SPE-172085-MS

Purification of Natural Gas by Using Carbon Molecular Sieve Membrane

Subrata Mondal, and Kean Wang, Department of Chemical Engineering, The Petroleum Institute, Sas Al Nakhl,Abu Dhabi, United Arab Emirates

Copyright 2014, Society of Petroleum Engineers

This paper was prepared for presentation at the Abu Dhabi International Petroleum Exhibition and Conference held in Abu Dhabi, UAE, 1013 November 2014.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contentsof the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflectany position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the writtenconsent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations maynot be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract

Carbon molecular sieve membrane (CMSM) is the promising candidate for natural gas purificationbecause of its excellent stability, per-selectivity and permeability. However, morphological design ofCMSM is very important for the specific application. In this project, two CMSM samples weresynthesized at different pyrolytic conditions and examined for separation of N2/CH4 gas pair. Adsorptionand permeation experiments were conducted to examine the separation performance of each membranesample. At the ambient conditions, a perm-selectivity of ~ 6 was found for (N2/CH4) pair while it is ~30for (CO2/N2) pair on the membrane pyrolyzed at the lower pyrolytic temperature of 800C. Whenpyrolytic temperature is increased to 1000C, however, these two selectivity values are changed to ~1.5(decrease) and ~ 100 (increase), respectively. Analysis revealed that both surface diffusion and molecularsieving play important roles in the overall gas permeation mechanisms, which result in the abnormalbehaviors in the selectivities of different gas molecules. It is concluded that the variation of pore size andthickness are critical for the design of surface flow selective carbon molecular sieve membranes.

Key words Natural gas carbon molecular sieve membrane N2 purification CO2 purification

IntroductionDue to its abundance in natural reserve and relatively clean burning nature, natural gas (NG, CH4 70%)is becoming increasingly more important as the source of energy and constitutes 23.8% of worlds energysupply in 2010[1]. Raw NG contains such impurities as CO2, H2S, H2O and N2, etc., which result in healthhazards, corrosion of equipment, and lower heating value. The concentration of these impurities variesconsiderably in NGs produced at different locations but should be removed to certain levels beforeshipment. Table 1 summarizes the composition of raw NGs and the US specification for commercialpipeline NG.

Currently, large-scale NG processing plants employ such technologies as adsorption (e.g. dehydration),absorption (e.g. amine wash for the removal of H2S and CO2), and cryogenic distillation (e.g. N2removal), etc.. Although these technologies are mature and effective, they are expensive, energy intensiveand environmentally unfriendly. For example, an N2 removal unit (NRU) generally operates at T -150C (P ~7 bar), while the amine washing unit requires constant maintenance and solvent replenishing[2, 3].

Membrane technology offers great advantages for NG processing. It is compact, easy to install and/orscale-up, and requires minimal resources such as space, energy input, labor and maintenance. Theseattributes make it particularly attractive in off-shore and remote gas fields [1-3]. Nowadays, membranemodules for CO2 removal are commercially available and is the dominant technology in off-shoreapplications while becoming more competitive in on-shore applications [4]. However, improvements areneeded to compete with the current technologies in large-scale, inland NG processing plants [1, 2, 4].

The majority of the commercial membrane modules comprise of thin polymeric films or hollow fibers[1, 5, 6]. Polymers possess good flexibility and are easy to process, as they can be easily cast into thin filmor spun into fine hollow-fibers. However, polymeric membranes are susceptible to degradation whenexposed to an aggressive feed because they work on a solution-diffusion mechanism. NG contains sourgases (CO2 & H2S), oil mist and moisture, which can cause plasticization and degradation of the polymermatrix. As a result, the polymeric membrane will swell, dilate, and consequently lose its selectivity [3].

Carbon Molecular Sieve Membrane or CMSM, has unique advantages for NG processing. It isproduced from the controlled pyrolysis of polymeric precursors, with rigid graphite microstructure,superior selectivity, and thermal stability. For air (O2/N2) separation, the perm-selectivity of CMSM is 10folds higher than in polymeric membranes at a similar flux [7].

CMSM has two other attributes: (1) its inert carbon surface is much less affected by aggressive feed(e.g. NG), and (2) its pore size distribution (PSD) is largely tunable via the thermal-treatment [5]. CMSMhas been fabricated in the configuration of hollow fiber (asymmetric)[8], film (symmetric)[9] andcomposite (supported)[10], and used to separate: air (N2/O2), NG (CH4/CO2, CH4/N2), hydrocarbons(C2H4/C2H6), syngas (CO2/H2), and flue gas (N2/CO2). CMSM pilot plants for air separation have beenreported [11]. CMSMs were prepared from the pyrolysis of such polymeric precursors as: polyimide (PI),phenolic resin, polyfurfury alcohol (PFFA), polyacrylonitrile (PAN), cellulose, polyetherimide (PEI),polyvinylidene chloride (PVDC). Table 2 summarizes some recent research results on CMSMs.

More research has been done on CO2 separation than on N2 separation (Table 2). The separation ofCH4/N2 is a challenging issue due to two reasons: a) N2 molecule is slightly smaller than CH4 (3.64 vs3.80) and possesses a higher diffusivity, or Di; but b) CH4 is more condensable than N2 and presents ahigher adsorption affinity, or bi. As the result, their permeabilities, which is defined as Pi bi Di, canbe very close to each other. To explain this issue further, Table 3 lists the molecular properties andadsorption properties (heat of adsorption, Hads; affinity, b0) on a carbon surface for CH4, N2, andCO2[19].

For N2 removal, membranes can be prepared to be selective (more permeable) to either N2 or CH4, withthe former based on the mechanism of molecular sieving (or size exclusion, Figure 1a) while the laterdictated by the surface flow (rate of surface diffusion, Figure 1b).

The difference of 0.16 between the dimension of CH4 and N2 molecules make it virtually impossiblefor a sieving membrane to achieve a good selectivity while keeping a reasonable flux [21]. Yet, CH4adsorbs more strongly in graphite pores than N2 (a higher value of heat of adsorption, Hads, as listed inTable 3). Thus, the adsorbed phase concentration can be much higher for CH4 than N2. The surfacediffusion (or surface flow), which is strongly concentration dependent, will result in a product enrichedwith CH4 (Figure 1b). This approach is more flexible and practical because we can tailor the pore size,and vary the pressure and temperature of feed to optimize the selectivity and flux of the CMSM. Thus,the surface flow selective CMSM is a very promising technology for N2 removal from NG.

Table 1The composition and commercial specification of NG [2, 3]

Species CH4 (%) CO2 (%) H2S (ppm) N2 (%) H2O (ppm) Hydrocarbons (C2-C4)

Raw NG Composition 70 - 95 0.5-20 4 - 1104 0.21-26 800-1200 diversified

Pipeline NG Specs 75 2 4 4 120 dew point -20C

2 SPE-172085-MS

Currently, polymeric (rubbery/glassy) membranes reported the CH4/N2 selectivity of 2-3.5[22], whilethe inorganic membranes (zeolite silica, carbon) reported selectivity 5[1]. For polymeric membranes,multiple-staged module systems were needed to reduce N2 content from 15% to below 4%, as the demounit designed by the Membrane Technology & Research (MTR) [1, 22].

Rao and Sircar are the pioneers who fabricated the surface flow selective membranes for hydrocarbon/gas separation. They coated PVDF on the inner surface of an porous support, followed by pyrolysis at600C under N2 flow [17]. The composite membrane demonstrated the surface flow selective mecha-nism and preferentially permeates the strongly adsorbed hydrocarbon gases (CH4, C2H6, etc.) rather thanthe small H2 molecules. A perm-selectivity of 6 was reported for C2H6/H2 (Table 2) [17]. They also brieflyinvestigated the permeation of CH4/N2 with their system. An ideal selectivity of 8.5 was calculated basedon the permeance of each pure gas, while a perm-selectivity of 2.9 was measured at the steady statepermeation of the binary gas mixtures [18]. The possible reason for the low selectivity is that the systemwas designated for syngas production rather than for NG purification.

In summary, the surface flow selective CMSM is a very promising but challenging technology for NGprocessing (CH4/N2 separation).

Table 2CMSMs for gas separation

Precursor Configuration Feed & Conditions

PerformancePermeance/Selectivity

permeability Ref.

Polyimide FilmHollow fiberHollow fiber

C2H4/C2H6 C3H6/C3H8CO2/CH4

35C, 50 bar35C, 4 bar35C, 8 bar

10 GPU10 GPU42 GPU

122023

[12, 13][13][11]

Polyimide Follow fiber dual layer O2/N2 CO2/CH4 35C, 1-5 bar35C, s1-5 bar

8 GPU13 GPU

2150

[8][8]

Phenolic resins SupportedSupportedSupported

O2/N2C3H6/C3H8CO2/CH4

35C, 1 bar20C, 4 bar25C, 1 bar

30 GPU100 Barrer~7 GPU

1215170

[14][15][10]

Cellophane paper FilmFilmFilm

O2/N2 CO2/CH4N2/CH4Calculated from pure gass-

30C, 2 bar30C, 2 bar30C, 2 barpermeability

~10 Barrer~10 Barrer0.05 Barrer

101704

[16][16][16]

PVDC Supported C2H6/H2CH4/H2CH4/N2 (calculated) CH4/N2(measured)

23C, 2 bar23C, 2 bar23C, 2 bar23C, 2 bar

7 Barrer1.3 Barrer1.3 Barrer-

658.82.9

[17][17][17][18]

Note: Permeance: 1 Gas permeation Unit (GPU) 110-6 (cm3-STP)/(cm2s cmHg);Permeability: 1 Barrer 110-10 (cm3-STPcm)/(cm2s cmHg).

Table 3The molecular properties of CH4, N2, and CO2.

MW(g/mol)

Kineticdiameter,

()[20]

/K[20]

(K)b0

[19]

(bar-1)Hads

[19]

(kJ/mol) Hevap (kJ/mol)Normal Boiling Point

(NBP, C)

CH4 16 3.80 148.6 2.910-7 22.7 8.2 -156

N2 28 3.64 101.5 4.110-7 19.0 5.6 -196

CO2 44 3.30 195.2 2.710-7 26.0 26.1 N/A

Figure 1Schematic diagram of two mechanisms: (a) molecular sieving, (b) surface flow.

SPE-172085-MS 3

TheoryAs CMSM can be both sieving selective and surface flow selective, a mathematical modeling can providethe guidelines on: i) the optimal pore size range for the separation [20, 23, 24]. The surface diffusivity,D, and the adsorption affinity, b0, of each species can be respectively represented as:

(1)

Where Ea is the activation energy for diffusion, is the diffusivity at zero loading and infinitetemperature, Hads is the heat of adsorption, M is molecular weight, and relates to the gas frequencyconstant [23]. The ideal permeability of component i is defined as:

(2)

The ideal perm-selectivity, ij, is then expressed as:

(3)

Where Ep Hads Ea is the activation energy for permeation. Both Hads and Ea are related to theadsorption potential (e.g. Lennard-Jones 10-4-3 potential), , in a local graphite pore (with the size of 2r)as:

(4)

Where z is the location in the local pore, is the inter-layer distance of graphite units, sk and sk arethe cross L-J parameters for adsorbatecarbon which are calculated using a Lorentz-Berthelot rule,respectively [20]. With eq.(3), the selectivity, ij, for a pore (with the size of 2r) can be calculated byminimizing the adsorption potential in eq.(4) with the LJ parameters listed in Table 3. Figure 2 shows thesimulated ideal CH4/N2 selectivity vs the pore size (centre centre) of the CMSM. We see that the optimalCH4/N2 selectivity, ij, occurs in pores of ~6.8 (or a wallwall pore size of ~4.5 ), and this selectivityis strongly dependent on the temperature (ij8.5@ 303K; but ij 12.5@ 253K). It should be pointedout that above simulation is oversimplified with such assumptions of: linear isotherm, equal molar ratio,homogeneous slit pore, activation energy is half of the heat of adsorption, CH4 and N2 are sphericalmolecules, Fickian type diffusion, surface flow dominant, etc.. However, it does give us the guidance forthe surface flow selective CMSM.

Figure 2The simulated CH4/N2 selectivity vs. pore size at different temperatures

4 SPE-172085-MS

Experiments

Two types of carbon molecular sieve membranes were prepared via the controlled pyrolysis ofKapton polyimide thin film under high vacuum. The highest pyrolysis temperature of 800 C (referredto as KP 800) and 1000C (referred to as KP 1000), respectively, with the step-wise thermal treatmentprogram was employed which is described in the literature [25]. Previous studies have shown that CMSMsderived under such conditions are pinhole-free, symmetric in structure and consists of mainly carbon.[9,26] The membrane is slab in geometry with a thickness of 125m and 23 m, respectively. Table 4 showsthe pyrolysis condition of the two membranes.Adsorption isotherms of CO2, N2, and CH4 gases were measured on the CMSM samples with a Cahn

2000 microbalance (Resolution: 10-3 g). The pure gas permeation of the each species were measuredrespectively with a high pressure time-lag device. The details of the equilibrium and permeationexperiments are given in the references. [9, 25, 27]

Results and discussionFigure 3 shows the adsorption isotherm of N2 and CH4 measured on KP 800 at ~30C, respectively.

It is seen in Figure 3 that, on KP800 (with an average pore size ~ 4.6A and a thickness of 125m), CH4has an adsorption capacity ~6 that of the N2 at high pressure (35 bar) and 4 time at low pressure (1.5 bar).We expect that, as temperature further decreases, CH4 becomes more significantly adsorbed and thisdifference would be larger. This observation is in agreement with our simulation results in Figure 2. KP1000 has an average pore size ~ 4A and thickness of ~23m [9], so both CH4 and N2 will be adsorbedwith a smaller capacity (less porous), but with a similar trend of temperature dependence, as shown inFigure 3.

Table 5 shows the permeabilities and selectivities of a few gas molecules on the two samples at ambientconditions. It can be seen that the membranes present reasonable sieving effect for gas molecules withdifferent kinetic diameters, suggesting that the CMSMs are predominantly microporous with no major

Table 4The pyrolysis parameters and the properties of the two membrane samples

Sample Temp. (C) Soaking Time(hr) Heating rate Weight loss Thickness (m) Density (g/cc)

KP1000 1000 1 10C/min ~30% 23 ~2.14

KP800 800 1 10C/min ~15% 125 ~1.99

Figure 3CH4 and N2 isotherms on KP 800

SPE-172085-MS 5

contribution from Knudsen diffusion or viscousflow in the overall mass transfer. This is in agree-ment with previous results from molecular probingexperiments on the CMSM prepared under similarconditions.[9, 28]

Let us have an in-depth analysis of these selec-tivity/permeation data. Figure 4 plotted the perm-selectivity of (N2/CH4, N2/CH4) on the two CMSMvs the Lennard-John diameters of CO2 and CH4,respectively.

We see that, compared with KP 800, KP 1000presents much higher perm-selectivity towards CO2(or a lower selectivity towards N2). This is expectedbecause CO2 is a small molecule with strongeradsorption affinity than N2. A smaller membranepore size of 4A will enhance the molecular sieving effect as well as the surface flow effect, thus resultingin a higher perm-selectivity.

For N2/CH4 pair, however, things are very different. Compared with CH4, N2 is a molecule withsmaller diameter but weaker adsorption affinity. Decreasing the pore size will have two effects: 1)increasing the molecular sieving separation, this will enhance the selectivity towards N2; however, 2) itwill also enhance the adsorption affinity in the smaller pores, therefor promote the surface flow. BecauseCH4 has a stronger adsorption affinity than N2, its surface flow increases more significantly in KP 1000(more sensitive to the variation of pore size due to its larger molecular dimension). The combination ofthe two mechanisms results in the abnormal behavior on KP 1000 for N2/CH4 pair, that is, a lowerpermeation selectivity towards N2 in the membranes with smaller pore size (or a higher selectivity towardsCH4). These results give quantitative evidence of the surface flow in the CMSMs. As the diffusion path(or membrane thickness) increases, the contribution of surface flow is also expected to increase. The exactoptimal pore size for the surface selective CMSM has to be further investigated.

ConclusionBoth molecular sieving and surface flow play an important role in the gas separation/permeation in carbonmolecular sieve membranes. As the pore size decreases from 4.6A to 4A, the contribution of the surfaceflow is significantly increased for CH4 molecules so that the apparent N2/CH4 selectivity decreases from~ 6 to ~ 1.5. The moderation of the pore size, membrane thickness, as well as the operation conditionscan fundamentally change the overall selectivity of N2/CH4 on the CMSM.

Table 5Gas permeation properties of CMSMs at ambient conditions

Gases N2 CH4 CO2 CF4

Kinetic diameter () 3.64 3.80 3.30 4.70

Permeability(Barrer)

KP 800KP 1000

7.1~0.05

1.24~0.03

273.64.2

0*0*

Selectivity(PN2/Pi)

KP 800KP 1000

11

5.71.7

0.0250.012

* Observation for 12 hours

Figure 4The perm-selectivity of N2/CH4 and N2/CO2 on two CMSMs

6 SPE-172085-MS

AcknowledgementAuthors would like to acknowledge Gas Research Committee (GRC) project code GRC10 (Fund code11810) for funding of this project.

Reference1. T.E. Rufford, S. Smart, G.C.Y. Watson, B.F. Graham, J. Boxall, J.C. Diniz da Costa, E.F. May,

The removal of CO2 and N2 from natural gas: A review of conventional and emerging processtechnologies, Journal of Petroleum Science and Engineering, 94-95 (2012) 123154.

2. W.P. A. J. Kidnay, Fundamentals of Natural Gas Processing, CRC Press, (2006).3. R.W. Baker, K. Lokhandwala, Natural gas processing with membranes: An overview, Industrial

and Engineering Chemistry Research, 47 (2008) 21092121.4. R.W. Baker, K. Lokhandwala, Natural gas processing with membranes: An overview, Ind Eng

Chem Res, 47 (2008) 21092121.5. A.F.I. S.M Saufi, Fabrication of carbon membranes for gas separationa review, Carbon, 42

(2004).6. T.J. Hughes, A. Honari, B.F. Graham, A.S. Chauhan, M.L. Johns, E.F. May, CO2 sequestration

for enhanced gas recovery: New measurements of supercritical CO2-CH4 dispersion in porousmedia and a review of recent research, International Journal of Greenhouse Gas Control, 9(2012) 457468.

7. W.J. Koros, R. Mahajan, Pushing the limits on possibilities for large scale gas separation: whichstrategies?, Journal of Membrane Science, 175 (2000) 181196.

8. T.-S.C. Lan Ying Jiang, Raj Rajagopalan, Dual-layer hollow carbon fiber membranes for gasseparation consisting of carbon and mixed matrix layers, Carbon, 45 7.

9. K.H. Hiroyuki Suda, Gas Permeation through Micropores of Carbon Molecular Sieve MembranesDerived from Kapton Polyimide, J. Phys. Chem. B, 101 (1997) 7.

10. T.A. Centeno, A.B. Fuertes, Supported carbon molecular sieve membranes based on a phenolicresin, Journal of Membrane Science, 160 (1999) 201211.

11. O. Karvan, J.R. Johnson, P.J. Williams, W.J. Koros, A Pilot-Scale System for Carbon MolecularSieve Hollow Fiber Membrane Manufacturing, Chemical Engineering and Technology, 36 (2013)5361.

12. M. Rungta, L.R. Xu, W.J. Koros, Carbon molecular sieve dense film membranes derived fromMatrimid (R) for ethylene/ethane separation, Carbon, 50 (2012) 14881502.

13. L. Xu, M. Rungta, M.K. Brayden, M.V. Martinez, B.A. Stears, G.A. Barbay, W.J. Koros,Olefins-selective asymmetric carbon molecular sieve hollow fiber membranes for hybrid mem-brane-distillation processes for olefin/paraffin separations, Journal of Membrane Science, 423-424 (2012) 314323.

14. W. Zhou, M. Yoshino, H. Kita, K.-i. Okamoto, Preparation and gas permeation properties ofcarbon molecular sieve membranes based on sulfonated phenolic resin, Journal of MembraneScience, 217 (2003) 5567.

15. M. Teixeira, M.C. Campo, D.A.P. Tanaka, M.A.L. Tanco, C. Magen, A. Mendes, Compositephenolic resin-based carbon molecular sieve membranes for gas separation, Carbon, 49 (2011)43484358.

16. M.C. Campo, F.D. Magalhes, A. Mendes, Carbon molecular sieve membranes from cellophanepaper, Journal of Membrane Science, 350 (2010) 180188.

17. M.B. Rao, S. Sircar, Nanoporous carbon membranes for separation of gas mixtures by selectivesurface flow, Journal of Membrane Science, 85 (1993) 253264.

SPE-172085-MS 7

18. M. Anand, M. Langsam, M.B. Rao, S. Sircar, Multicomponent gas separation by selective surfaceflow (SSF) and poly-trimethylsilylpropyne (PTMSP) membranes, Journal of Membrane Science,123 (1997) 1725.

19. C.J. Anderson, W.D. Tao, J.W. Jiang, S.I. Sandler, G.W. Stevens, S.E. Kentish, An experimentalevaluation and molecular simulation of high temperature gas adsorption on nanoporous carbon,Carbon, 49 (2011) 117125.

20. K. Wang, D.D. Do, Characterizing the micropore size distribution of activated carbon usingequilibrium data of many adsorbates at various temperatures, Langmuir, 13 (1997) 62266233.

21. T.E. Rufford, S. Smart, G.C.Y. Watson, B.F. Graham, J. Boxall, J.C.D. da Costa, E.F. May, Theremoval of CO2 and N-2 from natural gas: A review of conventional and emerging processtechnologies, Journal of Petroleum Science and Engineering, 94-95 (2012) 123154.

22. K.A. Lokhandwala, I. Pinnau, Z. He, K.D. Amo, A.R. DaCosta, J.G. Wijmans, R.W. Baker,Membrane separation of nitrogen from natural gas: A case study from membrane synthesis tocommercial deployment, Journal of Membrane Science, 346 (2010) 270279.

23. D.D. Do, K. Wang, Predictions of adsorption equilibria of nonpolar hydrocarbons onto activatedcarbon, Langmuir, 14 (1998) 72717277.

24. J. Chen, L.S. Loo, K. Wang, D.D. Do, The structural characterization of a CMS membrane usingAr sorption and permeation, Journal of Membrane Science, 335 (2009) 14.

25. C. Nguyen, D.D. Do, K. Haraya, K. Wang, The structural characterization of carbon molecularsieve membrane (CMSM) via gas adsorption, Journal of Membrane Science, 220 (2003) 177182.

26. K. Wang, H. Suda, K. Haraya, Permeation time lag and the concentration dependence of thediffusion coefficient of CO2 in a carbon molecular sieve membrane, Ind Eng Chem Res, 40 (2001)29422946.

27. K. Wang, L.S. Loo, K. Haraya, CO2 permeation in carbon membranes with different degrees ofcarbonization, Ind Eng Chem Res, 46 (2007) 14021407.

28. K. Wang, Diffusion anomaly and blind pore character in carbon molecular sieve membrane,Chemical Engineering Science, 62 (2007) 36543659.

8 SPE-172085-MS

Purification Of Natural Gas By Using Carbon Molecular Sieve MembraneIntroductionTheoryExperimentsResults and discussionConclusion

AcknowledgementReference