Enabling Kinetic Light Hydrocarbon Separation via Crystal Size

Subscriber access provided by UNIV OF WATERLOO

Article

Enabling kinetic light hydrocarbon separation via crystal size engineering of ZIF-8 Brian R. Pimentel, and Ryan P Lively Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.6b03199 • Publication Date (Web): 08 Nov 2016 Downloaded from http://pubs.acs.org on November 14, 2016

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 free 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 accessible to all readers and 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.

Industrial & Engineering Chemistry Research 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 33

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

Industrial & Engineering Chemistry Research

Enabling kinetic light hydrocarbon separation via crystal size engineering of ZIF-8 AUTHOR LIST Brian R. Pimentel and Ryan P. Lively*

ADDRESS School of Chemical & Biomolecular Engineering Georgia Institute of Technology 311 Ferst Drive NW, Atlanta, GA 30332

1 ACS Paragon Plus Environment


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 33

ABSTRACT

Recent increases in shale gas production provide an excellent opportunity for advancements in energy efficient separations of natural gas liquids, as these are increasingly used as fuel sources and chemical feedstock. Current fractionation schemes generally involve cryogenic distillation of C1-C4 hydrocarbons and are extremely energy intensive. Here, we describe the first steps towards a lower-energy, kinetic pressure swing adsorption cycle for the separation of ethane, propane, propylene, and butane using ZIF-8 as a diffusionally-selective adsorbent. Crystal engineering techniques were employed to control the diffusive time scale of the separation, allowing for multiple separations using the same adsorbent within reasonable process times. Equimolar separation of ethane/propane mixtures at 293 K exhibited separation factors of 2.7 in the gas phase under non-optimized conditions, which enhances the concentration of the feed mixture to 75 mol% propane. Separation performance was shown to improve to 3.8 at lower temperatures (81 mol% propane), which is attributed to differences in the activation energy of permeation of the two components. Propane/butane mixtures demonstrated a lower diffusive selectivity and almost negligible enhancement, while propylene/propane showed enhancement beyond ethane/propane due to a strong diffusive selectivity and sorption selectivities closer to unity. Single component adsorption and diffusion results were incorporated into a computational model of the system and was shown to be in relatively good agreement with experimental values. The model was used to predict separation system performance and recovery at various temperatures.

1. INTRODUCTION The last decade has seen dramatic increase in natural gas production in the United States from hydraulic fracturing of shale deposits and with it a growing supply of natural gas liquids as 2 ACS Paragon Plus Environment

Page 3 of 33

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


feedstock for fuel and the chemicals industry.1 Indeed, several ethane cracking facilities in North America have been announced looking to capitalize on this swell of hydrocarbons.2 This presents an opportunity to improve upon the highly energy-intensive cryogenic distillation trains currently utilized to separate natural gas liquids. The use of turbo-expanders to achieve sub-ambient temperatures and the subsequent enthalpic costs of boiling and condensing in distillation processes leaves significant room for improvements in the energy efficiency of natural gas liquid (NGL) separations.3-4 Although it is difficult to imagine the complete elimination of distillation columns from industrial separation trains, several alternative separation techniques could be used to de-bottleneck and improve efficiency of existing cold trains by providing a rough-cut of the incoming feed or in other hybrid adsorption-distillation configurations.5 Techniques such as membrane permeation and pressure-swing adsorption (PSA) systems are capable of NGL separations and are more energy efficient,4, 6 though have yet to see significant commercial application in this area. While inorganic membranes have demonstrated strong performance in light hydrocarbon separations,7-11 they tend to be expensive and difficult to produce in a defect-free manner at scale. Polymeric membranes are cheap and easy to mass produce, but lose selectivity due to plasticization by the strongly adsorbing species and therefore generally rely on solubility-driven selectivity.6, 12-15 Recent work using interfacial microfluidic processing has developed polymer-supported inorganic membranes with exceptional propane/propylene selectivities circumventing many of the traditional shortcomings of both technologies, although these membranes have not yet been produced beyond single fiber modules.16-17 Adsorptive separations take two forms, (1) equilibrium separations where one species adsorbs more strongly than the other, and (2) kinetic separations, where adsorption is governed by the



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 4 of 33

rate of uptake of each species and selectivity is inherently transient. Equilibrium PSA separation of hydrocarbons has largely focused on removal from hydrogen streams or olefin/paraffin separations, as the latter are some of the most difficult distillations in industry.18-22 The “heavier”, i.e. more strongly adsorbed component, is the adsorbed or retained species in equilibrium separations and therefore requires more complex cycles to recover at higher purities.23 Moreover, recovery of the heavy compound is typically at atmospheric pressure or under vacuum. Methods to collect the “heavy” compound at high purity without loss of partial pressure are needed. Diffusionally-selective materials, like carbon molecular sieves or cationexchanged zeolites, can be used in kinetic PSA schemes due to similarity in size between the micropore opening and the kinetic diameter of the gas molecule.23-26 Since diffusivity is generally inversely-proportional to molecule size, the higher carbon number compound in an NGL feed will be transiently enriched in the headspace over the sorbent, as the smaller (and lighter) compounds will more rapidly equilibrate with the sorbent. As of now, the only commercial application of kinetic separations exists in O2/N2 air separation using carbon molecular sieves and the ‘molecular gate’ process for CH4/N2,23, 27-28 while some research has explored CO2/CH4 and propane/propylene using carbon and cation-exchanged zeolites.29-31

Process cycles are governed by the diffusive time scale of the system, = /, where R is the

particle radius and D is the intracrystalline diffusivity. In many of the preferred adsorbents such as zeolites and carbons, large modification to particle size are difficult to achieve, leading to little control over process time. Metal-organic frameworks (MOFs) are commonly studied in gas adsorption literature due to their high surface area, pore volumes, and varied surface chemistry.32-36 Recently, several examples of small-pore MOFs have demonstrated strong diffusive selectivity for light gases in a


Page 5 of 33

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


variety of separations.37-43 Various researchers have also demonstrated control over the crystal size of these materials spanning several orders of magnitude.44-48 This so-called ‘crystal-size engineering’ opens the door for a much broader application of kinetic separation schemes, where previously prohibitively slow-diffusing gases can be separated via smaller crystals, while faster components can be separated by utilizing much larger crystals. Here, we demonstrate the use of a MOF material—ZIF-8—as a kinetic adsorbent for light gas mixtures relevant to the natural gas and chemical processing industry.

We measured the

temperature dependence of single component adsorption isotherms and diffusivity of light gases in ZIF-8. These measurements were utilized to build a model of the kinetic separation, and the results of this model were compared to proof-of-concept separation experiments. 2. MATERIALS AND METHODS 2.1 MATERIALS This work made use of the following chemicals: 1-Methylimidazole (99%, AlfaAesar), 2Methylimidazole (97%, AlfaAesar), Sodium Formate (98%, AlfaAesar), Methanol (ACS Grade, VWR), Zinc Nitrate Hexahydrate (99%, AlfaAesar). The following gases were used in this work: Nitrogen (UHP, Airgas), Ethane (99.99%, Airgas), Propane (99.5%, Tech Air), Propylene (99.5%, TechAir), n-Butane (99.5%, TechAir). All materials were used as received without any further purification. 2.2 MATERIAL SYNTHESIS 145 µm ZIF-8 ZIF-8 samples exhibiting 145 µm diffusive lengths were synthesized in a glass vial in methanol as per previous work.49 Briefly, 2.205 g of zinc nitrate hexahydrate were dissolved in



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 6 of 33

25 mL of methanol. 1.215 g of 2-methylimidazole and 0.504 g of sodium formate were dissolved in a separate 25mL 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 and sieved through 120 µm mesh. Samples were dried at 80 °C under vacuum overnight to remove adsorbed methanol. Under SEM, these crystals were shown to be either often cracked of broken, likely as a result of growing from a flat surface such as the vial wall. Due to these irregularities in geometry, these crystals were not used in diffusion measurements. 1.5 µm ZIF-8 ZIF-8 samples of 1.5 µm were synthesized in a glass vial in methanol as per previous work.46 Briefly, 0.734 g of zinc nitrate hexahydrate were dissolved in 50 mL of methanol. 0.810 g of 2methylimidazole and 0.810 g of 1-methylimidazole were dissolved in a separate 50 mL solution of methanol. The second solution was poured into the first and stirring was stopped after mixing. The vial was kept capped at room temperature for 24 h, then washed and filtered. Samples were dried at 80 °C under vacuum overnight to remove adsorbed methanol. 2.3 MATERIAL CHARACTERIZATION AND SINGLE COMPONENT MEASUREMENTS Crystals were characterized using powder X-ray diffraction (XRD, PANalytical X’pert PRO Multi-Purpose Diffractometer and PANalytical Empyrean with rotating sample stage), scanning electron microscopy (SEM, Hitachi SU8230 Cold Field Emission Scanning Electron Microscope), and cryogenic N2 physisorption (Microtrac BELSORP-max). All XRD patterns matched simulated structural reflections, while SEM imaging revealed sharply faceted rhombic dodecahedron crystals (Fig 1, Fig S1 & S2). N2 physisorption at 77K experiments revealed a high degree of microporosity within all ZIF-8 samples, with the characteristic ‘gate-opening’ step from 5 to 9 x10-3 P/P0 (Fig S3) and BET surface areas of 1350 m2/g (145 µm) and 1400


Page 7 of 33

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


m2/g (1.5 µm), and pore volumes of 0.64 cm3/g and 0.66 cm3/g respectively.50 From these values we believe that the samples are of similar quality and differences in internal microporosity should have minimal effect on their separation performance.

Figure 1. 145 µm diffusive length ZIF-8 crystal (left) and 1.5 µm radius ZIF-8 crystals showing sharp rhombic dodecahedral facets Single-component excess isotherms of ethane, propane, propylene and n-butane were measured via volumetric adsorption techniques in a Micromeritics HPVA-II from 253 K to 333 K, depending on the gas. Sub-ambient temperatures were accessed with the use of a cold head and associated Helium compressor from ColdEdge Technologies. Near-zero occupancy diffusion of propane and n-butane were also measured by pressure decay techniques, whereas ethane and propylene proved too fast to accurately measure. We assume that no difference in adsorption isotherms exist between crystals of different sizes. Diffusion measurements were carried out with approximately 10 mg of sample; the sample was dispersed on the walls of the sample cell with methanol to minimize interparticle resistance, and adsorption was calculated to be isothermal (over the time scales of diffusion) per the Ruthven-Lee isothermal criterion.51-52 Uptake curves were matched to the full analytical solution of sorbate diffusion from a finite volume into a sphere of a known particle size distribution.53 We



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 8 of 33

also confirmed that the uptake experiments were free of confounding valve resistances by following the procedure established by Brandani.54 The mathematical details of the Fickian diffusion curve fitting protocols are given in S3. 2.4 BATCH ADSORPTION OF MULTICOMPONENT MIXTURES Kinetic adsorption separation experiments

of ethane/propane, propane/butane,

and

propane/propylene mixtures were carried out in a homemade batch adsorption apparatus. In the unit, the evacuated sample (containing ZIF-8) was pressurized from the feed ballast for one second via a manual 3-way valve and then isolated for a set equilibration time (Scheme 1 top). A plot of the theoretical time-dependent adsorption behavior occurring during this initial exposure is shown in Scheme 1 (bottom). At predetermined time points, the sample volume was blown down into a gas chromatograph at atmospheric pressure and analyzed via flame ionization detector (FID). Blowdown to the GC was performed via a 1 second opening of the manual 3-way valve towards the GC then closed, in similar fashion to the dosing procedure. A real possibility of changes to the headspace composition exists via blowdown due to desorption, but for almost all time points investigated, the time of the desorption is much less (< 10x) than the allowed equilibration time, so we are comfortable with calling this an assumption of negligible error. The gas sampling line was flushed with inert prior to each time point to remove traces of old sample gases. All experiments were done in triplicate. Samples were evacuated overnight at room temperature before each initial time point after initial post-synthesis activation, and for ten times the previous exposure time before each new time point. Though the same feed ballast was used for all doses in a single run, the ballast pressure averaged 5 ± 0.5 bar over all time points for a given ballast composition. Feed composition was measured at the beginning and end of sample runs and was found to have no


Page 9 of 33

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


significant variation from the dosing composition. To illustrate the necessity of crystal size engineering, we began our experimental campaign with the use of 1.5 µm crystals, and later demonstrated the need for manipulation of the crystal dimension by duplicating one of the experiments with 145 µm samples and continuing to utilize these for further studies. It is important to point out that we directly utilized crystal size engineering to create diffusive length scales that permitted experimentation within reasonable time scales. Indeed, this same approach can be easily used to match larger process time scales. Scheme 1. (Top) Three step process schematic of experimental batch adsorption unit with (a) pressurization, (b) partial equilibration, and (c) headspace blowdown steps. (Bottom) Plot of (a)

t=0

(b)

t≤τ

(c)

Product enriched in “heavy” gas

t≥τ

adsorbed quantities versus square root time in a binary adsorption-diffusion process with a



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 10 of 33

fast/light component () and a slow/heavy component () where the blue line denotes the transition from kinetic to equilibrium control and Separation Factor plotted on the right axis ().

2.5 MULTICOMPONENT ADSORPTION MODELING FRAMEWORK The multicomponent adsorption and diffusion of gases in ZIF-8 was modeled using gPROMS Model Builder 4.0.0 to solve the transient PDE’s associated with the system. The mathematical framework was adapted from that put forth in previous published work

55

from a constant bulk

concentration step change to that of a depleting volume system via coupled mass balances between the adsorbed phase and the gas phase. This more accurately captures the time-dependent behavior of our system, especially in the cases of high component depletion where the driving force in the bulk begins to change dramatically.53 Equation 1 describes multicomponent intracrystalline diffusion, Equation 2 describes a competitive external surface concentration boundary condition, and Equation 3 provides the mass balance between the gas and adsorbed phase via an integration of the crystal concentration profile. Eq 2 uses a rapid exponential approach 1 − to minimize temporal discontinuities in the mass balance that could result from a step change in concentration. !

, , 1 = , #

, $%" = + =

"

&, ' (

1+

! Σ" ' (

1 −

(,, 0 − 6 , 4πr 9 . / 4 3 4 5 %" 3

Eq 1

Eq 2

Eq 3

Langmuir parameters, heats of adsorption, thermodynamic diffusivities and activation energies of diffusion were extracted from single component adsorption and diffusion measurements and 10 ACS Paragon Plus Environment

Page 11 of 33

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


used to predict mixture diffusion behavior, while values we were unable to measure were taken from previously published work. Model predictions include gas-phase separation factor and component recovery, as defined by fraction of dosed moles of component i remaining in the sample headspace. Modeling results were compared to experimental data and used to predict behavior under varying temperature and operating conditions. A complete mathematical description of the system model is provided in S1. 3. RESULTS AND DISCUSSION 3.1 SINGLE COMPONENT ADSORPTION AND DIFFUSION Single component gas isotherms show a typical Langmuir response in their uptake (Fig 2), with n-butane as the highest carbon number molecule having the steepest response. No evidence of flexible ‘gate-opening’ transitions were observed. Isosteric heats of adsorption (Fig 3) show an upward trend uncorrelated to sorbate latent heats of vaporization. This upward trend has been previously attributed to both non-polar interactions and changing adsorbate molar volume with system temperature.56 Increasing sorption coefficients (the secant line on the isotherm from the initial adsorption pressure to the equilibrium pressure) with increasing carbon number highlights the difficulty of adsorptive processes that are not based on equilibrium adsorption. Ideal selectivity within a particle of a diffusive process under linear isotherm conditions is given by the ratio of the leading terms of the short-time solution: 52 :; =

< = exp D− whereby substitution results in

GH F /

GI = GJ + EK

Eq 8

Eq 9

Evaluating the activation energy of permeation of ethane and propane, we see that propane yields an GH ≈ 15 kJ/mol, indicating an increase in permeation rates at higher temperatures.

Ethane however, yields GH ≈ 0 kJ/mol, showing little temperature dependence of permeation. This would indicate that the kinetic separation of ethane and propane would be more favorable at lower temperatures, i.e. the kinetic separation unit will more favorably adsorb ethane in short times, thereby generating a purified propane product closer to its original partial pressure in the feed. An interesting demonstration of this principle in membranes was conducted by Kapteijn et al.59

3.2 MIXED GAS BATCH ADSORPTION Batch adsorption of mixed gas feeds was employed as the simplest way to demonstrate the applicability of ZIF-8 to kinetic adsorption schemes. Separation factor (not selectivity) was chosen to report compositional changes, defined for the gas phase as:60 Q ,R Q,T

Enabling Kinetic Light Hydrocarbon Separation via Crystal Size

Recommend Documents