CO2 Plasticization Resistance Membrane for Natural Gas Sweetening Process: Defining Optimum Operating Conditions for Stable Operation

Membranes with a stable performance during the natural gas sweetening process application are highly demanded. This subject has been immensely explored due to several challenges faced by conventionally used polymeric membranes, especially the high tendency of plasticization and physical aging. In this study, polysulfone (PSf) hollow-fiber membrane was formulated and tested for its application in natural gas sweetening based on several compositions of CO2/CH4 mixed gas. The effects of operating conditions such as pressure, temperature and CO2 feed composition on separation performance were analyzed. The findings showed that the formulated membrane exhibited decreasing CO2 permeation trend with the increase in pressure. Conversely, the increase in operating temperature boosted the CO2 permeation. High productivity can be attained at higher operating temperatures with a reduction in product purity. Interestingly, since PSf has higher plasticization pressure, it was not affected by the change in CO2 percentage up to 70% CO2. The experimental study showed that the membrane material formulated in this study can be potentially evaluated at the field stage. Longer testing duration is needed with the real feed gas, appropriate pre-treatment based on the material limitations, and optimum operating conditions at the site to further confirm the membrane’s long-term lifetime, resistance, and stability.


Introduction
Gas separation is an attractive technology for a wide range of chemical processes. Gas separation membranes have been predicted to have a compound annual growth rate (CAGR) of 5.87% by 2027 [1]. Asia Pacific represents the largest and fastest growing market, with increasing consumption predicted from countries such as Japan, India, and China [1]. The global gas separation membrane market was valued at USD 957.47 million in 2021 [1], owing to the efforts of many governments to decrease greenhouse gas emissions in order to mitigate issues related to global warming [2]. Using a gas separation membrane, the permeate, which consists mainly of CO 2 gas, can be injected into depleted reservoirs to reduce its emission or improve oil and shale oil recovery. CO 2 is miscible with oil and can induce oil swelling hence reducing its viscosity. Many studies have shown that CO 2 can be used for enhanced oil and gas recovery and geological storage through lab experiments, field tests, molecular and reservoir simulation [3,4].
The membrane materials used for commercial application are mainly cellulose acetate, PSf, and polyimide. However, the classical polymeric membrane often suffers from plasticization, which leads to lower performance [5]. The use of polyimide polymer for the separation increased from 16 and was maintained at 17 for equimolar CO 2 and CH 4 , while CO 2 /CH 4 diffusivity selectivity was reduced from 2 to 1. 5. This resulted in the overall CO 2 /CH 4 perm-selectivity being reduced from 30.5 to 24.5. CO 2 -induced swelling caused a faster diffusion of CH 4 in the mixed gas case. CTA (Cellulose Tri-Acetate) was found to be less susceptible to the detrimental effects of CO 2 -induced swelling, which caused a reduction of the diffusivity-selectivity since its solubility-selectivity is more dominant. CTA permselectivity also displayed a weaker dependence on the CO 2 content in the mixture [23].
Miandoab performed a simulation of multi-component gas, including water, in humidified versus dry feed conditions. The membrane model shows that a simplistic model that does not account for such complex effects can deviate more than 20% from a more rigorous approach. The impact of competitive sorption and plasticization by CO 2 and H 2 O-induced free volume blocking was significantly greater, leading to fugacity-dependent permeabilities for CO 2 and CH 4 that were below those of the pure gas components. Relative to the simplified models, the new model predicts differences up to 2% and 18% in CH 4 recovery at low feed flow rates, and the difference in CO 2 removal can be as significant as 50% [24].
Genduso found that the permeability selectivity of the mixture strongly diverges from the pure-gas trend. At about 18 atm partial fugacity, almost 35% of the ideal permeability selectivity was lost, and it was assumed that competitive sorption was the cause for this loss of permeability selectivity. The data of mixed-gas solubilities at 50 mol% equilibrium concentration showed that competitive sorption strongly affects CH 4 mixed-gas solubility, that is, when partial pressures increase, the CH 4 mixed-gas solubility coefficients diverge from the pure-gas values. Because the effect of competitive sorption on the solubility coefficient of CO 2 is limited, they found that at 2 atm CO 2 partial pressure, the CO 2 /CH 4 solubility selectivity increased from~5 in the pure-gas state to~10 in the mixture [25].
Operating conditions such as pressure, temperature, and flow rate can influence the resistance of the membrane during its application which will, in turn, impact the working lifetime of the membrane. The product purity and productivity are determined by various factors such as membrane properties, which are also helped by selecting the optimum operating conditions [26]. The accurate choice of the best-operating conditions enables the membrane to function continuously with good performance for years with minimal hydrocarbon and permeance loss over time.
Most previous works reported were based on CO 2 and CH 4 separation using low pressure with single gas and simple binary gas. Multi-component gas mixture permeation properties were not widely studied due to the difficulty and risk of handling the operating parameters at a higher level and the need to analyze more than two components simultaneously [14]. Single gas testing has overlooked the effects of competitive sorption, plasticization, compaction, pore blocking, and aging, which upset the membrane performance when tested in the field under harsh conditions. Furthermore, raw natural gas usually contains heavy hydrocarbon [27], and the CO 2 contaminant concentration may vary from field to field in the range of 15% to 70%. For less than 15% CO 2 , the absorption technology is usually applied.
The objective of this research was to study the performance of a formulated glassy PSf membrane when subjected to various operating conditions. The aim was to identify the optimum setting that will be used for its field application with stable operation. Experiments using pure and mixtures of five gases typically found in natural gas with various compositions were carried out. Effects of operating temperature (25 to 45 • C), pressure (20 to 50 barg), and CO 2 composition (15%, 20%, 40% and 70%) variation were investigated. The effect of operating conditions on the rubbery membrane can be compared and is referred to the study conducted by Sadrzadeh [14]. The operating conditions employed for glassy and rubbery membranes are different since the main mechanism of transport is different. The former is governed by diffusivity, whereas the latter involves a solubility mechanism. Furthermore, the rubbery polymers is lacking in terms of selectivity and have lower overall performance [28].

Theory
The importance of operating temperature determination for the selected membrane material for gas separation application is related to its transport properties which are diffusion and solubility. One of the main vital phenomena that happens in gas separation membranes is the Joule Thompson effect. Theoretically, a gas flows through a membrane from the highpressure side (subscript 1) to the low-pressure side (subscript 2), and it is assumed to happen adiabatically (no heat transfer (q = 0) and the whole system is isolated) and isenthalpic (constant enthalpy). Hence, the internal energy (U) change of this process is: (1) The temperature change in this process is called JT coefficient (µ JT ) and is represented by the differential equation ( ∂T /∂P) H . The µ JT of CO 2 gas is 1.11 K/bar, while for CH 4 and N 2 it is 0.70 and 0.25 K/bar, respectively. This demonstrates that when CO 2 is removed at high pressure, it causes a significant temperature decrease. Researchers must monitor the temperature changes caused by CO 2 removal to ensure that the membrane performance is maintained. It is also crucial to ensure the membrane material of choice can sustain the possibility of liquid formation resulting from the temperature drop to the dew point of the feed gas. Therefore, studying membrane performance with respect to temperature variation is imperative.
The operating temperature and the gas diffusion coefficient relationship in polymeric membranes can be illustrated by the Arrhenius equation [18]: where D A denotes the diffusion coefficient (cm 2 /s) of gas A. D A0 , E D , T, and R are the pre-exponential factor, the activation energy of diffusion (in kJ/mol), the temperature (in K), and the universal gas constant, respectively. Here, the value of the R is 8.31 J/mol. Equally, the Van't Hoff equation describes solubility as a function of temperature [29].
S A0 and ∆H S are the pre-exponential factors and the enthalpy of the solution of gas A, respectively. The gas permeability can be stated as a function of temperature as follows: P A0 and E P are the pre-exponential factor, and the activation energy of permeation, respectively.
It is worth noting that as temperature rises, so do permeation and diffusion parameters, but the solubility relationship reverses [30].

Materials
The hollow fiber membranes were prepared from a dope solution composed of PSf from Solvay Advanced Polymers, and a mixture of solvent and alcohol bought from Sigma-Aldrich (St. Louis, MO, USA) with appropriate composition. The name and composition of the chemicals used in this study are not mentioned in accordance with the company's unpublished confidential information guideline. The polymer pellets were dried at 60 • C in a vacuum oven overnight before use, while solvents and non-solvent were used as received.

Membrane Characterization
FESEM (S4800, Hitachi, Japan) is used to study the resultant membrane morphology for the cross-sectional image. The sample is sputter-coated with an ultrathin conductive layer, namely titanium, at high vacuum evaporation to ensure the sample is electrically conductive. This pre-treatment could avoid the accretion of a static electric field at the sample induced by electron irradiation employed during the measurement process. The membrane samples are cut into square shape sheets of size 0.5 cm × 0.5 cm and attached to a sample stub using carbon tape. For cross-sectional morphology, the membranes are fractured in liquid nitrogen and placed on a sample stub using carbon tape. Scanning is conducted at a magnification between ×500 and ×10 k.

Membrane Performance Testing
The hollow fiber membrane module made from stainless steel (SS 316) is used in this study (refer to Figure 1). The membrane element is installed in the membrane casing that comprises of two removable parts. Rubber O-rings are used to ensure pressure-tight sealing between the tube sheet and the casing at both ends.
from Solvay Advanced Polymers, and a mixture of solvent and alcohol bought from Sigma-Aldrich (St. Louis, MO, USA) with appropriate composition. The name and composition of the chemicals used in this study are not mentioned in accordance with the company's unpublished confidential information guideline. The polymer pellets were dried at 60 °C in a vacuum oven overnight before use, while solvents and non-solvent were used as received.

Membrane Characterization
FESEM (S4800, Hitachi, Japan) is used to study the resultant membrane morphology for the cross-sectional image. The sample is sputter-coated with an ultrathin conductive layer, namely titanium, at high vacuum evaporation to ensure the sample is electrically conductive. This pre-treatment could avoid the accretion of a static electric field at the sample induced by electron irradiation employed during the measurement process. The membrane samples are cut into square shape sheets of size 0.5 cm × 0.5 cm and attached to a sample stub using carbon tape. For cross-sectional morphology, the membranes are fractured in liquid nitrogen and placed on a sample stub using carbon tape. Scanning is conducted at a magnification between ×500 and ×10 k.

Membrane Performance Testing
The hollow fiber membrane module made from stainless steel (SS 316) is used in this study (refer to Figure 1). The membrane element is installed in the membrane casing that comprises of two removable parts. Rubber O-rings are used to ensure pressure-tight sealing between the tube sheet and the casing at both ends. The prepared hollow fibers are put in the bundle and potted into Swagelok stainless steel tubing. The leak test is performed to ensure the integrity of the fiber and module prepared. Pure gas testing is conducted at low pressure (3 to 7 barg) using the bubble test method to determine the membrane ideal and intrinsic CO2 permeance and selectivity. Mixed gas testing is later conducted with variations of feed pressure (20 to 50 barg), temperature variation (25 to 45 °C), and CO2 percentage variation (15%, 20%, 40% and 70%) The prepared hollow fibers are put in the bundle and potted into Swagelok stainless steel tubing. The leak test is performed to ensure the integrity of the fiber and module prepared. Pure gas testing is conducted at low pressure (3 to 7 barg) using the bubble test method to determine the membrane ideal and intrinsic CO 2 permeance and selectivity. Mixed gas testing is later conducted with variations of feed pressure (20 to 50 barg), temperature variation (25 to 45 • C), and CO 2 percentage variation (15%, 20%, 40% and 70%) to determine the membrane performance close to real field environment conditions. Both pure and mixed gas measurements are determined using a permeation apparatus that allows the measurement of permeability and selectivity at various operating conditions. For the mixed gas testing, the feed, non-permeate and permeate stream gaseous composition are analyzed using gas chromatography. In this study, Micro GC Agilent 490 (Agilent Technologies, Santa Clara, CA, USA) is used for the gas composition measurement. Mass flow controller is also needed for the mass balance calculation. CO 2 permeance is calculated by using the following equations.
where: x, y = Component P x = Differential pressure of x component (barg) P FR x = Mean differential difference of component x between feed and retentate (barg) P F = Feed pressure (barg) The operating parameters are selected based on the real application needs. Operating pressure of 40 barg is usually applied in commercial field applications, and 20 barg has also been tested in this study as the lowest possible operating pressure, although it is not quite practical for the natural gas sweetening process; 50 bar is provided to provide a comparison with 40 barg data, and 30 barg data can be interpolated from 20 barg and 40 barg data.
This study focuses on mixed gas testing because the transport behavior of one component through the membrane is affected by the presence of other penetrants, which can be termed as coupling effect or competitive sorption. Permeation properties of multicomponent gas mixtures are still lacking in the literature due to the complexity of conducting the testing, difficulty in controlling the operating parameters, and simultaneous analysis of multi-component [14]. Computational molecular modeling could reduce the amount of tedious and complex testing that needs to be conducted since a thorough screening can be first performed through simulation [26].
Mixed gas compositions used in this study are listed in Table 1: For the long-term performance study, the membrane is subjected to 40% mixed gas composition at 40 barg operating pressure and 45 • C operating temperature using a system that can recycle the feed gas. Relative numbers are the value which is relative to the reference value. The relative change is used in this paper basically to compare the trends. This comparison is generally expressed as the ratio and this number does not have any units. The actual membrane performance in this study is not mentioned in accordance with the company's unpublished confidential information guideline.

Membrane Characterizations
The integrally skinned asymmetric membrane with a spongy structure and thin dense skin layer is observed in membrane cross-section images, as shown in Figure 2. The desired membrane structure for gas separation has successfully been formulated and fabricated for further performance evaluation, which will be explained in the next sections. No finger-like structure or macro void formation can be found, which is detrimental for highpressure applications. Macro void, a large pore, can frequently be detected in asymmetric membranes. However, this structure is only applicable for liquid separation, such as ultrafiltration, and it is also very useful in drug delivery systems, but for reverse osmosis process and gas separation, this macro void structure is not favorable [12]. For the long-term performance study, the membrane is subjected to 40% mixed gas composition at 40 barg operating pressure and 45 °C operating temperature using a system that can recycle the feed gas.
Relative numbers are the value which is relative to the reference value. The relative change is used in this paper basically to compare the trends. This comparison is generally expressed as the ratio and this number does not have any units. The actual membrane performance in this study is not mentioned in accordance with the company's unpublished confidential information guideline.

Membrane Characterizations
The integrally skinned asymmetric membrane with a spongy structure and thin dense skin layer is observed in membrane cross-section images, as shown in Figure 2. The desired membrane structure for gas separation has successfully been formulated and fabricated for further performance evaluation, which will be explained in the next sections. No finger-like structure or macro void formation can be found, which is detrimental for high-pressure applications. Macro void, a large pore, can frequently be detected in asymmetric membranes. However, this structure is only applicable for liquid separation, such as ultrafiltration, and it is also very useful in drug delivery systems, but for reverse osmosis process and gas separation, this macro void structure is not favorable [12].  The membrane performances in terms of CO2 permeance and CO2/CH4 selectivity have been studied based on several important parameters, as discussed in the following sections.  The membrane performances in terms of CO 2 permeance and CO 2 /CH 4 selectivity have been studied based on several important parameters, as discussed in the following sections.

Effect of the Feed Pressure on the CO 2 Permeation of the PSf Hollow Fibre Membranes at Fixed Operating Temperature
Feed pressure provides the key driving force for the gas separation membrane [2]. The determination of feed pressure is important for the equipment pipe rating, the pretreatment and post-treatment requirements, as well as the cost of booster and export compressors. Techno-economic evaluation needs to be performed before selecting the optimum operating pressure for the process design. Type 2 polymer exhibit decreasing permeability with increasing pressure, and the permeability of gas begins to increase as the pressure reaches the CO 2 plasticization pressure [31]. Too high operating pressure will result in a supercritical gas condition that should also be avoided. In this condition, CO 2 exhibits properties and behaviors between liquid and gas, where the density is liquidlike while the diffusivity, surface tension, and viscosity are gas-like. Supercritical CO 2 is considered an organic solvent with remarkable solvation power, which should be avoided when used with polymeric membranes.
In this experimental analysis, the impact of CO 2 permeance with respect to the CO 2 composition and feed pressure changes has been investigated at a fixed operating temperature of 45 • C. As the driving force is increased from 20 to 50 barg, it can be seen in Figure 3 that the CO 2 permeance reduced significantly for all the trends despite the variation in CO 2 %. This can be explained by several phenomena that took place in the membrane during the separation process, namely competitive sorption, pore blocking, site saturation, aging, and compaction, which greatly affect the membrane performance. These phenomena have overcome the higher-pressure driving force that is supposed to enhance the permeation. There are many mechanisms leading to free volume loss, and numerous theories have been invented to define the aging process in terms of free volume reduction [12]. This has negatively impacted the membrane and explains the declining trends shown in the graph, since the transport behavior of PSf is diffusivity dominant. In this case, if CO 2 plasticization is more dominant than aging and compaction, the CO 2 permeation trend should increase as the pressure increases, as is typical for cellulose-based membranes. This is because plasticization will swell the membrane, while aging and compaction will make the membrane denser. These two effects counterbalance each other, but for the cellulose case, which has lower plasticization pressure, the plasticization effect is greater than aging and compaction, as can be seen in their increasing permeability trends during the testing or operation.
However, in this case, since PSf has high plasticization pressure, aging and compaction are more dominant, leading to the decreasing trend of CO 2 permeation. The permeability coefficient of gases at higher feed pressure usually does not obey the dual mode sorption model due to plasticization phenomena happening in the polymer [32].
The trend for CO 2 permeance with pressure increment is subject to the polymer material. Ismail and Lorna [31] explained the three types of permeability trends for several types of glassy polymers. Type 1 is for polymers such as PSf and polycarbonate, which demonstrate a reducing permeability trend throughout the testing pressure up to 30 barg. Type 2 polymers exhibit decreasing permeability with increasing pressure, and the permeability of gas begins to increase as the pressure reaches the CO 2 plasticization pressure. An example of a type 2 polymer is polyimide which follows dual mode sorption behavior. Type 3 polymers such as cellulose acetate and polystyrene experience an increasing permeability trend from the beginning of the testing as they have low CO 2 plasticization pressure and behave as rubbery permeation behavior with low selectivity [33]. At the same CO 2 percentage, increasing the pressure from 20 to 50 barg, increased the CO 2 partial pressure and the CO 2 solubility of the membrane material. Furthermore, the CO 2 plasticization effect, dilates the membrane, which also counterbalances the shrinkage effect from aging, compaction, and site saturation. This, in turn, increases the permeability when the dilation is more than the shrinkage. In this case, it will be advantageous for Type 2 and Type 3 membrane materials if the selectivity is still within an acceptable level and does not drop significantly. The increasing trend of CO 2 permeation is helped by the solubility coefficient that is increased with the increment of pressure. In this condition, membrane densification happens and hinders diffusion, but solubility can still happen through the polymer matrix. As for the PSf, it has higher CO 2 plasticization pressure, thus, it follows the Type 1 trend when the pressure is increased, which explains the trend in Figure 3. greater than aging and compaction, as can be seen in their increasing permeability trends during the testing or operation. However, in this case, since PSf has high plasticization pressure, aging and compaction are more dominant, leading to the decreasing trend of CO2 permeation. The permeability coefficient of gases at higher feed pressure usually does not obey the dual mode sorption model due to plasticization phenomena happening in the polymer [32]. The trend for CO2 permeance with pressure increment is subject to the polymer material. Ismail and Lorna [31] explained the three types of permeability trends for several types of glassy polymers. Type 1 is for polymers such as PSf and polycarbonate, which demonstrate a reducing permeability trend throughout the testing pressure up to 30 barg. Type 2 polymers exhibit decreasing permeability with increasing pressure, and the permeability of gas begins to increase as the pressure reaches the CO2 plasticization pressure. An example of a type 2 polymer is polyimide which follows dual mode sorption behavior. Type 3 polymers such as cellulose acetate and polystyrene experience an increasing permeability trend from the beginning of the testing as they have low CO2 plasticization pressure and behave as rubbery permeation behavior with low selectivity [33]. At the same CO2 percentage, increasing the pressure from 20 to 50 barg, increased the CO2 partial pressure and the CO2 solubility of the membrane material. Furthermore, the CO2 plasticization effect, dilates the membrane, which also counterbalances the shrinkage effect from aging, compaction, and site saturation. This, in turn, increases the permeability when the dilation is more than the shrinkage. In this case, it will be advantageous for Type 2 and Type 3 membrane materials if the selectivity is still within an acceptable level and does not drop significantly. The increasing trend of CO2 permeation is helped by the solubility coefficient that is increased with the increment of pressure. In this condition, membrane densification happens and hinders diffusion, but solubility can still happen through the polymer matrix. As for the PSf, it has higher CO2 plasticization pressure, thus, it follows the Type 1 trend when the pressure is increased, which explains the trend in Figure 3. The same trend for selectivity is observed when the pressure is increased from 20 to 50 barg, where the trends show a decreasing trend for all CO 2 % variation, as displayed in Figure 4. The selectivity values for all the mixed gas testing are lower than pure gas values. This is because the pressure force also pushes the slower gas i.e., methane to permeate. The increasing pressure has resulted in both reductions of CO 2 permeance due to compaction and aging and reduction of selectivity for PSf membrane that has diffusivity as the transport dominant factor.

Effect of the Feed Pressure on the CO2/CH4 Selectivity of the PSf Hollow Fiber Membranes at Fixed Operating Temperature
The same trend for selectivity is observed when the pressure is increased from 20 to 50 barg, where the trends show a decreasing trend for all CO2% variation, as displayed in Figure 4. The selectivity values for all the mixed gas testing are lower than pure gas values. This is because the pressure force also pushes the slower gas i.e., methane to permeate. The increasing pressure has resulted in both reductions of CO2 permeance due to compaction and aging and reduction of selectivity for PSf membrane that has diffusivity as the transport dominant factor.

Effect of the Feed Temperature on the CO2 Permeation of the PSF Hollow Fiber Membranes at Fixed Operating Pressure
Feed temperature is one of the most significant parameters for membrane operation as it affects the transport properties. The impact is more prominent for diffusivity-dominan polymers, where diffusion increases with rising temperature, whereas for solubility-

Effect of the Feed Temperature on the CO 2 Permeation of the PSF Hollow Fiber Membranes at Fixed Operating Pressure
Feed temperature is one of the most significant parameters for membrane operation as it affects the transport properties. The impact is more prominent for diffusivity-dominan polymers, where diffusion increases with rising temperature, whereas for solubilitydominating membranes, the effect is reversed, with increasing temperature lowering solubility. As mentioned by Duthie diffusion at high temperatures is mostly contributed by Henry's law species and plasticization effects are conquered by Henry's law dissolution [6]. At lower temperatures, the movement of Langmuir component species also contributes to the total diffusion coefficient [6].
Operating temperature also considerably impacts the robustness of the membrane if the setting is below the feed gas dewpoint after the CO 2 removal and liquid are allowed to form during operation. The liquid is formed when condensation happens in the gas separation membrane during the separation of CO 2 from natural gas. There are two effects that will result in condensation. Firstly, the bulk CO 2 removal that permeates faster than the heavy hydrocarbon makes the gas heavier, and the dew point increases towards the operating temperature. Secondly, the gas cools down due to Joule Thompson (JT) effect as it passes through the membrane. The change in fluid temperature happens upon expansion (pressure decrease), causing a significant change to the temperature between the feed gas, retentate gas, and the permeate gas. The drastic temperature change results in liquid formation and greatly affects the permeation properties in which the permeance is reduced by pore blocking. On the other hand, the selectivity is enhanced due to the hydrocarbon permeation is also blocked by the presence of liquid. In many cases, polymers are contacted with liquids that penetrate their structure and plasticize it [7], and in this case, the selectivity will drop.
Experimental studies have been performed to explore the impact of operating temperature changes on the formulated membrane performance. By doing this investigation, the optimum temperature can be set to obtain the best performance while avoiding liquid formation by providing sufficient superheating. In this experiment, condensation cannot be simulated, and it requires heavier gases in the feed gas to demonstrate the liquid formation usually found in the field.
Based on Figure 5, when the temperature increases, the increasing trend can be seen for the CO 2 permeance for all cases of CO 2 percentage following the Arrhenius equation [2]. However, the mixed gas number is lower than the pure gas value. Similar findings have been reported by other researchers [2,31,34]. This is because, at higher temperatures, the diffusion increases as the gas molecule's kinetic energy and gas collision increases beyond its minimum activation energy, especially for the fast gas. This is advantageous for the polymers governed by diffusivity as the dominant transport behavior. Conversely, the solubility has an opposing trend with the temperature, where the higher the temperature, the lower the solubility. When the increase in diffusivity is higher than the decrease in solubility for a given polymer material, the overall permeability is higher.
At high temperatures, the polymer matrix is also hypothesized to increase segmental mobility resulting in easier penetration of the fast gas through the thin skin layer of an asymmetric membrane. This shows that a higher temperature can be set as this membrane operating temperature to gain the benefit of higher CO 2 permeance with a certain limit to control the other effect, such as aging. Further investigation is needed for the long-term effect under continuous exposure of the polymer to higher pressure and temperature. At higher operating temperatures, perm selectivity is shifted to a higher permeability value [35]. The best gas permselective membrane is reached when high selectivity combined with high permeability can be attained simultaneously [14].
Condensation can be prevented for membrane materials that are sensitive to liquid hydrocarbon by heating the feed gas and inter-step membrane (refer to Figure 6) to the optimum operating temperature. This is to provide a sufficient margin of superheat based on the predetermined dew point value that can prevent the membrane from being affected by the liquid formation. Therefore, a higher operating temperature operation is favorable for the membrane that is not robust to liquid and has diffusivity as the dominant transport mechanism. However, the impact of aging and other potential issues due to long-term operation at high temperatures needs to be further investigated.
be simulated, and it requires heavier gases in the feed gas to demonstrate the liquid formation usually found in the field. Based on Figure 5, when the temperature increases, the increasing trend can be seen for the CO2 permeance for all cases of CO2 percentage following the Arrhenius equation [2]. However, the mixed gas number is lower than the pure gas value. Similar findings have been reported by other researchers [2,31,34]. This is because, at higher temperatures, the diffusion increases as the gas molecule's kinetic energy and gas collision increases beyond its minimum activation energy, especially for the fast gas. This is advantageous for the polymers governed by diffusivity as the dominant transport behavior. Conversely, the solubility has an opposing trend with the temperature, where the higher the temperature, the lower the solubility. When the increase in diffusivity is higher than the decrease in solubility for a given polymer material, the overall permeability is higher. Figure 5. CO2 permeance result for mixed gas testing vs. variation of CO2 composition and feed temperature. Note: Pure gas testing condition is 25 °C temperature with 3 to 7 barg applied pressure.
At high temperatures, the polymer matrix is also hypothesized to increase segmental mobility resulting in easier penetration of the fast gas through the thin skin layer of an asymmetric membrane. This shows that a higher temperature can be set as this membrane operating temperature to gain the benefit of higher CO2 permeance with a certain limit to control the other effect, such as aging. Further investigation is needed for the long-term effect under continuous exposure of the polymer to higher pressure and temperature. At higher operating temperatures, perm selectivity is shifted to a higher permeability value Figure 5. CO 2 permeance result for mixed gas testing vs. variation of CO 2 composition and feed temperature. Note: Pure gas testing condition is 25 • C temperature with 3 to 7 barg applied pressure.
Polymers 2021, 13, x 12 of 19 [35]. The best gas permselective membrane is reached when high selectivity combined with high permeability can be attained simultaneously [14]. Condensation can be prevented for membrane materials that are sensitive to liquid hydrocarbon by heating the feed gas and inter-step membrane (refer to Figure 6) to the optimum operating temperature. This is to provide a sufficient margin of superheat based on the predetermined dew point value that can prevent the membrane from being affected by the liquid formation. Therefore, a higher operating temperature operation is favorable for the membrane that is not robust to liquid and has diffusivity as the dominant transport mechanism. However, the impact of aging and other potential issues due to long-term operation at high temperatures needs to be further investigated.

Effect of the Feed Temperature on the CO2/CH4 Selectivity of the PSf Hollow Fiber Membranes at Fixed Operating Pressure
When the temperature is increased from 25 to 45 °C, the selectivity decreases for all the CO2% variations, as shown in Figure 7. The decrease in selectivity can be explained by the kinetic energy and diffusivity increment for both fast and slow gas, i.e., CO2 and CH4, as the temperature is increased. The selectivity is reduced because the CH4 gas also has higher activation energy at higher temperatures, which increases permeation. The competition impact between CH4 and CO2 has resulted in a declining trend in selectivity value as the temperature increases [2]. This is why membranes with high selectivity at room temperature may produce low separation at elevated temperatures [35]. Low-temperature operation could minimize the amount of 'slow gases' permeation through the membrane since they have lower energy. There is a tradeoff between CO2 permeability and selectivity that needs to be considered when applying high-temperature application. If the selectivity value at high-temperature operation is still at an acceptable limit, then the setting can be used for the continuous membrane operation application.

Effect of the Feed Temperature on the CO 2 /CH 4 Selectivity of the PSf Hollow Fiber Membranes at Fixed Operating Pressure
When the temperature is increased from 25 to 45 • C, the selectivity decreases for all the CO 2 % variations, as shown in Figure 7. The decrease in selectivity can be explained by the kinetic energy and diffusivity increment for both fast and slow gas, i.e., CO 2 and CH 4, as the temperature is increased. The selectivity is reduced because the CH 4 gas also has higher activation energy at higher temperatures, which increases permeation. The competition impact between CH 4 and CO 2 has resulted in a declining trend in selectivity value as the temperature increases [2]. This is why membranes with high selectivity at room temperature may produce low separation at elevated temperatures [35]. Low-temperature operation could minimize the amount of 'slow gases' permeation through the membrane since they have lower energy. There is a tradeoff between CO 2 permeability and selectivity that needs to be considered when applying high-temperature application. If the selectivity value at high-temperature operation is still at an acceptable limit, then the setting can be used for the continuous membrane operation application.

Effect of the Feed Composition on the CO2 Permeation and Selectivity of the PSf Hollow Fiber Membranes
Based on Figure 8a,b, it could be summarized that the CO2 permeance trend is quite stable for the PSf membrane formulated in this study. This happens even though the CO2 partial pressure is increased with the increment of CO2 composition [26] from 15% to 70%, within the same operating pressure of 40 barg. Higher CO2 concentration at the same operating pressure leads to higher CO2 partial pressure, which is supposed to boost CO2 permeance to a certain extent. Plasticization is expected to happen following the CO2 concentration level as CO2 is a strong plasticizer agent. Using a polyimide membrane, Duthie found that as the concentration of CO2 is boosted to a higher value, the polymer swells due to polymer relaxation, leading to an increase in the diffusion coefficient [6]. However, for the PSf membrane prepared in this study, demonstrated higher polymer rigidity and strong plasticization resistance to CO2 gas, as well as a stable trend as the CO2% increased.

Effect of the Feed Composition on the CO 2 Permeation and Selectivity of the PSf Hollow Fiber Membranes
Based on Figure 8a,b, it could be summarized that the CO 2 permeance trend is quite stable for the PSf membrane formulated in this study. This happens even though the CO 2 partial pressure is increased with the increment of CO 2 composition [26] from 15% to 70%, within the same operating pressure of 40 barg. Higher CO 2 concentration at the same operating pressure leads to higher CO 2 partial pressure, which is supposed to boost CO 2 permeance to a certain extent. Plasticization is expected to happen following the CO 2 concentration level as CO 2 is a strong plasticizer agent. Using a polyimide membrane, Duthie found that as the concentration of CO 2 is boosted to a higher value, the polymer swells due to polymer relaxation, leading to an increase in the diffusion coefficient [6]. However, for the PSf membrane prepared in this study, demonstrated higher polymer rigidity and strong plasticization resistance to CO 2 gas, as well as a stable trend as the CO 2 % increased.
According to Fick's Law, diffusion increases as the concentration gradient increases, and as per Henry's law, the solubility increases as the partial pressure increases. It is noted that at higher CO 2 composition, the effect of plasticization will become more severe. However, since the material used in this study is PSf which is known to have very high plasticization pressure [36], the highest condition used in this experiment, which is 40 barg with 70% CO 2 , resulting in 28 barg of CO 2 partial pressure. This partial pressure is still below the plasticization pressure of PSf, which is around 34 barg, making the trend looks stable for both permeability and selectivity as the CO 2 increased from 15% to 70%. Under this condition, most membrane materials would have experienced certain degrees of plasticization. Based on Figure 9a,b, it could be summarized that the selectivity trend is quite stable for both variations of pressure (constant temperature) and variation of temperature (constant pressure) for the formulated PSf membrane used in this study. This happened even though the CO 2 partial pressure is increased with the increment of CO 2 composition [26] from 15% to 70%, within the same operating temperature of 45 • C. This shows the integrity and rigidity of the PSf membrane as the pressure increased, reflecting its higher plasticization pressure. According to Fick's Law, diffusion increases as the concentration gradient increases, and as per Henry's law, the solubility increases as the partial pressure increases. It is noted that at higher CO2 composition, the effect of plasticization will become more severe. However, since the material used in this study is PSf which is known to have very high plasticization pressure [36], the highest condition used in this experiment, which is 40 barg with 70% CO2, resulting in 28 barg of CO2 partial pressure. This partial pressure is still below the plasticization pressure of PSf, which is around 34 barg, making the trend looks stable for Based on Figure 9a,b, it could be summarized that the selectivity trend is quite stable for both variations of pressure (constant temperature) and variation of temperature (constant pressure) for the formulated PSf membrane used in this study. This happened even though the CO2 partial pressure is increased with the increment of CO2 composition [26] from 15% to 70%, within the same operating temperature of 45 °C. This shows the integrity and rigidity of the PSf membrane as the pressure increased, reflecting its higher plasticization pressure.
(a) (b) Figure 9. Relative selectivity vs. variation of CO2 feed composition; varies operating pressure (a) and varies operating temperature (b). Dotted line is the trendline as a guide. Figure 9. Relative selectivity vs. variation of CO 2 feed composition; varies operating pressure (a) and varies operating temperature (b). Dotted line is the trendline as a guide.

Long-Term Performance Study of the Membrane
A long-term performance study is performed to investigate the stability of the membrane performance over time. It has been well-established that thin films aged faster than thick films [26]. As plasticization is also a time-dependent phenomenon, the degree of plasticization needs to be further investigated, if any. Normally, physical aging takes place slowly over a long period of time, with the membrane gradually approaching equilibrium from the initial non-equilibrium state induced during the fabrication stage [12].
Based on Figure 10, throughout 30 days of testing with 40% mixed gas at 40 barg and 45 • C, the membrane performance is stable with only a minor decline due to com-paction and aging. The graph achieved a plateau trend on the 15th day, indicating the high stability of the membrane. White mentioned that the membrane films from multiple polymers, for example, PSf, polyimide Matrimid, and PPO (polyphenylene oxide) with 0.4 to 60 microns in thickness, were periodically tested in a laboratory for up to 10,000 h [26]. He had shown a substantial reduction in permeability trend when those membranes were stored at ambient conditions while not in testing instead of keeping them at constant conditions [26]. Ma and Koros proposed that the frequent changes in CO 2 partial pressure for the asymmetric hollow-fiber membranes affect the long-term performance of the membrane as compared to steady and constant CO 2 exposure which will yield more stable membrane performance [37]. This can be further investigated in field conditions, which will have some fluctuations due to operational issues and sudden trips or unplanned shutdowns. Constant condition is proposed to ensure long-term membrane performance; hence proper procedure needs to be in place if the fluctuation happens.
A long-term performance study is performed to investigate the stability of the membrane performance over time. It has been well-established that thin films aged faster than thick films [26]. As plasticization is also a time-dependent phenomenon, the degree of plasticization needs to be further investigated, if any. Normally, physical aging takes place slowly over a long period of time, with the membrane gradually approaching equilibrium from the initial non-equilibrium state induced during the fabrication stage [12].
Based on Figure 10, throughout 30 days of testing with 40% mixed gas at 40 barg and 45 °C, the membrane performance is stable with only a minor decline due to compaction and aging. The graph achieved a plateau trend on the 15th day, indicating the high stability of the membrane. White mentioned that the membrane films from multiple polymers, for example, PSf, polyimide Matrimid, and PPO (polyphenylene oxide) with 0.4 to 60 microns in thickness, were periodically tested in a laboratory for up to 10,000 h [26]. He had shown a substantial reduction in permeability trend when those membranes were stored at ambient conditions while not in testing instead of keeping them at constant conditions [26]. Ma and Koros proposed that the frequent changes in CO2 partial pressure for the asymmetric hollow-fiber membranes affect the long-term performance of the membrane as compared to steady and constant CO2 exposure which will yield more stable membrane performance [37]. This can be further investigated in field conditions, which will have some fluctuations due to operational issues and sudden trips or unplanned shutdowns. Constant condition is proposed to ensure long-term membrane performance; hence proper procedure needs to be in place if the fluctuation happens. As shown in Figure 11, some polymers with high separation performance in laboratory testing degrade more drastically than conventional polymers such as polyimide and cellulose acetate membrane [26,38]. This can be explained by the presence of contaminants in natural gas. This effect must be considered when one starts to formulate a new membrane. The contaminants in gas fields are H2S, heavy hydrocarbon [39], aromatic, mercury, water, and many others. Investigation using mixed gas with 5 components is still not enough and further study is needed with cyclic testing to test the membrane durability, robustness, and lifetime, which is critically important in maintaining the high performance of the membrane formulated in the laboratories. As shown in Figure 11, some polymers with high separation performance in laboratory testing degrade more drastically than conventional polymers such as polyimide and cellulose acetate membrane [26,38]. This can be explained by the presence of contaminants in natural gas. This effect must be considered when one starts to formulate a new membrane. The contaminants in gas fields are H 2 S, heavy hydrocarbon [39], aromatic, mercury, water, and many others. Investigation using mixed gas with 5 components is still not enough and further study is needed with cyclic testing to test the membrane durability, robustness, and lifetime, which is critically important in maintaining the high performance of the membrane formulated in the laboratories.
Heavy hydrocarbon contamination, even at low concentrations, is hypothesized to be the reason for some membrane failures and loss in performance. According to the dual mode sorption theory, highly condensable feed components, such as toluene, will contribute to severe competition impact due to their size and based on its critical temperature, it can be easily condensed at the operating pressure applied. Toluene can block permeation opportunities of CO 2 and made the membrane performance drop [39][40][41][42][43][44][45][46]. The impact of heavy hydrocarbon, including aromatic hydrocarbon, on the membrane system, has been lacking due to the complexity of the components present and the difficulty in characterizing their effects experimentally [39]. Heavy hydrocarbon, including aromatic hydrocarbon, also can severely swell and plasticize the polymer. Figure 11. Illustration of PI (Polyimide) and CA (Cellulpse Acetate) membrane trend going from pure gas testing to the field conditions testing results [26].
Heavy hydrocarbon contamination, even at low concentrations, is hypothesized to be the reason for some membrane failures and loss in performance. According to the dual mode sorption theory, highly condensable feed components, such as toluene, will contribute to severe competition impact due to their size and based on its critical temperature, it can be easily condensed at the operating pressure applied. Toluene can block permeation opportunities of CO2 and made the membrane performance drop [39][40][41][42][43][44][45][46]. The impact of heavy hydrocarbon, including aromatic hydrocarbon, on the membrane system, has been lacking due to the complexity of the components present and the difficulty in characterizing their effects experimentally [39]. Heavy hydrocarbon, including aromatic hydrocarbon, also can severely swell and plasticize the polymer.

Conclusions
Commonly, the permeation behavior of gases through membranes is related mainly to the properties of gases and membranes along with the operating conditions [14]. Based on the study results, it can be concluded that the formulated PSf membrane has a decrease in CO2 permeation as pressure rises due to gas diffusivity being blocked. Increasing the operating temperature can help to boost CO2 diffusion and permeation. High productivity can be attained at higher operating temperatures with a reduction in product purity. At higher partial pressure conditions, when the CO2 percentage is increased, the CO2 productivity and product purity remain stable due to the rigidity and integrity of the membrane produced. Interestingly, since PSf has higher plasticization pressure, it is unaffected by changes in CO2 percentage up to 70% CO2. The present study shows that the membrane material formulated has the potential for further evaluation at the field stage. Longer testing duration at the optimum operating conditions on site is needed with the real feed gas, which includes heavy hydrocarbon and aromatic components, to confirm the actual membrane lifetime, robustness, durability, reliability, and stability.

Conclusions
Commonly, the permeation behavior of gases through membranes is related mainly to the properties of gases and membranes along with the operating conditions [14]. Based on the study results, it can be concluded that the formulated PSf membrane has a decrease in CO 2 permeation as pressure rises due to gas diffusivity being blocked. Increasing the operating temperature can help to boost CO 2 diffusion and permeation. High productivity can be attained at higher operating temperatures with a reduction in product purity. At higher partial pressure conditions, when the CO 2 percentage is increased, the CO 2 productivity and product purity remain stable due to the rigidity and integrity of the membrane produced. Interestingly, since PSf has higher plasticization pressure, it is unaffected by changes in CO 2 percentage up to 70% CO 2 . The present study shows that the membrane material formulated has the potential for further evaluation at the field stage. Longer testing duration at the optimum operating conditions on site is needed with the real feed gas, which includes heavy hydrocarbon and aromatic components, to confirm the actual membrane lifetime, robustness, durability, reliability, and stability.