Preparation, morphologies and properties for flat sheet PPESK ultrafiltration membranes

Preparation, morphologies and properties for flat sheet PPESK ultrafiltration membranes

Journal of Membrane Science 270 (2006) 146–153 Preparation, morphologies and properties for flat sheet PPESK ultrafiltration membranes Yanbin Yun a,∗...

589KB Sizes 0 Downloads 19 Views

Journal of Membrane Science 270 (2006) 146–153

Preparation, morphologies and properties for flat sheet PPESK ultrafiltration membranes Yanbin Yun a,∗ , Yunhua Tian a , Guoling Shi b , Jiding Li b , Cuixian Chen b a

College of Material Science and Technology, Beijing Forestry University, 100083, China b Department of Chemical Engineering, Tsinghua University, Beijing, 100084, China

Received 31 December 2004; received in revised form 20 June 2005; accepted 27 June 2005 Available online 10 August 2005

Abstract Flat sheet poly(phthalazinone ether sulfone ketone) (PPESK) ultrafiltration membranes were prepared. The effects of PPESK concentration, solvents, additives and exposing time on the structure and performance of PPESK ultrafiltration membranes were investigated in more detail. The optimal preparation conditions were: 12 wt.% PPESK, NMP/DMAc mixed solvent (mass ratio = 1), 8 wt.% polyethylene glycol 400 (PEG400), 2.5 wt.% LiCl and 5 s exposing time. Under these conditions, the pure water flux and the rejection of ␥-globulins were 1139L/m2 h and 93.7% at the operation pressure of 0.1 MPa, respectively. Scanning electron microscope (SEM) micrographs showed that spongy structure could be formed while finger-like structure could be suppressed due to the longer exposing time or higher LiCl concentration. © 2005 Published by Elsevier B.V. Keywords: Poly(phthalazinone ether sulfone ketone); Ultrafiltration; Flux; Rejection; Morphology

1. Introduction Ultrafiltration (UF) is a pressure-driven process to separate macromolecules or colloids from the smaller species and an active research area in membrane separation. The requirements of the membrane materials varied with the feed property, operation condition, pretreatment, sterilization and cleaning method. For example, for biochemistry separation, the membrane is required to bear 120 ◦ C for sterilization; while for water treatment or chemical separation, the membrane is required to tolerate oxidant, acid, alkali, free chlorine etc. [1–5]. Ultrafiltration membranes can be prepared by organic polymer or inorganic materials. Currently, in the organic membrane market, ultrafiltration membrane materials include cellulose acetate (CA), polyacrylonitrile (PAN), polyvinylidene fluoride (PVDF), polysulfone (PS) and ∗ Corresponding author. Tel.: +86 13671310642/10 62338152; fax: +86 10 62770304. E-mail address: [email protected] (Y. Yun).

0376-7388/$ – see front matter © 2005 Published by Elsevier B.V. doi:10.1016/j.memsci.2005.06.050

polyethersulfone (PES) and so on. Comparing with the inorganic membranes, the cost of organic membranes is lower, whereas its tolerance capacities towards high temperature, solvent and chemical corrosive are not good enough generally. Hence, it is necessary to develop a novel organic ultrafiltration membrane, which possesses comprehensive properties and outstanding thermostability. PPESK is a rigid aromatic compound and has a small linear coefficient of expansion, which shows superior mechanical strength, chemical resistance and very high glass transition temperature (Tg ) in the range of 263–305 ◦ C [6–9]. These properties are very important to the performance and the life of membranes. Jian et al. synthesized PPESK and successfully prepared ultrafiltration membranes using it. The casting solution was composed of NMP, 10 wt.% PPESK and 15 wt.% butanone. The flux of 601 L/m2 h and the rejection of 19% for PEG (Mw 12,000) under operation pressure of 0.1 MPa were obtained when the evaporation time was 5 s and gelation bath temperature was 6 ◦ C [2]. The objective of this study was further to improve the properties of the asymmetric PPESK UF membranes, so the main

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

147

factors affecting the membrane structure and performance such as polymer concentration, solvents, organic additive, inorganic additive and exposing time were investigated in more detail, and optimized. 2. Experimental 2.1. Materials and methods PPESK used as the flat sheet ultrafiltration membrane material was supported by Dalian Polymer New Material Co., Ltd., it contained a sulfone/ketone ratio of 1/1. Chemical structure of PPESK was shown in Fig. 1. N-methyl-2pyrrolidone (NMP) and Dimethylacetamide (DMAc) were used as solvents. Tween80, PEG400 and LiCl were selected as additives. The rejection of UF membranes was characterized by bovine serum albumin (BSA, Mw 76,000) and ␥-globulins (Mw 150,000). All chemicals used in the experiments were analytical grade without further purification. The flat sheet membrane casting apparatus and the permeation test instrument were designed in our laboratory. After pretreated with pure water under 0.3 MPa pressure for 30 min, the membranes were characterized with the permeation test apparatus under operation pressure of 0.1 MPa and feed temperature of 25 ◦ C. The schematic diagram of permeation test was shown in Fig. 2. Two liters of 0.05 wt.% protein solutions were used to measure the rejection for each flat sheet membrane. To investigate the rejection efficiencies for proteins with different molecular weight, two kinds of proteins, namely, ␥-globulins and BSA were used. The proteins concentrations were determined by using a UVspectrophotometer (Agilent 8453, USA) at the wavelength of 280 nm. To avoid destroying the structure of the crosssections of flat membrane, the membrane samples for scanning electron microscopy (SEM) were firstly immersed in liquid nitrogen, then fractured, and finally sputtered with metallic gold to obtain an adequate contrast of the membrane fracture. A JSM-6301 field emission scanning electron microscope (JEOL Ltd.) was used to observe the membrane cross-section morphology. 2.2. Membrane preparation Membrane preparation steps were as follows: PPESK and additive were dissolved in solvent at about 40 ◦ C for

Fig. 2. The schematic diagram of permeation test: 1 pump; 2 bypass valve; 3 valve; 4 pressure gauge; 5 pressure adjustment valve; 6 membrane module; 7 permeate vessel; 8 electronic balance; 9 computer; 10 feed vessel; 11 draining valve.

72 h with vigorous stirring until the homogenous polymer solution was formed; after being filtered and degassed, the casting solution was cast on the non-woven fabrics using the flat sheet membrane casting apparatus and was precipitated by immersing it into gelation bath water until the membrane formed; in order to replace the solvent completely by nonsolvent water, the membrane was moved into another water bath at ambient temperature and further kept for 24 h. The specific preparation conditions were: (1) ambient humidity: 30%; (2) environmental temperature: 18 ◦ C; (3) casting speed: 5 cm/s; (4) thickness of the interval between scraper and non-woven fabrics: 0.135 mm; (5) exposing time: 5 s. 2.3. Membrane characterization The pure water flux (F) of the membrane is calculated as F = Q/At (1), where Q is the total permeate volume in each experiment; A denotes the membrane area; t represents the operation time. Rejection (R) is expressed as R = (1 − Cp /Cf ) × 100% (2), where Cp and Cf are the concentrations of the permeation and the feed, respectively.

Fig. 1. Chemical structure of PPESK.

148

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

3. Results and discussion 3.1. The influence of PPESK concentration on membrane properties When NMP was used as solvent and 8 wt.% PEG400 was selected as additive, the PPESK concentrations in each of the casting solutions were 10.5, 12.0, 13.5, 15.0 and 16.5 wt.%, respectively. Fig. 3 shows the effect of PPESK concentration on the membrane performance. It was noted from Fig. 3 that when the PPESK concentration increased from 10.5 to 16.5 wt.%, the rejection of BSA and ␥-globulins increased from 32.6 to 91.3% and from 80.5 to 97.7%, respectively. In the meantime, the pure water flux decreased from 1153 to 269 L/m2 h of 0.1 MPa. For the membranes prepared by phase inversion method, there were three types of pores: network pore, aggregate pore and phase separation pore. The network pore originated from the space between polymer segments in the network of a polymer aggregate, the aggregate pore from the interstitial space between polymer aggregates, and the phase separation pore from the polymer-poor phase which appears at the liquid–liquid phase separation. For the UF membranes the pore playing an important role is the phase separation pore [10]. The size of the phase separation pore depends the polymer concentration. When the polymer concentration is lower, the size of the pore is larger. With the increase of the polymer concentration, the pore size decreases rapidly, so the pure water flux decreases rapidly and the rejection increases significantly. When the NMP used as solvent, the effect of PPESK concentration on the membrane performance was consistent with this law. But considering the membrane mechanical strength, pure water flux, rejection and the bonding strength between membrane and non-woven fabrics, 12 wt.% PPESK was selected as the optimal polymer concentration. 3.2. The influence of solvent on membrane properties The effects of the DMAc, NMP and the mixture of NMP/DMAc (mass ratio = 1) on membrane performance

Fig. 3. Effects of PPESK concentration on membrane performance.

Fig. 4. Effects of solvent type on pure water flux.

Fig. 5. Effects of solvent type on rejection of BSA.

were investigated and plotted in Figs. 4–6. The concentration of organic additive PEG400 was kept at 8 wt.%. Fig. 4 indicated that the initial flux with DMAc solvent was the highest, but the flux then decreased at the sharpest rate when the PPESK concentration increased from 12 to 15 wt.%, and the flux changed slightly when PPESK concentration further increased from 15 to 16.5 wt.%. It was noted from Figs. 5 and 6 the rejection with DMAc solvent was the lowest. The results of NMP solvent were similar to that of NMP/DMAc mixed solvent when PPESK concentration increased from 10.5 to 16.5 wt.%. The SEM micrographs of three membranes were shown in Fig. 7. It was observed that

Fig. 6. Effects of solvent type on rejection of ␥-globulins.

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

149

Fig. 7. Cross-section SEM micrographs of membranes cast from 12 wt.% PPESK solution in: (a) solvent: DMAc; (b) solvent: NMP; (c) mixed solvent: NMP/DMAc (mass ratio = 1).

the membrane morphology was affected by the solvent. The membranes with DMAc solvent exhibited a spongy structure without clear skin layer on the membrane surface (Fig. 7a). The membranes with NMP solvent and with NMP/DMAc mixed solvent (mass ratio = 1) showed finger-like structure and obvious skin layers on the membrane surfaces (Fig. 7b and c). The interaction force between solvent and polymer resulted in the difference of thermodynamic state and gelation kinetics of the casting solution, and further changed the membrane morphology and performance. According to the solubility parameter principle, if solubility parameters of the polymer were closer to that of the solvent, the polymer was easier to dissolve in this solvent. The solubility parameters of PPESK, DMAc and NMP were 23.4, 22.7 and 22.9 MPa, respectively. So the solubility of NMP was better than that of DMAc for PPESK. This phenomenon was also observed from dissolved state of the casting solution in our experiments. When NMP was used as solvent, PPESK was easier to be dispersed in the solvent. A homogeneous phase system was formed with stable thermodynamics performance. When this casting solution was immerged in the gelation bath, phase separation was delayed significantly (phase separation delay time was long). The solvent on the surface layer was exchanged rapidly with gelation bath water, but the amount of the solvent outflow was larger than that of the gelation bath water inflow. At this time, the substrate solvent could not transfer to membrane surface timely, so that the polymer concentration on the surface layer became

very high, then the skin layer was easy to form. As a result, the pure water flux decreased and the rejection increased. In the case of PPESK/DMAc casting solutions, because their compatibility and thermodynamics stability were poor, so the phase separation could not only be delayed, but also be advanced. Due the poorer compatibility and poorer thermodynamic stability, when the polymer PPESK concentration was lower, the polymer aggregates could disperse in the DMAc solvent well, but the polymer aggregates were easier to form big agglomerates, and when the casting solution was immerged in the gelation bath, the phase separation happened immediately, so a not obvious skin layer with bigger pores was formed, the pure water flux was the highest and the rejection was the lowest; when the PPESK concentration was higher (for example, 15–16.5 wt.%), the system of PPESK/DMAc could become more instable in the thermodynamic, perhaps the phase separation happened earlier during the exposing time before immersed into the gelation bath due to absorbing some moisture from the air. So its law of the pure water flux changing with PPESK concentration was different from that of solvent NMP and mixed solvent NMP/DMAc (mass ratio = 1), and its membranes exhibited a spongy structure without clear skin layer on the membrane surface, and the rejection was the lowest, the mechanical strength and the bonding strength between the membrane and non-woven fabrics with DMAc solvent were the worst. The membrane performance of NMP/DMAc mixed solvent was situated between that of NMP solvent and that of DMAc solvent.

150

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

Fig. 11. Cross-section SEM micrographs of PPESK flat sheet membranes for different organic additive of 8 wt.%: (a) PEG400; (b) Tween80.

Fig. 8. Effects of concentration of PEG400 on membrane performance.

Fig. 9. Effects of concentration of Tween80 on membrane performance.

Although the performances of the membranes with NMP solvent were the best, but the cost of the NMP solvent was much more expensive than DMAc, so the mixture solvent NMP/DMAc (mass ratio = 1) was selected. 3.3. The influence of the additives on membrane properties Additives as forming pore agents play an important role in membrane preparation. They could change the solubility and the dissolution status of polymer. Furthermore, the chemical potential of solvent and the exchange rate between solvent and coagulating agent were also influenced and thereby the membrane morphology could be changed.

The influences of PEG400 and Tween80 as the organic additive were investigated. Four casting solutions containing 12 wt.% PPESK polymer were prepared. The concentrations of organic additive in each of the casting solutions were 8, 10, 12, and 14 wt.%, respectively. In the subsequent experiments, the solvent was mixed solvent NMP/DMAc (mass ratio = 1). Figs. 8 and 9 suggested that pure water flux increased and the rejection decreased when the organic additive concentration increased. The chemical structures of PEG400 and Tween80 are shown in Fig. 10. PEG400 was a watersolubility nonionic surfactant and had strong affinity with water. Hydrogen bond could be formed between PEG400 and DMAc. This would reduce activity of DMAc and increase thermodynamics stability of the casting solution, and improve phase separation trend. The characteristic of Tween80 was similar to that of PEG400, but steric hindrance of Tween80 was larger than that of PEG400. Therefore, the leakage rate of Tween80 from the casting solution was slower, phase separation delay time was longer, and skin layer was easier to form. Thus, pore formation capability of Tween80 was worse than that of PEG400. It was noted from Fig. 11 that PEG400 and Tween80 membranes exhibited a finger-like structure with the skin layer, but the skin layer of the membrane using Tween80 was thicker than that using PEG400. As a result, the pure water flux of PEG400 was higher than that of Tween80 at the same additive concentration. A mixture of LiCl/PEG400 was used to improve permeability of the PPESK ultrafiltration membrane. By using the mixture of LiCl/PEG400 as additive, four casting solutions with 12 wt.% PPESK were prepared with the mixture of NMP/DMAc (mass ratio = 1) as the solvent. For all casting solutions, the concentration of PEG400 was kept at 8 wt.%

Fig. 10. The chemical structures of PEG400 and Tween80: (a) PEG400; (b) Tween80.

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

151

Fig. 12. Effects of inorganic additive concentration on membrane performance.

Fig. 14. Effects of exposing time on membrane performance.

and the concentration of LiCl increased gradually from 0 to 2.5 wt.%. The results are shown in Figs. 12 and 13. Fig. 12 suggested that pure water flux increased sharply as the LiCl concentration increased. For example, the pure water flux increased from 625 to 955 L/m2 h when LiCl concentration increased from 0 to 1.0 wt.%. However, the rejection of BSA decreased gradually and then approached a steady state, and the rejection of ␥-globulins was nearly stable. The membrane structure was changed drastically by adding LiCl (Fig. 13). As the LiCl concentration increased gradually from 0 to 1.5 wt.%, the membrane pore diameter increased gradually from skin layer to bottom. Meanwhile, the skin layer was

very thin. This was caused by high nonsolvent concentration existed in the micronucleus of polymer-poor phase under the skin layer. As well-known, the nonsolvent phase could result in instantaneous phase separation, and cause the formation of sponge-like structure and large surface pore. When LiCl concentration was higher than 1.5 wt.%, the effect of LiCl concentration on the surface pore diameter was insignificant. However, pore diameter of support layer further increased. This is why the pure water flux increased and the rejection was unchangeable as LiCl concentration increased continuously above 1.5 wt.%. Though the membrane pore diameter changed, it was not enough to lead to the decrease of the rejection for ␥-globulins due to the large volume of ␥-globulins.

Fig. 13. Cross-section SEM micrographs of PPESK membranes for different concentration LiCl.

152

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

Fig. 15. Cross-section SEM micrographs of PPESK membranes for different exposing time.

3.4. The influence of exposing time on the membrane properties As outlined in the earlier literatures [1,2], the membranes after casting were allowed to stay in the air for a given time to evaporate solvent or absorb the moisture from the air. This time is called as the evaporation time or exposing time. Because the NMP and DMAc are the solvents with high boiling point and low volatility, so the time was called as the exposing time. It was expected that the exposing time might influence the membrane performances. To verify such an expectation, the membrane prepared with different exposing times were made and characterized. Keeping the other preparation conditions invariable, the membranes were prepared with the exposing times of 5, 60 and 120 s. Fig. 14 showed that the pure water flux decreased gradually as the exposing progressed, and then increased rapidly with the time increase after 60 s. It is well-known that solvents NMP and DMAc possess weak volatility, and rather strong hydroscopicity [11], so before immersed into gelation bath there were two processes on the nascent membrane surface: the solvents evaporation and absorbing moisture. In a short exposing time, the effect of the solvent evaporation was larger than that of absorbing moisture. The evaporation conduced to the increase of polymer concentration on the nascent membrane surface and in turn caused the decrease of the pure water flux. In a long exposing time, the effect of absorbing moisture was larger

than that of the solvent evaporation. A skin layer was formed firstly on the membrane surface. The solvent of inner membrane was broken away through the pores on the skin layer. Some pore wall would burst and pore connectivity increased. Thus, network structure was formed. At this time, surface pore diameter increased, the pure water flux increased rapidly and the rejection decreased. The cross-sectional SEM micrographs in Fig. 15 showed distinctly: (1) in a short exposing time, the membrane structure was regular finger-like structure with skin layer (Fig. 15 5 s); (2) when the exposing was extended, the finger-like structures changed from regular to irregular and had some blemish. There was a trend from finger-like to sponge-like (Fig. 15 60 s); (3) when the exposing was further extended, the membrane structure was sponge-like structure with incomplete skin layer (Fig. 15 120 s).

4. Conclusion In this study, a PPESK ultrafiltration membrane was prepared by using the phase inversion method. The optimal conditions were: 12 wt.% PPESK, mixture of NMP/DMAc (mass ratio = 1) as solvent, 8 wt.% polyethylene glycol 400 (PEG400) as first additive, 2.5 wt.% LiCl as second additive and exposing time of 5 s. The optimized PPESK ultrafiltration membrane had the pure water flux of 1139 L/m2 h and rejection of 93.7% for ␥-globulins under operation pressure of 0.1 MPa.

Y. Yun et al. / Journal of Membrane Science 270 (2006) 146–153

According to the observation of SEM micrographs, it was found that the longer exposing time or higher LiCl concentration could induce the formation of spongy structure and suppress the formation of finger-like structure. The membranes with DMAc solvent exhibited a spongy structure without clear skin layer on the membrane surface. The membranes with NMP solvent and with NMP/DMAc mixed solvent (mass ratio = 1) showed finger-like structure and obvious skin layers on the membrane surfaces. Acknowledgements This work was supported by the National Natural Science Foundation of China (No. 200276034), the Major State Basic Research Development Program of China (973 Program) (No. 2003CB615701) and the National High Technology Research and Development Program of China (863 Program) (No. 2003AA328020). References [1] R.W. Baker, E.L. Cussler, W. Eykamp, W.J. Koros, R.L. Riley, H. Strathmann, Membrane Separation System, Noyes Data Corporation, 1991.

153

[2] X. Jian, Y. Dai, G. He, G. Chen, Preparation of UF and NF poly(phthalazine ether sulfone ketone) membranes for high temperature application, J. Membr. Sci. 161 (1999) 185–191. [3] Y. Dai, X. Jian, S. Zhang, M.D. Guiver, Thermostable ultrafiltration and nanofiltration membranes from sulfonated poly(phthalazinone ether sulfone ketone), J. Membr. Sci. 188 (2001) 195– 203. [4] M.J.H. Snow, D. de Winter, R. Buckingham, J. Campbell, J. Wagner, New techniques for extreme conditions: high temperature reverse osmosis and nanofiltration, Desalination 105 (1996) 57– 61. [5] A. Erswell, C.J. Brouckaert, C.A. Buckley, The reuse of reactive dye liquors using charged ultrafiltration membrane technology, Desalination 70 (1988) 157–167. [6] Y.Z. Meng, A.S. Hay, X.G. Jian, S.C. Tjong, Synthesis of novel poly(phthalazinone ether sulfone ketone)s and improvement of their melt flow properties, J. Appl. Polym. Sci. 66 (1997) 1425– 1432. [7] X.G. Jian, Y. Dai, L. Zeng, R.X. Xu, Application of Poly(phthalazine ether sulfone ketone)s to gas membrane separation, J. Appl. Poly. Sci. 71 (1998) 2385–2390. [8] X.G. Jian, Y.Z. Meng, H.B. Zheng, Preparation of poly(phthalazinone ether sulfone), Chinese Patent 93109180.2 (1993). [9] X.G. Jian, Y.Z. Meng, H.B. Zheng, Preparation of poly(phthalazinone ether ketone), Chinese Patent 93109179.9 (1993). [10] Z.X. Zhu, T. Matsuura, Discussion on the formation mechanism of surface pores in RO, UF and MF membranes prepared by phase inversion process, J. Colloid Interface Sci. 147 (1991) 307–315. [11] T. Matsuura, S. Sourirajan, Advances in Reverse Osmosis and Ultrafiltration, Natl. Res. Council Can. (1989) 83–99.