Propylene Enrichment via Kinetic Vacuum Pressure Swing Adsorption

Subscriber access provided by UNIV OF DURHAM

Applications of Polymer, Composite, and Coating Materials

Propylene Enrichment via Kinetic Vacuum Pressure Swing Adsorption Using ZIF-8 Fiber Sorbents Brian R. Pimentel, and Ryan P Lively ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b08983 • Publication Date (Web): 01 Oct 2018 Downloaded from http://pubs.acs.org on October 1, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Propylene Enrichment via Kinetic Vacuum Pressure Swing Adsorption Using ZIF-8 Fiber Sorbents Brian R. Pimentel and Ryan P. Lively* School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, United States

1 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 29

KEYWORDS

Fiber Sorbents, Metal-Organic Frameworks, ZIF-8, Pressure Swing Adsorption, Kinetic Adsorption, Diffusion, Separation

ABSTRACT

This work presents the synthesis, characterization, and implementation of ZIF-8/Cellulose Acetate fiber sorbents for the cyclic kinetic adsorption separation of an equimolar propane/propylene feed. These fiber sorbents are packed as structured mass-transfer contactors and employed in a two-bed vacuum pressure swing adsorption cycle without propylene product purge. The unoptimized adsorption cycle process produces a high-pressure product of up to 81% propane purity at 31% recovery at 0 °C. The effects of adsorption step time, temperature, and feed rate are investigated and presented as a purity/recovery tradeoff. This work represents the first successful demonstration of a Metal-Organic Framework fiber sorbents in a pressure swing adsorption cycle, and the first use of MOFs in a kinetic separation cycle.


Page 3 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1.

INTRODUCTION

The separation of propane/propylene is one of the most difficult industrial distillation challenges. Due to the close boiling points of both of the components, the relative volatility of the mixture remains low and as a result, commercial towers require an excess of 120 stages to achieve polymer-grade (99.5%) propylene.1 The high reflux conditions necessary to achieve these purities combined with the cost of refrigeration cycles at sub-ambient temperatures result in an energy-intensive separation. Furthermore, the growing demand for plastic and the increased use of natural gas feedstock make this separation more important than ever.2 The reduction in energy costs of this chemical feedstock can be achieved by the implementation of next-generation technologies, such as adsorptive separation schemes. However, the high purity requirements of the propylene stream call for particularly selective sorbents to meet process specifications during the desorption step at reasonable recoveries.3 The production of propylene via adsorption generally involve sorbents with strong π-complexation interactions, such as AgNO3 or ion-exchanged zeolites, or a kinetically-selective sorbent (typically zeolite 4A).4-5 Previous reports of Pressure Swing Adsorption (PSA) and Vacuum Swing Adsorption (VSA) systems demonstrate varied performance,6-9 with some units achieving up to 95.9% recovery when combined with a second tail gas recovery unit.10 Given their high surface area and highly tunable adsorption properties, several reports of metal-organic frameworks (MOFs) with high selectivity for the olefin/paraffin pair exist,11-15 including a few demonstrating kinetic selectivity due to restrictive pore apertures intrinsic to the adsorbent.16-18 However, previous reports are typically confined to single component investigations, idealized batch adsorption experiments, or once-through breakthrough



Page 4 of 29

measurements, and therefore there has not yet been a report of a successfully implemented kinetic cyclic separation of propane/propylene using metal organic frameworks. Our previous work highlights the importance of matching the diffusive time scale of the sorbate-sorbent system to the adsorption process time,18 and demonstrates how manipulating the crystal size of ZIF-8 can result in a kinetically-selective system at reasonable process time scales. In brief, when the diffusive time scale of the sorbates, = / , is fast compared to the process time (cell equilibration time or bed residence time), the sorbent is rapidly equilibrated and demonstrates equilibrium selectivity towards the feed. However, when the diffusive time scales are comparable to the process time scale, and a significant difference between sorbate time scales exist ( ≥100x),19 the kinetically selective behavior is capturable within the sorbent. This transition between equilibrium and kinetic control can be induced by changing the term in the diffusive time scale, and to a lesser extent, by the temperature dependence of the term. The first part of this work demonstrates this transitional behavior by separation of an ethane/propane feed in packed beds of differently sized ZIF-8. As of now, relatively few MOFs have been demonstrated in a PSA or VSA operation, as these processes require materials to be produced at large scale and generally shaped prior to use.20-22 Fiber sorbents offer a scalable method to create structure sorbents by dispersing the material throughout an open polymer matrix.23-25 These structured mass transfer contactors result in sharper breakthrough fronts, faster cycle times, and allow for thermal modulation of exothermic adsorption processes.26-28 However, the water-based polymer phase inversion used in fiber spinning makes the use of MOFs without exceptional water stability difficult. Recently, UiO-66 was successfully incorporated in a cellulose acetate (CA) matrix for the removal of mercaptans


Page 5 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


from natural gas,29 while a post-synthetic modification approach enabled the growth of HKUST1 and ZIF-8 in CA fibers from hydroxy double salts via ZnO precursors.30 This works presents the synthesis and testing of kinetically selective ZIF-8/CA fiber sorbents in a 2-bed PVSA cycle for the proof-of-concept separation of an equimolar propane/propylene mixture without a heavy product (propylene) sweep. Although cycle optimization is not the goal of this work, the effects of adsorption step time, temperature, and feed rate are investigated and presented as a purity/recovery tradeoff to provide guidance for modeling work and future experimental studies. Product purities and recoveries are modest in these proof-of-concept cycles; Our high-level process analysis based on these experimental results suggest that this VPSA column can reduce the energy intensity of cryogenic distillation by 42 % in hybrid separation schemes (SI). This represents both the first demonstration of cyclic MOF fiber sorbent operation as well as the first cyclic operation of any MOF as a kinetic sorbent.

2.

MATERIALS AND METHODS

2.1 Materials The following were used as received without any further purification: Cellulose acetate (50,000 MW, Sigma Aldrich), polyvinylpyrrolidone (K30 Mw 40,000, TCI), N-methyl-2pyrrolidone (ACS grade, VWR), methanol (ACS grade, VWR), hexanes (ACS grade, VWR), Zn(NO3)2·6H2O (99+%, Alfa Aesar), 2-methylimidazole (97%, Alfa Aesar), ), sodium formate (98%, Alfa Aesar), Basolite Z1200 (Sigma-Aldrich), helium (UHP, Airgas), nitrogen (UHP, Airgas), 50% propylene in propane (certified blend, Airgas), 50% ethane in propane (certified blend, Airgas).



Page 6 of 29

2.2 ZIF-8 Synthesis Large ZIF-8 crystal samples of radii approximately 145 µm were synthesized in a glass vial in methanol as per previous work.31 Briefly, 2.205 g of zinc nitrate hexahydrate was dissolved in 25 mL of methanol. A total of 1.215 g of 2-methylimidazole and 0.504 g of sodium formate were dissolved in a separate 25 mL solution of methanol. The zinc solution was poured into the imidazole solution and stirring was stopped after mixing. The vial was kept capped at 90 °C for 24 h, then washed. Samples were dried at 80 °C under vacuum overnight to remove adsorbed methanol. Numerous synthesis batches were combined to yield over 50 g of sorbent material for fiber sorbent synthesis.

2.3 Fiber Sorbent Synthesis Fibers were spun in a dry-jet wet-quench spinning method. In brief, the solids were dispersed in solution using 80% of the total dope solvent by impeller and sonication (3 times for 60s) alternatingly. To this, a polymer solution consisting of 20% of the total dope solvent and dissolved polymer was added to maintain solid suspension and dispersion, and again sonicated (3 times for 60s) and stirred via impeller. To the mixed dope, the rest of the dry polymer was added and mixed via impeller for 4 h at 50 °C until homogenous. The overall dope composition is listed under Table 1. The mixture was loaded into syringe pumps and degassed over night at 50 °C. Due to the high solid volume and size of the particles present within the dope, the fibers were extruded through a 1/8” Swagelok adapter rather than a traditional coaxial spinneret. Fibers were quenched in a water bath and taken up by a rolling drum in a secondary water bath. The spinning conditions for these fibers are detailed in Table 2.


Page 7 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Cellulose acetate was chosen as the porous polymer support in this work due to its favorable spinning characteristics and low materials cost. A propane/propylene breakthrough of sorbentfree CA fibers shows no separation, minimal hydrocarbon uptake, and demonstrates the polymer’s role as a non-adsorbing support for the ZIF sorbent (Figure S2). Table 1. ZIF-8/CA fiber sorbent dope composition CA/ZIF-8 %) NMP

60.10

Water

10.16

Cellulose Acetate (Mw 50,000)

10.24

PVP (Mw 40,000)

3.97

ZIF-8

15.36

(wt

Table 2. Spinning conditions utilized in this work* CA/ZIF-8 Spinneret Temperature (°C)

60

Bath Temperature (°C)

50

Dope Flow Rate (ml/h)

350

Drum Uptake Rate (m/min)

6.5

Air Gap (cm)

3

Draw Ratio

6.7

*

Dope was extruded through 1/8” Swagelok tubing and not a traditional coannular die After being cut from the drum, fibers were solvent exchanged in a water bath for 3 days to remove left over NMP from the spinning solution. At this point, while still wet and lightly 7 ACS Paragon Plus Environment


Page 8 of 29

plasticized, the fibers were loaded into 12” long, ¼” Swagelok® tubing by threading the bundle of fibers from one end of the tube and out the other. This has been found to increase the success rate of loading unbroken fibers, though potentially leads to a lower packing fraction as fibers shrink somewhat after drying. Once inside Swagelok® tubing, the fibers were exchanged 3 times for 1 h in methanol, then hexanes, followed by drying at 110 °C under vacuum.

2.4 Material Characterization Crystals were characterized using X-ray diffraction (XRD; PANalytical X’pert PRO Alpha-1 diffractometer with rotating sample stage), scanning electron microscopy (SEM; Hitachi SU8230 cold-field-emission scanning electron microscope), and cryogenic N2 physisorption (Microtrac BELSORP-max).

2.5 Pressure Drop Experiments Pressure drop across fiber module beds were evaluated as a function of superficial velocity using a gas rotameter and a differential pressure gauge. Gas velocities were varied from 10 to 70 cm/s at room temperature using a N2 gas line as the probe fluid.

2.6 Fixed Bed Experiments Breakthrough experiments were carried out in a temperature controlled cabinet system built by L&C Science and Technology, connected to a Pfeiffer Omnistar mass spectrometer. Two 12” beds of differing particle size were prepared. The first bed consisted of Basolite Z1200 nanoparticles (~25-50 nm radii) pressed into pellets and then ground and sieved to approximately 500 µm size. Approximately 1.7 g of material was packed into the ¼” Swagelok module and


Page 9 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


held with a glass wool plug on either side. The second bed consisted of 2.1 g of the 145 µm ZIF8 crystals, without any further modification and again plugged by glass wool on either side. The samples were activated at 110 °C for 8 h prior to initial experiments and for 2 h in between sample runs. The bed was pre-pressurized with nitrogen while feed mixtures consisted of an equimolar ethane/propane stream diluted in helium at 298 K. The system was operated at 5 bar with a total hydrocarbon partial pressure of 0.9 bar at 0 °C. Total flowrate was set to 94 cc STP /min. Similar breakthrough experiments were performed using one of the packed fiber modules, where the hydrocarbon feed consisted of equimolar propane/propylene diluted in helium (80%) rather than ethane/propane, and the total bed pressure was 1 bar.

2.7 Vacuum Swing Adsorption Cycle Design and Operation Cyclic studies were carried out in in the same unit as the breakthrough experiments using a 2bed configuration of packed fiber modules. A simplified unit diagram is presented in Scheme 1 and bed characteristics are listed in Table 3. Samples were initially degassed at 110 °C for 12 h under vacuum, then equilibrated under approximately 10 ccSTP of the propane/propylene feed mixture at 1 bar and 40 °C to saturate the slower diffusing species, then brought down to 0 °C under the same flow. This aids in the rapid approach to cyclic steady state of kinetic separations, and therefore beds were subsequently not evacuated in between different cycle configurations. Each configuration was evaluated over 30-50 cycles and was generally seen to achieve a cyclic steady state after 15 cycles. Cycles were simple 4-step VPSA configurations consisting of (1) co-current feed pressurization at 2 bar, (2) co-current adsorption under feed flow at 2 bar, (3) product-end



Page 10 of 29

pressure equalization, (4) counter-current bed evacuation under full vacuum. No blowdown was necessary given the bed equilibration occurring at 1 bar. Due to the configurational limitations of the unit, the adsorbed product could not be directly measured and instead the cycle performance has been gauged by the high-pressure propane rich product, rather than the more valuable propylene rich stream. For similar reasons, the typical purge steps associated with kinetic production of high-purity propylene could not be incorporated, and therefore we present this work as a demonstration of the viability of the composite material towards kinetic separations rather than as a direct comparison with other published works.

Feed

Vacuum

Product/ Mass Spec Scheme 1. Simplified unit schematic of the 2-bed VPSA system used in this work.

Table 3. Bed characteristics used in this study


Page 11 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Bed Length (cm)

30

Bed Diameter (cm)

0.48

Number of Fibers (-)

11

Fiber Diameter (µm)

1000

Void Fraction (-)

0.5

Sorbent Mass per Bed(g)

0.6

Cycles were evaluated through the conventional metrics of purity, recovery, and productivity over the last 6 cycles. Purity of the high-pressure stream was determined via mass spectrometry while productivity and recovery at steady state were calculated as follows:

=

∫ , ∙ ∫ , + , ∙ ,

=

∫ , ∙ !"!# ∙ $%&'(

(1)

(2)

Purity and recovery of the propylene-rich desorption product was estimated via mass balance of the system.

3.

RESULTS AND DISCUSSION

3.1 Powder Sorbent Adsorption Characterization of the ZIF-8 materials via XRD suggests highly crystalline materials in both the case of the synthesized sorbent and the commercial Basolite® sample, although expectedly broader peaks due to the smaller crystallite domains in the latter material (Figure 1). N2 physisorption (Figure 2) experiments reveal samples of high microporous volume and surface area, with the characteristic “gate-opening” phenomenon visible in the low-pressure range. 11 ACS Paragon Plus Environment


Page 12 of 29

Reduction in crystal size has been known to shift the appearance of the gate-opening effect towards higher pressures,32 as seen in our nanocrystalline Basolite® data. Both sorbents exhibit very high surface area with 1450 m2 g-1 (0.63 cc/g) and 1680 m2 g-1 (0.63 cc/g) for the ZIF-8 crystals and the Basolite® sample respectively. Due to the non-adsorbing CA matrix, the fiber sorbent samples exhibit lower surface area and pore volume 960 m2 g-1 (0.35 cc/g), although the reduction corresponds to the weight loading of ZIF-8 within the non-adsorbing CA matrix. SEM images and crystal size distributions of the powder samples used in this study are available in the SI (Figures S3-S4).

Figure 1. Powder XRD patterns of the ZIF-8 samples studied in this work. From top to bottom: ZIF-8/CA fibers (green), Basolite® (blue), ZIF-8 crystals (red), simulated (black)33


Page 13 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. N2 physisorption at 77 K of ZIF-8 crystals (,), Basolite (7), and ZIF-8 fibers (!) The adsorption and diffusion of light hydrocarbons in ZIF-8 has been characterized in previous work. Due to the flexibility of the pore aperture linkers in ZIF-8, the crystallographic aperture of 3.4 Å is able to demonstrate significant diffusive selectivity between light hydrocarbons, but not perfect molecular sieving. From previously measured single component diffusivities, we estimate the diffusive selectivity in ZIF-8 of ethane over propane to be ~500 and that for propylene over propane to be ~100-140 respectively.31 For reference, zeolite 4A has an estimated propylene over propane diffusivity ratio of ~200.34 Equilibrium sorption coefficients increase with carbon number, where the strength of adsorption can be summarized as Ethane < Propylene ≈ Propane. Fixed bed binary breakthrough experiments of ethane/propane using pelletized Basolite® sorbent demonstrate the selective sorption of propane over ethane, with roll-up behavior exhibited by competitive sorption as the faster moving ethane front is displaced in the bed by the propane front (Figure 3). This demonstrates equilibrium control of the column, with a longer



Page 14 of 29

breakthrough time, and therefore greater capacity, for propane over ethane, facilitated by the rapidly equilibrating sub-micron sized crystals in the Basolite® pellets.

Figure 3. Packed bed breakthrough of equimolar ethane/propane mixture at 298 K and 0.9 bar (balance 4.1 bar He) using sub-micron ZIF-8 (Basolite) crystals packed into pellets. t=0 indicates initial helium tracer breakthrough. As demonstrated in our previous work,18 creating a sufficiently long diffusive time scales via crystal size engineering is necessary to observe kinetically selective behavior during adsorption cycles. If the time scale for diffusion is too short, as in the case of adsorption into Basolite® nanocrystals, the sorbent is quickly equilibrated and the system falls under thermodynamic control. Breakthrough experiments of ethane/propane feed using large (~145 µm) ZIF-8 crystals exhibits fundamentally different behavior than that of the pelletized nanocrystals (Figure 4). Propane is seen to breakthrough almost simultaneously with Helium, indicating slow sorption kinetics and only partial uptake of the component. In contrast, ethane, although a less strongly adsorbing species at equilibrium, is seen to breakthrough after propane. Faster intracrystalline


Page 15 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


diffusion, and therefore overall uptake rate, leads to greater ethane uptake within the time scale of the breakthrough; this is the guiding principle behind a kinetic separation. Some small amount of ethane is seen to breakthrough with propane, which may either be an artifact of shared m/z peaks in the mass spectrometer, or signs that ethane is still relatively mass transfer limited where some amount of the gas still “bleeds” through the bed. The relative broadness of the breakthrough front compared to the equilibrium breakthrough configuration suggest that the latter is present to a significant degree.

Figure 4. Packed bed breakthrough of equimolar ethane/propane mixture at 298 K and 0.9 bar (balance 4.1 bar He) using 145 µm ZIF-8 crystals demonstrating kinetically selective behavior. t=0 indicates initial helium tracer breakthrough. By controlling the diffusive time scale, we have demonstrated that the same material is able to operate under both thermodynamic and kinetic control, and as such, reverse the breakthrough selectivity of a system.



Page 16 of 29

3.2 Fiber Sorbent Characterization Sorbent loading of the composite fibers was determined by N2 physisorption at 77 K similar to previous work.30 The microporous volume of the sample was ascribed exclusively to the ZIF-8 within the fiber structure, and the percentage uptake compared to the neat powder was used to estimate the sorbent weight loading within the fibers. By this method, the fiber composites were determined to be composed of 57 wt% ZIF-8, closely resembling the expected 60 wt% from the dope formulation. The isotherm of the fibers and the powder sample are shown in Figure 2 and exhibit the same microporous behaviour and gate-opening phenomena, suggesting little difference in the sorbent characteristics. XRD was conducted on the fibers after spinning and compared to the powder samples, and revealed that no structural changes had occurred in the ZIF-8 material (Figure 1). The ability of ZIF-8 to survive the spinning process is not wholly unexpected, as ZIFs are one of the more stable MOF families and most ZIFs are not susceptible to degrading in either water or the spinning solvents employed. The differences in peak intensities are likely a result of statistical distribution of exposed faces, as much fewer crystals are contained in the fiber scan than in a powder scan. In SEM, ZIF-8 crystals of varying sizes can be seen within the cross-sectional images, although almost all the crystals are above 20 µm. Most of the sorbent seems to be present in large crystals with diameters greater than 100 µm (Figure 5). These crystals are well-dispersed throughout the fiber, as are the indentations of other crystals that came away on the other face of the fiber during sample fracture for SEM analysis. The large size of the crystals with respect to the overall fiber diameter at times gives an impression of maldistribution, but no significant


Page 17 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


clumping exists and few crystals are seen to be in “face to face” configurations. The average fiber diameter was measured to be 1000 µm, with some small variations () = 80 ,$) present due to disturbances of the spinning line by large volumes of solids and some variability in spinning conditions.

Figure 5. Cross-sectional image of ZIF-8 fiber sorbents demonstrating large crystals dispersed throughout the polymer matrix. Inset shows higher magnification of sorbent fibers. Higher magnification cross sections in the figure inset reveal a highly porous polymer substructure; it may be possible to further enhance the substructure transport by future modifications to the spinning dope composition. Due to the structured natured of the packed modules, fiber sorbent beds allow for higher gas velocities during adsorption at similar pressure drops when compared to pellet packed beds. The measurements of pressure drop across the fiber modules are presented in Figure 6. Four different 17 ACS Paragon Plus Environment


Page 18 of 29

modules constructed from the spun fibers are compared to predictions of the Ergun equation for particle sizes of similar dimensions and comparable bed porosity. This most fairly compares the diffusive length scales present in both systems, although packed beds typically use extrudates 3 to 8 times larger than our current fiber dimensions to mitigate excessive pressure drop.35-37

Figure 6. Pressure drop of fiber modules prepared in this work compared to a traditional Ergun calculation. Packed bed pressure drops were calculated at 50% porosity and a particle size of 1 mm to correspond with measured fiber bed characteristics. Modules 1 and 2 were used in the cyclic adsorption studies. As expected of structured contactors versus randomly packed particles, the pressure drop across the fiber modules is significantly smaller than that predicted by the Ergun equation, and increases much more slowly with increasing superficial velocity. The ability to employ higher fluid velocities results in an increase in overall mass transfer throughout the bed and improved


Page 19 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


cycle productivity, which can outweigh the lower mass of sorbent compared to extrudate packed beds.38 A largely unavoidable aspect of these particular fiber modules is their difficulty in packing, especially in the case of high solid fraction fibers or fibers supporting large particles, both of which tend to be brittle. This makes packing fractions in excess of 50-60% difficult to achieve for these fiber modules, and therefore the most structured state of the structured packing is rarely achieved in these experiments. The greater configurations available at lower void fractions and the space allowed for fibers to curl and curve while in the module lead to non-ideal pressure drop and mass transfer behavior. Lower packing fractions also increase the risk of channelling pathways and decrease product purity; development of reliable packing strategies for brittle fiber sorbents is necessary prior to industrial scale-up. Breakthrough analysis of the fiber bed (M1) under propylene/propane feed reveals behavior like that observed when using large ZIF-8 crystals in an ethane/propane system. In Figure 7, propane is seen to breakthrough before propylene due to slower diffusion of the olefin through ZIF-8. The initial low level of propane detected in the outlet prior to helium breakthrough is a result of the shared 28 m/z peak of nitrogen being partially attributed to propane. The broadness of the propane breakthrough compared to that of helium suggests that some adsorption is still occurring at time scales relevant to the experiment. Propylene in the bed still appears somewhat mass transfer limited given its broad breakthrough front that emerges shortly after helium. Since isotherms are nearly identical in ZIF-8 for these two gases, we can conclude that large ZIF-8 crystals present within the fiber sorbent enable separation under a kinetically selective regime. However, it does suggest that the diffusive length scale may be too large and some



Page 20 of 29

improvements in swing capacity could likely be made by having more rapid adsorption and higher surface area-to-volume ratios of the adsorbent crystals.

Figure 7. Fixed bed breakthrough of a propane/propylene mixture (0.2 bar) over packed ZIF8/CA fiber sorbent bundles diluted in helium (0.8 bar) at 1 bar total pressure and 0 °C. N2 (black, left), He (red, left), propane (green, right), propylene (blue, right). = 0 corresponds to initial feed flow to the bed.

3.3 Vacuum Swing Adsorption Cycle Operation The kinetic cyclic performance of the ZIF-8/CA fiber modules were evaluated using two fiber beds packed with 11 unbroken fibers of 1000 µm diameters, M1 and M2. Various conditions were tested to explore the parameter space, including pressurization and feed rate, feed time, and process temperature. Most experiments were conducted at 0 °C, with some at 15 and 35 °C. An example of the approach to cyclic steady state of the outlet concentration is presented in Figure 8. The experiment was begun immediately after investigating other parameters, therefore initial concentrations do not necessarily match feed conditions. Forgoing the use of a product storage 20 ACS Paragon Plus Environment

Page 21 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tank allows for faster approach to equilibrium in low productivity situations, but the lack of product tank volume as a buffer leads to some degree of noise in the outlet concentration. Process conditions and results are presented in Table 4.

Figure 8. Representative outlet concentration profile demonstrating the system approach to steady state. Temperature- 273 K, Pressurization– 77 cc/min, Feed- 19 cc/min, Feed Time- 60s Table 4. Summary of cyclic PSA results using kinetic ZIF-8 fiber sorbents Temperature (K)

Pressurizationd Feedd (cc/min) (cc/min)

Feed Time Puritya,b Recoverya (%) (s) (%)

Productivityd (cc/g min)

154

60

62 [76]

85

33

30

67 [65]

64

20

60

63 [77]

85

27

30

74 [58]

38

9

60

69 [65]

62

13

90

64 [68]

72

15

60

77 [58]

35

6

77 77

273 77 39

19



Page 22 of 29

39

60

71 [63]

60

10

19

60

79 [57]

34

4

19

90

77 [65]

49

6

10

120

81 [58]

31

3

39

60

70 [65]

61

13

19

60

76 [58]

34

5

39

60

72 [67]

62

10

19

60

79 [57]

29

3

39

60

69 [65]

60

13

19

60

77 [57]

30

5

39

60

71 [65]

60

10

19

60

78 [56]

26

3

39

77 288 39

77 308 39 a

Purity and recovery are based propane. bPropylene purity calculated from a mass balance, reported in [brackets]. dVolume units of cc are specifically ccSTP as measured by mass flow controller and corrected for thermal conductivity. A non-optimized parametric study of some of the cycle variables reveals opposing trends in purity and recovery, as is typical of most PSA systems. Increasing the feed flow rate during the adsorption step increased product recovery, as a greater portion of the total inlet gas (feed and pressurization) from the cycle is now creating high-pressure product. However, both the reduced residence time that results from the parameter change and the increased mass needing to be adsorbed decreases product purity, as the sorbent becomes more saturated and less time is permitted for propylene adsorption. Clearly, it is not only a matter of saturating the sorbent, as would be expected in equilibrium separations, since cycles with equivalent feed production (i.e. twice the feed time, half the feed rate) demonstrate higher purity at lower flow rates. The only discrepancy in this comparison is that longer cycle times also mean longer evacuation steps, as the 2-bed cycle used in this work is necessarily symmetric.


Page 23 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The effect of feed time on purity and production was similarly typical to that of equilibrium PSA systems (Figure 9A). As the sorbents become saturated at higher feed times, lower purity product is produced. However, product recovery increases since a greater portion of the cycle time is spent producing high pressure product relative to pressurization steps. Overall, we observe that cycles with longer bed residence times for the feed (i.e. slower pressurization and lower feed rates) generally produced a higher purity product (Figure 9B). This is likely due to propylene having longer time to adsorb into the crystal, and suggests that the diffusive time scale of the propylene in the system is still greater than our operating parameters. As we did not observe a maximum in purity with decreasing flow rate, where we would see a return to equilibrium control, we assume that system was still operating within the kinetic regime. Were propane given enough time to adsorb, it would displace propylene and move towards equilibrium control of the separation, while overly fast residence times do not allow for enough adsorption to significantly alter the void composition.

Figure 9. A) Effect of feed rate on purity and recovery of propane at 0 °C. Pressurization (77 cc/min) and cycle time (60 s) held constant. B) Effect of feed time on purity and recovery of propane at 0 °C. Pressurization (77 cc/min) and feed rate (39 cc/min) held constant



Page 24 of 29

The overall trends in purity and recovery can be visualized in a Pareto plot, shown in Figure 10. These demonstrate the common trade-off between purity and recovery exhibited in all pressure swing adsorption systems where the Pareto front was drawn based on available experimental data. The highest achievable purity in this configuration was 81% propane at 31% recovery, demonstrating the ability of this composite material to partially separate the difficult olefin paraffin mixture. At higher temperatures, we expect the diffusion coefficients of both components to increase, but propane more so than propylene given measured activation energies.39-40 This would result in decreased product purity but potentially an increase in productivity, since the mass transfer of the system would be faster overall.8 Experiments at higher temperatures only saw mild changes in performance, which indicate that this separation could potentially be effectively performed at room temperature. It is likely that the effect of changing kinetic selectivities in the system would be more pronounced with the use of smaller crystal dimensions (characteristic times more closely resembling reasonable cycle parameters), or in the measurement of the evacuated propylene product.

Figure 10. Pareto plots of propane (A) and propylene (B) streams from experimental conditions and results reported in Table 4. Propane streams were measured directly while propylene values were calculated via a mass balance of the system 24 ACS Paragon Plus Environment

Page 25 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4.

CONCLUSIONS

In this work, we present the first reported use of a metal-organic framework (ZIF-8) in a proofof-concept kinetic PSA cycle, along with its incorporation into a structured sorbent fiber contactor. The transition from an equilibrium to a kinetically controlled adsorption scheme is demonstrated through breakthrough experiments of powder packed beds by changing the diffusive length scale. The composite fiber sorbent material exhibits 57 wt% MOF loadings and lower pressure drop across the bed compared to random packing predictions. The sorbent was demonstrated to have survived the spinning process and maintained its large crystal domains which enable its application in diffusionally-selective adsorption of propane over propylene. In a simple 4-step cycle, these fiber beds were able to produce high pressure product of up to 81% propane from an equimolar mixture. Increasing feed rate or feed time was shown to decrease product purity and increase recovery, a function of both mass transfer limitations and sorbent saturation. Temperature was found to only have a slight effect on product outcome, likely due to the size of the crystals being employed. Further optimization of cycle and material design parameters are necessary in the future to achieve unit performance competitive with the state-of-the-art. SUPPORTING INFORMATION Cellulose acetate fiber spinning and breakthrough. ASPEN process modeling simulation of hybrid PSA/distillation separation unit. AUTHOR INFORMATION Corresponding Author *To whom correspondence should be addressed: Ryan P. Lively. Tl: +1 (404) 894-8795. E-mail: [email protected] 25 ACS Paragon Plus Environment


Page 26 of 29

Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding Sources This work was supported by the NSF. BRP was funded under GRFP DGE-1650044 while RPL was supported under NSF EEC 134219. ACKNOWLEDGMENTS The authors acknowledge Dr. Simon Pang, Stephen J.A. DeWitt, and Conrad Roos for their assistance with various aspects of material characterization. The authors also acknowledge Fengyi Zhang for his assistance with the TOC. REFERENCES 1. Eldridge, R. B., Olefin/Paraffin Separation Technology: A Review. Ind. Eng. Chem. Res. 1993, 32 (10), 2208-2212. 2. Waldheim, J., Outlook '18: Us Propylene Production Set to Expand in 2018. ICIS News 3 Jan, 2018, 2018. 3. Da Silva, F. A.; Rodrigues, A. E., Propylene/Propane Separation by Vacuum Swing Adsorption Using 13x Zeolite. AlChE J. 2001, 47 (2), 341-357. 4. Padin, J.; Rege, S. U.; Yang, R. T.; Cheng, L. S., Molecular Sieve Sorbents for Kinetic Separation of Propane/Propylene. Chem. Eng. Sci. 2000, 55 (20), 4525-4535. 5. Rege, S. U.; Padin, J.; Yang, R. T., Olefin/Paraffin Separations by Adsorption: Π‐ Complexation Vs. Kinetic Separation. AlChE J. 1998, 44 (4), 799-809. 6. Narin, G.; Martins, V. F. D.; Campo, M.; Ribeiro, A. M.; Ferreira, A.; Santos, J. C.; Schumann, K.; Rodrigues, A. E., Light Olefins/Paraffins Separation with 13x Zeolite Binderless Beads. Sep. Purif. Technol. 2014, 133, 452-475. 7. Rege, S. U.; Yang, R. T., Propane/Propylene Separation by Pressure Swing Adsorption: Sorbent Comparison and Multiplicity of Cyclic Steady States. Chem. Eng. Sci. 2002, 57 (7), 1139-1149. 8. Grande, C. A.; Rodrigues, A. E., Propane/Propylene Separation by Pressure Swing Adsorption Using Zeolite 4a. Ind. Eng. Chem. Res. 2005, 44 (23), 8815-8829. 9. Grande, C. A.; Gascon, J.; Kapteijn, F.; Rodrigues, A. E., Propane/Propylene Separation with Li-Exchanged Zeolite 13x. Chem. Eng. J. 2010, 160 (1), 207-214. 26 ACS Paragon Plus Environment

Page 27 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


10. Grande, C. A.; Poplow, F.; Rodrigues, A. E., Vacuum Pressure Swing Adsorption to Produce Polymer-Grade Propylene. Sep. Sci. Technol. 2010, 45 (9), 1252-1259. 11. Bloch, E. D.; Queen, W. L.; Krishna, R.; Zadrozny, J. M.; Brown, C. M.; Long, J. R., Hydrocarbon Separations in a Metal-Organic Framework with Open Iron(Ii) Coordination Sites. Science 2012, 335 (6076), 1606-1610. 12. Xiang, S.; Zhou, W.; Gallegos, J. M.; Liu, Y.; Chen, B., Exceptionally High Acetylene Uptake in a Microporous Metal-Organic Framework with Open Metal Sites. J. Am. Chem. Soc. 2009, 131, 12415-12419. 13. Bae, Y. S.; Lee, C. Y.; Kim, K. C.; Farha, O. K.; Nickias, P.; Hupp, J. T.; Nguyen, S. T.; Snurr, R. Q., High Propene/Propane Selectivity in Isostructural Metal–Organic Frameworks with High Densities of Open Metal Sites. Angew. Chem. Int. Ed. 2012, 51 (8), 1857-1860. 14. Cadiau, A.; Adil, K.; Bhatt, P.; Belmabkhout, Y.; Eddaoudi, M., A Metal-Organic Framework–Based Splitter for Separating Propylene from Propane. Science 2016, 353 (6295), 137-140. 15. Bao, Z.; Alnemrat, S.; Yu, L.; Vasiliev, I.; Ren, Q.; Lu, X.; Deng, S., Adsorption of Ethane, Ethylene, Propane, and Propylene on a Magnesium-Based Metal–Organic Framework. Langmuir 2011, 27 (22), 13554-13562. 16. Eddaoudi, M.; Kim, J. Y.; Rosi, N. L.; Vodak, D.; Wachter, J.; O'Keeffe, M.; Yaghi, O. M., Systematic Design of Pore Size and Functionality in Isoreticular Mofs and Their Application in Methane Storage. Science 2002, 295, 469. 17. Li, K.; Olson, D. H.; Seidel, J.; Emge, T. J.; Gong, H.; Zeng, H.; Li, J., Zeolitic Imidazolate Frameworks for Kinetic Separation of Propane and Propene. J. Am. Chem. Soc. 2009, 131 (30), 10368-10369. 18. Pimentel, B. R.; Lively, R. P., Enabling Kinetic Light Hydrocarbon Separation Via Crystal Size Engineering of Zif-8. Ind. Eng. Chem. Res. 2016, 55 (48), 12467-12476. 19. Krishna, R., Tracing the Origins of Transient Overshoots for Binary Mixture Diffusion in Microporous Crystalline Materials. PCCP 2016, 18 (23), 15482-15495. 20. Ribeiro, A. M.; Campo, M. C.; Narin, G.; Santos, J. C.; Ferreira, A.; Chang, J.-S.; Hwang, Y. K.; Seo, Y.-K.; Lee, U.-H.; Loureiro, J. M., Pressure Swing Adsorption Process for the Separation of Nitrogen and Propylene with a Mof Adsorbent Mil-100 (Fe). Sep. Purif. Technol. 2013, 110, 101-111. 21. Plaza, M.; Ribeiro, A. M.; Ferreira, A.; Santos, J.; Lee, U.-H.; Chang, J.-S.; Loureiro, J. M.; Rodrigues, A., Propylene/Propane Separation by Vacuum Swing Adsorption Using Cu-Btc Spheres. Sep. Purif. Technol. 2012, 90, 109-119. 22. Plaza, M. G.; Ferreira, A. F. P.; Santos, J. C.; Ribeiro, A. M.; Müller, U.; Trukhan, N.; Loureiro, J. M.; Rodrigues, A. E., Propane/Propylene Separation by Adsorption Using Shaped Copper Trimesate Mof. Microporous Mesoporous Mater. 2012, 157, 101-111. 23. Lively, R. P.; Chance, R. R.; Kelley, B. T.; Deckman, H. W.; Drese, J. H.; Jones, C. W.; Koros, W. J., Hollow Fiber Adsorbents for Co2 Removal from Flue Gas. Ind. Eng. Chem. Res. 2009, 48 (15), 7314-7324. 24. Lively, R. P.; Leta, D. P.; DeRites, B. A.; Chance, R. R.; Koros, W. J., Hollow Fiber Adsorbents for Co2 Capture: Kinetic Sorption Performance. Chem. Eng. J. 2011, 171 (3), 801810. 25. Lively, R. P.; Bessho, N.; Bhandari, D. A.; Kawajiri, Y.; Koros, W. J., Thermally Moderated Hollow Fiber Sorbent Modules in Rapidly Cycled Pressure Swing Adsorption Mode for Hydrogen Purification. Int. J. Hydrogen Energy 2012, 37 (20), 15227-15240. 27 ACS Paragon Plus Environment


Page 28 of 29

26. DeWitt, S. J.; Sinha, A.; Kalyanaraman, J.; Zhang, F.; Realff, M. J.; Lively, R. P., Critical Comparison of Structured Contactors for Adsorption-Based Gas Separations. Annual Review of Chemical and Biomolecular Engineering 2018, (0). 27. Fan, Y.; Kalyanaraman, J.; Labreche, Y.; Rezaei, F.; Lively, R. P.; Realff, M. J.; Koros, W. J.; Jones, C. W.; Kawajiri, Y., Co2 Sorption Performance of Composite Polymer/Aminosilica Hollow Fiber Sorbents: An Experimental and Modeling Study. Ind. Eng. Chem. Res. 2015, 54 (6), 1783-1795. 28. Ohs, B.; Lohaus, J.; Marten, D.; Hannemann-Tamás, R.; Krieger, A.; Wessling, M., Optimized Hollow Fiber Sorbents and Pressure Swing Adsorption Process for H2 Recovery. Ind. Eng. Chem. Res. 2018, 57 (14), 5093-5105. 29. Chen, G.; Koros, W. J.; Jones, C. W., Hybrid Polymer/Uio-66 (Zr) and Polymer/Nay Fiber Sorbents for Mercaptan Removal from Natural Gas. ACS Appl. Mater. Inter. 2016, 8 (15), 9700-9709. 30. Pimentel, B. R.; Fultz, A. W.; Presnell, K. V.; Lively, R. P., Synthesis of Water-Sensitive Metal–Organic Frameworks within Fiber Sorbent Modules. Ind. Eng. Chem. Res. 2017, 56 (17), 5070-5077. 31. Zhang, C.; Lively, R. P.; Zhang, K.; Johnson, J. R.; Karvan, O.; Koros, W. J., Unexpected Molecular Sieving Properties of Zeolitic Imidazolate Framework-8. J. Phys. Chem. Lett. 2012, 3 (16), 2130-2134. 32. Zhang, C.; Gee, J. A.; Sholl, D. S.; Lively, R. P., Crystal-Size-Dependent Structural Transitions in Nanoporous Crystals: Adsorption-Induced Transitions in Zif-8. J. Phys. Chem. C 2014, 118 (35), 20727-20733. 33. Park, K. S.; Ni, Z.; Côté, A. P.; Choi, J. Y.; Huang, R.; Uribe-Romo, F. J.; Chae, H. K.; O’Keeffe, M.; Yaghi, O. M., Exceptional Chemical and Thermal Stability of Zeolitic Imidazolate Frameworks. Proc Natl Acad Sci U S A 2006, 103 (27), 10186-10191. 34. Grande, C.; Rodrigues, A., Adsorption Kinetics of Propane and Propylene in Zeolite 4a. Chem. Eng. Res. Des. 2004, 82 (12), 1604-1612. 35. Thomas, W. J.; Crittenden, B. D., Adsorption Technology and Design. ButterworthHeinemann: 1998. 36. Ruthven, D. M., Principles of Adsorption and Adsorption Processes. Wiley: New York, 1984. 37. Ruthven, D.; Farooq, S.; Knaebel, K., Pressure Swing Adsorption. VCH: New York. 38. Kalyanaraman, J.; Fan, Y.; Lively, R. P.; Koros, W. J.; Jones, C. W.; Realff, M. J.; Kawajiri, Y., Modeling and Experimental Validation of Carbon Dioxide Sorption on Hollow Fibers Loaded with Silica-Supported Poly (Ethylenimine). Chem. Eng. J. 2015, 259, 737-751. 39. Verploegh, R. J.; Nair, S.; Sholl, D. S., Temperature and Loading-Dependent Diffusion of Light Hydrocarbons in Zif-8 as Predicted through Fully Flexible Molecular Simulations. J. Am. Chem. Soc. 2015, 137 (50), 15760-15771. 40. Liu, D.; Ma, X.; Xi, H.; Lin, Y. S., Gas Transport Properties and Propylene/Propane Separation Characteristics of Zif-8 Membranes. J. Membr. Sci. 2014, 451, 85-93. TOC GRAPHIC


Page 29 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



Propylene Enrichment via Kinetic Vacuum Pressure Swing Adsorption

Recommend Documents