Morphology Effect of Zinc Oxide Nanoparticles on the Gas Separation Performance of Polyurethane Mixed Matrix Membranes for CO2 Recovery from CH4, O2, and N2

The effect of the morphology and content of zinc oxide nanoparticles (ZnO-NPs) on the physicochemical, mechanical, and gas transport properties of the polyurethane (PU) mixed matrix membranes (MMMs) with respect to CO2 recovery from CH4, O2, and N2 was studied. The MMMs based on PU with spherical and rod-shaped ZnO-NPs at various loadings, namely, 0.05, 0.1, 0.5, 1, and 2 wt. %, were prepared with membrane density control and studied using AFM, wettability measurements, surface free energy calculation, gas separation and mechanical testing. To evaluate the resistance of the ZnO-NPs to agglomeration in the polymer solutions, zeta potential was determined. The ZnO-NPs with average cross sectional size of 30 nm were obtained by plasma-enhanced chemical vapor deposition (PECVD) from elemental high-purity zinc in a zinc-oxygen-hydrogen plasma-forming gas mixture. It was established that the spherical ZnO-NPs are promising to improve the gas performance of PU-based MMMs for CO2 recovery from natural gas, while the rod-shaped NPs better demonstrate their potential in capturing CO2 in flue gases.


Introduction
Nowadays, one of the most promising and energy efficient approaches to carbon dioxide recovery from natural [1][2][3] and flue [3][4][5][6] gases is membrane gas separation providing the process in the absence of phase transitions at room temperature.
To implement the membrane gas separation technology, inorganic and polymeric materials are used. As a rule, inorganic membranes have exceptional chemical, mechanical and thermal stability, exhibit higher gas flows and can withstand higher pressures in contrast to polymeric ones [7]. However, their higher cost and ability to be processed into modules for large-scale applications are the main challenges in using these materials for gas separation [7,8]. In this context, polymeric membranes are the mainstay of commercialization due to their ease of processing, manufacturing, and availability. However, the use of polymeric membranes does not always solve the problem of obtaining the optimal ratio of selectivity and permeability. So, researchers resort to using various composite and hybrid PU Elastollan ® 1190 A were purchased from BASF (Ludwigshafen, Germany). Tetrahydrofuran (THF) were supplied from Chimreactive (Nizhny Novgorod, Russia) and purified by distillation before using. Liquids for wettability tests and density measurements (diiodomethane (99%), glycerol (99%), and ethylene glycol (99.5%)) were purchased from Merck (Darmstadt, Germany).
Zinc of 5N purity (Changsha Rich Nonferrous Metals Co., Ltd., Hunan, Changsha, China), high purity hydrogen (99.9999%) and high purity oxygen 6.0 (99.9999%) (Horst Technologies Ltd., Dzerzhinsk, Nizhny Novgorod region, Russia) were used as components of a plasma-forming mixture. Hydrogen was used as a carrier gas, which also acted as a temperature stabilizer and regulator of the growth of nanostructures.

Plasma-Chemical Synthesis of ZnO-NPs
Spherical and rod-shaped ZnO-NPs were obtained by plasma-enhanced chemical vapor deposition (PECVD) under optical emission spectrometry control from elemental high-purity zinc in a zinc-oxygen-hydrogen plasma-forming gas mixture. The installation for the plasma-chemical synthesis was described in [60,61].
As the first step, the installation was evacuated to a pressure of 1 × 10 −3 Pa for several hours to remove traces of nitrogen and water from the walls of the reactor. After that, the source with zinc was heated to 370 • C in the case of the spherical NPs and to 470 • C in the case of the rod-shaped ones. The temperature of the collecting powder tank was maintained at a level of 250 • C. Until the system entered the operating mode, the powder tank was closed with a magnetic diaphragm of a special design. The total gas flow through the plasma-chemical reactor was set equal to 30 mL min −1 at a total pressure in the system of 0.1 Pa. The plasma discharge power was 100 W in the case of the spherical NPs and 70 W in the case of the rod-shaped ones.

Preparation of Membranes
PU membranes without and with loading ZnO-NPs were fabricated by corresponding polymer solutions casting using an automatic casting machine MemcastPlus (POROME-TER, Nazareth, Belgium) onto a glass substrate under ambient conditions (20 • C, 10 5 Pa). The polymer membranes were easily peeled off the substrates after solvent evaporation and desiccated under vacuum for reaching the constant mass. The thickness of the obtained membranes was about 60 µm.
THF volume of required to obtain 5 wt. % PU solutions was divided into two parts. In the first one, the polymer was dissolved at 70 • C with the help of a heated stirrer; in the second one, ZnO-NPs were dispersed in an ultrasonic bath for 10 min. After that, both parts of THF were mixed with stirring and further ultrasonication for 10 min. The content of ZnO-NPs (0.05, 0.1, 0.5, 1, and 2 wt. %) was calculated relative to the weight of PU.
As a result, the following series of samples was obtained: PU (without adding ZnO-NPs), PU/ZnO-NPs(sph) in the case of the spherical NPs, and PU/ZnO-NPs(rod) in the case of the rod-shaped ones with ZnO-NPs loadings of 0.05, 0.1, 0.5, 1, and 2 wt. %.

Zeta Potential
The resistance of the ZnO-NPs to agglomeration in the polymer solutions was evaluated by determining zeta potential. The measurements were conducted with a generation zeta potential instrument Stabino ® II (Microtrac Inc., Montgomeryville, PA, USA) using a streaming potential method. The prepared solution of PU/ZnO-NPs in THF was additionally ultrasonicated for dispersion and degassing. The samples were measured for 10 min with a step of 10 s; the obtained values were averaged for each time interval.

Density Measurements
To control the densities of the membranes, the "flotation" method was used. The measurements were conducted as follows: a graduated cylinder (25 cm 3 ) with ground joint and the plug was filed to a half with a mixture of ethylene glycol (ρ = 1.11 g/cm 3 @ 20 • C) and glycerol (ρ = 1.26 g/cm 3 @ 20 • C) in the volume ratio 1:1. These liquids were chosen because they do not react with the measured samples and do not cause swelling and solvation [62]. After that, a piece of a membrane sample was put down into the cylinder. Based on the position of the membrane sample in the liquid, more of a low density or a high-density liquid was added and mixed with the glass mixer so that a membrane sample takes an equilibrium position in the middle of the liquid column according to the graduation. Then a density of the obtained mixture of liquids was measured using a pycnometer (10 cm 3 ). Density was measured five times on each membrane.

Atomic Force Microscopy
The surface of the membranes was studied by atomic force microscopy (AFM) using a scanning probe microscope SPM-9700 (Shimadzu, Kyoto, Japan). AFM scanning was performed using a tapping mode by silicon vibrating cantilevers PointProbe FMR-20 (NanoWorld Innovative Technologies, Neuchâtel, Switzerland) with a stiffness coefficient of 1.3 N/m and a typical tip radius of no more than 8 nm (guaranteed-no more than 12 nm); a tip height was 15 µm. After AFM scanning, an arithmetic average roughness height (R a ) and a mean roughness depth (R z ) were obtained. A base length was 10 µm. Processing of the obtained AFM images and their analysis were performed using a software SPM Manager ver. 4.02 (Shimadzu, Kyoto, Japan).

Scanning Electron Microscopy and Energy Dispersive X-ray Spectroscopy
The stoichiometry and size-morphological characteristics of the ZnO-NPs were studied by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) using an electron microscope JSM-IT300LV (JEOL, Peabody, MA, USA) with an electron probe diameter of about 5 nm and a probe current of less than 0.5 nA (operating voltage 20 kV). SEM scanning was performed using low-energy secondary electrons and backscattered electrons under a low vacuum to eliminate the charge.

Wettability Measurements and Surface Free Energy Calculation
Wettability tests were carried out with using three test liquids with different surface tensions (Table 1): water, glycerol, and diiodomethane. A drop of test liquid was applied on a membrane surface in an enclosed area filled with the corresponding liquid's vapor. After reaching the equilibrium state, a drop image was taken, and the contact angle of wetting (θ) was determined using ImageJ software with contact angle plugin. The measurements were conducted at 20 • C. The results were collected for a series of five drops with contact angle deviations did not exceed ±1 • . Total surface free energy and its polar and dispersive components were calculated based on wettability test results using the Owens-Wendt method [63,64].

Gas Separation Experiments
The pure gas permeability coefficients (P) of CO 2 , CH 4 , O 2 , and N 2 through the membranes were measured by an experimental setup ( Figure 1) equipped with an automatic computing system based on a software-logic controller V130-33-RA22 (Unitronics, Airport City, Israel) at the initial transmembrane pressure of 110 kPa and ambient temperature (20 • C) in a constant volume mode. Each single-gas test was repeated at least three times. The selectivity (α) of the polymeric membranes was calculated as the ratio of the single gas permeability coefficients. Total surface free energy and its polar and dispersive components were calculated based on wettability test results using the Owens-Wendt method [63,64].

Gas Separation Experiments
The pure gas permeability coefficients (P) of CO2, CH4, O2, and N2 through the membranes were measured by an experimental setup ( Figure 1) equipped with an automatic computing system based on a software-logic controller V130-33-RA22 (Unitronics, Airport City, Israel) at the initial transmembrane pressure of 110 kPa and ambient temperature (20 °C) in a constant volume mode. Each single-gas test was repeated at least three times. The selectivity (α) of the polymeric membranes was calculated as the ratio of the single gas permeability coefficients.

Mechanical Testing
The mechanical properties, namely, breaking strength (σ, MPa) and elongation at break (ε, %), were measured using Universal Tensile Machine Z005 (ZWICK, Ennepetal Germany) in tension elongation experiments at a tension rate of 50 mm/min under ambient conditions. The polymer samples were prepared as a flat rectangular shape (50 mm × 10 mm) with original film thickness. The data were collected for ten measurements for each membrane sample and processed with ZWICK Tensile Machine software testControl II (ZWICK, Ennepetal, Germany). Figure 2 shows the SEM images of the spherical and rod-shaped ZnO-NPs obtained by the PECVD method; the average size of the NPs in cross section 30 nm with coefficients of variation of 17% and 23%, respectively.

Mechanical Testing
The mechanical properties, namely, breaking strength (σ, MPa) and elongation at break (ε, %), were measured using Universal Tensile Machine Z005 (ZWICK, Ennepetal, Germany) in tension elongation experiments at a tension rate of 50 mm/min under ambient conditions. The polymer samples were prepared as a flat rectangular shape (50 mm × 10 mm) with original film thickness. The data were collected for ten measurements for each membrane sample and processed with ZWICK Tensile Machine software testControl II (ZWICK, Ennepetal, Germany). Total surface free energy and its polar and dispersive components were calculated based on wettability test results using the Owens-Wendt method [63,64].

Gas Separation Experiments
The pure gas permeability coefficients (P) of CO2, CH4, O2, and N2 through the membranes were measured by an experimental setup ( Figure 1) equipped with an au tomatic computing system based on a software-logic controller V130-33-RA22 (Unitron ics, Airport City, Israel) at the initial transmembrane pressure of 110 kPa and ambien temperature (20 °C) in a constant volume mode. Each single-gas test was repeated at leas three times. The selectivity (α) of the polymeric membranes was calculated as the ratio o the single gas permeability coefficients.

Mechanical Testing
The mechanical properties, namely, breaking strength (σ, MPa) and elongation a break (ε, %), were measured using Universal Tensile Machine Z005 (ZWICK, Ennepetal Germany) in tension elongation experiments at a tension rate of 50 mm/min under am bient conditions. The polymer samples were prepared as a flat rectangular shape (50 mm × 10 mm) with original film thickness. The data were collected for ten measurements for each membrane sample and processed with ZWICK Tensile Machine software testCon trol II (ZWICK, Ennepetal, Germany).  The EDS analysis data for the obtained ZnO-NPs are shown in Table 2.

Zeta Potential of Polymer Solutions
According to the zeta potential measurements (Figure 3), the negative charge in the polymer solutions was declined due to loading ZnO-NPs. Moreover, the charge drop was more noticeable for the rod-shaped NPs in contrast to the spherical ones. However, in all cases, the value of the zeta potential did not decrease below 25 mV modulo, so, it could be concluded that the ZnO-NPs are well resistant to agglomeration in the polymer solutions and the latter have a high degree of stability [65]. The EDS analysis data for the obtained ZnO-NPs are shown in Table 2.

Zeta Potential of Polymer Solutions
According to the zeta potential measurements (Figure 3), the negative charge in the polymer solutions was declined due to loading ZnO-NPs. Moreover, the charge drop was more noticeable for the rod-shaped NPs in contrast to the spherical ones. However, in all cases, the value of the zeta potential did not decrease below 25 mV modulo, so, it could be concluded that the ZnO-NPs are well resistant to agglomeration in the polymer solutions and the latter have a high degree of stability [65].

Membrane Surface
The AFM data of the membranes based on the various PU solutions with ZnO-NPs are shown in Figure 4.

Membrane Surface
The AFM data of the membranes based on the various PU solutions with ZnO-NPs are shown in Figure 4.
The surface of the samples is loosely packed but with low roughness. It is noteworthy that, first, the surface irregularities decreased with loading ZnO-NPs of both types and, then, increased with the growth in the NPs content above 0.5 wt. %. This effect could be related to the compensation of voids in the polymer due to the NPs. Moreover, the smoothing of the membrane surface was more expressed in the case of loading the spherical ZnO-NPs in contrast to the rod-shaped ones. This difference might be due to the varying degrees of resistance to agglomeration of the NPs with different shapes in the initial polymer solutions (Figure 3). The higher the stability of the system, the lower the degree of the NPs agglomeration and the smaller the perturbations introduced into the surface relief formation of the membranes.
It should be noted that the results of the AFM study for the reverse side of the polymer membranes were similar ( Figure S1), which indicated the absence or minimal presence of the NPs sedimentation process during film formation.
According to the wettability measurements, it was found that loading the ZnO-NPs into the polymer matrix caused not only a change in the relief of the membrane surface but also a redistribution of polar and nonpolar fragments on it. This led to a change in the nature of the surface wettability of the PU membranes (Table 3)  The surface of the samples is loosely packed but with low roughness. It is noteworthy that, first, the surface irregularities decreased with loading ZnO-NPs of both types and, then, increased with the growth in the NPs content above 0.5 wt. %. This effect could be related to the compensation of voids in the polymer due to the NPs. Moreover, the smoothing of the membrane surface was more expressed in the case of loading the spherical ZnO-NPs in contrast to the rod-shaped ones. This difference might be due to the varying degrees of resistance to agglomeration of the NPs with different shapes in the initial polymer solutions (Figure 3). The higher the stability of the system, the lower the degree of the NPs agglomeration and the smaller the perturbations introduced into the surface relief formation of the membranes.
It should be noted that the results of the AFM study for the reverse side of the polymer membranes were similar ( Figure S1), which indicated the absence or minimal presence of the NPs sedimentation process during film formation. It is noteworthy that with an increase in the NPs' loading into the polymer matrix, the wettability of the membranes' surface with the polar liquids (water, glycerol) mainly decreased and then increased again. The opposite tendency was observed in the case of nonpolar diiodomethane. Such behavior could be associated with a combination of two possible factors, namely, the ZnO presence on the surface of the membranes and a change in their roughness. Probably, the first factor contributed due to the self-energy of the ZnO surface, and the second contributed due to the redistribution of the PU polar and non-polar segments during a physical change in the relief of the membrane.
Based on the wettability data, the specific surface free energy (with its polar and dispersive components) of the polymeric membranes was calculated using the Owens-Wendt method ( Figure 5). It is noteworthy that with an increase in the NPs' loading into the polymer matrix, the wettability of the membranes' surface with the polar liquids (water, glycerol) mainly decreased and then increased again. The opposite tendency was observed in the case of nonpolar diiodomethane. Such behavior could be associated with a combination of two possible factors, namely, the ZnO presence on the surface of the membranes and a change in their roughness. Probably, the first factor contributed due to the self-energy of the ZnO surface, and the second contributed due to the redistribution of the PU polar and non-polar segments during a physical change in the relief of the membrane.
Based on the wettability data, the specific surface free energy (with its polar and dispersive components) of the polymeric membranes was calculated using the Owens-Wendt method ( Figure 5). With loading spherical ZnO-NPs into the PU matrix, a decrease in the polar surface energy component was observed with a minimum at 0.5 wt. % of the NPs. The same minimum was observed with loading the rod-shaped ZnO-NPs. However, the maximum values slightly exceeded the values of the polar energy for the pure PU membrane.
At the same time, the dispersion surface energy component in the case of both NPs' types decreased with respect to the pure PU membrane. Such tendency might be due to an increase in the fractional free volume (FFV) [66] of the polymer membranes with loading the ZnO-NPs. This fact is indirectly confirmed by the theory described in the works [63,67] and a decrease in membrane density after modification (Table 4).
It should be noted that the membrane densities' decrease with loading the ZnO NPs into the PU matrix was probably associated with an intensification of the NPs' agglomeration from an increase in their content. With an increase of the NPs' agglomeration, a forced rearrangement of PU macromolecules probably occurred, which caused a decrease in the density of the membranes.
The essence of the theory mentioned above lies in the inversely dependence between the dispersion component of the surface free energy in polymeric membranes and their FFV, which has the directly dependence with the total gas permeability [63,67]. The fact is that the dispersion component is understood as the sum of van der Waals and other non-sitespecific interactions between a polymer surface and an applied liquid. I.e., the dispersion component has only physical origins and depends on the morphological/supramolecular structure of the polymer. The looser the packed structure, the smaller the dispersion component of its surface energy. At the same time, the polar component of the surface free energy in polymeric membranes is understood as the sum of site-specific interactions such as electrostatic, hydrogen bonding, etc., between a polymer surface and an applied liquid. I.e., a change in the polar component can be associated with the corresponding redistribution of polar/nonpolar segments of polymer chains in membranes. Thus, an increase in the polar surface energy indicates an increase in the polar segment proportion on the polymer Membranes 2022, 12, 577 9 of 15 surface, and vice versa. In this case, the presence of polar segments of the polymer chain on the membrane surface will directly affect its interaction with separated gases. At that, plasticizing gases are more susceptible to this effect than non-plasticizing ones, since they are prone to site-specific interactions.

Gas Separation Properties of Membranes
As exhibited in Figure 6, at all membrane samples, the gas permeability of CO 2 was higher than those of CH 4 , O 2 , and N 2 . Such tendency was probably associated with intrinsic affinity of ZnO which could render its NPs to interact with CO 2 more than the other three gases and consequently led higher adsorption capacity for CO 2 [58,68]. In addition to the affinity of the gases to the membrane filler, the difference in the kinetic diameters of the gas molecules could also made a contribution (Table 5).
However, the different degrees of the changes in the permeability with the ZnO NPs loading into the PU matrix indicated that the affinity of the gases to the membrane filler had the greatest contribution to the formation of the gas permeability.  In addition to the affinity of the gases to the membrane filler, the difference in the kinetic diameters of the gas molecules could also made a contribution (Table 5). However, the different degrees of the changes in the permeability with the ZnO NPs loading into the PU matrix indicated that the affinity of the gases to the membrane filler had the greatest contribution to the formation of the gas permeability.
Notably, loading the ZnO-NPs into the PU matrix caused an increase in the CO 2 permeability of the polymer membranes with a further decrease below the permeability of pure PU (Figure 6). The maximum growth occurred at 0.5 wt. % NPs loading. Moreover, the peaks of the CO 2 permeability for the membranes were 30% and 38% in the case of the spherical and rod-shaped ZnO-NPs, respectively.
The maximum increase in the permeability of other gases was also observed at 0.5 wt. % NPs loading into the polymer matrix. However, the peaks in the case of these gases were relatively modest ( Figure 6).
The gas permeability data correlate well with the changes in the specific surface free energy of the membranes ( Figure 5) in the context of the theory described above. At that, the maximum permeability of the gases coincided with the minimum of the polar component for the surface energy of the membranes. I.e., a lower value of the polar energy provided less resistance to the interaction between the gases and composite media. As noted earlier, the degree of such influence is higher for the gases prone to site-specific interactions, such as CO 2 and CH 4 .
However, the changes in the membrane densities should also be taken into account ( Table 4). The density of the membranes with the loading rod-like ZnO NPs into the PU matrix exhibited greater degradation than in the case of the spherical ones. This fact could been a reason of the higher CO 2 and CH 4 permeability values observed with the loading rod-like ZnO NPs. The opposite tendency for the O 2 and N 2 permeabilities could been observed due to the larger molecule kinetic diameter of these gases (Table 5) compared to CO 2 and the lower propensity to site-specific interactions compared to CH 4 .
As a result of the different degrees of the changes in the permeability with the ZnO NPs loading into the PU matrix, the membranes' selectivity increased after modification (Figure 7). For both forms of the NPs, the growth peak occurred at 0.5 wt. % NPs loading. It is noteworthy that, the largest increase was observed for CO 2 /CH 4 in the case of the spherical ZnO-NPs and amounted to 8%. In the case of the rod-like ZnO-NPs, the CO 2 /CH 4 selectivity remained virtually unchanged. However, the CO 2 /N 2 and CO 2 /O 2 selectivities of the PU/ZnO-NPs(rod) membranes increased by 36%, as well as ones increased by 10% and 19%, respectively, in the case of the spherical NPs. noted earlier, the degree of such influence is higher for the gases prone to site-specific interactions, such as CO2 and CH4. However, the changes in the membrane densities should also be taken into account ( Table 4). The density of the membranes with the loading rod-like ZnO NPs into the PU matrix exhibited greater degradation than in the case of the spherical ones. This fact could been a reason of the higher CO2 and CH4 permeability values observed with the loading rod-like ZnO NPs. The opposite tendency for the O2 and N2 permeabilities could been observed due to the larger molecule kinetic diameter of these gases (Table 5) compared to CO2 and the lower propensity to site-specific interactions compared to CH4.
As a result of the different degrees of the changes in the permeability with the ZnO NPs loading into the PU matrix, the membranes' selectivity increased after modification (Figure 7). For both forms of the NPs, the growth peak occurred at 0.5 wt. % NPs loading. It is noteworthy that, the largest increase was observed for CO2/CH4 in the case of the spherical ZnO-NPs and amounted to 8%. In the case of the rod-like ZnO-NPs, the CO2/CH4 selectivity remained virtually unchanged. However, the CO2/N2 and CO2/O2 selectivities of the PU/ZnO-NPs(rod) membranes increased by 36%, as well as ones increased by 10% and 19%, respectively, in the case of the spherical NPs.

Mechanical Properties of Membranes
Mechanical tests showed that loading the ZnO-NPs into the PU matrix improved tensile strength and elongation at break of the MMMs with a further decrease in these values (Figure 8).

Mechanical Properties of Membranes
Mechanical tests showed that loading the ZnO-NPs into the PU matrix improved tensile strength and elongation at break of the MMMs with a further decrease in these values ( Figure 8).

Mechanical Properties of Membranes
Mechanical tests showed that loading the ZnO-NPs into the PU matrix improved tensile strength and elongation at break of the MMMs with a further decrease in these values ( Figure 8). The greatest degradation of the mechanical characteristics for the polymeric membranes was observed with loading the rod-shaped ZnO-NPs. Probably, that was due to The greatest degradation of the mechanical characteristics for the polymeric membranes was observed with loading the rod-shaped ZnO-NPs. Probably, that was due to the lower resistance of these NPs to agglomeration ( Figure 3) and their elongated shape, which might prevent compact embedding the NPs into the PU matrix. Nevertheless, at an optimal (according to the gas transport tests) ZnO-NPs loading of 0.5 wt. %, the tensile strength of the polymer membranes exceeded one for pure PU.

Conclusions
In this work, the effect of the morphology of the ZnO-NPs and their content in the polymeric matrix on the physicochemical, mechanical, and gas transport properties of the PU MMMs with respect to CO 2 recovery from CH 4 , O 2 , and N 2 was studied.
The MMMs based on PU and the spherical and rod-shaped ZnO-NPs with various NPs loadings, namely, 0.05, 0.1, 0.5, 1, and 2 wt. % relative to the weight of PU, were prepared with membrane density control and studied using AFM, wettability measurements, surface free energy calculation, gas separation and mechanical testing. To evaluate the resistance of the ZnO-NPs to agglomeration in the polymer solutions, zeta potential was determined. Each sample demonstrated a high degree of stability.
It was shown that loading the ZnO-NPs into the PU matrix improved the CO 2 permeability of the polymer membranes with a further decrease below the permeability of pure PU. The maximum growth corresponded to 0.5 wt. % NPs loading. The peaks of the CO 2 permeability for the MMMs were 30% and 38% in the case of the spherical and rodshaped ZnO-NPs, respectively. The permeability peaks of other gases were also observed at 0.5 wt. % NPs loading into the polymer matrix but were relatively modest. The selectivity of the membranes increased too with loading the ZnO-NPs into the polymer matrix. The growth peaks corresponded to 0.5 wt. % NPs loading. It is noteworthy that, the largest increase was observed for CO 2 /CH 4 in the case of the spherical ZnO-NPs and amounted to 8%. In the case of the rod-like ZnO-NPs, the CO 2 /CH 4 selectivity remained virtually unchanged. However, the CO 2 /O 2 and CO 2 /N 2 selectivities of the PU/ZnO-NPs(rod) membranes increased by 36%, as well as ones increased by 19% and 10%, respectively, in the case of the spherical NPs.
According to mechanical testing, it was shown that loading the ZnO-NPs to the PU matrix improved tensile strength and elongation at break of the MMMs with a further decrease in these values. The greatest degradation of the mechanical properties for the MMMs was observed in the case of loading the rod-shaped ZnO-NPs. Nevertheless, at an optimal (based on the gas separation data) 0.5 wt. % NPs loading, the tensile strength of the polymer membranes exceeded one for pure PU.
To sum up, it was established that use of spherical ZnO-NPs is promising for the improvement of the gas performance of PU-based MMMs for CO 2 recovery from natural gas, while the rod-shaped NPs better demonstrate their potential in capturing CO 2 in flue gases.