Effect of Non-Solvents Used in the Coagulation Bath on Morphology of PVDF Membranes

The aim of this paper was to prepare a poly (vinylidene fluoride) (PVDF) membrane using different non-solvents in the coagulation bath for the phase inversion method. In order to increase the mechanical strength of membranes, facing the pressure of work, was used a macro-porous polyester support. The morphology and structure of the resulting membranes were evaluated by scanning electron microscopy, porosity measurements, water and 1-octanol uptake, contact angle, pure water flux, hydraulic permeability and hydraulic resistance. The morphology and pure water flux changed significantly using ethanol (symmetric membrane) and/or water (asymmetric membrane) as the non-solvent. The symmetric membrane presented a high hydrophobic surface (water contact angle ~136°) and a higher pure water flux and porosity than the asymmetric membrane, which presented a lower hydrophobicity surface (water contact angle ~90°). The morphologies obtained suggest different applications.


Introduction
Membrane processes are increasingly important to treat drinking water and wastewater.Substantial advances in the preparation of membranes and the composition of membrane materials have led to this breakthroughs 1 .The cost of membranes continues to fall and it is possible to predict that membranes will be in wide use in water treatment facilities 2 .
Poly (vinylidene fluoride) (PVDF) is a very attractive polymer, exhibiting chemical and mechanical resistances, widely used at to synthesize a lot of membrane operations 3 .
The key advantages of PVDF include its highly hydrophobic nature and excellent chemical resistance against corrosive chemicals such as acids, bases, oxidants and halogens.Typically, PVDF comprises crystalline phases along with amorphous and/or rubbery regions which respectively provide the thermal stability and flexibility of the membranes.Compared to other hydrophobic materials such as poly (tetrafluoroethylene) (PTFE) and polypropylene (PP), PVDF has better processability since it is soluble in common organic solvents 4 .With these superior properties, PVDF has been considered for utilization in various membrane-based separations e.g.micro/ultra/nano-filtration 5,6 , membrane distillation [7][8][9] , pervaporation 10 and membrane gas absorption 11,12 .
The phase inversion technique is commonly used to prepare PVDF membranes.Membrane precipitation is induced by means of liquid-liquid demixing and/or crystallization.The membranes formed often exhibit characteristics from both types of phase separations, cellular pores or interlinked crystalline particles 3 .
In the phase inversion process, there are many preparation parameters affecting polymer precipitation and ultimately the morphology of the formed membrane.The polymer concentration, the use of additives [13][14][15] , the solvent 14 , the composition and the temperature of the coagulation bath 3,16,17 , the composition of the casting solution 13,18 and the temperature of the casting solution 19,20 are the most influential and the most frequently discussed in the literature.
The hydrophobic PVDF membranes are extensively used for water treatments as a form of flat sheet or of hollow fiber, because the membrane offers high void volumes, has well defined porosity and it is chemically inert 21 .Kuo et al. 22 investigated the permeation flux and rejection coefficient of the porous PVDF membranes with different hydrophobicity, in a direct contact membrane distillation system and obtained high permeation flux and high salt rejection coefficient which were close to the commercial membrane.
In the present study we have focused our attention on the effect of non-solvents on the morphology, hydrophobicity and pure water flux of the PVDF membranes.The PVDF membranes obtained were characterized by density, porosity, water contact angle, pure water flux, hydraulic permeability, hydraulic resistance and scanning electron microscope (SEM).
(DMF) from Vetec.The non-solvents used were distilled water and ethanol from Simoquímica.The support used was woven macroporous 100% polyester with a thickness of 180 µm purchased from Notredame Trade and Imports.

Membrane preparation
Membranes were prepared by the immersion precipitation method.The dope solution was prepared at 60 ± 2 °C by dissolving the PVDF (20 wt.(%)) in DMF (80 wt.(%)) for 24 hours.The polymer solution (with 0.2 mm of thickness) was cast on polyester support on a glass plate and was immersed into a coagulation bath.Two types of non-solvents were used to prepare the A and B membranes.The non-solvent utilized in the A membrane coagulation bath was only deionized water for 4 hours.The B membrane was immersed into ethanol for 2 seconds, and then was immersed into water coagulation bath for 4 hours.After precipitation, the membranes were dried at room temperature for 48 hours.

Scanning electron microscope (SEM)
The membrane morphology was examined by a Shimadzu (Model SSX-550) scanning electron microscope (SEM).In SEM studies, the PVDF membrane was carefully removed from the support (because it was not possible to fracture the support) and the sample was fractured in liquid nitrogen and coated with Au by sputtering.

Density of the membranes
Membrane density was determined according to ASTM D792-08 23 .Membrane density can be related to membrane porosity.

Porosity
The method used to determine porosity was proposed by Wang et al. 24 The porosity was measured by soaking the membrane into 1-octanol for 2 hours and then the membrane surface was dried by filter paper.The membrane was weighed before and after absorption of the 1-octanol.The porosity was calculated using Equation 1.
where ε is the porosity of the membrane (%), m p is the mass of membrane (g), m b is the mass of absorbed 1-octanol (g), ρ p is the density of membrane (g.cm - ), and ρ b is the density of 1-octanol (g.cm -3 ).

Water and 1-octanol uptake measurements
The water uptake (WU) and 1-octanol uptake (OU) were determined by soaking the membrane samples in deionized water and 1-octanol, respectively and separately, until it reached the condition of sample saturation, at 25 ± 2 °C.Samples were removed and the excess liquid adhered to both sides of their surface was quickly blotted by filter paper.The water uptake and the 1-octanol uptake were determined as the weight gain of the membrane in the wet state compared to the dry state using Equation 2.
( ) where W w is weight of wet membranes (g) and W d is weight of dry membrane (g).The same equation is used for both calculations.These tests, along with water contact angle measurement were used for the purpose of verifying the hydrophobicity of the membranes.

Water contact angle measurement
Water contact angle of membrane surface was measured by using the sessile drop method, and all the contact angle data were an average of five measurements on different locations of membrane surface.A water droplet was introduced on the surface of membranes, and the contour of the water drop was recorded.
In this method, by optical microscopy, we obtain the profile of a drop deposited on a horizontal surface.The image profile of the drop obtained by the digital camera is analyzed by software (Surftens) adjusting the diameter of the drop and the contact angle with the surface.The shape of the drop deposited on a surface is determined by the interaction of three forces that are positioned along the contact line formed by each pair of the three phases: the line between solid and liquid phase (γ SL ), between solid and vapor phase (γ SV ), and between the liquid and vapor phase (γ LV ), as in Figure 1.All three forces define the overall shape of the drop, and the surface tension (energy per unit area).

Normalized flux
The permeated flux of water through the membranes was normalized to the temperature of 20 °C, considering the coefficient of permeability of the membrane as a constant.The normalization was calculated using Equation 4. .
Figure 1.Forces acting on a drop deposited on a solid surface.
where J 20 is the membrane flux in the temperature of 20 °C (L/m 2 h), J T is the membrane flux at temperature T (L/m 2 h), η 20 is the dynamic viscosity of the water at a temperature of 20 °C (N.s/m 2 ) and η T is the dynamic viscosity of the water at temperature T ( N.s/m 2 ).

Compaction
The prepared PVDF membranes were compacted with deionized water at 9 bar until stable flux in a permeation system of 0.063 m 2 area.After compactation, the pure water flux at different transmembrane pressures (ranging 0-9 bar) were measured under steady state condition using Equation 5.
where J w is pure water flux ( L/m 2 h), V is the volume of permeate (L), A is effective membrane area (m 2 ) and ∆T is sampling time (h).Hydraulic permeability was calculated from the slope of linear relationship between the pure water flux (J w ) and transmembrane pressure, using Equation 6.
where P m is hydraulic permeability (L/m 2 h.bar) and ∆P is transmembrane pressure (bar).
The membrane resistance was evaluated from the slope obtained by plotting water flux versus transmembrane pressure difference, using Equation 7.
where R m is membrane hydraulic resistance (bar.m 2 .h/L),∆P is transmembrane pressure (bar) and J w is water flux (L/m 2 h).

Scanning Electron Microscope (SEM)
SEM micrographs of the A membrane coagulated in pure water bath and B membrane coagulated in ethanol and in pure water at room temperature are shown in Figure 2. To evaluate the structure of membranes by SEM, it was necessary to remove the macroporous support to enable the cryogenic fracture of the samples.The skin membranes obtained from the different non-solvents coagulation bath, presented greater differences caused by the variation of the coagulation bath.
The A membrane exhibit an asymmetric structure showed in Figure 2a-c.A dense thin top layer was formed, namely "skin" on the surface, and a sponge-like structure was observed in the sublayer of resulting membrane.
The membrane coagulated in ethanol for 2 seconds and then immersed in pure water (Figure 2d-f) presented a symmetric structure, without a dense top layer.The B membrane exhibits a porous structure throughout the cross section.Concerning PVDF membranes, Bottino et al. 29 reported that the cavities are separated by walls composed of discrete polymer globules.The porous sublayer structure, presented by the B membrane, is globular and the size of globules appears to be fairly uniform over the entire membrane cross section.
The morphologies presented in this work, was similar to those observed by Yong et al. 30 where the PVDF membrane precipitated from water (harsh bath) exhibited an asymmetric structure and using 1-octanol (soft bath) in the coagulation bath, the membrane presented a uniform microporous structure, composed of spherical particles.
Li et al. 31 attributed the morphology obtained in the B membrane, to the coagulation ability of the coagulant.Ethanol (∆δ c-p = 10.94MPa 1/2 ) presents poorer coagulation ability than water (∆δ c-p = 33.43MPa 1/2 ) for PVDF solution.The weakened coagulation ability slows the precipitation rate during the immersion process.Coagulation plays an important role in the formation of membranes by phase inversion processes.According to Tomaszewska 32 , porous structure, like a sponge type, which was observed in B membrane, is formed by a slow coagulation.In other hand, when an instantaneous coagulation occurs, a dense skin layer is formed.
According to some authors 30,33 , a dense top layer observed in A membrane, is formed because solvent in the casting solution desolvates into the coagulation bath before the coagulant diffuses into the casting solution.The dense top layer stops growing when enough coagulant medium diffuses into the sublayer solution to create pores.On the other hand, if the coagulant medium diffuses into the casting solution before solvent in the casting solution desolvates into the coagulation bath, the coagulant medium may create pores in the membrane surface.The porous surface occurs when a high ratio of the non-solvent inflow compared to the solvent outflow.

Density, porosity and thickness of the membranes
The porosity of each membrane was calculated using Equation 1 and they are reported in Table 1.The difference of porosity may be explained by variation of the non-solvent used in the coagulation bath.The porosity of A membrane prepared using water as coagulation bath was lower than B membrane prepared using ethanol and than water.The porosity difference was small because it takes the entire membrane structure into account, not only the top layer.Studies indicate that in general the top layer thickness of the membrane increases and the porosity decreases with increasing molar volume of the external non-solvent species, due to their lower diffusion rate 3 .However Lin et al. 34 found the opposite result, as well as the results found in our present study, which were attributed to a stronger interaction between the polymer and the non-solvent (i.e.thermodynamic factors).The difference found between the A and B membrane morphology and the absence of the dense top layer is shown in Figure 2. The density of A membrane was slightly lower than B membrane, however this small difference, may be disregarded.The contribution of the PVF fraction was small, because the measures were carried out to the supported membranes (PVDF + polyester support).

Water and 1-octanol uptake measurements and water contact angle measurement
The water and 1-octanol uptake maximum and the average contact angle of the membranes prepared are reported in Table 2.This test was carried out in order to verify the greater hydrophobicity of B membrane, as expected, according to the literature 22 .The condition of sample saturation in water occurred within 48 hours.For the samples in 1-octanol this was observed within 24 hours.Evaluating the results of Table 2, the B membrane presented a lower water absorption capacity with higher porosity, thus proving that it is more hydrophobic than the A membrane.The A membrane presented lower porosity, but higher water absorption.This conclusion can also be evidenced by the contact angle measures, shown in Figure 3, where the B membrane (Figure 3b), is significantly more hydrophobic than the A membrane (Figure 3a).The higher hydrophobicity is also proved by the higher absorption of 1-octanol.
Li et al. 31 concluded that hydrophobicity is related with the surface porosity for PVDF membrane.As discussed above, the use of ethanol, in the coagulation bath, increased the porosity of the surface, resulting in a membrane with greater hydrophobicity.

Compaction and pure water flux
The membranes were subjected to hydraulic compaction at a constant transmembrane pressure of 9 bar for 4 hours until a steady-state flux was attained.The water flux was calculated using Equation 5 from the experimental permeate flux rate measured at 5-minute intervals.Through the permeability experiments it was also possible observed that there was good adhesion between the support and PVDF film when the membrane was subjected to high operating pressures.
The effect of compaction time on the pure water flux for the A and B membranes is shown in Figure 4.It was observed that the pure water flux decreased gradually due to compaction with time and, after about 2 hours of compaction, it was reached a steady-state value.This is due to the fact that the pore walls become closer, denser and uniform resulting in reduction of pore size and flux during compaction 35 .
The higher flux presented by B membrane is due to the higher porosity as discussed in the preceding section.
Transmembrane pressure versus pure water fluxes are plotted as shown in Figure 5.In general, the pure water flux increases with increasing transmembrane pressure and the increase in the operating pressure increase the driving

Conclusions
The use of different non-solvents in the preparation of PVDF membranes caused significant differences in morphology and permeability characteristics of the membranes.The results show that: • Different morphologies, asymmetric and symmetric membranes, can be obtained by varying the non-solvent in the coagulation bath; • The A membrane morphology prepared in the water coagulation bath presented lower porosity and lower hydrophobicity compared to B membrane prepared in the ethanol and than water coagulation bath; • The pure water flux and hydraulic permeability of B membrane are higher than A membrane; and • Hydraulic resistance is the opposite of hydraulic permeability, thus the B membrane offers lower resistance to water flux.It was concluded that the PVDF membranes prepared by the phase inversion method using different non-solvents significantly affect the membrane structure and the membrane properties.

Hydraulic permeability (Pm) and membrane hydraulic resistance (R m )
Membrane hydraulic permeability is significant, particularly for membranes used in a pressure-driven separation process.Permeability was determined from the slope of the plots in Figure 5 using Equation 6and is represented in Table 3.It was observed that the hydraulic permeability (Pm) is higher in B membrane, which confirms that the B membrane is more porous than the A membrane.
Membrane hydraulic resistance (Rm) is the intrinsic resistance of the membrane determined using pure water as feed and obtained by Equation 7 [36] .The A membrane offers higher hydraulic resistance compared to B membrane.This result is in agreement with that of pure water flux experiments.The Rm of the membranes was calculated from the slope of the plot and given in Table 3.It was seen that

Figure 3 .
Figure 3. Water contact angle measurement of the A and B PVDF membranes, respectively.

Figure 4 .
Figure 4. Effect of time on pure water flux of A and B PVDF membranes.

Figure 5 .
Figure 5.Effect of transmembrane pressure on pure water flux of A and B PVDF membranes.

Table 1 .
Properties of PVDF membranes prepared in different non-solvent coagulation baths.

Table 2 .
Water and 1-octanol uptake and contact angle measurement of the PVDF membranes.

Table 3 .
Hydraulic permeability and hydraulic resistance of the PVDF membranes.