Preparation and properties of flat-sheet membranes from poly(vinylidene fluoride) for membrane distillation

Preparation and properties of flat-sheet membranes from poly(vinylidene fluoride) for membrane distillation

DESALINATION ELSEVIER Desalination 104 (1996) 1-11 Preparation and properties of flat-sheet membranes from poly(vinylidene fluoride) for membrane di...

1024KB Sizes 0 Downloads 17 Views

DESALINATION ELSEVIER

Desalination 104 (1996) 1-11

Preparation and properties of flat-sheet membranes from poly(vinylidene fluoride) for membrane distillation M. T o m a s z e w s k a Technical University of Szczecin, Institute of Inorganic Chemical Technology, ul. Pulaskiego 10, 70-322 Szczecin, Poland, Tel. 48-91-330352

Abstract

The flat-sheet membranes from poly(vinylidene fluoride) were prepared by the phase inversion process. The effects of the casting solution composition, exposure time prior to coagulation and temperature of the coagulation bath on properties of prepared membranes were investigated. LiC1 was used as a modifying agent. The membrane structure was studied by scanning electron microscopy. For all prepared membranes an asymmetric structure, sometimes without a dense skin layer, was observed. The porosity of prepared membranes before the drying process varied from 72-88%. After drying the membranes become hydrophobic. A contact angle of water droplet on the membrane surface was 107 ° . The nitrogen permeability varied from 12-2,205 m3/m2 d, depending on the preparation conditions. The maximum pore size, LEPw and mechanical properties were also determined. The membrane distillation process of 1-2% aqueous NaC1 solution was applied as a final test of membrane performance. The permeate flux up to 233 dm3/m2 d was achieved at the temperature of the feed and permeate of 333 K and 293 K, respectively. A chloride elimination in the permeate higher than 99% was reached.

Keywords: Microporous and porous membranes; Membrane preparation and structure; Membrane distillation

1. Introduction Membrane distillation (MD) is a separation process based on evaporation through the porous hydrophobic membrane [1, 2]. The presence of only a vapor phase in the membrane pores is a necessary condition for MD. Thus, the hydrophobicity of the memPresented at the 7th International Symposium on Synthetic Membranes in Science and Industry, Ttibingen, Germany, August 29 - September 1, 1994.

*Corresponding author,

brane, i.e., its non-wettability, plays an essential role in this process. Several polymers such as polytetrafluoroethylene (PTFE), polypropylene (PP), and poly(vinylidene fluoride) (PVDF), with low s u r f a c e e n e r g y , h a v e the n e c e s s a r y hydrophobic properties [3]. Moreover, these polymers exhibit excellent chemical resistance and good physical and thermal stability. The hydrophobic membranes are prepared in a d i f f e r e n t way, d e p e n d i n g on p o l y m e r properties. The PTFE membranes are formed by the stretching and heating process [4, 5]. The membranes from PP or P V D F are

0011-9164/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved.

PII S0011-9164(96)00020-3

2

M. Tomaszewska / Desalination 104 (1996) 1-11

prepared by the thermal phase separation process [6, 7]. The latter can be formed by a wet phase inversion process [8-15]. In this process a polymer solution is cast on a suitable support and immersed into a nonsolvent bath. The structure and properties of prepared membranes mainly depend on the rate of the membrane precipitation. These effects of different solvents and system solvent-nonsolvent on the membrane morphology were investigated [10]. The water soluble p o l y m e r s such as p o l y v i n y l pirrolidone (PVP), polyethylene glycol (PEG) and LiC1 were used as an additive to the casting solution [8-13]. The PVDF merebranes prepared by the wet phase inversion process were hydrophilic and they were used for ultrafiltration. Drying of the membranes restored their h y d r o p h o b i c properties, associated with hydrophobic properties of PVDFpolymer [13]. The PVDFmembranes prepared from a solution with the addition of LiCI were applied for industrial wastewater treatment by MD [13]. The investigation has shown, that the traces of PEG and PVP remained in the membranes and reduced their

(DMF) or dimethylacetamide (DMA) at elevated temperature. The polymer concentrations in casting solutions in DMF ranged from 8-15 wt%, but in the case of DMA, was kept at 8 wt%. LiC1 was used as an additive to the casting solution. The solutions were cast on a glass plate at 293 K with a knife gap of 0.1 ram. The casting time was 15 s, so generally all the prepared membranes were exposed 15 s, unless otherwise reported. The plate with a thin film was immersed into a water bath at 277 K, unless otherwise reported. The precipitated membranes were removed from the coagulation bath and rinsed with running water to remove residues of solvent and LiC1 from the membranes. Then, the membranes were dried, fiat spread out on a glass to avoid their shrinkage. After drying the membranes become hydrophobic. A contact angle of water droplet on the membrane surface was 107 °. The dry membranes were 35-45 ~tm thick.

hydrophobicity. The aim of this work was to obtain PVDF hydrophobic membranes for membrane distillation, exhibiting both a high permeate flux and a high degree of salt separation (rejection). The effect of solvent, polymer concentration, LiC1 addition to the casting solution, exposure time prior to a coagulation step and the temperature of a coagulation bath on the morphology, mechanical properties and MD performance of the resulting membranes were investigated,

The structures of membrane cross-sections were investigated by a Cambridge Stereoscan 180 scanning electron microscope (SEM). The membrane specimens were frozen in liquid nitrogen, broken to obtain crosssections, and coated with gold for SEM observation. Moreover, the membranes were

2. Experimental

2.1. Membrane preparation Membranes were prepared from poly (vinylidene fluoride) (PVDF) supplied by Kureha Chemical Co. Ltd., with an average degree of polymerization 1000. The polymer was dissolved either in dimethylformamide

2.2. Membrane characterization

characterized by the determination of porosity, maximum pore size (pore radius), pore distribution, LEPw (liquid entry pressure of water), gas permeability and mechanical properties. The membrane porosity was determined by gravimetric method, determining the weight of liquid contained in the membrane pores. Porosity was determined both for wet membranes and for membranes after drying. Dry membranes were prewetted by immersion in isopropyl alcohol followed by water to remove the alcohol. The maximum pore size was estimated by the bubble-point method and

M. Tomaszewska / Desalination 104 (1996) 1-11

calculated according to the La Place equation, The pore size distribution was determined through the combined bubble-point method and gas permeability [16]. The liquid entry pressure of water was estimated according to the procedure described in [17]. The mechanical properties of prepared m e m b r a n e s were d e t e r m i n e d by the measurements with the Instron Test, using the strain rate of 5 mrn/min. The tensile strength was calculated dividing the force at break across the cross-section area of the specimens. The performance of prepared membranes was determined by nitrogen permeability and in MD tests, The membrane distillation process was the final test of the PVDF membranes performance. A 1-2% aqueous sodium chloride solution was used as a feed. A simplified scheme of a set-up for MD studies is shown in [4]. The main element of the system was a cell consisting of two compartments, warm and cold, separated by the hydrophobic membrane. The vapor transferred through the membrane was condensed directly in the distillate, in the cold compartment. The MD tests were carried out at the feed temperature of 333 K, and the permeate temperature kept at 293 K. The feed recirculation rate was 4 cm3/s. The permeate flux and chloride elimination coefficient were determined every hour.

3

casting solution. The proposed mechanism is applicable to different polymer/solvent/nonsolvent systems. A porosity of the membrane for MD should be high enough to obtain a considerable permeate flux. A higher porosity is often associated with increasing pore sizes, but this factor also favors a membrane wettability. Thus, parameters applied for the MD membrane preparation should be carefully chosen. The effects of PVDF concentration in the casting solution on the properties of the resultant membrane are presented in Figs. 1 and 2. The performed studies showed that membranes prepared from casting solutions of higher polymer concentration exhibited lower porosity and nitrogen permeability. This is associated with the closer structures of membrane cast from the solutions of higher polymer concentration. Fig. 1 also presents drying effects on porosity. The applied drying procedure slightly decreased the membrane porosity. The SEM micrographs (Fig. 3) show that the s o l v e n t a f f e c t s the membrane morphology, DMA cast membranes exhibit a spongy structure containing macrovoids, without clearly outlined skin layers (see Fig.

I00-

3. R e s u l t s a n d d i s c u s s i o n

90

The morphology of membranes prepared by the phase inversion process mainly depends on the precipitation rate of the polymer from a casting solution upon immersion in the non-solvent solutions. Porous structure, a sponge type is formed by a slow coagulation. The high rate of precipitation leads to the formation of cavities and macrovoids in the membrane. After immersion of the polymer film, a dense skin layer is formed. This skin layer acts as a barrier for diffusion of the coagulation medium, inwards, and the solvent out from the

80 70

porosity !%1

casting solution PVDFIDMF

60 ,o 40

~

,*.,m.,,b,... dry membrane .

7

9

.

. 11

. 13

15

PVDF concentration [wt.%]

Fig. 1.

Effect of PVDF concentration in casting

solutions on the porosity of the resultant membrane: (a)

wet, (b) dry.

4

M. Tomaszewska / Desalination 104 (1996) 1-11

N 2 p e r m e a b i l i t y [m3/m2d]

4o-~,~;

casting solution PVDF/DMF

× messu,e to ~P.

l* *"""" is =P, 30 t 20

*

lo

f o~7

~

. .

9

.

11

.

-"-----...~ .

13

15

PVDFconeentratlon Iwt.%l

Fig. 2. Effect o f P V D F c o n c e n t r a t i o n in casting solutions on the gas flux of the resultant membrane.

3a). The m e m b r a n e cast from the DMF solution presents macrovoids layers located directly underneath a dense skin of 0.9 ~tm (see Fig. 3b). The dense skin caused that the membrane cast from DMF exhibited 3-fold lower gas permeability than the cast from DMA (compare Figs. 2 and 4). These differences in the structure depend rather on solvent-water interaction d u r i n g the immersion of thin film into a water bath, than on PVDF-solvent interaction [10], and a greater tendency of water to mix with DMA than with DMF [11]. The membranes were precipitated in the water bath with different temperatures ranging from 275-293 K. As it can be seen from Fig. 4, the nitrogen permeability decreased nearly 2-fold for m e m b r a n e s prepared at higher temperatures. The effect of bath temperature on MD p e r f o r m a n c e o f the resultant membranes is shown in Fig. 5. The permeate flux was lower nearly 2-fold for membranes precipitated at 293 K, than for those prepared at 277 K. Probably, upon immersion of a polymer film in a bath with a higher temperature, a denser skin layer forms, which determines lower gas and vapor

transport through the membrane. For further studies the bath temperature of 277 K was chosen. The addition of LiCI to a casting solution drastically altered the membrane structure (Fig. 6). The cross-section of asymmetric PVDF membranes consist of an upper layer with large macrovoids with length of several microns, reaching up to the m e m b r a n e surface. The bottom spongy layer contains the rarely deposited macrovoids. The addition of higher amounts of LiC1 caused the formation of larger cavities, extended for about 50% of the membrane thickness. The addition of LiC1 to the casting solution increased the rate of PVDF precipitation during the immersion step, hence a more open structure of the membrane is formed. More rapid precipitation of the polymer from solutions with LiC1 addition is associated with high tendency of the additive to mix with water and the interaction of LiCl-solvent and LiCL-PVDF [11]. As it can be seen from Fig. 6, the upper layer altered with increasing LiC1 amounts in casting solutions and the membrane made from PVDF:LiCI:DMA = 8:3:89 solution was highly defective. Fig. 7a presents the cross-section of the membrane cast from solution PVDF:LiCI: DMA = 8:2:90 precipitated at 283 K. The skin layer is denser than for the membrane precipitated at 277 K, which can explain its lower N2 permeability (see Fig. 4). Fig. 7b presents the cross-section of the membrane cast from the same solution but exposed to the air for 60 s prior to immersion in the bath. The comparison of Figs. 6c and 7b shows that exposure to air, even at temperature of 293 K, m a k e s the m e m b r a n e denser. This observation is in contrast to that described in [9]. This effect can be explained in terms of increase of polymer concentration in the top layer of the cast film by partial solvent evaporation. As a consequence, a denser structure is formed. The effect of LiC1 contents in the casting solution on p o r o s i t y o f the resultant membrane is presented in Fig. 8. According

M. Tomaszewska / Desalination 104 (1996) 1-11

5

Fig. 3. SEM of cross-section of membrane cast from 8 wt% PVDF solution in: (a) DMA, (b) DMF; (A) top layer, (B) bottom layer.

to SEM observations with an increase of the additive amount, the resultant membrane became more porous. The curves in Fig. 8 also show that the applied drying procedure decreased porosity by about 3-5%. The pore distribution of the dry P V D F membranes is presented in Fig. 9.

The estimated pore radius of the membrane cast from solution P V D F : D M A = 8:92 ranged from 54.3 to 122 nm, with the biggest fraction of pore size at 69.8 nm (Fig. 9a). The addition of 0.5 g LiC1 per 8 g of PVDF in a casting solution rapidly increased an average and maximum pore size. The pore

6

M. Tomaszewska / Desalination 104 (1996) 1-11

N2Permeability [m3/m2d] too casting solution PVOF/DMA

80-

6°40-

.__ : 20-,-, ....... SAP, ~, ....... ,o~. *e,:..... ,s~. 0 i 273 278

--,

i

I

283

288

i

293 IKI

bath temperature

Fig. 4. Effect of bath temperature on the gas flux of the

resultant membrane, ~ermeate flux Idm3/rn2dl

12o casting solution PVDFIDMA=8/92

loo-

at-

60~-" 40273

. . . . . 275 277 27g 281

. 283

. 285

. . . 287 289 291

bath temperature

293

[K]

Fig. 5. Effect of bath temperature on membrane distillation p e r f o r m a n c e of the resultant m e m b r a n e : feed = 1-2% aqueous NaC1 solution, TFeed = 333 K, and Tpermeate = 293 K.

radius ranged from 244.3-488.6 nm. The pore fraction of 325.7 nm was above 50% (Fig. 9b). The additional amounts of LiC1 slightly changed the pore distributions (Figs. 9c and 9d). The m a x i m u m pore size (calculated f r o m L E P w ) increased with increasing LiC1 amounts in the casting solution (see Fig. 10). The pore size

c a l c u l a t e d for the dry and p r e w e t t e d m e m b r a n e s are similar, but they were different from the pore size o f the wet membranes. The applied drying procedure reduced the maximum pore radius close to 2fold. The membranes obtained from the casting solutions with greater amounts o f LiCI exhibited lower values of LEPw (Fig. 11). These results confirm the SEM observations. Such more open structures exhibited higher nitrogen permeability, which increased at higher pressure differences (Fig. 12). The comparison of data presented in Figs. 4 and 12 shows that gas permeability of membranes significantly increased when LiC1 was added to the casting solution. The mechanical properties of membranes are the result of its structure. As can be seen in Fig. 13, LiCl's presence in the casting solution drastically decreased the strength at break of the prepared membranes. These results are in agreement with alterations of the membrane morphology (see Figs. 1 and 6). Bigger cavities have r e d u c e d the m e c h a n i c a l resistance of membranes. The membrane distillation process of 1-2% aqueous NaC1 solution was applied as a final test of the membrane performance. The temperature of a feed and cold solution was kept at 333 K and 293 K, respectively. The obtained permeate flux as a function of the LiCI content in the casting solution is presented in Fig. 14. As can be seen, the addition of 0.5 g LiC1 per 8 g of PVDF in DMA solution increased the permeate flux very strong. Additional amounts of LiC1 in a casting solution caused a gradual increase of permeate flux. The observed changes are in a g r e e m e n t with the a b o v e p r e s e n t e d membrane morphology. The membranes with more open structure and thinner skin layer allowed to obtain larger permeate flux. Fig. 15 p r e s e n t s the p e r f o r m a n c e of membranes with different time of exposure to air upon immersion in a coagulation bath. The longer exposure time clearly decreased the obtained permeate flux.

M. Tomaszewska / Desalination 104 (1996) 1-11

7

Fig. 6. SEM of cross-section of membrane cast from 8 wt% PVDF solution in DMA. Effect of LiC1 contents, wt%: (a)

0.5, (b) 1, (c) 2, (d) 3; (A) top layer.

8

M. Tomaszewska / Desalination 104 (1996) 1-11

Fig. 7. SEM of cross-section of m e m b r a n e cast from PVDF:LiCI:DMA = 8:2:90 solution: (a) precipitated at 283 K, (b) precipitated at 277 K, exposure time prior to precipitation 60 s; (A) top layer.

porosity

90

[%J

maximum

1.585- ~

pore radius

casting

[urn]

solution

. , ~

1

80"

casting solution PVDFILiCl/DMA

0.5

75-

" wet membrane 70 0

,

,

,

0.5

1

1.5

,

X dry membrane ,

2

2.5

LiCI c o n t e n t

~(~ prewette(W membrm~e

0

3

[wt.%]

Fig. 8. Effect of LiCI contents in 8 wt% PVDF solution in D M A on the porosity of the resultant m e m b r a n e : (a) wet, (b) dry.

0

,

,

,

0.5

1

1.5

4-~,r

2

2.5

LiCI c o n t e n t

3 [wt.%]

Fig. 10. Effect of LiC1 contents in 8 wt% PVDF solution in D M A on the m a x i m u m pore size of the resultant membrane.

M. Tomaszewska / Desalination 104 (1996) 1-11

a)

b)

1*ractlon [%1

'° I

frectlon [%]

-T

]

{

~.ngl

.lutllon ~

L

'of

I

51,1

ll.O

81,4

I

c~l~ion

ILl 122.2 161,9 144,3 pore radlui lnel

.....

i

196,4 llT,l 244,3 i7i,l

c)

,

q

5al,6 977,1 pori radius [nil

1 ° , 7 150,9 468,6

d)

fraction [%]

liacUon [%]

', '

l 60 c

I

,



54.3

i

:.,.

'°i

48,9

9

!

!

--

-io,L~

F'*~'uc'I°=~ ='P"'I --

L i 60

m

i

267.4 305.4 325.7 34.9 375.8 407.2 444.2 4mLO SglL9 610.6 pore

radius

fnml

.

.

,o

.

i

.

,

r'°~"cl°=~ = ~"'° .

.

,

.

i'........ ill.s

257.2

271.4

287.4

i

305.4

$25.7

349

375,8

407,2

pore radius

444.2

[rim 1

Fig. 9. The pore size distribution of PVDF membranes.

LEPw [kPa]

N2 permeability [m~m2dl

/

290-t

casting solution

240- ~

2300 casting solution PVDFILICIIDMA 1800

190

-

1300140-

,o_l

.oo- ~ ;

40

i p/ellurl

300 0

0.5

1

1.5

2 2.5 LiCI c o n t e n t [ w t . %

Fig. 11. Effect of LiC1 contents in 8 wt% PVDF solution in D M A on LEP W of the resultant membrane,

0

0.5

1

1.5

3 . 2 kPll

2 2.5 LiCI content I w t . % ]

Fig. 12. Effect of LiCI contents in 8 wt% PVDF solution in D M A on the gas flux of the resultant membrane.

10

M. Tomaszewska / Desalination 104 (1996) 1-11 pressure I M P a ] ,

casting solution PVDF/LICI/DMA

1.6 -

~ 0.80

Idm3lm2d]

k ~ 180

casting solution PVDF/LiCIIDMA = 8/2/9(~

~

160-140_

1.2-

0.4

) e rme a t e flux

200

120 -

¥ 0.5

, 1

, 1.5

, , 2 2.5 3 LiCI c o n t e n t { w t . % ]

Fig. 13. Effect of Lie1 contents in 8 wt% PVDF solution in D M A on the mechanical properties (the strength at break) of the resultant membrane, :)ermeate

, 30

15

, 45 exposure time

60 [s]

Fig. 15~ Effect of exposure time on the m e m b r a n e distillation p e r f o r m a n c e of the resultant m e m b r a n e : feed = 1-2% aqueous NaCI solution, TFeed = 333 K, and Tpermeate = 293 K.

flux [dm:~m2dl /

240-

1 O0

,

~

"

4. C o n c l u s i o n s

190 -

casting solution PVDF/LICI/DMA

14o

90

_.,

40 0

0.5

. 1

.

. 1.5

. 2

.

LiCIcontent 2.5

3

3.5 [wt.%] Fig. 14. Effect of Lie1 contents in 8 wt% P V D F s o l u t i o n in D M A on the m e m b r a n e d i s t i l l a t i o n performance of the resultant membrane: feed = 1-2% aqueous NaC1 solution, TFeed = 333 K, and Tpermeate = 293 K.

The chloride elimination coefficient in the permeate was practically 100% for all prepared membranes. Only the membranes precipitated from solution PVDF:LiCI:DMA = 8:3:89 rejected chloride in 98%. The hydrophobicity of the prepared membranes w a s maintained during membrane distillation tests.

The characteristics and properties of PVDF membranes obtained by the wet phase inversion process strongly depends on the composition of casting solution and the temperature of coagulation bath as well. The structure of the resultant membranes markedly changed with Lie1 addition. The prepared membranes maintained hydrophobicity during M D t e s t s . The PVDF membranes can be used for the membrane distillation process.

Acknowledgements This work was supported by the State Committee for Scientific Research, RKH/BW 1995, Poland.

References [1]

E. Drioli and Y. Wu, M e m b r a n e distillation: An experimental study, Desalination 53 (1985) 330.

M. Tomaszewska / Desalination 104 (1996) 1-11

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

E. Drioli, Y. Wu, and V. Calabro, Membrane distillation in the treatment of aqueous solutions, J. Membr. Sci. 33 (1987) 277. M. Mulder (Ed.), Basic Principles of Membrane Technology, Kluwer Academic Publishers, Dordrecht. The Netherlands, 1991. M. Tomaszewska, Concentration of the extraction fluid from sulfuric acid treatment of phosphogypsum by membrane distillation, J. Membr. Sci. 78 (1993) 277. N. Chlubek and M. Tomaszewska, Some properties of hydrophobic membranes for membrane distillation, Env. Prot. Eng. 15 (1989) 95. K.E. Kinzer, G.H. Shipman, Y.F. Yang, D.R. Lloyd, and H.S. Tseng, Thermally-induced phase separation membranes: structure-formation relationships, Proceedings Intl. Congress on Membranes and Membrane Processes, Tokyo, 1987, pp. 344-345. D.R. Lloyd, K.E. Kinzer, and H.S. Tseng, Microporous membrane formation via thermallyinduced phase separation. I: Solid-liquid phase separation, J. Membr. Sci. 52 (1990) 239. T. Uragami, M. Fujimoto, and M. Sugihara, Studies on syntheses and permeabilities of special polymer membranes: 24. Permeation characteristics of poly(vinylidene fluoride) membranes, Polymer 21 (1980) 1047. A. Bottino, G. Capannelli, and A. Bottino, Factors affecting the structure and properties of asymmetric polymeric membranes, In: E. Drioli and M. Nakagaki (Eds.), Membranes and Membrane Processes, Plenum Press, New York, 1986, pp. 163-178.

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

11

A. Bottino, G. Camera-Roda, G. Capannelli, and S. Munari, The formation of microporous polyvinylidene difluoride membranes by phase separation, J. Membr. Sci. 57 (1991) 1. A. Bottino, G. Capannelli, S. Munari, and A. Turturro, High performance ultrafiltration membranes cast from LiC1 doped solutions, Desalination 68 (1988) 167. S. Munari, A. Bottino, G. Camera-Roda, and G. Capannelli, Preparation of ultrafiltration membranes. State of the art, Desalination 77 (1990) 85. Y. Wu, Y. Kong, J. Liu, J. Zhang, and J. Xu. An experimental study on membrane distillationcrystallization for treating wastewater in taurine production, Desalination 80 (1991) 235. P. Abaticchio, A. Bottino, G. Camera-Roda, G. Capannelli, and S. Munari, Characterization of ultrafiltration polymeric membranes, Desalination 78 (1990) 235. A. Bottino, G. Capannelli, and S. Munari, Effect of coagulation medium on properties of sulfonated polyvinylidene fluoride membranes. J. Appl. Polym. Sci. 30 (1985) 3009. W. Kujawski, P. Adamczak, and A. Narebska, A fully automated system for the determination of pore size distribution in microfiltration and ultrafiltration membranes, Separation Sci. and Techn. 24 (1989) 495. A.C.M. Franken and S. Ripperger, Terminology for membrane distillation, European Society for Membrane Science and Technology, 1988.