- Email: [email protected]
Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane
Accepted Manuscript Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane Yuting Zhang, Shengze Chen, Rui Shi, Peng Du, Xufeng Qiu, Xuehong Gu PII: DOI: Reference:
S1383-5866(18)30451-9 https://doi.org/10.1016/j.seppur.2018.04.066 SEPPUR 14562
To appear in:
Separation and Purification Technology
Received Date: Revised Date: Accepted Date:
5 February 2018 24 April 2018 24 April 2018
Please cite this article as: Y. Zhang, S. Chen, R. Shi, P. Du, X. Qiu, X. Gu, Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane, Separation and Purification Technology (2018), doi: https://doi.org/10.1016/j.seppur.2018.04.066
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane
Yuting Zhang, Shengze Chen, Rui Shi, Peng Du, Xufeng Qiu, Xuehong Gu*
State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, Jiangsu National Synergetic Innovation Center for Advanced Materials, Nanjing Tech University, 5 Xinmofan Road, Nanjing 210009, PR China
*Corresponding author: Xuehong Gu Tel.: (86)25-83172268 Fax: (86)25-83172268 E-mail: [email protected]
1
Abstract All-silica Decadodecasil 3R (DD3R) zeolite membranes were prepared on four-channel ceramic hollow fibers by seeding with ball-milled Sigma-1 zeolites and hydrothermal synthesis method. Effect of synthesis time on membrane structure and morphology was investigated and the membrane quality was evaluated by CO2/CH4 binary gas separation. The as-prepared membranes were further used in pervaporation dehydration of acetic acid (AcOH) for the first time. Although behaving strong hydrophobicity, the defect-free DD3R zeolite membrane exhibited a good water-selective permeability based on molecular-sieving mechanism. A water permeation flux of 0.58 kg m-2 h-1 as well as a separation factor of 800 was achieved in dehydration of 70 wt% water/AcOH mixture at 368 K. Both experimental and molecular simulation studies were explored to investigate the influence of pervaporation condition on membrane separation performance. The results showed that both high temperature and feed water content were beneficial for improving the water flux and separation factor. The long-term pervaporation test revealed that hydrophobic DD3R zeolite membrane behaved stable permeation flux and separation factor in dehydration of AcOH, even in the presence of inorganic acid, confirming the excellent acid-resistance of its all-silica framework structure.
Keywords: DD3R zeolite membrane; hollow fiber; pervaporation; acetic acid
2
1. Introduction Acetic acid (AcOH) is an important chemical intermediate in the synthesis of vinyl acetate, terephthalic acid, cellulose esters and esters. The separation of water from AcOH is a critical process in the industrial production of AcOH. Azeotropic distillation and extractive distillation have been developed for AcOH dehydration but they are energy-intensive due to close relative volatility of water and AcOH in the aqueous solution. [1, 2] Pervaporation (PV) is a promising membrane technique for liquid separation, especially for azeotropic and close-boiling mixtures. During the last decades, zeolite membranes have been widely accepted for pervaporation due to their high permeation flux, good thermal stability and superior swelling-resistance. As the only commercialized zeolite membrane, NaA zeolite membrane has a uniform pore size of 0.42 nm and strong hydrophilicity, which exhibits excellent separation performance in dehydration of organic solvents such as methanol, ethanol and iso-propanol [3-5]. Nowadays, more than 200 industrial dehydration plants based on NaA zeolite membranes have been set up around the world [6, 7]. However, NaA zeolite membrane suffers from low stability against acid due to its low Si/Al ratio (about 1) of the zeolite framework, which results in aluminum dissolution of the membrane in acidic solution [8, 9]. To improve the acid-resistance, many types of zeolite membranes with higher Si/Al ratios (e.g. CHA with >2, T with 3-4, ZSM-5 and MOR with >3) have been developed [10-15]. De la Iglesia et al. [10] carried out both NaA and MOR zeolite membranes into esterification reaction of acetic acid with ethanol. They found that MOR zeolite membrane showed a better 3
acidic resistance and maintained an improved conversion for 5 days while the NaA zeolite membrane was unstable under such acidic condition. Generally, with an increase in Si/Al ratio of zeolite framework, the hydrophilicity of zeolite membrane decreases whereas the acid-resistance simultaneously increases. Wang et al. [12] prepared high-flux T zeolite membranes on hollow fiber supports and used in PV dehydration of 98 wt% propanol/water mixture under weak acid (pH=4). The permeate water content maintained at about 98 wt% for about 200 h. Zhu et al. [13] prepared MOR zeolite membranes on alumina tubes by a rapid microwave-assisted synthesis method. The membrane synthesized for 3 h maintained stable flux after immersed in high AcOH content solution for 59 days. Yang et al. [14] prepared a ZSM-5 membrane for dehydration of AcOH by tuning aluminum spatial distribution in the membrane. The membrane behaved a stable water flux of 2.2 kg m-2 h-1 while separation selectivity was very high. Wang et al. [15] prepared acid-proof Ge-ZSM-5 zeolite membrane for dehydrating AcOH solution. The water flux was nearly constant (about 0.57 kg m-2 h-1) in the acetic acid concentration range of 30-98 wt%. Nevertheless, allowing for the complexity of the real acidic solution, not only the tolerance to weak organic acid but also to strong inorganic acid is worth to concern. Cui et al. [16] tested the stability of T zeolite powder immersed in 1.0 M HCl aqueous solution for 1 h. They found that the treated zeolite crystal decomposed a lot in the acid solution. Zhu et al. [17] prepared MOR zeolite membrane with higher Si/Al ratio of 5 and adopted it in esterification of acetic acid and alcohol with H2SO4 as the catalyst. They did not also find that the membrane had good stability in the presence 4
of H2SO4. It seems that such zeolite membranes containing aluminum can hardly withstand strong inorganic acids. DD3R zeolite membrane is a type of zeolite membrane with pore size of 0.36 nm×0.44 nm, which is promising to behave water-selective separation based on molecular-sieving mechanism. Moreover, the all-silica framework structure makes it possible to be stable against strong acids. Until now, there have been very limited studies on DD3R zeolite membrane and most of them have been focused on CO2 separation [18-22]. Kuhn et al. [23] used a tubular DD3R zeolite membrane for dehydrations of methanol and ethanol. The membrane exhibited a water flux of 2 kg m-2 h-1 and a separation factor of 1500 for 18 wt% water/ethanol mixture at 373 K, confirming that DD3R zeolite membrane is a high-potential candidate for solvent dehydration. Although NGK Insulators, Ltd., has published several literatures on DD3R zeolite membrane in the last decade [18, 19], it has been still great challenge to prepare high-quality DD3R zeolite membrane reproducibly because the competitive impurity crystals such as Dodecasil 1H (DOH) and Sigma-2 (SGT) (Fig.S1) which have different topological structures are very easily grown in the membrane and thermal-induced expansion of zeolite membrane during the high-temperature calcination often causes intercrystalline defects [22]. To address the issue, we developed a secondary-growth method using ball-milled Sigma-1 zeolite (a structural analogue of DD3R with aluminium content in the framework) seeds followed by low-temperature calcination in mixing ozone/oxygen atmosphere in our previous work [24]. The as-synthesized membranes achieved high CO2/CH4 separation 5
selectivity of over 300. Recently, hollow fiber supported zeolite membranes have been widely attractive due to their virtues of large packing density (as high as 1000 m2 m-3) and high permeation flux [25-28]. As compared to polymeric hollow fiber supports, ceramic hollow fiber supports are more thermally and chemically stable and therefore suitable for hydrothermal synthesis of zeolite membranes. However, as the conventional single-channel ceramic hollow fiber supports are very fragile, the industrialization of the supported zeolite membranes have been still limited. To overcome this issue, we proposed a novel four-channel ceramic hollow fiber support with larger outer diameter and cross-linking configuration [29, 30]. Such structure made the obtained supported zeolite membranes possess not only excellent separation performance but also robust mechanical strength in pervaporation and gas separation [31, 32]. Following up our previous work, we further prepared DD3R zeolite membranes on ceramic four-channel hollow fiber supports with different qualities. The DD3R zeolite membranes were firstly used for pervaporation dehydration of AcOH. Inorganic acids such as HCl and H2SO4 were also included in AcOH for stability evaluation. Influences of temperature and feed water content on PV performance were studied experimentally and theoretically. PV stability of DD3R zeolite membrane in acid environment was tested for a long operation time.
2. Experimental 2.1 Membrane preparation 6
DD3R zeolite membranes were prepared on four-channel ceramic hollow fiber supports by the secondary growth method. The supports were prepared by the combined phase-inversion and sintering method in our laboratory [29, 30]. The obtained supports, as shown in Fig. 1, had outer/inner diameter (O.D./I.D.) of about 3.6 mm/0.9 mm, average pore size of 0.5 μm and porosity of ~40%. The outer surface of support was seeded with ball-milled Sigma-1 zeolite particles (average particle size 0.5 μm) by dipping into an aqueous seed suspension (1 wt%) for 15 s. The detailed preparation process of ball-milled Sigma-1 zeolite seeds were described in our previous work [24]. The seeded supports were dried at 373 K overnight. The synthesis precursor was prepared by mixing 1-adamantanamine (ADA, Sigma-Aldrich), colloidal silica (Ludox SM-30, Sigma-Aldrich), ethylenediamine (EDA, Sigma-Aldrich) and deionized water with a molar composition of ADA: SiO 2: EDA: H2O=3:100:50:4000. After stirring for one hour, the synthesis mixture was loaded in a Teflon-lined autoclave and used for membrane synthesis by immersing the seeded support. The hydrothermal crystallization was carried out at 413 K for a certain time, and subsequently the as-synthesized membranes were washed and dried overnight. Detemplation of DD3R zeolite membrane was conducted in a stainless-steel module in mixing ozone/oxygen atmosphere at 473 K for 80 h. The heating and cooling rates were controlled at 1 K min-1. The detailed procedure was also described in our previous work [24]. The morphologies of DD3R zeolite membranes were characterized by Field Emission Scanning Electron Microscopy (FE-SEM, S-4800, Hitachi). The crystal 7
phases were determined by X-ray diffraction (XRD, MiniFlex 600, Rigaku) with Cu Kα radiation in the 2θ rang of 5-50°.
2.2 Gas separation Before PV separation, the qualities of DD3R zeolite membranes were evaluated by binary CO2/CH4 gas separation. The membrane was mounted in a stainless steel module, with each end sealed with a silicone O-ring. The feed pressure was fixed at 0.2 MPa (absolute pressure) by a back pressure regulator, and the permeate side was operated under atmospheric pressure. An equimolar CO2/CH4 flow was fed at a rate of 80 mL min-1 and a helium stream (50 mL min-1) was used to sweep the permeate side. A gas chromatography (GC, GC-2014, Shimadzu) equipped with thermal conductivity detector (TCD) and a packed column of HAYESEP-DB 200, was used to analyze gas compositions. The gas permeance of component i (Pm,i, mol m-2 s-1 Pa-1) is defined as Pm,i=Fi/∆p
(1)
where Fi is permeation flux through the membrane, mol m-2 s-1, ∆p is the pressure drop between both sides of the membrane, Pa. The CO2/CH4 separation selectivity (SCO2/CH4) is calculated from the gas permeances of CO2 and CH4 by SCO2/CH4= Pm,CO2/Pm,CH4
(2)
2.3 PV test The PV experimental apparatus for evaluating separation performance of DD3R 8
zeolite membranes is similar to that in our previous work [11, 12]. One end of the hollow fiber DD3R zeolite membrane was sealed with silicone and the other end was connected to a vacuum line. The effective membrane area was about 4 cm2. The membrane was immersed in the feed tank, which was filled with approximately 2 L of feed liquid mixture. The inside of the membrane was evacuated by a vacuum pump, and the pressure was maintained below 200 Pa. The permeated vapors were collected using cold traps cooled by liquid nitrogen. Both the feed and permeate were analyzed by a gas chromatograph (GC-2014, Shimadzu) equipped with a thermal conductivity detector and a capillary column of PEG-20M (Nanjing Tech). The PV performance of each membrane was determined by separation factor (α) and water permeation flux (J), which are respectively defined as follows: Ji=mi/(A·t)
w/ e
y w / xw y e / xe
(3) (4)
where mi is the water mass (kg) permeated over a time period of t (h); A is the effective membrane area (m2); yw and ye are the weight fractions of water and ethanol in the permeate, respectively; and xw and xe are the weight fractions of water and ethanol in the feed, respectively.
2.4 Molecular simulation For permeations of water/AcOH binary mixtures in DD3R zeolite layer, their adsorption behaviors in DD3R zeolite were studied by molecular simulation using the Accelrys Materials Studio software on a work station as described in our previous 9
work [28]. All the molecular simulations were conducted in 1×2×1 DD3R zeolite unit cells with periodic boundary condition, the formula of the supercell is Si480O960. As shown in Fig. 2, the DD3R zeolite topology structure consists of three types of cages: pentagonal dodecahedra (512 cage), decahedron (435661 cage) and 19-hedron (435126183 cage). A two-dimensional channel system is formed by connecting the 19-hedron cavities through a single 8-ring with an aperture of 0.36 nm×0.44 nm. The COMPASS (condensed-phase optimized molecular potentials for atomistic simulation studies) force field was adopted and the partial atomic charges associated with each atomic species were set as -0.445 for O, +0.89 for Si. The saturation fugacities of water and AcOH (fs,w and fs,A) were calculated by Aspen Plus software using Wilson equation. Their corresponding saturation adsorption loadings (As,w and As,A) were simulated by the GCMC (Grand canonical ensemble Monte Carlo) method using the Sorption module. The electrostatic potential energy and van der Waals potential energy were calculated by the Ewald summation and atom-based techniques, respectively. The cutoff distance for van der Waals potential energy calculation was taken as 12 Å, which was about half the length of the zeolite cell. The spline width and the buffer width were set at the default values, 1.0 Å and 0.5 Å. Here, the equilibration steps were set to 4,000,000 and the production steps were set to 4,000,000 Monte Carlo moves. The self-diffusion behaviors of water/AcOH mixtures were simulated by the EMD (equilibrium molecular dynamics) method. The EMD simulation was carried out by using the Discover module in cases of adsorption simulation results. The 10
adsorbent model and the calculation of non-bonded interaction were the same as those used in adsorption simulations. A smart minimize method with an ultrafine convergence level and equilibration run was carried out for optimization before running the production steps of the MD simulation. The step time was set to 1.0 fs. The total simulation time was about 1600 ps. The initial time of 400 ps was used to perform the MD equilibration for balance, and the rest simulation time was special for the MD production and the trajectories were saved every 1000 steps in the productions simulations for the purpose of analysis. The temperature was controlled using a NVT ensemble and Nose thermostat. The self-diffusion coefficient of component can be calculated by EMD simulation based on the Einstein equation [28]:
Ds,i
2 1 d Na lim ri t ri 0 6 N a t dt i 1
where Na is the number of diffusing molecules, and
ri t ri 0
(5) 2
is the average
of the mean squared displacement (MSD) of the diffusing molecules.
3. Results and discussion 3.1 Membrane preparation DD3R zeolite membranes with different qualities were hydrothermally synthesized at the same synthesis temperature (413 K) for different periods of 40 h, 44 h and 48 h. Fig. 3 shows XRD patterns of the membranes synthesized for different time. For the membrane synthesized for 40 h, the characteristic peaks of DD3R zeolite at 7.5°, 15.5° and 17° are observable but their intensities are very weak, indicating the immature growth of DD3R zeolite membrane. When the synthesis time 11
was 44 h, the intensities of the characteristic peaks become obvious, implying that the membrane had good crystallization quality. However, as the synthesis time was further prolonged, the characteristic peaks of DD3R zeolite nearly disappear. Instead, three characteristic peaks of SGT zeolite at 9.01°, 19.56° and 19.76° occurred. Fig. 4 presents surface and cross-section SEM images of the DD3R zeolite membranes synthesized for different synthesis time. As being consistent with the observation on XRD results, a loose-packing zeolite layer with large voids and defects occur on the surface of the membrane synthesized for 40 h, the thickness of zeolite layer is only about 1 μm. A well-intergrown DD3R zeolite membrane with uniform thickness (about 5 μm) is observed after synthesized for 44 h, indicating good quality of the membrane. As the synthesis time was 48 h, ball-shaped SGT zeolite crystals with large crystal size covered on membrane surface. Table 1 lists CO2/CH4 gas separation results of different DD3R zeolite membranes in this work and literatures. It can be seen that CO2 permeance of the membranes in this work decreased as the synthesis time increased from 40 to 48 h due to the increased membrane thickness. Particularly, the membrane M2 synthesized for 44 h showed the highest CO2/CH4 separation selectivity of 679 while its CO2 permeance is 1.0×10-7 mol m-2 s-1 Pa-1. As compared to other membranes in literatures [18, 19, 22], M2 also exhibited an extremely high separation selectivity as well as a good CO2 permeance. The membranes were further adopted for PV dehydrations of 10 wt% water/AcOH mixture at 348 K. Since the pore size of DD3R zeolite membrane (0.36 nm×0.44 nm) is between the kinetic diameters of water (0.29 nm) 12
and AcOH (0.436 nm), the membranes shown in Table 2 behaved water-selective for separation of H2O/AcOH. M2 exhibited a total flux of 0.17 kg m-2 h-1 as well as a separation factor of 213 in PV dehydration of AcOH/water mixture at 348 K. Table 2 also lists PV results of two zeolite membranes with high Si/Al ratios (Ge-ZSM-5 and hydrophilic silicalite-1) for comparison [15, 33]. As compared, DD3R zeolite membrane showed moderate separation factor due to its strong hydrophobicity. However, its separation factor would be improved by more careful manipulation in membrane synthesis or posttreatment on membrane surface e.g. hydrophilic modification in further work.
3.2 Pure water and AcOH permeation fluxes M2 membrane was further used for PV permeation of pure water and AcOH permeation. During the intervals of PV operation, the membrane was also immersed in water or AcOH. As shown in Fig. S2, the pure water and AcOH fluxes of the membrane at 368 K were about 0.91 and 0.1 kg m-2 h-1 respectively, indicating the possibility of the membrane for separation between water and AcOH. It is noted that although the membrane was immersed in pure water and AcOH for a long time, both water and AcOH fluxes varied by less than 5% at the same temperature. Such highly stable performance was mainly owed to the all-silica structure of DD3R zeolite membrane. Fig. 6 shows the simulated adsorption isotherms and diffusion curves of pure water and AcOH in DD3R zeolite at 368 K. As shown in Fig. 6a, DD3R zeolite 13
preferred to adsorb AcOH than water due to its strong hydrophobicity. The saturation adsorption loadings of water and AcOH were 0.44 and 1.84 mol kg-1 respectively. It can be seen from Fig. 6b that the water molecules performed much larger MSD than that of AcOH. The self-diffusion coefficients of water and AcOH (Dw and DA) were 5.69×10-10 and 4.93×10-11 m2 s-1 respectively. The self-diffusion coefficient of AcOH was very low for its large molecule size in confined zeolitic pores. However, the small molecular diameter of water made it move more freely in the zeolite. In addition, there was little interaction between the water molecules and the hydrophobic zeolite channels.
3.3 PV dehydration of water/AcOH mixture 3.3.1 Effect of temperature Temperature is a significant factor to determine adsorption loading and diffusivity of component in zeolite membrane, which affects the membrane separation performance. Fig. 6a shows PV results of M2 as a function of temperature in dehydration of 10 wt% water/ethanol mixture. As expected, high temperature was beneficial for improving the permeation flux. It can be explained from Table 3 that on one side high temperature increased the saturation fugacity of water in the feed side, which enhanced its driving force for permeation through the membrane; on the other side, the self-diffusion coefficient of water was accelerated at high temperature. As shown in Fig. 6b, both water and AcOH fluxes followed the Arrhenius relationship with temperature. The activation energies of water and AcOH are 23.56 kJ mol-1 and 14
0.57 kJ mol-1 respectively. Generally, in the adsorption–diffusion model of molecular transport through a zeolite membrane, the apparent activation energy (EA) comprises a contribution of both the activation energy for diffusion (ED) as well as the heat of adsorption (ΔHS), EA=ED−ΔHS. When EA>0, it indicates that the activation energy of diffusion dominates over the heat of adsorption. It can be also seen from the table that the saturation adsorption loading of water decreased gradually with the temperature whereas that of AcOH was nearly insensitive to the temperature. Due to the simultaneous water adsorption, the saturation adsorption loading of AcOH in the mixture was lowered to 1.67 mol kg-1 at 368 K as compared with its pure component (1.84 mol kg-1). But, it is interesting to find that the saturation adsorption loading of water in the mixture (0.49 mol kg-1) was higher than that of its pure component (0.44 mol kg-1). Fig. 8 shows the radial distribution function (RDF) of hydrogen atoms in water molecules and oxygen atoms in AcOH molecules in DD3R zeolite at 368 K. Two peaks at about 1.9 Å and 3.3 Å represent the hydrogen bonds between hydrogen atoms of water and carbonyl and hydroxyl oxygen atoms of AcOH, respectively. We considered that such interaction between water and AcOH molecules in the mixture could be the main reason for improving adsorption of water but weakening that of AcOH in the zeolite. It can be also seen from Fig. 7 that high PV temperature was beneficial not only for permeation flux but also for separation factor. At 368 K, the permeation flux was 0.26 kg m-2 h-1 while the separation factor was 345. Such temperature dependency of separation factor for DD3R zeolite membrane was different from those occurred on 15
some hydrophilic zeolite membranes, e.g. NaA zeolite membrane [31], T zeolite membrane [16] and MOR zeolite membrane [34]. It is well known that pervaporation through zeolite membrane can be explained by the adsorption-diffusion mechanism. Water molecules are preferred to adsorb in hydrophilic zeolite membrane pores and sometimes condensate on the surface of small intercrystalline defects. The elevated temperature significantly reduced the adsorption coverage of water on membrane surface and in zeolite cages, which would open pores for diffusion of large organic molecules and thus decrease the separation selectivity. But, towards hydrophobic DD3R zeolite membrane, although the saturation adsorption selectivity of water over AcOH decreased from 0.41 to 0.29 as the temperature climbed up from 348 to 368 K, the corresponding selectivity in self-diffusion coefficient increased more significantly from 17.8 to 28.6. Therefore, the approximate water/AcOH separation selectivity calculated based on the adsorption-diffusion model was improved at higher temperature. It is noted that although the calculated self-diffusion coefficient of water in DD3R zeolite was close to that in NaA zeolite [28], the actual permeation flux of DD3R zeolite membrane was much lower than the expected result. Kuhn et al. [35] studied diffusion behaviors of water/methanol mixture in DD3R zeolite via MD simulation. They also found that the experimental flux was one order of magnitude lower than the predicted flux based on the simulated diffusivities. They attributed the deviation between the simulated and experimental results to many factors, e.g. the assumption of thermodynamic correction factor, the perfect crystalline system, the 16
silanol group present in the actual zeolite and so on. Not only these, as shown in Fig. 9, the diffusion tensors in the directions x, y and z of water were different due to the different channel types of DD3R zeolite. The anisotropic self-diffusion coefficients followed an order of Dy>Dx>Dz, especially Dz was nearly close to zero. This agreed with the topology structure of DD3R zeolite which has a two-dimensional channel system in the xy-plane. The diffusion of molecules along the direction z were hindered by the pentagonal and decahedron cages with small 4-ring and 5-ring windows which are inaccessible for water and AcOH molecules. We suggested that high permeation flux would be achieved on the oriented DD3R zeolite membrane that the 8-ring pore channels are aligned perpendicularly to the support surface since the 2D channel is the fastest diffusion pathway in DD3R crystal. Similar claim was also proposed by Choi et al. on DD3R zeolite membrane for CO2 separation [36]. However, according to the XRD and SEM results, the used DD3R zeolite membrane in this study was randomly oriented, which could be one reason for the large derivations between the experimental and simulated diffusivities and activated energies.
3.3.2 Effect of feed water content Fig. 10 shows PV results of M2 in dehydrations of water/AcOH mixtures with different feed water contents at 368 K. As the feed water content climbed up from 10 to 70 wt%, the permeation flux of the membrane increased gradually from 0.26 to 0.58 kg m-2 h-1. It can be seen from Table 3 that high feed water content led to high saturation fugacity of water, which helped to improve its driving force for permeation 17
through the membrane. When the feed water content varied from 10 to 70 wt%, the saturation adsorption loading of water increased largely from 0.39 to 1.07 mol kg-1 whereas that of AcOH declined a little from 1.67 to 1.57 mol kg-1. It is interesting to find that the self-diffusion coefficient of water decreased gradually under the feed water content from 10 to 50 wt%. Such phenomenon could be owed to the obviously increased adsorption loading of water which reduced vacant coverage in DD3R zeolite. The further increase in the self-diffusion coefficient of water at the feed water content of 70 wt% could be attributed to the large reduction in the adsorption loading of AcOH, which released more vacant sites for water diffusion. On the contrary, the self-diffusion coefficient of AcOH decreased continuously with the feed water content due to increasing adsorption loading of water molecules in DD3R zeolite, which reduced vacant coverage for AcOH diffusion. Not only permeation flux, high separation factor was also achieved at high feed water content. When the feed water content was 70 wt%, the separation factor was as high as 800 while the permeate water content was 99.9 wt%. The improvement in separation factor at high feed water content was mainly caused by the simultaneous increases in adsorption and diffusion selectivities of water over AcOH. As the feed water content raised from 10 to 70 wt%, the adsorption selectivity increased from 0.23 to 0.68 while the selectivity in self-diffusion coefficient increased from 28.6 to 98.9.
3.3.3 Long-term PV stability As mentioned above, the stability of zeolite membrane in acidic solution is a 18
significant concern for practical application. The hollow fiber supported DD3R zeolite membrane (M2) was further used in dehydration of 10 wt% water/AcOH mixture at 348 K to evaluate its acid stability. Two kinds of typical strong inorganic acids (0.1 M hydrochloric acid and 0.05 M sulfuric acid) were added into the feed mixture sequentially. During the PV operation intervals, the used membrane was dried and then soaked in the 10 wt% water/AcOH solution at room temperature. As shown in Fig. 11, the membrane exhibited stable permeation flux of about 0.185 kg m-2 h-1 and separation factor of around 250 during the whole PV test despite high-content AcOH in the feed mixture. It is also interesting to observe that the membrane still maintained good and stable separation performance even in the presences of HCl and H2SO4, indicating its excellent durability to withstand inorganic acid. Table 4 lists PV results of the membrane in dehydration of 10 wt% water/AcOH mixtures with different HCl contents. It seems that the addition of hydrochloric acid in the mixture did not affect the separation performance obviously. It should be noted that there was no report on successful pervaporation dehydration of solvents including strong acids using other types of zeolite membranes so far. van Veen et al. [37] tested the stability of HybSi® pervaporation membrane in the presence of nitric acid with different contents. They found that the maximum concentration of nitric acid to be withstood for the membrane was below 0.08 mol L-1. Comparatively, DD3R zeolite membrane exhibited superior acid-resistance. Fig. 12 shows SEM images of the used membrane after long-term PV test in the presences of hydrochloric acid and sulfuric acid. The surface morphology of the used membrane was nearly the same with the fresh one. 19
No obvious structural change occurred on the membrane surface, which should be attributed to the pure silica crystalline structure.
4. Conclusions High-quality hollow fiber supported DD3R zeolite membranes were prepared and successfully used in pervaporation dehydration of acetic acid. Both high temperature and feed water content were beneficial for improving permeation flux and separation factor of the membrane. Due to the strong hydrophobicity, the membranes behaved morderate permeation flux and separation factor. However, the pure-silica zeolite membranes exhibited a very stable separation performance during the pervaporation in acidic solution, even in the presence of strong acids. It could be helpful to increase the permeation flux largely by grafting hydrophilic groups on the membrane surface. On the other hand, it is highly encouraged to fabricate oriented DD3R zeolite membranes with the 8-ring pore channels aligned perpendicularly on the support surface. We suggest that such highly-stable separation performance makes DD3R zeolite membrane have great potential in application of dehydration of acidic solutions.
Acknowledgement This work is sponsored by the National Natural Science Foundation of China (21490585 and 21776128), the National High-tech R&D Program of China (2015AA03A602), the “Six Top Talents” and “333 Talent Project” of Jiangsu 20
Province, State Key Laboratory of Materials-Oriented Chemical Engineering (ZK201602), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).
21
References [1] K. Makita, Y. Hirota, Y. Egashira, K. Yoshida, Y. Sasakib, N. Nishiyama, Synthesis of MCM-22 zeolite membranes and vapor permeation of water/acetic acid mixtures, J. Membr. Sci. 372 (2011) 269-276. [2] U. S. Toti, M. Y. Kariduraganavar, K. S. Soppimath, T. M. Aminabhavi, Sorption, diffusion, and pervaporation separation of water–acetic acid mixtures through the blend membranes of sodium alginate and guar gum-grafted-polyacrylamide, J. Appl. Polym. Sci. 83 (2002) 259–272. [3] J. Gascon, F. Kapteijn, B. Zornoza, V. Sebastián, C. Casado, J. Coronas, Practical approach to zeolitic membranes and coatings: State of the art, opportunities, barriers, and future perspectives, Chem. Mater. 24 (2012) 2829-2844. [4] Z. Yang, Y. Liu, C. Yu, X. Gu, N. Xu, Ball-milled NaA zeolite seeds with submicron size for growth of NaA zeolite membranes, J. Membr. Sci. 392-393 (2012) 18-28. [5] C. Yu, Y. Liu, G. Chen, X. Gu, W. Xing, Pretreatment of isopropanol solution from pharmaceutical industry and pervaporation dehydration by NaA zeolite membranes, Chin. J. Chem. Eng. 19(2011) 904-910. [6] Y. Morigami, M. Kondo, J. Abe, H. Kita, K. Okamoto, The first large scale pervaporation plant using tubular-type module with zeolite NaA membrane, Sep. Purif. Technol. 25 (2001) 251-260. [7] N. Rangnekar, N. Mittal, B. Elyassi, J. Caro, M. Tsapatsis, Zeolite membranes-a review and comparison with MOFs, Chem. Soc. Rev. 44 (2015) 7128-7154. 22
[8] Y. Hasegawa, T. Nagase, Y. Kiyozumi, T. Hanaoka, F. Mizukami, Influence of acid on the permeation properties of NaA-type zeolite membranes, J. Membr. Sci. 349 (2010) 189-194. [9] X. Cai, Y. Zhang, L. Yin, D. Ding, W. Jing, X. Gu, Electrochemical impedance spectroscopy for analyzing microstructure evolution of NaA zeolite membrane in acid water/ethanol solution, Chem. Eng. Sci. 153 (2016) 1-9. [10] O. Iglesia, R. Mallada, M. Menendez, J. Coronas, Continuous zeolite membrane reactor for esterification of ethanol and acetic acid, Chem. Eng. J. 131 (2007) 35–39. [11] J. Jiang, X. Wang, Y. Zhang, D. Liu, X. Gu, Fabrication of pure-phase CHA zeolite membranes with ball-milled seeds at low K+ concentration, Micropor. Mesopor. Mater. 215 (2015) 98-108. [12] X. Wang, Z. Yang, C. Yu, L. Yin, C. Zhang, X. Gu, Preparation of T-type zeolite membranes using a dip-coating seeding suspension containing colloidal SiO2, Micropor. Mesopor. Mater. 197 (2014) 17-25. [13] M.-H. Zhu, S.-L. Xia, X.-M. Hua, Z.-J. Feng, N. Hu, F. Zhang, I. Kumakiri, Z.-H. Lu, X.-S. Chen, H. Kita, Rapid preparation of acid-stable and high dehydration performance mordenite membranes, Ind. Eng. Chem. Res. 53 (2014) 19168-19174. [14] J. Yang, L. Li, W. Li, J. Wang, Z. Chen, D. Yin, J. Lu, Y. Zhang, H. Guo, Tuning aluminum spatial distribution in ZSM-5 membranes: a new strategy to fabricate high performance and stable zeolite membranes for dehydration of acetic acid, 23
Chem. Commun. 50 (2014) 14654-14657. [15] X. Wang, X. Deng, Z. Bai, X. Zhang, X. Feng, W. Huang, The synthesis of super-hydrophilic and acid-proof Ge–ZSM-5 membranes by simultaneous incorporation of Ge and Al into a Silicalite-1 framework, J. Membr. Sci. 468 (2014) 202–208. [16] Y. Cui, H. Kita, K.-I. Okamoto, Zeolite T membrane: preparation, characterization, pervaporation of water/organic liquid mixtures and acid stability, J. Membr. Sci. 236 (2004) 17-27. [17] M.-H. Zhu, I. Kumakiri, K. Tanaka, H. Kita, Dehydration of acetic acid and esterification product by acid-stable ZSM-5 membrane, Micropor. Mesopor. Mater. 181 (2013) 47-53. [18] T. Tomita, K. Nakayama, H. Sakai, Gas separation characteristics of DDR type zeolite membrane, Micropor. Mesopor. Mater. 68 (2004) 71-75. [19] S. Himeno, T. Tomita, K. Suzuki, K. Nakayama, K. Yajima, S. Yoshida, Synthesis and permeation properties of a DDR-type zeolite membrane for separation of CO2/CH4 gaseous mixtures, Ind. Eng. Chem. Res. 46 (2007) 6989-6997. [20] J. van den Bergh, W. Zhu, J. Gascon, J.A. Moulijn, F. Kapteijn, Separation and permeation characteristics of a DD3R zeolite membrane, J. Membr. Sci. 316 (2008) 35-45. [21] J. van den Bergh, A. Tihaya, F. Kapteijn, High temperature permeation and separation characteristics of an all-silica DDR zeolite membrane, Micropor. Mesopor. Mater. 132 (2010) 137-147. 24
[22] S. Yang, Z. Cao, A. Arvanitis, X. Sun, Z. Xu, J. Dong, DDR-type zeolite membrane synthesis, modification and gas permeation studies, J. Membr. Sci. 505 (2016) 194-204. [23] J. Kuhn, K. Yajima, T. Tomita, J. Gross, F. Kapteijn, Dehydration performance of a hydrophobic DD3R zeolite membrane, J. Membr. Sci. 321 (2008) 344-349. [24] L. Wang, C. Zhang, X. Gao, L. Peng, J. Jiang, X. Gu, Preparation of defect-free DDR zeolite membranes by eliminating template with ozone at low temperature, J. Membr. Sci. 539 (2017) 152–160. [25] P. S. Lee, M. S. Lim, A. Park, H. Park, S. -E. Nam, Y. Park, A zeolite membrane module composed of SAPO-34 hollow fibers for use in fluorinated gas enrichment, J. Membr. Sci. 542 (2017) 123–132. [26] H. Yan, N. Ma, Z. Zhan, Z. Wang, Fabrication of zeolite NaA membranes on hollow fibers using nano-sized seeds exfoliated from mesoporous zeolite crystals, Micropor. Mesopor. Mater. 215 (2010) 244-248. [27] V. Sebastian, R. Mallada, J. Coronas, A. Julbe, R. A. Terpstra, R. W. J. Dirrix, Fabrication of zeolite NaA membranes on hollow fibers using nano-sized seeds exfoliated from mesoporous zeolite crystals, J. Membr. Sci. 355 (2010) 28–35. [28] P. Ye, Y. Zhang, H. Wu, X. Gu, Mass transfer simulation on pervaporation dehydration of ethanol through hollow fiber NaA zeolite membranes, AIChE J. 62 (2016) 2468-2478. [29] Z. Shi, Y. Zhang, C. Cai, C. Zhang, X. Gu, Preparation and characterization of α-Al2 O3 hollow fiber membranes with four-channel configuration, Ceram. Int. 41 25
(2015) 1333-1339. [30] C. Cai, Y. Zhang, C. Zhang, X. Gu, Microstructure modulation of α-Al2O3 hollow fiber membranes with four-channel geometric configuration, Asia-Pac. J. Chem. Eng. 11 (2016) 949–957. [31] D. Liu, Y. Zhang, J. Jiang, X. Wang, C. Zhang, X. Gu, High-performance NaA zeolite membranes supported on four-channel ceramic hollow fibers for ethanol dehydration, RSC Adv. 5 (2015) 95866–95871. [32] Y. Chen, Y. Zhang, C. Zhang, J. Jiang, X. Gu, Fabrication of high-flux SAPO-34 membrane on α-Al2O3 four-channel hollow fibers for CO2 capture from CH4, J. CO2 Util. 18 (2017) 30–40. [33] T. Masuda, S. Otani, T. Tsuji, M. Kitamura, S. Mukai, Preparation of hydrophilic and acid-proof silicalite-1 zeolite membrane and its application to selective separation of water from water solutions of concentrated acetic acid by pervaporation, Sep. Purif. Technol. 32 (2003) 181-189. [34] G. Li, E. Kikuchi, M. Matsukata, Separation of water-acetic acid mixtures by pervaporation using a thin mordenite membrane, Sep. Purif. Technol. 32 (2003) 199-206. [35] J. Kuhn, J. M. Castillo-Sanchez, J. Gascon, S. Calero, D. Dubbeldam, T. J. H. Vlugt, F. Kapteijn, J. Gross, Adsorption and diffusion of water, methanol, and ethanol in all-silica DD3R: Experiments and simulation, J. Phys. Chem. C 113 (2009) 14290-14301. [36] E. Kim, W. Cai, H. Baik, J. Nam, J. Choi, Synthesis and sonication-induced 26
assembly of Si-DDR particles for close-packed oriented layers, Chem. Commun. 49 (2013) 7418-7420. [37] H. M. van Veen, M. D. A. Rietkerk, D. P. Shanahan, M. M. A. van Tuel, R. Kreiter, H. L. Castricum, J. E. ten Elshof, J. F. Vente, Pushing membrane stability boundaries with HybSi® pervaporation membranes, J. Membr. Sci. 380 (2011) 124-131.
27
Table 1 Comparison in CO2/CH4 gas separation results between different DD3R zeolite membranes. Membrane/Support Synthesis temperature
Membrane
CO2/CH4 separation
Reference
(K)/Time (h)
thickness (μm)
CO2 permeance (10-7 mol m-2 s-1 Pa-1)
CO2/CH4 selectivity
DD3R/Tube
423/48
5-10
0.7
280
[18]
DD3R/Tube
423/48
2-3
3.0
200
[19]
DD3R/Disc
433/24
10
~1.8
92
[22]
M1/Hollow fiber
413/40
~1
2.9
1.2
This work
M2/Hollow fiber
413/44
5
1.0
679
This work
M3/Hollow fiber
413/48
7
0.17
34
This work
28
Table 2 Comparison in PV results of acetic acid dehydration between different zeolite membranes.
a
Membrane
Feed mixture
PV temperature (K)
Total flux (kg m-2 h-1)
αwater/AcOH
Reference
Ge-ZSM-5
2 wt% water/AcOH
353
0.57
-a
15
Hydrophilic silicalite-1
2 wt% water/AcOH
353
~2×10-4
-a
33
M1
10 wt% water/AcOH
348
0.57
3.6
This work
M2
10 wt% water/AcOH
348
0.17
213
This work
M3
10 wt% water/AcOH
348
0.08
55
This work
The permeate acetic acid concentration was too low to be detected.
Table 3 Calculated saturation fugacity, adsorption loading and self-diffusion coefficients of 10 wt% water/AcOH mixture under different temperatures. Temperature
Water
AcOH
(K)
fs,w (kPa)
As,w (mol kg-1)
Ds,w (m2 s-1)
fs,w (kPa)
As,A (mol kg-1)
Ds,A (m2 s-1)
348
12.21
0.70
1.76×10-10
17.67
1.68
9.83×10-12
29
358
18.33
0.59
2.60×10-10
25.95
1.68
1.10×10-11
363
22.25
0.47
3.00×10-10
31.17
1.67
1.15×10-11
368
26.83
0.49
3.86×10-10
37.21
1.67
1.35×10-11
30
Table 4 Calculated saturation fugacity, adsorption loading and self-diffusion coefficients of water/AcOH mixtures under different feed water contents at 368 K. Feed water
Water
AcOH
content (wt%)
fs,w (kPa)
As,w (mol kg-1)
Ds,w (m2 s-1)
fs,w (kPa)
As,A (mol kg-1)
Ds,A (m2 s-1)
10
26.83
0.49
3.86×10-10
37.21
1.67
1.35×10-11
30
52.50
0.76
2.45×10-10
23.17
1.67
3.83×10-12
50
65.31
0.92
2.00×10-10
14.10
1.60
3.00×10-12
70
74.34
1.07
2.64×10-10
7.03
1.57
2.67×10-12
31
Table 5 PV results of M2 in dehydration of 10 wt% water/AcOH mixtures with different HCl contents at 348 K. HCl content in mixture (mol L-1)
Total flux (kg m-2 h-1)
αwater/AcOH
0
0.19±0.02
250±50
0.01
0.19±0.02
312±30
0.05
0.19±0.01
266±21
0.10
0.21±0.01
249±36
32
Figure captions Fig. 1
Cross-section and outer surface images of four-channel ceramic hollow fiber support.
Fig. 2
Schematic diagram of 1×2×1 unit cells of DD3R zeolite: left, (100) view and right, (001) view.
Fig. 3
XRD patterns of DD3R zeolite membranes synthesized for different time.
Fig. 4
Surface (a, c, e) and cross-section (b, d, f) SEM images of DD3R zeolite membrane synthesized for different time: 40 h (a, b); 44 h (c, d); 48 h (e, f).
Fig. 5
Simulated adsorption isotherms (a) and MSD vs. t curves (b) of pure water and AcOH in DD3R zeolite at 368 K.
Fig. 6
PV results of M2 for dehydration of 10 wt% water/AcOH mixture under different temperatures (a) and Arrhenius plot of water and AcOH fluxes (b).
Fig. 7
RDF of hydrogen atoms (water) and oxygen atoms (AcOH) in DD3R zeolite at 368 K..
Fig. 8
MSD of isotropic and anisotropic self-diffusion coefficient of water in DD3R zeolite for 10 wt% water/AcOH mixture at 368 K.
Fig. 9
PV results of M2 for dehydration of water/AcOH mixtures at 368 K under different feed water contents.
Fig. 10
Long-term PV results of M2 for dehydration of 10 wt% water/AcOH 33
mixture at 348 K. Fig. 11
Surface (a) and cross-section (b) SEM images of the used DD3R zeolite membrane after long-term PV test.
34
Fig. 1 Cross-section and outer surface images of four-channel ceramic hollow fiber support.
35
Fig. 2 Schematic diagram of 1×2×1 unit cells of DD3R zeolite: left, (100) view and right, (001) view.
36
Fig. 3 XRD patterns of DD3R zeolite membranes synthesized for different time.
37
Fig. 4 Surface (a, c, e) and cross-section (b, d, f) SEM images of DD3R zeolite membrane synthesized for different time: 40 h (a, b); 44 h (c, d); 48 h (e, f).
38
Fig. 5 Simulated adsorption isotherms (a) and MSD vs. t curves (b) of pure water and AcOH in DD3R zeolite at 368 K.
39
Fig. 6 PV results of M2 for dehydration of 10 wt% water/AcOH mixture under different temperatures (a) and Arrhenius plot of water and AcOH fluxes (b).
40
Fig. 7 RDF of hydrogen atoms (water) and oxygen atoms (AcOH) in DD3R zeolite at 368 K.
41
Fig. 8 MSD of isotropic and anisotropic self-diffusion coefficient of water in DD3R zeolite for 10 wt% water/AcOH mixture at 368 K.
42
Fig. 9 PV results of M2 for dehydration of water/AcOH mixtures at 368 K under different feed water contents.
43
Fig. 10 Long-term PV results of M2 for dehydration of 10 wt% water/AcOH mixture at 348 K.
44
Fig. 11 Surface (a) and cross-section (b) SEM images of the used DD3R zeolite membrane after long-term PV test.
45
Graphic abstract
46
Supporting information
Fig. S1 Framework structures of (a) DD3R, (b) DOH and (c) SGT zeolites.
Fig. S2 Pure water and AcOH fluxes through DD3R zeolite membrane (M2) at 368 K.
47
Graphical abstract
48
Highlights 1. High-quality DD3R zeolite membrane was fabricated on hollow fiber support. 2. DD3R zeolite membrane was firstly used in pervaporation dehydration of acetic acid. 3. The membrane showed excellent acid-resistance in the pervaporation separation.
49
S1383-5866(18)30451-9 https://doi.org/10.1016/j.seppur.2018.04.066 SEPPUR 14562
To appear in:
Separation and Purification Technology
Received Date: Revised Date: Accepted Date:
5 February 2018 24 April 2018 24 April 2018
Please cite this article as: Y. Zhang, S. Chen, R. Shi, P. Du, X. Qiu, X. Gu, Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane, Separation and Purification Technology (2018), doi: https://doi.org/10.1016/j.seppur.2018.04.066
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Pervaporation dehydration of acetic acid through hollow fiber supported DD3R zeolite membrane
Yuting Zhang, Shengze Chen, Rui Shi, Peng Du, Xufeng Qiu, Xuehong Gu*
State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, Jiangsu National Synergetic Innovation Center for Advanced Materials, Nanjing Tech University, 5 Xinmofan Road, Nanjing 210009, PR China
*Corresponding author: Xuehong Gu Tel.: (86)25-83172268 Fax: (86)25-83172268 E-mail: [email protected]
1
Abstract All-silica Decadodecasil 3R (DD3R) zeolite membranes were prepared on four-channel ceramic hollow fibers by seeding with ball-milled Sigma-1 zeolites and hydrothermal synthesis method. Effect of synthesis time on membrane structure and morphology was investigated and the membrane quality was evaluated by CO2/CH4 binary gas separation. The as-prepared membranes were further used in pervaporation dehydration of acetic acid (AcOH) for the first time. Although behaving strong hydrophobicity, the defect-free DD3R zeolite membrane exhibited a good water-selective permeability based on molecular-sieving mechanism. A water permeation flux of 0.58 kg m-2 h-1 as well as a separation factor of 800 was achieved in dehydration of 70 wt% water/AcOH mixture at 368 K. Both experimental and molecular simulation studies were explored to investigate the influence of pervaporation condition on membrane separation performance. The results showed that both high temperature and feed water content were beneficial for improving the water flux and separation factor. The long-term pervaporation test revealed that hydrophobic DD3R zeolite membrane behaved stable permeation flux and separation factor in dehydration of AcOH, even in the presence of inorganic acid, confirming the excellent acid-resistance of its all-silica framework structure.
Keywords: DD3R zeolite membrane; hollow fiber; pervaporation; acetic acid
2
1. Introduction Acetic acid (AcOH) is an important chemical intermediate in the synthesis of vinyl acetate, terephthalic acid, cellulose esters and esters. The separation of water from AcOH is a critical process in the industrial production of AcOH. Azeotropic distillation and extractive distillation have been developed for AcOH dehydration but they are energy-intensive due to close relative volatility of water and AcOH in the aqueous solution. [1, 2] Pervaporation (PV) is a promising membrane technique for liquid separation, especially for azeotropic and close-boiling mixtures. During the last decades, zeolite membranes have been widely accepted for pervaporation due to their high permeation flux, good thermal stability and superior swelling-resistance. As the only commercialized zeolite membrane, NaA zeolite membrane has a uniform pore size of 0.42 nm and strong hydrophilicity, which exhibits excellent separation performance in dehydration of organic solvents such as methanol, ethanol and iso-propanol [3-5]. Nowadays, more than 200 industrial dehydration plants based on NaA zeolite membranes have been set up around the world [6, 7]. However, NaA zeolite membrane suffers from low stability against acid due to its low Si/Al ratio (about 1) of the zeolite framework, which results in aluminum dissolution of the membrane in acidic solution [8, 9]. To improve the acid-resistance, many types of zeolite membranes with higher Si/Al ratios (e.g. CHA with >2, T with 3-4, ZSM-5 and MOR with >3) have been developed [10-15]. De la Iglesia et al. [10] carried out both NaA and MOR zeolite membranes into esterification reaction of acetic acid with ethanol. They found that MOR zeolite membrane showed a better 3
acidic resistance and maintained an improved conversion for 5 days while the NaA zeolite membrane was unstable under such acidic condition. Generally, with an increase in Si/Al ratio of zeolite framework, the hydrophilicity of zeolite membrane decreases whereas the acid-resistance simultaneously increases. Wang et al. [12] prepared high-flux T zeolite membranes on hollow fiber supports and used in PV dehydration of 98 wt% propanol/water mixture under weak acid (pH=4). The permeate water content maintained at about 98 wt% for about 200 h. Zhu et al. [13] prepared MOR zeolite membranes on alumina tubes by a rapid microwave-assisted synthesis method. The membrane synthesized for 3 h maintained stable flux after immersed in high AcOH content solution for 59 days. Yang et al. [14] prepared a ZSM-5 membrane for dehydration of AcOH by tuning aluminum spatial distribution in the membrane. The membrane behaved a stable water flux of 2.2 kg m-2 h-1 while separation selectivity was very high. Wang et al. [15] prepared acid-proof Ge-ZSM-5 zeolite membrane for dehydrating AcOH solution. The water flux was nearly constant (about 0.57 kg m-2 h-1) in the acetic acid concentration range of 30-98 wt%. Nevertheless, allowing for the complexity of the real acidic solution, not only the tolerance to weak organic acid but also to strong inorganic acid is worth to concern. Cui et al. [16] tested the stability of T zeolite powder immersed in 1.0 M HCl aqueous solution for 1 h. They found that the treated zeolite crystal decomposed a lot in the acid solution. Zhu et al. [17] prepared MOR zeolite membrane with higher Si/Al ratio of 5 and adopted it in esterification of acetic acid and alcohol with H2SO4 as the catalyst. They did not also find that the membrane had good stability in the presence 4
of H2SO4. It seems that such zeolite membranes containing aluminum can hardly withstand strong inorganic acids. DD3R zeolite membrane is a type of zeolite membrane with pore size of 0.36 nm×0.44 nm, which is promising to behave water-selective separation based on molecular-sieving mechanism. Moreover, the all-silica framework structure makes it possible to be stable against strong acids. Until now, there have been very limited studies on DD3R zeolite membrane and most of them have been focused on CO2 separation [18-22]. Kuhn et al. [23] used a tubular DD3R zeolite membrane for dehydrations of methanol and ethanol. The membrane exhibited a water flux of 2 kg m-2 h-1 and a separation factor of 1500 for 18 wt% water/ethanol mixture at 373 K, confirming that DD3R zeolite membrane is a high-potential candidate for solvent dehydration. Although NGK Insulators, Ltd., has published several literatures on DD3R zeolite membrane in the last decade [18, 19], it has been still great challenge to prepare high-quality DD3R zeolite membrane reproducibly because the competitive impurity crystals such as Dodecasil 1H (DOH) and Sigma-2 (SGT) (Fig.S1) which have different topological structures are very easily grown in the membrane and thermal-induced expansion of zeolite membrane during the high-temperature calcination often causes intercrystalline defects [22]. To address the issue, we developed a secondary-growth method using ball-milled Sigma-1 zeolite (a structural analogue of DD3R with aluminium content in the framework) seeds followed by low-temperature calcination in mixing ozone/oxygen atmosphere in our previous work [24]. The as-synthesized membranes achieved high CO2/CH4 separation 5
selectivity of over 300. Recently, hollow fiber supported zeolite membranes have been widely attractive due to their virtues of large packing density (as high as 1000 m2 m-3) and high permeation flux [25-28]. As compared to polymeric hollow fiber supports, ceramic hollow fiber supports are more thermally and chemically stable and therefore suitable for hydrothermal synthesis of zeolite membranes. However, as the conventional single-channel ceramic hollow fiber supports are very fragile, the industrialization of the supported zeolite membranes have been still limited. To overcome this issue, we proposed a novel four-channel ceramic hollow fiber support with larger outer diameter and cross-linking configuration [29, 30]. Such structure made the obtained supported zeolite membranes possess not only excellent separation performance but also robust mechanical strength in pervaporation and gas separation [31, 32]. Following up our previous work, we further prepared DD3R zeolite membranes on ceramic four-channel hollow fiber supports with different qualities. The DD3R zeolite membranes were firstly used for pervaporation dehydration of AcOH. Inorganic acids such as HCl and H2SO4 were also included in AcOH for stability evaluation. Influences of temperature and feed water content on PV performance were studied experimentally and theoretically. PV stability of DD3R zeolite membrane in acid environment was tested for a long operation time.
2. Experimental 2.1 Membrane preparation 6
DD3R zeolite membranes were prepared on four-channel ceramic hollow fiber supports by the secondary growth method. The supports were prepared by the combined phase-inversion and sintering method in our laboratory [29, 30]. The obtained supports, as shown in Fig. 1, had outer/inner diameter (O.D./I.D.) of about 3.6 mm/0.9 mm, average pore size of 0.5 μm and porosity of ~40%. The outer surface of support was seeded with ball-milled Sigma-1 zeolite particles (average particle size 0.5 μm) by dipping into an aqueous seed suspension (1 wt%) for 15 s. The detailed preparation process of ball-milled Sigma-1 zeolite seeds were described in our previous work [24]. The seeded supports were dried at 373 K overnight. The synthesis precursor was prepared by mixing 1-adamantanamine (ADA, Sigma-Aldrich), colloidal silica (Ludox SM-30, Sigma-Aldrich), ethylenediamine (EDA, Sigma-Aldrich) and deionized water with a molar composition of ADA: SiO 2: EDA: H2O=3:100:50:4000. After stirring for one hour, the synthesis mixture was loaded in a Teflon-lined autoclave and used for membrane synthesis by immersing the seeded support. The hydrothermal crystallization was carried out at 413 K for a certain time, and subsequently the as-synthesized membranes were washed and dried overnight. Detemplation of DD3R zeolite membrane was conducted in a stainless-steel module in mixing ozone/oxygen atmosphere at 473 K for 80 h. The heating and cooling rates were controlled at 1 K min-1. The detailed procedure was also described in our previous work [24]. The morphologies of DD3R zeolite membranes were characterized by Field Emission Scanning Electron Microscopy (FE-SEM, S-4800, Hitachi). The crystal 7
phases were determined by X-ray diffraction (XRD, MiniFlex 600, Rigaku) with Cu Kα radiation in the 2θ rang of 5-50°.
2.2 Gas separation Before PV separation, the qualities of DD3R zeolite membranes were evaluated by binary CO2/CH4 gas separation. The membrane was mounted in a stainless steel module, with each end sealed with a silicone O-ring. The feed pressure was fixed at 0.2 MPa (absolute pressure) by a back pressure regulator, and the permeate side was operated under atmospheric pressure. An equimolar CO2/CH4 flow was fed at a rate of 80 mL min-1 and a helium stream (50 mL min-1) was used to sweep the permeate side. A gas chromatography (GC, GC-2014, Shimadzu) equipped with thermal conductivity detector (TCD) and a packed column of HAYESEP-DB 200, was used to analyze gas compositions. The gas permeance of component i (Pm,i, mol m-2 s-1 Pa-1) is defined as Pm,i=Fi/∆p
(1)
where Fi is permeation flux through the membrane, mol m-2 s-1, ∆p is the pressure drop between both sides of the membrane, Pa. The CO2/CH4 separation selectivity (SCO2/CH4) is calculated from the gas permeances of CO2 and CH4 by SCO2/CH4= Pm,CO2/Pm,CH4
(2)
2.3 PV test The PV experimental apparatus for evaluating separation performance of DD3R 8
zeolite membranes is similar to that in our previous work [11, 12]. One end of the hollow fiber DD3R zeolite membrane was sealed with silicone and the other end was connected to a vacuum line. The effective membrane area was about 4 cm2. The membrane was immersed in the feed tank, which was filled with approximately 2 L of feed liquid mixture. The inside of the membrane was evacuated by a vacuum pump, and the pressure was maintained below 200 Pa. The permeated vapors were collected using cold traps cooled by liquid nitrogen. Both the feed and permeate were analyzed by a gas chromatograph (GC-2014, Shimadzu) equipped with a thermal conductivity detector and a capillary column of PEG-20M (Nanjing Tech). The PV performance of each membrane was determined by separation factor (α) and water permeation flux (J), which are respectively defined as follows: Ji=mi/(A·t)
w/ e
y w / xw y e / xe
(3) (4)
where mi is the water mass (kg) permeated over a time period of t (h); A is the effective membrane area (m2); yw and ye are the weight fractions of water and ethanol in the permeate, respectively; and xw and xe are the weight fractions of water and ethanol in the feed, respectively.
2.4 Molecular simulation For permeations of water/AcOH binary mixtures in DD3R zeolite layer, their adsorption behaviors in DD3R zeolite were studied by molecular simulation using the Accelrys Materials Studio software on a work station as described in our previous 9
work [28]. All the molecular simulations were conducted in 1×2×1 DD3R zeolite unit cells with periodic boundary condition, the formula of the supercell is Si480O960. As shown in Fig. 2, the DD3R zeolite topology structure consists of three types of cages: pentagonal dodecahedra (512 cage), decahedron (435661 cage) and 19-hedron (435126183 cage). A two-dimensional channel system is formed by connecting the 19-hedron cavities through a single 8-ring with an aperture of 0.36 nm×0.44 nm. The COMPASS (condensed-phase optimized molecular potentials for atomistic simulation studies) force field was adopted and the partial atomic charges associated with each atomic species were set as -0.445 for O, +0.89 for Si. The saturation fugacities of water and AcOH (fs,w and fs,A) were calculated by Aspen Plus software using Wilson equation. Their corresponding saturation adsorption loadings (As,w and As,A) were simulated by the GCMC (Grand canonical ensemble Monte Carlo) method using the Sorption module. The electrostatic potential energy and van der Waals potential energy were calculated by the Ewald summation and atom-based techniques, respectively. The cutoff distance for van der Waals potential energy calculation was taken as 12 Å, which was about half the length of the zeolite cell. The spline width and the buffer width were set at the default values, 1.0 Å and 0.5 Å. Here, the equilibration steps were set to 4,000,000 and the production steps were set to 4,000,000 Monte Carlo moves. The self-diffusion behaviors of water/AcOH mixtures were simulated by the EMD (equilibrium molecular dynamics) method. The EMD simulation was carried out by using the Discover module in cases of adsorption simulation results. The 10
adsorbent model and the calculation of non-bonded interaction were the same as those used in adsorption simulations. A smart minimize method with an ultrafine convergence level and equilibration run was carried out for optimization before running the production steps of the MD simulation. The step time was set to 1.0 fs. The total simulation time was about 1600 ps. The initial time of 400 ps was used to perform the MD equilibration for balance, and the rest simulation time was special for the MD production and the trajectories were saved every 1000 steps in the productions simulations for the purpose of analysis. The temperature was controlled using a NVT ensemble and Nose thermostat. The self-diffusion coefficient of component can be calculated by EMD simulation based on the Einstein equation [28]:
Ds,i
2 1 d Na lim ri t ri 0 6 N a t dt i 1
where Na is the number of diffusing molecules, and
ri t ri 0
(5) 2
is the average
of the mean squared displacement (MSD) of the diffusing molecules.
3. Results and discussion 3.1 Membrane preparation DD3R zeolite membranes with different qualities were hydrothermally synthesized at the same synthesis temperature (413 K) for different periods of 40 h, 44 h and 48 h. Fig. 3 shows XRD patterns of the membranes synthesized for different time. For the membrane synthesized for 40 h, the characteristic peaks of DD3R zeolite at 7.5°, 15.5° and 17° are observable but their intensities are very weak, indicating the immature growth of DD3R zeolite membrane. When the synthesis time 11
was 44 h, the intensities of the characteristic peaks become obvious, implying that the membrane had good crystallization quality. However, as the synthesis time was further prolonged, the characteristic peaks of DD3R zeolite nearly disappear. Instead, three characteristic peaks of SGT zeolite at 9.01°, 19.56° and 19.76° occurred. Fig. 4 presents surface and cross-section SEM images of the DD3R zeolite membranes synthesized for different synthesis time. As being consistent with the observation on XRD results, a loose-packing zeolite layer with large voids and defects occur on the surface of the membrane synthesized for 40 h, the thickness of zeolite layer is only about 1 μm. A well-intergrown DD3R zeolite membrane with uniform thickness (about 5 μm) is observed after synthesized for 44 h, indicating good quality of the membrane. As the synthesis time was 48 h, ball-shaped SGT zeolite crystals with large crystal size covered on membrane surface. Table 1 lists CO2/CH4 gas separation results of different DD3R zeolite membranes in this work and literatures. It can be seen that CO2 permeance of the membranes in this work decreased as the synthesis time increased from 40 to 48 h due to the increased membrane thickness. Particularly, the membrane M2 synthesized for 44 h showed the highest CO2/CH4 separation selectivity of 679 while its CO2 permeance is 1.0×10-7 mol m-2 s-1 Pa-1. As compared to other membranes in literatures [18, 19, 22], M2 also exhibited an extremely high separation selectivity as well as a good CO2 permeance. The membranes were further adopted for PV dehydrations of 10 wt% water/AcOH mixture at 348 K. Since the pore size of DD3R zeolite membrane (0.36 nm×0.44 nm) is between the kinetic diameters of water (0.29 nm) 12
and AcOH (0.436 nm), the membranes shown in Table 2 behaved water-selective for separation of H2O/AcOH. M2 exhibited a total flux of 0.17 kg m-2 h-1 as well as a separation factor of 213 in PV dehydration of AcOH/water mixture at 348 K. Table 2 also lists PV results of two zeolite membranes with high Si/Al ratios (Ge-ZSM-5 and hydrophilic silicalite-1) for comparison [15, 33]. As compared, DD3R zeolite membrane showed moderate separation factor due to its strong hydrophobicity. However, its separation factor would be improved by more careful manipulation in membrane synthesis or posttreatment on membrane surface e.g. hydrophilic modification in further work.
3.2 Pure water and AcOH permeation fluxes M2 membrane was further used for PV permeation of pure water and AcOH permeation. During the intervals of PV operation, the membrane was also immersed in water or AcOH. As shown in Fig. S2, the pure water and AcOH fluxes of the membrane at 368 K were about 0.91 and 0.1 kg m-2 h-1 respectively, indicating the possibility of the membrane for separation between water and AcOH. It is noted that although the membrane was immersed in pure water and AcOH for a long time, both water and AcOH fluxes varied by less than 5% at the same temperature. Such highly stable performance was mainly owed to the all-silica structure of DD3R zeolite membrane. Fig. 6 shows the simulated adsorption isotherms and diffusion curves of pure water and AcOH in DD3R zeolite at 368 K. As shown in Fig. 6a, DD3R zeolite 13
preferred to adsorb AcOH than water due to its strong hydrophobicity. The saturation adsorption loadings of water and AcOH were 0.44 and 1.84 mol kg-1 respectively. It can be seen from Fig. 6b that the water molecules performed much larger MSD than that of AcOH. The self-diffusion coefficients of water and AcOH (Dw and DA) were 5.69×10-10 and 4.93×10-11 m2 s-1 respectively. The self-diffusion coefficient of AcOH was very low for its large molecule size in confined zeolitic pores. However, the small molecular diameter of water made it move more freely in the zeolite. In addition, there was little interaction between the water molecules and the hydrophobic zeolite channels.
3.3 PV dehydration of water/AcOH mixture 3.3.1 Effect of temperature Temperature is a significant factor to determine adsorption loading and diffusivity of component in zeolite membrane, which affects the membrane separation performance. Fig. 6a shows PV results of M2 as a function of temperature in dehydration of 10 wt% water/ethanol mixture. As expected, high temperature was beneficial for improving the permeation flux. It can be explained from Table 3 that on one side high temperature increased the saturation fugacity of water in the feed side, which enhanced its driving force for permeation through the membrane; on the other side, the self-diffusion coefficient of water was accelerated at high temperature. As shown in Fig. 6b, both water and AcOH fluxes followed the Arrhenius relationship with temperature. The activation energies of water and AcOH are 23.56 kJ mol-1 and 14
0.57 kJ mol-1 respectively. Generally, in the adsorption–diffusion model of molecular transport through a zeolite membrane, the apparent activation energy (EA) comprises a contribution of both the activation energy for diffusion (ED) as well as the heat of adsorption (ΔHS), EA=ED−ΔHS. When EA>0, it indicates that the activation energy of diffusion dominates over the heat of adsorption. It can be also seen from the table that the saturation adsorption loading of water decreased gradually with the temperature whereas that of AcOH was nearly insensitive to the temperature. Due to the simultaneous water adsorption, the saturation adsorption loading of AcOH in the mixture was lowered to 1.67 mol kg-1 at 368 K as compared with its pure component (1.84 mol kg-1). But, it is interesting to find that the saturation adsorption loading of water in the mixture (0.49 mol kg-1) was higher than that of its pure component (0.44 mol kg-1). Fig. 8 shows the radial distribution function (RDF) of hydrogen atoms in water molecules and oxygen atoms in AcOH molecules in DD3R zeolite at 368 K. Two peaks at about 1.9 Å and 3.3 Å represent the hydrogen bonds between hydrogen atoms of water and carbonyl and hydroxyl oxygen atoms of AcOH, respectively. We considered that such interaction between water and AcOH molecules in the mixture could be the main reason for improving adsorption of water but weakening that of AcOH in the zeolite. It can be also seen from Fig. 7 that high PV temperature was beneficial not only for permeation flux but also for separation factor. At 368 K, the permeation flux was 0.26 kg m-2 h-1 while the separation factor was 345. Such temperature dependency of separation factor for DD3R zeolite membrane was different from those occurred on 15
some hydrophilic zeolite membranes, e.g. NaA zeolite membrane [31], T zeolite membrane [16] and MOR zeolite membrane [34]. It is well known that pervaporation through zeolite membrane can be explained by the adsorption-diffusion mechanism. Water molecules are preferred to adsorb in hydrophilic zeolite membrane pores and sometimes condensate on the surface of small intercrystalline defects. The elevated temperature significantly reduced the adsorption coverage of water on membrane surface and in zeolite cages, which would open pores for diffusion of large organic molecules and thus decrease the separation selectivity. But, towards hydrophobic DD3R zeolite membrane, although the saturation adsorption selectivity of water over AcOH decreased from 0.41 to 0.29 as the temperature climbed up from 348 to 368 K, the corresponding selectivity in self-diffusion coefficient increased more significantly from 17.8 to 28.6. Therefore, the approximate water/AcOH separation selectivity calculated based on the adsorption-diffusion model was improved at higher temperature. It is noted that although the calculated self-diffusion coefficient of water in DD3R zeolite was close to that in NaA zeolite [28], the actual permeation flux of DD3R zeolite membrane was much lower than the expected result. Kuhn et al. [35] studied diffusion behaviors of water/methanol mixture in DD3R zeolite via MD simulation. They also found that the experimental flux was one order of magnitude lower than the predicted flux based on the simulated diffusivities. They attributed the deviation between the simulated and experimental results to many factors, e.g. the assumption of thermodynamic correction factor, the perfect crystalline system, the 16
silanol group present in the actual zeolite and so on. Not only these, as shown in Fig. 9, the diffusion tensors in the directions x, y and z of water were different due to the different channel types of DD3R zeolite. The anisotropic self-diffusion coefficients followed an order of Dy>Dx>Dz, especially Dz was nearly close to zero. This agreed with the topology structure of DD3R zeolite which has a two-dimensional channel system in the xy-plane. The diffusion of molecules along the direction z were hindered by the pentagonal and decahedron cages with small 4-ring and 5-ring windows which are inaccessible for water and AcOH molecules. We suggested that high permeation flux would be achieved on the oriented DD3R zeolite membrane that the 8-ring pore channels are aligned perpendicularly to the support surface since the 2D channel is the fastest diffusion pathway in DD3R crystal. Similar claim was also proposed by Choi et al. on DD3R zeolite membrane for CO2 separation [36]. However, according to the XRD and SEM results, the used DD3R zeolite membrane in this study was randomly oriented, which could be one reason for the large derivations between the experimental and simulated diffusivities and activated energies.
3.3.2 Effect of feed water content Fig. 10 shows PV results of M2 in dehydrations of water/AcOH mixtures with different feed water contents at 368 K. As the feed water content climbed up from 10 to 70 wt%, the permeation flux of the membrane increased gradually from 0.26 to 0.58 kg m-2 h-1. It can be seen from Table 3 that high feed water content led to high saturation fugacity of water, which helped to improve its driving force for permeation 17
through the membrane. When the feed water content varied from 10 to 70 wt%, the saturation adsorption loading of water increased largely from 0.39 to 1.07 mol kg-1 whereas that of AcOH declined a little from 1.67 to 1.57 mol kg-1. It is interesting to find that the self-diffusion coefficient of water decreased gradually under the feed water content from 10 to 50 wt%. Such phenomenon could be owed to the obviously increased adsorption loading of water which reduced vacant coverage in DD3R zeolite. The further increase in the self-diffusion coefficient of water at the feed water content of 70 wt% could be attributed to the large reduction in the adsorption loading of AcOH, which released more vacant sites for water diffusion. On the contrary, the self-diffusion coefficient of AcOH decreased continuously with the feed water content due to increasing adsorption loading of water molecules in DD3R zeolite, which reduced vacant coverage for AcOH diffusion. Not only permeation flux, high separation factor was also achieved at high feed water content. When the feed water content was 70 wt%, the separation factor was as high as 800 while the permeate water content was 99.9 wt%. The improvement in separation factor at high feed water content was mainly caused by the simultaneous increases in adsorption and diffusion selectivities of water over AcOH. As the feed water content raised from 10 to 70 wt%, the adsorption selectivity increased from 0.23 to 0.68 while the selectivity in self-diffusion coefficient increased from 28.6 to 98.9.
3.3.3 Long-term PV stability As mentioned above, the stability of zeolite membrane in acidic solution is a 18
significant concern for practical application. The hollow fiber supported DD3R zeolite membrane (M2) was further used in dehydration of 10 wt% water/AcOH mixture at 348 K to evaluate its acid stability. Two kinds of typical strong inorganic acids (0.1 M hydrochloric acid and 0.05 M sulfuric acid) were added into the feed mixture sequentially. During the PV operation intervals, the used membrane was dried and then soaked in the 10 wt% water/AcOH solution at room temperature. As shown in Fig. 11, the membrane exhibited stable permeation flux of about 0.185 kg m-2 h-1 and separation factor of around 250 during the whole PV test despite high-content AcOH in the feed mixture. It is also interesting to observe that the membrane still maintained good and stable separation performance even in the presences of HCl and H2SO4, indicating its excellent durability to withstand inorganic acid. Table 4 lists PV results of the membrane in dehydration of 10 wt% water/AcOH mixtures with different HCl contents. It seems that the addition of hydrochloric acid in the mixture did not affect the separation performance obviously. It should be noted that there was no report on successful pervaporation dehydration of solvents including strong acids using other types of zeolite membranes so far. van Veen et al. [37] tested the stability of HybSi® pervaporation membrane in the presence of nitric acid with different contents. They found that the maximum concentration of nitric acid to be withstood for the membrane was below 0.08 mol L-1. Comparatively, DD3R zeolite membrane exhibited superior acid-resistance. Fig. 12 shows SEM images of the used membrane after long-term PV test in the presences of hydrochloric acid and sulfuric acid. The surface morphology of the used membrane was nearly the same with the fresh one. 19
No obvious structural change occurred on the membrane surface, which should be attributed to the pure silica crystalline structure.
4. Conclusions High-quality hollow fiber supported DD3R zeolite membranes were prepared and successfully used in pervaporation dehydration of acetic acid. Both high temperature and feed water content were beneficial for improving permeation flux and separation factor of the membrane. Due to the strong hydrophobicity, the membranes behaved morderate permeation flux and separation factor. However, the pure-silica zeolite membranes exhibited a very stable separation performance during the pervaporation in acidic solution, even in the presence of strong acids. It could be helpful to increase the permeation flux largely by grafting hydrophilic groups on the membrane surface. On the other hand, it is highly encouraged to fabricate oriented DD3R zeolite membranes with the 8-ring pore channels aligned perpendicularly on the support surface. We suggest that such highly-stable separation performance makes DD3R zeolite membrane have great potential in application of dehydration of acidic solutions.
Acknowledgement This work is sponsored by the National Natural Science Foundation of China (21490585 and 21776128), the National High-tech R&D Program of China (2015AA03A602), the “Six Top Talents” and “333 Talent Project” of Jiangsu 20
Province, State Key Laboratory of Materials-Oriented Chemical Engineering (ZK201602), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).
21
References [1] K. Makita, Y. Hirota, Y. Egashira, K. Yoshida, Y. Sasakib, N. Nishiyama, Synthesis of MCM-22 zeolite membranes and vapor permeation of water/acetic acid mixtures, J. Membr. Sci. 372 (2011) 269-276. [2] U. S. Toti, M. Y. Kariduraganavar, K. S. Soppimath, T. M. Aminabhavi, Sorption, diffusion, and pervaporation separation of water–acetic acid mixtures through the blend membranes of sodium alginate and guar gum-grafted-polyacrylamide, J. Appl. Polym. Sci. 83 (2002) 259–272. [3] J. Gascon, F. Kapteijn, B. Zornoza, V. Sebastián, C. Casado, J. Coronas, Practical approach to zeolitic membranes and coatings: State of the art, opportunities, barriers, and future perspectives, Chem. Mater. 24 (2012) 2829-2844. [4] Z. Yang, Y. Liu, C. Yu, X. Gu, N. Xu, Ball-milled NaA zeolite seeds with submicron size for growth of NaA zeolite membranes, J. Membr. Sci. 392-393 (2012) 18-28. [5] C. Yu, Y. Liu, G. Chen, X. Gu, W. Xing, Pretreatment of isopropanol solution from pharmaceutical industry and pervaporation dehydration by NaA zeolite membranes, Chin. J. Chem. Eng. 19(2011) 904-910. [6] Y. Morigami, M. Kondo, J. Abe, H. Kita, K. Okamoto, The first large scale pervaporation plant using tubular-type module with zeolite NaA membrane, Sep. Purif. Technol. 25 (2001) 251-260. [7] N. Rangnekar, N. Mittal, B. Elyassi, J. Caro, M. Tsapatsis, Zeolite membranes-a review and comparison with MOFs, Chem. Soc. Rev. 44 (2015) 7128-7154. 22
[8] Y. Hasegawa, T. Nagase, Y. Kiyozumi, T. Hanaoka, F. Mizukami, Influence of acid on the permeation properties of NaA-type zeolite membranes, J. Membr. Sci. 349 (2010) 189-194. [9] X. Cai, Y. Zhang, L. Yin, D. Ding, W. Jing, X. Gu, Electrochemical impedance spectroscopy for analyzing microstructure evolution of NaA zeolite membrane in acid water/ethanol solution, Chem. Eng. Sci. 153 (2016) 1-9. [10] O. Iglesia, R. Mallada, M. Menendez, J. Coronas, Continuous zeolite membrane reactor for esterification of ethanol and acetic acid, Chem. Eng. J. 131 (2007) 35–39. [11] J. Jiang, X. Wang, Y. Zhang, D. Liu, X. Gu, Fabrication of pure-phase CHA zeolite membranes with ball-milled seeds at low K+ concentration, Micropor. Mesopor. Mater. 215 (2015) 98-108. [12] X. Wang, Z. Yang, C. Yu, L. Yin, C. Zhang, X. Gu, Preparation of T-type zeolite membranes using a dip-coating seeding suspension containing colloidal SiO2, Micropor. Mesopor. Mater. 197 (2014) 17-25. [13] M.-H. Zhu, S.-L. Xia, X.-M. Hua, Z.-J. Feng, N. Hu, F. Zhang, I. Kumakiri, Z.-H. Lu, X.-S. Chen, H. Kita, Rapid preparation of acid-stable and high dehydration performance mordenite membranes, Ind. Eng. Chem. Res. 53 (2014) 19168-19174. [14] J. Yang, L. Li, W. Li, J. Wang, Z. Chen, D. Yin, J. Lu, Y. Zhang, H. Guo, Tuning aluminum spatial distribution in ZSM-5 membranes: a new strategy to fabricate high performance and stable zeolite membranes for dehydration of acetic acid, 23
Chem. Commun. 50 (2014) 14654-14657. [15] X. Wang, X. Deng, Z. Bai, X. Zhang, X. Feng, W. Huang, The synthesis of super-hydrophilic and acid-proof Ge–ZSM-5 membranes by simultaneous incorporation of Ge and Al into a Silicalite-1 framework, J. Membr. Sci. 468 (2014) 202–208. [16] Y. Cui, H. Kita, K.-I. Okamoto, Zeolite T membrane: preparation, characterization, pervaporation of water/organic liquid mixtures and acid stability, J. Membr. Sci. 236 (2004) 17-27. [17] M.-H. Zhu, I. Kumakiri, K. Tanaka, H. Kita, Dehydration of acetic acid and esterification product by acid-stable ZSM-5 membrane, Micropor. Mesopor. Mater. 181 (2013) 47-53. [18] T. Tomita, K. Nakayama, H. Sakai, Gas separation characteristics of DDR type zeolite membrane, Micropor. Mesopor. Mater. 68 (2004) 71-75. [19] S. Himeno, T. Tomita, K. Suzuki, K. Nakayama, K. Yajima, S. Yoshida, Synthesis and permeation properties of a DDR-type zeolite membrane for separation of CO2/CH4 gaseous mixtures, Ind. Eng. Chem. Res. 46 (2007) 6989-6997. [20] J. van den Bergh, W. Zhu, J. Gascon, J.A. Moulijn, F. Kapteijn, Separation and permeation characteristics of a DD3R zeolite membrane, J. Membr. Sci. 316 (2008) 35-45. [21] J. van den Bergh, A. Tihaya, F. Kapteijn, High temperature permeation and separation characteristics of an all-silica DDR zeolite membrane, Micropor. Mesopor. Mater. 132 (2010) 137-147. 24
[22] S. Yang, Z. Cao, A. Arvanitis, X. Sun, Z. Xu, J. Dong, DDR-type zeolite membrane synthesis, modification and gas permeation studies, J. Membr. Sci. 505 (2016) 194-204. [23] J. Kuhn, K. Yajima, T. Tomita, J. Gross, F. Kapteijn, Dehydration performance of a hydrophobic DD3R zeolite membrane, J. Membr. Sci. 321 (2008) 344-349. [24] L. Wang, C. Zhang, X. Gao, L. Peng, J. Jiang, X. Gu, Preparation of defect-free DDR zeolite membranes by eliminating template with ozone at low temperature, J. Membr. Sci. 539 (2017) 152–160. [25] P. S. Lee, M. S. Lim, A. Park, H. Park, S. -E. Nam, Y. Park, A zeolite membrane module composed of SAPO-34 hollow fibers for use in fluorinated gas enrichment, J. Membr. Sci. 542 (2017) 123–132. [26] H. Yan, N. Ma, Z. Zhan, Z. Wang, Fabrication of zeolite NaA membranes on hollow fibers using nano-sized seeds exfoliated from mesoporous zeolite crystals, Micropor. Mesopor. Mater. 215 (2010) 244-248. [27] V. Sebastian, R. Mallada, J. Coronas, A. Julbe, R. A. Terpstra, R. W. J. Dirrix, Fabrication of zeolite NaA membranes on hollow fibers using nano-sized seeds exfoliated from mesoporous zeolite crystals, J. Membr. Sci. 355 (2010) 28–35. [28] P. Ye, Y. Zhang, H. Wu, X. Gu, Mass transfer simulation on pervaporation dehydration of ethanol through hollow fiber NaA zeolite membranes, AIChE J. 62 (2016) 2468-2478. [29] Z. Shi, Y. Zhang, C. Cai, C. Zhang, X. Gu, Preparation and characterization of α-Al2 O3 hollow fiber membranes with four-channel configuration, Ceram. Int. 41 25
(2015) 1333-1339. [30] C. Cai, Y. Zhang, C. Zhang, X. Gu, Microstructure modulation of α-Al2O3 hollow fiber membranes with four-channel geometric configuration, Asia-Pac. J. Chem. Eng. 11 (2016) 949–957. [31] D. Liu, Y. Zhang, J. Jiang, X. Wang, C. Zhang, X. Gu, High-performance NaA zeolite membranes supported on four-channel ceramic hollow fibers for ethanol dehydration, RSC Adv. 5 (2015) 95866–95871. [32] Y. Chen, Y. Zhang, C. Zhang, J. Jiang, X. Gu, Fabrication of high-flux SAPO-34 membrane on α-Al2O3 four-channel hollow fibers for CO2 capture from CH4, J. CO2 Util. 18 (2017) 30–40. [33] T. Masuda, S. Otani, T. Tsuji, M. Kitamura, S. Mukai, Preparation of hydrophilic and acid-proof silicalite-1 zeolite membrane and its application to selective separation of water from water solutions of concentrated acetic acid by pervaporation, Sep. Purif. Technol. 32 (2003) 181-189. [34] G. Li, E. Kikuchi, M. Matsukata, Separation of water-acetic acid mixtures by pervaporation using a thin mordenite membrane, Sep. Purif. Technol. 32 (2003) 199-206. [35] J. Kuhn, J. M. Castillo-Sanchez, J. Gascon, S. Calero, D. Dubbeldam, T. J. H. Vlugt, F. Kapteijn, J. Gross, Adsorption and diffusion of water, methanol, and ethanol in all-silica DD3R: Experiments and simulation, J. Phys. Chem. C 113 (2009) 14290-14301. [36] E. Kim, W. Cai, H. Baik, J. Nam, J. Choi, Synthesis and sonication-induced 26
assembly of Si-DDR particles for close-packed oriented layers, Chem. Commun. 49 (2013) 7418-7420. [37] H. M. van Veen, M. D. A. Rietkerk, D. P. Shanahan, M. M. A. van Tuel, R. Kreiter, H. L. Castricum, J. E. ten Elshof, J. F. Vente, Pushing membrane stability boundaries with HybSi® pervaporation membranes, J. Membr. Sci. 380 (2011) 124-131.
27
Table 1 Comparison in CO2/CH4 gas separation results between different DD3R zeolite membranes. Membrane/Support Synthesis temperature
Membrane
CO2/CH4 separation
Reference
(K)/Time (h)
thickness (μm)
CO2 permeance (10-7 mol m-2 s-1 Pa-1)
CO2/CH4 selectivity
DD3R/Tube
423/48
5-10
0.7
280
[18]
DD3R/Tube
423/48
2-3
3.0
200
[19]
DD3R/Disc
433/24
10
~1.8
92
[22]
M1/Hollow fiber
413/40
~1
2.9
1.2
This work
M2/Hollow fiber
413/44
5
1.0
679
This work
M3/Hollow fiber
413/48
7
0.17
34
This work
28
Table 2 Comparison in PV results of acetic acid dehydration between different zeolite membranes.
a
Membrane
Feed mixture
PV temperature (K)
Total flux (kg m-2 h-1)
αwater/AcOH
Reference
Ge-ZSM-5
2 wt% water/AcOH
353
0.57
-a
15
Hydrophilic silicalite-1
2 wt% water/AcOH
353
~2×10-4
-a
33
M1
10 wt% water/AcOH
348
0.57
3.6
This work
M2
10 wt% water/AcOH
348
0.17
213
This work
M3
10 wt% water/AcOH
348
0.08
55
This work
The permeate acetic acid concentration was too low to be detected.
Table 3 Calculated saturation fugacity, adsorption loading and self-diffusion coefficients of 10 wt% water/AcOH mixture under different temperatures. Temperature
Water
AcOH
(K)
fs,w (kPa)
As,w (mol kg-1)
Ds,w (m2 s-1)
fs,w (kPa)
As,A (mol kg-1)
Ds,A (m2 s-1)
348
12.21
0.70
1.76×10-10
17.67
1.68
9.83×10-12
29
358
18.33
0.59
2.60×10-10
25.95
1.68
1.10×10-11
363
22.25
0.47
3.00×10-10
31.17
1.67
1.15×10-11
368
26.83
0.49
3.86×10-10
37.21
1.67
1.35×10-11
30
Table 4 Calculated saturation fugacity, adsorption loading and self-diffusion coefficients of water/AcOH mixtures under different feed water contents at 368 K. Feed water
Water
AcOH
content (wt%)
fs,w (kPa)
As,w (mol kg-1)
Ds,w (m2 s-1)
fs,w (kPa)
As,A (mol kg-1)
Ds,A (m2 s-1)
10
26.83
0.49
3.86×10-10
37.21
1.67
1.35×10-11
30
52.50
0.76
2.45×10-10
23.17
1.67
3.83×10-12
50
65.31
0.92
2.00×10-10
14.10
1.60
3.00×10-12
70
74.34
1.07
2.64×10-10
7.03
1.57
2.67×10-12
31
Table 5 PV results of M2 in dehydration of 10 wt% water/AcOH mixtures with different HCl contents at 348 K. HCl content in mixture (mol L-1)
Total flux (kg m-2 h-1)
αwater/AcOH
0
0.19±0.02
250±50
0.01
0.19±0.02
312±30
0.05
0.19±0.01
266±21
0.10
0.21±0.01
249±36
32
Figure captions Fig. 1
Cross-section and outer surface images of four-channel ceramic hollow fiber support.
Fig. 2
Schematic diagram of 1×2×1 unit cells of DD3R zeolite: left, (100) view and right, (001) view.
Fig. 3
XRD patterns of DD3R zeolite membranes synthesized for different time.
Fig. 4
Surface (a, c, e) and cross-section (b, d, f) SEM images of DD3R zeolite membrane synthesized for different time: 40 h (a, b); 44 h (c, d); 48 h (e, f).
Fig. 5
Simulated adsorption isotherms (a) and MSD vs. t curves (b) of pure water and AcOH in DD3R zeolite at 368 K.
Fig. 6
PV results of M2 for dehydration of 10 wt% water/AcOH mixture under different temperatures (a) and Arrhenius plot of water and AcOH fluxes (b).
Fig. 7
RDF of hydrogen atoms (water) and oxygen atoms (AcOH) in DD3R zeolite at 368 K..
Fig. 8
MSD of isotropic and anisotropic self-diffusion coefficient of water in DD3R zeolite for 10 wt% water/AcOH mixture at 368 K.
Fig. 9
PV results of M2 for dehydration of water/AcOH mixtures at 368 K under different feed water contents.
Fig. 10
Long-term PV results of M2 for dehydration of 10 wt% water/AcOH 33
mixture at 348 K. Fig. 11
Surface (a) and cross-section (b) SEM images of the used DD3R zeolite membrane after long-term PV test.
34
Fig. 1 Cross-section and outer surface images of four-channel ceramic hollow fiber support.
35
Fig. 2 Schematic diagram of 1×2×1 unit cells of DD3R zeolite: left, (100) view and right, (001) view.
36
Fig. 3 XRD patterns of DD3R zeolite membranes synthesized for different time.
37
Fig. 4 Surface (a, c, e) and cross-section (b, d, f) SEM images of DD3R zeolite membrane synthesized for different time: 40 h (a, b); 44 h (c, d); 48 h (e, f).
38
Fig. 5 Simulated adsorption isotherms (a) and MSD vs. t curves (b) of pure water and AcOH in DD3R zeolite at 368 K.
39
Fig. 6 PV results of M2 for dehydration of 10 wt% water/AcOH mixture under different temperatures (a) and Arrhenius plot of water and AcOH fluxes (b).
40
Fig. 7 RDF of hydrogen atoms (water) and oxygen atoms (AcOH) in DD3R zeolite at 368 K.
41
Fig. 8 MSD of isotropic and anisotropic self-diffusion coefficient of water in DD3R zeolite for 10 wt% water/AcOH mixture at 368 K.
42
Fig. 9 PV results of M2 for dehydration of water/AcOH mixtures at 368 K under different feed water contents.
43
Fig. 10 Long-term PV results of M2 for dehydration of 10 wt% water/AcOH mixture at 348 K.
44
Fig. 11 Surface (a) and cross-section (b) SEM images of the used DD3R zeolite membrane after long-term PV test.
45
Graphic abstract
46
Supporting information
Fig. S1 Framework structures of (a) DD3R, (b) DOH and (c) SGT zeolites.
Fig. S2 Pure water and AcOH fluxes through DD3R zeolite membrane (M2) at 368 K.
47
Graphical abstract
48
Highlights 1. High-quality DD3R zeolite membrane was fabricated on hollow fiber support. 2. DD3R zeolite membrane was firstly used in pervaporation dehydration of acetic acid. 3. The membrane showed excellent acid-resistance in the pervaporation separation.
49