, between the coated CS-based layer and the porous PES substrate, accounting for the deterioration of separation of the 10 wt% AM-4/ILCS composite membranes (Table 1 and Figure 3). On the other hand, the selective layer thickness observed by SEM presents lower values than those of the freshly made composite membranes, except for 5 wt.% AM-4/ILCS composite membrane. Apparently the ETS-10 and zeolite A filled membranes experienced some compaction after all the separation runs, which might be also explaining their increase in selectivity as with polymers with high fractional free volume [57]. CS-based biopolymers show free volume in water-swollen state, and this can be recovered once dried, just by wetting the membranes again, in batch or continuous mode [25,28].
ATR-FTIR is a characterization technique that allows gaining knowledge of the physicochemical characterization properties of the surface as well as the functional groups present on the top layer of a composite membrane. Figure 5 represents the ATR-FTIR spectra of the 5 wt.% filled membranes. The spectra were measured after gas separation experiments. No large differences were observed as a function of the type of filler embedded in the ILCS coated selective layer of the membrane (region 950 -700 cm -1 ). This verifies again that the interaction between the Na + -ion exchangeable selected filler and the ILCS matrix is generally good, as observed in the SEM images. The characteristic OH band corresponding to the presence of structural water in the polymer matrix, is present in the region of 3800-3000 cm -1 , which confirmed the water uptake (WU) in Table 1 are maintained during gas separation experiments. The bands between 1700-600 cm -1 are correlated to the presence of the IL in the structure [29]. The ATR-FTIR spectra of the 10 wt.% filled membranes were similar and so they are not included in Figure 5 for clarification. The EIS analyses provide insight for predicting the diffusion resistance of the coated ion exchange membrane layers overcoming the influence of the porous substrate [58]. The equivalent circuit used for the Z fit of Nyquist plots was CP 1 -|R 2 -CPE 2 |-CPE 3 as described in a previous work [59], where CPE 1 and CPE 3 correspond to the capacitance of the double layer at the interfaces PES support/electrode and the CS based membrane/electrode, respectively; CPE 2 corresponds the bulk membrane-PES capacitance and R 2 corresponds to the bulk membrane-PES resistance. We assume that at high frequencies the double layer by hydroxyl ions traveling to/from the interface is negligible or smaller [59]. The plots of the mixed matrix composite membranes show a feature similar to ionic conductive materials [60]. The fitting of the figures of merit is shown in Figure 6 was very accurate. The resistances of ILCS based composite membranes were obtained from the fitting of the above equivalent circuit for the determination of R 2 . The resistances given by these fittings were 2450, 72 and 410 Ω for the AM-4, ETS-10 and Zeolite A with 5 wt.% loading, respectively. It may be of notice that the resistance of the TMC-untreated PES support accounts for only 16 Ω, while the TMC-treated PES support 70 Ω, so it does significantly contribute to the intrinsic membrane resistance. Furthermore, the ILCS/PES membrane only accounted for a bulk membrane resistance of 550 Ω. The increasing resistance observed in AM-4, ETS-10 and Zeolite A-filled membrane at 10 wt.% loading is correlated with the extent of detachment of the coated layer from the porous PES support observed in Figure 4, which led to higher CO 2 and CH 4 permeance and thus, lower selectivity, due to some clustering of particles in the selective membrane layer matrix upon increasing filler loading, commented above. Using the wet thickness of the freshly activated membranes filled with 5 wt.% AM-4, ETS-10 and Zeolite A, the ionic conductivity has values of 9.34 · 10 -4 , 0.047 and 0.00409 mS/cm, respectively. Those values are higher than those previously measured for ILCS based membranes without the modified PES support [59,60]. It is remarkable the higher ionic conductivity of the ETS-10 and zeolite A/ILCS composite membranes than those of the laminar titanosilicate filled membranes. Zeolite A is a small pore zeolite with breathable water and CO 2 affinity of its own that is interacting closely with the CS matrix even at 10 wt.% loading. The zeolite A into poly(trimethyl-1-silyl-propyne) allowed thus increasing both CO 2 permeability and CO 2 /CH 4 selectivity in the presence of relative humidity in the feed stream [37]. Accordingly, the titanosilicate-filled membranes provided a lower resistance to the ETS-10 membrane than the AM-4 filled membranes, which can be linked to the differences in hydrophilicity observed in Table 1. In addition, these resistances increased with increasing filler loading of all three fillers (with values of 8260, 4860 and 3640 Ω, for the AM-4, ETS-10 and Zeolite A, respectively) in the selective membrane matrix. For comparison, a summary of the current state-of-the-art of biopolymer-based membranes, including fabrication and characterization, for CO 2 /CH 4 separation has been carried out and reviewed in Table 2. The use of characterization techniques like X-ray photoelectron spectroscopy (XPS) employed to characterize water and ion transport in ion exchange membranes have revealed the layer formations within the active membrane matrix and the compatibility with the support by correlating diffusion impedances and transport resistances with varying layer thicknesses [58,61,62]. For example, Prasad et al. used XPS to validate the qualitative and quantitative nature of the chemical structure of graphene/chitosan MMM during CO 2 separation performance [63]. Nguyen et al. correlated the CO 2 and CH 4 transport through cellulose diacetate membranes with crystallinity and membrane thickness by a semi-empiric equation [64], which is instrumental in the development of industrial asymmetric membranes. Besides, these design and fabrication challenges are tuned up by the MMM approach highly explored for conventional polymers, consisting on blending them with other biopolymers or hybridizing with small amount of organic or inorganic fillers [27,52,65] and coating on a compatible porous support [32,55]. The main issue in mixed matrix or composite membranes are a good adhesion or compatibility between the components and the support determining the producibility, selectivity and stability performance [17,39,41].
These techniques can be added to the common selection of analytical techniques reported usually in the synthesis and characterization of membranes for CO 2 /CH 4 separation. PLA Easy Fil™ -White membrane DSC, SEM, FTIR CO 2 /CH 4 separation at room temperature. Ideal selectivity of CO 2 /CH 4 = 285 attained at a CO 2 permeability of 70 barrer. [12] Polyurethane/PES XRD, DSC, TGA, SEM, solution viscosity, porosity (water uptake), mechanical properties (ASTM D638), pressure decay sorption Separation of CO 2 /CH 4 . Single gas permeation [20] Sodium alginate and sodium carboxymethylcellulose modification of Pebax/PES membranes FESEM, XRD, TGA-DTG-DSC, Single gas permeation of CO 2 and CH 4 at 4-7 bar.

Membrane unit separation performance validation
The CO 2 /CH 4 separation performance of the mixed matrix composite membranes with different filler type and loading in the selective layer was validated in the range from 20 vol.% to 70 vol.% in CO 2 and from 80vol.% to 30 vol.% in CH 4 , in order to evaluate the behavior of the membranes in different target situations for CO 2 and CH 4 simultaneous recovery. From the experimental results on CO 2 and CH 4 permeance, the target variables purity and recovery of CO 2 and CH 4 in permeate and retentate streams, respectively, were calculated using equations (8) and (9). Figure 7 collects the CO 2 purity and recovery for the 5 wt.% filled membranes in the permeate, whereas the CH 4 purity and recovery of these membranes is represented in Figure 8. Figure 9 and Figure 10 show the corresponding results for the 10 wt.% filled composite membranes. Please refer to the supporting information for the detailed tables of experimental performance results.
The CO 2 purity in the permeate lied between 80 and 99% for all the membranes in the whole concentration range under study, which stands for the CO 2 selective character of the ILCS-based coated layer [38], but the CO 2 recovery in the permeate stream was low. This is attributed to the low values of experimental stage-cut in the laboratory scale experiments (0.02 -0.05), which implies a significant increase mainly in two variables: (i) the CO 2 purity in the permeate stream, and (ii) the CH 4 recovery in the retentate stream. To achieve more significant increases in the CO 2 recovery in the permeate, and CH 4 purity in the retentate streams, higher stage-cuts would be required. The recovery values obtained for the zeolite A/ILCS-are lower than those of the ETS-10/ILCS composite membranes, and the CO 2 recovery of the AM-4/ILCS membrane lies in between. This observation agrees with the ionic and hydrophilic character of the selective layers discussed above. The trend is reverse upon increasing filler loading (Figure 7(c) and Figure 7(d)) where the appearance of non-idealities in the interface between the high-aspect ratio AM-4 titanosilicate filler and the ILCS matrix is observed and not when ETS-10 nanoparticles were used as filler. The slightly higher CO 2 purity obtained for the zeolite A/ILCS composite membrane confirms as well the role of the water and CO 2 adsorptive cavities of the small pore hydrated zeolite [37]. For the process scaling-up, higher values of stage-cut would be necessary in order to attend to the purity and recovery of both components simultaneously, although higher values of stage-cut also imply higher membrane area requirements (therefore, higher costs). The process optimization would focus on these issues, as these results gave a clear idea on the balance between the different objectives to be covered for the technoeconomic optimization of the membrane separation process.
Nevertheless, those CO 2 purity and recovery in the permeate led to a linear increase of the CH 4 purity and high values of CH 4 recovery (> 99.01%) in the retentate stream, as observed in Figure 8. The membrane material did not apparently influence the CH 4 purity and recovery in the retentate stream. This is probably due to the small amount of filler loading (5 wt.%) in the bulk of the ILCS/PES membrane configuration. This is confirmed by the CH 4 recovery in the retentate obtained for the 10 wt.% filled membranes, where the ETS-10 and zeolite A-based membranes behaved similarly while layered AM-4 filled membrane attained slightly lower values, in agreement with their intrinsic hydrophilic, ionic character and differential morphology, as discussed above. Other observations derived from Figure 7 and Figure 8 is that the agreement of the experimental and simulated purity and recovery values for the 5 wt.% loaded mixed matrix composite membranes is generally very close, with relative errors lower than 10% (AARE), especially when the ETS-10 or the zeolite A were used as fillers of the ILCS layer matrix. As expected, the CO 2 purity of the 10 wt. % AM-4/ILCS composite membrane deviates from the experimentally observed value at low CO 2 concentration in the feed. This can be attributed to the higher water uptake and swelling degrees (Table 1) imparted by the high aspect ratio lamellar AM-4 filler than the other 3-dimensional ETS-10 and zeolite A filler particles used in this work. The lamellar AM-4 particles thus provide a lower permeance and stage-cut at 5 wt.% loading, while the cluster aggregating and layer detachment observed at increasing filler loading deteriorated the separation properties of the AM-4/ILCS composite membranes at 10 wt.%. Usually, layered materials are filled in at low loadings in the mixed matrix membranes because of the highaspect ratio providing contact with the polymer matrix and the diminished particle size upon exfoliation of the nanosheets leading to agglomeration at higher loadings [53,69]. Figure 9 collects the deviation between simulated and experimental CO 2 and CH 4 purity and recovery in the permeate and retentate streams, calculated by equation (10). As expected, only the AM-4 filled mixed matrix composite membrane surpassed values of AARE larger than 10%, due to the lower flux and stage-cut conducing to higher sensitivity than the other membranes tested in this work. Please refer to the Supporting information for the full recollection of CO 2 and CH 4 purity and recovery values and their AARE in both streams.

Conclusions
This work validates the mixed gas separation performance of novel chitosan-based composite membranes for CO 2 /CH 4 separation in a whole range of feed concentrations, to evaluate the potential in different applications. Mixed matrix composite membranes were prepared by hybridizing chitosan biopolymer solution with a 5 wt.% [emim][acetate] (IL) and 5, 10 wt.% loading of Na + -exchangeable fillers in different morphologies and compositions, as layered AM-4 titanosilicate, 3D ETS-10 titanosilicate and small-pore zeolite 4A, influencing in their CO 2 permeance and CO 2 /CH 4 separation factor. The membranes were characterized regarding their water uptake and swelling, film thickness and physicochemical properties of the coated layer. The membranes were used to validate a previously developed cross-flow mathematical model focusing in two target objectives, the purity and recovery of permeate and retentate simultaneously. The CO 2 purity in the permeate was highly dependent on the composition, dispersion, and adhesion of the selective membrane layer components among them and with the porous PES support.
Further work is being performed regarding the optimization and techno-economic assessment of the best CS-based composite membranes to maximize both CO 2 and CH 4 recovery in the permeate and retentate streams, that is, the 5 wt.% ETS-10 and zeolite A filled ILCS-based mixed matrix composite membranes studied in this work. The effect of thickness and blending with other biopolymers with higher intrinsic permeability on membrane and process configuration are also being undertaken in a future work.

Acknowledgements
This research was funded by the Spanish Ministry for Science and Innovation, grant numbers PID2019-108136RB-C31 and PID2019-108136RB-C32 of the Spanish Research Agency (AEI/10.13039/501100011033). A.T.C. also acknowledges the Ministry for the Early-Stage researcher contract (FPI grant no. PRE2020-09765/AEI/10.13039/501100011033) and the Scientific Committee of the International Separation and Purification Technology Congress (ISPT 2022, Elsevier) celebrated online in December 2022 for the Best Poster Award. Alba Pérez Cañaveras and the General Analytical Services at the University of Alicante are also gratefully acknowledged.