REMOVAL OF CO ₂ , N ₂ , AND H ₂ S FROM GAS MIXTURES CONTAINING SAME

FIELD OF THE INVENTION

[0001] The present invention relates to the removal of one or more of the gases CO ₂ , N ₂ and H ₂ S from gas mixtures containing at least one of said gases with use of an 8-ring zeolite having a Si: Al ratio from about 1 : 1 to about 1000: 1. The preferred gas mixture is a natural gas feedstream and the preferred 8-ring zeolite is DDR.

BACKGROUND OF THE INVENTION

[0002] Gas separation is important in various industries and can typically be accomplished by flowing a mixture of gases over an adsorbent that preferentially adsorbs a more readily adsorbed component relative to a less readily adsorbed component of the mixture. One of the more important gas separation techniques is pressure swing adsorption (PSA). PSA processes rely on the fact that under pressure gases tend to be adsorbed within the pore structure of the microporous adsorbent materials or within the free volume of a polymeric material. The higher the pressure, the more gas is adsorbed. When the pressure is reduced, the adsorbed gas is released, or desorbed. PSA processes can be used to separate gases from a mixture of gases because different gases tend to fill the micropore, or free volume, of the adsorbent to different extents. For example, if a gas mixture such as natural gas is passed under pressure through a vessel containing polymeric or microporous adsorbent that fills with more nitrogen than it does methane, part or all of the nitrogen will stay in the sorbent bed, and the gas coming out of the vessel will be enriched in methane. When the bed reaches the end of its capacity to adsorb nitrogen, it can be regenerated by reducing the

_ _

pressure, thereby releasing the adsorbed nitrogen. It is then ready for another cycle.

[0003] Another important gas separation technique is temperature swing adsorption (TSA). TSA processes also rely on the fact that under pressure gases tend to be adsorbed within the pore structure of the microporous adsorbent materials or within the free volume of a polymeric material. When the temperature of the adsorbent is increased, the gas is released, or desorbed. By cyclically swinging the temperature of adsorbent beds, TSA processes can be used to separate gases in a mixture when used with an adsorbent that selectively adsorbs one or more of the components in the gas mixture relative to another.

[0004] Adsorbents for PSA systems are usually very porous materials chosen because of their large surface area. Typical adsorbents are activated carbons, silica gels, aluminas and zeolites. In some cases a polymeric material can be used as the adsorbent material. Though the gas adsorbed on the interior surfaces of microporous materials may consist of a layer of only one, or at most a few molecules thick, surface areas of several hundred square meters per gram enable the adsorption of a significant portion of the adsorbent's weight in gas.

[0005] Different molecules can have different affinities for adsorption into the pore structure or open volume of the adsorbent. This provides one mechanism for the adsorbent to discriminate between different gases. In addition to their affinity for different gases, zeolites and some types of activated carbons, called carbon molecular sieves, may utilize their molecular sieve characteristics to exclude or slow the diffusion of some gas molecules into their structure. This provides a mechanism for selective adsorption based on the size of the molecules and usually restricts the ability of the larger molecules to be adsorbed. Either of these mechanisms can be employed to selectively fill the micropore structure of an adsorbent with one or more species from a multi-

component gas mixture. The molecular species that selectively fill the micropores or open volume of the adsorbent are usually referred to as the "heavy" components and the molecular species that do not selectively fill the micropores or open volume of the adsorbent are usually referred to as the "light" components.

[0006] An early teaching of a PSA process having a multi-bed system is found in U.S. Patent No. 3,430,418 wherein a system having at least four beds is described. This '418 patent describes a cyclic PSA processing sequence that includes in each bed: (1) higher pressure adsorption with release of product effluent from the product end of the bed; (2) co-current depressurization to intermediate pressure with release of void space gas from the product end thereof; (3) countercurrent depressurization to a lower pressure; (4) purge; and (5) repressurization. The void space gas released during the co-current depressurization step is commonly employed for pressure equalization purposes and to provide purge gas to a bed at its lower desorption pressure. Another conventional PSA processes using three sorbent beds is disclosed in U.S. Patent No. 3,738,087. Conventional PSA processes are typically able to recover only one of the key components (i.e., light or heavy) at high purity and are unable to make a complete separation and separate both components with a high recovery. The light component usually has a low recovery factor. Recovery of the light component usually drops even lower when the feed gas is introduced at higher pressures (i.e., pressures above 500 psig).

[0007] While there are various teachings in the art with respect to new adsorbent materials, new and improved adsorbent contactors, and improved rapid cycle PSA equipment, none of these to date are capable of removing a commercially feasible amount of contaminants such as CO ₂ , N ₂ and H ₂ S from natural gas, particularly when relatively large amounts of such contaminants are present, particularly at high pressures. This is an important issue because natural

_ _

gas is usually produced at high pressures (500-5000 psi) and methane acts as a light component in the adsorption process. It is preferred that as much of the acid gases H ₂ S and CO ₂ be removed from natural gas as possible. In all natural gas separations, methane is a valuable component and acts as a light component in swing adsorption processes. While various processes exist for removing CO ₂ , H ₂ S, and N ₂ from natural gas there remains a need for processes and materials that will perform this recovery more efficiently, at lower costs, and at higher purified methane product streams.

SUMMARY OF THE INVENTION

[0008] In accordance with the present invention there is provided a process for the removing a first gas component selected from the group consisting of CO ₂ , N ₂ , and H ₂ S from a gas mixture containing said first gas component and a second gas component, which process comprises:

a) conducting said gas mixture to a swing adsorption gas separation unit wherein the gas separation unit contains at least one adsorbent contactor comprised of:

i) a gas inlet end; and

ii) a gas outlet end;

wherein the gas mixture is introduced into said gas inlet end and a product stream exits the gas outlet end, which is in fluid communication with said gas inlet end, by a plurality of open flow channels, wherein the surface of the open flow channels are comprised of an adsorbent material comprised of an 8-ring zeolite that has a selectivity for said first gas component over said second gas component greater than 5, wherein at least a portion of said first gas component

_ _

is adsorbed into said adsorbent material, thereby resulting in said product stream which is depleted of said first gas component;

b) collecting said product stream;

c) desorbing the adsorbed gases from said adsorbent material, thereby resulting in a waste gas stream rich in said first gas component; and

d) collecting said waste gas stream.

[0009] In a preferred embodiment, the adsorbent contactor is a parallel channel contactor comprising structured (engineered) adsorbents in which substantially parallel flow channels are incorporated into the adsorbent structure.

[0010] In another embodiment, the open flow channels have a channel gap from about 5 to about 1000 microns.

[0011] In still another embodiment the adsorbent material is comprised of an 8-ring zeolite that has a Si to Al ratio of from about 1 :1 to about 1000:1.

[0012] In another embodiment, the 8-ring zeolite is a DDR framework zeolite; and in yet and the embodiment, the 8-ring zeolite is selected from Sigma-1 and ZSM-58.

[0013] In another embodiment the contactor has less than about 10% of its open pore volume in pores with diameters greater than about 20 angstroms and less than about 1 micron.

- O -

BRIEF DESCRIPTION OF THE FIGURES

[0014] Figure 1 hereof is a representation of one embodiment of a parallel channel contactor of the present invention in the form of a monolith directly formed from the microporous adsorbent of the present invention and containing a plurality of parallel channels.

[0015] Figure 2 hereof is a cross-sectional representation along the longitudinal axis of the monolith of Figure 1.

[0016] Figure 3 hereof is another representation of an embodiment of a parallel channel contactor of the present invention in the form of a coated monolith where the adsorbent layer is coated onto the channel wall.

[0017] Figure 4 hereof is an electron micrograph of the surface of a DDR film that is suitable to act as the adsorbent layer of the contactors of the present invention.

[0018] Figure 5 hereof is a representation of another embodiment of the present invention wherein the parallel channel contactor is in the form of a coated monolith for TSA applications where the adsorbent layer is coated onto the channel walls of a preformed monolith.

[0019] Figure 6 hereof is another representation of a parallel channel contactor of the present invention but in the form of an array of hollow fibers.

[0020] Figure 7 hereof is yet another representation of a parallel channel contactor of the present invention but in the form of a hollow fiber contactor for TSA applications.

[0021] Figure 8 hereof is another representation of a hollow fiber contactor for TSA as shown in Figure 7 but with the outer surfaces of the housing for the contactor rendered transparent. Dotted lines are used to indicate the edges of the outer surface.

[0022] Figure 9 hereof is a representation of an embodiment of the present invention wherein the parallel contactor is constructed from parallel laminate sheets.

[0023] Figure 10 hereof is a schematic diagram of a preferred five steps PSA/RCPSA process for treating a stream containing about 20 vol.% CO ₂ and about 80 vol.% CH ₄ .

[0024] Figure 11 hereof is a schematic diagram of a preferred five steps PSA/RCPSA process for treating a stream containing about 2 vol.% N ₂ and about 98 vol.% CH ₄ .

DETAILED DESCRIPTION OF THE INVENTION

[0025] The present invention is directed to novel adsorbents for use in swing adsorption processes for the removal of an undesirable gas selected from the group consisting of CO ₂ , N ₂ and H ₂ S from a gas mixture containing same and at least one desirable gas. The adsorbents are preferably selected from the 8-ring zeolites having an Si:Al ratio of about 1 :1 to about 1000:1, preferably from about 10:1 to about 500: 1, and more preferably from about 50:1 to about 300:1. The more preferred 8-ring zeolites are DDR, Sigma- 1 and ZSM-58. It should be noted that the term "Si/ Al ratio" as used herein means the molar ratio of silica to alumina of the zeolite structure.

_ .

[0026] The 8-ring zeolite will be present in the swing adsorption apparatus in any suitable form. For example, it can be in the form of beads that are then packed to form an adsorbent bed through which the gas to be treated is passed. Adsorbent beads, or aggregates, for swing adsorption processes are well known in the art and they can be of any suitable shape including spherical or irregular. Adsorbent aggregates are typically formed by adhering microporous zeolite crystals together with binder materials. The micropores are due to the crystalline structure of the zeolite, in this case 8-ring zeolites. Binders are typically dense materials that do not have adsorptive properties, but which are used to bind the zeolite crystals. In order to function effectively, the size of binder particles must be smaller than the size of the individual zeolite crystals.

[0027] The term "adsorbent contactor" as utilized herein includes both structured and unstructured adsorbent contactors. The preferred contactors of the present invention are a type of structured adsorbent contactor entitled herein as "parallel channel contactors". "Parallel channel contactors" are defined herein as a subset of adsorbent contactors comprising structured (engineered) adsorbents in which substantially parallel flow channels are incorporated into the adsorbent structure. These flow channels may be formed by a variety of means, many of which are described herein and in addition to the adsorbent material, the adsorbent structure may contain items such as, but not limited to, support materials, heat sink materials, void reduction components, etc., which are described more fully herein. It should also be noted that unless otherwise noted herein or by reference to specific "geometric shapes" (in which case would apply only to structured adsorbent contactors), that all preferred embodiments as described in this application, such as, but limited to, contactor voidages, separation components and efficiencies, operating conditions, preferred materials, etc., apply to both structured and unstructured adsorbent contactors of the present invention as described herein.

. .

[0028] It is preferred that the 8-ring zeolite adsorbent of the present invention be an integral part of an engineered contactor, preferably a parallel channel contactor having a plurality of flow channels so that a gas can pass from one end of the contactor, through the flow channels, and exit the opposite end of the contactor. The preferred contactors of the present invention are those suitable for use in all types of thermal swing adsorption (TSA) and all of the various types of pressure swing adsorption processes, including conventional pressure swing adsorption (PSA), and partial pressure swing or displacement purge adsorption (PPSA) technologies. These swing adsorption processes can be conducted with rapid cycles, in which case they are referred to as rapid cycle thermal swing adsorption (RCTSA), rapid cycle pressure swing adsorption (RCPSA), and rapid cycle partial pressure swing or displacement purge adsorption (RCPPSA) technologies. The term "swing adsorption processes" shall be taken to include all of these processes (i.e., TSA, PSA, PPSA, RCTSA, RCPSA, and RCPPSA) including combinations of these processes. Such processes require efficient contact of a gas mixture with the zeolite adsorbent material.

[00291 The structure of parallel channel contactors, including fixed surfaces on which the adsorbent or other active material is held, provides significant benefits over previous conventional gas separation methods, such as vessels containing adsorbent beads or extruded adsorbent particles. These swing adsorption technologies are all well known to those having ordinary skill in the art and they can be applied to remove a variety of target gases from a wide variety of gas mixtures. It is possible to significantly improve the recovery percentage of the light component of a gas mixture by use of the present invention. The light component is taken to be the species, or molecular component, or components that are not preferentially taken up by the adsorbent in the adsorption step of the process. Recovery of the light component is defined as the time averaged molar flow rate of the light component in the product

stream divided by the time averaged molar flow rate of the light component in the feedstream. Similarly, recovery of the heavy component is defined as the the time averaged molar flow rate of the heavy component in the product stream divided by the time averaged molar flow rate of the heavy component in the feedstream.

[0030] It will be understood that the term PSA, unless preceded by the term "conventional" or "rapid cycle" refers collectively to all pressure swing adsorption processes including conventional PSA, RCPSA and PPSA. In PSA processes, a gaseous mixture is conducted under pressure for a period of time over a first bed of a solid sorbent that is selective, or relatively selective, for one or more components, usually regarded as a contaminant, that is to be removed from the gaseous mixture. The components that are selectively adsorbed are referred to as the heavy component and the weakly adsorbed components that pass through the bed are referred to as the light components. It is possible to remove two or more contaminants simultaneously but for convenience, the component or components, that are to be removed by selective adsorption will be referred to in the singular and referred to as a contaminant or heavy component.

[0031] Unless otherwise noted, the term "selectivity" as used herein is based on binary (pairwise) comparison of the molar concentration of components in the feed stream and the total number of moles of these components adsorbed by the particular adsorbent during the adsorption step of the process cycle under the specific system operating conditions and feedstream composition. For a feed containing component A, component B, as well as additional components, an adsorbent that has a greater "selectivity" for component A than component B will have at the end of the adsorption step of the swing adsorption process cycle a ratio:

U _A = (total moles of A in the adsorbent) / (molar concentration of A in the feed)

_ _

that is greater than the ratio:

U _B = (total moles of B in the adsorbent) / (molar concentration of B in the feed)

Where U _A is the "Adsorption Uptake of component A" and UB is the "Adsorption Uptake of component B".

Therefore for an adsorbent having a selectivity for component A over component B that is greater than one:

Selectivity = U _A /U _B (where U _A > U _B ).

Amongst a comparison of different components in the feed, the component with the smallest ratio of the total moles picked up in the adsorbent to its molar concentration in the feed is the lightest component in the swing adsorption process. This means that the molar concentration of the lightest component in the stream coming out during the adsorption step is greater than the molar concentration of that lightest component in the feed. The adsorbent contactors of the present invention have a selectivity for a first component (e.g., component A) over a second component (e.g., component B) of at least 5, more preferably a selectivity for a first component over a second component of at least 10, and most preferably a selectivity for a first component over a second component of at least 25.

[0032] Examples of components are molecules such as molecular nitrogen, N ₂ , or compounds, such as carbon dioxide, CO ₂ , and methane, CH ₄ . In a preferred embodiment of the present invention, the adsorbent contactor has a selectivity for CO ₂ over CH ₄ of at least 5, more preferably a selectivity for CO ₂ over CH ₄ of at least 10, and most preferably a selectivity for CO ₂ over CH ₄ of at least 25. In another preferred embodiment of the present invention, the adsorbent contactor has a selectivity for N ₂ over CH ₄ of at least 5, more preferably a selectivity for N ₂ over CH ₄ of at least 10, and most preferably a

_ _n _

selectivity for N ₂ over CH ₄ of at least 25. In yet another preferred embodiment of the present invention, the adsorbent contactor has a selectivity for H ₂ S over CH ₄ of at least 5, more preferably a selectivity for H ₂ S over CH ₄ of at least 10, and most preferably a selectivity for H ₂ S over CH ₄ of at least 25.

[0033] In a preferred embodiment of the present invention, the adsorbent has a "kinetic selectivity" for two or more gas components. As used herein, the term "kinetic selectivity" is defined as the ratio of single component diffusion coefficients, D (in m ² /sec), for two different species. These single component diffusion coefficients are also known as the Stefan-Maxwell transport diffusion coefficients that are measured for a given adsorbent for a given pure gas component. Therefore, for example, the kinetic selectivity for a particular adsorbent for component A with respect to component B would be equal to D _A / D _B . The single component diffusion coefficients for a material can be determined by tests well known in the adsorptive materials art. The preferred way to measure the kinetic diffusion coefficient is with a frequency response technique described by Reyes et al. in "Frequency Modulation Methods for Diffusion and Adsorption Measurements in Porous Solids", J. Phys. Chem. B. 101, pages 614-622, 1997. In a kinetically controlled separation it is preferred that kinetic selectivity (i.e., D _A /D _B ) of the selected adsorbent for the first component (e.g., Component A) with respect to the second component (e.g., Component B) be greater than 5, more preferably greater than 20, even more preferably greater than 50.

[0034J In another preferred embodiment of the present invention, the adsorbent has an "equilibrium selectivity" for two or more gas components. As used herein, the term "equilibrium selectivity" is defined in terms of the slope of the single component uptake into the adsorbent (in μmole/g) vs. pressure (in torr) in the linear portion, or "Henry's regime", of the uptake isotherm for a given adsorbent for a given pure component. The slope of this line is called herein the Henrys constant or "equilibrium uptake slope", or "H". The "equilibrium

_ _

selectivity" is defined in terms of a binary (or pairwise) comparison of the Henrys constants of different components in the feed for a particular adsorbent. Therefore, for example, the equilibrium selectivity for a particular adsorbent for component A with respect to component B would be H _A /H _B . It is preferred that in an equilibrium controlled separation the equilibrium selectivity (i.e., H _A /H _B ) of the selected adsorbent for the first component (e.g., Component A) with respect to the second component (e.g., Component B) be greater than 5, more preferably greater than 20, even more preferably greater than 50.

[0035] In the PSA process, the gaseous mixture is passed over a first adsorption bed in a first vessel and a light component enriched product stream emerges from the bed depleted in the contaminant, or heavy component, which remains sorbed in the bed. After a predetermined time or, alternatively when a break-through of the contaminant or heavy component is observed, the flow of the gaseous mixture is switched to a second adsorption bed in a second vessel for the purification to continue. While the second bed is in adsorption service, the sorbed contaminant, or heavy component is removed from the first adsorption bed by a reduction in pressure. In some embodiments, the reduction in pressure is accompanied by a reverse flow of gas to assist in desorbing the heavy component. As the pressure in the vessels is reduced, the heavy component previously adsorbed in the bed is progressively desorbed to a heavy component enriched product stream. When desorption is complete, the sorbent bed may be purged with an inert gas stream, e.g., nitrogen or a purified stream of process gas. Purging may also be facilitated by the use of a higher temperature purge gas stream.

[0036] After the first bed has been regenerated so that it is again ready for adsorption service, the flow of the gaseous mixture is switched from the second bed to the first bed, and the second bed is regenerated. The total cycle time is the length of time from when the gaseous mixture is first conducted to the first bed in a first cycle to the time when the gaseous mixture is first conducted to the

first bed in the immediately succeeding cycle, i.e., after a single regeneration of the first bed. The use of third, fourth, fifth, etc. vessels in addition to the second vessel can serve to increase cycle time when the adsorption cycle time for the bed is shorter than the cycle times for the desorption & purging cycles for the bed.

[0037] Conventional PSA suffers from several inherent disadvantages. For example, conventional PSA units are typically more costly to build and operate and are significantly larger in size for the same amount of target gas that needs to be recovered from a target-gas containing gas stream, such as natural gas, as compared to RCPSA. Also, a conventional PSA unit will generally have cycle times in excess of one minute, typically in excess of 2 to 4 minutes due to time limitations required to allow diffusion of the components through the larger beds utilized in conventional PSA and the equipment configuration involved. In contrast, RCPSA generally has a total cycle times of less than one minute. The total cycle times of RCPSA may be less than 30 seconds, preferably less than 15 seconds, more preferably less than 10 seconds, even more preferably less than 5 seconds, and even more preferably less than 1 second. Further, the rapid cycle pressure swing adsorption units can make use of substantially different sorbents, such as, but not limited to, structured materials such as monoliths, laminates, and hollow fibers.

[0038] RCPSA can enable a significant increase in process intensification (e.g., higher operating frequencies and gas flow velocities) when compared to conventional PSA. RCPSA typically utilizes a rotary valving system to conduct the gas flow through a rotary adsorber module that contains a number of separate adsorbent bed compartments or "tubes", each of which is successively cycled through the sorption and desorption steps as the rotary module completes the cycle of operations. The rotary sorber module is normally comprised of multiple tubes held between two seal plates on either end of the rotary sorber module

_ _χ 5 _

wherein the seal plates are in contact with a stator comprised of separate manifolds wherein the inlet gas is conducted to the RCPSA tubes and processed purified product gas and the tail gas exiting the RCPSA tubes are conducted away from the rotary sorber module. By suitable arrangement of the seal plates and manifolds, a number of individual compartments or tubes may pass through the characteristic steps of the complete cycle at any given time. In contrast, with conventional PSA, the flow and pressure variations, required for the RCPSA sorption/desorption cycle, changes in a number of separate increments on the order of seconds per cycle, which smoothes out the pressure and flow rate pulsations encountered by the compression and valving machinery. In this form, the RCPSA module includes valving elements angularly spaced around the circular path taken by the rotating sorption module so that each compartment is successively passed to a gas flow path in the appropriate direction and pressure to achieve one of the incremental pressure/flow direction steps in the complete RCPSA cycle. One key advantage of the RCPSA technology is a significantly more efficient use of the adsorbent material. The quantity of adsorbent required with RCPSA technology can be only a fraction of that required for conventional PSA technology to achieve the same separation quantities and qualities. As a result, the footprint, investment, and the amount of active adsorbent required for RCPSA is typically significantly lower than that for a conventional PSA unit processing an equivalent amount of gas.

[0039] The present invention can be used in PPSA, RCPSA or hybrid PPSA or RCPPSA processes where a gas or liquid is purged through the bed to help desorb molecules. In a PPSA process desorption of the adsorbed species is accomplished by passing a gas or liquid through the contactor to desorb molecules taken up during an adsorption step. An example of a gas that may be used is steam. In hybrid PPSA processes, the desorption of molecules from the contactor is accomplished by use of a thermal or pressure swing and part of the desorption is accomplished with a purge.

- -

[0040] Improvements in the recovery of the light component are especially important for processes used to remove impurities from natural gas streams, particularly high pressure natural gas streams. It is desirable to recover the impurities, also referred to as the "heavy component(s)", and the methane-rich product, also referred to as the "light component", at as high a pressure as practical for operability in natural gas processing. As previously mentioned, the present invention can be used to obtain methane recovery of greater than about 80 vol.%, more preferably greater than about 85 vol.%, even more preferably greater than about 90 vol.%, and most preferably greater than about 95 vol.%, even when the natural gas is fed at high pressures, such as at inlet pressures greater than about 50 psig, preferably at inlet pressures greater than about 150 psig, more preferably at inlet pressures greater than about 450 psig, even more preferably at inlet pressures greater than about 600 psig and most preferably at inlet pressures greater than about 1200 psig. The present invention can be used even when the gas stream is at an exceptionally high pressure of up to about 7000 psig. The composition of natural gas streams directly from an underground field (raw natural gas) will vary from field to field. Non-limiting examples of components that comprise a raw natural gas stream include water, condensates (higher molecular weight organics), methane, ethane, propane, butane, CO ₂ , N ₂ , He, H ₂ S, Hg, and mercaptans. Water and condensates are typically removed and the condensates sent to a petroleum refinery. In order to produce a gas that can be introduced into a pipeline for sale to residential and commercial fuel markets contaminants, such as N ₂ , Hg, mercaptans, and the acid gases CO ₂ and H ₂ S must to removed to acceptable levels. The levels and impurity types vary from gas field to gas field and in some cases can comprise the majority of molecules in the produced gas. For example, it is not uncommon for some natural gas fields to contain from about 0 to 90 vol.% CO ₂ , more typically from about 10 vol.% to about 70 vol.% CO ₂ .

[0041] The present invention also provides a method to increase recovery of the light component by conditioning the temperature and pressure of gas fed to the contactor. During the adsorptive step of well designed kinetically controlled swing adsorption processes, the amount of heavy component in the micropores of the 8-ring zeolites can be approximately computed from the adsorption isotherm of the heavy component in equilibrium with its local gas phase concentration in the contactor. In well designed equilibrium controlled swing adsorption processes the amount of heavy component in the micropores can be approximately computed from the competitive adsorption isotherm of the heavy and light components in equilibrium with their local gas phase concentration in the contactor. These approximations are possible because in well designed swing adsorption processes, the contactor provides good mass transfer characteristics between the gas phase and the adsorbed phase in the micropores of the contactor. The maximum attainable loading of the heavy component in the macropores or free volume of the contactor is called q _s (units for q _s are milli- mole/m ³ of the microporous or polymeric material). At low pressures the adsorption isotherm for the heavy component usually obeys Henry's Law and the amount of heavy component adsorbed, q _Heavy* in the microporous or polymeric material is:

SW = ^K Heavy P _Hmy Qs (in milli-mole/m ³⁾

where K _He avy is the Henry's constant and Pπ _eavy is the partial pressure of the heavy component. The Henry's constant, K _Heavy depends on temperature and usually varies according to the equation:

&H

K Heavy = K _n e R T (in pascals .- ^" U )

_ j _g _

where K ₀ is a pre-exponential and δH is the heat of adsorption (in joule/mole).

[0042] To improve selectivity and recovery for either a kinetically or equilibrium controlled swing adsorption processes the inlet temperature and pressure should be chosen such that at the end of the adsorption step the loading of the heavy component in the micropores near the point at which feed is introduced to the contactor should be greater than 0.15 q _s and preferably greater than 0.3 q _s and even more preferably greater than 0.6 q _s . This requirement places a lower bound on the inlet pressure and a maximum bound on the inlet temperature. With increasing loading of the heavy component in the micropores of the adsorbent the amount of material that is selectively adsorbed in the contactor is increased and the amount of material that can be selectively released in the desorption step is increased. Increasing the loading significantly beyond this range reduces the recovery of the light component because the slope of the adsorption isotherm tends to decrease with increasing pressure. To maximize the recovery of the light component it is also preferred that near the point at which feed is introduced to the contactor the slope of the adsorption isotherm for the heavy component be large enough so that:

⁹ 1 Heavy „

~z-r > a K _Heavy q _s

° "Heavy where α = 1/50, more preferably α = 1/25, and even more preferably α = 1/8. This inequality places a maximum bound on the inlet pressure and a minimum bound on the inlet temperature. As such these requirements define a window (i.e., maxima and minima) for feed pressure and temperature in which the recovery of the light component is optimized.

[0043J This window is especially important in natural gas separations because some natural gas is usually produced at pressures ranging from 1,500 to 7,000 psi. These feed pressures are usually too high to fall within the optimum

_ _

recovery window for methane which acts as a light component in swing adsorption separation. It is also possible to reduce the feed pressure with a simple expansion nozzle, however this technique wastes energy. It is possible to access the optimum light component recovery window for most heavy component (such as CO ₂ , N ₂ , and H ₂ S) separations by preconditioning the natural gas with a turboexpander that recovers the energy from the gas expansion. Energy recovered from gas expansion can then be used for power generation or to help recompress separated acid gas components (such as CO ₂ or H ₂ S) so that they can be disposed of in underground formations. Underground formations that are suitable for disposal /sequestration of CO ₂ and H ₂ S include aquifers that have a top seal that prevents significant loss of injected acid gas components, oil reservoirs, gas reservoirs, depleted oil reservoirs and depleted gas reservoirs. Typically, the separated CO ₂ and H ₂ S has to be recompressed to pressures greater than 2,000 psi and often to pressures greater than 5,000 psi to be injected into these types of underground formations. Thus, it is important to be able to reuse energy recovered from a turboexpander for recompression. The cost of a turboexpander is also less than a gas fired turbine producing the same amount of power. As such it is economically advantageous to use a turboexpander to capture energy from gas expansion used to condition natural gas for the optimum methane recovery window. When a turboexpander is used, the energy can either be recovered with a shaft coupled electric generator or with a shaft coupled compressor. It can be advantageous to pass the gas coming out of the turboexpander through a heat exchanger before introducing it into the swing adsorption process in order to access the operating window that maximizes methane recovery. Gas coming out of a turboexpander can be too cold to be in the optimum recovery window because the work is recovered in an isentropic expansion. Typically, a heat exchanger will be run so that the gas temperature is increased before entering a swing adsorption process. These considerations are especially important when the swing adsorption is a PSA or RCPSA process.

[0044] The kinetic selectivity of 8-ring zeolites of the present invention allows CO ₂ to be rapidly transmitted into zeolite crystals while hindering the transport of methane so that it is possible to selectively separate CO ₂ from a mixture Of CO ₂ and methane. For the removal Of CO ₂ , from natural gas, this specific class of 8-ring zeolite materials has a Si/Al ratio from about 1 : 1 to 1000: 1, preferably from about 10:1 to 500:1, and more preferably from about 50: 1 to 300: 1. This preferred class of 8-ring zeolites that are suitable for use herein allow CO ₂ to access the internal pore structure through 8-ring windows in a manner such that the ratio of single component diffusion coefficients Of CO ₂ and methane (i.e., D _CO2 /D _CH _ _I ) is greater than 10, preferably greater than about 50, and more preferably greater than about 100 and even more preferably greater than 200. Single component diffusion coefficients are taken to be transport diffusion coefficients measured for a pure gas in the Henry's law regime of the adsorption isotherm. The loading of molecules in the zeolite is low in the Henry's law regime and in this regime the Fickian and Stephan-Maxwell diffusion coefficients are nearly equal. The diffusivity of a porous crystalline material for a particular sorbate is conveniently measured in terms of its diffusion time constant, D/r ² , wherein D is the Fickian diffusion coefficient (m ² /sec) and the value "r" is the radius of the crystallites (m) characterizing the diffusion distance. In situations where the crystals are not of uniform size and geometry, "r" represents a mean radius representative of their corresponding distributions. One way to measure the time constant and diffusion coefficient is from analysis of standard adsorption kinetics (i.e., gravimetric uptake) using methods described by J. Crank in "The Mathematics of Diffusion", 2nd Ed., Oxford University Press, Great Britain, 1975. Another way to measure the time constant and diffusion coefficient is from analysis of zero length chromatography data using methods described by D. M. Ruthven in "Principles of Adsorption and Adsorption Processes", John Wiley, NY (1984) and by J. Karger and D. M. Ruthven in "Diffusion in Zeolites and Other Microporous

_ _

Solids", John Wiley, NY (1992), which is incorporated herein by reference. A preferred way to measure the time constant and diffusion coefficient is with a frequency response technique described by Reyes et al. in "Frequency Modulation Methods for Diffusion and Adsorption Measurements in Porous Solids", J. Phys. Chem. B. 101, pages 614-622, 1997, which is also incorporated herein by reference. At temperatures below 100 ⁰ C the single component diffusion coefficient of methane is found to be more than a hundred fold less than that of CO ₂ . From the measured activation energies of the diffusion coefficients, at temperatures up to about 300 ⁰ C, the diffusion coefficient of methane is computed to be more than five fold less than that of CO ₂ . An example of a preferred 8-ring zeolite that can be used in the practice of the present invention is DDR. Isotypic framework structures of DDR include Sigma- 1 and ZSM-58, which are also suitable for use herein. Resistance to fouling in swing adsorption processes that remove CO ₂ from natural gas is another advantage offered by this class of 8-ring zeolite materials.

[0045] In many instances, nitrogen also has to be removed from natural gas or gas associated with the production of oil. In some cases this is because of the high nitrogen levels (>2 vol%) in the produced gas, and in other cases nitrogen removal is needed in order to liquefy natural gas. It may also be advantageous to separate nitrogen from flash gas that occurs in LNG production so that the methane and hydrocarbon products can be used as fuel. Another application is the purification of gas from LNG boil-off so that the methane and hydrocarbon products can be recovered or used as fuel. When recovered, it may be advantageous to re-liquefy the methane and hydrocarbon and returned them back to the LNG cargo. In all of these applications it is desirable to selectively adsorb the nitrogen to obtain high recovery of a purified methane product from nitrogen containing gas. There have been very few molecular sieve sorbents with significant equilibrium or kinetic selectivity for nitrogen separation from methane. For N ₂ separation from natural gas it is also preferred to formulate the

. ^ _

adsorbent with a class of 8-ring zeolite materials that has a kinetic selectivity. The kinetic selectivity of this class of 8-ring materials allows N ₂ to be rapidly transmitted into zeolite crystals while hindering the transport of methane so that it is possible to selectively separate N ₂ from a mixture of N ₂ and methane. For the removal OfN ₂ , from natural gas, this specific class of 8-ring zeolite materials has a Si/Al ratio from about 1 : 1 to 1000: 1, preferably from about 10: 1 to 500:1, and more preferably from about 50:1 to 300:1. This preferred class of 8-ring zeolites that are suitable for use herein allow N ₂ to access the internal pore structure through 8-ring windows in a manner such that the ratio of single component diffusion coefficients OfN ₂ and methane (i.e., D^/D _C I _M ) is greater than 5, preferably greater than about 20, and more preferably greater than about 50 and even more preferably greater than 100. Non-limiting examples of 8-ring zeolites suitable for use in swing adsorption processes to remove N ₂ from natural gas include DDR framework zeolites. Preferred DDR zeolites are Sigma- 1 and ZSM-58. Resistance to fouling in swing adsorption processes the remove N ₂ from natural gas is another advantage offered by this class of 8-ring zeolite materials.

[0046] In other instances, it is also desirable to remove H ₂ S from natural gas which can contain from about 0.001 vol% H ₂ S to about 70 vol% H ₂ S. In this case, it can be advantageous to formulate the adsorbent with stannosilicates as well as the aforementioned class of 8-ring zeolites that has kinetic selectivity. The kinetic selectivity of this class of 8-ring materials allows H ₂ S to be rapidly transmitted into zeolite crystals while hindering the transport of methane so that it is possible to selectively separate H ₂ S from a mixture OfH ₂ S and methane. For the removal OfH ₂ S, from natural gas, this specific class of 8-ring zeolite materials has a Si/Al ratio from about 1:1 to 1000:1, preferably from about 10:1 to 500:1, and more preferably from about 50:1 to 300:1. This preferred class of 8-ring zeolites that are suitable for use herein allow H ₂ S to access the internal pore structure through 8-ring windows in a manner such that the ratio of single

component diffusion coefficients Of H ₂ S and methane (i.e., is greater than 5, preferably greater than about 20, and more preferably greater than about 50 and even more preferably greater than 100. Resistance to fouling in swing adsorption processes the remove H ₂ S from natural gas is another advantage offered by this class of 8-ring zeolite materials. In some applications the H ₂ S has to be removed to the ppm or sub-ppm levels. To achieve such extensive removal Of H ₂ S it can be advantageous to use a PPSA or RCPPSA process.

[0047] The 8-ring adsorbents of the present invention can be used in both equilibrium and kinetically controlled swing adsorption processes to improve the recovery of CO ₂ , N ₂ or H ₂ S from a gas mixture. In one embodiment of the present invention, the walls of the open flow parallel channels are comprised of the adsorbent.

[0048] The adsorbent contactors of the present invention may optionally contain a thermal mass (heat transfer) material to help control heating and cooling of the adsorbent of the contactor during both the adsorption step and desorption step of a pressure swing adsorption process. Heating during adsorption is caused by the heat of adsorption of molecules entering the adsorbent. The optional thermal mass material also helps control cooling of the contactor during the desorption step. The thermal mass can be incorporated into the flow channels of ithe contactor, incorporated into the adsorbent itself, or incorporated as part of the wall of the flow channels. When it is incorporated into the adsorbent, it can be a solid material distributed throughout the adsorbent layer or it can be included as a layer within the adsorbent. When it is incorporated as part of the wall of the flow channel, the adsorbent is deposited or formed onto the wall. Any suitable material can be used as the thermal mass material in the practice of the present invention. Non-limiting examples of such materials include metals, ceramics, and polymers. Non-limiting examples of preferred metals include steel, copper, and aluminum alloys. Non-limiting

_ _

examples of preferred ceramics include silica, alumina, and zirconia. An example of a preferred polymer that can be used in the practice of the present invention is polyimide. Depending upon the degree to which the temperature rise is to be limited during the adsorption step, the amount of thermal mass material used can range from about 0 to about 25 times the mass of the microporous adsorbent of the contactor.

[0049] A preferred range for the amount of thermal mass in the contactor is from about 0 to 5 times the mass of the microporous adsorbent of the contactor. A more preferred range for the amount of thermal mass material will be from about 0 to 2 times the mass of the microporous adsorbent material, most preferably from about 0 to 1 times the mass of the microporous material of the contactor. In a preferred embodiment, an effective amount of thermal mass is incorporated into the contactor. The effective amount of thermal mass is an amount sufficient to maintain the thermal rise of the adsorbent during the adsorption step to less than about 100 ⁰ C. In a preferred embodiment, the amount of thermal mass incorporated into the contactor is an amount sufficient to maintain the thermal rise of the adsorbent during the adsorption step to less than about 50 ⁰ C, and more preferably to less than about 10 ⁰ C.

[0050] The channels, also sometimes referred to as "flow channels" or "gas flow channels" are paths in the contactor that allow gas flow through. Generally, flow channels provide for relatively low fluid resistance coupled with relatively high surface area. Flow channel length should be sufficient to provide the desired mass transfer zone, which is, at least, a function of the fluid velocity, and the ratio of surface area to channel volume. The channels are preferably configured to minimize pressure drop in the channels. In many embodiments, a fluid flow fraction entering a channel at the first end of the contactor does not communicate with any other fluid fraction entering another channel at the first end until the fractions recombine after exiting at the second end. It is important that there be channel uniformity to ensure that substantially all of the channels

are being fully utilized, and that the mass transfer zone is substantially equally contained. Both productivity and gas purity will suffer if there is excessive channel inconsistency. If one flow channel is larger than an adjacent flow channel, premature product break through may occur, which leads to a reduction in the purity of the product gas to unacceptable purity levels. Moreover, devices operating at cycle frequencies greater than about 50 cycles per minute (cpm) require greater flow channel uniformity and less pressure drop than those operating at lower cycles per minute. Further, if too much pressure drop occurs across the bed, then higher cycle frequencies, such as on the order of greater than 100 cpm, are not readily achieved.

[0051] The dimensions and geometric shapes of the parallel channel contactors of the present invention can be any dimension or geometric shape that is suitable for use in swing adsorption process equipment. Non-limiting examples of geometric shapes include various shaped monoliths having a plurality of substantially parallel channels extending from one end of the monolith to the other; a plurality of tubular members; stacked layers of adsorbent sheets with and without spacers between each sheet; multi-layered spiral rolls, bundles of hollow fibers, as well as bundles of substantially parallel solid fibers. The adsorbent can be coated onto these geometric shapes or the shapes can, in many instances, be formed directly from the adsorbent material with a suitable binder. An example of a geometric shape formed directly from the adsorbent plus binder would be the extrusion of a zeolite/polymer composite into a monolith. Another example of a geometric shape formed directly from the adsorbent would be extruded or spun hollow fibers made from a zeolite/polymer composite. An example of a geometric shape that is coated with the adsorbent would be a thin flat steel sheet that is coated with an 8-ring zeolite film. The directly formed or coated adsorbent layer can be itself structured into multiple layers or the same or different adsorbent materials. Multi-layered adsorbent

- ZO -

sheet structures are taught in United States Patent Application Publication No. 2006/0169142, which is incorporated herein by reference.

[0052] The dimensions of the flow channels can be computed from considerations of pressure drop along the flow channel. It is preferred that the flow channels have a channel gap from about 5 to about 1,000 microns, preferably from about 50 to about 250 microns. As utilized herein, the "channel gap" of a flow channel is defined as the length of a line across the minimum dimension of the flow channel as viewed orthogonal to the flow path. For instance, if the flow channel is circular in cross-section, then the channel gap is the internal diameter of the circle. However, if the channel gap is rectangular in cross-section, the flow gap is the distance of a line perpendicular to and connecting the two longest sides of the rectangular (i.e., the length of the smallest side of the rectangle). It should also be noted that the flow channels can be of any cross-sectional configuration. Preferred embodiments are wherein the flow channel cross-sectional configuration is either circular, rectangular or square. However, any geometric cross-sectional configuration may be used, such as but not limited to, ellipses, ovals, triangles, or various polygonal shapes. In other preferred embodiments, the ratio of the adsorbent volume to flow channel volume in the adsorbent contactor is from about 0.5: 1 to about 100: 1, and more preferably from about 1 :1 to about 50:1.

[0053] In some RCPSA applications, the flow channels are formed when adsorbent sheets are laminated together. Typically, adsorbent laminates for RCPSA applications have flow channel lengths from about 0.5 centimeter to about 10 meter, more typically flow channel lengths from about 10 cm to about 1 meter and a channel gap of about 50 to about 250 microns. The channels may contain a spacer or mesh that acts as a spacer. For laminated adsorbents, spacers can be used which are structures or material, that define a separation between

_ _2? _

adsorbent laminates. Non-limiting examples of the type of spacers that can be used in the present invention are those comprised of dimensionally accurate: plastic, metal, glass, or carbon mesh; plastic film or metal foil; plastic, metal, glass, ceramic, or carbon fibers and threads; ceramic pillars; plastic, glass, ceramic, or metal spheres, or disks; or combinations thereof. Adsorbent laminates have been used in devices operating at PSA cycle frequencies up to at least about 150 cpm. The flow channel length may be correlated with cycle speed. At lower cycle speeds, such as from about 20 to about 40 cpm, the flow channel length can be as long as or longer than one meter, even up to about 10 meters. For cycle speeds greater than 40 cpm, the flow channel length typically is decreased, and may vary from about 10 cm to about 1 meter. Longer flow channel lengths can be used for slower cycle PSA processes. Rapid cycle TSA processes tend to be slower than rapid cycle PSA processes and as such longer flow channel lengths can also be used with TSA processes.

[0054] The overall adsorption rate of the swing adsorption processes is characterized by the mass transfer rate from the flow channel into the adsorbent. It is desirable to have the mass transfer rate of the species being removed (i.e., the heavy component) high enough so that most of the volume of the adsorbent is utilized in the process. Since the adsorbent selectively removes the heavy component from the gas stream, inefficient use of the adsorbent layer can lower recovery of the light component and/or decrease the purity of the light product stream. One way to improve the efficiency of the adsorbent layer is to have an adsorbent of uniform thickness where the thickness of the adsorbent layer is set by the mass transfer coefficients of the heavy component and the time of the adsorption and desorption steps of the process. The thickness uniformity can be assessed from measurements of the thickness of the adsorbent or from the way in which it is fabricated. It is preferred that the uniformity of the adsorbent be such that the standard deviation of its thickness is less than about 25% of the average thickness. More preferably, the standard deviation of the thickness of the

_ _2g _

adsorbent thickness is less than about 15 % of the average thickness. It is even more preferred that the standard deviation of the adsorbent thickness be less than about 5% of the average thickness.

[0055J Calculation of these mass transfer rate constants is well known to those having ordinary skill in the art and may also be derived by those having ordinary skill in the art from standard testing data. D. M. Ruthven & C. Thaeron, Performance of a Parallel Passage Absorbent Contactor, Separation and Purification Technology 12 (1997) 43-60, which is incorporated herein by reference, clarifies many aspects of how the mass transfer is affected by the thickness of the adsorbent, channel gap and the cycle time of the process. Also, U.S. Patent No. 6,607,584 to Moreau et al., which is also incorporated by reference, describes the details for calculating these transfer rates and associated coefficients for a given adsorbent and the test standard compositions used for conventional PSA.

[0056] A figure of merit for the mass transfer through the adsorbent layer is a time constant, τ _a , for transport of the heavy component computed at each point in the adsorbent. For a planar adsorbent sheet with thickness in the x direction, and the y and z directions being in the plane of the sheet, the time constant τ _a of the heavy component is:

τ _a [x,y,z] = Minimum[L _path ² / D _path ] (in sec)

where D _path is the average transport diffusion coefficient of the heavy component along a path from the feed channel to the point (x,y,z) and L _path is the distance along the path. There are many possible trajectories or paths from the feed channel to each point (x,y,z) in the adsorbent. The time constant is the minimum of the possible time constants (L _path ² / D _path ) along all possible paths from the feed channel to the (x,y,z) point in the adsorbent. If there is a solid

material in the adsorbent (such as that which may be included for heat management) there will be no transport within it and (x,y,z) points within it are not included in the computation. The transport diffusion coefficient of each species is taken to be the single component Stefan-Maxwell diffusion coefficient for each species. The average transport diffusion coefficient along the path, D _path j is the linearly averaged diffusion coefficient along the path. A linear averaging is sufficient to provide a diffusion coefficient characterizing the path. When the heavy component has many species the diffusion coefficient, D _path , is also compositionally averaged. The diffusion coefficient depends on temperature and it may depend on pressure and temperature. To the extent that the diffusion coefficient changes, it must be averaged for the temperature and pressure changes occurring during a cycle. For an adsorbent to be efficient, the averaged thickness of the adsorbent layer preferably is chosen such that the time constant for at least half the points (or volume) in the adsorbent that is not a dense solid is less than the cycle time of the process. More preferably, the average thickness of the adsorbent layer is chosen such that the time constant for at least 75 % of the points (or volume) in the adsorbent that is not a dense solid is less than the cycle time of the process. Even more preferably the average thickness of the adsorbent layer is chosen such that the time constant for at least 75 % of the points (or volume) in the adsorbent that is not a dense solid is less than about 25 % of the cycle time of the process.

[0057] The contactors of the present invention can better be understood with reference to the Figures hereof. Figure 1 hereof is a representation of a parallel channel contactor of the present invention in the form of a monolith formed directly from a microporous adsorbent plus binder and containing a plurality of parallel flow channels. A wide variety of monolith shapes can be formed directly by extrusion processes. An example of a cylindrical monolith 1 is shown schematically in Figure 1 hereof. The cylindrical monolith 1 contains a plurality of parallel flow channels 3. These flow channels 3 can have channel

gaps from about 5 to about 1,000 microns, preferably from about 50 to about 250 microns, as long as all channels of a given contactor have substantially the same size channel gap. The channels can be formed having a variety of shapes including, but not limited to, round, square, triangular, and hexagonal. The space between the channels is occupied by the adsorbent 5. As shown the channels 3 occupy about 25% of the volume of the monolith and the adsorbent 5 occupies about 75% of the volume of the monolith. The adsorbent 5 can occupy from about 50% to about 98% of the volume of the monolith. The effective thickness of the adsorbent can be defined from the volume fractions occupied by the adsorbent 5 and channel structure as:

_» . _, . . _ _ . , , 1 ~, , _ . Volume Fraction Of Adsorbent

Effective Thickness Of Adsorbent = — Channel Diameter

2 Volume Fraction Of Channels

[0058] For the monolith of Figure 1 hereof the effective thickness of the adsorbent will be about 1.5 times the diameter of the feed channel. When the channel gap is in a range from about 50 to about 250 microns it is preferred that the thickness of the adsorbent layer, in the case wherein the entire contactor is not comprised of the adsorbent, be in a range from about 25 to about 2,500 microns. For a 50 micron channel gap, the preferred range of thickness for the adsorbent layer is from about 25 to about 300 microns, more preferred range from about 50 to about 250 microns. Figure 2 is a cross-sectional view along the longitudinal axis showing feed channels 3 extending through the length of the monolith with the walls of the flow channels formed entirely from adsorbent 5 plus binder.

[0059] When the monolith is used in a gas separation process that relies on a kinetic separation (predominantly diffusion controlled) it is advantageous for the microporous adsorbent particles to be substantially the same size. It is preferred that the standard deviation of the volume of the individual microporous

_ _

adsorbent particles be less than 100 % of the average particle volume for kinetically controlled processes. In a more preferred embodiment the standard deviation of the volume of the individual microporous adsorbent particles is less than 50% of the average particle volume. The particle size distribution for zeolite adsorbents can be controlled by the method used to synthesize the particles. It is also possible to separate presynthesized microporous adsorbent particles by size using methods such as a gravitational settling column. It may also be advantageous to use uniformly sized microporous adsorbent particles in equilibrium controlled separations.

[0060] There are several ways that monoliths can be formed directly from a structured microporous adsorbent. For example, when the microporous adsorbent is a zeolite, the monolith can be prepared by extruding an aqueous mixture containing effective amounts of a solid binder, zeolite adsorbent, solid heat control particles, and polymer. The solid binder can be colloidal sized silica or alumina that is used to bind the zeolite and solid heat control particles together. The effective amount of solid binder will typically range from about 0.5 to about 50% of the volume of the zeolite and solid heat control particles used in the mixture. If desired, silica binder materials can be converted in a post processing step to zeolites using hydrothermal synthesis techniques and, as such, they are not always present in a finished monolith. A polymer is optionally added to the mixture for rheology control and to give green extrudate strength. The extruded monolith is cured by firing it in a kiln where the water evaporates and the polymer burns away, thereby resulting in a monolith of desired composition.

[0061] Another method by which a monolith can be formed directly from the 8-ring zeolite is by extruding a polymer and 8-ring zeolite. Non-limiting examples of polymers suitable for the extrusion process include epoxies and

_ ^ _

thermoplastics, and curable polymers such as silicone rubbers that can be extruded without an added solvent.

[0062] Figure 3 hereof is a representation of a parallel channel contactor 101 of the present invention in the form of a coated monolith where an adsorbent layer is coated onto the walls of the flow channels of a preformed monolith. For the parallel channel contactors of this Figure, an extrusion process is used to form a monolith from a suitable non-adsorbent solid material, preferably a metal such as steel, a ceramic such as cordurite, or a carbon material. By the term "non-adsorbent solid material" we mean a solid material that is not to be used as the selective adsorbent for the parallel channel contactor. An effective amount and thickness of a ceramic or metallic glaze, or sol gel coating, 119 is preferably applied to effectively seal the channel walls of the monolith. Such glazes can be applied by slurry coating the channel walls, by any suitable conventional means, followed by firing the monolith in a kiln.

[0063] Another approach is to apply a sol gel to the channel walls followed by firing under conditions that density the coating. It is also possible to use vacuum and pressure impregnation techniques to apply the glaze or sol gel to the channel walls. In such a case, the glaze or sol gel will penetrate into the pore structure of the monolith 117. In all cases, the glaze seals the wall of the channel such that gas flowing thorough the channel is not readily transmitted into the body of the monolith. An adsorbent layer 105 is then uniformly applied onto the sealed walls of the channels. The adsorbent layer 105 reduces the opening, or bore, of the channels, thus the flow channel 103 used in swing adsorption processes is the open channel left inside of the coating. These flow channels 103 can have channel gaps as previously defined. The adsorbent layer 105 can be applied as a coating, or layer, on the walls of the flow channels by any suitable method. Non-limiting examples of such methods include fluid phase coating techniques, such as slurry coating, slip coating, hydrothermal film formation,

hydrothermal coating conversion, and hydrothermal growth. When non- hydrothermal coating techniques are used, the coating solutions should include at least the microporous adsorbent or polymeric particles, a viscosifying agent such as polyvinyl alcohol, heat transfer (thermal mass) solids, and optionally a binder. The heat transfer solid may not be needed because the body of the monolith 101 can act to as its own heat transfer solid by storing and releasing heat in the different steps of the separation process cycle. In such a case, the heat diffuses through the adsorbent layer 105 and into the body of the monolith 101. If a viscosifying agent, such as polyvinyl alcohol, is used it usually burns away when the coating is cured in a kiln. It can be advantageous to employ a binder such as colloidal silica or alumina to increase the mechanical strength of the fired coating.

[0064] If a hydrothermal film formation method is employed, the coating techniques used can be very similar to the way in which zeolite membranes are prepared. An example of a method for growing a zeolite layer is taught in US Patent No. 7,049,259, which is incorporated herein by reference. Examples of crack and grain boundaries in as-grown zeolite films are shown in high resolution scanning electron micrograph Figure 4 hereof. The zeolite film of Figure 4 is comprised of Sigma- 1 zeolite which has a framework structure that is isotypic with DDR. The film is about 25 micrometers thick with cracks 151 that are about 100 to about 300 angstrom wide, which cracks are readily visible on the surface of the film. Besides the cracks and gaps there are grain boundaries between crystals. These grain boundaries can connect to the crack structure and aid in transport of molecules to the zeolite crystals that are deeper in the film.

[0065] Figure 5 hereof is a representation of a parallel channel contactor of the present invention in the form of a coated monolith 201 for TSA applications where the adsorbent layer is coated onto the channel of a preformed monolith comprised of non-adsorbent material. When TSA or RCTSA processes are

_ _

performed the contactor will preferably have paths, or separate channels, that can be used to heat and cool the adsorbent. For TSA or RCTSA processes, the parallel channel contactor can be configured in a configuration similar to a shell and tube heat exchanger with the adsorbent coated on the tube walls of the heat exchanger. In this Figure, an extrusion process is used to form a monolith from a suitable non-adsorbent material including a metal such as steel, or a ceramic such as cordierite, or a carbon. A ceramic or metallic glaze or sol gel coating 219 is applied to seal the channel walls of the monolith. As previously mentioned, such glazes can be applied by slurry coating the channel walls followed by curing by firing. A sol gel can also be applied to the channel walls and then fired under conditions that densify the coating. As previously mentioned, it is also possible to use vacuum and pressure impregnation techniques to apply the glaze or sol gel. In this case the glaze or sol gel will penetrate into the pore structure of the monolith 217. In all cases the glaze seals the wall of the channel such that gas flowing thorough the channel is not readily transmitted into the body of the monolith. It may also be desirable to impregnate the pore structure of the monolith 217 with a solid material before the channel walls are sealed. Alternate rows of channels are sealed at their ends 215 in order to provide for TSA operation. At the opposite end of the monolith these same rows of channels are also sealed. Slots (223 and 225) are cut through the monolith at both ends to provide flow access to the sealed rows of channels 215. Sealing surfaces 219 are provided at both ends of the monolith as well as in the middle of the monolith 221. In operation, the monolith will be mounted in a module in a manner that seals the ends of the channels as well as the middle of the monolith. Any suitable technology can be used to seal the ends of the channels including metallic welds, gasketing with materials such as rubbers or carbons, and the use of adhesives such as inorganic cements and epoxies. The module is configured so that a heating/or cooling fluid can be flowed through the channels sealed at the ends 215 by introducing it through the slots 223 and removing it through slots 225. The heating/or cooling fluid will undergo heat

_ _

exchange with fluid flowing through the channels that are open at the end of the module. These modifications to the monolith convert it into a heat exchanger. It will be understood that there are various other ways in which heat exchangers can be produced or configured. Non-limiting examples of such other ways include shell and tube heat exchangers, fiber film heat exchangers and printed circuit heat exchangers, all of which are well known in the art.

[0066] Feed channels 203 can have channel gaps from about 5 to about 1,000 microns, preferably from about 50 to about 250 microns. When the feed channel gap 203 is in a range from 50 to about 250 microns it is preferred that the thickness of the adsorbent layer 205 be in a range form about 25 to about 2,500 microns. For a 50 micron feed channel gap 203 the preferred range of thickness for the adsorbent layer is from 25 to 300 microns and a more preferred range is from 50 to 250 microns. The techniques previously discussed above can be used to coat the adsorbent layer into the monolith.

[0067] Figure 6 hereof is a schematic of a parallel channel contactor of the present invention in the form of a substantially parallel array of hollow fibers embedded in a matrix material 331. A wide variety of hollow fibers can be formed directly using conventional spinning and extrusion processes. The contactor of Figure 6 is formed from an array of hollow fibers 301. The bores 303 of the fibers are used as flow channels. These flow channels 303 can also have channel gaps from about 5 to about 1,000 microns, preferably from about 50 to about 250 microns as previously mentioned. Also as previously mentioned, the walls of the fibers contain an adsorbent layer 305. When the feed channel gap 303 is in a range from 50 to about 250 microns it is preferred that the thickness of the adsorbent layer 305 be in a range form about 25 to 2,500 microns.

- JO -

[0068] Various methods known in the art can be used to produce the adsorbent layer 305 in the fiber. For example, the hollow polymer fibers. Hollow fibers that are composites of polymers and adsorbents can be formed in both spinning and extrusion processes. These processes often form the fiber from a dope containing the polymer, adsorbent particles, and often a solvent. In some cases, the surface of the adsorbent particle is functionalized to promote adhesion between the polymer matrix and the adsorbent particle.

[0069] It is also possible to produce hollow fibers of 8-ring zeolites by extrusion. In these processes, the zeolite is mixed with a polymer or an oligomeric viscosifying agent, such as a lower molecular weight polyvinyl alcohol. Optionally, a solvent such as water, alcohol, or liquid hydrocarbon can be added to the dope. It is also optional to use a binder material, such as colloidal silica or colloidal alumina that can be added to this dope. Solid particles, such as alumina or aluminum can also be added to the dope. The dope is then extruded and from the green state the final ceramic body is produced. This fiber, in the green state, can then be placed into a kiln and fired to form the final fiber comprised of zeolite, and optionally binder and solid thermal mass particles. Alternatively, the fiber in the green state can be placed in a hydrothermal synthesis reactor to produce a final fiber comprised of zeolite, and optionally binder and solid particles. Another method to produce a zeolite fiber is by hydrothermally growing a zeolite coating on a solid polymer fiber that burns away during the calcinations step.

[0070] The fibers can be formed into a substantially parallel array to form a contactor of the present invention. One method to do this is with an embedding or potting process that surrounds the entire length of the fibers with a matrix material 325. To visualize the array in Figure 6 the end of the matrix material 351 has been rendered transparent along with the face 321 of the embedded hollow fiber bundle. In many instances, it can be advantageous to coat the

exterior of the fiber with a material that acts as a diffusion barrier 315. Non- limiting examples of materials that can act as diffusion barriers include sputter deposited metal and ceramic films, evaporated metal and ceramic films, metal and ceramic films formed by chemical vapor deposition, coated composites of polymers and solids (such as clays) and coatings of polymers that have low diffusion coefficients. To act as a diffusion barrier, the effective diffusion coefficient of the coating should be less than about 1/10 the average diffusion coefficient in the adsorbent layer and preferably less than about 1/1000 the average diffusion coefficient in the adsorbent layer. When a diffusion barrier is used, the gas in the feed channel is effectively contained in the feed channel and adsorbent layer. This can eliminate the need for a supporting matrix around the fibers, thus lowering the mass of the contactor, and in some cases allowing for the cycle time in the process to be decreased (i.e., rapid cycle operation).

[0071] Another fabrication method suitable for use herein is to coat the adsorbent inside the prefabricated fiber, such as a hollow glass fiber, hollow silica fiber or hollow polymer fiber. Coating methods previously described can be used to form an adsorbent layer inside of a prefabricated fiber. When the prefabricated fiber is made from glass, or silica, the final product has a built in diffusion barrier 315.

[0072] When there is no diffusion barrier on the fiber it is advantageous to space the fibers closely together with the distance between adjacent fibers less than about 5 fiber diameters, preferably less than about 1.5 fiber diameters. When there is a diffusion barrier on the outer surface of the fibers, it can be advantageous to embed only the ends 351 and 353 of the fiber bundle in the matrix material. In this case, the matrix material only has to support the fibers and not have substantial gas flow through the material. It can be composed of polymer, metal or ceramic or combinations thereof. It is preferred that the matrix be nonporous and requirements for having an adsorbent in the matrix

_ ₃ g .

material can be relaxed or eliminated. Requirements for spacing between fibers can be less critical than when the entire length of the fiber is potted or embedded. The matrix material can be applied selectively to the ends of the fiber bundles by any suitable method known in the art. Non-limiting examples of such methods include potting, embedding, casting, electroplating, or electroless plating processes. To avoid plugging the end of the fibers the end of the fibers can be filled with a material that can be readily removed after the matrix is applied. Non-limiting examples of materials that can be readily removed include polymers, metals, salts and inorganics that can be selectively dissolved or etched away after the matrix material has been applied. Grinding, machining and polishing methods can also be used to open the ends of the fibers. Other methods to pot or embed the ends of the fibers are similar to those used to form hollow fiber membrane modules. When the ends of the fiber bundle are potted with a matrix material it is advantageous to place the contactor into an operational PSA, RCPSA, RCPPSA or PPSA module in a manner such that most of the feed gas flows through the bore of the fiber. One way to ensure that the flow goes through the bore of the fiber is to place a fiberous packing, or inram, between the matrix material at the ends 351 and 353 and the interior of the PSA, RCPSA, RCPPSA or PPSA module. Another way is to bond the ends of the contactor to the interior of the swing adsorption module.

[0073] Figures 7 and 8 hereof are representations of a parallel channel contactor of the present invention in the form of a hollow fiber contactor for a TSA process where the adsorbent layer 405 comprises part of the wall of the fiber with a center feed channel 403. In Figure 10, the outer surfaces of the housing for the contactor 401 are rendered transparent with only dotted lines indicating the edges of the outer surface. The hollow fibers used in this example have a diffusion barrier 415 on their exterior surface and can be manufactured using techniques described for Figure 4 hereof. The ends of the fiber bundle are potted or embedded in a matrix material 417. This potted array is then sealed

_ _

into a tubular housing 401. Sealing surfaces 419 are provided at the ends of the tubular housing 401. A sealing surface 421 is also provided in the middle of the housing. Slots 423 and 425 are cut through the wall near the ends of the tubular housing to allow for the flow of heating and cooling fluids.

[0074] In operation, the tubular housing is mounted in a TSA or RCTSA module in a manner that seals the ends of the channels as well as the middle of the monolith. Any suitable sealing technology can be used. Non-limiting examples of sealing technologies that can be used in the practice of the present invention include metallic welds, gasketing with materials such as rubbers or carbons, and the use of adhesives such as inorganic cements or epoxies. The module is configured so that a heating or cooling fluid can be flowed inside the hollow tubular housing 401 by introducing it though slots 423 and removing it through slots 425. The heating and cooling fluid will undergo heat exchange with fluid flowing through the hollow fibers which are open at the end of the module. With these sealing arrangements, the tubular housing 401 containing the parallel array of hollow fibers becomes a heat exchanger suitable for use in TSA processes.

[0075] Figure 9 hereof is an exploded view of a parallel channel contactor of the present invention in which the parallel channels are formed from laminated sheets containing adsorbent material. Laminates, laminates of sheets, or laminates of corrugated sheets can be used in PSA RCPSA, PPSA or RCPPSA processes. Laminates of sheets are known in the art and are disclosed in US Patent Applications US20060169142 Al and US7094275 B2 which are incorporated herein by reference. When the adsorbent is coated onto a geometric structure or components of a geometric structure that are laminated together, the adsorbent can be applied using any suitable liquid phase coating techniques. Non-limiting examples of liquid phase coating techniques that can be used in the practice of the present invention include slurry coating, dip

coating, slip coating, spin coating, hydrothermal film formation and hydrothermal growth. When the geometric structure is formed from a laminate, the laminate can be formed from any material to which the adsorbent of the present invention can be coated. The coating can be done before or after the material is laminated. In all these cases the adsorbent is coated onto a material that is used for the geometric shape of the contactor. Non-limiting examples of such materials include glass fibers, milled glass fiber, glass fiber cloth, fiber glass, fiber glass scrim, ceramic fibers, metallic woven wire mesh, expanded metal, embossed metal, surface-treated materials, including surface-treated metals, metal foil, metal mesh, carbon-fiber, cellulosic materials, polymeric materials, hollow fibers, metal foils, heat exchange surfaces, and combinations of these materials. Coated supports typically have two major opposing surfaces, and one or both of these surfaces can be coated with the adsorbent material. When the coated support is comprised of hollow fibers, the coating extends around the circumference of the fibers. Further support sheets may be individual, presized sheets, or they may be made of a continuous sheet of material. The thickness of the substrate, plus applied adsorbent or other materials (such as desiccant, catalyst, etc.), typically ranges from about 10 micrometers to about 2000 micrometers, more typically from about 150 micrometers to about 300 micrometers.

[0076] Metallic mesh supports can provide desirable thermal properties of high heat capacity and conductivity which "isothermalize" a PSA, RCPSA, PPSA or RCPPSA cycle to reduce temperature variations that degrade the process when conducted under more adiabatic conditions. Also, metal foils are manufactured with highly accurate thickness dimensional control. The metal foil may be composed of, without limitation, aluminum, steel, nickel, stainless steel or alloys thereof. Hence there is a need for a method to coat metal foils with a thin adsorbent layer of accurately controlled thickness, with necessary good adhesion. One method for doing this is by hydrothermal synthesis.

Coating procedures used can be very similar to the way in which zeolite membranes are prepared as discussed above. Another method of coating a metal foil is with thick film coating is slip casting, or doctor blading. An aqueous slurry of prefabricated zeolite particles, binder (for example colloidal silica or alumina), viscosifying agent such as a polymer like polyvinyl alcohol is cast for example onto a metal foil and fired to remove the polymer and cure the binder and zeolite. The product, after firing, is then a bound zeolite film on a metal foil typically containing about 30 to about 40 volume % voids. To make a suitable adsorbent layer, the voids are filled in a subsequent step by coating the bound zeolite film with a polymer or by introducing a liquid into the voids of the bound zeolite film.

[0077] Another method for coating metal foils with prefabricated zeolite crystals, or microporous particles, is electrophoretic deposition (EPD). EPD is a technique for applying high quality coatings of uniform thickness to metal substrates. The method can be used to apply organic and inorganic particulate coatings on electrically conductive substrates. Slurry compositions containing prefabricated zeolites, or microporous particles, may be electrophoretically applied to a rigid support material, such as by using the method described in Bowie Keefer et al.'s prior Canadian patent application No. 2,306,311, entitled "Adsorbent Laminate Structure," which is incorporated herein by reference.

[0078] Some contactor geometric shapes will require that the adsorbent be applied to the channel surface in a layer using a colloidal binder material or that an entire geometric shape be comprised of the adsorbent plus colloidal binder and containing a plurality of parallel channels. When a colloidal binder is used, the selection of the colloidal material depends on the particular adsorbent used. Colloidal materials capable of functioning as a binder and/or which form a gel are preferred. Such colloidal materials include, without limitation, colloidal silica-based binders, colloidal alumina, colloidal zirconia, and mixtures of

_ ^ _

colloidal materials. "Colloidal silica" refers to a stable dispersion of discrete amorphous silicon dioxide particles having a particle size ranging from about 1 to about 100 nanometers. Suitable colloidal silica materials also can be surface modified, such as by surface modification with alumina. Colloidal silicas are commercially available from a number of companies, such as Eka Chemicals, Grace Davison, Nalco, etc. Ludox is one example of a colloidal silica binder suitable for use in the preparation of the contactors of the present invention. Ludox can be obtained in a number of formulations, including HS30 and HS40. Ludox can be used alone, or in combination with other materials, such as Odorlok which is comprised of about 14% alumina, 3% calcium oxide, 1 percent magnesium oxide, and about 4% crystalline silica, the remainder being inert ingredients, according to the Zeolite-Material Safety Data Sheet (Canada). Odorlok apparently bridges between zeolite particles, adding to the strength of the laminate produced relative to those formulations using solely colloidal silica as the binder. Another type of colloidal binder suitable for use herein include clay materials, such as palygorskite (also known as attapulgite), which are hydrated magnesium aluminum silicates. Also, inorganic binders may be inert; however, certain inorganic binders, such as clays, used with zeolite adsorbents may be converted in-situ from kaolin binders to zeolite so that the zeolite is self- bound with minimal inert material. Organic binders used to bind activated carbon particulates in laminated structures may be pyrolyzed to form a useful carbonaceous adsorbent.

[0079] Returning to Figure 9, using the methods described in for Figure 1 hereof, illustrates an embodiment of the present invention wherein a microporous adsorbent film 505, preferably DDR, is hydrothermally grown on each of both faces of flat metal foils 509, which is preferably fabricated from a corrosion resistant metal such as stainless steel. The separate metal foils 509 with the adsorbent films 505 are fabricated to form a parallel channel contactor

_ _

501. Spacers of appropriate size may placed between the metal foils during contactor fabrication so that the channel gap 503 is of a predetermined size. Preferably about half of the volume of the feed channels 503 are filled with a spacer that keeps the sheets substantially evenly spaced apart.

[0080] The heat capacity of the metal foils 509 limits the thermal excursions in the process. When CO ₂ is adsorbed in the adsorbent, heat is released in the amount of the heat of adsorption. This warms the adsorbent films and as the film warms, its temperature rises above that of the metal foils and heat diffuses into the metal foil where it is stored. Desorption Of CO ₂ from the adsorbent is an endothermic process and heat must be supplied in an amount equal to the heat of adsorption. When CO ₂ desorbs, the temperature of the films falls below that of the metal foils and heat stored in the metal foils flows into the films. The thermal excursion of the adsorbent film is less than +/-1O ⁰ C with the contactor dimensions and the process as described in Example 1.

[0081] The present invention can better be understood with reference to the following examples that are presented for illustrative purposes and not to taken as limiting the invention.

Example 1

[0082] With the parallel channel contactor described for Figure 9 hereof, a PSA/RCPSA cycle with five steps is operated to produce a product stream containing about 20 vol.% CO ₂ and about 80 vol.% CH ₄ . Overall methane recovery for the PSA/RCPSA cycle is computed to be about 95 vol.%. Figure 10 hereof is a schematic diagram of five different steps in a preferred PSA/RCPSA cycle suitable for use in this invention. In the first step 611 a parallel channel contactor PSA/RCPSA cycle is pressurized with high pressure product gas 687. This pressurization raises the pressure in the parallel channel

_

contactor and fills the contactor with the purified product containing about 20 vol.% CO ₂ and about 80 vol.% CH ₄ . In a second step 621 a high pressure 51 atmosphere (atm) feed gas 671 is conducted through the parallel channel contactor. During this step 621 the DDR adsorbent layer adsorbs CO ₂ from the flowing feed gas 671. A purified product 625 flows out of the end of the contactor. The feed gas 671 is flowed at a rate such that as the product 625 emerges from the parallel channel contactor as a concentration front moves through the contactor. Ahead of the front the gas has a composition near that of the product 625. Behind the front the gas has a composition near that of the feed 671. The second step 621 is stopped before this front completely breaks-through the end of the contactor. The amount of feed which emerges from the contactor before this step is halted determines in part the product purity.

[0083] At this point, a third step of the cycle 631 is initiated which serves to purge the contactor of feed gas trapped in the contactor channels. The third step 631 also acts, in part, as a partial pressure displacement purge of the contactor. Valves are opened at the top and bottom of the contactor. A pressurized CO ₂ rich stream 633 flows into the top of the module and gas originally contained in the flow channel 503 of the structured parallel channel contactor flows out 639. The gas fed into the top of the module 633 is a CO ₂ -rich gas produced in later steps 4 and 5 that has been compressed 675 to a pressure slightly greater than the feed pressure (51 atm.). The composition of the gas fed through the top of the contactor is substantially equal to that of the CO ₂ reject stream 681, containing 97.5 vol.% CO ₂ and 2.5 vol.% CH ₄ . The gas exiting out the bottom of the contactor has a composition nearer to that of the feed gas 671 (70 vol.% CO ₂ and 30 vol.% CH ₄ ).

[0084] As the gas stream entering the module 633 displaces the gas in the flow channels, a compositional front moves from top to bottom of the module. The third step 631 is stopped and a fourth step 641 is begun before, or shortly

after, this front breaks through the bottom of the module. The fourth step 641 lets the pressure of the contactor down to an intermediate pressure and recovers some of the CO ₂ for recompression. In the design discussed in this example, the intermediate pressure is about 22 atm. In the fourth step, a CO ₂ -rich stream 649 exits the module at a pressure of about 22 atm. This stream is split into two streams 679 and 681. Stream 679 is fed to compressor 675 and stream 681 is rejected from the process at a pressure of about 22 atm. Stream 633 that was used to rinse the contactor in the third step of the process 631, is comprised of the gas stream 679 that emerges from compressor 675. As the pressure in the contactor drops towards the outlet pressure of about 22 atm., the flow in streams 679 and 681 decreases. When the flow in these streams has fallen to approximately ¹ A of the initial value step 4 is stopped and a step 5 is begun. In the fifth step of the process 651, the module pressure is dropped to about 5 atm. and a CO ₂ -rich stream is recovered 685. This stream 685 can optionally be fed through a compressor 677 that raises the stream pressure to about 22 atm. The stream is then combined with stream 681 and a CO ₂ -HCh stream 683 is rejected from the process at a pressure of about 22 atmospheres.

[0085] To improve the operation of the process, as well as the pressure at which CO ₂ is recovered, gas may be recovered in the fifth step 651 using a multi- step process in which the contactor pressure is decreased in a series of pressure equalization steps. Gas from these pressure equalization steps can be recovered as individual gas streams and recompressed. In an example with two pressure equalization steps, one portion of the CO ₂ -rich gas is recovered at a pressure of about 12 atmospheres while the rest is recovered at about 5 atmospheres.

[0086] It is also possible to decrease the module pressure in step four 641 using a series of pressure equalization steps. Again, each pressure equalization step can be used to form a gas stream that can either be rejected from the process in stream 683 or recompressed to form stream 633. If pressure equalization

. ,

- 4θ -

steps are employed, it is advantageous to design them to maximize the pressure at which the CO ₂ reject streams are captured.

[0087] Modeling used to predict the performance of the parallel channel contactor uses isotherms for CO ₂ and CH ₄ that were measured with well known gravimetric uptake methods, PVT (pressure, volume, temperature) methods, and with analysis of single component gas transport data in DDR membranes. A statistical isotherm shape was found to best describe the single component isotherms for CO ₂ and CH ₄ . The best fits to the measured isotherms for CO ₂ and CH ₄ give saturation capacities of 6 and 5 molecules per cage, respectively, in the DDR zeolite framework. These values correspond to a maximum loading of 5 milli-moles/gram (of DDR) for CO ₂ and 4.16 milli-moles/gram (of DDR) for CH ₄ . These saturation capacities are consistent with the physical expectation that the maximum possible loading would correspond to CO ₂ and CH ₄ filling the pores at a liquid density. A single parameter K, that is analogous to the Henry's constant in a Langmuir isotherm, describes the shape of the statistical isotherm. The K values used for modeling are:

25 * 10' ^J ° ^Ule

Mole

K _COi = 1.93 x 10 - -1100 „ R R TT (in pascals ^' )

,7.8 x 10 ³ ^ Mole

K _CHt = 4.25 x lQ- ¹⁰ e ^{R T} (in pascals ^"1 )

where R is the molar gas constant and T is the temperature in Kelvin.

[0088] Over a wide range of conditions (less than about 50% of saturation capacity loading), the shape of the statistical and Langmuir isotherms are very similar. For simple modeling of the process given in this example, the statistical

isotherm can be supplanted by an equivalent Langmuir isotherm. For modeling competitive adsorption effects, competitive adsorption isotherms can be derived from the single component statistical isotherms using well known techniques.

[0089] The single component Stefan-Maxwell transport diffusion coefficients for CO ₂ and CH ₄ used in the modeling hereof were:

7 4 * 10 ³ — Mole

= 5.70 _{l o} -io

X e R T (in m ² /sec)

13 4 , 10 ³ ^ Mole

= 0.48 _{l o} -io R T '

X e (in m ² /sec)

where R is the molar gas constant and T is the temperature in Kelvin.

[0090] It is seen that there is a large difference in the diffusion coefficients of CO ₂ and CH ₄ and a smaller difference in the isotherms when these transport parameters are evaluated at a given temperature. From a 50/50 molar mixture of CO ₂ and CH ₄ the isotherms slightly favor CO ₂ adsorption and dramatically favor the diffusional transport Of CO ₂ into the DDR crystals. By controlling the time scale of the adsorption step 621 and the purge displacement step 631, it is possible to take advantage of this difference in diffusion coefficients and improve the selectivity of the process. By controlling these time steps, a kinetic separation of CO ₂ and CH ₄ can be achieved that takes advantage of differences in diffusivity of these molecules. The class of 8-ring zeolites preferred for the removal of CO ₂ from natural gas will have a large difference in CO ₂ and CH ₄ diffusion coefficients. This example illustrates a particular RCPSA cycle that can be tuned to achieve a kinetic separation of CO ₂ and CH ₄ , however, other swing adsorption cycles are possible. A parameter that can be used to evaluate the ability of a given material to produce a kinetic separation is the ratio of

- -

diffusion coefficients for the components that are to be separated and the diffusion coefficients are evaluated at the temperature and pressure of the intended process,

[0091] It is preferred that the material be chosen to have a value of K for CO ₂ and methane separation greater than 10 at the operating temperature. More preferably the material is chosen to have a value of K greater than 25 at the operating temperature. More preferably the material is chosen to have a value of K greater than 50 at the operating temperature.

[0092] In order to take advantage of the intrinsic kinetic selectivity of the preferred class of 8-ring zeolite materials for removal of CO ₂ from natural gas, the crystals forming the contactor are preferably substantially the same size. If they have widely different sizes, some will substantially fill with CH ₄ during the adsorption step 621, resulting in increased methane loss during the desorption step 651. It is therefore preferred that the standard deviation of the volume of the individual crystallites in the DDR film forming the adsorbent layer 505 (as shown in Figure 9) be less than 100 % of the volume of an average crystallite in order to increase methane recovery in the process. In a more preferred embodiment, the standard deviation of the volume of the crystallites in the DDR film forming the adsorbent layer 505 is less than 50% of the volume of an average crystallite. In the most preferred embodiment, the standard deviation of the volume of the crystallites in the DDR film forming the adsorbent layer 505 is less than 10% of the volume of an average crystallite. The most preferred embodiment was chosen to model the PSA cycle described in this example.

_ _

[0093] With this type of adsorbent, the time for steps 621 and 631 is set by the average crystal size in the adsorbent. It is preferred that the time step be chosen so that adsorbed CO ₂ in the DDR has time to equilibrate with gaseous CO ₂ in the feed channel 503 but the methane does not have time to equilibrate. The time for CO ₂ to achieve 90% approach to equilibrium (following a change in surface concentration) within a single DDR crystal that has rapid diffusion path to the gas in the contactor is;

τ ₉₀ = 0.183 I ² Z Dc ₀₂

where r is the average DDR crystal radius and D _C o ₂ is the diffusion coefficient OfCO ₂ at the operating temperature. It is preferred that the time for steps 621 and 631 be in a range from 0.5 τ ₉₀ to 10 τ ₉₀ and it is more preferred that the time for steps 621 and 631 be in a range from 1 τ ₉₀ to 5 τ ₉₀ . For modeling, a time step of 1.5 τ ₉ o was chosen for steps 621 and 631. The numerical value of the time step is then set by the crystal size. It is preferred that the average DDR crystal size be in a range from about 0.005 μm to about 100 μm. It is more preferred that the average DDR crystal size be in a range from about 0.5 μm to about 50 μm and it is most preferred that the average DDR crystal size be in a range from about 1 μm to about 10 μm. For modeling an average DDR crystal size of 1 μm was used.

[0094] Several different treatments of the molecular transport into and out of the adsorbent layer were developed and results of the different modeling approximations were compared. The most exact treatment solved the time dependent fundamental multi-component transport equations into and out of the DDR zeolite layer at all points along the feed channel for every time step in the process. For this model the DDR film was idealized as fins along the side of the channel with 200 angstrom gaps between the fins. The pores formed by these

200 angstrom gaps occupied 5% of the volume of the adsorbent layer. On a grid encompassing 500,000 points, the fundamental time dependent transport equations were solved for this geometry. Three pressure equalization steps were used in the blow-down step 651 for process modeling. Pressure equalizations at 15, 10 and 5 atmospheres were employed. It was determined that the thermal excursion in the adsorption step was 1O ⁰ C. The methane recovery was computed as the ratio of the methane molar flow rate in purified product stream 625 to that in the feed stream 671 entering the process. The model using the most exact treatment of transport showed a 96% methane recovery. The average pressure at which molecules were recovered in streams 681 and 685 was found to be 12.5 atmospheres.

[0095] This modeling approach provides a much more exact solution than the linear driving force (LDF) models that are conventionally used to model PSA processes. Using knowledge of the most exact solution, a simpler model has been constructed that can readily be used by one skilled in the art to compute methane recoveries. In the adsorption step 621, the simple model separately treats the equilibration of CO ₂ and CH ₄ in the feed channel within the DDR crystals forming the adsorbent layer. The amount Of CO ₂ adsorbed in the DDR crystals is taken to be 80% of the amount that might be expected if the CO ₂ adsorbed in the DDR were fully equilibrated with the gaseous feed entering the process 671 at a temperature that is 10 ⁰ C higher than the feed temperature. The amount Of CH ₄ adsorbed in the DDR crystals is 1% of the amount that might be expected if the CH ₄ adsorbed in the DDR has fully equilibrated with the gaseous feed entering the process 671 at a temperature that is 10 ⁰ C higher than the feed temperature. For the simple model, the CO ₂ purge used in step 631 displaces all of the methane left in the feed channel into stream 639. With these approximations, the methane recovery from the process is predicted to be 95%.

[0096] Optionally when the CO ₂ reject stream 683 is sequestered it is preferred to capture the CO ₂ at a pressure that is more than 1/10 of the partial CO ₂ pressure in the feed. In a more preferred embodiment the pressure at which the CO ₂ is captured is more than ¹ A of the CO ₂ partial pressure in the feed.

Example 2

[0097] This example illustrates use of a parallel contactor of the present invention in a separation process that produces relatively high pressure products and high methane recoveries from N ₂ containing natural gas stream. In processing natural gas, the amount of N ₂ that has to be removed depends on the concentration in the field and the way in which the gas is transported to market (i.e., liquefied natural gas vs. pipeline). This example will consider a natural gas stream containing a small amount (< 5%) of impurities (for example H ₂ O and mercury compounds) other than N ₂ . These impurities are removed in initial processing steps using conventional separation techniques. The gas stream fed to the parallel contactor of the present invention has a composition of 70% CH ₄ and 30% N ₂ . The flowing gas stream is fed to the contactor at a pressure of 100 atmospheres and a temperature of 50 ⁰ C.

[0098] The contactor is comprised of laminated flat sheets of the type described above and a schematic diagram of the type of sheet used in the present example is shown in Figure 11 hereof. Using the methods described above a 50 μm thick DDR film 505 with an Si/Al ratio greater than 100 is hydrothermally grown on each of both faces of a 100 μm thick flat metal foil 509 (for example stainless steel). The metal foils 509 with the DDR films 505 are laminated together 501 to form a parallel channel contactor. During lamination, spacers are placed between the metal foils so that the channel gap 503 is 50 μm across. Approximately half the volume of the feed channels 503 are filled with spacers that keep the sheets substantially evenly spaced 50 μm apart.

_ _

[0099] The heat capacity of the metal foil 509 limits the thermal excursions in the process. When N ₂ is adsorbed in DDR it releases heat in the amount of the heat of adsorption. This warms the DDR film. The DDR film warms to a temperature above that of the metal foil and heat diffuses into the metal foil where it is stored. Desorption OfN ₂ from DDR is an endothermic process and heat must be supplied in the amount of the heat of the adsorption. When N ₂ desorbs the temperature of the DDR film falls below that of the metal foil and heat stored in the foil flows into the DDR film. With the contactor dimensions and the process described in this example, the thermal excursion of the DDR film is expected to be less than +/- 5°C. Due to the smaller heat of adsorption for N ₂ (compared to CO ₂ ) this temperature rise is less than that of the description for Example 1 above.

[00100] The DDR film is comprised of individual DDR crystals, pores (including grain boundaries) and macropores. In this example, the crystals in the DDR film are substantially of the same size. Most of the open volume in the film is comprised of mesoporous cracks with characteristic widths of 200 angstroms.

[00101] Using this parallel channel contactor, a PSA/RCPSA cycle with five different steps is operated to produce product stream containing 2% N ₂ and 98% CRj. Overall methane recovery for the PSA/RCPSA cycle is computed to be 91%. Figure 11 hereof shows a schematic diagram of the five different steps in the PSA/RCPSA cycle. In the first step 711 a parallel channel contactor PSA/RCPSA cycle is pressurized with high pressure product gas 725. This pressurization raises the pressure in the parallel channel contactor and fills the contactor with the purified product containing 2% N ₂ and 98% CH ₄ . In a second step 721 a high pressure 100 atm feed gas 771 is flowed through the contactor. During this step 721 the DDR adsorbent layer removes N ₂ from the flowing feed

_ .

gas 771. A purified product 725 flows out of the end of the contactor. The feed gas 771 is flowed at a rate such that as the product 725 emerges from the contactor a concentration front moves through the contactor. Ahead of the front the gas has a composition near that of the product 725. Behind the front the gas has a composition near that of the feed 771. Before this front completely breaks through the end of the contactor the second step 721 is stopped. The amount of feed which emerges from the contactor before this step is halted determines in part the product purity.

[00102] At this point, a third step of the cycle 731 is initiated which serves to purge the contactor of feed gas trapped in the contactor channels. The third step 731 also acts in part as a partial pressure displacement purge of the contactor. Valves are opened at the top and bottom of the contactor. A pressurized N ₂ rich stream 733 flows into the top of the module and gas originally contained in the flow channel 503 (of Figure 11 hereof) of the structured parallel channel contactor flows out 739. The gas fed into the top of the module 733 is a N ₂ rich gas produced in later steps (4 and 5) that has been compressed 775 to a pressure slightly greater than the feed pressure (100 atm.). Composition of the gas fed in through the top of the contactor is nearly that of the N ₂ reject stream 781. The gas exiting out the bottom of the contactor has a composition nearer to that of the feed gas 771 (30% N ₂ and 70% CH ₄ ).

[00103] As the gas stream entering the module 733 displaces the gas in the feed channel a compositional front moves from top to bottom of the module. Before or shortly after this front breaks through the bottom of the module the third step 731 is stopped and a fourth step 741 is begun. The fourth step 741 lets the pressure of the contactor down to an intermediate pressure and recovers some of the N ₂ for recompression. In the design discussed in this example the intermediate pressure is 30 atm. In the fourth step a N ₂ rich stream 749 exits the module at a pressure of 30 atm. This stream is split into two streams 779 and

_ _

781. Gas in stream 779 is fed to a compressor 775 and gas in stream 781 is rejected from the process at a pressure of 30 atm. In an optimization of this process a pressure in step 741 is chosen that minimizes the amount of gas flowing in stream 781. Stream 733 that was used to rinse the contactor in the third step of the process 731 is comprised of the gas stream 779 that emerged from the compressor 775. As the pressure in the contactor drops towards the outlet pressure of 30 atm, the flow in streams 779 and 781 decrease. When the flow in these streams has fallen to approximately ¹ A of the initial value the fourth step is stopped and a fifth step is begun. In the fifth step of the process 751 the module pressure is dropped to 1.2 atm and a N ₂ rich stream is recovered 785.

[00104] To improve the operation of the process, as well as the pressure at which N ₂ is recovered, gas may be recovered in the fifth step 751 using a multi- step process in which the contactor pressure is decreased in a series of pressure equalization steps. In an example with two pressure equalization steps, one portion of the N ₂ rich gas is recovered at a pressure of 12 atmospheres while the rest is recovered at 1.2 atmospheres.

[00105] It is also possible to decrease the module pressure in step four 741 with a series of pressure equalization steps. Again each pressure equalization step can be used to form a gas stream that can either be rejected from the process in stream 783 or recompressed to form stream 733. If pressure equalization steps are employed it is advantageous to design them to maximize the pressure at which the N ₂ reject streams are captured.

[00106] The modeling used to predict the performance of the contactor used isotherms for N ₂ and CH ₄ that were measured with well known gravimetric uptake methods, PVT (pressure, volume, temperature) methods, and with analysis of single component transport data in DDR membranes. A statistical isotherm shape was found to best describe the single component isotherms for

_ _

N ₂ and CH ₄ . The best fits to the measured isotherms for N ₂ and CH ₄ give saturation capacities of 5 molecules per cage in the DDR zeolite framework. These values correspond to a maximum loading of 4.17 milli-moles/ gram (of DDR). These saturation capacities are consistent with the physical expectation that the maximum possible loading would correspond to N ₂ and CH ₄ filling the pores at a liquid density. A single parameter K, that is analogous to the Henry's constant in a Langmuir isotherm, describes the shape of the statistical isotherm. The K values used for modeling are:

9.6 J lO ¹ --*"*

Mole

K _N = 3.79 x lO ^"9 e ^{R T} (in pascals")

Mole

K _CHt = 4.25 x lO ^"10 e ^{R T} (in pascals ^"1 )

where R is the molar gas constant and T is the temperature in Kelvin.

[00107] Over a wide range of conditions the shape of the statistical and Langmuir isotherms are very similar. For simple modeling of the process given in this example, the statistical isotherm can be supplanted by an equivalent Langmuir isotherm. For modeling competitive adsorption effects, competitive adsorption isotherms can be derived from the single component statistical isotherms using well known techniques.

[00108] The single component Stefan-Maxwell transport diffusion coefficients for N ₂ and CH ₄ used in the modeling were

- -

1.5 , 10» *ϊ* Mole

D _Ni = 0.48 * It) ^"10 e R T (in m ² /sec)

Mole

D _CHt = 0.48 X lO ^"10 e ^{R T} (in m ² /sec)

where R is the molar gas constant and T is the temperature in Kelvin.

[00109] When these transport parameters are evaluated at a given temperature, it is seen that there is a large difference in the diffusion coefficients ofN ₂ and CH ₄ and a smaller difference in the isotherms. From a 50/50 molar mixture ofN ₂ and CH ₄ , the isotherms slightly favor CH ₄ adsorption but dramatically favor the diffusional transport OfN ₂ into the DDR crystals. By controlling the time scale of the adsorption step 721 and the purge displacement step 731, it is possible to take advantage of this difference in diffusion coefficients and improve the selectivity of the process. By controlling these time steps a kinetic separation of N ₂ and CH ₄ can be achieved that takes advantage of differences in diffusivity of these molecules. The preferred class of 8-ring zeolite materials for removal ofN ₂ from natural gas will have a large difference in N ₂ and CH ₄ diffusion coefficients. This example illustrates a particular RCPSA cycle that can be tuned to achieve a kinetic separation of N ₂ and CH ₄ ; however, other swing adsorption cycles are possible. A parameter that can be used to evaluate the ability of a given material to produce a kinetic separation is the ratio of the single component diffusion coefficients for the components. The ratio is evaluated at the temperature of the intended process,

D,

K =

D ₁ CH ₄

Values of K for DDR for nitrogen and methane separation at several different temperatures are given in the table below:

[00110] For the preferred class of 8-ring zeolite materials for removal OfN ₂ from natural gas K _N2/CH4 is a function of temperature. It is preferred that the material be chosen to have a value of K _N2/CH4 greater than 5 at the operating temperature. More preferably the material is chosen to have a value of K _N2/CH4 greater than 20 at the operating temperature. Even more preferably the material is chosen to have a value of K _N2/CH4 greater than 50 at the operating temperature.

[00111] In order to take advantage of the intrinsic kinetic selectivity of the preferred class of 8-ring zeolite materials for removal OfN ₂ from natural gas, the crystals forming the contactor are preferably the same size. If they have widely different sizes some will substantially fill with CH ₄ during the adsorption step 721, resulting in increased methane loss during the desorption step 751. To increase methane recovery in the process it is then preferred that the standard deviation of the volume of the individual crystallites in the DDR film forming the adsorbent layer 505 be less than 100 % of the volume of an average crystallite. In a more preferred embodiment the standard deviation of the volume of the crystallites in the DDR film forming the adsorbent layer 505 is less than 50% of the volume of an average crystallite. In the most preferred embodiment the standard deviation of the volume of the crystallites in the DDR film forming the adsorbent layer 505 is less than 10% of the volume of an

average crystallite. The most preferred embodiment was chosen to model the PSA cycle described in this example.

[00112] With this type of adsorbent, the time for steps 721 and 731 is set by the average crystal size in the adsorbent. It is preferred that the time step be chosen so that adsorbed N ₂ in the DDR has time to equilibrate with gaseous N ₂ in the feed channel 503 but the methane does not have time to equilibrate. The time for N ₂ or CH ₄ achieve 90% approach to equilibrium (following a change in surface concentration) within a single DDR crystal that has rapid diffusion path to the gas in the contactor is;

τ ₉₀ = 0.183 r ² / D _N2 and τ ₉₀ = 0.183 i^ / Dαw

where r is the average DDR crystal radius and D _M2 and D _CH4 are the diffusion coefficients of N ₂ and CH ₄ at the operating temperature. If there are no external mass transfer limitations, the equilibration times for different crystallite sizes are given in the table below:

[00113] It is preferred that for N ₂ the time for steps 721 and 731 be in a range from 0.5 τ ₉₀ to 10 τ go and it is more preferred that the time for steps 721 and 731 be in a range from 1 τ go to 5 τ go. For modeling, a time step of 1.5 τ go was chosen for steps 721 and 731. The numerical value of the time step is then set

_ _

by the crystal size. It is preferred that the average DDR crystal size be in a range from 0.005 μm to 100 μm. It is more preferred that the average DDR crystal size be in a range from 0.5 μm to 50 μm and it is most preferred that the average DDR crystal size be in a range from 1 μm to 10 μm. For modeling an average DDR crystal size of 5 μm was used.

[00114] A simplified modeling approach similar to that described above was used. In the adsorption step 721, the simple model separately treats the equilibration ofN ₂ and CH ₄ in the feed channel with the DDR crystals forming the adsorbent layer. The amount ofN ₂ adsorbed in the DDR crystals is taken to be 80% of the amount that might be expected if the N ₂ adsorbed in the DDR has fully equilibrated with the gaseous feed entering the process 771 at a temperature that is 10 ⁰ C higher than the feed temperature. The amount Of CH ₄ adsorbed in the DDR crystals is 1% of the amount that might be expected if the CH ₄ adsorbed in the DDR has fully equilibrated with the gaseous feed entering the process 771 at a temperature that is 10 ⁰ C higher than the feed temperature. In the simple model, gas filling the mesopores and macropores in the laminate at the end of the adsorption step 721 is not recovered. Also, for the simple model the N ₂ purge used in step 731 displaces all of the methane left in the feed channel into stream 739. With these approximations the methane recovery from the process is predicted to be 91%.