Process optimization of camel dung-derived biogas upgrading through ionic liquid–incorporated polyethersulfone membranes

Abstract

This study is focused on optimization of the biogas upgradation process by using ionic liquid–embedded polyethersulfone (PES) membranes. Process optimization using response surface methodology (RSM) via central composite design (CCD) was conducted to model and optimize the synthesis and separation performance of these membranes. The membranes were synthesized with varying loadings of [emim][Tf₂N] ionic liquid by the solution casting technique. Feed pressure and ionic liquid loading were taken as the independent variables, whereas CO₂ permeability, CH₄ permeability, and CO₂/CH₄ selectivity were chosen as the process responses. At optimum IL concentration (20%) and feed pressure (5 bar), CO₂ permeability of PES/[emim][Tf₂N] membrane was found to be 119.45 barrer with CO₂/CH₄ selectivity of 31.07 and overall desirability of 0.52. From SEM analysis, a dense and defect-free structure of PES/[emim][Tf₂N] was observed, which confirmed successful incorporation of the ionic liquid within the polymer matrix. Similarly, uniform distribution of [emim][Tf₂N] ionic liquid in PES matrix was confirmed by EDX analysis. These results confirm that PES/[emim][Tf₂N] membranes have a huge potential to upgrade biogas and enrich biomethane for circular bioeconomy and renewable energy goals.

Biogas; Camel; CO₂ separation; IL; membrane; [emim][Tf₂N]

1. INTRODUCTION

The push toward renewable and decentralized energy systems has drawn renewed attention in recent years to the valorisation of livestock and agricultural-based wastes via anaerobic digestion. In arid and semi-arid regions, including Saudi Arabia, where camel husbandry remains culturally and economically prominent, the large volumes of camel dung generated represent a substantive potential feedstock for biogas production [¹]. Although much of the research on biogas from animal manure has focused on cattle or pig substrates, several studies and industry efforts point to the viability of using camel manure for methane-rich biogas under conditions appropriate for desert-rangeland systems [²]. Moreover, the use of camel dung is already being explored as an alternative fuel in the United Arab Emirates in cement-plant operations [³].

In the context of Saudi Arabia’s drive toward circular-bioeconomy and waste-to-energy solutions, tapping camel-manure anaerobic digestion aligns with national ambitions of reducing fossil-fuel dependence, valorising agricultural by-products, and achieving decentralized renewable-energy generation. Recent assessments of domestic waste-based biomethane potential in Saudi Arabia indicate large resource volumes and underline the strategic importance of such substrates in the national energy transition [⁴]. However, the specific application of camel-dung-derived biogas remains under-represented in the literature, particularly in conjunction with advanced biogas-upgrading technologies.

Raw biogas generated from anaerobic digestion typically contains methane (CH₄), carbon dioxide (CO₂), and some minor quantities of hydrogen sulfide (H₂S), water vapour, and trace gases [⁵]. To upgrade the raw biogas into biomethane that meets pipeline-injection or vehicle-fuel specifications (typically > 95% CH₄ or higher calorific value), removal of CO₂ (and other impurities) is required [⁶, ⁷]. Among available upgrading technologies (absorption, adsorption, cryogenic, membrane separation), membrane-based upgrading offers attractive features for decentralized systems: compact footprint, modularity, relatively low energy consumption, and simpler operation compared to conventional solvent-based or cryogenic systems [⁵]. Indeed, membranes dedicated to CO₂/CH₄ separation have been the focus of numerous studies.

In membrane gas-separation technologies, the integration of ionic liquids (ILs) into polymeric membranes (through blending, embedding, or supported ionic liquid membranes) has become a promising solution to increase their separation performance [⁸, ⁹]. Ionic liquids have unique characteristics such as high CO₂ affinity, low vapour pressure, tunable chemistry, and compatibility with polymeric matrices that make them a suitable candidate to improve CO₂ permeability, CO₂/CH₄ (or CO₂/N₂) selectivity, and minimize plasticization or ageing problems in gas separation membranes [¹⁰]. According to the literature, the IL-modified membranes may enhance permeability and selectivity by several important factors. Incorporation of ILs like 1-ethyl-3- methylimidazolium bis(trifluoromethylsulfonyl)imide [emim][Tf₂N]) into polymeric membranes not only improve compatibility and free volume but also add specific CO₂-affinity sites into the polymeric membrane. Imidazolium-based ILs demonstrate strong quadrupolar and Lewis acid-base interactions with CO₂ which highly enhance the CO₂ solubility in comparison with other gases [¹¹]. In one study of polyethersulfone (PES)/SAPO-34 mixed matrix membrane, [emim][Tf₂N] was added at about 20 wt.% and produced a clear enhancement in CO₂ permeability and CO₂/N₂ selectivity compared to the unmodified membrane [¹²]. The mechanism appears to involve improved filler (zeolite) dispersion, better polymer-filler/IL compatibility (reducing interfacial defects), and, in some cases, the so-called anti-plasticization effect at low IL loadings, leading to improved kinetic sieving [¹⁰]. Although most of these studies focus on CO₂/N₂ separation or pure-gas systems, the underlying approach is directly relevant for CO₂/CH₄ separation in biogas upgrading.

Response surface methodology (RSM) is a powerful statistical analysis package that has been extensively employed in modeling and optimization of various processes to improve system performance and efficiency in processes [¹³]. It presents the liberty to compare and contrast the impact of several inputs or process variables on outputs or process responses of a system at the same time [¹⁴]. The Central Composite Design (CCD) is one of the popular designs of RSM. CCD can be used to build a design of experiment (DoE) that incorporates factorial, axial, and centre points in order to establish a detailed correlation between the independent variables and the response variables [¹⁵]. RSM-CCD has been successfully applied to optimize membrane-based methane enrichment (purity and recovery). By using this approach, optimum pressure and stage-cut conditions have been identified that maximize CH₄ yield with minimal losses [¹⁶]. Recent studies combine RSM with predictive modelling to optimize membrane fabrication parameters (polymer concentration, casting and conditioning temperatures), validating the RSM approach for tailoring CO₂ separation membranes [¹⁷]. In another study, JUSOH et al. [¹⁸] employed RSM-based CCD to optimize the separation performance of ZIF-8/6FDA-durene MMM. The developed models were found to be reliable for optimizing CO₂/CH₄ separation of these MMMs. Recent studies have demonstrated the importance of integrating material innovation with process optimization for sustainable resource recovery in decentralized systems. For example, ZHANG et al. [¹⁹] have shown that source separation coupled with process optimization can significantly enhance the recovery of value-added resources from waste streams in rural environments, highlighting the potential of localized waste-to-resource strategies. Similarly, LIU et al. [²⁰] reported the scalable fabrication of ultraselective and highly permeable aromatic polyamide membranes for optimized separation of antibiotic from water streams.

The present work focuses on the development and optimisation of a PES membrane embedded with [emim][Tf₂N] ionic liquid (PES/[emim][Tf₂N]) for upgrading biogas produced from camel dung under Saudi Arabian conditions. The study has employed response surface methodology (RSM), implemented via Design- Expert^® software, to optimise ionic liquid concentration and feed pressure with respect to separation performance (CO₂ permeation, CH₄ permeation, and CH₄ purity). No study is reported in the literature on the use of ionic liquid embedded polymeric membranes for process optimization of the biogas upgradation process. By combining the context of camel-dung-derived biogas with the advanced membrane strategy of IL embedding and statistical optimisation, this work aims to contribute both to the waste-to-energy value chain and the membrane-separation domain.

2. MATERIALS AND METHODS

2.1. Materials

The ionic liquid 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([emim][Tf₂N], purity ≥ 98%) was sourced from Sigma-Aldrich, Germany. Polyethersulfone (PES) flakes (ULTRASON^® E-6020 P) were obtained from BASF, Germany. The solvent used in this study was 1-methyl-2-pyrrolidone (NMP, EMPLURA^® grade), and it was purchased from Merck. High-purity CO₂ and CH₄ gases were supplied by a local vendor. All materials were used as received, without additional purification.

2.2. Membrane synthesis

Casting solutions were prepared by dissolving 17 wt.% PES (dried overnight at 110 °C) in NMP, followed by the addition of a predetermined amount of ionic liquid (0–40 wt.% relative to polymer weight). Each batch consisted of 50 g of solution, mixed in Duran^® bottles. The mixtures were stirred gently for 24 hours until a uniform, transparent solution was achieved and then allowed to rest for approximately 30 minutes to release entrapped air bubbles. The degassed solution was cast onto a clean glass plate, using a casting knife adjusted to 300 µm to obtain dense, flat-sheet films. Solvent evaporation was carried out in two stages: initially at 90 °C for 20 hours and subsequently at 130 °C for 4 hours to ensure complete drying. Dried membranes were peeled from the glass plates and stored for characterization.

2.3. Gas permeation studies

Gas permeation measurements for the synthesized membranes were conducted according to the design matrix shown in Tables 1 and 2. A detailed description of the permeation setup and its operating procedure is reported elsewhere [²¹]. The permeability of an individual gas species was determined using [²²]:

(1) $$P_{CO}{}_{{}_2}=\frac{N_{CO}{}_{{}_2}l}{\mathrm{\Delta }{P}_{CO}{}_{{}_2}}$$

where N is the permeation flux of the gas through the membrane, l is the membrane thickness, and Dp represents the transmembrane pressure difference. The same expression was used to calculate methane permeability. The membrane’s ideal selectivity, α, was obtained from [²²]:

(2) $${\alpha }_{CO}{}_{{}_2/C{H}_4}=\frac{P_{CO}{}_{{}_2}}{P_{C{H}_4}}$$

2.4. Central composite design (CCD)

The combined influence of feed pressure and ionic liquid loading on the performance of the biogas upgrading process was evaluated using response surface methodology (RSM). Design-Expert^® software (v.13.0) was used to generate a design matrix based on the central composite design (CCD) approach. A randomized set of experiments was produced by the software that adequately covered the ranges of the selected variables and allowed for statistical optimization of the process. Table 1 presents the selected factors along with their ranges and levels.

Table 1
Selected operating factors and their coded/actual limits used in the design matrix.

LEVEL	CODED LEVEL	IONIC LIQUID LOADING (%)	FEED PRESSURE (BAR)
Low	-1	0	5
Mid	0	20	15
High	+1	40	25

Feed pressure and IL content were chosen as the independent variables, while CO₂ and CH₄ permeability, and CO₂/CH₄ selectivity of ionic liquid embedded membranes were taken as response parameters. These responses were used to assess how the two factors act individually and jointly to influence biogas enrichment. In total, 13 experimental runs were carried out, as listed in Table 2, and the values of measured responses are also summarized in Table 2.

Table 2
Experimental design matrix and results.

RUN	FACTOR A: FEED PRESSURE (BAR)	FACTOR B: IONIC LIQUID LOADING (%)	RESPONSE 1: CO2 PERMEABILITY (BARRER)	RESPONSE 2: CH4 PERMEABILITY (BARRER)	RESPONSE 3: CO2/CH4 SELECTIVITY
1	5	40	255.30	6.07	42.03
2	25	0	2.42	0.15	15.91
3	15	20	110.68	3.31	33.39
4	5	0	2.86	0.20	13.80
5	25	20	101.65	2.67	38.02
6	5	20	117.19	3.91	29.94
7	15	0	2.59	0.17	14.72
8	15	20	109.3	3.24	33.73
9	25	40	227.65	4.60	49.46
10	15	20	112.8	3.45	32.69
11	15	20	110.32	3.31	33.32
12	15	20	109.74	3.2	34.29
13	15	40	236.84	5.17	45.75

The data collected were fitted to the proposed model equations using Design-Expert^®, which also estimated the corresponding regression coefficients. The adequacy and statistical significance of the models were examined through analysis of variance (ANOVA). Goodness-of-fit between predicted and experimental results was assessed using the coefficient of determination (R²). Three-dimensional response surface plots were constructed for each response in order to visualize the interactions between variables. Finally, desirability-based numerical optimization was performed, and confirmation experiments were conducted at the predicted optimal conditions to validate the accuracy and reliability of the developed RSM models.

2.5. Membrane characterization

The morphology of the synthesized ILPMs was examined using field emission scanning electron microscopy (FESEM, ZEISS SUPRA™ 55VP). Surface images were obtained from randomly selected regions. For cross-sectional imaging, membrane strips were immersed in liquid nitrogen for about 30 seconds to induce brittle fracture and then mounted on stainless steel stubs. Samples were sputter-coated with a thin layer of gold using an Emitech K550X coater before observation. To verify the presence and distribution of the ionic liquid within the polymer matrix, energy-dispersive X-ray (EDX) analysis was performed on membrane surfaces. Thermogravimetric analysis (TGA) was conducted using a Pyris-1 TGA system (PerkinElmer, Inc.). Samples were heated from 30 °C to 800 °C at 10 °C/min under a nitrogen atmosphere to evaluate thermal stability. Fourier Transform Infrared (FTIR) spectroscopy (PerkinElmer Spectrum One) was used to observe the interaction between the polymer and the ionic liquid. Spectra were collected in transmittance mode in the range of 400–4000 cm⁻¹.

3. RESULTS

3.1. Development of predictive models using CCD-based response surface analysis

The experimental data shown in Table 2 were used to generate quadratic predictive models as given by Equations 3–5. For this purpose, response surface methodology (RSM) was used by employing a central composite design. This statistical method allowed systematic analysis of the influences of the two crucial process variables, viz., feed pressure and ionic liquid loading, on the CO₂ and CH₄ permeability of the membrane together with the CO₂/CH₄ selectivity. Using CCD, it is possible to determine the factors that have the most significant impact on the biogas upgrading performance and which interactions are statistically significant to optimize the processes [¹³]. The resulting empirical models in terms of the actual operating variables are as follows:

(3) $${\text{CO}}_{\text{2}}\text{Permeability}=\text{4}.\text{18}–0.\text{2}0\times \text{P}+\text{5}.\text{36}\times \text{L}–0.0\text{3}\times \text{P}\times \text{L}+0.00\text{5}\times {\text{P}}^{\text{2}}+0.0\text{27}\times {\text{L}}^{\text{2}}$$

(4) $${\text{CH}}_{\text{4}}\text{Permeability}=0.\text{39}–0.0\text{2}\times \text{P}+0.\text{21}\times \text{L}–0.00\text{1}\times \text{P}\times \text{L}+0.000\text{3}\times {\text{P}}^{\text{2}}+0.00\text{1}\times {\text{L}}^{\text{2}}$$

(5) $${\text{CO}}_{\text{2}}{\text{/CH}}_{\text{4}}\text{Selectivity}=\text{12}.\text{85}+0.0\text{7}\times \text{P}+\text{1}.0\text{1}\times \text{L}+0.00\text{6}\times \text{P}\times \text{L}+0.00\text{2}\times {\text{P}}^{\text{2}}–0.00\text{8}\times {\text{L}}^{\text{2}}$$

where P represents feed pressure, and L denotes ionic liquid loading. By these equations, it is possible to estimate values of the expected responses at any combination of the two factors within the range studied.

All three responses were analyzed using analysis of variance (ANOVA), and the findings were summarized in Tables 3–5. The P-values for the regression terms were below 0.05 in each case, confirming that the models are statistically significant [²³]. Very high F-values, i.e., 5084.02 for CO₂ permeability, 462.93 for CH₄ permeability, and 415.52 for CO₂/CH₄ selectivity, further demonstrate the strength and reliability of the fitted models [²⁴]. Such high ratios indicate that the probability of obtaining these values by chance is extremely low (less than 0.01%). The goodness of fit was examined using the coefficient of determination (R²) and adjusted R² values, all of which exceeded 0.98 (Tables 3–5). This reflects an excellent agreement between the experimental measurements and the predicted responses [²⁴]. In addition, the lack of fit was found to be insignificant for every model, confirming that the selected functional form adequately describes the experimental trends [²⁵].

Table 3
Variance analysis summary for the CO₂ permeability response model.

SOURCE	SUM OF SQUARES	df	MEAN SQUARE	F-VALUE	P-VALUE
Model	85365.65	5	17073.13	5084.02	< 0.0001	significant
A-Pressure	317.44	1	317.44	94.53	< 0.0001
B-IL Loading	84468.78	1	84468.78	25153.01	< 0.0001
AB	185.02	1	185.02	55.09	0.0001
A2	0.79	1	0.79	0.23	0.6414
B2	324.14	1	324.14	96.52	< 0.0001
Residual	23.51	7	3.36
Lack of Fit	16.16	3	5.39	2.93	0.16	not significant
R2	0.9997
Adjusted R2	0.9995
Predicted R2	0.9774

Table 4
Variance analysis summary for the CH₄permeability response model.

SOURCE	SUM OF SQUARES	df	MEAN SQUARE	F-VALUE	P-VALUE
Model	41.90	5	8.38	462.93	< 0.0001	significant
A-Pressure	1.27	1	1.27	70.34	< 0.0001
B-IL Loading	39.08	1	39.07	2158.68	< 0.0001
AB	0.49	1	0.49	27.54	0.001
A2	0.003	1	0.003	0.1855	0.679
B2	0.93	1	0.938	51.85	0.0002
Residual	0.12	7	0.018
Lack of Fit	0.09	3	0.030	3.31	0.1388	not significant
R2	0.9970
Adjusted R2	0.9948
Predicted R2	0.9773

Table 5
Variance analysis summary for the CO₂/CH₄ selectivity response model.

SOURCE	SUM OF SQUARES	df	MEAN SQUARE	F-VALUE	P-VALUE
Model	1530.87	5	306.17	415.52	< 0.0001	significant
A-Pressure	51.73	1	51.73	70.20	< 0.0001
B-IL Loading	1435.63	1	1435.63	1948.36	< 0.0001
AB	7.11	1	7.11	9.65	0.01
A2	0.22	1	0.22	0.30	0.60
B2	32.96	1	32.96	44.73	0.0003
Residual	5.16	7	0.73
Lack of Fit	3.79	3	1.26	3.68	0.12	not significant
R2	0.9966
Adjusted R2	0.9942
Predicted R2	0.9739

Figure 1 shows plots of the comparison of the actual and predicted values of CO₂ permeability, CH₄ permeability, and the CO₂/CH₄ selectivity. The close clustering of the data points around the diagonal line shows the robustness of the developed CCD-RSM models. It further shows the suitability of these modes in determining the combined effects of feed pressure and ionic liquid loading on the biogas upgradation process [²⁶]. Figure 2 shows the residual plots of all the experiment runs. In both scenarios, the residuals are randomly distributed along the zero line, which means that there is neither a systematic trend nor an abnormal pattern. This randomness confirms that the experimental runs were properly randomized and the fitted models adequately captured the underlying behaviour of the biogas upgradation process. Thus, the distribution of the residuals confirms the validity and reliability of the regression models obtained by the RSM approach.

Figure 1
Predicted vs. actual plots of process responses.

Figure 2
Residual plots of process responses.

Although the developed quadratic models have shown high F-values and R² values, this behavior is physically understandable for the present system. The experimental design space was very narrow as it focused on two dominant variables i.e. feed pressure and ionic liquid loading. Both factors have a strong influence on gas transport behavior in dense ionic-liquid glassy polymeric membranes. In these systems, the permeability and selectivity are governed by solution-diffusion mechanism which lead to limited experimental deviation and high correlation coefficients. Moreover, the close agreement between adjusted R² and predicted R² values confirms that the models possess strong predictive capability, and no overfitting of the experimental data was observed. In addition to residual randomness, the strong agreement between predicted and experimental values observed in the confirmatory experiments further supports the adequacy and robustness of the developed response surface models.

3.2. Effect of process parameters on membrane performance

Figure 3(a)–(c) shows the effect of feed pressure and ionic liquid loading on the CO₂ permeability, CH₄ permeability, and CO₂/CH₄ selectivity. The results of the 3D response surface plots obtained with the help of quadratic models (Equations 3 and 4) indicate a steady decrease in the CO₂ and CH₄ permeability with the feed pressure increase. At the lowest pressure studied (5 bar), the maximum permeabilities were observed for both gases, and then a slow decrease was observed. This is typical of glassy polymers, in which high pressures generally lead to densification of the polymer chains and a decrease in fractional free volume, which decreases gas transportation rates [¹⁴]. However, the opposite trend is observed in ionic liquid loading. The increased loading of [emim][Tf₂N] contributes to an increase in the permeability of both gases. This increase is explained by the fact that the ionic liquid is able to disrupt the packing of the polymer chains and provides the formation of extra free volume pathways which facilitate the diffusion of the gases. However, the dominant mechanism in IL-containing glassy polymers is solubility-controlled transport rather than diffusivity enhancement alone. The incorporation of [emim][Tf₂N] introduces CO₂-philic sorption domains due to strong quadrupolar and Lewis acid–base interactions between CO₂ molecules and the ionic liquid components [²⁷]. These interactions preferentially increase CO₂ solubility relative to CH₄ and enable simultaneous improvement in permeability and selectivity, as observed in Figure 3(a)–(c). These trends are consistent with reports for imidazolium-based IL membranes [¹¹].

Figure 3
3D plots of the effect of process parameters on process responses (a) CO2 permeability, (b) CH4 permeability and (c) CO2/CH4 selectivity.

The pressure dependence of transport reflects a balance between mechanical compression and sorption effects. Increasing pressure generally suppresses permeability through densification of the PES matrix and reduced segmental mobility. However, in PES/[emim][Tf₂N] membranes, strong CO₂–IL interactions partially counteract this effect by maintaining accessible sorption sites, resulting in a weaker pressure sensitivity for CO₂ permeability compared with CH₄. In cases of CH₄, much lower affinity for the ionic liquid phase is present, and the transport remains predominantly diffusivity controlled. Therefore, CH₄ permeability shows a more significant decline with pressure. Similar behavior has been reported for IL-modified membranes where CO₂–IL interactions suppress classical plasticization and stabilize separation performance under pressure [²⁸]. The distinct pressure responses of CO₂ and CH₄ further confirm that ionic liquid incorporation enhances solubility selectivity rather than merely increasing free volume within the polymer matrix.

In the case of CO₂/CH₄ selectivity, both ionic liquid loading and feed pressure have a positive effect, as can be seen in Figure 3(c). The selectivity is greater at higher pressure because of higher CO₂ sorption in the PES matrix and plasticization resistance of the synthesized PES/[emim][Tf₂N] membrane [⁷]. Similarly, increased IL loading enhances selectivity through an increased interaction between CO₂ gas and the ionic liquid. These combined effects demonstrate that membrane performance is governed by a coupled transport mechanism in which diffusivity reduction from compression competes with solubility enhancement from ionic liquid domains. These results highlight the synergistic roles of pressure and ionic liquid content in tuning membrane performance for biogas upgrading applications.

3.3. Optimization analysis

RSM-based models were used to optimize the process of biogas upgrading using the synthesized PES-ionic liquid membranes. Ionic liquid loading was positively correlated with all response measures of the process, as shown by the parametric study, whereas feed pressure had a negative effect on CO₂ and CH₄ permeability and a positive effect on CO₂/CH₄ selectivity. These opposite impacts indicate the necessity to establish a balance between the ionic liquid loading in membranes and the operating feed pressure so that biomethane enrichment may be efficient and cost-effective [²⁹].

The optimization criteria had a number of practical considerations. It is more economical to reduce the concentration of ionic liquids since they are rather costly additives. Similarly, biogas plants typically operate at low pressures to reduce compression costs and maintain safer operating conditions; therefore, the feed pressure was preferred to be kept as low as possible. Conversely, the process responses viz., CO₂ permeability and CO₂/CH₄ selectivity were set to be maximized, as higher permeability improves throughput while higher selectivity enhances methane purity in the upgraded biogas [¹⁴].

Following these constraints, Design Expert^® generated a single optimal solution, as presented in Table 6. Feed pressure had the greatest desirability (0.99) as shown in Figure 4, implying that it has a stronger impact on the overall optimization criteria than CH₄ permeability which had the lowest desirability (0.379), which implied that it is quite challenging to reduce the transport of CH₄ and maximize the CO₂/CH₄ selectivity simultaneously. In general, the system had a desirability of 0.52 and was optimized under the conditions that were applied.

Table 6
Optimized conditions obtained by process optimization using RSM.

NO.	PRESSURE	IL LOADING	CO2 PERMEABILITY (BARRER)	CH4 PERMEABILITY (BARRER)	CO2/CH4 SELECTIVITY	DESIRABILITY
1	5	20	119.45	3.82	31.07	0.52

Figure 4
Overall desirability distribution illustrating factor contributions toward the optimized biogas upgrading outcome.

Figure 5 shows the ramp plot that pictures the suggested optimal setting of both the process factors and the responses. The spots on every ramp are the estimated local optimum of feed pressure, ionic liquid loading, and every parameter of performance of the membranes, showing how the desired operating space is set within the economic and performance limits.

Figure 5
Optimization ramp outlining the chosen solution and predicted performance values.

The confirmatory experiments were carried out using the optimal conditions described in Table 6. The experimental results obtained from these validation runs are presented in Table 7. To ensure reliability, three independent membrane samples containing the optimized ionic liquid loading were synthesized and tested at the optimized feed pressure. The corresponding statistical evaluation of these confirmatory experiments is provided in Table 8. A strong agreement was observed between the predicted values from the optimization analysis (Table 8) and the experimentally measured values (Table 8), demonstrating the accuracy and robustness of the developed RSM models. Based on these findings, it can be concluded that the PES/[emim][Tf₂N] membrane containing 20% ionic liquid and operated at a feed pressure of 5 bar delivers the most favorable performance for camel-dung biogas upgrading.

Table 7
Measured response values obtained under the recommended optimum settings.

NO.	CO2 PERMEABILITY (BARRER)	CH4 PERMEABILITY (BARRER)	CO2/CH4 SELECTIVITY
1	117.32	3.74	31.37
2	122.46	4.03	30.38
3	120.89	3.65	33.12

Table 8
Comparison of model-predicted and experimentally observed values at optimum conditions.

ANALYSIS	ACTUAL MEAN	ACTUAL MEDIAN	STD DEV	N	SE PRED	95% PI LOW	DATA MEAN	95% PI HIGH
CO2 permeability (barrer)	119.45	119.45	1.83	3	1.67	115.51	120.23	123.40
CH4 permeability (barrer)	3.82	3.82	0.13	3	0.12	3.53	3.80	4.11
CO2/CH4 selectivity	31.07	31.07	0.86	3	0.78	29.22	31.62	32.91

It should be noted that the gas permeation data reported in this study are based on single-gas measurements of CO₂ and CH₄, and therefore represent ideal permeability and ideal selectivity values. Under real biogas conditions, mixed-gas effects such as competitive sorption and diffusion coupling are also observed that lead to deviations from ideal behavior particularly in glassy polymer membranes modified with CO₂-philic additives such as ionic liquids. In mixed-gas systems, the preferential sorption of CO₂ can suppress CH₄ transport while simultaneously reducing the apparent CO₂/CH₄ selectivity relative to single-gas values, as commonly described by dual-mode sorption theory. Nevertheless, previous studies have shown that at low operating pressures, single-gas measurements provide a reliable basis for identifying relative performance trends and optimal membrane compositions, because plasticization and competitive diffusion effects are less pronounced. Accordingly, the optimized membrane composition identified in this work is expected to remain effective for practical biogas upgrading applications.

3.4. Membrane characterization

The morphology of the PES/[emim][Tf₂N] membrane at the optimized ionic liquid loading of 20% is presented in Figure 6. The morphology of the neat PES membrane is also shown for comparison. Both membranes exhibit a dense, non-porous, and defect-free structure, characteristic of glassy PES-based materials. Importantly, no visible phase separation or pore formation is detected in the IL-modified membrane, indicating that the ionic liquid is uniformly dispersed within the polymer matrix. This homogeneous distribution suggests strong compatibility between PES and [emim][Tf₂N], and the ability of the PES matrix to act as the host while maintaining its structural integrity [³⁰]. Furthermore, the absence of measurable thickness variations or surface defects across multiple randomly selected regions suggests uniform film formation during solution casting and solvent evaporation.

Figure 6
Cross-sectional images of (a) pure PES and (b) PES/[emim][Tf2N] membrane at 20% loading.

Figure 7 presents the EDX spectra and elemental mapping for both the neat PES membrane and the PES/[emim][Tf₂N] membrane containing 20% ionic liquid. PES membrane displays characteristic peaks for carbon, oxygen, and sulfur at approximately 0.2, 0.5, and 2.5 keV, respectively, which is consistent with previously reported spectra for PES [³¹]. In contrast, the PES/[emim][Tf₂N] membrane shows additional peaks corresponding to nitrogen at 0.39 keV and fluorine at 0.57 keV. These findings confirm the successful incorporation of [emim][Tf₂N] ionic liquid into the PES polymer matrix. The elemental mapping further demonstrates a uniform distribution of fluorine and nitrogen throughout the membrane, indicating that the ionic liquid is well dispersed and does not form segregated domains. This observation confirms the presence of the ionic liquid within the membrane matrix and demonstrates a uniform elemental distribution at the micrometer scale, consistent with homogeneous incorporation of [emim][Tf₂N] into the PES matrix. This homogeneous incorporation supports the morphological observations and reinforces the compatibility between PES and [emim][Tf₂N] at the optimized loading.

Figure 7
EDX mapping and spectral results of PES and PES/[emim][Tf2N] membrane at 20% loading.

It is necessary to point out that SEM and EDX analyses provide structural and compositional information at the micrometer scale and are therefore semi-quantitative in nature. Within the resolution limits of these techniques, no phase-separated domains, pores, or localized ionic-liquid-rich regions were observed across the membrane surface or cross section. This indicates the effective macroscopic homogeneity of the PES/[emim][Tf₂N] membranes. While nanoscale ionic liquid domains cannot be fully observed by SEM/EDX, any significant microstructural heterogeneity would result in irregular gas permeation behavior. Therefore, the observed reproducibility of membrane performance and low experimental variability (±7%) points towards uniform ionic liquid distribution within the polymer matrix.

Figure 8 presents the FTIR spectra of the neat PES membrane and the PES/[emim][Tf₂N] membrane containing 20% ionic liquid. The characteristic aromatic bands of PES are clearly visible in the range of 1479–1586 cm⁻¹, corresponding to benzene ring stretching [²¹]. Additional peaks associated with the sulfone group (S=O) appear between 1147–1299 cm⁻¹, while the asymmetric aromatic ether linkage (C–O–C) is evident in the 1255–1011 cm⁻¹ region. These spectral features match well with previously reported FTIR signatures of PES, confirming that the polymer backbone remains intact.

Figure 8
FTIR spectra of PES and PES/[emim][Tf2N] membrane at 20% loading.

The spectrum of the PES/[emim][Tf₂N] membrane exhibits all major PES peaks, indicating that the incorporation of the ionic liquid does not induce any chemical modification of the polymer. Two additional absorption bands at approximately 760 and 858 cm⁻¹ are attributed to the EMIM cation [²², ³⁰], confirming the presence of the ionic liquid within the membrane structure. Overall, the FTIR results suggest that [emim][Tf₂N] is physically dispersed within the PES matrix without the formation of new chemical bonds.

The thermal degradation profiles of the synthesized membranes are shown in Figure 9. No apparent weight loss was observed at lower temperatures for both membranes which indicates the absence of moisture or residual solvent. The major decomposition step occurs between 400 °C and 600 °C, corresponding to the simultaneous thermal degradation of both the PES matrix and the ionic liquid. PES/[emim][Tf₂N] membrane display slightly lower thermal stability compared to neat PES, since [emim][Tf₂N] begins to decompose earlier than the polymer. This observation is consistent with the reported degradation range of [emim][Tf₂N] (418–448 °C) [³²], which aligns well with the trend in Figure 9. Additionally, PES/[emim][Tf₂N] membrane shows a higher final char residue than the pure PES membrane, further confirming the presence of the ionic liquid within the membrane matrix.

Figure 9
TGA profile of PES and PES/[emim][Tf2N] membrane at 20% loading.

4. CONCLUSIONS

In this study, we have successfully utilized high-performance PES/[emim][Tf₂N] ionic liquid-polymeric membranes for process optimization of the biogas upgradation process produced from camel dung. The membranes were synthesized by a simple physical blending technique and systematically optimized using response surface methodology with a central composite design. From the statistical analysis, it was observed that ionic liquid loading has the most significant influence in enhancing CO₂ permeability and improving CO₂/CH₄ selectivity. On the other hand, feed pressure showed an inverse relationship with gas permeabilities for both gases but a positive impact on CO₂/CH₄ selectivity. From the optimization analysis, it was found that 20 wt.% loading of [emim][Tf₂N] ionic liquid gave the best separation performance in terms of CO₂ permeability and CO₂/CH₄ selectivity at 5 bar feed pressure for the biogas upgradation process. This membrane has shown excellent agreement between predicted and experimental values. While the reported separation performance is based on ideal single-gas measurements, the identified optimum operating conditions are expected to remain valid for mixed-gas biogas systems, as supported by established dual-mode sorption behavior. Characterization results confirm that this membrane exhibited a dense and homogenous structure with uniform distribution of ionic liquid on the polymer matrix, as deduced by SEM and EDX analyses, respectively. FTIR analysis confirmed the physical blending without the formation of intermediates or crosslinks, and TGA results confirmed the excellent thermal stability of PES/[emim][Tf₂N] membrane. The optimized PES/[emim][Tf₂N] membranes are promising to be used in the designs of decentralized biogas upgrading units in rural and desert societies where camel husbandry is common and simple, and robust separation technologies are required. This work not only advances the field of ionic-liquid-based membranes but also contributes to transforming local waste streams into valuable clean- energy resources within Saudi Arabia.

5. ACKNOWLEDGMENTS

This research was funded by the “Camel Research Grant Program” offered by the Saudi Ministry of Culture. All opinions expressed herein belong to the researchers and do not necessarily reflect those of the Ministry of Culture.

6. BIBLIOGRAPHY

• ^[1] ABDEL-RAHMAN, E.H., ALZAHRANI, F.S., SULIEMAN, A.M.E.H., “Potential of camel dung as promising organic manure in Saudi Arabia”, Advancements in Life Sciences, v. 7, n. 4, pp. 202–207, 2020.
• ^[2] ALHARBI, M., ALSEROURY, F., ALKTHAMI, B., “Biogas production from manure of camel and sheep using tomato and rumen as co-substrate”, Journal of Ecological Engineering, v. 24, n. 11, pp. 54–61, 2023. doi: https://doi.org/10.12911/22998993/170984.
  » https://doi.org/10.12911/22998993/170984
• ^[3] ZACHARIAS, A., How camel waste is fuelling the UAE’s economy, Abu Dhabi, Then National News, 2019.
• ^[4] RAHMAN, M.M., “The potential of biomethane produced from waste landfill to supplement renewable energy in Saudi Arabia”, Engineering, Technology & Applied Scientific Research, v. 14, n. 6, pp. 18264–18270, 2024. doi: https://doi.org/10.48084/etasr.8985.
  » https://doi.org/10.48084/etasr.8985
• ^[5] AGUILLOSO, G., ARPIA, K., KHAN, M., et al., “Recent advances in membrane technologies for biogas upgrading”, Engineering Proceedings, v. 67, n. 1, pp. 57, 2024. doi: https://doi.org/10.3390/engproc2024067057.
  » https://doi.org/10.3390/engproc2024067057
• ^[6] TOMCZAK, W., GRYTA, M., DANILUK, M., et al., “Biogas upgrading using a single-membrane system: a review”, Membranes, v. 14, n. 4, pp. 80, 2024. doi: https://doi.org/10.3390/membranes14040080. PubMed PMID: 38668108.
  » https://doi.org/10.3390/membranes14040080
• ^[7] JUSOH, N., YEONG, Y.F., LOCK, S.S.M., et al., “Biomethane generation from biogas upgrading by means of thin-film composite membrane comprising Linde T and fluorinated polyimide: optimization of fabrication parameters”, RSC Advances, v. 10, n. 6, pp. 3493–3510, 2020. doi: https://doi.org/10.1039/C9RA06358G. PubMed PMID: 35497748.
  » https://doi.org/10.1039/C9RA06358G
• ^[8] ZIA UL MUSTAFA, M., BIN MUKHTAR, H., MD NORDIN, N.A.H., et al., “Recent developments and applications of ionic liquids in gas separation membranes”, Chemical Engineering & Technology, v. 42, n. 12, pp. 2580–2593, 2019. doi: https://doi.org/10.1002/ceat.201800519.
  » https://doi.org/10.1002/ceat.201800519
• ^[9] FRIESS, K., IZÁK, P., KÁRÁSZOVÁ, M., et al., “A review on ionic liquid gas separation membranes”, Membranes, v. 11, n. 2, pp. 97, 2021. doi: https://doi.org/10.3390/membranes11020097. PubMed PMID: 33573138.
  » https://doi.org/10.3390/membranes11020097
• ^[10] MANNAN, H.A., NASIR, R., MUKHTAR, H., et al. “Role of ionic liquids in eliminating interfacial defects in mixed matrix membranes”, In: Goh, K.L., Aswathi, Silva, R.T., et al. (eds.), Interfaces in Particle and Fibre Reinforced Composites, Sawston, Woodhead Publishing, pp. 269-309, 2020. doi: https://doi.org/10.1016/B978-0-08-102665-6.00011-X.
  » https://doi.org/10.1016/B978-0-08-102665-6.00011-X
• ^[11] KARUNAKARAN, M., VILLALOBOS, L.F., KUMAR, M., et al., “Graphene oxide doped ionic liquid ultrathin composite membranes for efficient CO2 capture”, Journal of Materials Chemistry. A, Materials for Energy and Sustainability, v. 5, n. 2, pp. 649–656, 2017. doi: https://doi.org/10.1039/C6TA08858A.
  » https://doi.org/10.1039/C6TA08858A
• ^[12] CARDOSO, J.S., LIN, Z., BRITO, P., et al., “The functionalization of PES/SAPO-34 mixed matrix membrane with [emim][Tf2N] ionic liquid to improve CO2/N2 separation properties”, Inorganics, v. 11, n. 11, pp. 447, 2023. doi: https://doi.org/10.3390/inorganics11110447.
  » https://doi.org/10.3390/inorganics11110447
• ^[13] SALEEMI, S., MANNAN, H.A., RIAZ, T., et al., “Optimizing synergistic silica–zinc oxide coating for enhanced flammability resistance in cotton protective clothing”, Fibers, v. 12, n. 5, pp. 44, 2024. doi: https://doi.org/10.3390/fib12050044.
  » https://doi.org/10.3390/fib12050044
• ^[14] MASOR, A., NASIR, R., MANNAN, H.A., et al., “Performance optimization of polyetherimide‐zeolite 4A mixed matrix membranes for carbon dioxide/methane separation process using response surface methodology”, Materialwissenschaft und Werkstofftechnik, v. 54, n. 5, pp. 571–585, 2023. doi: https://doi.org/10.1002/mawe.202200220.
  » https://doi.org/10.1002/mawe.202200220
• ^[15] MAQSOOD, K., ALI, A., ILYAS, S.U., et al., “Multi-objective optimization of thermophysical properties of multiwalled carbon nanotubes based nanofluids”, Chemosphere, v. 286, n. Pt 2, pp. 131690, 2022. doi: https://doi.org/10.1016/j.chemosphere.2021.131690. PubMed PMID: 34352553.
  » https://doi.org/10.1016/j.chemosphere.2021.131690
• ^[16] SEMAN, S.Z.A., IDRIS, I., ABDULLAH, A., et al., “Optimizing purity and recovery of biogas methane enrichment process in a closed landfill”, Renewable Energy, v. 131, pp. 1117–1127, 2019. doi: https://doi.org/10.1016/j.renene.2018.08.057.
  » https://doi.org/10.1016/j.renene.2018.08.057
• ^[17] CARRILLO-SÁNCHEZ, L., TÉLLEZ, C., CORONAS, J., “Membrane preparation assisted by integration of machine learning and response surface methodology for CO2 separation”, Journal of Membrane Science, v. 736, pp. 124708, 2025. doi: https://doi.org/10.1016/j.memsci.2025.124708.
  » https://doi.org/10.1016/j.memsci.2025.124708
• ^[18] JUSOH, N., YEONG, Y.F., LAU, K.K., et al., “Transport properties of mixed matrix membranes encompassing zeolitic imidazolate framework 8 (ZIF-8) nanofiller and 6FDA-durene polymer: optimization of process variables for the separation of CO2 from CH4”, Journal of Cleaner Production, v. 149, pp. 80–95, 2017. doi: https://doi.org/10.1016/j.jclepro.2017.02.069.
  » https://doi.org/10.1016/j.jclepro.2017.02.069
• ^[19] ZHANG, X., XU, L., TIAN, S., et al., “Novel approach based on source separation for conversion of fresh urine to valuable resources in rural villages”, Energy & Environmental Sustainability, v. 1, n. 1, pp. 100005, 2025. doi: https://doi.org/10.1016/j.eesus.2025.02.003.
  » https://doi.org/10.1016/j.eesus.2025.02.003
• ^[20] LIU, H., LIANG, L., TIAN, F., et al., “Scalable preparation of ultraselective and highly permeable fully aromatic polyamide nanofiltration membranes for antibiotic desalination”, Angewandte Chemie International Edition in English, v. 63, n. 23, pp. e202402509, 2024. doi: https://doi.org/10.1002/anie. 202402509. PubMed PMID: 38588046.
  » https://doi.org/10.1002/anie.202402509
• ^[21] ABDUL MANNAN, H., MUKHTAR, H., SHIMA SHAHARUN, M., et al., “Polysulfone/poly(ether sulfone) blended membranes for CO2 separation”, Journal of Applied Polymer Science, v. 133, n. 5, pp. app.42946, 2016. doi: https://doi.org/10.1002/app.42946.
  » https://doi.org/10.1002/app.42946
• ^[22] MANNAN, H.A., NASIR, R., MUKHTAR, H., et al., “Enhanced structural and separation performance of PES nanocomposite membranes via imidazolium IL and modified silica”, Journal of Membrane Science, v. 738, pp. 124893, 2026. doi: https://doi.org/10.1016/j.memsci.2025.124893.
  » https://doi.org/10.1016/j.memsci.2025.124893
• ^[23] ABDULRAHMAN, A., ALI, A., ALFAZAZI, A., “Synthesis and process parameter optimization of biodiesel from jojoba oil using response surface methodology”, Arabian Journal for Science and Engineering, v. 46, n. 7, pp. 6609–6617, 2021. doi: https://doi.org/10.1007/s13369-020-05302-y.
  » https://doi.org/10.1007/s13369-020-05302-y
• ^[24] QADIR, D., NASIR, R., MANNAN, H.A., et al., “Optimization and performance studies of NFDK membrane for ionic separation from aqueous solutions”, Chemicke Zvesti, v. 76, n. 3, pp. 1815–1831, 2022. doi: https://doi.org/10.1007/s11696-021-01990-4.
  » https://doi.org/10.1007/s11696-021-01990-4
• ^[25] HANBAZAZAH, A.S., ALI, A., ALSAADY, M., et al., “Optimization and experimental analysis of sustainable solar collector efficiency under the influence of magnetic nanofluids”, Applied Nanoscience, v. 12, n. 12, pp. 3859–3870, 2022. doi: https://doi.org/10.1007/s13204-022-02533-3.
  » https://doi.org/10.1007/s13204-022-02533-3
• ^[26] NAIDUA, T., QADIR, D., NASIR, R., et al., “Utilization of moringa oleifera and nanofiltration membrane to treat palm oil mill effluent (POME)”, Materialwissenschaft und Werkstofftechnik, v. 52, n. 3, pp. 346–356, 2021. doi: https://doi.org/10.1002/mawe.202000084.
  » https://doi.org/10.1002/mawe.202000084
• ^[27] AKHTAR, F.H., KUMAR, M., VOVUSHA, H., et al., “Scalable synthesis of amphiphilic copolymers for CO2- and water-selective membranes: effect of copolymer composition and chain length”, Macromolecules, v. 52, n. 16, pp. 6213–6226, 2019. doi: https://doi.org/10.1021/acs.macromol.9b00528.
  » https://doi.org/10.1021/acs.macromol.9b00528
• ^[28] KARUNAKARAN, M., KUMAR, M., SHEVATE, R., et al., “CO2-Philic Thin Film Composite Membranes: Synthesis and Characterization of PAN-r-PEGMA Copolymer”, Polymers, v. 9, n. 7, pp. 219, 2017. doi: https://doi.org/10.3390/polym9070219. PubMed PMID: 30970944.
  » https://doi.org/10.3390/polym9070219
• ^[29] ALI, A., ILYAS, S.U., DANISH, M., et al., “Multi-objective optimization of thermophysical properties of f–Al2O3 nano-dispersions in heat transfer oil”, SN Applied Sciences, v. 3, n. 2, pp. 230, 2021. doi: https://doi.org/10.1007/s42452-021-04256-6.
  » https://doi.org/10.1007/s42452-021-04256-6
• ^[30] ZIA-UL-MUSTAFA, M., MUKHTAR, H., NORDIN, N.A.H.M., et al., “Effect of imidazolium based ionic liquids on PES membrane for CO2/CH4 separation”, Materials Today: Proceedings, v. 16, pp. 1976–1982, 2019. doi: https://doi.org/10.1016/j.matpr.2019.06.076.
  » https://doi.org/10.1016/j.matpr.2019.06.076
• ^[31] HAIDER, M.S., SHAO, G.N., IMRAN, S.M., et al., “Aminated polyethersulfone-silver nanoparticles (AgNPs-APES) composite membranes with controlled silver ion release for antibacterial and water treatment applications”, Materials Science and Engineering C, v. 62, pp. 732–745, 2016. doi: https://doi.org/10.1016/j.msec.2016.02.025. PubMed PMID: 26952479.
  » https://doi.org/10.1016/j.msec.2016.02.025
• ^[32] LIU, S.-H., CHEN, C.C., ZHANG, B., et al., “Fire and explosion hazards of 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide”, RSC Advances, v. 10, n. 38, pp. 22468–22479, 2020. doi: https://doi.org/10.1039/D0RA01821J. PubMed PMID: 35514589.
  » https://doi.org/10.1039/D0RA01821J

Publication Dates

• Publication in this collection
  27 Apr 2026
• Date of issue
  2026

History

• Received
  05 Dec 2025
• Accepted
  27 Feb 2026