Sustainable Separations of C4‐Hydrocarbons by Using Microporous Materials

Abstract Petrochemical refineries must separate hydrocarbon mixtures on a large scale for the production of fuels and chemicals. Typically, these hydrocarbons are separated by distillation, which is extremely energy intensive. This high energy cost can be mitigated by developing materials that can enable efficient adsorptive separation. In this critical review, the principles of adsorptive separation are outlined, and then the case for C4 separations by using zeolites and metal–organic frameworks (MOFs) is examined. By analyzing both experimental and theoretical studies, the challenges and opportunities in C4 separation are outlined, with a focus on the separation mechanisms and structure–selectivity correlations. Zeolites are commonly used as adsorbents and, in some cases, can separate C4 mixtures well. The pore sizes of eight‐membered‐ring zeolites, for example, are in the order of the kinetic diameters of C4 isomers. Although zeolites have the advantage of a rigid and highly stable structure, this is often difficult to functionalize. MOFs are attractive candidates for hydrocarbon separation because their pores can be tailored to optimize the adsorbate–adsorbent interactions. MOF‐5 and ZIF‐7 show promising results in separating all C4 isomers, but breakthrough experiments under industrial conditions are needed to confirm these results. Moreover, the flexibility of the MOF structures could hamper their application under industrial conditions. Adsorptive separation is a promising viable alternative and it is likely to play an increasingly important role in tomorrow's refineries.


Introduction
Valorization of C 4 hydrocarbons (butane, 1-butene, 2-butene, isobutene, and 1,3-butadiene) is an important research topic because of the market size and versatility of these bulk chemicals. [1] Compared with ethene and propene, the main challenge in C 4 compounds is their upgrading to high-value end products. [2] Large amounts of C 4 hydrocarbons are coproduced with ethylene in steam cracking, as well as alongside gasoline in the fluid catalytic cracking (FCC) process. [3] Moreover,t here is increasingi nteresti nv alorizing the C 4 hydrocarbons that arise from coal liquefaction and biomass refining. This is especially important in China,w here coal is the main carbon resource. [4,5] Typical C 4 streamsf rom FCC, steam cracking, and the methanol-to-olefins (MTO) processes contain mainly 1-butene, 2butene, isobutene, and 1,3-butadiene (see Table 1). The actual ratios vary, due to different compositions of the raw materials. [2] Currently,o nly 1,3-butadiene, isobutene, and 1-butene are on the market as intermediates with standardized product purities. [2] 1,3-Butadiene is mainly used as amonomer in the manufacture of synthetic rubbers and elastomers. [2] It is also used as am onomer fors tyrene-butadiene (S/B)l atex, acrylonitrilebutadiene-styrene (ABS) resins, and high-impact polystyrene (HIPS). [2] Gaseous isobutene is another important petrochemical building block. About 15 milliont ons per year of isobutene are derived from oil and converted into fuels, plastics, and elas-tomers. [6] The demand for n-butenes is also high because of the large markets for alkylate gasoline, detergenta lcohols, synthetic lubricants, and plasticizers. [6] For most of these applications, and especially for polymerization, the purity of the C 4 components is critical. Scheme 1s hows the valorization chain of C 4 hydrocarbons.
In large-scale chemical processes, separation and purification accountsf or much of the costs (both capital expenditure (CapEx)a nd operating expenditure (OpEx)). This is mainly due to the high energy demando ft hese unit operations. In most of these processes, it is the separation units that incur mosto f the environmental burden in terms of CO 2 footprint and energy costs.
Unfortunately,t he physical properties of C 4 isomersa re similar (see Ta ble 2). The boiling points of 1-butene (266.92 K) and isobutene (266.25 K) are practically identical. Separating such compounds by distillation is extremelyc ostly. [2,7] 1,3-Butadiene is usually separated by extractived istillation with acetonitrile and N,N-dimethylformamide. [8] Isobutene is removedu nder mild acid catalysis upon which it forms selectively MTBEo r tert-butanol. [9] The challenging step is the separation of 1butene from the isomers of 2-butene. High-purity 1-butene is crucial in thep roduction of linear low-density polyethylene (LLDP). [10] Isobutene is usually absorbed by using molecular sieves. [11] The isomers of 2-butenea re not furtherseparated because they react analogously in further processing by dehydrogenation, oligomerization, or alkylation. [3] Ap ossible separation could,h owever,l ead to new applications, including the productiono fh igh-performance polymers. [4] Hence, developing Petrochemical refineries must separate hydrocarbon mixtures on al arge scale for the production of fuels and chemicals.T ypically,t hese hydrocarbons are separated by distillation, which is extremelye nergy intensive. This high energy cost can be mitigated by developing materials that can enablee fficient adsorptive separation. In this criticalr eview,t he principles of adsorptive separation are outlined, and then the case for C 4 separations by using zeolites and metal-organic frameworks (MOFs)i se xamined. By analyzing both experimental and theoretical studies, the challenges and opportunities in C 4 separation are outlined, with af ocus on the separation mechanisms and structure-selectivity correlations. Zeolites are commonly used as adsorbents and, in somec ases,c an separate C 4 mix-tures well. The pore sizes of eight-membered-ring zeolites,f or example, are in the order of the kinetic diameters of C 4 isomers. Althoughz eolitesh ave the advantage of ar igid and highly stable structure, this is often difficult to functionalize. MOFs are attractive candidatesf or hydrocarbon separation because their pores can be tailored to optimize the adsorbateadsorbent interactions. MOF-5a nd ZIF-7 show promising results in separating all C 4 isomers, but breakthrough experiments under industrial conditions are needed to confirm these results. Moreover,t he flexibility of the MOF structures could hamper their application under industrial conditions. Adsorptive separation is ap romising viable alternative and it is likely to play an increasingly important role in tomorrow's refineries. Table 1. Composition of C 4 streams obtained from FCC, MTO, and steam cracking. [2,5]  more sustainable processes for C 4 separation can create substantialvalue for these largemarkets. Separation technologies based on adsorption by using microporous materials are alternative energy-efficientp urification methods. [12] Microporous materials can separatec ompounds by using their physicalp roperties, such as kinetic diameter,p olarizability,a cid-base nature,c oordinative properties, permanent dipole moment, and quadrupole moment. [13] This can give advantages in terms of product recovery andp urity,a s well as energy costs. [7,14,15] Herein, we look at the state of the art of C 4 separation by using microporous materials, with recommendations for possible industrial applications
Zeolites are microporous aluminosilicateswith awell-defined crystalline structure. All zeolites have rigid skeletons. They are highly porous anda ct as molecular sieves. [22] Because their pore sizes are in the same order of magnitudea st he sizes of small gas molecules, some zeolitesa re attractive for molecular separation, for example, as packed beds or microporous membranes. [22] To day,t he term zeolite has broadened to include all microporous silica-based solids with crystalline walls. Consequently,i ta lso includes materials in which some of the silicon ions are substituted by other elements. [18] The silicon/aluminum ratios may vary to give hydrophilic and -phobic zeolite structures. This is because aluminum ions induce an overall negative chargei nt he zeolite framework, which is counterbalanced by nonframework cations. These, in turn, may interact with specific adsorbate molecules. The result is ah ighers electivity of aluminum-rich zeolitest owards certain molecules. [23][24][25] MOFs are ar elativelyn ew class of microporous materials. [17] They are built from metal ions (or clusters of metal ions) linked by organic ligands. [16] MOFs have an advantage over zeolites because chemicalm odification of the organic linkersc an pro-vide tailored materials for specific applications.F or example, the length of the organic linker often defines the pore size of ag iven material. [26] Currently, there are about1 0000 experimentally known MOFs (versus < 300 zeolite types). [16] The main disadvantage of MOFs is their low thermals tability( typically stable to 350-400 8C, rarely 500 8C). This rules out high-temperature processes, but MOFs can be used for gas storage, separation, and purification. [27] Unlike zeolites, which are alwaysr igid, MOFs can be flexible, responding dynamically to guest molecules or to external stimuli, such as pressure and temperature. [13] With their exceptionally high porosity and relative simples elf-assembly,M OFs are interesting for both fundamental studies and practical applications. [12] Rigid MOFs have permanent porosity and well-definedp ores or channels, similart oz eolites; [16] this makes them good molecular sieves. [28] Pore size is usually the dominating parameter in separating small molecules. The molecular kinetic diameter is ak ey factor for separation efficiency.F or larger molecules, both the size and nature of the pores (hydrophilic/-phobic, aliphatic/aromatic) are important. [13] Flexible MOFs have an added advantage. For example, in gate-opening transitions, a non-porous materialc an becomea no pen, microporous one. [29] Similarly," breathing" occurs when the pores expand (or contract) reversibly. [30] In both cases, changing the framework structure will change the adsorption capacity. [31] Combining this flexibility with af unctionals urfacec an increaset he selec-Scheme1.Tree view of the industrial applications of C 4 streams. Polymersand polymera pplicationsare highlighted in orange;fuels and fuel additives are highlightedi ng reen. LPG = liquefied petroleumgas;M TBE = methyl-tert-butyl ether;ETBE = ethyl-tert-butyl ether;B HT = butylated hydroxytoluene; BHA = butylated hydroxyanisole;SBR = styreneb utadiener ubber. Table 2. Physicalproperties of C 4 hydrocarbons. [14,18,43] CompoundB.p. [K] Kinetic tivity of gas separation processes. [2] Lewis acidic MOFs have unsaturated metal centers that act as Lewis acid sites. Similar to aluminum-rich zeolites, these MOFs interact with certain sorbate molecules through specific interactions. This makest hem attractive for separating molecules with coordinativea ctive groups. [13] 3

. Adsorptive Separation
Adsorptive separation is as ustainable process that is widely used by the chemicali ndustry. [11] The process implies the separation of am olecular mixture based on differences in adsorption-desorption behavior of the distinct components in the mixture.I ns uch ap rocess, the mixture is first contacted with an adsorbent material under specific conditions to allow the selectiver emoval of one or more components. [32] For gasphase separations, the regenerationo ft he adsorbent is usually achievedb yc hanging the pressure or temperature of the system,s uch processes are known as pressure swing adsorption (PSA) or temperature swing adsorption (TSA), respectively. The main advantageo fs uch processes is that they can be operated at low adsorbent loading because the selectivity between components in the gas phase is greatest in the Henry's law region. For liquid-phases eparations, ad esorbent is required that displaces the adsorbed speciesp referentially from the adsorbent. Economic viability of both gas-and liquidphase adsorptive separations requires adsorbent materials that facilitateh igh separation selectivity,h igh adsorption capacity, and short duration cycles. [32] Adsorptive separation by using porous materials is ac ombination of steric, equilibrium, and kinetic separations. [12] The size of the adsorptivem olecules limits the range of pore and/or window accessibility.G enerally,asmallerp ore results in stronger interaction with the adsorbent. However,i ft he pores are too narrow (relative to the size of the adsorbates), repulsive forces increase and the interaction weakens. [17] Steric separation prevents certainc omponents of am ixture from entering the pores. Such size/shape exclusioni sc ommon in zeolites and rigid MOFs. Here, both the cross-sectional size and shape of the adsorbate affect the selectivea dsorption.T he former is known as the kinetic diameter or collisiond iameter; this is the intermolecular distance of the closest approachf or molecules colliding with zero initial kinetic energy.
If the pores are large enough for all components of am ixture to pass, preferential adsorption can occur.T his is known as the thermodynamic equilibrium effect. [12] The strength of the interaction depends on the surface of the adsorbent and properties of the adsorbate. These are polarizability,m agnetic susceptibility,a cid-base nature, coordinativep roperties, permanent dipole moment, and quadrupole moment. [12,13] Kinetic separation, also known as partial molecular sieve action, is an alternative when equilibrium separation is not feasible. Althought he amounts of different components of am ixture adsorbed at equilibrium are similar, some components may diffuse faster than others.T he different diffusing rates may be used to separatet he components. For kinetic separation, the pore diameter of the adsorbent needs to be between the kinetic diameters of the two molecules to be separated. [12] The adsorption quantity of ac omponent at ag iven temperature is measured by an adsorptioni sotherm. [12] This isotherm relates the amount of substance adsorbed at equilibrium to the pressure of the adsorptivei nt he mixture phase. [12,13] For flexible MOFs, the adsorption isotherms cannot be classified according to the IUPAC schemeb ecauses uch MOFs can undergo structuralc hanges when guest molecules enter. [13] Consequently,t he isotherms showd istinct steps and hysteresis in the adsorption and desorption phases. [33] Adsorption is an exothermic process, whereas desorptioni s endothermic. Thus, the temperature changes within the adsorbent during adsorption/desorption. This temperature is a key variable in determining local adsorption equilibria and ultimatelyg overns the separationp erformance of the material. [34] The isothermic heat of adsorption determines the variation range of the temperature change that takes place during adsorptionp rocesses. High isosteric adsorption heats imply a strong interaction between guest moleculesa nd the host. Therefore, the strength of the interaction needs to be optimized to reach high adsorption capacities. [2] Most model studies on mixture separations rely on singlecomponent isotherms.Y et, in reality,p ore blockinga nd cooperativee ffects between different components play ak ey role. Breakthrough curves of multicomponent mixtures give valuable information about the separation efficiency of am aterial towardsagas mixture. These curvesa re measured by flowing the mixture throughat hermostated bed of adsorbent and monitoringt he effluent (by GC or MS, for example). Ideally,t he adsorbate to be removed shouldb es trongly adsorbed and not be detected in the effluent until saturation. Once saturation is reached, the adsorbentisr egenerated. [13]

SeparatingC 4 Hydrocarbons with Zeolites
The separationo fC 4 hydrocarbons is reported on several zeolite structures. [35,36] Thep ore size of these materials range from approximately 2nmd own to the order of the kinetic diameters of the C 4 isomers. We discuss the performance of various zeolite structures in terms of their pore size. First, we analyze the Faujasites (FAUs;w hichh ave the largest pores), then medium-pore-sized 10-membered-ring (10MR)M FI-typez eolites, and finally 8-membered-ring (8MR) zeolites.
FAUz eolitesc onsist of sodalite cages interconnected in such aw ay that 15 diameter supercages are accessible through 7.4 diameter windows in at etrahedral arrangement (Figure 1a). [25] Their composition is Na x Al x Si 192Àx O 384 (0 x 96). These zeolites exist as high-silicatez eolite Xo ra sh igh-aluminate zeolite Y. [24] The former contains between 77 and 96 aluminum ions per unit cell, whereas the latter has < 77 aluminum ions per unit cell. Because aluminum ions induce the presence of negative charges in the framework (which are then counterbalanced by nonframework metal cations),the hydrophilicity of FAUz eolites increases as the silicon/aluminum ratio decreases. [24,25,38] We refer to zeolite XasM -X and to zeolite Ya sM -Y,i nw hich Md enotes the nonframework metal cat- ions. Although the pores of FAUz eolitesa re much larger than the kinetic diameterso ft he C 4 isomers( see Ta ble2), they can still separatet hese isomers by using differences in the dipole moments or electric polarizabilities.

n-Butane
Harlfinger et al. studied the C 4 single-component adsorption isotherms and integral heats of adsorption on Na-X (see structure in Figure 1a). [38] This zeolite has ac omposition of Na 68 Al 68 Si 124 O 260, with aS i/Al ratio of about 1.8. Figure 1b and c showst he isotherms at 303 Ka nd the heat of adsorption of C 4 hydrocarbons. The adsorption order is cis-2-butene > 1-butene > trans-2-butene > butane. [38] This order reflects differences in dipole moments, electric polarizabilities, and molecular geometries. cis-2-Butene adsorbs preferentially owing to the arrangement of methyl groups to one side of the molecule and to the magnitude and direction of its dipolem oment. 1-Butene has aw eaker interaction with the zeolite than that of cis-2-butene because it has at erminal double bond. Indeed, trans-2-butene shows the weakesti nteraction because it has no dipole moment and its double bond is sterically hindered. [38] The differenceb etween the isothermal behavioro f the isomeric butenes and that of butane is relatedt ot he absence of ad ouble bond in the latter. [37] The same study also determined that the adsorption order of cis-2-butene, 1butene, trans-2-butene, and butane remained unchanged for alkali modifications of the zeolite.H arlfinger et al. also determined that, for alkali modifications of the zeolite,t he adsorption order of cis-2-butene,1 -butene, trans-2-butene, and butaner emained unchanged. [38] It was expected that exchanging Na + with smaller (Li + )o rl arger (K + ,R b + ,a nd Cs + )c ations would result in ac hange in the electric field strength in the interior of the zeolitea nd, hence, in ac hange in heterogeneity of the zeolite surface. Because of the larger charge-to-radius ratio, greater interactions can be expected for Li + than those for Na + .A ccordingly,t hese interactions should become smaller for K + ,R b + ,a nd Cs + .T hese studies demonstrated that cis-2butene, irrespective of the type of cation involved and because of its specific geometry and dipole moment, had the most favorable arrangement in the large cavity of the zeolite, compared with both 1-butene and trans-2-butene. [38] Lamia et al. determined the single-component adsorption isotherms of isobutane and 1-butene on Na-13X. [24,39,40] This zeolite has the compositionN a 88 Al 88 Si 104 O 384 and al ower Si/Al ratio than that of the structure reportedb yH arlfinger et al. [38] Figure2 shows that the extracted saturation capacity for1butene is higher than that of isobutene on Na-13X. This is due to the kinetic diameter of 1-butene( 4.83 ), which is smaller than that of isobutane (5.28 ). Although both molecules can accesst he supercageso ft he zeolite throught he 7.4 diameter windows, the number of isobutane molecules per cage is lower.F igure 2a lso shows the simulated equilibrium adsorption isotherms of isobutene and 1-butene in Na-13X,a sd etermined by Granato et al. [23] We see that the determined set of Lennard-Jones parameters successfully reproduces the equilibrium adsorption properties of 1-butene and isobutene. In principle, the proposed extended force field can be used to predict the adsorption properties of mixtures of C 4 isomerso nz eolite 13X. [24] The differencesb etween the single-component adsorption isothermso ft he C 4 isomers reported by Harlfinger et al. [38] and Lamia et al. [39] are small.

Reviews
Tielens et al. showedt hat the adsorption capacity of Na-Y zeolitesw as even lower than that of Na-X zeolites. [22] They measuredt he heats of adsorption of 1-butene, cis-2-butene, trans-2-butene, and isobutene on Na-Y with aS i/Al ratio of 3.8. Ta ble 3l ists the Henry constants, adsorption enthalpies, and adsorption entropies of the isomers, as well as the adsorption enthalpies reportedb yT ielens et al. [22] The Henry constants decrease in the order isobutene > cis-2-butene > 1-butene > trans-2-butene > butane. This trend was also reported by Harlfinger et al., [38] albeit with lower values due to using az eolite with al ower aluminum content (and thus, fewer interaction sites).
The examples discussed above exclude the 1,3-butadiene isomer because in 1976 Priegnitz patented ap rocess for separating 1,3-butadiene by selective adsorption on az eolite Xa dsorbentc ontaining sodium or potassium as nonframeworkc ations. [41] However,t he adsorption capacity of the zeolite at low diene partial pressures cannots atisfy today's purity requirements. [42] Studies with zeolite Yw ere performed, in which 100 %o ft he Na + nonframework cationsw ere exchanged for transition-metal ions. These ions can participate in both s bondingt oc arbon and p complexation. [43] Thanks to the p bonds, one may achieve high selectivity andh igh capacity for the separation of C 4 -butene isomerso nt ransition-metalmodified zeolites. These bonds are weak enought ob eb roken by simply raising the temperature or decreasing the pressure. [44] This was the first application of p-complexation adsorbents in the separation of C 4 hydrocarbons, and showed that Ag-Y was more selectivef or 1-butene and 1,3-butadiene over butane (see Figure 3a,b).
Studying the influence of the Ag + ion contenti nA g-Y on 1,3-butadiene and 1-butene adsorption shows that the adsorption decreases when the Si/Al ratio increases. [45] This is because fewer Ag + cations are availablei nt he zeolites with ah igher Si content (Figure 3c). The adsorption of 1,3-butandiene on Ag + -Na + mixed ion-exchangedz eolites (AgNa-Y) was also studied. At 1bar (= 10 5 Pa), adsorption of 1,3-butadiene and1butene was still comparable to that of Ag-Y when 70 %o ft he Ag + ions were exchanged with Na + .T he problem is that C 4 streamsc ontain traces of H 2 S, C 2 H 2 ,a nd H 2 ;a ll of which can poison Ag + ions. Therefore, aC u-Y zeolite was used instead of Ag-Y and its stability under H 2 and H 2 Se xposure was studied. [46] The results showed that exposure to H 2 Sa nd H 2 had no effect on the 1,3-butadiene uptake. Furthermore, H 2 exposure had no effect on 1-butene uptake, whereas H 2 Se xposure slightly decreased the 1-buteneu ptake on Cu-Y (becauseH 2 S adsorbs irreversibly on the framework).N otably,i na ll experiments,t he 1,3-butadienea nd 1-butene uptakes on Cu-Y were higher than that on Ag-Y because of the higherp ore volume of Cu-Y.A lthough these resultsa re promising, there are no reports of mixed-gas breakthrough experiments on Cu-Y.
One mesoporous materialt hat gives additional insight into copper-based adsorbentsi sC u-Fe/MCM-41, which contains ferrous/cuprous (Fe 2 + /Cu + )i ons. MCM-41 has separate channel pores with ad iameter of 28 (Figure 4a). [47,48] Single-component adsorption isotherms (Figure 4b)a nd breakthrough experiments (Figure 4c)b oth showed that Fe 2 + ion speciess tabilized the Cu + speciesi nt he framework, increasing 1-butene/ butanes eparation. Indeed, Fe 2 + species may prevent oxidation and reduction of Cu + speciesd uring adsorbentp reparation, the separation process, and in the presence of H 2 . [48] In the Cu-Yf ramework, the Cu + /Cu 2 + ratio was 0.5 and incorporating  [23] (open symbols)and experimental data of Lamiaetal. [24] (closeds ymbols)for 1-butene adsorption isotherms (a) and isobutane adsorption isotherms (b) on zeolite 13X. Reproduced with permission from The American Chemical Society. Table 3. Experimental Henry constants( K'), adsorption enthalpies, and adsorption entropiesoft he butene isomersonz eolite Y. [22] Compound Fe 2 + ions into the zeolite also improved the separation of 1butene from butane. The MFI-type zeolites silicate-1 and ZSM-5 are among the most studied and most widely used zeolites. [49] Figure 5s hows ad rawing of their zigzag channels along the x direction that are intersected by straight channels along the y direction. Both channels are defined by 10MRs. The straight channels are approximately elliptical in shape, with a5 .3 5.6 cross section, whereas the zigzag channels have a5 .1 5.5 cross section. [50] Because the cross sections are in the order of the kinetic diameters of isobutane and isobutene, researchers aim to separatemixtures of these compounds.
Fernandez et al. studied anM FI membrane prepared from silicate-1. [51] This framework is highly hydrophobic ands table up to 400 8Cd ue to the high silicon/aluminum ratio. [50] For  single-component loadings on the membrane at 363 K, the self-diffusion coefficient of butane (D butane )i st hree orders of magnitude larger than that of isobutane (D isobutane ). In particular, at al oading of four molecules per unit cell, the values were D butane = 6 10 À9 m 2 s À1 and D isobutane = 2 10 À12 m 2 s À1 .M atsufuji et al. [52] and Vroon et al. [53] reporteds imilar values for single-gas permeances through MFI membranes. However, Fernandez et al. reported that in an equimolar mixture of butane/isobutane the diffusionc oefficient of butane was two orders of magnitude lower than that in the single-component measurement, whereas isobutane diffuseds lightly faster. [51] Configurational-bias Monte Carlo simulations showed that the butane molecules could be located either along the straight channels or in the zigzag channels of the membrane. Isobutane was located preferentially at the intersections of the straight and zigzag channels of the MFI membrane. Thus,t he intersections provide more space for isobutane and probably serve as traffic junctions. In thee quimolar mixture, the transport of butane along the straight channels in the y direction is halted because isobutaneblocks the intersections.
Caro and co-workers developeda nd patented aZ SM-5 membrane prepared from tetraethylorthosilicate (TEOS) instead of silicate-1. [54] It showed highf luxes for 1-butene but reduced selectivity for 1-butene over isobutene, only slightly compared with membranes prepared from other silica sources. This was attributed to the presence of ethanol in the synthesis batch (originating from TEOS hydrolysis). SEM studies on silicate-1-MFI membranes from synthesis batches with and without ethanol indicated that the crystal size of all MFI membranesw as reducedw ith increasinga lcohol concentration. Smaller crystals have larger intercrystalline grain boundaries, and additional narrow non-zeolite pores may form in the intercrystalline boundaries of the ZSM-5 membranes.T hese pores increaset he 1-butene permeance in mixtures of 1-butene/ isobuteneg ases. [55] Voße tal. reportedp ermeation tests by using an undiluted equimolar mixture of 1-butene/isobutene at 403 Ka nd an MFI membrane prepared from TEOS. [56] Their studies showed that the mixture separation factor decreased from 10 to 5w hen the pressure difference, Dp,a cross the membrane increased from 1t o2 0bar.T his significant Dp is relevant to the practical operational pressure. The pressure of the equimolar undiluted feed was up to 21 bar and the permeateh ad ap ressure of 1bar.T his drop in the separation factor impedes practical applications.T he isobutene flux increases more steeply than that of 1-butene ( Figure 6), which causes al oss of selectivity with increasing pressure. Consequently,t he 1-butenet oi sobutene ratio in the permeate lessens with increasing Dp and the selectivity for 1-butene decreases. Chmelike tal. ran similart ests on butane/isobutane separation over MFI membranes prepared from silicate-1, and reached similarconclusions. [57] All of thesee xamples usedM FI-typez eolites to separate butanef rom isobutane and 1-butene from isobutene.A part from the adsorption equilibrium of pure butane and 1-butene, Wang et al. also studied the separation of their mixtures on ZSM-5 zeolites. [15] Adsorption isotherms were measured for pure and binary mixtures of 1-butene and butane at 300 Ka nd over ap ressure range from 10 À4 to 1bar.T he zeolites used were an all-silicon ZSM-5 and ZSM-5 with Si/Al ratios of 120:1, 50:1, and 20:1, respectively (ion exchange was achieved with ammonium nitrate, setting protons as the nonframework cations). All four ZSM-5 zeolitess electively adsorbed 1-butene over butane.M oreover,t he selectivity for 1-butene increased at lower silicon/aluminum ratios. This can be explained by the  . a) Decrease of the mixture separation factor, a,f or an undiluted equimolar mixture of 1-butene/isobutene through an MFI membrane at 403 K. b) Fluxes of 1-butene (opensymbols) and isobutene( filled symbols) from an equimolar mixture through an MFI membraneat4 03 K. In both cases, the feed pressure was increased upt o21bar,whereas the permeate pressure was constant at 1bar.The threedata points at each Dp were derived from three independent membranep reparation and permeation tests. [56] ChemSusChem 2017, 10,3947 -3963 www.chemsuschem.org 2017 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim presenceo fm ore available sites in zeolites with small silicon/ aluminumratios. [15] Most experiments for C 4 separation use 8MR zeolites. The pore sizes of these zeolitesa re smaller than those of the FAUand MFI-typez eolites, andm atch more closely with the kinetic diameters of C 4 isomers. Here, we discussC 4 isomer separation by using SAPO-17, DD3R,S i-(CHA), ITQ-32, and RUB-41 zeolites. With the exception of SAPO-17, these zeolites have high silicon/aluminum ratios and are therefore hydrophobic. SAPO-17 is made by substituting framework atoms (P,A l) in AlPO 4 -17 for silicon. This leads to the formation of one Brønsted center per unit cell. [58] The 8MR channel system of SAPO-17 has elliptical pore aperturesw iths izes of 3.6 5.1 .R ichter et al. performed single-component adsorption experiments (5 %o fe ach C 4 isomer in H 2 )a nd showed that trans-2-butene was selectively adsorbed on SAPO-17 at low temperatures. [19] In contrast, AlPO 4 -17 has nearly the same adsorption capacity for all three butene isomers. It was concluded that SAPO-17w as different because of the presence of silicon and, hence,the modification of lattice properties associated with the presence of Brønsted acid sites. [59] Discriminating 1-butene and cis-2-butene by SAPO-17 is the result of differentiated electrostatic interactions between the negatively charged anion lattice and butenes with different polarity. trans-2-Butene has no permanent dipole moment, unlike cis-2-butenea nd 1-butene. The orientation of 1-butene and cis-2-butene dipolest owards the electrostatic fielda tt he pore entrance is unfavorablef or entering thep ores, whereas trans-2-butene can enter the pore system easily. [19] Zhu et al., [17] Gücüyener et al., [14] and Jansen and coworkers. [25] ran single-component breakthroughe xperiments, multicomponent experiments, andp erformed molecular modeling studies with 1-butene, cis-2-butene, trans-2-butene, and 1,3-butadiene by using the hydrophilic decadodecasil 3R (DD3R) as adsorbent.T he DD3R structure is stable at high temperatures and formed by three different types of cages. Au nit cell of DD3Rc onsists of six 10 [17,22] The 8MRs are (at least in theory) wide enough for someC 4 isomers. [22] Figure 7a shows the differentc ages and the framework of DD3R. [17] The breakthrough measurements showed that, at temperatures between 303 and 373 Ka t1bar,D D3R is accessible to trans-2-butene and 1,3-butadiene. 1-Butene and cis-2-butenea re excluded from the framework (see Figure 7b,c). [17,18] This was explained through the critical diameter of the adsorptivem olecules. The critical diameters of trans-2-butene (4.31 )a nd 1,3-butadiene (4.31 )a re slightly smallert han the free cross diameter of the 8MR and these molecules can enter into the cavities. However, the critical diameter of cis-2-butene (4.94 )i sl arger than the window size and the critical diameter of 1-butene (4.46 )i s comparable to it. [17] Molecular modeling studies support this theory. [17] When the 8MRs are indeed the smallest passage, the energy barriero fd iffusion is simply the differencebetweenthe energy of component i in the cage, E icage ,a nd in the ring, E iring . The permeabilities of the C 4 isomers calculated from these barriers were in the order of trans-1,3-butadiene > trans-2-butene > cis-1,3-butadiene > cis-2-butene > 1-butene, which was consistent with the order reported by Zhu et al. [17] and Gücüyener et al. [14] Performing both breakthrough ands ingle-component adsorptione xperiments,C asty et al. [60] and Palomino et al. [61] showedt hat all-silica8 MR zeolites Si-(CHA) and ITQ-32 had similar adsorption behavior to that of SAPO-17 and DD3R.P ure silica CHA and ITQ-32 adsorbed trans-2-butene quickly at temperatures of 273 and 298 Ka nd pressures of 2a nd 0.3 bar. These zeolites showed little or no adsorption for cis-2-butene and 1-butene, even after an hour of contactt ime. The Si-(CHA) consists of an 8MR channel system with window sizes of 3.50 4.17 ,w hereas ITQ-32 consists of interconnected 8MR and 12MR channels. [62] The 8MR channels have window sizes of 3.5 4.5 ,w hereas the 12MR channels have ad iameter of 6.3 .S imilar adsorption behavior of Si-(CHA) and ITQ-32 indicates that in ITQ-32 the 8MRw indows are the limiting factor for C 4 diffusion.
All of these zeolites prefer trans-2-butene over 1-butene and cis-2-butene, owing to its smaller kinetic diameter (4.31 vs. 4.94 ). These studies, however,w ere performed in the gas phase. [14] Liquid-phase adsorption studies were reported only for channels (4.0 6.5 ). [14,15] Figure 8s hows the skeletal model of the RUB-41 structure, highlighting the projection along the 8MRc hannels. Wang et al. ran single-component adsorptioni sotherms of isobutane, 1-butene, and trans-2-butene on RUB-41 up to ap ressure of about 0.8 bar. [15] Their results were consistentw ith the previous results on other 8MR zeolites: trans-2-butene adsorbed preferentially to isobutane. Tijsebaert et al. performed single-component adsorption isotherms on trans-2-butene, cis-2-butene, 1-butene,a nd isobutene in cyclohexane. [7] The resultsi ndicated that trans-2butene and cis-2-butene were much more strongly adsorbed than 1-butene or isobutene. RUB-41 zeolite prefers, therefore, both trans-2-butene and cis-2-butene over 1-butene. This contrasts the order observed for other 8MRz eolites. Because the adsorption isotherms for RUB-41 were recorded after sufficiently long times,i tw as concluded that the performance of RUB-41 for 2-butenes over 1-butene was due to thermodynamic rather than kinetic effects. It may be that the 2-butenes are more efficiently packed inside the pores than 1-butene, or that 1-butene might lose more of its conformational entropy in the pores. Breakthrough experiments with two binary butene mixtures were also performed. The first mixture consisted of cis-2butene and 1-butene and the second mixture consisted of trans-2-butene and 1-butene. In both cases, 1-butene eluted first (Figure 9). These results confirmt hat the separation can be performed in the liquid phase using RUB-41.
Li et al. designed MOF-5, the first rigid MOF that showed permanent porosity after being fully desolvated or heatedu p to 573 K. [65] It consists of Zn 4 Ou nits connected by linear 1,4benzenedicarboxylate struts to form ac ubic network. [66] One unit cell of the framework consists of eight ZnO 4 clusters ( Figure 10) and encloses al argec avity with ad iameter of 18.5 .M ertens and co-workers prepared aM OF-5-CSA-coated columnt os tudy its potentiala pplication in the separation of commercial natural gas and butaneg as components. [64] This MOF-5-CSA-coated columnw as compared with ac ommonly used commercialc olumn, Agilent HP PLOT S. Te sts were run by using an aturalg as sample that consisted mainly of C 1 -C 4 alkanes( methane 97.1 %, ethane 1.7 %, propane 0.7 %, isobutene 0.2 %, and butane 0.3 %). [64] The individual components were clearly baseline separated by both columns. However,the MOF-5-CSA-coated column separated all five components more rapidly and without any other performance loss (0.8 min total separation time). The commercial columnr equired 0.13 min longera nd the distribution of peaks was nonuniform.
To demonstrate the separation powero ft he MOF-5-CSA column, additional C 4 components were added to the natural gas samples. [64] The results showedt hat even different butene isomersw ere easily baseline separated, despite their similar vapor pressures and smalla mounts. The minimal separation time for all of the components was < 4min. [64] The MOF-5-CSA columnwas operated over ap eriod of five months, performing more than 300 chromatographic separations (in the range 40-50 8C) without any discernable loss of separation power. [64] Further studies are required to test whether the MOF-5 column maintains its selectivity when used for larger scale separations. ZIF-7 belongs to the zeolite imidazolate frameworks (ZIFs) group of compounds,asubfamily of MOFs named after the resemblance of the metal-imidazolate-metal bond anglesw ith that of the SiÀOÀSi angles of zeolites. [67] This flexible MOF is formed by Zn 2 + metal-ion clusters linked through benzimidazole (BIM). It has six-membered-ring (6MR) pore openings with, in the optimized structure in vacuum, ad iameter of 3 (see Figure 11 for details).
Gascon and co-workersm easured single-component adsorption isothermso fb utane, 1-butene, cis-2-butene, and trans-2butene at 298, 338, and 373 Ko nZ IF-7 (see Figure 12). [33] All isotherms showedt he typical characteristics of af lexible host material, including distinct steps and hysteresis in the adsorption and desorption branches. At 298 Ka nd 1bar,t he MOF showedahighers aturation adsorption capacity for trans-2butene over butanea nd cis-2-butene over 1-butene. [33] At 338 K, ZIF-7 showed 25 %higher saturation adsorption capacity for trans-2-butene and cis-2-butene compared with that of 1butene. [33] At ah igher temperature, about3 73 K, ZIF-7 showeda highers aturation adsorption capacity for cis-2-butene and butaneo ver 1-butene and trans-2-butene. [33] The main conclusion is that the separation of the C 4 isomers on ZIF-7 is temperatured ependent, which enables the separation of the isomers studied.  Nair and co-workers used two types of linkerst of ine-tune the pore size, hydrophilicity,a nd organophilicity of ZIFs. [68] They demonstrated this through adsorption and diffusion measurements of hydrocarbons, alcohols, and water by using mixed-linker ZIF-8 x -90 100Àx materials with al arge range of crystal sizes. Varying the mixed-linker composition parameter (x) allows continuous control of n-butane, isobutane, butanol, and isobutanol diffusivities over two to three orders of magnitude. It also allows control of water and alcohol adsorption, especially at low activities.
Cu 3 (BTC) 2 (BTC = 1,3,5,-benzene-tricarboxylate), also known as HKUST-1, is aL ewis acid MOF that hasb een studied extensively. [69,70] The main structuralf eature of Cu 3 (BTC) 2 is aC u 2 + dimer with aC u 2 + ÀCu 2 + distance of 2.63 .T welve carboxylate oxygen atomsf rom the two BTC ligandsb ind to the four coordination sites of each of the three Cu 2 + ions. In addition to the carboxylate ligands, one water molecule points towards the centero ft he pore andi sc oordinated to the copper center. When the coordinated water molecules are removed in vacuum, accessible Cu 2 + centers are created that can act as Lewis acid sites. These paddle wheel units form af acecentered crystal latticet hat possesses at hree-dimensional channel system with ab imodal pore size distribution. [71] The larger pores are hydrophilic and have ad iameter of about 9 , which define the 12 paddle wheel subunits that form ac uboctahedron. As maller pore system of tetrahedron-shaped side pockets,with ad iameter of about 5 ,a re formedb yf our benzene rings ( Figure 14).
The latter system is accessible from the large pores through windows with ad iameter of about 3.5 . [20] Hartman et al. measured single-component adsorption isotherms of isobutene and isobutane at different temperatures with Cu 3 (BTC) 2 as an adsorbant. [20] Figure 15 as hows the highresolution isobutane and isobutene adsorption isothermsa t 303 K. Figure 15 bd isplays the breakthrough curvesf or the separation at the same temperature. Initially, isobutene and isobutene are completely removed from the feed stream.A fter about 90 min on stream,f irst isobutene breaks through and the partial pressure at the adsorber outlet rises to p isobutane / p isobutene = 1.6. After about 140 min on stream, isobutene breaks through and the adsorber inlet concentration is reached after about 150 min. This behavior is explainedi nt erms of overshootingt hrough the partial displacement of isobutane by isobutene from the adsorption sites in Cu 3 (BTC) 2 .T he adsorption  enthalpy of isobutene is only 5kJmol À1 higher than that of isobutene. [20] This suggests that there are no strong p interactions between isobutenea nd the coppers ite. Instead, van der Waals interactions dominate. [72] Alaerts et al. studied the separation of binary equimolar mixtures of cis-butene and1 -butene, cis-butenea nd trans-butene, and 1-butene and trans-butene on Cu 3 (BTC) 2 in liquid hexane. [21] The separation factors for these mixturesa re 1.8, 1.9, and2 .4, respectively.T hese values indicate the preference of Cu 3 (BTC) 2 for cis-2-buteneo ver 1-butene and trans-2butene. To tal uptakes varied from 12 to 21 wt %. The high preferenceo fC u 3 (BTC) 2 for butenes in hexane reflects strongc ompetition from the aliphatic solvent, with adsorption driven by p interactions with the Lewis acid sites. This is remarkable when compared with the findings of Hartmann et al., [20] who found that the gas-phase separation of isobutane from isobutene was driven by van der Waalsi nteractions. Notably,A laerts et al. ran their experiments in the liquid phase, and therefore, at ah igher pressure. [21] M-MOF-74, also knowna sM -CPO-27 or M 2 (dobdc) (M = transition-metal ion;d obdc = 2,5-dioxido-1,4-benzene dicarboxylate) also has coordinatively unsaturated metal sites. [73,74] Fe-MOF-74 has ah oneycombs tructure with open metal sites towards large pores of about 15 in diameter (Figure16). [75] All Fe 2 + ions within this structure are coordinatively unsaturated, and the distance between them varies from 7t o8 in both laterala nd vertical directions. Such al arge pore volume could accommodate relatively large molecules, such as C 4 alkenes.
Kim et al. reported aD FT study showingt hat the Fe-MOF-74 structure wasapromisingc andidate for 1-butene separation. [76] Binding energy calculations showedt hat 1-butene boundp referentially over isobutene, cis-2-butene, and trans-2butene. [76] Particularly,1 -butene had a1 3-24 kJ mol À1 higher binding energy than those of the other isomers, which indicated that selective adsorption of 1-butene on the MOF should be feasible. Experimental proof is needed because blockingo r cooperative effectso fo ther components in am ixture could    [7] ChemSusChem 2017, 10,3947 -3963 www.chemsuschem.org 2017 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim lead to different results. Theoretical calculations indicated that 1-butene could approach the metal binding sites more closely than the other butene isomers, enabling stronger bondinga nd p-back-bonding interactions between 1-butene and MOF-74. [76] Potential p complexation is significantly hindered sterically for 2-butenes and the adsorption depends largely on van der Waals interactions. [76] 6. Summary and Outlook The number of studies published on C 4 separation by using microporous materials is relativelys mall. Most of these studies focus on single-componentg as adsorption experiments. These experimentsa lone are insufficient for assessing the separation of mixtures of C 4 isomers because blocking and cooperative effects in mixtures may lead to differentb ehavior. [13] Mosto ft he experimentsw ere run in the gas phase, at pressures up to about 1bar,w hereas industry usually uses higher pressures to minimize process costs.Ideally,s eparation should be studied in the liquid phase. There is ac lear need for developing laboratory equipment that will enable in operando adsorption separation studies.
Regarding zeolites, an industrial C 4 separation is unlikely with larger pore FAUz eolitesa nd MFI membranes. In the Na-FAUf rameworks, all isomerscan enter the large cages and specific interactions between the Na + cations and the frameworks are too weak to effectively separatet he isomers. [23,24,37,40] The separation process is mainly related to differences in the dipole moments and electrical polarizabilities of the C 4 components. Al ow silicon/aluminum ratio provides more interaction sites and 1-butene and 2-butene are adsorbed preferentially. The adsorption of these molecules decreases considerably for zeoliteswith ahigh silicon/aluminum ratio.
MFI-type materials have medium size pores and are hydrophobic due to the high silicon/aluminum ratio. They were studied for butane andi sobutane separation, which is ad iffusion-controlled process. However,t he selectivity of MFI-type materials is still too low at practicallyr elevant pressures. This is because the tunneli ntersections of MFI membranes are blockedb yt he larger iso isomer,w hichh alts linear isomers. [57] Conversely,s eparation with 8MR zeolitesm ay be feasible because their pore sizes are in the order of the kinetic diameters of the isomers. These zeolites (except for RUB-41) adsorb trans-2-butene preferably over 1-butene and cis-2-butene, in accordance with the criticald iameter values. [14] Additionally,R UB-41 can separate 1-butene from the 2-butene isomersd ue to thermodynamic effects. [7] That said, the selectivities of the 8MR zeolitesa re still too low to meet high-purity olefin demands, even after many adsorption/desorption cycles.
MOFs are attracting increasing attention because their flexibility and unsaturated metal centers might provide increased selectivity for gas separation. [2] Indeed, ac olumn coated with MOF-5 showed remarkable performance by baseline separating all butene isomers. [62] The results reported for the flexible ZIF-7 framework are also promising. The framework shows temperature-dependent preferences for different C 4 isomers. [32] However,o nly single-component isotherms over al ow-pressure range were reported. Breakthrough experiments at higher practically relevant pressures are neededb ecause slight changes in external stimuli can cause considerable changes in the framework and degree of flexibility.M oreover,t he flexibility of MOFs could be problematic in an industrial application because the framework has to face permanent stress through heating, outgassing, and cooling during adsorption. [31,72] Concerning Lewis acidic MOFs, the liquid-phase separation in aliphatic solvents is an attractive option because in the presence of an aliphatic solventbutenes are preferably adsorbed. [21] An attractive, but much less explored, methodf or C 4 separations is separation through p complexation. The advantage of this method arises from strong interactions being established between adsorbent and adsorbate, which are stronger than those involving only van derW aals interactions. Such strong interactions allow highers electivity anda dsorption capacity to be achieved. Importantly,t he bonds formed by p complexation are stillw eak enough to be broken by changing the parameters of aseparation process, forexample, pressure or temperature. Therefore, separation by p complexation could be a simple and efficient process, particularly by PSA or TSA. Indeed, some studies werer eportedo nA g + -o rC u + -modified zeolites. [60] So far,t he most promising resultsw ere obtained for Cu-Y frameworks, but one must bear in mind that this better performance comesa taprice. [42] One ton of sodium costs about $2000, whereas at on of copper costs about $6000 (2015 prices).T he use of copperi nsteado fs odium would increase the adsorbentc osts. Nevertheless, this might be compensated for by the fact that separations by PSA, especially in the gas phase, are relativelys imple andh ighly efficient in terms of product purity and recovery. This review illustrates that most studies on C 4 separations have mainly focused on zeolites, whereas the potential of MOFs is largely unexplored. We believe that MOFs, with their unique tailoredf unctionalities, hold the key to sustainable C 4 separation processes in the coming decade. We foresee that separation by p complexation with MOFs will offer advantages for adsorptive separations due to the possibility of tuning the interactions established betweena dsorbent and adsorbate. Furthermore, experimental studies should also focus on both low-and high-pressures eparations. Low-pressure separations are enthalpyd riven, but at higherl oadings the separations become entropic in nature,w hen molecules are highly structural ordered. Consequently,e ntropy-driven separations would allow for separations by using high loading, and hence, increaseefficiency.