Postextraction Separation, On-Board Storage, and Catalytic

Review pubs.acs.org/CR

Postextraction Separation, On-Board Storage, and Catalytic Conversion of Methane in Natural Gas: A Review Dipendu Saha,*,† Hippolyte A. Grappe,‡ Amlan Chakraborty,§ and Gerassimos Orkoulas† †

Chemical Engineering Department, Widener University, 1 University Place, Chester, Pennsylvania 19013, United States RMX Technologies, 835 Innovation Drive, Suite 200, Knoxville, Tennessee 37932, United States § Entegris Inc., 10 Forge Park, Franklin, Massachusetts 02038, United States ‡

ABSTRACT: In today’s perspective, natural gas has gained considerable attention, due to its low emission, indigenous availability, and improvement in the extraction technology. Upon extraction, it undergoes several purification protocols including dehydration, sweetening, and inert rejection. Although purification is a commercially established technology, several drawbacks of the current process provide an essential impetus for developing newer separation protocols, most importantly, adsorption and membrane separation. This Review summarizes the needs of natural gas separation, gives an overview of the current technology, and provides a detailed discussion of the progress in research on separation and purification of natural gas including the benefits and drawbacks of each of the processes. The transportation sector is another growing sector of natural gas utilization, and it requires an efficient and safe on-board storage system. Compressed natural gas (CNG) and liquefied natural gas (LNG) are the most common forms in which natural gas can be stored. Adsorbed natural gas (ANG) is an alternate storage system of natural gas, which is advantageous as compared to CNG and LNG in terms of safety and also in terms of temperature and pressure requirements. This Review provides a detailed discussion on ANG along with computation predictions. The catalytic conversion of methane to different useful chemicals including syngas, methanol, formaldehyde, dimethyl ether, heavier hydrocarbons, aromatics, and hydrogen is also reviewed. Finally, direct utilization of methane onto fuel cells is also discussed.

CONTENTS 1. Introduction 2. Environmental Impacts of Using Natural Gas 3. Separation Needs for Natural Gas 3.1. Water Removal (Dehydration) from Natural Gas 3.2. Acid Gas Removal (Sweetening of Natural Gas): Separation of Carbon Dioxide, Hydrogen Sulfide, and Other Sulfur-Bearing Species 3.3. Separation from Inert (N2) 3.4. Separation of Heavier Hydrocarbons (Natural Gas Liquid (NGL) and Liquefied Petroleum Gas (LPG)) 4. Processes Employed for Natural Gas (Methane) Separation and Enrichment 4.1. Classical Techniques 4.1.1. Chemical and Physical Absorption 4.1.2. Cryogenic Fractionation 4.2. Modern Techniques 4.2.1. Adsorption Fundamentals and Computations 4.2.2. Solid Sorbents for Sulfur Removal 4.2.3. Membrane Separation 5. Natural Gas as Transportation Fuel 5.1. High Indigenous Sources and Improvement in Extraction Technology © 2016 American Chemical Society

5.2. Suitability as Fuel in Engines 5.3. Safer Fuel and Low Environmental Pollution 5.4. On-Board Storage Strategy of Natural Gas 5.4.1. CNG and LNG 5.4.2. Adsorbed Natural Gas (ANG) 6. Catalytic Conversion of Methane 6.1. Catalytic Conversion Strategies 6.1.1. The Indirect Method 6.2. Catalytic Conversion to Methanol 6.3. Catalytic Conversion of Other Hydrocarbons and Chemicals 6.4. Catalytic Conversion to Hydrogen 6.5. Fuel Cells 7. Conclusive Remarks Author Information Corresponding Author Notes Biographies Acknowledgments References

11436 11438 11440 11441

11441 11441

11441 11442 11442 11442 11443 11443

11464 11464 11464 11464 11465 11473 11473 11473 11474 11477 11479 11481 11482 11482 11482 11482 11483 11483 11483

11443 11452 11453 11464 Received: February 2, 2016 Published: August 24, 2016

11464 11436

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

trillion cubic feet).1 Although worldwide natural gas consumption has increased over the years, the consumption in the U.S. from 2010 to 2011 rose only by 2.2%.3 Worldwide, there are very large reserves of natural gas, mostly concentrated in northern Russia, Middle East, and North America, including the U.S. (Figure 2). Russia, Iran, and Qatar hold 56% of the world’s reserves, and the members of the Organization of Petroleum Exporting Courtiers (OPEC) control the rest of the world’s reserves. Sources of natural gas can be classified into two categories: (a) conventional and (b) nonconventional. In conventional sources, natural gas is recovered as free gas from either crude oil or easily accessible rock formations, such as carbonate, sandstone, or siltstones. Nonconventional sources of natural gas include (i) shale gas, (ii) tight gas, (iii) coal bed methane (CBM) or coal seam gas (CSM), and (iv) methane hydrate. The term shale gas refers to methane gas reserves trapped inside rock formations. In shale gas formation, methane is either adsorbed in the micropores of rock, into the insoluble moiety of sedimentary rock, such as kerogen, or confined in the narrow fractures of rock. The rock formations that hold the shale gas are characterized by low permeability, and therefore the gas is released very slowly upon exposure. Because of technological advancements, shale gas has become the key nonconventional source of natural gas and serves as a vast reservoir of methane for a few western countries such as the U.S. Usually, shale gas is excavated by the so-called hydraulic or high-pressure fracture technology in which nonconventional horizontal drilling is employed through hundreds of meters for economic recovery of methane. The details of hydraulic fracture and the shale gas recovery process are beyond the scope of this Review and can be found elsewhere.6 Although tight gas is very similar to shale gas, the rock formations have much less permeability than that of shale gas. Typically, the porosity for tight gas rocks is less than 10%, and the permeability is less than 0.1 millidarcy. Hence, costly hydraulic fracturing is required. Coal bed methane (CBM) or coal seam gas (CSM) comprises methane adsorbed in the pores of coal or confined in the crevices between the coal layers and is usually located close to the surface of the earth. CBM is also an important source of natural gas in North America and Australia. CBM is formed by either biogenic or thermogenic processes.7 In the biogenic

1. INTRODUCTION Because of advancements in technology of excavation, remarkably low emissions as compared to other fossil fuels, and vast available reserves, natural gas has gained attention as the key source of conventional energy. Natural gas is primarily methane. It did not gain considerable attention until recently, when the potential of this valuable resource was completely realized followed by the development of associated technology to fully utilize its benefits. Natural gas can be utilized in different areas of residential, commercial, industrial, electrical, and transport sectors. According to the Energy Information and Administration (EIA), it is estimated that industrial, electric power, and residential sectors serve as the largest consumers of natural gas and utilize 32%, 30%, and 21% of overall consumption, respectively.1 The other areas are commercial and transport sectors that consume 14% and 3%, respectively. The current worldwide usage of natural gas as a source of energy is 22%, preceded by oil (30%) and coal (27%)2 (Figure 1). According to the International Energy outlook for 2011, the

Figure 1. World consumption of different kinds of energy.3

projected increase in global utilization of natural gas is 44% from the reference year 2007 to year 2035 (from 108 to 156

Figure 2. Worldwide natural gas reserves. Reprinted with permission from refs 4 and 5. 11437

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

Figure 3. U.S. natural gas production trends.

commonly associated with methane in natural gas are carbon dioxide, nitrogen, moisture, oxygen, carbon monoxide, heavier hydrocarbons, sulfur compounds such as hydrogen sulfide, and others. Thus, it is not possible to assign a single composition to natural gas; rather, a wide range of compositions that can be used for identification purposes. The following table provides an entire gamut of natural gas compositions excavated from different parts of the world.9−13

process, methane is formed during the coalification of organic (plant) matter followed by the thermogenic process in which additional methane is produced at elevated temperatures (50− 150 °C) along with other gases such as CO2, N2, and H2O. Excavation of CBM requires a less stringent protocol in which vertical drilling is performed in the coal bed, and the gas is released when the hydrostatic pressure is reduced. Methane hydrate is a well-known “host−guest” type of compound in which the methane is trapped within the tiny cavities of ice crystals, and the overall growth of the system depends on the thermodynamic equilibrium of the system.8 Methane hydrate is one of the largest sources of methane. For instance, 1 m3 of ideal methane hydrate contains 0.8 m3 of water and 160 m3 of methane at equilibrium temperature and pressure. Natural methane hydrates are formed in large amounts in shallow lithosphere, continental sedimentary rocks in polar regions, and in several oceanic segments where the temperature is below 2 °C. Despite the large reserves of natural gas within methane hydrates, methane is never commercially produced from hydrates, and it is unlikely that it will attract commercial attention in the near future due to difficult technological and economic barriers. Figure 3 shows natural gas production in the U.S. from different sources. Although natural gas consists primarily of methane, it is often associated with many other constituents of compositions that vary from source to source (Table 1). Species that are most

2. ENVIRONMENTAL IMPACTS OF USING NATURAL GAS One of the key attractive features of using natural gas in electricity, heat generation, transportation, or other industrial sectors is the low emission of greenhouse gases (GHGs), such as CO2 and other toxic pollutants. Because of the low C/H ratio, methane in natural gas always generates the lowest CO2 emissions as compared to its counterparts. On an average basis, burning natural gas produces less than one-half CO2 per unit of generated electricity as compared to that of conventional sources.14 In terms of CO2 generation upon burning, it has been demonstrated that natural gas produces 55.9 kg CO2/GJ of energy as compared to lignite (102 CO2/GJ), anthracite (91.3 CO2/GJ), fuel oil (78.5 CO2/GJ), and diesel (73.3 CO2/ GJ). 15 Besides GHG emissions, numerous other toxic substances are also generated when burning fossil fuel for energy generation purposes. The pollutants other than CO2 include total hydrocarbon (THC), carbon monoxide (CO), nitrogen oxides (NOX), sulfur oxides (SOX), particulate matter (PM), polyaromatic hydrocarbons (PAHs and nitro-PAHs), carbonyl compounds and light aromatics (formaldehyde, acetone, benzene, xylene, acetaldehyde, etc.), and mutagens or genotoxic materials. Furthermore, particles smaller than 0.1 μm cause health issues by creating stress on the epithelial cell lining of lung, leading to irritation and inflammation.16,17 According to the World Health Organization (WHO), urban air pollution is the most significant environmental risk factor and a significant contributor to the death toll in Europe. Exposure to particulate matter (PM) is responsible for 100 000 premature deaths per year.18,19 Carbon monoxide (CO) is acutely toxic to humans and other mammals as it binds with hemoglobin irreversibly and lowers its oxygen-carrying capacity, leading to death. All of the PAHs are essentially carcinogenic, and affect areas such as the skin, lung, bladder, liver, and

Table 1. Compositions of Natural Gas components

volume fractions

methane ethane propane N-butane iso-butane N-pentane hexane nitrogen oxygen carbon dioxide carbon monoxide helium hydrogen sulfide

55−98% 2−4% 0.5−2% 0.15−0.3% 0.1−0.2% 0.02−1% 0.01−0.2% 1−5% 0.09−0.3% 0.3−1% traces ( O2 > Ar > CH4 > N2 > Xe, which brings an interest for natural gas purification.366

Nonporous, or dense, polymeric membranes are the most common used membrane for gas separation (Table 8),368 but they are marginalized for natural gas because of their instability in the presence of condensable gases342 and their low permeability.288 Their functionality follows the solutiondiffusion model.368 Anisotropic (or asymmetric) membranes have two distinct layers at least: a thick porous layer and a thin dense layer, the “skin”. The thickest layer serves as a substrate for the thin selective layer. When exposed to a gas mixture, all of the penetrants diffuse easily through the porous layer. The selection occurs in the thin dense layer. In the case of natural gas separation, CO2 diffuses more quickly than CH4. Althuluth et al.374 reported gas sweetening by supported ionic liquid membrane (SILM) on two γ-alumina layers. The ionic liquid that was employed was [emim][FAP]. The pure gas permeability of different natural gas components was examined through this membrane at a transmembrane pressure of 0.7 MPa and at a temperature of 313 K. The CO2/CH4 selectivity was found to be 9.69. It was reported that the membrane performance was adversely affected by the presence of moisture. Diffusivity values of other gas components including CO2, CH4, C2H6, and C3H10 were modeled with suitable 11458

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

Table 9. Performance of Mixed Matrix Membranes in CO2 Separation from Natural Gas Reported by Various Researchersa polymer matrix polysulfone (Psf) Matrimid 5218 PDMCb 6FDA-DAMc a

membrane thickness (cm)

pressure (bar)

temperature (°C)

CO2/CH4 selectivity

MCM-41 (silica 10%) CMS (17%)

0.0056

4.0

35

23.0

6.6

0.0051

3.4

35

44.8

10.3

6.59 × 10−11

382

AR-SSZ-13 zeolite (25%) ZIF-90 (15%)

0.0047

4.4

35

38.9

153.0

1.08 × 10−9

383

0.0069

2.0

25

37.0

720.0

3.49 × 10−9

384

inorganic filler

342 b

CO2 permeability (barrer)

CO2 permeance (mol/m2 s Pas)

ref

3.94 × 10−13

381

c

Adapted from Muhammad et al. PDMC = (poly acrylamide-co-diallyldimethyl-ammonium chloride). 6FDA-based polyimide (6FDA = 2,2bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride, DAM = diaminomesitylene).

addition of a nonporous nanosized-fumed silica into a glassy polymer with high free volume379 and later confirmed by Ahn et al. for the addition of silica into a low free volume glassy PSf matrix.380 The increase in the free volume by addition of a foreign substance is proven to be beneficial for larger penetrants, such as CH4. However, the increase in permeability is often compensated by the lowering of selectivity.380 Thin films of MOFs provide some potential in membrane applications.385 When used in a membrane, MOFs are mainly integrated in a polymer matrix. As compared to other fillers, MOFs give the opportunity of infinite possible structures and compositions with different combinations of active metal sites and organic linkers. MOFs can also be adhered easily to polymer matrices.386 Because of their unique structure (inorganic metals interconnected by organic fillers), MOFs are characterized by a wide variety of pore shape, size, and surface chemistry. Furthermore, the countless possible organic ligands make them highly tunable for gas sorption and gas selectivity, even after post synthetic modification process. There are several possibilities to employ MOFs for CO2 gas separation. Adjusting their pore size to the smaller kinetic diameter of CO2 (3.3 Å) is an efficient method to separate them from other gases like CH4.386 ZIF-8387 is a well-known MOF with a cross-section of its wider pore of 3.4 Å (Figure 17). It has a stable crystal structure with a sodalite-type

equations, except CH4, which demonstrated a much higher diffusivity. The mixed gas permselectivity of CO2/CH4 (50/ 50%, v/v) was reported to be much lower than the ideal permselectivity (α = 1.15 versus 3.12). Because of such opposing adverse effects, it was concluded that such SILM configuration may not be an ideal choice for CO2 separation from natural gas. 4.2.3.3.3. Hybrid Matrix Asymmetric Membranes. Creating membranes with a high permeation rate requires a small resistance to fluids that can be achieved either through a capillary (using hollow fibers) or through flat sheets. Irrespective of the selected geometry, high permeation can be achieved with the thin design only. Typical thicknesses for a membrane are in the range of 0.1−0.5 μm.292 An anisotropic bulk membrane is composed of two noticeable layers: a thin one, termed as the “skin”, and a thicker layer, termed as “substrate”. The skin part guarantees the separation, whereas the substrate confirms the mechanical properties and the shape of the membrane. Although these two parts can be made of the same material, most advanced membranes are composed of more complex materials and/or structures.375 4.2.3.3.4. Polymer/Inorganic Mixed Matrix Membranes (MMMs). Despite efforts spent into tailoring the polymer structure, the creation of new materials that surpass the upper bound is a difficult task.338,339 To circumvent this difficulty, recent research is more focused on forming and evaluating novel membranes. A common approach comprises the construction of mixed matrix membranes by incorporating microporous materials that have high permeability and permselectivity into an industrially feasible polymer matrix. Microporous materials can be made of either organic (e.g., carbon molecular sieve) or inorganic compounds (e.g., zeolites).339,376 Thus, the heterogeneous hybrid membrane is made of a continuous polymer phase and a dispersed filler phase.377 It potentially combines the advantages of each medium: mechanical performances (flexibility and strength), feasibility (homogeneity, dimensions), with limited costs (controlled recourse to expensive microporous particles). Gas transport through MMMs is considerably influenced by different properties including compatibility, interfacial defects, morphological characteristics, and the process of membrane formation.376 Nevertheless, the transport properties can be modified by addition of nanoparticles into the composite structure of the MMMs (Table 9). Addition of carbon nanotubes (CNTs, 10 Å microporous loaded at 10 wt %) into a polysulfone matrix with porous zeolites significantly influenced the CO2/CH4 separation. Permeability increased from 3.9 to 5.2 barrer, whereas the selectivity reduced from 24 to 18.378 This result is attributed to the increase in free volume (FV). Similar results were also obtained by Pinnau et al. for the

Figure 17. SEM imaging of ZIF membrane on titania support. Reproduced with permission from ref 387. Copyright 2009 American Chemical Society.

morphology with openings in the three spatial directions.388,389 When laid on a titania substrate, a ZIF-8-based membrane can be fabricated by a microwave-assisted solvothermal process in methanol solution. Gas separation experiments on such membrane were conducted using several pure gases and one gas pair (H2/CH4 at 50/50). It was observed that the permeance is a function of the molecular size (kinetic diameter) of the penetrants, and, therefore, the smallest species (H2) has the highest permeance. CO2, which has a kinetic diameter only slightly lower than the pore window of ZIF-8, diffuses almost 3 times less than H2. On the other hand, penetrants that have higher kinetic diameters than the cross section of ZIF-8 11459

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

4.2.3.4. Membrane Limitations. 4.2.3.4.1. Intrinsic Limitations of Membranes Robeson Tradeoff: Permeability versus Permselectivity. In most cases, the membrane presents either a good permeability or a good selectivity, but not both. There seems to be an upper limit that only exceptional materials can cross. Robeson described this phenomenon in 1991 by gathering the permeability and the permselectivity data of various gas pairs from the literature.402 He noticed that the data of the best materials were following a noticeable straight line on a ln(α12) versus ln(P1) plot. He interpreted this line as an upper bound and concluded that there is a tradeoff between permeability and permselectivity (Table 10). An example of this plot is shown in Figure 18 for the gas pair CO2/CH4.403 Other similar plots for different pairs of penetrant gases can be found in Robeson’s works (Figure 19).404,405

marginally pass through the membrane. It is also believed that the lattice of the MOF is not rigid enough to prevent its distension; consequently, it also allows the partial diffusion of larger penetrants.390 In the case of H2/CH4 mixed-gas permeance, the permeation of H2 becomes slightly lower as compared to that of pure H2 due to the partial and competitive diffusion of CH4. This phenomenon is not observed in other inorganic materials such as zeolite.391 The selectivity of the gas pair CO2/CH4 can be improved by functionalizing the pore surfaces with acidic open metal sites. These sites behave as Lewis acids and are often referred to as coordinately unsaturated metal sites (CUMs).392,393 Different functionalities, such as −CF3394 or SO2,395 can be grafted on the surface of the MOFs to provide an improved efficiency in the selective adsorption of the CO2/CH4 mixture. Henke et al.396 used flexible alkyl ether groups as additional functionalities to bind to the rigid phenyl ring of Zn2(bmebdc)2(bipy). The CO2/CH4 selectivity of their design was around 9. It is concluded that the polarization took the predominant role in the molecular sieving effect. Such additional polarization effect can also be achieved by employing a charged framework structure. On the basis of computer simulations, the zeolite-like material (rho-ZMOF) could reach a selectivity of 80 for the gas pair CO2/CH4.397 The high performance of MMMs filled with MOFs requires an optimized interfacial morphology between the matrix and the fillers. More research has to be conducted to improve morphologies, particle/polymer, and the uniformity of the distribution in the composite structure.398 This would allow the researchers to maximize the potential of MOF technology and transform it into a reliable process for gas separation/ purification such as zeolites. 4.2.3.3.5. Polymeric Membranes with Nanoparticles Embedded. Efficient separation of natural gas, or any other gas mixture, using a membrane process requires controlling the free volume (FV) embedded into its bulk. It can be achieved by using nanoparticles to create nanocomposite structures.400 In this type of structure, a proper distribution of the particles is one of the main factors. It contributes to the permeability because it rearranges the voids and defects at the interface between the matrix and the particles. Takahashi et al. demonstrated that voids can create channels through the membrane. Thus, treatment with silica fumes may help to disperse the nanoparticles and to mitigate the void formation.401 The use of a coupling agent (silane) between the nanoparticle and the matrix is another approach to reduce the void and the agglomeration of the fillers, and even the distribution. It is possible to couple fillers chemically to the polymer matrix. In a recent study, Li et al.377 tested the insertion of carbon nanotubes (CNTs) and graphene oxide (GO) fillers in a MMM exposed to two gas pairs (CO2/CH4 and CO2/N2). Up to a certain concentration, nanoparticles can provide benefits to the membrane. Doping a MMM with CNTs improves its permeability. The best permeability was observed at 8 wt% CNTs for both gas mixtures. Meanwhile, doping a MMM with GO nanosheets leads to higher selectivity. The best selectivity was achieved at 2 wt % GO for both gas mixtures. The addition of a single filler is favorable to either the permeability or the selectivity. On the other hand, cumulating both CNTs and GO nanosheets provides MMMs with better permeability and selectivity. In this case, the best properties were observed at 5 wt % of CNTs and 5 wt % of GO.377

Table 10. Experimental Data Points Close to the Empirical Upper Bound for CO2/CH4 Separation Gathered by Robeson in 2008a polymer PVSH doped polyaniline polypyrrole 6FDA/PMDA (25/75)TAB polyimide TADATO/DSDA (1/1)DDBT poly(diphenyl acetylene) 3a polyimide 6FDA-TMPDA/DAT (1:1) polyimide 6FDA-TMPDA/DAT (3:1) polyimide PI-5 poly(diphenyl acetylene) 3e poly(diphenyl acetylene) 3f polyimide 6FDA-TMPDA polyimide 6FDA-durene 6FDA-based polyimide (8) PIM-7 PIM-1 PTMSP PTMSP a

P(CO2) 0.029 3.13 45 110 130.2 187.6 190 290 330 555.7 677.8 958 1100 2300 19 000 29 000

α(CO2/CH4)

ref

2200 140

404 405

60

406

47.8 38.9 33.9 33.9 31.5 27.5 22.7 20.18 24 17.7 18.4 4.42 4.46

407 408 415 409 414 414 415 410 411 412 419 413 420

Reproduced with permission from ref 403. Copyright 2008 Elsevier.

Figure 18. Prediction of CO2/CH4 performance in hypothetical mixed-matrix membranes along with known polymers. Reproduced with permission from ref 399. Copyright 2010 Royal Society of Chemistry. 11460

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

the volume reduction is a function of the material,314 its polymer package, and its thickness. Material exhibiting high FFV experiences a faster aging rate as compared to others.418,419 4.2.3.4.3. Limitations by Wear and Tear or Excessive Process Conditions. The environment and the application of the membrane may have a strong impact on its performances and reliability. Particular attention must be given to CO2 because it has a higher critical temperature and is significantly more soluble in glassy polymers than other gases such as CH4, He, N2, or Ar.420 4.2.3.4.4. Plasticization. Dry cellulose acetate (CA) membranes have been used for natural gas separation for decades. The values of their selectivity were significantly below expected theoretical values because of plasticization, caused by a variety of condensable gases. CO2, H2S, and heavier hydrocarbons were already at the origin of this phenomenon. Plasticization is a major limitation of membrane performance.290,300,375,421 It has the undesirable effect of filling the FV of the polymer matrix,420,422−424 causing the matrix to swell. Swelling eases the permeation of other penetrants, resulting in a significant loss of selectivity of the membrane.425 Another consequence of plasticization is the lowering of Tg and the increase in softness and ductility.420,426 This phenomenon, induced by CO 2 , was already described in 1979 in polycarbonate.427 It must be noted that the plasticization is also thickness dependent: ultrathin dense polyimide films (0.5− 1.8 μm) experience stronger plasticization effects as compared to thicker films. Plasticization caused by crude natural gas on a membrane can be quite detrimental. An example of plasticization of an asymmetric CA membrane has been reported by Funk et al.428 The gas mixture was composed of 29.3 mol% CH4, 44.9 mol% CO2, 16.9 mol% H2S, higher hydrocarbons, various organics, N2, and H2O. At about 38 °C and at a pressure of 14.6 atm (215 psia), the selectivity of the CA membrane was reduced from 18 to 14.429 To limit the effect of plasticization, several techniques have been developed, including physical blending, formation of semi-interpenetrating networks, physical cross-linking effect such as hydrogen bonding, hyperbranched structure, thermal cross-linking, UV cross-linking, and chemical cross-linking. Thermal annealing can be effective for suppressing CO2-induced plasticization.430 Swelling and plasticization did not manifest for a strongly crosslinked 6FDA−DABA polyimide under 20 atm of CO2/CH4 (50/50), and the selectivity remained high at 70 atm.320 However, excessive cross-linking increases the brittleness of the membrane, making overcross-linked membranes unsuitable at high pressure.300 Cross-linking Matrimid films by thermal treatment at 350 °C is a successful method to suppress CO2induced plasticization. 15−30 min of treatment was proven to be quite enough to protect the membrane from a single gas permeation or a CO2/CH4 mixture at high pressure.425 4.2.3.4.5. Conditioning. In addition to the plasticization phenomenon, high soluble gases contained in natural gas deteriorate membrane performance by conditioning.431 This problem is strongly detrimental for a glassy polymer membrane exposed to high pressure CO 2 followed by a partial depressurization.333 Indeed, a large amount of penetrants can significantly change the backbone chain configuration of the polymers. For example, high condensation of CO2 in polyimide membranes provokes a rise in diffusion and solubility due to volume expansion. The membrane performances are irreversibly impacted by conditioning,314 even after total desorption of

Figure 19. Upper bound correlation of CO2/CH4. With time and the discovery of new materials, the upper bound shifts, but its slope remains constant. Blue points represent data for thermally rearranged polymers (TR). These polymers are benzoxazole-phenylene or benzothiazole-phenylene and are considered to be in the class of molecular sieving materials produced by thermal reaction. Reproduced with permission from ref 410. Copyright 2008 Elsevier.

The upper bound shown in Figure 18 can be described by the following equation: ln α12 = ln β12 − λ12 ln P1

(21)

where α12 is the coefficient of separation, and P1 is the permeability of the best permeate of the considered gas pair. Freeman revisited this equation and proposed the following model (eq 22): ⎧ ⎡⎛ ⎞ 2 ⎤ ⎡⎛ ⎞ 2 ⎤ ⎪ ⎛S ⎞ d d ln α12 = −⎢⎜ 1 ⎟ − 1⎥ ln P1 + ⎨ln⎜ 1 ⎟ − ⎢⎜ 2 ⎟ − 1⎥ ⎪ ⎝S ⎠ ⎢⎣⎝ d 2 ⎠ ⎥⎦ ⎥⎦ ⎢⎣⎝ d1 ⎠ 2 ⎩ ⎫ ⎛ ⎞⎪ ⎛1 − a ⎞ ⎟ − ln S ⎟⎬ × ⎜b − f ⎜ 1 ⎪ ⎝ RT ⎠ ⎝ ⎠ ⎭

(22)

In the case of glassy polymers-based membranes, a and b are constants (a = 0.64, and b = 11.5 cm/s), and f is a function of the polymers. From eqs 21 and 22, it can be deduced that (i) the slope of the upper bound λ12 is only a function of the gas kinetic diameters; and (ii) the intercept is a strong function of the nature of the penetrant and the polymer. Thus, the pair of gas imposes the slope of the upper bound. 4.2.3.4.2. Physical Aging. Because glassy polymers are solids that have been “immobilized” at nonequilibrium states by cooling below their Tg, they experience slow progressive evolution toward a natural equilibrium.414 This phenomenon is caused by a hydrostatic internal stress that is strong enough to produce a bulk creep reduction (gradual contraction of the polymer). This contraction is possible due to the reduction of the FV. This phenomenon is known as physical aging,415 or volume relaxation.416 It is a time-dependent and reversible phenomenon417 that can be observed on membranes stored in safe condition with no gas pressure.334 The membrane material becomes stiffer and more brittle, thereby resulting in changes in the properties of permeability and selectivity.314 The pace of 11461

DOI: 10.1021/acs.chemrev.5b00745 Chem. Rev. 2016, 116, 11436−11499

Chemical Reviews

Review

the penetrants.431 Conditioning effects depend on the nature of the penetrants, process conditions, and prior history of the system. Polysulfone and polycarbonate exposed to N2, Ar, or CH4 exhibit almost no sorption or conditioning as compared to CO2. However, once polymers like polysulfone or polycarbonate have been exposed to CO2 (Figure 20), the solubility of gases such as N2, Ar, and CH4 increases significantly (by 21% for Ar, 20% for N2, and 14% for CH4 in the case of polysulfone).427,432

Table 11. Comparison of Amines and Membranes for CO2 Removal Systemsa amines Operating Issues comfort level of user loss of hydrocarbon low CO2 specification low H2S specification consumption of energy operating cost maintenance cost ease of operation environmental impact dehydration

very familiar

still considered new technology

very low

losses depend upon conditions

meets (ppm levels) meets (