Strategies to stabilize silver salt in composite Pebax2533©/Ag(NH3)2OH and Pebax©2533 [Ag(15-crown-5-ether)] membranes for enhanced ethylene/ethane separation

ABSTRACT Facilitated transport membranes containing silver ions have attracted much attention thanks to their high performance in olefin/paraffin separation. However, the poor long-term stability of such membranes remains a drawback that must be overcome to enable their practical use. In this work, silver ions were incorporated as ion complexes in the Pebax®2533 matrix to increase their stability as a function of time. Two complexing agents were evaluated: ammonia and 15-crown-5-ether. Membranes of Pebax®2533/Ag(NH3)2OH and Pebax®2533/[Ag(15-crown-5-ether)]BF4 maintained stable selectivity for more than thirty days of observation without compromising permeability. Although complexation decreases the performance of silver compared to free ions, it significantly increases the long-term performance in ethylene/ethane separation. the complexing agents weaken the intensity and strength of Ag+/ethylene-π complexation, but prevent the reduction of Ag(I) to Ag(0). In long-term single gas permeation tests, the Pebax®2533/Ag(NH3)2OH membrane showed selectivity in the range of 12.3 to 9 for 39 days, while the Pebax®2533/[Ag(15crown-5-ether)]BF4 membrane showed selectivity around 9 for 30 days. Mixed gas permeation studies allowed to identify conditions to maintain the separation performance above the upper bound for more than 40 days, under continuous membranes exposure to mixed gases. Increased stability compared to membranes reported in the literature was achieved.


Introduction
Ethylene/ethane separation is one of the most important processes in the petrochemical industry. [1][2][3] Ethylene is largely used as starting material for the synthesis of several chemicals and to produce polyethylene. Separation of light olefins is usually performed by cryogenic distillation processes that are energy-intensive and expensive. [4] Other technologies have been investigated to overcome this issue. Membrane separation processes are among those that have attracted much attention thanks to the fact that they require lower energy input. [3,5,6] Ethylene and ethane are compounds with very similar size and condensability; therefore, their separation by membrane-based mechanisms is very challenging. [7][8][9][10][11][12][13][14][15] Several types of membranes were tested for this separation, such as poly(phenylene oxide) (PPO) and its copolymers, [16] 6FDA-based polyimides, [17] carbon molecular sieves, [12,16,17] gas-liquid membrane contactor, [18,19] and protic ionic liquid composite membranes. [20] The performance of conventional membranes in terms of selectivity for C 2 H 4 /C 2 H 6 separation is not as high as that achieved for other gas pairs like CO 2 /CH 4 or CO 2 /N 2 . [4] Transition metal ions interact reversibly with the double bond of olefins. [20,21] The introduction of metal ions into polymeric membranes can therefore increase considerably the separation performance. These membranes belong to the class of the "facilitated transport" membranes where an agent acts as a carrier for one of the two components to be separated. [22,23] Good performance was obtained by addition of AgBF 4 to Pebax®2533 in the weight ratio polymer/AgBF 4 (1:4) [24] but these membranes exhibited low stability over time. The metal/olefin interaction, investigated and described by Dewar et al., [1,24] is commonly known as "π-bond complexation" (Fig. S1). According to this model, the metal/olefin interaction takes place by the formation of a σ-bond between the vacant outermost s atomic orbital of the metal with the full π molecular orbital of the olefin. A second π bond occurs by the back-donation of electrons from the full outer 4d atomic orbital shell of the metal to the vacant π (antibonding) molecular orbital of the olefin. The ability of the metal ions to act as carrier for the olefin depends on the intensity of the πcomplexation between the metals and olefins. The electronegativity of the ion is the parameter that most influences the intensity and the strength of the πcomplexation. Copper (I) and especially silver (I) are the most used metal ions as carrier.
In the preparation of facilitated transport membranes, an important aspect is the compatibility between the polymer and the salt carrier. Pebax®2533, a commercial rubbery polyamide-12/poly-(tetramethyleneoxide) block copolymer, showed high compatibility with AgBF 4 . [22,[25][26][27][28] The various nonbonding electrons of oxygen atoms present in the structure of Pebax®2533 are able to coordinate the silver ions, enabling a high solubility of the salt in the polymer matrix. [1] However, the main drawback regarding the use of silver and copper salts, is represented by their instability over time. In particular, silver ions that are more light-sensitive, undergo reduction to silver nanoparticles, resulting in aggregated particles. [29] Several studies have been conducted to better understand the reduction mechanism in order to improve the stability of silver ions. [30,31] Addition of a small amount of a phthalate plasticizer in PVP/AgBF 4 membranes [32][33][34] an acid such as HBF 4 in POZ/AgBF 4 membranes, [35] a nonionic surfactant like n-octyl-β-D-glucopyranoside (8G1) in PVP/AgBF 4 membranes, [36] or salts blending [37] improved membrane stability. The addition of Al(NO 3 ) 3 in POZ/AgBF 4 and PEO/AgBF 4 membranes give stable separation performance for 14 days,- [38,39] while in Pebax 1657®/AgBF 4 membranes, longterm stability was observed for more than 100 h. [40] Silver reduction can be suppressed by the coordination of Ag + via complex formation or in the presence of ionic liquids. [41,42] Complexation of Ag + with NH 3 , reported by Kan et al., [43] inhibits the reducing effect of polyvinylpyrrolidone. Accordingly, the silver ions in the Ag(NH 3 ) 2 OH complex, embedded in the PVP matrix coordination with NH 3 groups, showed high stability. [43] In addition, crown ethers can bind with metal ions and form strong complexes. The 15-crown-5-ether has a cavity with the diameter in the range of 170-220 pm, while the silver ions have an ionic diameter of 253 pm. However, the 15crown-5-ether is able to form a strong complex with the silver ion, immobilizing it. [44] Even though research activity in developing membranes for olefins separation based on facilitated transport mediated by metal ions has been quite intensive, composite membranes' behavior in long time operation has not been fully clarified yet.
Therefore, in the present work, a systematic study of Ag + complexation with NH 3 or 15-crown-5-ether embedded in Pebax®2533 composite membranes was carried out with the aim to elucidate their performance and stability in long-term ethylene/ethane separation. Membranes' morphology and chemical properties were evaluated by SEM and elemental analysis, respectively. Permeation measurements were performed using single and mixed gases. Membrane performance as a function of time was evaluated also by continuously exposing the membrane to ethylene/ethane mixture. Membranes were monitored for some 75 days. Increased membranes performance stability was obtained.
Nitrogen, oxygen, methane, helium, hydrogen, and carbon dioxide (purity 99.99+ %), used for the permeation tests were purchased from Pirossigeno (Italy), while ethane, ethylene, and their mixtures with a certified composition were purchased from SIAD and SAPIO (Italy).

Dense membranes
A dense Pebax®2533 membrane was prepared by solution casting from an 8 wt% Pebax®2533 solution in ethanol, prepared at 78°C, followed by slow solvent evaporation at room temperature and ambient pressure, as previously described in Bernardo et al. [45] Thin-film composite (TFC) membranes Pebax®2533 was dissolved at a concentration of 2 wt% in ethanol by heating the polymer pellets at 80-85°C under reflux and magnetically stirring for at least 2 h.
The ammonia silver complex was prepared according to the procedure described by Kan et al. [43] (Scheme 1): 0.160 g of AgNO 3 was dissolved in 0.160 g of water and subsequently 0.102 g of aqueous NH 3 solution (32 wt%) was added. After 1 min, additional 0.204 g of aqueous NH 3 solution (32 wt%) was added, obtaining the Ag(NH 3 ) 2 OH complex. The solution of the Ag(NH 3 ) 2 OH complex was added dropwise to 2 g of the Pebax®2533 solution under stirring.
The final solutions were coated on a porous support by an Elcometer film applicator with a casting gap of 100 microns. Polysulfone (50 kDa) and polyethersulfone (20 kDa) were used as support for preparing [Pebax®2533/(Ag(NH 3 ) 2 OH)] and Pebax®2533/[Ag(15crown-5-ether)]BF 4 membranes, respectively. Membranes were prepared by solvent evaporation at room temperature for 24 h in the dark. Each support was coated twice to guarantee a defect-free membrane. It should be noted that two different porous supports were used only because of the availability of materials. Since they serve only as mechanical supports for the TFCs, there is no reason to believe that the porous support significantly affects the stability or selectivity of the thin film. Therefore, the use of one substrate or the other should not affect the main conclusions.

Membrane morphology
Membrane morphology was studied by scanning electron microscopy (SEM), using a Phenom Pro X desktop SEM (Phenom World BV), equipped with backscattering detector. Analyses were carried out at an acceleration voltage of 10 kV for the imaging and 15 kV for the elemental analysis, without samples sputter coating to highlight the presence of the heavier Ag atoms.

Single gas permeability measurements
Permeation experiments of pure gases were carried out at 35°C and feed pressure of 1 bar in a fixed volume/pressure increase apparatus. [46,47] Before each experiment, the membrane sample was carefully evacuated (≤10 −2 mbar) to remove previously dissolved gas species. For a better understanding of their properties, membranes were tested for a set of permanent gases (e.g., He, N 2 , CH 4 , CO 2 ), besides ethylene and ethane.
The permeance, defined as the ratio of the permeability (P) and the membrane thickness (l), was calculated basing on Eq. (1), from the slope of the pressuretime curve in the linear trait, where the permeate pressure is negligible compared to the feed pressure, and the permeate pressure increases linearly [48] : where p t is the permeate pressure at time t, p 0 the starting pressure, (dp/dt) 0 is the baseline slope, R the universal gas constant, T is the absolute temperature, A is the exposed membrane area, V P is the permeate volume, V m is the molar volume of a gas at conditions of 0°C and 1 atm, and p f is the feed pressure. The ideal selectivity, given by the ratio of permeance of two gases, was determined for each composite sample as a measure of the membrane integrity. For the thick isotropic reference membrane based on the neat Pebax®2533, measurements were carried out in the time lag mode, [47] allowing for the determination of the permeability coefficients and the diffusion coefficient (D). The solubility coefficient, S, for the gas in the polymer matrix was evaluated indirectly, assuming the validity of the solution-diffusion permeation model: The gas permeability is expressed in Barrer (1 Barrer = 10 −10 cm 3 cm cm −2 s −1 cmHg −1 = 3.35 × 10 −16 mol m m −2 s −1 Pa −1 ). The ideal selectivity for a pair of gases, A and B, was calculated as the ratio of the individual single gas permeability. It can be decoupled into solubility-selectivity and diffusivity-selectivity: The thickness of dense films was measured using a digital micrometer (Mitutoyo, model IP65).
In the case of the composite membranes, owing to the thin selective layers present on the top of porous support, the time-lag of a fixed permeating species is too short to be measured. Therefore, for composite membranes only the linear portion with a constant slope of the permeate pressure vs. time curve was used to evaluate the gas permeance. The permeation tests on TFC membranes were performed using an area of 2.14 cm 2 , while 11.3 cm 2 was used for the dense Pebax®2533 reference membrane.

Mixed-gas permeability measurements
Mixed gas permeation experiments using different C 2 H 4 :C 2 H 6 compositions (25, 50, 75, mol% C 2 H 4 ) were carried out to measure the mass transport properties, permeance and selectivity. All measurements were performed in steady state at 35°C under a feed pressure in the range of 3-6 bar and at atmospheric permeate pressure. The concentration gradient method was used to carry out the experiments, forcing a part of the feed stream to permeate the membrane under a pressure gradient and measuring both the permeate and retentate flow rates. Membrane separation properties were periodically monitored to evaluate their trend as a function of time. After each test, the membrane sample was left inside the module under exposure of a C 2 H 4 :C 2 H 6 = 50:50 molar mixture at 35°C and 4.7 bar on both membrane sides. The experimental apparatus used for carrying out the gas permeation experiments is schematically shown in Fig. 1.
The core of the system was the membrane module placed inside the furnace. The flat sheet membrane was placed in a stainless-steel module with two inlets/exits: the feed and retentate on one side and the permeate and sweep on the other side. In this work, no sweep gas was used and the sweep and permeate connections were joined together as a single exit.
The mixed gas permeation measurements were performed on samples with a membrane area of 3.8 cm 2 . The pre-mixed gases were fed to the membrane by means of mass flow controllers. Transmembrane pressure was regulated by a back-pressure controller and a manometer on the retentate line. The presence of a manometer also on the feed line permitted the measurement of eventual pressure drops. The retentate and permeate flow rates were measured by means of two soap bubble flow meters and their composition was analyzed with a micro-gas-chromatograph (Agilent 3000). Results were reported in terms of Permeance (expressed in GPU, 1 GPU = 10 −6 cm 3 (STP)/(cm 2 s cmHg)), calculated as the ratio of the permeating flux and driving force: The mixed gas selectivity for a pair of gases, A and B, was calculated as the ratio of their permeances measured in mixed gases experiments: Figure 1. Experimental setup for the mixed-gas permeation measurements. Adapted from Cersosimo et al. [49] with permission from Elsevier.
To assure negligible partial pressure profiles along the membrane module length, a stage cut lower than 2% was maintained during the experiments.

Membrane morphology and chemical properties
SEM images of the composite membranes are illustrated in Fig. 2, which also reports a comparison with neat Pebax®2533 membrane (Fig. 2a,d). Thin-film composite Pebax®2533 membrane containing Ag(NH 3 ) 2 OH (Fig. 2b,e) or [Ag(15-crown-5-ether)]BF 4 (Fig. 2c,f) showed uniform distribution of silver complexes on the membrane surface; some larger cluster was observed in the latter case (Fig. 2c). The silver appears as brighter points on the upper surface region due to stronger back-scattering by the heavier atoms and it is confirmed by the Ag + peaks in the EDX spectrum (Fig. 2g-i). In the case of the Pebax®2533 [Ag(15-crown-5-ether)]BF 4 membrane, the silver complex visibly deposits also in the bulk of the membrane because of leaching of the solution through the porous skin during coating. The bright spots (Fig. 2f) suggest that [Ag(15-crown-5-ether)]BF 4 crystallizes upon evaporation of the solvent, whereas in the other membrane, the Ag(NH 3 ) 2 OH is not visible in the porous support (Fig. 2e), probably because of the narrower pore size of the support used.

Pure gas permeability
Initially, membranes were tested with a series of light gases to determine the effect of the silver complexes on the overall performance. Thin-film composite membranes containing silver complex were compared to pristine Pebax (Table 1). The results confirmed that the silver complexes had a strong impact on the performance of the Pebaxbased membranes, even for inert gases. The silver complexes drastically reduced the gas permeability of the TFC membranes. For instance, the 79 µm thick neat Pebax®2533 membrane had a similar CO 2 permeance as the TFC membranes containing Ag + complex (having thickness around 1 m). On the other hand, C 2 H 4 /C 2 H 6 selectivity of the TFC membranes containing the Ag + complex increased about 10-fold compared to the neat Pebax®2533 film. Furthermore, the selectivity of the TFC membranes for the CO 2 /CH 4 , He/N 2 and He/CH 4 gas pairs also increased compared to that of the neat Pebax®2533 film, which further proved that membranes became more sizeselective when silver complexes were loaded into Pebax®2533. This was likely due to changes of the local chain mobility as an effect of the complexation of polyether chain segments with silver ions. As a result, the selectivity of the doped membranes was very different from that of neat Pebax®2533. Indeed, the addition of silver complexes significantly depressed the permeation of ethane, enhancing the ethylene/ ethane selectivity. Table 1 also shows that the Pebax®2533/[Ag(15crown-5-ether)]BF 4 membrane causes a strong increase in CO 2 /N 2 selectivity compared to the bare Pebax®2533 membrane. Considering that diffusion coefficients of CO 2 and N 2 are generally very similar in all types of polymers, [50] it is expected that changes in gas solubility of CO 2 and N 2 play the most important role in increasing the selectivity. However, further investigation is needed to reach an unambiguous explanation of this phenomenon.
The initial selectivity of the membranes with complexed Ag + was somewhat lower than the value reported in the literature for the equivalent salt without complexing agent Pebax®/AgBF 4 (1:4). [24] As reported by Merkel et al., the selectivity and permeability for ethylene/ ethane in the Pebax®2533/AgBF 4 membranes depend on the amount of AgBF 4 incorporated in the polymer matrix and the best performance was obtained with the polymer/Ag + weight ratio (1:4). [24] We did not study non-complexed Pebax®2533, because Merkel et al. showed a rapid decline in selectivity with time. Instead, we studied Pebax®1657, which is stiffer than Pebax®2533 and is therefore expected to have some additional advantage in the ethylene/ethane separation in terms of size-selectivity. Nevertheless, since the polymer is only 20 wt% of the final membrane and 80% is the silver complex, the role of the polymer is expected to be of secondary importance.
In the present work, the complexation of the same polymer/silver salt changed the electronic density and the availability of the orbitals of Ag + , owing to the πcomplexation and, consequently, a different silver-ethylene interaction was established. Therefore, as for the results obtained with the two different complexes, a decrease of the selectivity was observed (α≈12 for Ag(NH 3 ) 2 OH and α ≈ 9 for Ag(15-crown-5-ether)), compared to the membranes with uncomplexed Ag+, which presented an initial selectivity of 34. [24] The Pebax®2533/Ag(NH 3 ) 2 OH (1:4) membrane was evaluated over a period of 39 days, and between tests it was kept in the dark (Table S1). The separation performance for ethylene/ethane showed a higher stability during the investigated period compared to Pebax®1657 with free Ag + . In fact, the selectivity decreased from 12.3to 9.0 (Fig. 3), while in the same period the selectivity of Pebax®1657 membrane with free Ag + decreased from 32 to about 1. This confirms that the silver ions are highly stabilized by complexation with ammonia and that the complex is a very effective carrier for ethylene. Interestingly, the ethylene permeance increased from 0.88 to 4.79 GPU in 33 days (Fig. 4a). This latter behavior would suggest that in this period there is an increase in the number of active silver ions that act as carrier for the facilitated transport of ethylene.
For the Pebax®2533/[Ag(15-crown-5-ether)]BF 4 membrane, the stability of the separation performance was evaluated for 56 days (Table S2). In this period, the ethylene/ethane selectivity remained constant around the initial value (α≈ 9) for 30 days (Fig. 3), which highlighted the stabilizing effect of the 15-crown-5-ether toward the silver ions. After this time, the selectivity of the membrane decreased more rapidly, reaching a value of less than 2 after 56 days (data point not shown).
For the Pebax®2533/[Ag(15 crown-5-ether)]BF 4 membrane, an increase of ethylene permeance from 3.04 to 9.07 GPU was observed in the period 1-17 days after the casting (Fig. 4b). Between the 30th and 40th day the ethylene permeance halved, due to the reduction of silver ions. However, in the last 10 days an increase of permeance for ethylene and ethane with loss in selectivity was observed. Probably, with the reduction of silver ions to nanoparticles the cohesive energy density of the polymeric chains in the polymer matrix decreases, which instead increases significantly in the presence of the Ag +. [51] Consequently, the membrane becomes more permeable and less selective. The Pebax®1657/AgBF 4 (1:4) membrane has a high initial selectivity α ≈ 32 (Fig. 3), but this value falls quickly to α ≈ 5 after 20 days, while the neat Pebax®2533 membrane has a very low stable selectivity α ≈ 1.18. Hence, the two prepared membranes are stable over at least thirty days.
Upon aging, also the permeance and selectivity of the inert light gases approach the values of the neat polymer membrane (Fig. S2). This indicates that the Ag + /polymer complex is lost due to reduction of the Ag + to metallic silver, with the strongest time-dependence for the Pebax® 2533/Ag(NH 3 ) 2 OH (1:4) membrane. The selectivities of the Pebax®2533/[Ag(15crown-5ether)]BF 4 membrane are lower, but they remain more stable over time. Although the precise mechanism is unknown, the significantly higher He/CH 4 selectivity over the entire time span of the experiment must be due to increased size-selectivity, since both gases are fully inert and nonpolar. It cannot be excluded that this is partly due to a non-negligible contribution of helium permeation through the thin PES and PSf skin of the porous support membranes, but it must be stressed that none of the above effects can be attributed to physical aging or to the low membrane thickness, because physical aging is characteristic only for glassy polymers and not for elastomers. Likewise, the permeability and the diffusion coefficient of Pebax® 2533 were found to be substantially thickness-independent in four membranes ranging from ca. 40 to 230 µm, [52] while for the somewhat stiffer Pebax® 1657, even thin-film composite membranes with an effective thickness on PAN hollow fiber supports of approximately 5 µm have the same selectivity as thick dense films of the same polymer. [53] Mixed gas permeation A different piece of the same sample of Pebax 2533/Ag(NH 3 ) 2 OH (1:4) membrane characterized with single gases was also tested (32 days after preparation) with three different gas mixtures. At this time, the membrane still showed good separation performance, well above the Robeson upper bound (Fig. 5); the C 2 H 4 permeance decreased with increasing molar fraction in the feed, in contrast to the C 2 H 6 permeance, which showed a slight increase (Fig. 6). Both permeances are higher than those measured in the single gas tests, which is most likely due to the different measurement conditions, with a vacuum on the permeate in the time lag setup and atmospheric permeate pressure in the mixed gas setup, or to the different sample conditions due to the thorough evacuation with consequent dehydration of the sample in the time lag setup before measurement. The mixture selectivity tended to decrease with the increase of C 2 H 4 fraction in the feed and was always higher than the ideal selectivity measured with single gases in the time lag setup. In contrast to the pure gas permeation tests, where all gases (CO 2 , CH 4 , N 2 , He, C 2 H 6 , C 2 H 4 ) were measured sequentially and for which continuous exposure is not feasible, during mixed gas permeation measurements, the membranes were constantly in contact with the gas mixture. In particular, during the entire measurement cycle, between subsequent measurements, the  [45] assuming the theoretical membrane thickness, calculated on the basis of the casting thickness and solids content in solution.  membrane was left inside the membrane module, where it was constantly exposed to a C 2 H 4 :C 2 H 6 = 50:50 mixture at 35°C and 4.7 bar on both membrane sides. Separation properties were monitored regularly (until 75 days after membrane preparation), by feeding the same mixture under a ΔΔP of 3.7 bar. The C 2 H 4 permeance decreased with time, in contrast to the C 2 H 6 permeance, which tended to increase (Fig. 7). Consequently, the selectivity decreased during the whole measurement period, going from 10.7 to 2.4. This can be attributed to the decrease in facilitated transport due to reduction of silver ions; however, it must be highlighted that the membrane maintained the separation performance above the upper bound [45] for more than 40 days under continuous exposure to mixed gases (Fig. 5).

Conclusions
Facilitated transport membranes based on Pebax®2533 with complexation of silver ions by ammonia and 15crown-5-ether were prepared and tested for ethylene/ ethane separation. Single gas experiments showed that the thin-film composite membranes exhibited more stable performance compared to literature data. Silver complexation with ammonia yielded membranes with selectivity (α) in the range of 12-9 over 39 days of measurement, and silver complexation by 15-crown-5-ether yielded membranes with a selectivity of 9 over 30 days of measurement.
Ethylene permeance initially increased over time for both membranes: from 0.88 GPU to 4.79 GPU in 33 days for the Pebax®2533/Ag(NH 3 ) 2 OH and from 3.04 GPU to 9.07 GPU in 17 days for the Pebax®2533/[Ag(15-crown -5-ether)]BF 4 . The increase in permeance implies that an increasing amount of Ag + is available to act as carrier over time. Although complexation reduces the effectiveness of silver ions in π-complexation and lowers the membrane selectivity compared to Pebax® 2533-based membranes containing free Ag + , it can keep the silver ions more stable. This approach allowed to obtain facilitated transport membranes with longer lifetime than other methods for stabilizing the Ag + . [16][17][18][19][20][21]54] The stabilization of the silver ions is attributed to the coordination interaction with the lone pairs of the oxygen of 15-crown-5-ether and the nitrogen of NH3 that is established in the complexes' formation. Pebax®2533-based membranes, loaded with silverammonia complex, were tested with mixed gases for 43 days (from 32nd to 75th after their preparation) and kept the separation performance above the upper bound for more than 40 days under continuous exposure to mixed gases. The SEM images have shown better silver distribution in the Ag(NH 3 )2OH complex and this could be the reason for the better gas transport performance of Pebax®2533/Ag(NH 3 )2OH compared to Pebax®2533/[Ag(15-crown-5-ether)]BF 4 membrane. Future work should focus on avoiding leaching of Ag(NH 3 )2OH into the porous support membrane to further enhance the Ag + concentration in the selective dense film, and thus increase both the permeability and the stability. Systematic studies on the effect H 2 in the feed mixture, which could be responsible for reduction of the Ag + to metallic Ag, or of trace amounts H 2 S, which tend to react irreversibly with the Ag + complex, could further increase our fundamental understanding on the stabilization of the Ag + by complex formation.

Acknowledgements
SABIC is acknowledged for providing financial support in frame of a collaborative project. Phenom-World B.V., Eindhoven (NL), is gratefully acknowledged for providing a Phenom Pro X desktop SEM for evaluation. Arkema is gratefully acknowledged for providing a free sample of Pebax®2533.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Statement of Novelty
In this study the complexation of silver ions embedded in a polymeric matrix was investigated to improve silver ion complex stability in facilitated transport membranes for ethylene/ ethane separation. Although the complexation of silver ions with NH 3 and 15crown-5-ether is known in literature, such complexes have not yet been incorporated into polymers for the preparation of membranes.
Mixed matrix membranes of each of the two silver complexes with Pebax®2533 were prepared and their performance in ethylene/ethane separation was studied for long time operation, with both single and mixed gases. Mixed gas permeation studies allowed to identify conditions to maintain the separation performance above the upper bound for more than 40 days, under continuous membranes' exposure to mixed gases.
Increased stability compared to membranes reported in the literature was achieved.