Novel thermally cross-linked polyimide membranes for ethanol dehydration via pervaporation

Author’s Accepted Manuscript Novel thermally cross-linked polyimide membranes for ethanol dehydration via pervaporation Sheng Xu, Yan Wang

www.elsevier.com

PII: DOI: Reference:

S0376-7388(15)30149-6 http://dx.doi.org/10.1016/j.memsci.2015.08.055 MEMSCI13942

To appear in: Journal of Membrane Science Received date: 2 June 2015 Revised date: 20 August 2015 Accepted date: 25 August 2015 Cite this article as: Sheng Xu and Yan Wang, Novel thermally cross-linked polyimide membranes for ethanol dehydration via pervaporation, Journal of Membrane Science, http://dx.doi.org/10.1016/j.memsci.2015.08.055 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 galley proof before it is published in its final citable 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.

Novel thermally cross-linked polyimide membranes for ethanol dehydration via pervaporation

Sheng Xu and Yan Wang*

1

Key Laboratory for Large-Format Battery Materials and System, Ministry of

Education, School of Chemistry and Chemical Engineering, Huazhong University of Science and Technology, Wuhan 430074, China

* Corresponding author. Tel.: 86 027-87793436; fax: 86 027-87543632. E-mail address: [email protected] (Yan Wang)

1

Abstract In this work, two novel carboxyl-containing polyimides, 2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane acid

dianhydride-4,4'-diaminodiphenylmethane/3,5-diaminobenzoic

(6FDA-MDA/DABA,

FMD)

and

3,3',4,4'-benzophenone

dianhydride-4,4'-diaminodiphenylmethane/3,5-diaminobenzoic

tetracarboxylic acid

(BTDA-MDA/DABA, BMD), are synthesized via chemical and thermal imidization methods, respectively, and employed as pervaporation membranes for ethanol dehydration. Chemical structures of the two polyimides are examined by FTIR and TGA to confirm the successful synthesis. A post thermal treatment of the polyimide membranes with the temperature range of 250 to 400 oC is applied, and its effects on the membrane morphology and separation performance are studied and characterized by FTIR, TGA, WXRD, solubility and sorption test. It is believed that the thermal treatment of the carboxyl-containing polyimide membrane at a relative low temperature only leads to the physical annealing, while it may cause the decarboxylation-induced cross-linking at a higher temperature. In addition, the operation temperature in pervaporation is also varied and shown to be an important factor to affect the final membrane performance. Performance benchmarking shows that the developed polyimide membranes both have superior pervaporation performance to most other flat-sheet dense membranes. This work is believed to shed useful insights on polyimide membranes for pervaporation applications.

2

Graphic Abstract for Novel thermally cross-linked polyimide membranes for ethanol dehydration via pervaporation Sheng Xu and Yan Wang Key Laboratory for Large-Format Battery Materials and System, Ministry of Education, School of Chemistry and Chemical Engineering, Huazhong University of Science and Technology, 430074, Wuhan, China 500

FMD 450

400 50 350 45 300

40

80

250 200 (origin)

250

300

350

400

1500

BMD



Total flux (g/m2h)

70

1200

60 900 50 600 40 300

30 20

0 200 (origin)

250

300

350

Separation factor (Water/ethanol)

Total flux (g/m2h)

55

Separation factor (Water/ethanol)

60

400

Thermal treatment temperature (oC)

Keywords: pervaporation

membrane;

carboxyl-containing

polyimide;

thermal

treatment;

cross-linking; ethanol dehydration

3

1. Introduction

Pervaporation, which combines the permeation and evaporation processes of liquid components across the membrane, is a membrane-based separation technology for molecular-level liquid/liquid separations, and has been commercialized in the modern industry over the past decades. Compared to conventional separation technologies, pervaporation presents many attractive advantages such as energy efficiency, environmental benignity and high separation efficiency especially for the separation of azeotropic liquid mixtures [1–3]. Based on the solution-diffusion transport mechanism, the membrane is the key factor in order to obtain a satisfactory pervaporation performance.

For bioalcohol dehydration, polymeric membranes are generally used because of their low cost and high processability as compared to inorganic membranes. In the early time, most researches were concentrated on the hydrophilic polymeric membrane materials such as polyvinyl alcohol, chitosan and sodium alginate etc. [4–7], all of which show high permeation flux and good selectivity towards water. However, grave limitations exist for the wide applications of above membrane materials due to their weak mechanical strength, poor stability and severe membrane swelling in the feed solutions.

Alternatively, polyimide, an excellent candidate as pervaporation membrane material, has drawn more and more attention in recent years because of their superior physicochemical properties to most conventional hydrophilic polymer materials mentioned above [8–11], such as higher thermal resistance, greater mechanical 4

properties and better operation stability. Polyimide is a typical glassy polymer usually prepared by condensation polymerization of dianhydride and diamine monomers. The facial routes of polyimide synthesis include two strategies based on its solubility in organic solvent [2,12]. For the synthesis of both soluble and insoluble polyimides, the first step is to form a polyamic acid precursor via addition polymerization of dianhydrides and diamines. Then the second step  imidization can be conducted with heating or chemical reagent to convert the polyamic acid to the corresponding polyimide. Obviously, the properties of the corresponding polyimide mainly depend on chemical structures of the reaction monomers. Therefore, the molecular design of the polyimide can be carried out via selection of suitable monomers, thus to further manipulate the separation performance of the polyimide membrane easily.

However, in spite of the low swelling property of most polyimide materials, it still remains a big problem to the membrane stability during the pervaporation process. Therefore, a number of modification methods of polyimide membranes have been exploited to improve their physicochemical properties and obtain a more stable separation performance of the membranes. One of the most efficient modifications is cross-linking, which can be implemented by irradiation, thermal treatment or a chemical crosslinker [13]. The selectivity of the resulted membrane is therefore improved, while the corresponding flux is reduced to some extent accordingly. The most common method for polyimide modification is diamino cross-linking, which has been extensively investigated and demonstrated to be a useful way to enhance the gas separation [14–16] and pervaporation performance [17,18] of polyimide membranes. In addition, diol [18–20] and thermal cross-linking [18,21–24] have also been reported to modify the carboxylic-containing polyimides efficiently. 5

Among them, thermal crosslinking is particularly attractive since no external cross-linker is employed, where it involves a decarboxylation-induced cross-linking reaction at a relative high temperature. To our best knowledge, only a few researches on thermally cross-linked polyimide membranes have been investigated for the membrane separation. And systematic studies on the effect of thermal cross-linking temperature on the membrane morphology and separation performance of carboxyl-containing polyimide membranes are also limited [22–24]. Koros and co-workers

developed

2,2'-bis(3,4-dicarboxyphenyl)

hexafluoropropane

dianhydride-2,4,6-trimethyl-1,3-diaminobenzene/3,5-diaminobenzoic

acid

(6FDA-DAM/DABA) membranes for gas separation and found that membranes can be thermally cross-linked at a high temperature (15 oC above glass transition temperature) within a short time [21] or at low temperatures (much below the glass transition temperature) for a period long enough [22,23]. Maya et al found that 2,2'-bis(3,4-dicarboxyphenyl)

hexafluoropropane

dianhydride-4,4'-oxydianiline/3,5-diaminobenzoic

acid

(6FDA-ODA/DABA)

membranes could be thermally cross-linked during the pyrolysis process at temperatures above Tg, resulting in an enhanced permeability and an improved plasticization resistance to CO2 [24]. On the other hand, Le et al employed the thermally-cross-linked

2,2'-bis(3,4-dicarboxyphenyl)

hexafluoropropane

dianhydride-1,5-naphthalene diamine/3,5-diaminobenzoic acid (6FDA-NDA/DABA) as pervaporation membranes for ethanol dehydration, and achieved a high flux and a comparable separation factor [20].

In this work, two novel polyimides with similar chemical structures, i.e., 6

6FDA-MDA-DABA and BMD-MDA-DABA polyimides, are synthesized via chemical and thermal imidization methods, respectively, and employed as pervaporation membranes for ethanol dehydration. Their chemical structures are characterized by FTIR and TGA to confirm the successful synthesis. Effects of further thermal treatment temperature with a range of 250 to 400 oC are investigated on the morphology and performance of the polyimide membranes and characterized by FTIR, TGA, WXRD, solubility and swelling test. The impact of operation temperature is also carried out. We believe this work could make important contributions to the fabrication and modification of polyimide membranes.

2. Experimental

2.1. Materials

2,2'-bis (3,4-dicarboxyphenyl) hexafluoropropane dianhydride (6FDA) was purchased from Aladdin and purified by vacuum sublimation. 3,3',4,4'-benzophenone tetracarboxylic dianhydride (BTDA) from Shanghai Flute Cypress Chemical Technology Co., Ltd. was purified by recrystallization from acetic anhydride (Ac2O) and

dried

in

a

vacuum

oven

at

180

o

C.

The

diamine

monomers,

4,4'-diaminodiphenylmethane (MDA) and 3,5-diaminobenzoic acid (DABA), provided by Aladdin, were purified via recrystallization from ethanol and vacuum sublimation, respectively. All of the above monomers were used immediately for polyimide polymerization after purification. 1-methyl-2-pyrrolidinone (NMP) and Ac2O were evaporated and subsequently dehydrated with 4 Å molecular sieves prior to use. Triethylamine (TEA), N,N-dimethylformamide (DMF) and methanol of 7

analytical grade were supplied by Sinopharm Chemical Reagent Co., Ltd. and used as received. The deionized water was supplied by Wuhan PinGuan.

2.2. Preparation of nascent polyimide dense membranes

2.2.1. Preparation of 6FDA-MDA/DABA polyimide dense membrane

6FDA-MDA/DABA polyimide was synthesized by a two-step route via chemical imidization (Fig. 1.) according to the previous literature [25]. In brief, one diamine, MDA, was completely dissolved in NMP under mechanical agitation, followed by the addition of the stoichiometric amounts of another diamine, DABA, and the dianhydride, 6FDA with the MDA/DABA/6FDA mass ratio of 7/3/10. Then the addition polymerization was conducted with stirring for 24 hours to form a 15 wt% polyamic acid (PAA) solution. Subsequently, TEA and Ac2O (with a molar ratio of 1:4 to 6FDA), acted as the catalyst and dehydrating agent respectively, were added to the solution, and subsequently stirred for another 24 hours to produce polyimide. The whole synthesis process was carried out at ambient temperature under nitrogen protection. The resulted polyimide solution was precipitated in methanol solution, collected by filtration, washed with water, and dried in a vacuum oven at 200 oC for 24 hours to obtain FMD solid.

6FDA-MDA/DABA dense membrane (denoted as FMD-origin) was fabricated by using a 15 wt% 6FDA-MDA/DABA polyimide solution in DMF. The prepared solution was filtered to remove any existing impurities in order to obtain defect-free membranes, degased for overnight to remove bubbles, and then cast onto a glass plate 8

with a casting knife at the thickness of 200 μm. The as-cast membrane was then placed in a vacuum oven at 50 oC for 18 hours to evaporate the solvent slowly. Subsequently, the temperature was increased to 200 oC at a rate of 2.5 oC/mim, maintained for 4 hours to remove the residual solvent, and naturally cooled to room temperature. The resulted FMD membrane was peeled off from the glass plate carefully by being immersed in hot water and further dried at 80 oC for 24 hours under vacuum. The thickness of as-fabricated FMD-origin membranes was about 13–17 μm measured by a Mitutoyo micrometer.

2.2.2. Preparation of BTDA-MDA/DABA polyimide dense memrbane

BTDA-MDA/DABA polyimide was also synthesized from a two-step route but via thermal imidization (Fig. 1.) following a previous work by Xu et al. [26]. The synthesis of PAA precursor was carried out using the similar method as above except that BTDA was used as the dianhydride here. The obtained PAA solution was filtered and cast onto a glass plate with a casting knife at the thickness of 200 μm directly. Then the as-cast membrane was placed in a vacuum oven with the following temperature procedure: (1) 75 oC for 18 hours to evaporate most solvent slowly, (2) 75 oC to 200 oC with a heating rate of 2.5 oC/min, (3) 200 oC for 4 hours to induce the thermal imidization of BTDA-MDA/DABA polyimide membrane and remove the residual solvent, (4) 200 oC to room temperature with the temperature cooled down naturally. The resultant BMD membrane (denoted as BMD-origin) was also peeled off from the glass plate carefully by immersion in hot water and dried at 80 oC for 24 hours under vacuum. The measured thickness was about 14–19 μm.

9

2.3. Thermal treatment of polyimide dense membranes

The original FMD and BMD membranes were thermally treated using the same protocol. In brief, The membrane was wrapped in an aluminum foil and treated in a muffle furnace (AnHui BEQ Equipment Technology CO. Ltd, MF-1100C) by increasing the temperature to a specific temperature (250, 300, 350, 400 oC) with a heating rate of 5 oC/min and then maintaining the temperature for 30 min before cooling down naturally to ambient temperature.

2.4. Pervaporation study

A static pervaporation cell of laboratory scale was used for the pervaporation test and its schematic design was depicted elsewhere [27]. The membrane was loaded in the stainless steel permeation cell with an effective area of 13.5 cm2. The feed solution of about 350 mL was added into the cell contacting the upper surface of membranes, while the bottom surface of the membrane (permeate side) was under vacuum (less than 2 mbar) maintained by a vacuum pump. For all pervaporation tests, there was a 2-hour conditioning for each membrane sample to reach a steady state. Subsequently, at least two permeate samples for all the tested membranes were collected in a cold trap cooled with liquid nitrogen. The composition of the feed and the permeant was determined by Agilent Technologies GC 7890 A equipped with a TCD detector using an Agilent J&W column and nitrogen carrier gas. The feed composition was 85/15 wt% ethanol/water, and varied less than 0.5 wt% during the entire experiment and therefore was considered constant because of the large volume of the feed solution compared with the permeate. The operation temperature was 40 oC in most cases, unless 10

specially noted elsewhere. The permeation flux (J) and the separation factor (α) are calculated according to the following Eqs. (1) and (2): J

Q At

(1)

where Q represents the total mass of component in the permeate (g), A stands for the effective membrane area (m2), and t is the operation time (h). 1/2 

yw,1 / yw,2 xw,1 / xw,2

(2)

where subscripts 1 and 2 refer to water and alcohol, respectively, while yw and xw represent the weight fractions of the component in the permeate and feed, respectively.

The flux and the separation factor were further converted into two intrinsic performance indexes, i.e., permeability (Pi) and selectivity () to exclude the effect of the external driving force, according to the following Eqs. (3) and (4): Pi 

Pi Ji  l xn,i i pisat  yn ,i p p

1/ 2 

P1 P2

(3)

(4)

where the membrane permeability of the component i, Pi, is the product of its diffusivity and solubility coefficients, l is the membrane thickness, xn,i and yn,i are mole fractions of the component i in the feed and permeate, γi is the activity coefficient, pisat and pp are the saturated vapor pressure of the component i in the feed temperature and the permeate pressure, respectively. pisat and γi were obtained with the AspenTech DISTIL software [28].

2.5. Characterizations 11

The molecular weight of 6FDA-MDA/DABA polyimide was measured by gel permeation chromatography (GPC) (Agilent 1100 series) with polystyrene standards and a refractive index detector. The samples were prepared by being dissolved in tetrahydrofuran (THF) with a concentration of 20 mg/mL. The intrinsic viscosity of 6FDA-MDA/DABA polyimide was tested with a 0.5 g/dl FMD/NMP solution at 25 °C using an Ubbelohde capillary viscometer.

The chemical structures of the PAA precursors and corresponding polyimide membranes were characterized by Fourier transform infrared spectroscopy (FTIR) (Bruker Tensor 27) in a range of 4000–400 cm-1, with an average of 16 scans and a resolution of 2 cm-1. The UV (ultraviolet) spectra were recorded with a Shimadzu UV-3600 spectrophotometer in the range of 200–900 nm. PAA precursors (FMD-PAA and BMD-PAA) were obtained by drying the PAA solution under vacuum at 75 oC for 3 days. The thermal decomposition behavior of original and thermally-treated membranes was recorded on a PerkinElmer TGA7 analyzer at a heating rate of 20 o

C/min with the temperature ranging from 50 to 800 oC under an argon atmosphere.

Wide-angle X-ray diffraction (WAXD) (PAnalytical X’pert Pro diffractometer) was conducted by using Cu Kα radiation with a wavelength 1.54 Å, in the range of 10–40o. The average inter-chain spacing distance (d-spacing) value was evaluated by Bragg’s equation as follows: n  2d sin 

(5)

where n is an integer (n=1 in this study), λ represents the X-ray radiation wavelength, d stands for the d-spacing of the polymer chains, and θ refers to the X-ray diffraction angle of the peak. 12

The solubility test was conducted to examine the cross-linking degrees of FMD and BMD membranes with various thermal treatment temperatures. The original and thermally-treated membrane samples with similar shapes and weights were immersed in NMP at 50 oC for a given period, and the dissolution progress was recorded by a conventional camera.

For sorption test (swelling test), pre-dried FMD and BMD membrane strips were weighed and immersed in pure water and pure ethanol liquids, respectively. After being soaked in the liquids at a specific temperature for one month (the weight of the immersed membrane samples showed no notable change), the samples were taken out, blotted by tissue papers and weighed. The swelling degree was calculated according to the following Eq. (6):

Swelling degree (mg /g membrane) =

Wwet  Wdry Wdry

(6)

where Wwet and Wdry represents the weight of the wet membranes at equilibrium and the original dry membranes, respectively.

The water contact angle on the membrane surface was measured by a Kruss ZSA25 Contact Angle Goniometer at ambient temperature, where the deionized water droplets were automatically dropped by a microsyringe onto the top surface of the membrane. For each membrane, at least 8 points on the membrane surface were measured to obtain the average result.

The tensile strength of the polyimide membranes was characterized by an Instron 13

5542 Material Testing Instrument at room temperature (25 oC). Each sample was clamped at the both ends with an initial dimension of 15x5 mm, and stretched vertically at a speed of 10 mm/min until breakage. At least three samples were tested for each membrane to get the average value.

3. Results and discussion

3.1. Characterizations of original polyimides

Due to the poor solubility of BMD polyimide in most common solvents, the GPC test and viscosity test are only carried out for FMD polyimide to get its molecular weight and inherent viscosity. The molecular weight of FMD is 8.07×104 (Mw) with a polydispersity of 2.3 obtained from GPC. And its inherent viscosity measured by Ubbelohde viscometer is 0.545, which is in the general range of 0.4–1.2 for polyimides with similar structures, depending on the molecular weight of the synthesized polymer [29–33].

Chemical structures of the synthesized PAA precursors and corresponding polyimides are confirmed by FTIR spectra as shown in Fig. 2. In the spectra of FMD-PAA (Fig. 2(a)), several characteristic peaks of amide groups are observed: 1650 cm-1 (C=O stretching), 1550 cm-1 (C–N stretching). And the peak of –CF3 in 6FDA moiety is about 1150 cm-1. On the other hand, there also exit some small characteristic peaks of imide groups at 1785 cm-1 (symmetric C=O stretching vibration), 1725 cm-1 (asymmetric C=O stretching vibration), 1354 cm-1 (C–N stretching), 1087 cm-1 (C–N– C transverse stretching), and 714 cm-1 (C–N–C out-of-plane bending), which may be 14

due to the partial induced imidization reaction during the 3-day drying procedure at 75 o

C for the solvent removal. In terms of the spectra of FMD polyimide, several

characteristic peaks of imide groups as mentioned above are observed; and there are no obvious peaks around 1650 and 1550 cm-1 characteristic of amide groups, indicating the successful conversion of PAA to polyimide. Based on above analysis, it is believed that FMD polyimide is successfully synthesized in this work. It is noted that the broad peak at 3200–3800 cm-1 are observed in both spectra of FMD-PAA and FMD, which is probably induced by the combination of the broad peak of O–H in H2O (3400–3800 cm-1) and carboxyl groups (3200–3400 cm-1). Similarly, in the spectra of both BMD-PAA precursor and BMD polyimide, corresponding characteristic peaks of imide groups at 1780, 1721, 1377, 1094 and 720 cm-1 are observed; O–H bond vibration in DABA can be also found at 3200–3400 cm-1. But no obvious peaks at about 1650 and 1550 cm-1 characteristic of amide groups can be seen in BMD spectra, which indicates that the polyamic acid has also been imidized into BMD polyimide.

TGA characterization of both original polyimide membranes also proves the successful synthesis from another aspect, as shown in Fig. 3. As noted, there is a weight decrease in the initial stage probably caused by the removal of residual moisture or solvent. Besides, both TGA curves exhibit two obvious weight loss steps. The first obvious weight loss stage is in the regions of 250–400 oC and 300–400 oC respectively for FMD and BMD polyimides, which should be caused by decarboxylation. Similar thermal behavior can be also observed for other carboxylcontaining polyimides [18,22]. The weight loss of this stage is about 2.1 wt% and 3.3 wt% for FMD and BMD respectively, which is roughly consistent with the theoretical 15

contents of carboxylic groups in the original polyimides. The second decomposition stage, starting at about 450 oC and 470 oC for FMD and BMD membranes respectively, should be due to the degradation of polymer backbones, and are in the range of the Tds of the reported polyimide materials [18,22,24]. In summary, the TGA results also demonstrate the successful imidization reaction.

3.2. Characterizations of polyimide membranes thermally-treated at different temperatures

As mentioned in the previous section, for carboxyl-containing polyimides, thermal cross-linking may occur at a high temperature to result in the elimination of carboxyl groups, as shown in Fig. 4. Two carboxyl groups in adjacent DABA moieties can react at high temperatures to result in an anhydride intermediate. Subsequently, the anhydride is decarboxylated by releasing one CO2 molecule and one CO molecule, resulting in two phenyl free radicals at sufficiently high temperature. Then the resultant neighboring phenyl free radicals interact to form the cross-linked polyimide [21,22]. FTIR characterization of the original and thermally-treated membranes in Fig. 5 confirms this hypothesis. Due to the minor content of DABA structure in both polyimides, the weak characteristic peak of C=O bond vibration is partially overlapped by that of imide groups. Therefore, the only obvious difference reflected in the FTIR spectra of original polyimide and thermally-treated polyimides is the peak intensity change of –OH groups in the range of 3200–3400 cm-1. As shown in Fig. 5, the peak intensities for both FMD and BMD polyimides exhibit slightly decreasing trends, which implies the increase in the decarboxylation degree with the increase in the thermal treatment temperature [18–24]. 16

To further examine the decarboxylation-induced cross-linking with different treated temperature, TGA characterizations of polyimide membranes thermally-treated at different temperatures were also conducted, and the results are shown in Fig. 3. It is noted that obvious carboxyl decomposition stage in the range of 250–400 oC can be found in TGA curves of original PI membrane and those thermally-treated with lower temperatures, while can’t be observed clearly in the spectra of those thermally treated at higher temperatures. In another word, the weight loss of the carboxyl decomposition stage diminishes with an increase in the thermal treatment temperature, which indicates a higher degree of decarboxylation and a higher cross-linking degree of the membranes. In addition, the main decomposition temperatures (Td) of both FMD and BMD membranes show an increasing trend with an increase in the thermal treatment temperature from 250 to 400 oC, which should be ascribed to the stable structure

of

cross-linked

polyimide

with

decarboxylation-induced

chemical

cross-linking by thermal treatment. Specially, for both two polyimides with high thermal treatment temperature up to 400

o

C, there only exists the backbone

decomposition stage, illuminating the complete decarboxylation of the polyimide structures.

For an amorphous glassy polymer material, the d-spacing between polymeric chains can be estimated according to the WXRD diffraction patterns, that is, the higher 2θ value indicates the smaller d-spacing, and vice versa. From Fig. 6, it is observed that the changes of d-spacing values (2θ values) of both polyimides show similar trends. Specifically, the d-spacing of thermally-treated FMD membrane decreases from 5.55 to 5.50 Å with the thermal temperature increasing to 300 oC, and then increases to 17

5.78 Å with the further increase of the thermal temperature to 400 oC. Similarly, for thermally-treated BMD membrane, the d-spacing decreases from 4.22 to 4.10 Å with the thermal temperature increasing to 300 oC and then increases to 4.21 Å with its further increase to 400 oC. The variation in the d-spacing change is probably ascribed to the combined result of two opposite effects, i.e., physical annealing and chemical cross-linking, on the carboxyl-containing polyimides with thermal treatment [22]. In contrast to conventional thermal treatment which only tightens the polymer chain packing, the thermal cross-linking in this work may result in the increase in d-spacing of the carboxyl-containing polyimide membranes. Similar phenomena were also reported in many previous works [18–24]. In addition, the d-spacing value of BMD-origin (4.22 Å) is smaller than that of FMD-origin (5.55 Å), resulting in the more polymer chain packing of BMD-origin, which is because that the van der Waals volume of the bridging group C=O in BTDA (the dianhydride moiety for BMD synthesis) is lower than that of C(CF3)2 group in 6FDA (the dianhydride moiety for FMD synthesis). Its effect on the membrane performance will be further discussed in the following section.

In addition, the solubility tests at 50 oC of the original and thermally-treated polyimide membranes in NMP are also carried out. Fig. 7 displays the dissolution progress of FMD and BMD membranes thermally treated under different temperature with time. We observe that FMD-origin begins to dissolve in NMP after roughly 8 min, and is completely dissolved within 25 min; the dissolution behavior of FMD-250 is almost the same as that of FMD-origin, probably ascribed to weak physical annealing or the negligible cross-linking degree at a low temperature, if there is any; and FMD-300 starts to dissolve after 15 min and completely dissolve after about 30 18

min, which is possibly because of the low thermal cross-linking; as to FMD-350 membrane, only partially dissolution phenomenon is observed after immersion for 2 hours and still not fully dissolved even after 2 months, which demonstrates that the cross-linking degree is much higher than that of FMD-300 membrane; and because of the high cross-linking degree, FMD-400 is almost totally insoluble in NMP even after 2 months. The solubility phenomena of thermally-treated FMD membranes are in good agreement with the above FTIR, TGA and WXRD results. On the other side, since the original BMD membrane is not soluble in NMP, no obvious weight changes of all thermally-treated BMD membranes are observed after immersion in NMP for 2 months either. The different solubilities of BMD and FMD membranes are believed to be caused by their different chemical structures. As we can see that, the bridging group C=O in BTDA is much less flexible than that of C(CF3)2 group in 6FDA, which should be the reason leading to the poor solubility of the synthesized BMD polyimide.

However, sorption results of both FMD and BMD membranes in pure ethanol is higher than that in water, which indicates that they actually both exhibit sorption selectivity towards ethanol, as shown in Table 1. Similar sorption result of another polyimide membrane (6FDA-durene-DABA) was also reported in the previous work, where isopropanol sorption is higher than that of water [34]. In addition, sorption of BMD in pure water is higher than that of FMD, while sorption of BMD in pure ethanol is lower than that of FMD, probably because of higher hydrophilicity of the bridging group C=O in BTDA than that of C(CF3)2 group in 6FDA. With the increase in the thermal treatment temperature, water sorption of both FMD and BMD membranes show down-and-up trends because of the reduced hydrophilicity (decarboxylation effect) and higher fractional free volume (decarboxylation-induced 19

cross-linking).

For separation membranes, the mechanical property is an essential factor to determine their potential for industrial applications. The effects of thermal treatment temperature on the mechanical properties of the original and thermally-treated FMD and BMD polyimide membranes are illuminated in Table 2. For both membranes, the mechanical properties show similar trends. It can be seen that the elongation at break shows insignificant change with a thermal treatment temperature up to 350 oC, but remarkably decreases when the thermal treatment temperature reaches 400 oC, presumably

contributed

to

the

increased

brittleness

from

the

highly

decarboxylation-induced cross-linking and physical annealing at high thermal temperature. Similar phenomenon has also been reported in previous literature [18]. In addition, it is also observed that, the tensile strength firstly increases with the increase in the thermal treatment temperature up to 350 oC and then decreases after that. As aforementioned, with relatively low thermal treatment temperature, i.e., 250 and 300 o

C in this study, the physical annealing results in a densified morphology and

therefore an enhanced tensile strength. When the thermal temperature turns to 350 oC, thermal cross-linking occurs, bringing about a higher brittleness and lower tensile strength of the membrane. On the whole, the results show that the mechanical properties of the membranes can be significantly enhanced when they are thermally treated at relatively high temperature, i.e., 350 oC in this work. In addition, it is also observed that BMD membranes exhibit much higher tensile strength than that of FMD membranes. It is probably because the more efficient polymer chains packing of BMD than that of FMD as discussed above [35].

20

3.3. Pervaporation performance of original and thermally-treated polyimide membranes

Table 3 shows the pervaporation performance of original and thermally-treated polyimide membranes for ethanol dehydration. For original membranes, BMD-origin shows a lower flux and a higher separation factor than those of FMD-origin. The lower flux of BMD-origin may be mainly due to the larger d-spacing compared with FMD-origin as discussed above. The higher separation factor of BMD-origin may be ascribed to the combined effect of the more efficient polymer chain packing (smaller d-spacing value), and the stronger anti-swelling ability (confirmed by the solubility test).

For the effect of the thermal treatment on the membrane performance, the separation performance of common polymeric membranes is generally influenced by the physical annealing effect. However, for carboxyl-containing polyimide in this study, the effect of decarboxylation-induced cross-linking also plays a significant role on the pervaporation performance.

For FMD membranes, it can be seen that the flux first decreases with the increase in the thermal treatment temperature to 300 oC and then increases with the temperature further increasing to 400 oC. From Table 3, it can be seen that the flux change of BMD related membranes exhibits a similar trend. As discussed in Section 3.2, at lower thermal treatment temperature (up to 300 oC), since there is no obvious cross-linking reaction, the flux is only influenced by physical annealing, where thermal treatment promotes the polymer chain packing and reduces the d-spacing 21

(free volume) of the polymer matrix, and leads to a lower flux with an increase in the thermal temperature. But for membranes with higher thermally-treated temperature at 350 and 400 cross-linking

o

C, both effects of physical annealing and thermally-induced

exist.

Opposite

to

the

effect

of

physical

annealing,

the

decarboxylation-induced cross-linking brings out a higher d-spacing among polymer chains as illustrated by the XRD results (in Fig. 6), resulting in a higher permeation flux.

On the other hand, the corresponding separation factors of both FMD and BMD polyimide membranes show up-and-down trends with the increase in the thermal temperature, with the highest separation factors at thermal treatment temperatures of 300 and 250 oC respectively. It can be ascribed to the combined effects of the enhanced anti-swelling property, the formation of charge transfer complexes (CTCs) and the changed membrane hydrophilicity. Firstly, both physical annealing and chemical cross-linking effects via thermal treatment can enhance the membrane anti-swelling property in the feed solution during pervaporation and therefore its separation performance. Secondly, the CTC formation in the thermally-treated polyimide membrane can promote polymer chain packing effectively and thus improve membrane selectivity. It can be proved by the color change in membranes as shown in Fig. 8 and red shift in UV absorption bands [2,14,18] as listed in Table 4 with an increase in the thermal treatment temperature. It is noted that red shift in UV absorption bands is negligible initially and become prominent with the temperature increase, indicating the effect of CTCs is more significant at higher thermal treatment temperatures. In another aspect, the decrease in the membrane hydrophilicity of the thermally treated membrane may explain the decline in the separation factor with the 22

further increase in the thermal treatment temperature. As shown in Fig. 9, the water contact angle of both hydrophilic polyimide membranes ( < 90 oC) increases with the increase in the thermal treatment temperature because of the smoother surface. According to Wenzel Equation [36] (Eq. 6), the smoothness increase of the hydrophilic surface will lead to an increase in the water contact angle.

cosW   cosY

(6)

where W is the measured water contact angle on a rough surface, Y is the contact angle on a smooth surface of the same material, and  refers to the roughness ratio of the surface (larger than 1). At the same time, the decarboxylation-induced cross-linking causes the diminishment of the hydrophilic COOH groups on the membrane surface, also resulting in a decrease in the hydrophilicity.

3.4. Effect of operation temperature on pervaporation performance

In this section, FMD-400 and BMD-400 are studied to investigate the effect of the operation temperature on the pervaporation performance. Permeability and selectivity are also employed to illuminate the intrinsic separation performance by decoupling the change of external driving forces [37,38] with the temperature change. The results are shown in Figs. 10 and 11, respectively. For both membranes, the total flux and individual fluxes of water and ethanol all increase while the separation factor decreases with an increase in the operation temperature; on the other hand, the total permeability and individual permeabilities of water and ethanol and the selectivity follow the decreasing trends.

As well known, the increase in the operation temperature has complicated effects on 23

the pervaporation performance. Firstly, because the saturated vapor pressure increases in the feed side with the temperature increase hence, the driving forces for the feed components to diffuse through the membrane are enhanced since the permeate pressure remains near zero under vacuum in this study. Secondly, a high operation temperature will weaken interactions among feed components and the membrane material, including the solubilities of feed components in the membrane and the swelling effect on the membrane by feed components. Thirdly, the increase in the operation temperature may facilitate the thermal motion of polymer chains and increase the fractional free volume in the polymeric membrane. Based on the analysis above, the permeability decrease with the increase in the operation temperature should be primarily due to the predominant solubility reduction of both water and ethanol molecules and the mitigated swelling effect. It is proved by the sorption test (swelling test) of FMD-400 and BMD-400 membranes in water and ethanol liquids as shown in Fig. 12. It can be observed that the swelling degree of both polyimide membranes in both water and ethanol decreases with an increase in the liquid temperature. Therefore, the flux increase should be mainly ascribed to the enhanced driving forces of the feed components, according to the relationship between the flux and permeability as described in Eq. (3).

It is also noted that separation factor and selectivity both decrease with an increase in operation temperature. As shown in Figs. 10 and 11, the increasing magnitude of ethanol flux is greater than that of water flux with the increase in the operation temperature, indicating the transport rate of ethanol molecules through the membrane is more sensitive to the operation temperature compared with that of water molecules. Therefore, the separation factor of the membrane towards water decreases with an 24

increase in the operation temperature. On the other hand, the decreasing magnitude of water permeability is larger than that of ethanol permeability with an increase in the operation temperature, which should be ascribed to the sharper solubility (sorption ability) drop of water molecules in the membrane than that of ethanol molecules, as shown in Fig. 12. Hence, the selectivity of the membrane towards water decreases with increasing the operation temperature.

The relationships between flux or permeability and operation temperature can generally be expressed using the Arrhenius Equations [39]:

 E  J = J 0 exp   J   RT 

(7)

 E  P = P0 exp   P   RT 

(8)

where Jo and P0 are the pre-exponential factors of flux and permeability, EJ and EP are the apparent activation energies of flux and permeability, R is the universal gas constant, and T (K) is the operation temperature. EJ and Ep can be calculated according to Eqs. (7) and (8) with the least square method.

As illuminated in Figs. 13 and 14, the good linearity between logarithmic flux or permeability and reciprocal temperature indicates that the experimental data matches the Arrhenius equation well. In general, a higher value of the apparent activation energy for the transport of molecules through the membrane implies a more sensitive behavior with the temperature change. From the data listed in Table 5, it can be seen that the apparent activation energies of ethanol flux and permeability are all higher than those of water (i.e., EJ,Eth > EJ,W, EP,Eth > EP,W), which is consistent with the above speculation for the performance change with the temperature, i.e., the transport rate of 25

ethanol molecules through the membrane is more sensitive to the operation temperature compared with that of water molecules. Table 5 also shows that the activation energies of permeability for water and ethanol are both negative, which reconfirms the above hypothesis that the solubility decline of water molecules in the membrane is severer than that of ethanol molecules.

3.5. Benchmarking

To evaluate the pervaporation performance of the two novel polyimide membranes developed in this work, Table 6 gives a performance benchmarking of a series of polyimide dense membranes for ethanol dehydration under similar operation conditions. Compared to most other polyimide membranes with either low flux (or permeability) or low separation factor (or selectivity), the two polyimide membranes developed in this study show superior separation performance, which may be contributed by the large fractional free volume and high anti-swelling property derived from the thermally cross-linked structure. The BMD membrane thermally treated at 400 C can achieve an excellent separation factor of 636 with a permeation flux of 64.5 g/m2 h for dehydrating the feed solution of 85/15 wt% ethanol/water at 40 o

C.

Further performance enhancement of the two polyimide membranes can be achieved by balancing the separation factor and the permeation flux through suitable morphology optimization and/or material modification. Possible modification routes include: (1) apply further hydrophilization modification of two polyimide membranes; (2) develop BMD-inorganic hybrid membranes by incorporating nanoparticles with 26

high fractional free volume into BMD membrane; (3) or employ the composite membrane morphology with the BMD polyimide as the selective layer, and other polymeric material with higher permeation rate as the supporting layer.

4. Conclusion

In this work, two novel carboxyl-containing polyimides, 6FDA-MDA-DABA and BTDA-MDA-DABA, are synthesized and employed as pervaporation membranes for ethanol dehydration. The following conclusions can be made from this study:

(1) Organic-soluble FMD and organic-insoluble BMD polyimides are successfully synthesized via chemical and thermal imidization methods, respectively. Various characterization techniques (FTIR, TGA, etc.) are employed to confirm the successful synthesis.

(2) Thermal treatment of the polyimide membranes at various temperatures demonstrates different effects on the membrane morphology and separation performance of both FMD and BMD membranes. The thermal treatment at a relative low temperature only leads to the physical annealing and results in a densified membrane morphology, which slightly reduces the flux and enhances the separation factor. However, it may induce the thermally cross-linking besides the physical annealing at a high thermal treatment temperature and brings about a higher factional free volume and d-spacing in the membrane morphology, resulting in a higher flux and up-and-down separation factor by the combined effects of thermal cross-linking and physical annealing. 27

(3) The effect of operation temperature on the separation performance of both FMD-400 and BMD-400 membranes show the similar tread, i.e., the flux increases while the separation factor, the permeability and the selectivity all decrease with the increase in the operation temperature, because of the combined effects of the enhanced driving force, mitigated interactions between feed components and membrane, as well as the larger fractional free volume.

(4) Compared with other polyimide membranes, the novel FMD and BMD polyimide membranes possess relative superior separation performance because of the desirable molecular structure and effective decarboxylation-induced cross-linking by thermal treatment.

Acknowledgement

The authors thank the financial support from Huazhong University of Science and Technology China (Grant nos. 0124013041, 2014YQ012), and National Natural Science Foundation of China (Grant No. 21306058) and “Thousand Youth Talent Plan”. Special thanks are given to Dr. Ngoc Lieu Le in King Abdullah University of Science and Technology (KAUST), Prof. Sixue Cheng in Wuhan University (China) and Prof. Lu Shao in Harbin Institute of Technology (China) for their valuable suggestions and help. We would also like to thank the Analysis and Testing Center, Huazhong University of Science and Technology for their assistance in material characterizations.

28

List of abbreviations and symbols

Abbreviation

Full name

6FDA

2,2'-bis (3,4-dicarboxyphenyl) hexafluoropropane dianhydride

6FDA-DAM/DABA

2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride2,4,6-trimethyl-1,3-diaminobenzene/3,5-diaminobenzoic acid

6FDA-MDA/DABA,FMD

2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride4,4'-diaminodiphenylmethane/3,5-diaminobenzoic acid

6FDA-NDA/DABA

2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride1,5-naphthalene diamine/3,5-diaminobenzoic acid

6FDA-ODA/DABA

2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride4,4'-oxydianiline/3,5-diaminobenzoic acid

Ac2O

acetic anhydride

BTDA

3,3',4,4'-benzophenone tetracarboxylic dianhydride

BTDA-MDA/DABA,BMD 3,3',4,4'-benzophenone tetracarboxylic dianhydride4,4'-diaminodiphenylmethane/3,5-diaminobenzoic acid DABA

3,5-diaminobenzoic acid

DMF

N,N-dimethylformamide

FTIR

Fourier transform infrared spectroscopy

GPC

gel permeation chromatography

MDA

4,4'-diaminodiphenylmethane

NMP

1-methyl-2-pyrrolidinone

PAA

polyamic acid

TEA

triethylamine

THF

tetrahydrofuran 29

TGA

thermogravimetric analysis

UV

ultraviolet

WAXD

wide-angle X-ray diffraction

Symbol

Full name

Unit

A

effective area of the membrane

m2

d

average intersegmental distance of polymer chains

Å

EJ,Eth

apparent activation energies of ethanol flux

kJ/mol

EJ,W

apparent activation energies of water flux

kJ/mol

EP,Eth

apparent activation energies of ethanol permeability

kJ/mol

EP,W

apparent activation energies of water permeability

kJ/mol

J

flux

g/m2·h

JEth

ethanol flux

g/m2·h

J0

pre-exponential factor for the component flux

g/m2·h

JW

water flux

g/m2·h

l

membrane thickness

m

n

an integer in the Bragg’s equation

P

permeability of the component i

g·m/m2·h·KPa

PEth

ethanol permeability

g·m/m2·h·KPa

P0

pre-exponential factor for the component permeability

g·m/m2·h·KPa

pp

permeate pressure

KPa

pisat

saturated vapor pressure of the component i

KPa

PW

water permeability

g·m/m2·h·KPa

r

roughness ratio

30

R

universal gas constant

J/K·mol

R²

R-squared value of the trend line

t

operation time

T

operating temperature

Td

thermal decomposition temperature

o

xn,i

mole fraction of the component i in the feed

mol%

xw,i

weight fraction of component i in the feed

wt%

yn,i

mole fraction of the component i in the permeate

mol%

yw,i

weight fraction of component i in the permeate

wt%



separation factor



selectivity

γi

activity coefficient of the component i

θ

X-ray diffraction angle of the peak

o

θW

observed contact angle on a rough surface

o

θY

contact angle on a smooth surface

o

λ

Cu K radiation wavelength of wide-angle X-ray diffraction

Å

h o

C or K

C

Subscripts 1

component 1 with higher or preferred permeability

2

component 2 with lower or less preferred permeability

1/2

(separation efficiency), to separate component 1 over component 2

i

component i in the feed

n

mole-based

w

weight-based

31

Superscripts p

permeate side

sat

saturated vapor pressure

32

Reference [1] L.M. Vane, A review of pervaporation for product recovery from biomass fermentation processes, J. Chem. Technol. Biotechnol. 80 (2005) 603–629. [2] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Review: Polyimides membranes for pervaporation and biofuels separation, Prog. Polym. Sci. 34 (2009) 1135–1160. [3] X. Feng, R.Y.M. Huang, Liquid separation by membrane pervaporation: a review, Ind. Eng. Chem. Res. 36 (1997) 1048–1066. [4] V.V. Namboodiri, L.M. Vane, High permeability membranes for the dehydration of low water content ethanol by pervaporation, J. Membr. Sci. 306 (2007) 209–215. [5] S. Xiao, X. Feng, R.Y.M. Huang, Trimesoyl chloride crosslinked chitosan membranes for CO2/N2 separation and pervaporation dehydration of isopropanol, J. Membr. Sci. 306 (2007) 36–46. [6] Y.-L. Liu, C.-H. Yu, K.-R. Lee, J.-Y. Lai, Chitosan/poly(tetrafluoroethylene) composite membranes using in pervaporation dehydration processes, Journal of Membrane Science 287 (2007) 230–236. [7] R.Y.M. Huang, R. Pal, G.Y. Moon, Characteristics of sodium alginate membranes

for

the

pervaporation

dehydration

of

ethanol–water

and

isopropanol–water mixtures, J. Membr. Sci. 160 (1999) 101–113. [8] S. Xiao, X. Feng, R.Y.M. Huang, Synthesis and properties of 6FDA-MDA copolyimide membranes: effects of diamines and dianhydrides on gas separation and pervaporation properties, Macromol. Chem. Phys. 208 (2007) 2665–2676. [9] W.F. Guo, T.-S. Chung, Study and characterization of the hysteresis behavior of polyimide membranes in the thermal cycle process of pervaporation separation, 33

J. Membr. Sci. 253 (2005) 13–22. [10] X. Qiao, T.-S. Chung, Fundamental characteristics of sorption, swelling, and permeation of P84 co-polyimide membranes for pervaporation dehydration of alcohols, Ind. Eng. Chem. Res., 44 (2005) 8938–8943. [11] Y. Wang, L. Jiang, T. Matsuura, T.S. Chung, S.H. Goh, Investigation of the fundamental differences between polyamide-imide (PAI) and polyetherimide (PEI) membranes for isopropanol dehydration via pervaporation, J. Membr. Sci. 318 (2008) 217–226. [12] D.-J. Liaw, K.-L. Wang, Y.-C. Huang, K.-R. Lee, J.-Y Lai, C.-S. Ha, Advanced Polyimide Materials: Syntheses, Physical Properties and Applications, Prog. Polym. Sci., 37 (2012) 907–974. [13] K. Vanherck, G. Koeckelberghs, I.F.J. Vankelecom, Crosslinking polyimides for membrane applications: A review, Prog. Polym. Sci. 38 (2013) 874–896. [14] L. Shao, T.-S. Chung, S.H. Goh, K.P. 1,3-cyclohexanebis(methylamine)

modification

Pramoda, The effects of on

gas

transport

and

plasticization resistance of polyimide membranes, J. Membr. Sci. 267 (2005) 78–89. [15] L. Shao, L. Liu, S.-X. Cheng, Y.-D. Huang, J. Ma, Comparison of diamino cross-linking in different polyimide solutions and membranes by precipitation observation and gas transport, J. Membr. Sci. 312 (2008) 174–185. [16] C. Cao, T.-S. Chung, Y. Liu, R, Wang, K.P. Pramoda, Chemical cross-linking modification of 6FDA-2,6-DAT hollow fiber membranes for natural gas separation. J. Membr. Sci. 216 (2003) 257–68. [17] X.Y. Qiao, T.-S. Chung, Diamine modification of P84 polyimide membranes for pervaporation dehydration of isopropanol, AIChE J. 52 (2006) 3462–3472. 34

[18] N.L. Le, Y. Wang, T.-S. Chung, Synthesis, Cross-linking modifications of 6FDA-NDA/DABA polyimide membranes for ethanol dehydration via pervaporation, J. Membr. Sci. 415-416 (2012) 109–121. [19] J.D. Wind, C. Staudt-Bickel, D.R. Paul, W.J. Koros, The effects of cross-linking chemistry on CO2 plasticization of polyimide gas separation membranes, Ind. Eng. Chem. Res. 41 (2002) 6139–6148. [20] J.D. Wind, C. Staudt-Bickel, D.R. Paul, W.J. Koros, Solid-state covalent cross-linking of polyimide membranes for carbon dioxide plasticization reduction, Macromolecules 36 (2003) 1882–1888. [21] A.M. Kratochvil, W.J. Koros, Decarboxylation-induced cross-linking of a polyimide for enhanced CO2 plasticization resistance, Macromolecules 41 (2008) 7920–7927. [22] W. Qiu, C.-C. Chen, L. Xu, L. Cui, D.R. Paul, W.J. Koros, Sub-Tg cross-linking of a polyimide membrane for enhanced CO2 plasticization resistance for natural gas separation, Macromolecules 44 (2011) 6046–6056. [23] C.-C. Chen, W. Qiu, S.J. Miller, W.J. Koros, Plasticization-resistant hollow fiber membranes for CO2/CH4 separation based on a thermally crosslinkable polyimide, J. Membr. Sci. 382 (2011) 212–221. [24] E.M. Maya, A. Tena, J. de Abajo, J.G. de la Campa, A.E. Lozano, Partially pyrolyzed membranes (PPMs) derived from copolyimides having carboxylic acid groups. Preparation and gas transport properties, J. Membr. Sci. 349 (2010) 385–392. [25] C. Cao, R. Wang, T.-S. Chung, Y. Liu, Formation of high-performance 6FDA-2,6-DAT asymmetric composite hollow fiber membranes for CO2/CH4 separation, J. Membr. Sci. 209 (2002) 309–319. 35

[26] Y. Xu, C. Chen, J. Li, Experimental study on physical properties and pervaporation performances of various polyimide membranes, Chem. Eng. Sci. 62 (2007) 2466–2473. [27] Y. Wang, M. Gruender, T.S. Chung, Pervaporation dehydration of ethylene glycol through polybenzimidazole (PBI)-based membranes. 1. Membrane fabrication, J. Membr. Sci. 363 (2010) 149–159. [28] Y. Wang, T.S. Chung, B. Neo, M. Gruender, Processing and engineering of pervaporation dehydration of ethylene glycol via dual-layer polybenzimidazole (PBI)/polyetherimide (PEI) membranes, J. Membr. Sci. 378 (2011) 339–350. [29] L. Shao, T.-S. Chung, G. Wensley, S.H. Goh, K.P. Pramoda, Casting solvent effects

on

morphologies,

gas

transport

properties

of

a

novel

6FDA/PMDA-TMMDA copolyimide membrane and its derived carbon membranes, J. Membr. Sci. 244 (2004) 77–87. [30] C.-L. Li, K.-R. Lee, Dehydration of ethanol/water mixtures by pervaporation using soluble polyimide membranes, Polym. Int. 55 (2006) 505–512. [31] Q.Y. Zhang, S.H. Li, W.M. Li, S.B. Zhang, Synthesis and properties of novel organosoluble polyimides derived from 1,4-bis 4-(3,4-dicarboxylphenoxy) triptycene dianhydride and various aromatic diamines, Polymer 48 (2007) 6246–6253. [32] Y.K. Ong, H. Wang, T.-S. Chung, A prospective study on the application of thermally rearranged acetate-containing polyimide membranes in dehydration of biofuels via pervaporation, Chem. Eng. Sci. 79 (2012) 41–53. [33] N.L. Le, T.-S. Chung, High-performance sulfonated polyimide/polyimide/ polyhedral oligosilsesquioxane hybrid membranes for ethanol dehydration applications, J. Membr. Sci. 454 (2014) 62–73. 36

[34] P. Salehian, M.L. Chua, M. Askari, G.M. Shi, T.-S. Chung, In situ regulation of micro-pore to design high performance polyimide membranes for pervaporation dehydration of isopropanol, J. Membr. Sci. 493 (2015) 299–310. [35] N.W. Li, Z.M. Cui, S.B. Zhang, W. Xing, Sulfonated polyimides bearing benzimidazole groups for proton exchange membranes, Polymer 48 (2007) 7255–7263. [36] B.J. Ryan, K.M. Poduska, Roughness effects on contact angle measurements, Am. J. Phys. 76 (2008) 1074–1077. [37] Y. Wang, T.S. Chung, M. Gruender, Sulfonated polybenzimidazole membranes for pervaporation dehydration of acetic acid, J. Membr. Sci. 415–416 (2012) 486–495. [38] W.F. Guo, T.S. Chung, T. Matsuura, Pervaporation study on the dehydration of aqueous butanol solutions: a comparison of flux vs. permeance, separation factor vs. selectivity, J. Membr. Sci. 245 (2004) 199–210. [39] X.S. Feng, R.Y.M. Huang, Estimation of activation energy for permeation in pervaporation processes, J. Membr. Sci. 118 (1996) 127–131. [40] L. H. Wang, J. D. Li, Z. P. Zhao, C. X. Chen, Synthesis and characterization of soluble polyimides derived from 4,4'-diamino-3,3'-dimethyldiphenylmethane and their pervaporation performances. J. Macromol. Sci. A: Pure Appl. Chem. 43 (2006) 305–314.

37

List of Tables Table 1. Sorption results of original and thermally-treated FMD and BMD membranes in pure water and pure ethanol at 50 oC, respectively. Table 2. Mechanical properties of original and thermally-treated FMD and BMD membranes. Table 3. Effect of the thermal treatment temperature on the pervaporation performance of original and thermally-treated FMD and BMD membranes for ethanol dehydration. Table 4. UV absorption result of original and thermally-treated FMD and BMD membranes. Table 5. Apparent activation energies of flux and permeability for FMD-400 and BMD-400 membranes. Table 6. A comparison of pervaporation performance of polyimide dense membranes for ethanol dehydration.

38

Fig. 1. Synthesis route of FMD and BMD polyimides by chemical and thermal imidization.

39

(a) 1785 1725 1354 1087 714

Transmission

FMD-PAA

FMD

1650 1550

4000

3500

3000

2500

2000

1500

1000

500

Wavenumber (cm-1)

(b) 1780 1721 1354 1094 720

Transmission

BMD-PAA

BMD

1650 1550

4000

3500

3000

2500

2000

1500

1000

500

-1

Wavenumber (cm )

Fig. 2. FTIR spectra of (a) FMD-PAA and FMD polyimide and (b) BMD-PAA and BMD polyimide.

40

(a) 100

Weight (%)

90

FMD-origin FMD-250 FMD-300 FMD-350 FMD-400

80

70

60 0

100

200

300

400

500

600

700

800

600

700

800

Temperature (oC)

(b) 100

Weight (%)

90

BMD-origin BMD-250 BMD-300 BMD-350 BMD-400

80

70

60 0

100

200

300

400

500

Temperature (oC)

Fig. 3. TGA curves of (a) FMD and (b) BMD membranes thermally treated at different temperatures.

41

Fig. 4. Cross-linking mechanism of the carboxylic-containing polyimides under thermal treatment.

42

(a)

FMD-origin

Transmission

FMD-250 FMD-300 FMD-350

FMD-400

3800

3600

3400

3200

3000

2800

3000

2800

Wavenumber (cm-1)

(b)

Transmission

BMD-origin

BMD-250

BMD-300

BMD-350 BMD-400

3800

3600

3400

3200 -1

Wavenumber (cm )

Fig. 5. FTIR spectra of (a) FMD and (b) BMD membranes thermally treated at different temperatures.

43

(a) 5.55 Å

FMD-origin

5.54 Å

Intensity

5.50 Å

FMD-250 5.62 Å

FMD-300

5.78 Å

FMD-350 FMD-400

10

15

20

25

30

35

40

2θ (º)

(b) 4.22 Å

4.21 Å

BMD-origin

Intensity

BMD-250 4.10 Å

4.18 Å

BMD-300

4.21 Å

BMD-350 BMD-400

10

15

20

25

30

35

40

2θ (º)

Fig. 6. WXRD spectra of (a) FMD and (b) BMD membranes thermally treated at different temperatures.

44

(a)

0 minute

20 minutes

3 days

2 months

0 minute

20 minutes

3 days

2 months

(b)

Fig. 7. Solubility test of (a) FMD and (b) BMD membranes thermally treated at different temperatures. (Samples 1–10 are FMD-origin, FMD-250, FMD-300, FMD-350, FMD-400, BMD-origin, BMD-250, BMD-300, BMD-350, BMD-400, sequentially.)

45

(a)

(b)

FMD-origin

FMD-250

FMD-350

FMD-300

FMD-400

BMD-origin

BMD-250

BMD-350

BMD-300

BMD-400

Fig. 8. Pictures of (a) FMD and (b) BMD membranes thermally treated at different temperatures.

46

(a) 90

Contact angle (o)

85 80 75 70 65 60

origin

250

300

350

400

FMD membrane

(b) 90

Contact angle (o)

85 80 75 70 65 60

origin

250

300

350

400

BMD membrane

Fig. 9. Water contact angles of (a) FMD and (b) BMD membranes thermally treated at different temperatures.

47

(a) 350

Total flux (g/m2h)

250 90 200 60 150

30

100 30

40

50

60

70

8 Water flux Ethanol flux

120 6 90 4 60 2 30

0

80

Ethanol flux (g/m2h)

300 120

150

Water flux (g/m2h)

Total flux Separation factor

Separation factor (Water/Ethanol)

150

0 30

40

50

60

70

80

Temperature (oC)

Temperature (oC)

(b)

200

150

160

100

120

50

80

0

40 30

40

50

60

Temperature (oC)

70

80

6

200

3.0 Water permeability Ethanol permeability

200

2.5

150

2.0

100

1.5

50

1.0

0

0.5 30

40

50

60

70

80

Ethanol permeability (106 g/mhKpa)

250

Water permeability (10 g/mhKpa)

240 Total permeability Selectivity

Selectivity (Water/Ethanol)

Total permeability (106 g/mhKpa)

250

Temperature (oC)

Fig.10. Effect of the operation temperature on the pervaporation performance of FMD-400 membrane: (a) flux and separation factor and (b) permeability and selectivity.

48

(a)

700

100

600

90

500

80

400

70

300

60

200

50

100 30

40

50

60

70

3.0 Water flux Ethanol flux

Water flux (g/m2h)

Total flux (g/m2h)

110

120

2.5

90 2.0 60

1.5 1.0

30 0.5 0

80

Ethanol flux (g/m2h)

800 Total flux Separation factor

Separation factor (Water/Ethanol)

120

0.0 30

40

Temperature (oC)

50

60

70

80

Temperature (oC)

(b)

300

250 200

200

150 100 100 50

0 30

40

50

60

Temperature (oC)

70

80

1.0

350 0.8 300 250

0.6

200 0.4

150 100

0.2

50 0

0.0 30

40

50

60

70

80

Ethanol permeability (106 g/mhKpa)

400 300

Water permeability Ethanol permeability

400

6

350

450

Water permeability (10 g/mhKpa)

500 Total permeability Selectivity

Selectivity (Water/Ethanol)

Total permeability (106 g/mhKpa)

400

Temperature (oC)

Fig. 11. Effect of the operation temperature on the pervaporation performance of BMD-400 membrane: (a) flux and separation factor and (b) permeability and selectivity.

49

(a) 120

Water swelling degree (‰)

Water swelling degree Ethanol swelling degree

15 90 10 60 5

0

Ethanol swelling degree (‰)

20

30 30

40

50

60

70

80

Operation temperature (oC)

(b) 100

Water swelling degree (‰)

Water swelling degree Ethanol swelling degree

80 30 60 20 40 10 20

0

Ethanol swelling degree (‰)

40

0 30

40

50

60

70

80

Operation temperature (oC)

Fig. 12. Sorption tests of (a) FMD-400 and (b) BMD-400 membranes at different operation temperatures.

50

(a) 6

4 Water Ethanol

3 5 ln(JW) = -3.3396×(1000/T) + 14.5948

2

4 1 3

ln(JEth) = -6.1170×(1000/T) + 19.5027

ln (JEth)

ln (JW)

R2 = 0.9874

0

R2 = 0.9958

2

-1 2.90

2.95

3.00

3.05

3.10

3.15

3.20

1000/T (K-1)

(b) 6.5

2.0 Water Ethanol

6.0 5.5

1.5

ln(PW)=2.7433×(1000/T) -0.3575

5.0 1.0 4.5 ln(PEth) = 1.7353×(1000/T) -4.7905

4.0

R2 = 0.9647

ln (PEth)

ln (PW)

R2 = 0.9955

0.5

3.5 3.0

0.0 2.90

2.95

3.00

3.05

3.10

3.15

3.20

1000/T (1/K) Fig. 13. Arrhenius plots of (a) the individual flux and (b) the individual permeability against the reciprocal temperature with FMD-400 membrane.

51

(a) 6

3 Water Ethanol

2

ln(JW) = -1.7884×(1000/T) + 9.8796

5

R2 = 0.9958

4 0

ln (JEth)

ln (JW)

1

ln(JEth) = -4.1101×(1000/T) + 12.5152

3

-1

R2 = 0.9796

2

-2 2.90

2.95

3.00

3.05

3.10

3.15

3.20

1000/T (K-1)

(b) 6 Water Ethanol

0.3

0.0

ln (PW)

ln(PW)=3.3619×(1000/T) -5.1556

R2 = 0.9996

-0.3

ln (PEth)

5

4 ln(PEth) = 1.5918×(1000/T) -5.3251

-0.6

R2 = 0.9886

-0.9 3 2.90

2.95

3.00

3.05

3.10

3.15

3.20

1000/T (1/K) Fig. 14. Arrhenius plots of (a) the individual flux and (b) the individual permeability against the reciprocal temperature with BMD-400 membrane. ciprocal temperature with BMD-400 membrane.

52

Table 1. Sorption results of original and thermally-treated FMD and BMD membranes in pure water and pure ethanol at 50 oC, respectively.

Membrane

Water sorption

Ethanol sorption

(mg water/g membrane)

(mg ethanol/g membrane)

FMD-origin

7.4±0.5

106.7±6.2

FMD-250

6.9±0.4

96.9±4.2

FMD-300

5.9±0.6

98.2±5.1

FMD-350

6.1±0.2

91.9±3.5

FMD-400

6.6±0.5

87.5±5.4

BMD-origin

13.3±0.2

38.1±2.1

BMD-250

12.1±0.3

56.1±3.5

BMD-300

11.4±0.8

47.5±3.7

BMD-350

13.1±0.4

49.6±4.2

BMD-400

14.1±0.5

49.5±1.6

53

Table 2. Mechanical properties of original and thermally-treated FMD and BMD membranes.

Membrane

Elongation at break

Tensile strength

(%)

(MPa)

FMD-origin

10.2±1.1

89.2±3.4

FMD-250

7.7±0.8

90.5±6.3

FMD-300

10.0±0.4

114.2±7.5

FMD-350

9.8±1.4

119.6±3.3

FMD-400

6.1±0.5

84.2±5.1

BMD-origin

10.2±4.2

148.1±10.2

BMD-250

10.2±1.1

156.5±5.8

BMD-300

11.0±3.8

178.8±11.8

BMD-350

11.5±1.3

180.8±9.3

BMD-400

7.0±0.5

137.2±6.6

54

Table 3. Effect of the thermal treatment temperature on the pervaporation performance of original and thermally-treated FMD and BMD membranes for ethanol dehydration.

Membrane

Water in permeate

Total flux 2

Separation factor

(wt%)

(g/m ·h)

(H2O/EtOH)

FMD-origin

98.36

48.6

340

FMD-250

98.52

47.5

377

FMD-300

98.60

42.3

451

FMD-350

98.53

46.7

380

FMD-400

98.21

53.2

298

BMD-origin

98.98

36.9

550

BMD-250

99.52

30

1175

BMD-300

99.34

25.6

853

BMD-350

99.32

43.4

828

BMD-400

99.09

63.4

636

Operation condition: feed solution – 85/15 wt% ethanol/water; permeate pressure – less than 2 mbar; operation temperature – 40 oC.

55

Table 4. UV absorption results of original and thermally-treated FMD and BMD membranes.

Membrane

λUV (nm)

Δλ (nm)

FMD-origin

350

-

FMD-250

355

5

FMD-300

357

7

FMD-350

366

16

FMD-400

385

35

BMD-origin

403

-

BMD-250

403

0

BMD-300

403

0

BMD-350

417

14

BMD-400

423

20

56

Table 5. Apparent activation energies of flux and permeability for FMD-400 and BMD-400 membranes.

Membrane

EJ,W (kJ/mol)

EJ,Eth (kJ/mol)

EP,W (kJ/mol)

EP,Eth (kJ/mol)

FMD-400

27.76

50.86

-22.81

-14.43

BMD-400

14.87

34.17

-27.95

-13.23

57

Table 6. A comparison of pervaporation performance of polyimide dense membranes for ethanol dehydration.

Feed ethanol Permeabilit Temperatur Flux concentratio Separatio y Selectivit Referenc Membrane e (g/m n n factor (106 g/m h y e o 2 ( C) h) (wt%) Kpa) BPDA-ODA/DABA 90 75 18.5 1800 40.9 2995 [40] 3287.1 48 BHTDA-BATB 90 35 282 27 [30] BTDA–ODA 88.9 65 17.9 432 51.3 214 [26] BTDA–MDA 88.9 45 14.7 237 112.7 122 [26] BTDA–MDA 88.9 65 27.1 302 79.4 150 [26] 6FDA-NDA/DABA 158. (7:3), 85 25 107 1749.0 63 [18] 7 treated at 425 oC, 30min 6FDA-NDA/DABA (9:1), cross-linked with 85 25 75.4 156 840.2 92 [18] EDA and treated at 100 o C 6FDA-NDA/DABA (9:1), cross-linked with 85 25 50.8 632 576.7 372 [18] XDA and treated at 200 o C 6FDA-NDA/DABA (9:1), cross-linked with 183. 85 25 33 1880.1 19 [18] BDM and treated at 300 4 o C 6FDA-MDA/DABA 220. 85 70 71 206.6 38 (7:3) 7 BTDA-MDA/DABA 85 70 41.1 121 39.3 65 (7:3) 6FDA-MDA/DABA (7:3), treated at 400 oC, 85 40 55.3 298 193.7 170 30min 6FDA-MDA/DABA This 135. (7:3), treated at 400 oC, 85 70 134 113.4 73 study 4 30min BTDA-MDA/DABA (7:3), treated at 400 oC, 85 40 64.5 636 265.7 364 30min BTDA-MDA/DABA 106. (7:3), treated at 400 oC, 85 70 337 104.2 182 8 30min

Highlights 

Two novel carboxyl-containing polyimides are synthesized and thermally cross-linked.



FMD and BMD membranes are applied for pervaporation dehydration of ethanol. 58



Effects of thermal treatment temperature on physicochemical properties and pervaporation performance.



Effects of operation temperature on membrane performance.

59