Preparation of a poly(DMAEMA-co-HEMA) self-supporting microfiltration membrane with high anionic permselectivity by electrospinning

2.1 Materials

The monomers, HEMA (analytic grade), DMAEMA (analytic grade), were obtained from Aldrich. The monomers were purified by distillation under the reduced pressure. Other reagents, such as 2, 2′-azobis (isobutyronitrile) (AIBN, chemical grade), N, N′-dimethylformamide (DMF, analytic grade), methanol and diethyl ether, were purchased from Guoyao Chemicals Co. Ltd. (Beijing, China) and used as received. Distilled water (DI) was used throughout the experiment.

2.2 Synthesis of poly(DMAEMA-co-HEMA)

The ratios of HEMA/DMAEMA (w/w) were 2/1, 4/1, 5/1 and 6/1. As an example, the process of copolymerization with HEMA/DMAEMA=4/1 was described as follows: HEMA (8.0 g), DMAEMA (2.0 g) and methanol (85 ml) were added into a 150 ml three-necked flask equipped with an inlet for a condenser, an inlet for a thermometer and an inlet for the charging reagents. The mixture of reagents was purged with N₂ to deoxygenate for 30 min at room temperature. Afterwards it was heated to 60°C, AIBN (0.1 g) dissolved in 5 ml methanol was added. The polymerization was carried out at 60°C with a magnetic stirrer for 10 h, wherein positive N₂ pressure was maintained throughout. The resulting solution was then precipitated by adding diethyl ether, and the copolymer was dried in a vacuum oven at 60°C for 24 h.

The copolymer was characterized by ¹H NMR spectroscopy (AVANCE AV-500, Bruker Ltd. Co., Britain) with CDCl₃ as a solvent and Fourier transform infrared (FTIR) spectroscopy (NICOLETAVATAR 370, Nicolet Ltd. Co., Tokyo, Japan) by using a KBr tablet as a sample loader. The thermal property of copolymer was investigated with a differential scanning calorimeter (DSC8523, Rigaku Co. Ltd., Tokyo, Japan) in N₂ atmosphere. The scanning rate was 20°C/min.

2.3 Preparation of an anion exchange membrane by electrospinning

Poly (DMAEMA-co-HEMA) was dissolved in DMF to produce various spinning solutions. The equipment and operation are described elsewhere (17).

The as-prepared nanofiber was evaluated by scanning electron microscopy (SEM) (JSM-5310, JEOL, Tokyo, Japan). The diameter of nanofiber was calculated from the SEM photos (17), (18). The photos were magnified 10 fold at first and then printed. The diameters of about 100 microfibers were selected and measured with an ordinary ruler. The error was about 0.5%.

2.4 Crosslinking the membrane and quaternization of the amines

Quaternization of amine groups was performed analogous to staining the sample of the amino substance with iodomethane vapor for transmission electron microscopy (TEM) (17), (18). The membrane (40×40 mm) was placed on a scaffold settled in a 100 ml beaker. Three milliliters of CH₃I was poured into the beaker without flooding the membrane, and then the beaker was sealed with plastic film. The quaternization of amine was conducted in the dark at room temperature for a week. As for the crosslinking reaction, the apparatus and device were the same as described above, except that a 500 ml beaker was used. A hundred milliliters of formaldehyde aqueous solution (30%) was charged into the beaker without flooding the membrane, and then the beaker was sealed with a plastic film (19). The beaker was placed in a water bath and the crosslinking reaction was carried out at 60°C.

2.5 Water sorption of the anion exchange membrane

The water sorption was equilibrated in deionized water for 12 h. After the excess water on the sample surface was wiped off, the sample was weighed (m_wet). The dry weight (m_dry) of the sample was determined after drying to a constant weight under a vacuum at 60°C. The water sorption, φ_w, was calculated as following.

φω= mwet−mdrymdry×100%

2.6 Determination of permselectivity

The transport number of gegenions and the value of permselectivity were determined by measuring the membrane potential. We made the equipment used ourselves. Its structure is shown in Figure 1. The size of the plastic cell was φ 30 mm×100 mm. The diameter of the connector was φ 5 mm.

Figure 1:

Schematic diagram of equipment for the determination of permselectivity.

Different concentrations of salt, for instance, 0.1 mol/l KCl and 0.2 mol/l KCl, were injected into two cells at 25°C, respectively, and we then measured the potential using a digital potentiometer (SDC-II, Sangli Electronics Co., Nanjing, China). The relation between the membrane potential (E_m) and the transport number of gegenion (t¯g) through the membrane are as follows:

t¯g=Em+Em02Em0

where t¯g is the transport number of the gegenion through the membrane, Em0, is the ideal potential (16.1 mV) and E_m, is the membrane potential measured. The permselectivity of the membrane was calculated as:

P=t¯g−tgtg0¯−tg=t¯g−tg1−tg

where P is the permselectivity, t¯g, is the transport number of gegenion through the membrane and t_g, is the transport number of gegenions in the solution found in the relevant manual.

2.7 Mechanical property

The mechanical properties of membranes were tested with a gage length of 50 mm and crosshead speed of 10 mm/min by electronic fiber strength tester (Series IX Automated Materials Testing System, Instron Corporation). Ten samples were tested and the mean values are used in the article. All samples were conditioned in a laboratory environment for 24 h before testing.

3.1 Characterization of the copolymer

Both the DMAEMA and HEMA are the methacrylate ester. Although the copolymerization reactivity ratios of the two monomers were not found, theoretically it is rational that their copolymerization reactivity ratios are similar. Teoh et al. reported that, in the ATRP copolymerization, the copolymerization reactivity ratios of (1) HEMA and (2) DMAEMA were near unity (r₁=1.08, r₂=1.12) in the polar solvents (20). Therefore, the composition of copolymer was close to the feed ratio of monomers. The random copolymer of P(DMAEMA-co-HEMA) was characterized by FTIR and ¹HNMR as shown in Figure 2. The broad peak at 3450 cm⁻¹ was attributed to the vibrations of the hydroxyl groups. A peak at 2932 was assigned to the asymmetric vibration of CH₃. A peak at 1650 cm⁻¹ was the typical stretch vibration of the ester carbonyl group. The absorption band at 1190 cm⁻¹ corresponded to the symmetric stretching vibration of C-N band associated with the DMAEMA unit.

Figure 2:

FTIR spectra of poly(DMAEMA-co-HEMA).

A representative ¹H NMR spectrum of poly (DMAEMA-co-HEMA) in d-CHCl₃ is shown in Figure 3. The peak, e.g. (δ=4.1 ppm) belongs to the α-protons of the ester groups in both DMAEMA and HEMA; peak f (δ=3.7 ppm), protons of ethyl groups in HEMA; peak d (δ=2.6 ppm), protons of ethyl groups in DMAEMA; peak c (δ=2.3 ppm) and protons of methyl groups in the tertiary amine of DMAEMA.

Figure 3:

¹H NMR spectra of poly(DMAEMA-co-HEMA) in d-CHCl₃ with peaks assigned.

A representative DSC diagram of copolymer (HEMA/DMAEMA=4/1) is shown in Figure 4. It is clear that there is only one T_g existing in the figure. The single peak of T_g usually reflects the homogeneous distribution of components in a polymer blend, namely that no domain or phase formed solely by one kind of polymer blocks exists in the blend. In the case that two polymeric units are immiscible, it can be a criterion to prove the formation of copolymer. Therefore, this result indicated that the random copolymer of poly(HEMA-co-DMAEMA) was successfully prepared. Moreover, as shown in Figure 4, T_g was about 320 K. However, Martin-Gomis et al. investigated the bulk copolymerization of HEMA and DMAEMA, and the thermal properties of copolymer with various compositions (21). They reported that T_g of poly(HEMA-co-DMAEMA) with 20% DMAEMA was about 330 K. The difference was likely the result of an experimental error, i.e. the determination of baseline (Figure 4). Anyway, it was not the main intention of this paper to accurately determine the T_g of the copolymer. Therefore, to some extent, T_g of ca 320 K supported the formation of a copolymer with the same composition as the feed ratios.

Figure 4:

DSC diagram of copolymer (HEMA/DMAEMA=4/1).

3.2 Determination of the optimal electrospinning parameters

As we know, electrospinning of poly (DMAEMA-co-HEMA) was carried out for the first time in this paper. Therefore, first of all, the electrospinning parameters should be obtained. In the preliminary experiments, it was observed that the microfibers were prepared without beads only when the total concentration of the copolymer was higher than 40 wt%. However, when the concentration was higher than 50 wt%, the distribution of the microfiber diameters became significantly more broad. Moreover, the averaged diameter of nanofibers was overwhelmingly dependent to the concentration, with less regard to the composition of the copolymer. These were the features of electrospinning systems applied in this paper, and also, the reasons that the following parameters were selected for the optimization of conditions to prepare the uniform microfibers.

The quantitative evaluation and statistic analysis of the effects of solution properties and processing parameters were investigated with the orthogonal experiment design. Four relevant factors were investigated: (A) solution concentration, (B) voltage, (C) flow rate and (D) tip to collector distance. Three levels were set for each factor, and the boundary values for the levels were determined in pilot experiments to ensure microfiber formation without droplets. Solution concentrations of 42 wt%, 44 wt% and 46 wt% were used. The feed rates of 0.7, 1.0 and 1.3 ml/h with a syringe pump were employed. The applied voltage ranged from 10 to 14 kV, and the distance from the collector to the needle tip ranged from 12 to 16 cm. All electrospinning experiments were carried out at 25°C and relative humidity of 40%. The surface morphologies of microfibers were examined by SEM. The fiber diameter averaged over at least 100 microfibers was determined by SEM micrographs. The aim of orthogonal experiments was to analyze the major factors affecting the diameter of microfibers by L₉ (3⁴) orthogonal table arrangements. The electrospinning parameters of the orthogonal experiment, fiber diameters and its standard deviation are listed in Table 1. As shown in Table 1, K value represented the average of three values at each level for each factor. R_d, Rs² represents the difference of maximal and minimal microfiber diameters and its standard deviation, respectively.

As shown in Table 1, it was obvious that, for the microfiber diameter, the significance levels are listed in the order: solution concentration >voltage >collector distance >flow rate. According to the results of the range analysis, the solution concentration played the most important role on the fiber diameter. Hence, in order to reduce the average diameter of microfibers, the best combination of the level of each factor was A₁B₁C₁D₂ according to the relationship between factors and fiber diameters. However, the standard deviation of microfibers must be taken into account as it represents the uniformity of microfibers. Therefore, overall the best combination was A₁B₁C₂D₁. Namely, the proper conditions for electrospinning were: solution concentration, 42 wt%; voltage, 10 kV; flow rate, 1.0 ml/h; tip-collector distance, 12 cm. Under these conditions, the averaged diameter of microfibers was about 1200 nm and deviation was 0.05.

3.3 Membrane characterization

The final experimental microfibrous membrane was obtained with the optimal electrospinning parameters when the electrospinning time was 30 min. As shown in Figure 5A, the as-prepared microfibrous membrane was not directly applied for the separation process because of its loose structure. Accordingly, the microfibrous membrane was fumed in formaldehyde atmosphere at 60°C, so that the microfiber copolymer was crosslinked due to the reaction of hydroxyl groups in HEMA and formaldehyde. Obviously, after being crosslinked, the membrane became denser (Figure 5B). To further elucidate the microscopic structure of the membrane, the typical surface morphologies at pre- and post-treatment were observed by SEM and are shown in Figure 5C and D, respectively. It is clear that the crosslinked microfibers became flat and curved (Figure 5D). Such a structure implies that the crosslinking reaction may occur in both intra-microfiber copolymers and inter-microfiber copolymers.

Figure 5:

Optical and SEM photographs before and after crosslinking (A: optical photograph before crosslinking, B: optical photograph after crosslinking, C: SEM photograph before crosslinking, D: SEM photograph before crosslinking).

3.4 Water sorption

Amino groups of DMAEMA are the active groups in the membrane. Hence, theoretically the more DMAEMA the better performance can be anticipated. In this paper, four molar ratios of monomers were employed to carry out the copolymerization, i.e. HEMA/DMAEMA=2/1, 4/1, 5/1 and 6/1. The difference in diameters of microfibers with these ratios was insignificant because, as discussed above, the diameter was dominantly determined by the concentration. However, the difference in water sorption was very big. As shown in Figure 6, the content of absorbed water decreased as the composition of HEMA increased. Meanwhile, the water sorption decreased with the increase of crosslinking time, regardless of the ratio of the monomers. This result was normal because HEMA mainly provided the hydroxyl groups for the crosslinking. The water sorption of a membrane electrospun with HEMA/DMAEMA=2/1 was not measurable because it almost dissolved in water, even though it was crosslinked for 32 h. These results indicated that the crosslinking reaction occurred in microfibrous membranes because the hydrophilicity decreased with the increase of crosslinking time. It was coincident to the results reported on PVA, although the water sorption in this paper was much higher (8). As we know, the hydrophilicity of anion exchange membranes strongly impacts to the perselectivity of anions, namely that decreasing hydrophilicity of membranes favors the permeation of less hydrated anions. Hence, for the application, the HEMA/DMAEMA ratio of 4/1 was the best and the crosslinking time should be shorter than 28 h.

Figure 6:

Water sorption of the cross-linked membranes vs. the crosslinking time.

3.5 Mechanical properties

The mechanical properties of microfibrous membranes with HEMA/DMAEMA=4/1 were measured and summarized in Table 2. It is clear that the tensile strength and modulus increased with the increase of crosslinking time. The elongation at the break at first increased as the crosslinking time increased from 20 to 24 h, and then decreased with the increase of crosslinking time. These results indicated that the density of crosslinking points in the membrane increased with the increase of crosslinking time. At 24 h, the membrane was the strongest. When the crosslinking time exceeded 24 h, the membrane became more and more brittle. Therefore, it was determined that 24 h was the proper time of crosslinking.

The mechanical properties of various microfibrous mats (22), (23) or single microfibers (24) have been investigated by many researchers. For example, Wang et al. (5) reported that the tensile strength of PVA microfibrous mat crosslinked with glutaraldehyde in acetone for 24 h was close to 8 MPa, whilst the uncrosslinked one was about 6 MPa. In contrast, the tensile strength of PAN and PVDF microfibrous mats were about 4 and 2 MPa at the break, respectively. The tensile strength of crosslinked PVA was much higher than that of the crosslinked poly (DMAEMA-co-HEMA) microfibrous mat shown in Table 1. However, we must point out, it is not rational to compare the data of tensile strength given in different papers as the mechanical test of the microfibrous mat is dependent on the precise measurement of the section area of the mat. The microfibrous mat is soft and sponge-like. A screw micrometer is usually applied to measure the thickness. However, the contact pressure of the micrometer on the surface of the mat is determined artificially. The difference in pressure gives a different number of thickness. Therefore, in reality the data of tensile strength given by different measurements is not comparable, even though, we think that the different crosslinking methods might result in different strengths. For example, in the reference, the glutaraldehyde could directly contact with the hydroxyl groups in acetone (5). However, in the present paper, the above method was not applicable because the copolymer was soluble in both water and acetone. Hence, the formaldehyde vapor was applied. By this means, the crosslinking reaction was dependent on the condensation and adsorption of formaldehyde in the mat. Reminiscent of the water sorption and big diameter of microfiber, it was possible that the crosslinking reaction mainly occurred on the surface of the fiber. And also, the bigger diameter of nanofiber decreased the density of inter-fiber contact points, in turn decreases the tensile strength. Accordingly, the crosslinking method an be improved in the future in order to further increase the tensile strength.

In the references, microfibrous mat was only applied as the supporting materials (2), (3), (4), (5), (6), (7). The active material was needed to be grafted on the microfibrous mat by interface polymerization. Therefore, comparatively, the method described in this paper was simple and effective to prevent the active materials from exfoliation in practical use.

3.6 Permselectivity

The crosslinked and quaternized membrane was directly employed for the membrane potential (E_m) measurement without using other supporting layers such as a non-woven PET mat. Moreover, it was sustained for a week without dissolution. It indicated that the strength of this membrane was strong enough to be a self-supporting microfiltration membrane. KCl, KNO₃ and K₂SO₄ were selected to evaluate the permeability of anions at 25°C, and the results are listed in Table 3. The permeabilities of NO₃⁻, Cl⁻ and SO₄²⁻ were 94, 91 and 87%, indicating a good permselectivity of membrane. Obviously, the smaller radius of hydrated anions, the faster anions migrate in solution.