Pressure Swing Adsorption Separation of H2S/CO2/ CH4 Gas Mixtures with Molecular Sieves 4A, 5A, and 13X

Home , Membrane gas separation, Molecular sieve

Separation Science and Technology

ISSN: 0149-6395 (Print) 1520-5754 (Online) Journal homepage: http://www.tandfonline.com/loi/lsst20

Pressure swing adsorption separation of H2S/CO2/ CH4 gas mixtures with molecular sieves 4A, 5A, and 13X

Howell H. Heck, Merilyn L. Hall, Rudy dos Santos & Manolis M. Tomadakis

To cite this article: Howell H. Heck, Merilyn L. Hall, Rudy dos Santos & Manolis M. Tomadakis (2018) Pressure swing adsorption separation of H2S/CO2/CH4 gas mixtures with molecular sieves 4A, 5A, and 13X, Separation Science and Technology, 53:10, 1490-1497, DOI: 10.1080/01496395.2017.1417315 To link to this article: https://doi.org/10.1080/01496395.2017.1417315

Published online: 26 Dec 2017.

Submit your article to this journal

Article views: 101

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=lsst20 SEPARATION SCIENCE AND TECHNOLOGY 2018, VOL. 53, NO. 10, 1490–1497 https://doi.org/10.1080/01496395.2017.1417315

Pressure swing adsorption separation of H2S/CO2/CH4 gas mixtures with molecular sieves 4A, 5A, and 13X Howell H. Hecka, Merilyn L. Hallb, Rudy dos Santosb, and Manolis M. Tomadakisb aDepartment of Civil Engineering, College of Engineering, Florida Institute of Technology, Melbourne, FL, USA; bDepartment of Chemical Engineering, College of Engineering, Florida Institute of Technology, Melbourne, FL, USA

ABSTRACT ARTICLE HISTORY Pressure swing adsorption experiments were carried out for the separation of equimolar mixtures Received 18 October 2016 of carbon dioxide and methane containing small amounts of hydrogen sulfide, utilizing 4A, 5A, Accepted 11 December 2017 and 13X molecular sieves. High-purity methane of zero or nearly zero hydrogen sulfide concen- KEYWORDS tration was produced in the adsorption stage with 13X and 5A sieves, at high product recovery H2S; CO2;CH4; molecular rates; high-purity carbon dioxide was obtained with the same sieves in the desorption stage. sieves; pressure swing Zeolite 4A was found capable of raising considerably the hydrogen sulfide concentration in the adsorption accumulated desorption product (vs. the adsorption feed) at high recovery rates too. Adsorption selectivity values derived from the experimental results for all three gas pairs were in line with some theoretical predictions and experimental data of the literature.

Introduction different gases including biogas, using a molecular modeling approach. The results indicated that the 13X and 5A The separation of ternary gas mixtures of H S, CO ,and 2 2 zeolites were the top candidates for H SandCO removal CH has been the subject of a number of experimental and 2 2 4 from H S/CO /CH biogas systems, based on selectivity, theoretical investigations, mostly using various types of 2 2 4 – whereas 4A was also ranked among the best four adsor- membranes as the separation agent.[1 4] Adsorption has bents for this separation. The model employed pressures also been studied experimentally and theoretically as a up to 4 MPa and a temperature of 303 K, with a biogas method for separating these gases, often from biogas or consisting of 70% CH ,29.8%CO, and 0.2% H S. natural gas, using activated carbons or carbon molecular 4 2 2 – The removal of H S from its ternary mixtures with sieves,[5 8] mesoporous silica,[9] multilayer graphene 2 CH and CO has also been studied by Cosoli et al.,[14] nanostructures,[10] or dolomite suspensions.[11] Asmall 4 2 who used grand canonical Monte Carlo (GCMC) simu- number of literature studies focused on pressure swing lations to model biogas adsorption on different zeolites, adsorption (PSA), some of them using 4A, 5A, and – namely, FAU NaX, FAU NaY, LTA, and MFI. The 13X[12] or similar zeolites,[13 15] and others utilizing differ- model used a biogas with partial pressures of H S ran- ent sieves such as Si-CHA,[16] naturally occurring zeolites,- 2 ging from 0.01 to 1 kPa at 1 atm and 298 K. The results [17,18] or a mixture of activated carbon and 13X zeolite.[19] indicated that zeolite FAU NaY exhibits the highest Other studies investigated the separation of H S, CH , and/ 2 4 selectivities for H S over CH and CO (ranging from or CO from natural gas by means of both activated carbon 2 4 2 2 300 to 1000 for H S/CH and 10 to 100 for H S/CO ), and membranes,[20,21] or gas hydrate crystallization.[22] 2 4 2 2 whereas MFI exhibits the lowest values of these pairwise Literature reviews of the employed membrane separation- selectivities (ranging from 0.10 to 0.20 for H S/CH and [23,24] and biogas purification[25] processes have also been 2 4 0.25 to 1.8 for H S/CO ). Cosoli et al.[15] coupled the reported. 2 2 GCMC with molecular dynamics simulations to further To our knowledge, only one journal article reported on investigate the adsorption of low-pressure H S and its the separation of ternary mixtures of H S, CO ,andCH 2 2 2 4 selective adsorption behavior toward CO and CH for using molecular sieves 4A, 5A, and 13X: Peng and Cao[12] 2 4 each of the previously mentioned four sieves. Zeolite carried out a theoretical investigation of 18 types of por- FAU NaY was once again found to be the best option ous materials, including carbons, zeolites, and metalor- for H2S removal from the simulated biogas mixtures. ganic frameworks, for the removal of H2SandCO2 from

CONTACT Manolis M. Tomadakis [email protected] College of Engineering, Florida Institute of Technology, 150 W. University Blvd., Melbourne, 32901 Florida, United States. Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/lsst. © 2018 Howell H. Heck, Merilyn L. Hall, Rudy dos Santos, and Manolis M. Tomadakis SEPARATION SCIENCE AND TECHNOLOGY 1491

Gholampour and Yeganegi[26] created a constricted slit Table 1. Molecular sieve properties.[27]. pore model for nanoporous carbon (NPC) and applied it Sieve type in a molecular simulation study of the separation of 4A 5A 13X Characteristic pore size (Å) 4 5 10 binary and ternary mixtures of CH4,CO2, and H2S. Pellet size (in.) 1/16 1/16 1/8 Adsorption selectivities were also calculated for both Bed porosity (dim/less) 0.37 0.38 0.41 binary and ternary mixtures of the gases and compared Bulk density (lb/ft3) 41 43.4 44 Heat capacity (Btu/lb°F) 0.23 0.23 0.23 to literature values for the corresponding selectivities of zeolites 13X, ZIF-8, NaX, LTA, NaY, MFI, and Si-CHA. This study is the first experimental investigation of composition, and adsorption takes place. The column is the separation of ternary mixtures of H2S, CH4, and then blown down to atmospheric pressure, and the bed is CO2 by PSA using molecular sieves 4A, 5A, or 13X. purged with nitrogen for desorption to occur. Throughout Equimolar mixtures of CH4 and CO2 are utilized, con- the adsorption and desorption steps, temperatures and flow – taining 0.1 0.7% H2S, undergoing adsorption at pres- rates are recorded at regular intervals, and gas samples are sures of 50–90 psig. In addition to the measured drawnfromthesamplingportsontheinletandoutlet adsorption and desorption product compositions and streams with pressure-tight valved syringes and are ana- recovery values, pairwise adsorption selectivities are lyzed by the gas chromatograph. Figure 1 illustrates the four also derived and compared to theoretical predictions steps of the PSA process, highlighting the main products and experimental data of the literature. obtained in the adsorption and desorption stages. A total of nine PSA experiments were carried out: five with molecular sieve 13X, two with 4A, and two Experiments with 5A. Table 2 presents the type of molecular sieve, The PSA apparatus consists of an adsorption column, a column pressure, feed flow rate, and H2S feed concen- reciprocating compressor, and a Gow-Mac 600 gas tration employed in each adsorption run. As shown in chromatograph. The adsorption bed is 2 feet long and the table, the adsorption feed pressure was maintained 3 inches in internal diameter, made of stainless steel, at 70 psig for seven of the nine runs, while the other and fitted with thermocouples, pressure gages, rota- two experiments were performed at a lower (50 psig) or meters, and sampling ports on all inlet and outlet higher (90 psig) pressure, to investigate the effect of lines. The feed gases are 99.5% chemically pure H2S, pressure on process performance. Similarly, the H2S 99.8% industrial grade CO2, and 99% commercial grade feed concentration was assigned values of 0.1%, 0.5%, CH4; ultrahigh purity 99.998% nitrogen is used as the and 0.7% in various runs, so that the effect of this purge gas. The solid adsorbents are molecular sieves 4A parameter on process performance is also examined. by UOP and 5A and 13X by W. R. Grace; Table 1 presents some characteristic sieve properties. Results and discussion First, the column is pressurized to the desired adsorption pressure, using methane. Next, the mixture of H2S, CO2, Figure 2 presents the CO2 breakthrough curves for experi- and CH4 is supplied to the bed at the predetermined ments A1, A3, A5, and A8. Three of the four curves are

Figure 1. PSA process steps highlighting main adsorption and desorption products in bold. 1492 H. H. HECK ET AL.

Table 2. Adsorption process parameters. breakthrough times of all gases (tH2S, tCO2, tCH4) for Adsorption feed each run; as indicated in the table, breakthrough times Pressure Flow rate H2S content were in some cases too small to be detected in time Sieve Run# (psig) (gmol/min) (mol %) with the manual syringe sampling system. The smallest 13X A1 70 0.66 0.7 A2 70 0.65 0.5 breakthrough times of all gases were obtained with A3 70 0.65 0.1 molecular sieve 4A, which encounters limited rates of A4 90 0.56 0.1 A5 50 0.60 0.1 adsorption, as also concluded from other results of this 4A A6 70 0.65 0.1 study. The two gases supplied at equimolar rates, CO A7 70 0.65 0.5 2 5A A8 70 0.65 0.5 and CH4, are found to have significant differences in A9 70 0.65 0.1 breakthrough times and amounts adsorbed. CH4 breaks through much faster in all seven 13X and 5A runs and adsorbs less or much less than CO2 in all runs. This is primarily due to the significant difference in the shape of the two molecules since the linear carbon dioxide is admitted readily to the pores of the zeolite particles, unlike the tetrahedral methane.

Adsorption selectivity

An adsorption selectivity factor was introduced by Ruthven,[28] equivalent to the relative volatility for dis- tillation: Sij = (xi/yi)/(xj/yj), where x and y represent the mole fractions of gas species i, j in the adsorbed phase and the gas phase, respectively. Using the mole fraction data from Tables 2 and 3, and accounting also for the equimolar supply of CO2 and CH4 to the adsorption bed, the values of all three pairwise selectivity factors are readily calculated and presented in Table 3. The H2S/CO2 selectivity values for the 4A zeolite are practically equal to those measured by Tomadakis et al.[27] Figure 2. CO2 breakthrough curves for adsorption runs A1, A3, A5, and A8. for adsorption from a binary mixture of the two gases on the same sieve. This is not the case for the other two sieves, where the H2S/CO2 selectivity values reported in quite steep and indicate completion of CO2 breakthrough that study vary from 0.9 to 11.9, whereas in this work 30–45 min into the run. The exception to that is the curve they are found to be not higher than 1.5. This should be for Experiment A1, the only run that utilized fresh adsor- attributed to the presence of methane, which may bent, which encountered the longest breakthrough curve, compete more with hydrogen sulfide than carbon diox- in agreement with similar observations reported by ide for adsorption in those sieves, due in part to its [27] Tomadakis et al. for H2S/CO2 adsorption in the same molecular shape being somewhat similar to that of the molecular sieves. It should be noted that the adsorption former (i.e., highly nonlinear) but very different than product concentration of the two major components the linear carbon dioxide molecule. This is in line with (CH4 and CO2) approached the feed gas composition (i. the results of the experimental investigation by e., the 50/50 limit) at long process times in all runs as Mofarahi and Gholipour,[29] who report ideal (non- anticipated by basic adsorption theory and intuition. The competitive) CO2/CH4 selectivity data for adsorption maximum temperature rise from the ambient tempera- on zeolite 5A that are in agreement with the 5A selec- ture of 23–25°C varied from 12°C to 17°C for the 5A and tivity values obtained in this work for the two gases. – 13X adsorption runs, but only 4 7°C for the 4A runs, due The authors measured the CO2/CH4 selectivity at var- to the lower rates of adsorption obtained with this rela- ious pressures and temperatures (namely, 1–6 bar and tively small-pore sieve, in line with temperature variations 273–343 K) and report a value of about 2.2 at 70 psig encountered in commercial PSA processes. (5.8 bar) and 303 K, in line with our own selectivity Table 3 presents experimental data collected or data reported on Table 3 for adsorption of the two derived from the adsorption runs, including the pres- gases at the same pressure and temperature (from a surization time (tp), total process time (ta), and the mixture containing also small amounts of H2S). The SEPARATION SCIENCE AND TECHNOLOGY 1493

Table 3. Experimental data collected or derived from adsorption runs. Time (min) Adsorbed gas (mol %) Adsorption selectivity

Sieve Run# tp ta tH2S tCO2 tCH4 H2SCO2 CH4 H2S/CO2 H2S/CH4 CO2/CH4 13X A1 7 120 7 22 10 1.1 63.9 35.0 1.2 2.2 1.8 A2 7 18 17 17 8 1.0 85.1 13.9 1.2 7.2 6.1 A3 6 26 9 17 9 0.1 52.6 47.2 1.0 1.1 1.1 A4 8 29 11 21 <11 0.2 95.0 4.8 1.1 20.8 19.8 A5 5 44 16 21 <8 0.3 99.7 –a 1.5 –– 4A A6 2 30 4 5 <4 0.2 79.6 20.2 1.3 5.0 3.9 A7 2 28 <5 <5 <5 0.9 50.6 48.5 1.8 1.9 1.0 5A A8 8 32 <9 21 9 0.7 64.0 35.3 1.1 2.0 1.8 A9 7 33 18 15 <9 –a 62.9 37.1 ––1.7 aThe gas was not detected in the adsorption outlet stream.

H2S/CH4 and CO2/CH4 selectivity values reported on in 4A and 5A molecular sieves. The authors computed Table 3 are not much different from each other as the H2S selectivity over CO2 and CH4 with the H2S feed anticipated from the fact that the H2S/CO2 selectivities content ranging from 0.01% to 1%, and encountered a are generally close to 1. All three sieves are found to small or negligible effect of that parameter on the have a higher or much higher affinity for H2S and CO2 selectivity for all but the smaller H2S feed contents – than CH4, as is also evident from direct inspection of (0.01 0.1%), where a significant effect was observed. the adsorbed amounts of the three gases in Table 3 and Their predictions for the H2S/CO2 selectivity of the comparison to the corresponding feed gas concentra- MFI zeolite at 0.1% and 0.5% H2S feed are 0.87 and tions in Table 2. 1.23, very close to the experimental value of 1.1 Peng and Cao[12] computed the three pairwise selectiv- obtained in this study for the 5A molecular sieve, for ities for adsorption of H2S/CO2/CH4 gas mixtures (0.2% a 0.5% H2S feed content. The corresponding selectivity H2S, 29.8% CO2,70%CH4) in molecular sieves 4A, 5A, predictions for the other three zeolites are moderately and 13X at 303 K over a broad pressure range (0.1– higher than those measured in our work (3.6–7.1 for 40 atm), and encountered some interesting trends on LTA and FAU NaX, and 15–30 for FAU NaY). The [14] the selectivity variation with pressure, which are discussed H2S/CH4 selectivities computed by Cosoli et al. for in the following paragraphs with reference to the results of the MFI sieve for the same H2S feed concentrations are other studies. As in our investigation, the highest selectiv- lower than those measured in our study (0.13–0.15), [12] ities obtained by Peng and Cao are those for H2Sover and those for the other three sieves are considerably – – CH4. However, for the pressures employed in our work higher (56 107 for LTA and FAU NaX, and 316 355 (50–90 psig or 4.4–7.1 atm), the authors report selectivity for FAU NaY), still significantly smaller than the cor- values that are moderately or significantly higher than responding predictions by Peng and Cao.[12] Given that those measured in our study for equimolar CO2/CH4 the bulk of our results were obtained at 70 psig feeds. McEwen et al.[30] processed experimentally (5.8 atm) but those of Cosoli et al.[14] at 1 atm, these obtained pure component isotherms using the Ideal observations are in agreement with the trends encoun- [12] Adsorbed Solution Theory to derive CO2/CH4 adsorption tered in the numerical study by Peng and Cao, who selectivities for zeolite 13X at 1 bar and 298 K. They predicted a sharp increase in the H2S/CH4 selectivity of encountered an exponential increase in selectivity with some zeolites (13X and 5A) as the pressure drops from CH4 mole fraction; for equimolar CO2/CH4 feed mix- about 10 atm to 1 atm and below, but a drop in the tures, such as those used in our study, they report a selectivity of another sieve (4A) over the same pressure selectivity value of 16, which is within the range of values range. It should be noted that Peng and Cao[12] also measured in our work. For a feed of 70% CH4,similarto predicted a practically constant H2S/CO2 selectivity for [12] that used by Peng and Cao, the authors derived a CO2/ the 5A sieve over the same pressure range. The H2S/ [14] CH4 selectivity value of about 18.5, five times smaller than CH4 selectivity values reported by Cosoli et al. are [12] that reported in ref. for the same pressure and practi- higher than the H2S/CO2 selectivities derived in the – cally same temperature. same study over the 0.1 0.5% H2S feed range for [14] Cosoli et al. simulated the H2S removal from CO2 three of the four sieves utilized. The same trend is and CH4 in biogas mixtures by adsorption at 1 atm in observed in the corresponding numerical results by various commercial zeolites, namely, FAU (NaX, NaY), Peng and Cao[12] and the experimental selectivities which are Faujasite-type sieves like 13X; LTA, with obtained in this work. pore diameters mostly higher than 6 Å; and MFI, Gholampour and Yeganegi[26] developed a con- whose pore sizes are primarily in the range 4–6Å,as stricted slit pore model to simulate the adsorption of 1494 H. H. HECK ET AL.

binary and ternary gas mixtures of H2S, CO2, and CH4 corresponding best adsorption product composition for in NPC for the pressure range of 0.1–20 atm. Their six experimental runs carried out with these two sieves. results indicate a negligible nonsystematic effect of the As shown in the table, complete or nearly complete presence of the third component on the selectivity of removal of H2S was achieved in all runs for short the other two gases of the mixture, as well as a very periods of time, along with nearly complete purification weak effect of pressure on selectivity. The authors of CH4. This is a promising result that can be utilized in explored the applicability of their model to other adsor- rapid-PSA system design for commercial application of bents as well, including zeolites such as 13X, NaX, NaY, the separation process. It should be noted that the CH4 LTA, and MFI, and encountered the closest agreement product concentration is limited by the presence of with a Si-CHA zeolite H2S/CH4 selectivity measure- nitrogen in the system (since the methane feed consists ment of the literature. A comparison of the slit pore of 99% CH4 and 1% nitrogen and hydrocarbons). The model predictions at 70 psig (5.8 atm) with our experi- average molar ratio of CH4 to CO2 in the adsorption mental results obtained at the same pressure indicates gas product accumulated over 30-min runs, using equi- good agreement for all three pairwise selectivities. molar feeds of the two gases, is 4.6 to 1 for the 13X Specifically, the values predicted by the model for the sieve, 3.3 to 1 for the 5A, and 1.9 to 1 for the 4A zeolite. H2S/CO2 (2% H2S), H2S/CH4 (0.2% H2S), and CO2/ CH4 (10% CO2) selectivities, for the full range of con- – stricted pore diameters (5 Å 20 Å), vary in the range Desorption product composition 1.5–2.2, 2.7–5.3, and 2.2–4.0, respectively, in line with the corresponding values measured in this study for The composition of the gas product collected in each molecular sieves 4A, 5A, and 13X, with pore sizes in desorption run is presented in Table 5, which shows practically the same range. Both studies encounter the also the supply flow rate of nitrogen purge gas in each highest selectivities for H2S over CH4, which are no run, and its ratio to the corresponding total adsorption more than 50% higher than those for CO2 over CH4, feed flow rate (P/F); this ratio is shown to vary in the and no more than twice higher, in most cases, than range 1–1.2, i.e., the optimum range for efficient adsor- [32] those of H2S over CO2. The much higher H2S/CH4 and bent regeneration reported by Yang. The nitrogen CO2/CH4 selectivity values obtained in our work at 90 purge gas was not included in the composition analysis psig are not predicted by the slit pore model of since it can be readily removed from the desorption Gholampour and Yeganegi.[26] Nonetheless, our 90 product through other separation processes, such as psig CO2/CH4 selectivity value of 19.8 (obtained with cryogenic techniques. As deduced from this data and a 13X sieve) is in line with the corresponding value of illustrated in Fig. 3, the average H2S concentration in [31] – about 23 reported by Ghoufi et al. for adsorption in the desorption product %H2S in the total moles of Faujasite NaY at the same pressure from an equimolar desorption product obtained with each sieve – is higher mixture of the two gases; the authors combined experi- than that in the adsorption feed for all sieves, but much mental techniques with GCMC simulations and derived higher for zeolite 4A, where it improves by one full CO2/CH4 selectivity values ranging from 24 to 14 for order of magnitude (i.e., from 0.3% to 2.9%). pressures varying from 1 to 30 bar. Nonetheless, CO2 dominates the desorption product, with molar ratios to CH4 that vary from about 1000–5000 (5A) to 5–1000 (13X) and 1.5–30 (4A), Methane product purity and concentrations as high as 98.6–99.6% in five 13X The experimental results indicate that two of the three and 5A runs. Table 5 presents also the duration of each molecular sieves tested (13X and 5A) are capable of desorption experiment, ranging from 62 to 83 min for producing high-purity methane (98% or more) of zero most runs; the blowdown step that preceded each des- or nearly zero H2S concentration. Table 4 presents the orption run lasted less than 1 min in all cases.

Table 4. Highest methane purity adsorption product. Product composition (%) H2S feed Pressure Time of highest Sieve Run# (%) (psig) CH4 Purity (min) H2SCO2 CH4 13X A2 0.5 70 10–21 0 0.1–0.3 94–98 A3 0.1 70 12–17 0 0.2–0.7 98 A4 0.1 90 12–19 0–0.01 0.6–1.3 95–98 A5 0.1 50 9–12 0.005–0.03 0.2–0.8 95–98 5A A8 0.5 70 13–17 0 0.0–0.02 98 A9 0.1 70 9–16 0.006–0.04 0.1–0.6 97–98 SEPARATION SCIENCE AND TECHNOLOGY 1495

Table 5. Desorption process parameters and product Table 6. Adsorption and desorption product recovery. composition. Adsorption recovery (%) Desorption recovery (%)

Product Sieve Run# H2SCO2 CH4 H2SCO2 composition (%) Purge flow rate Flow rate Time 13X A1/D1 25 37 65 12 20 Sieve Run# (gmol/min) ratio (P/F) (min) H2SCO2 CH4 A2/D2 0.3 15 86 54 80 13X D1 0.64 1.0 98 1.0 94.2 4.8 A3/D3 10 30 36 9 59 D2 0.70 1.1 63 0.8 99.1 0.1 A4/D4 6 19 96 100 100 D3 0.70 1.1 62 (0.04) 98.9 1.1 A5/D5 13 37 100 100 51 D4 0.70 1.2 83 0.8 82.7 16.5 4A A6/D6 1 22 80 100 46 D5 0.70 1.2 75.5 0.5 99.2 0.3 A7/D7 6 48 51 97 29 5A A8/D8 12 21 56 43 36 4A D6 0.70 1.1 129 1.9 57.9 40.2 –a D7 0.70 1.1 71 5.6 91.3 3.1 A9/D9 100 29 57 45 5A D8 0.63 1.0 69 1.3 98.6 0.1 aNot determined since adsorbed amount was not detected (Table 3). D9 0.69 1.1 60 0.4 99.6 (0.02)

such high rates are indeed achieved in most experi- Product recovery ments, topped by 100% in 13X and 4A runs D4, D5, The fractional recovery of gas products in the adsorp- and D6, and 97% in 4A run D7. The average adsorption tion and desorption stages is one of the factors deter- product recovery rate of CH4 with each molecular sieve mining the separation efficiency of a PSA process. The used in this work ranges from 60% to 70% (60% for 5A, adsorption recovery of a gas species is defined as the 67% for 4A, and 70% for 13X), much higher than that fraction of the total mass of the gas supplied to the of any other gas in this process. Similarly, the average column that is recovered as product in the adsorption H2S desorption recovery varies from 23% to 72% (23% outlet stream. Similarly, the desorption recovery of a for 13X, 44% for 5A, and 72% for 4A), encountering gas is defined as the fraction of the total mass of the gas the highest values among all three gases. loaded to the bed in the adsorption stage that is recovered as product in the desorption outlet stream. Table 6 presents the adsorption recovery rates obtained for Pressure effect each of the three gases in the feed mixture, as well as the desorption recovery rates for the two main deso- The effect of adsorption pressure on separation effi- – rption gas products, for all experimental runs of this ciency was investigated through runs A3 A5, all three carried out with a 13X sieve and 0.1% H2S feed. Table 7 study. Since CH4 is the only useful product in the summarizes the strongest systematic effects of pressure adsorption stage, high CH4 product recovery in that on process performance encountered in these runs. As step is essential. Indeed, very high CH4 product recovery values are achieved in most runs, topped by 96% is evident in this table, the H2S and CO2 adsorption and 100% in 13X runs A4 and A5. Similarly, high recovery rates decrease with pressure since higher amounts of these gases are adsorbed at higher pres- desorption recovery rates are desired for H2S, the main desorption gas product; as is evident in Table 6, sures. The CO2 desorption recovery increases with adsorption pressure, indicating that a higher fraction of the adsorbed CO2 is desorbed from the bed for a higher adsorption–desorption pressure swing. High pressure appears to favor much more the adsorption of CH4 rather than that of CO2. Specifically, the adsorbed amount of CH4 recovered in the desorption product is much higher at 90 psig than at any lower pressure, thus reducing the CO2 desorption product purity to 82.7% from its highest value of 99.2% obtained at 50 psig.

Table 7. Pressure effect on process parameters (13X sieve/0.1% H2S feed). Adsorption Desorption recovery (%) product (%) Pressure Desorption CO2 Run# (psig) H2SCO2 % recovery CO2 CH4 A4 90 6 19 100 82.7 16.5 Figure 3. Average H2S concentration in desorption product A3 70 10 30 59 98.9 1.1 versus adsorption feed. A5 50 13 37 51 99.2 0.3 1496 H. H. HECK ET AL.

H2S feed concentration effect the 13X sieve, 3.3 to 1 for the 5A, and 1.9 to 1 for the 4A zeolite. The average H S concentration in the accu- The effect of H S feed concentration on process per- 2 2 mulated desorption product is higher than that in the formance was studied through runs A1–A3 (performed adsorption feed for all sieves, but much higher for at 70 psig with a 13X sieve), A6–A7 (70 psig/4A sieve), zeolite 4A, where it improves by one full order of and A8–A9 (70 psig/5A sieve). Naturally, the H S feed 2 magnitude (2.9% vs. 0.3%). This is another promising concentration is found to affect systematically the con- result, supporting the potential of multistage PSA sys- centration of H S in both the adsorbed gas and the 2 tems in providing hydrogen fuel production processes desorption gas product. The average H S concentration 2 with a concentrated H S feed stream. The average H S in the adsorbed gas for all three sieves is 0.9% for a feed 2 2 concentration in the desorption product for all three of 0.5% H S (hence improved by a factor of 1.8) and 2 sieves is 2.4% for a feed of 0.5% H S, which represents 0.15% for a feed of 0.1% H S (improved by a factor of 2 2 an improvement by a factor of 4.8. For a feed of 0.1% 1.5). The corresponding average H S concentrations in 2 H S, the corresponding average H S concentration is the desorption product are 2.4% and 0.4% for feeds of 2 2 0.4%, an improvement by a factor of 4. 0.5% and 0.1% H S, respectively, therefore improved by 2 High CH recovery rates were achieved in most factors of 4.8 and 4. In both cases, we observe a 20% 4 adsorption experiments, averaging at 60–70% for all stronger H S enrichment effect at the higher feed con- 2 sieves, and topped by 100% in certain 13X runs. centration level (1.8/1.5 = 4.8/4 = 1.2). The most pro- Similarly, high H S recovery rates were achieved in nounced improvement in H S concentration (in the 2 2 most desorption experiments, reaching 100% in some collected desorption product vs. the feed) is encoun- 13X and 4A runs; the highest average H S desorption tered with the 4A sieve, where the concentration is 2 recovery rate among all sieves was 72%, achieved with raised on the average by one full order of magnitude, zeolite 4A. The strongest pressure effects on process as also illustrated in Fig. 3. parameters were encountered for the H2S and CO2 adsorption recovery, CO2 desorption recovery, and CO and CH desorption product concentrations. Conclusions 2 4 Specifically, a higher adsorption pressure resulted in a A series of adsorption and desorption experiments were lower H2S and CO2 adsorption recovery, higher CO2 carried out to evaluate the potential of PSA in separat- desorption recovery, and a higher relative concentra- ing ternary gas mixtures of methane, carbon dioxide, tion of CH4 to CO2 in the desorption gas product; most and hydrogen sulfide. Three types of solid adsorbents effects can be attributed to the higher adsorption rates were utilized for this purpose, namely, a 4A molecular achieved at higher pressures for all three gases. sieve by UOP, and 5A and 13X sieves by W. R. Grace. Pairwise adsorption selectivity values derived from The H2S feed volume fraction was varied in the range the experimental results for all three pairs of gases in 0.1% to 0.7%, to determine the effect of H2S concentra- the ternary mixture were found to be in line with some tion on PSA process efficiency. The adsorption pressure theoretical predictions and experimental data reported was also varied in the range of 50–90 psig, to evaluate in the literature. its effect on process performance. The experimental results indicate that molecular Funding sieves 13X and 5A are capable of producing high purity methane (up to 98%) of zero or nearly zero H2S con- This work was supported by Florida Center for Solid and centration, for short periods of time in the adsorption Hazardous Waste Management (Grant ID: UF-EIES- stage of the process, and high-purity carbon dioxide 0332005-FIT). (up to 99.6%) in the desorption stage. These are promising results that can be utilized in conventional or rapid PSA system design for commercial application of References the separation process. The achieved methane purity is [1] Vaughn, J.T.; Koros, W.J. (2014) Analysis of feed limited by the presence of nitrogen in the CH4 gas – stream acid gas concentration effects on the transport cylinder supply (99% CH4 1% nitrogen and hydrocar- properties and separation performance of polymeric bons) and could reach higher yet levels in commercial membranes for natural gas sweetening: A comparison systems. between a glassy and rubbery polymer. The Journal of – The average molar ratio of CH4 to CO2 in the Membrane Science, 465 (1): 107 116. adsorption gas product collected over 30-min runs, [2] Maghsoudi, H.; Soltanieh, M. (2014) Simultaneous using equimolar feeds of the two gases, is 4.6 to 1 for separation of H2S and CO2 from CH4 by a high silica SEPARATION SCIENCE AND TECHNOLOGY 1497

CHA-type zeolite membrane. The Journal of the all-silica DD3R zeolite. Adsorption, 19 (1): 1045– Membrane Science, 470 (1): 159–165. 1053. [3] Kraftschik, B.; Koros, W.J.; Johnson, J.R.; Karvan, O. [17] Pande, D.R.; Fabiani, C. (1989) Feasibility studies on (2013) Dense film polyimide membranes for aggressive the use of a naturally occurring molecular sieve for sour gas feed separations. The Journal of Membrane methane enrichment from biogas. Gas Separation and Science, 428 (1): 608–619. Purification, 3 (1): 143–147. [4] Makaruk, A.; Miltner, M.; Harasek, M. (2013) Biogas [18] Axente, D.; Abrudean, M.; Bâldea, A. (1983) Adsorption on desulfurization and biogas upgrading using a hybrid Romanian natural clinoptilolite. Zeolites, 3 (3): 259–260. membrane system – modeling study. Water Science [19] Verma, V.S.; Bora, B.J.; Sarkar, A.; Saha, U.K. (2014) and Technology, 62 (2): 326–332. Experimental investigation of a duel fuel diesel engine [5] Bansal, T.; Tripathi, N.; Chawla, G. (2013) Upgradation run on scrubbed biogas using the method of adsorp- of biogas using combined method of alkaline water tion. ASME ESDA Conference Proceedings, 2 (1): 1–9. scrubbing and adsoption through carbon molecular [20] Dussadee, N.; Reansuwan, K.; Ramaraj, R. (2014) sieve. International Journal of ChemTech Research,5 Potential development of compressed bio-methane (2): 886–890. gas production from pig farms and elephant grass [6] Monteleone, G.; De Francesco, M.; Galli, M.; silage for transportation in Thailand. Bioresource Naticchioni, V. (2011) Deep H2S removal from biogas Technology, 155 (1): 438–441. for molten carbonate fuel cell (MCFC) systems. [21] Rautenbach, R.; Welsch, K. (1994) Treatment of land- Chemical Engineering Journal, 173 (1): 407–414. fill gas by gas permeation – pilot plant results and [7] Pilarczyk, E.; Henning, K.D.; Knoblauch, K. (1987) comparison to alternatives. The Journal of Membrane Adsorptive purification of biogas by the PSA process. Science, 87 (1): 107–118. Energy from Biomass and Wastes, 10 (1): 1673–1674. [22] Castellani, B.; Rossi, F.; Filipponi, M.; Nicolini, A. [8] Ritter, J.A.; Yang, R.T. (1987) Equilibrium adsorption (2014) Hydrate-based removal of carbon dioxide and of multicomponent gas mixtures at elevated pressures. hydrogen sulphide from biogas mixtures: experimental Industrial and Engineering Chemistry Research, 26 (1): investigation and energy evaluations. Biomass and 1679–1686. Bioenergy, 70 (1): 330–338. [9] Belmabkhout, Y.; De Weireld, G.; Sayari, A. (2009) [23] Chen, X.Y.; Vinh-Thang, H.; Ramirez, A.A.; Rodrigue, Amine-bearing mesoporous silica for CO2 and H2S D.; Kaliaguine, S. (2015) Membrane gas separation removal from natural gas and biogas. Langmuir,25 technologies for biogas upgrading. RSC Advances,5 (23): 13275–13278. (1): 24399–24448. [10] Chen, J.; Li, W.; Li, X.; Yu, H. (2012) Improving biogas [24] Basu, S.; Khan, A.L.; Cano-Odena, A.; Liu, C.; Vankelecom, separation and methane storage with multilayer gra- I.F.J. (2009) Membrane-based technologies for biogas phene nanostructure via layer spacing optimization separations. Chemical Society Reviews, 39 (1): 750–768. and lithium doping: A molecular simulation investiga- [25] Abatzoglou, N.; Boivin, S. (2009) A review of biogas tion. Environmental Science and Technology, 46 (1): purification processes. Biofuels, Bioproducts and 10341–10348. Biorefining, 3 (1): 42–71. [11] Makareviciene, V.; Sendzikiene, E. (2015) [26] Gholampour, F.; Yeganegi, S. (2014) Molecular simula- Technological assumption for biogas purification. tion study on the adsorption and separation of acidic Environmental Technology, 36 (14): 1745–1750. gases in a model nanoporous carbon. Chemical [12] Peng, X.; Cao, D. (2013) Computational screening of Engineering Science, 117 (1): 426–435. porous carbons, zeolites, and metal organic frameworks [27] Tomadakis, M.M.; Heck, H.H.; Jubran, M.E.; Al-Harthi, for desulfurization and decarburization of biogas, natural K. (2011) Pressure-swing adsorption separation of H2S gas, and flue gas. AIChE Journal,59(8):2928–2942. from CO2 with molecular sieves 4A, 5A, and 13X. [13] Papurello, D.; Borchiellini, R.; Bareschino, P.; Chiodo, Separation Science and Technology, 26 (1): 428–433. V.; Freni, S.; Lanzini, A.; Pepe, F.; Ortigoza, G.A.; [28] Ruthven, R.M. (1984) Principles of Adsorption and Santarelli, M. (2014) Performance of a solid oxide Adsorption Processes; Wiley: New York, NY. fuel cell short-stack with biogas feeding. Applied [29] Mofarahi, M.; Gholipour, F. (2014) Gas adsorption Energy, 125 (1): 254–263. separation of CO2/CH4 system using zeolite 5A. [14] Cosoli, P.; Ferrone, M.; Pricl, S.; Fermeglia, M. (2008) Microporous and Mesoporous Materials, 200 (1): 1–10. Hydrogen sulphide removal from biogas by zeolite [30] McEwen, J.; Hayman, J.-D.; Ozgur Yazaydin, A. (2013) adsorption. Part I. GCMC molecular simulations. A comparative study of CO2,CH4, and N2 adsorption Chemical Engineering Journal, 145 (1): 86–92. in ZIF-8, Zeolite-13X and BPL activated carbon. [15] Cosoli, P.; Ferrone, M.; Pricl, S.; Fermeglia, M. (2008) Chemical Physics, 412 (1): 72–76. Hydrogen sulfide removal from biogas by zeolite [31] Ghoufi, A.; Gaberova, L.; Rouquerol, J.; Vincent, D.; adsorption. Part II. MD simulations. Chemical Llewellyn, P.L.; Maurin, G. (2009) Adsorption of CO2, Engineering Journal, 145 (1): 93–99. CH4 and their binary mixture in faujasite NaY: A combi- [16] Maghsoudi, H.; Soltanieh, M.; Bozorgzadeh, H.; nation of molecular simulations with gravimetry-mano- Mohamadalizadeh, A. (2013) Adsorption isotherms metry and microcalorimetry measurements. Microporous and ideal selectivities of hydrogen sulfide and carbon and Mesoporous Materials, 119 (1): 117–128. dioxide over methane for the Si-CHA zeolite: compar- [32] Yang, R.T. (1997) Gas Separation by Adsorption ison of carbon dioxide and methane adsorption with Processes; Imperial College Press: London, UK.