- Email: [email protected]
Nylon 66 system
EUROPEAN POLYMER JOURNAL
European Polymer Journal 42 (2006) 356–367
www.elsevier.com/locate/europolj
Fine structure and crystallinity of porous Nylon 66 membranes prepared by phase inversion in the water/formic acid/Nylon 66 system Dar-Jong Lin, Chi-Lin Chang, Chih-Kang Lee, Liao-Ping Cheng
*
Department of Chemical and Materials Engineering, Tamkang University, Taipei 25137, Taiwan, ROC Received 3 May 2005; received in revised form 4 July 2005; accepted 13 July 2005 Available online 24 August 2005
Abstract Microporous Nylon-66 polymeric membranes were prepared by immersion–precipitation from a ternary system, water/formic acid/Nylon 66. Depending upon the precipitation conditions, membranes with morphologies that reflect the sequence of liquid–liquid demixing (as characterized by cellular pores) and crystallization (as characterized by crystal particles) events during the course of precipitation were obtained. The details of the membrane morphologies were disclosed using a low voltage field emission scanning electron microscope (FESEM) at very high resolutions. In particular, nano-scale fine structures such as dendritic crystal elements, nano-pores, nano-grains, branching lamellae, etc., which were rarely presented in the membrane literature. Wide angle X-ray diffraction analyses indicated that Nylon66 crystallized into ÔaÕ structure in all prepared membranes. Crystallinities were determined by appropriate deconvolution of the diffraction peaks. The results indicated that membranes prepared by a well-dissolved casting dope had a somewhat higher crystallinity than those prepared by incipient dopes being in metastable states with respect to crystallization. This observation was confirmed by Fourier transform infrared spectroscopy and DSC thermo analyses. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Polyamide; Immersion–precipitation; Cellular pores; Spherulites; Crystallinity
1. Introduction Nylon 66, a polyamide derived from 1,6-hexamethylene diamine and adipic acid, is a semicrystalline polymer, which possesses good thermal stability and mechanical strength, and is considered to be an important engineering thermoplastic [1–3]. Porous polyamide membrane has been commercialized for many years
*
Corresponding author. E-mail address: [email protected] (L.-P. Cheng).
and is nowadays widely used in fine-separation processes [4,5]. Microporous membranes are often manufactured by the so-called immersion–precipitation process [6], in which a polymer solution is cast on a substrate and then immersed in a nonsolvent bath to induce polymer precipitation by means of crystallization and/or liquid– liquid demixing. Unlike a nonporous Nylon 66 film, which has a water contact angle of ca. 60°, a skinless microporous Nylon 66 membrane is water wettable; i.e., water drops can penetrate into the membrane matrix within a few seconds. This property is associated with the porous morphology of the membrane. It can be
0014-3057/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.eurpolymj.2005.07.007
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
produced by a fine-tuned immersion–precipitation process, which involves precipitation of an incipient casting dope in a soft coagulation bath [7,8]. Because the casting dope is metastable with respect to crystallization and the bath contains a large amount of solvent, crystallization dominates over liquid–liquid demixing to yield a structure consists of interlinked stick-like crystalline particles. The formation mechanism of crystallizable polyamide membranes has been investigated by a number of authors. On the basis of phase diagram and mass transfer modeling, Bulte et al. obtained the initial diffusion paths and concentration profiles for various immersion conditions for the preparation of Nylon 46 membranes. These calculated results correlated well with the membrane morphologies [9,10]. Cheng et al. took into consideration the effects of dopeÕs aggregation state (e.g., a supersaturated dope may be prepared which contains a high population of pre-nucleation embryo) and bath strength (e.g., a soft bath containing significant solvent may be used) on the relative level of liquid–liquid demixing and crystallization during the course of precipitation [8,11,12]. A mathematical model was developed to illustrate the concentration evolution of nonsolvent, solvent, and polymer within the membrane prior to phase separation. In conjunction with the phase diagram, these results help to understand the relation between morphology and immersion conditions. Recently, Thomas et al. used confocal microscope to observe scattering profiles of casting dopes during immersion–precipitation, and found that precipitation occurred initially at some distance from the membrane–bath interface, which then propagates both upward and downward through the membrane [13]. Techniques other than immersion– precipitation have also been used to prepare porous polyamide membranes. For example, Kho et al. introduced compressed carbon dioxide as an antisolvent to precipitate Nylon 6 from 2,2,2-trifluoroethanol solutions to form porous membranes. In this way, pressure became an additional effective variable for tailoring the membrane microstructure [14]. Despite that Nylon-66 membrane formation mechanism has been discussed previously [8,11,12,15], the issues concerning nano-scale fine structure of the membrane such as twisted lamellae, nano-grains, etc., and thermal behavior and crystallinity of the membrane
357
have never been presented. These characteristics were investigated using low voltage field emission scanning electron microscopy (LVFESEM, fine structure), differential scanning calorimetry (DSC, thermal behavior and crystallinity), Fourier transform infrared spectroscopy (FTIR, crystallinity) and wide angle X-ray diffraction (WAXD, crystallinity and crystal structure) [16–18]. In particular, a spectrum deconvolution technique based on Gauss and Lorentzian functions were employed to analyze both WAXS and IR spectra, and to determine the degree of crystallinity of the membrane. The results were found to be consistent with those reported in the literature [17,19,20].
2. Experimental 2.1. Materials Poly(hexamethylene adipamide) (Nylon 66, Zytel 101, Du Pont, measured intrinsic viscosity = 2.68 g/dl, Mv = 87,000 g/mole) was received in pellet form. Formic acid (Acros, 99%) was used as the solvent and distilled–deionized water was used as the nonsolvent for Nylon 66. All materials were used as received. 2.2. Membrane formation A dope solution was uniformly spread on a glass surface using a casting knife. It was then immersed at 25 °C into a coagulation bath composed of water and formic acid. After precipitation has completed (typically a few min), a white solid membrane was obtained. The nonsolvent and the residual solvent in the nascent membrane were removed by a sequence of washing steps. Finally, the formed membrane was held tightly in a press between two sheets of filter papers and then dried at 40 °C. The immersion conditions for various membranes are summarized in Table 1. Dope ‘‘A’’ was a homogeneous solution consisting only of polymer and solvent. Dope ‘‘B’’ contained a substantial amount of nonsolvent. It was in a supersaturated state with respect to crystallization (i.e., its composition is located below the crystallization line) and would gel upon standing at 25 °C for an extended period of time. However, prior
Table 1 Preparation conditions and properties of various membranes Code
AI AII BI BII
Dope composition (%) Water
FA
N-66
0 0 12 12
75 75 63 63
25 25 25 25
Bath FA content (%)
10 40 10 40
Wettability (s) Top
Bottom
300 300 264 62
300 2 15 19
Melting point (°C)
DH (J/g)
264.0 265.8 264.7 265.5
86.4 85.8 76.1 74.0
358
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
to gelation, the dope could be cast and immersed into a nonsolvent bath to form porous membranes. Bath ‘‘I’’ was a typical harsh nonsolvent for many polymers. A Nylon 66 dope precipitated very rapidly in this bath. Bath ‘‘II’’ was a very soft precipitant. The exchange of solvent and nonsolvent proceeded slowly in it, leading to a delayed phase separation situation. 2.3. Membrane characterization Several methods were adopted to characterize the formed membranes: (1) Morphologies of the membranes including top, bottom, and cross sectional views were observed using a low voltage FESEM (Leo 1530, GMBH) at high magnifications to disclose the nano-scale fine structure. The cross section of the membrane was obtained by fracturing the membrane in liquid nitrogen. To minimize artifacts caused by metal coating, only a very thin layer (ca. 0.5–1.5 nm) of Pt–Pd alloy was plated on the sample, using a sputter coater equipped with a Quartz Crystal Microbalance thickness controller. Since the conductive layer was extremely thin, low electron acceleration voltage (e.g., 2.5 kV) could be employed so as to obtain a stable image while reducing sample damage. (2) Differential Scanning Calorimeter (DSC, model 2010, TA Instrument Ltd., USA) was used to measure the melting temperature (Tm) and heat of fusion (DHf) of both Nylon 66 pellets and membranes. The instrument was first calibrated with an indium standard before running the tests. An appropriate amount, typically 5 mg, of a dried sample was sealed in an aluminum pan and placed in the heating chamber together with an empty reference pan. The temperature was raised from 25 °C to 290 °C at a constant rate (20 °C/min) in nitrogen atmosphere. The Tm and DHf of the sample were determined from the thermograms. (3) Wide angle X-ray diffraction (D8 Advanced, Bruker, Germany) experiments were carried out to determine the crystal structure and the crystallinity of the membranes. The wave length of the X-ray radiation ˚ (copper Ka line), the width of the slit was was 1.54 A 0.6 mm, and the scanning rate was 0.05°/s with a resolution of 0.01°. The crystallinity of the membrane sample was determined by decomposition of the diffraction peaks into contributions from amorphous halo (broad) and crystalline peaks (sharp) following a method described in the literature [21]. The curve fitting scheme incorporated Gaussian with Lorenztian functions in a mixed function, Mix(Gauss + Lorentz): MixðGauss þ LorentzÞ ¼ xðGaussÞ þ ð1 xÞðLorentzÞ x ¼ fraction of Gauss In the present work, x was taken to be 0.5. Such is justifiable by the fact that the crystallinity thus determined
matched reasonably well with the crystallinity data reported in the literature (30–45%) and those from FTIR and DSC methods in the present research [17,19,20]. The areas under the amorphous and crystalline regions were calculated by appropriate integrations and the area ratio of the crystalline component to the total diffraction peak defined the crystallinity [21]. (4) Infrared absorption spectra of various membranes were taken using an Fourier Transform Infrared Spectrophotometer (Nicolet spectrometer 550. USA). For each spectrum, at least 256 scans were performed to achieve an adequate signal-to-noise ratio. All IR spectra were carried out with a built-in Ominic software and curve fitting was also carried out with the GRAMS/ AITM software using the Mix(Gauss + Lorentz) function as in the deconvolution of XRD diffractograms. (5) Water wettability of the membranes was determined using the method described by Pall [22]. A drop of water (2 ll) was deposited on the top or bottom surface of the membrane and the time needed for the drop to be completely absorbed was recorded. (6) Pure water permeation fluxes of the prepared membranes were measured with parallel-flow type cell (effective area = 7.84 cm2) at different trans-membrane pressures over the range suitable for micro-filtration.
3. Results and discussion 3.1. Dope, bath, and phase equilibrium isotherms The phase diagram at 25 °C of the ternary system, water/formic acid/Nylon 66, is shown in Fig. 1 [11]. The crystallization line and liquid–liquid demixing line are indicated. An originally dissolved dope below the
FORMIC ACID 0.00
1.00
crystallization line 0.25
A 0.75
liquid-liquid demixing line
B 0.50
0.50
II 0.75
0.25
I 1.00 0.00
WATER
0.25
0.50
0.75
0.00 1.00
NYLON-66
Fig. 1. Ternary phase diagram (weight fraction) for water– formic acid–Nylon 66 system at 25 °C.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
359
crystallization line will eventually crystallize. The region of liquid–liquid demixing is defined by the binodal. A solution within this region will separate into two liquid phases in equilibrium. Since the binodal is below the crystallization line, both types of phase separations can occur for solutions within the binodal region. During the course of precipitation, these two-phase separation events compete to yield membranes with a wide variety of porous structures. In this report, membranes are prepared using two representative dopes and baths whose compositions cover the widest possible range for Nylon membrane formations. Their compositions are shown in Fig. 1 and Table 1 as dopes A, B and baths I, II. Other immersion conditions can be considered to be the intermediate cases of the present immersion conditions. 3.2. Morphology of Nylon 66 membrane Fig. 2 presents the cross sectional views of membrane ‘‘AI’’, which was prepared by immersion–precipitation of dope ‘‘A’’ from bath ‘‘I’’. At a low magnification, Fig. 2(a) shows that the cross section contains porous domains in the polymer matrix. The pores are largely of spherical shape (ca. 3–5 lm dia.) and they are separated with limited pore–pore interconnections. Such morphological features are typical of amorphous membranes precipitated from a harsh bath such as water. The pores are known to be produced by the liquid– liquid demixing process. However, for the precipitation of a Nylon 66 casting dope, both liquid–liquid demixing and crystallization can occur (referring to the phase diagram). Fig. 2 demonstrates an immersion case wherein liquid–liquid demixing predominates, and it fixes the overall porous structure prior to crystallization, which takes place in the gel surrounding the already grownup liquid pores. It appears that the long induction time associated with polymer crystallization (nucleation and growth) gives rise to this sequence of phase separation events. Some more detailed description of the evolution of the liquid phases that eventually leads to the cellular structure can be found in the literature [7,8,11,12]. Occurrence of crystallization during the precipitation process can be demonstrated by the XRD, DSC, and SEM techniques. High resolution SEM images are shown in Fig. 2(b) and (c), whereas the XRD and DSC analyses will be discussed in later sections. Fig. 2(b) presents the image of a few cellular pores that are formed just underneath the top surface. Unlike normal amorphous membranes whose pore walls are generally smooth, the pore surfaces are corrugated containing very small voids of several tens nm. A high magnification image of the region near the top surface (marked rectangle in Fig. 2(b)) is shown in Fig. 2(c). The crystals are highly branched. In addition to sticklike crystal elements, some sheaf-like crystals (e.g., arrow S in the figure) are observed with their ends splaying
Fig. 2. SEM photomicrographs of the cross section of the Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 5000X, (b) 20,000X, (c) 150,000X.
out and twisting to form granular entities (e.g., arrow G). Crystal lamellae can also be identified (e.g., arrow L) with thickness of ca. 6–10 nm.
360
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
The top surface of membrane ‘‘AI’’, as shown in Fig. 3(a), develops into a dense nonporous skin. The fact that the interfacial polymer concentration increases sharply soon after contact with a harsh bath (water) is
responsible for the formation of the dense skin. During precipitation, a stiff gel layer was formed at the top surface which precluded the possibility of nucleation of liquid micelles in this region. However, because this layer was supersaturated with respect to crystallization (in the area below the crystallization line), it crystallized to form a skin packed by large crystal grains (10–20 lm) [7,8]. A magnified view of three neighboring crystals is shown in Fig. 3(b). The nucleation center (marked ‘‘N’’) can clearly be identified, emanating from which crystal grew outwards following the habits of spherulite until they impinged on adjacent crystals with linear boundaries. Apparently, substantial branching of crystal elements have occurred during their growth. A very high magnification image of Fig. 3(b) near point ‘‘N’’ is shown in Fig. 3(c). Stick-like crystals (200–500 nm long) are observed with a little twisted appearance and each contained several branching points. Furthermore, the central part of each polygonal grain bulged up making the crystal looked like a ‘‘hill’’ in Figs. 2(b) and 3(b). The morphology of the bottom surface of membrane ‘‘AI’’ is presented in Fig. 4. It is composed of large spherulites (ca. 5 lm) in the shape of a truncated sphere. Cellular pores are not in evidence at this surface, which is consistent with the fact that the bottom surface stayed outside of the binodal for a substantial long time, during which crystals completed their growth to form fully developed spherulites. As a result of their growth against the glass plate, the spherulites are truncated and exhibited a flattened feature in Fig. 4(a). As shown in the magnified view in Fig. 4(b), the spherulites contact each other with linear boundaries. The branching feature of the crystal is again clearly demonstrated. The width of the lamella can be measured to be ca. 7 nm at a higher magnification. 3.3. The effect of formic acid content in the bath
Fig. 3. SEM photomicrographs of the top surface of Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 1500X, (b) 20,000X, (c) 200,000X.
When precipitation was carried out in a soft bath that contained a high solvent concentration, the rate of solvent–nonsolvent exchange across the membrane– bath interface could be significantly lowered. As a result, the diffusion path of the casting solution approached the binodal very slowly. In that case, crystallization may occur together with (or even earlier) than liquid–liquid demixing to yield a morphology bearing characteristics from both types of phase separation mechanisms. One such example is demonstrated in Fig. 5, for which the membrane was prepared by immersion of dope ‘‘A’’ in bath ‘‘II’’. Fig. 5(a) shows the cross sectional morphology, of which the upper half is composed of cellular pores and defected spherulites, and the lower half is dominated by large full spherulites (ca. 10 lm). A magnified view of the upper half region is shown in Fig. 5(b). Interestingly, liquid–liquid demixing left its imprint as large holes on the surfaces of the spherulites. It was
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
361
Fig. 4. SEM photomicrographs of the bottom surface of Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 1500X, (b) 20,000X.
possible that liquid–liquid demixing occurred first to form micelles and soon after crystallization took place around the cell walls. On the other hand, crystallization might occur first and during spherulitic growth liquid– liquid demixing commenced nearby. The growing liquid micelles pushed aside the interfacial soft gel, which afterward merged into the growing spherulites. In either case, it can be certain that the two types of phase separations competed intensively during the precipitation process. The surfaces of the spherulites are very porous, as clearly manifested in the high magnification image in Fig. 5(c) (region near point ‘‘A’’ in Fig. 5 (b)). The pores (e.g., arrow ‘‘P’’) are very small (ca. 80–100 nm) and are thought to be produced by impurity rejection (solvent and nonsolvent) activities of growing crystals. From central towards the bottom of Fig. 5(c) shows a large piece of crystal element that demonstrates interesting features such as winding, bending, branching, and splaying that sometimes observed in an intermediate-stage spherulite.
Fig. 5. SEM photomicrographs of the cross section of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 1000X, (b) 10,000X, (c) 200,000X.
The top surface of membrane ‘‘AII’’ is shown in Fig. 6. Unlike the corrugated surface shown in Fig. 2(b), this surface is relatively flat. Although it has the appearance of a skin, it is not as dense and there
362
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
Fig. 6. SEM photomicrographs of the top surface of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 5000X, (b) 50,000X.
can be observed random crevices between crystal grains; some of them even become large holes (e.g., 3 lm). At a magnified view, Fig. 6(b) exhibits the fine structure of a few crystal grains. The polygonal shape, linear boundary, and dendritic lamellae are typical of a planar spherulite in polyamide films. This implies that the crystal nuclei were created and grown two-dimensionally in the gel layer just underneath the top surface where liquid–liquid phase separation was prohibited by the interfacial boundary conditions [7,8,12]. The sizes of the crystals on this top surface (ca. 3–5 lm) are considerably smaller than that on membrane ‘‘AI’’ (cf. Fig. 3). This is because the top gel layer that contacted the 40% formic acid bath was soft and thus more nuclei could be generated. The bottom surface of the membrane, as shown in Fig. 7, is composed of well-developed yet truncated spherulites, as in the bottom surface of membrane ÔAIÕ. Fig. 7(b) shows the image of a spheruliteÕs surface, form which it can be seen that ribbon-like lamellae are curled and branched with small voids distributed among them.
Fig. 7. SEM photomicrographs of the bottom surface of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 1500X, (b) 100,000X.
3.4. The effect of water content in the dope As an incipient casting solution, such as dope ‘‘B’’, was employed, crystallization dominated the precipitation process to yield a membrane packed by nearly equal-sized crystallites. Two such cases were studied, viz. membranes ‘‘BI’’ and ‘‘BII’’. The SEM photomicrographs of membrane ‘‘BI’’ are illustrated in Fig. 8. It can be seen that the cross section is very uniform without any large cellular pores. Fig. 8(b) shows the sheaf-like crystals (1.5 lm) consisting the cross section, which interlock to form a continuous network coexisting with the porous channels. The voids are interconnected to form numerous channels within the polymer matrix. Compared with the membrane prepared by dope ‘‘A’’ immersed in the same bath (cf. Fig. 2), it is clear that the relative level of crystallization vs. liquid–liquid phase separation depends strongly on the dope condition. Dope ‘‘A’’ was a true solution in which polymer chains
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
363
Fig. 8. SEM photomicrographs of membrane ‘‘BI’’ prepared by immersion of dope B in bath I. (a) Cross section 1000X, (b) cross section 150,000X, (c) top surface 100,000X, (d) bottom surface 50,000X.
were well solvated by formic acid through hydrogen bonding [23]. Conversely, dope ÔBÕ was in a metastable state with respect to crystallization in which the polymer coils were on the edge of collapsing. That is, they may have associated into an aggregated structure, and thus could crystallize upon slight concentration fluctuation rapidly after immersion of the dope into the bath. The top surface of membrane ÔBIÕ is shown in Fig. 8(c). It exhibits a skin-like structure that is derived from the top gel layer contacting the harsh bath, water. The boundaries between the polygonal spherulites are not as well-defined as those in membrane ÔAIÕ (Fig. 3(a) and (b)), and some of them became troughs (20–50 nm) between crystal grains. The size of the spherulites are quite small, ca. 20–30 times smaller than those in Fig. 3(a), implying that there was a much higher nucleation density on this top surface than on membrane ÔAIÕ, which is consistent with the dopeÕs aggregation state for making these membranes. The fine structure of the bottom surface is shown in Fig. 8(d).
The stick-like crystallites with their ends slightly branched, i.e., preliminary spherulites, are in evidence, in contrast to the large truncated full spherulites in membrane AI (Fig. 4). The pores here are very small (ca. 100–300 nm), which is associated with the flatten effect of the crystallites at the bottom surface. Membrane ÔBIIÕ was prepared by immersion of an incipient dope (‘‘B’’) in a soft bath (‘‘II’’). Fig. 9(a) shows the cross section of the membrane. It is as uniform as membrane ‘‘BI’’. The high magnification image of this cross section resembles Fig. 8(b) also consisting of sheaf-like spherulites, and thus is not shown here. The top surface of membrane ‘‘BII’’ is shown in Fig. 9(b). Unlike membrane ‘‘BI’’ shown in Fig. 8(c), it is very porous and the shape of the crystals resemble those in membraneÕs cross section. That is, membrane ‘‘BII’’ is skinless symmetric membrane. The water flux data of membrane ‘‘BII’’ are shown in Fig. 10 together with that of membrane ‘‘BI’’. The trans-membrane pressure was set to the range, 0.2–1.0 bar, for common micro-filtration operations. As is expected, water fluxes
364
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
face, its water flux is much lower than that of membrane ‘‘BII’’. With a uniform microporous structure and relatively high water flux, membrane ‘‘BII’’ can be applied to aqueous micro-filtration applications. 3.5. Crystallinity and thermal behavior Semicrystalline Nylon 66 is usually regarded as a two-phase system consisting of crystalline and amorphous regions. This is also the case for porous Nylon 66 membranes. Three different methods, XRD, FTIR, and DSC were adopted to examine the crystalline characters (crystal type, crystallinity, thermal behavior) of the membranes, in addition to the morphological studies shown previously. The XRD patterns of the four membranes (AI, AII, BI, BII) are shown in Fig. 11. For each pattern, two strong reflection peaks at 2h equal to 20.12° and 23.94° and a weak peak at 13.41° are observed, which correspond to the (1 0 0), (0 1 0, 1 1 0) doublet, and (0 0 2) planes of a-type Nylon 66 crystals oriented in a triclinic unit cell [23–25]. The diffraction pattern can be decomposed into a broad amorphous halo and sharp peaks from reflections of the crystalline regions by a curve fitting technique, as described in the experimental section. As an example, the decomposed results of membrane ÔBIIÕ is shown in Fig. 11 (the two solid curves under pattern ‘‘BII’’), from which the crystallinity was calculated to be 30.1%, a value close to that of Nylon 66 film or fiber reported in the literature [17,19,23,26]. The crystallinity data of the other membranes are listed in Table 2, as were obtained from the same decomposition procedure. It is interesting to find Fig. 9. SEM photomicrographs of membrane ‘‘BII’’ prepared by immersion of dope B in bath II. (a) Cross section 3000X, (b) top surface 100,000X.
(100)
Water flux (g/cm2 min)
0.5
BI BII
(010) (110)
0.4
0.3
(002)
AI
0.2
AII 0.1
BI 0.0 0.2
0.4
0.6
0.8
BII
1.0
∆P (bar)
Fig. 10. Pure water fluxes of membranes ‘‘BI’’ and ‘‘BII’’.
12
14
16
18
20
22
24
26
28
2 Theta (deg.)
of both membranes increase with increasing pressure. However, because membrane ‘‘BI’’ has a denser top sur-
Fig. 11. X-ray (Cu Ka, k = 0.154056 nm) diffraction spectrum of Nylon 66 membranes.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
that the crystallinities of the membranes prepared from dope ÔAÕ are higher than those from dope B (viz., membrane AII AI, BI BII). Given the fact that the morphologies of membranes ÔAIÕ and ÔAIIÕ were largely derived from liquid–liquid demixing, it can be certain that crystallization has occurred in the soft gel-phase after the fast liquid–liquid demixing process. Some evidences have been demonstrated earlier as dendritic crystal elements on the walls of the cellular pores. IR absorption spectra of the membranes were also used to determine the relative levels of crystalline and amorphous phases in the membranes [17,19,23,26]. For a-type Nylon-66 crystals, the relation between crys-
Table 2 Comparison of the crystallinity values obtained from three different methods Code
Degree of crystallinity (Xc) DSC
XRD
FTIR
AI AII BI BII
43.9 43.5 38.6 37.5
38.3 39.1 30.6 30.1
39.8 42.1 36.7 34.5
1200
1190
1180
1170
1160
1180cm-1
1210
Wavenumbers (cm-1) BI
1190
1180
Wavenumbers (cm-1)
1190
1180
1170
1160
BII
1199cm-1
Absorbance
Absorbance
1180cm-1
1200
1200
Wavenumbers (cm-1)
1199cm-1
1210
AII
1199cm-1
Absorbance
Absorbance
1180cm-1
1210
tallinity and dichroic ratio, R = A119/A1180, are available from the literature, where A119 and A1180 are the areas of the integrated absorbances at bands 1199 and 1180 cm1, respectively. The band at 1199 cm1 is a characteristic of the a-crystalline phase whereas the band at 1180 cm1 represents the amorphous component [23]. Curve fitting of the infrared spectra in the range between 1160 and 1215 cm1 for Nylon 66 membranes are shown in Fig. 12. The calculated crystallinities for various membranes are summarized in Table 2. Obviously, the FTIR results for membranes ÔAIÕ and ÔAIIÕ are close to those from XRD measurements; yet, for membranes ÔBIÕ and ÔBIIÕ, the FTIR results are somewhat larger than those from XRD. However, both techniques indicate that membranes ÔAIÕ and ÔAIIÕ have higher crystallinities than membranes ÔBIÕ and ÔBIIÕ. DSC thermal analyses of various membranes were performed and the results are shown in Fig. 13. The scanning rate was 20 °C/min. All thermograms exhibit a major melting peak typical of the melting of normal Nylon 66 crystal and a less significant peak or shoulder at a lower temperature representing the melting of small less stable crystalline units. The measured major melting peak temperatures (Tm), as summarized in Table 1, are all close to 265 °C, a value typical of Nylon 66 polymer
AI
1199cm-1
365
1170
1160
1180cm-1
1210
1200
1190
1180
1170
Wavenumbers (cm-1)
Fig. 12. Curve fitting of the infrared spectra in the range between 1160 and 1215 cm1.
1160
366
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
Fig. 13. DSC thermograms of the Nylon 66 membranes.
determined by thermal analysis [17,23]. The heat of fusion (DHf) were determined and apparent crystallinities were calculated using DHf0 (197 J/g) of perfect crystalline Nylon 66 taken from literature [17]. Although the double peak in the thermograms suggest a high possibility of an irreversible melting process involving re-crystallization, the calculated crystallinities, as presented in Table 2, are largely in agreement with the results from XRD and FTIR measurements. This implies a low level re-crystallization during thermal analysis or the heat absorbed by re-crystallization was compensated by the heat released during melting of the formed crystals. This, however, require further investigation and is beyond the scope of the present work.
4. Conclusions In this article, nano-scale morphology and crystalline characters of Nylon 66 membranes prepared by immersion–precipitation of two representative dopes in both harsh and soft baths are discussed. During the precipitation process, both liquid–liquid demixing and crystallization occurred and these phase separation mechanisms competed to yield structures which reflected their level of influences. When a well-dissolved casting dope (i.e., dope ÔAÕ) was adopted, liquid–liquid demixing was important regardless of the bath type, and the formed membrane demonstrated a porous morphology contained many cellular pores. In this case, crystallization might either leave its imprint on the pore wall in terms of dendritic crystal elements, such as in membrane ‘‘AI’’, or develop into spherulites in specific regions of the membrane, such as the lower half of the cross section of membrane ‘‘AII’’. The fine structure of a full spherulite indicated a porous surface consisting of twisted
lamellae (<10 nm thickness). When an incipient dope containing a high population of pre-nucleation aggregates (e.g., dope ‘‘B’’) was employed, crystallization dominated the precipitation process, and a membrane with a uniform porous cross section was produced. The crystallites have a sheaf-like or stick-like appearance, being typical of early stage spherulites. If the bath was harsh, the formed membrane still had a skin (‘‘BI’’); yet, if the bath was soft, the formed membrane (‘‘BII’’) became skinless. This membrane was a promising candidate for micro-filtration applications. XRD and DSC analyses of the membranes showed that the crystals in the membranes were of a-type with a melting temperature of ca. 265 °C. The crystallinities of the formed membranes, as determined by FTIR, XRD, and DSC analyses, were over the range 30–45% consistent with the value of Nylon 66 polymer reported in the literature.
References [1] Tohgo K, Fukuhara D, Hadano A. Compos Sci Technol 2001;61:1005. [2] Albano C, Sciamanna R, Gonzalez R, Papa R, Navarro O. Eur Polym J 2001;37:851. [3] Murthy NS, Wang ZG, Akkapeddi MK, Hsiao BS. Polymer 2002;43:4905. [4] Persson A, Jonsson AS, Zacchi G. J Membr Sci 2003;223:11. [5] Castilho LR, Anspach FB, Deckwer WD. J Membr Sci 2002;207:253. [6] Mulder M. Basic principles of membrane technology. Dordrecht: Kluwer Academic Publisher; 1991. [7] Shih CH, Gryte CC, Cheng LP. J Appl Polym Sci 2005;96:944. [8] Cheng LP, Dwan AW, Gryte CC. J Polym Sci: Polym Phys 1995;33:211. [9] Bulte AMW, Mulder MHV, Smolders CA, Strathmann H. J Membr Sci 1996;121:37. [10] Bulte AMW, Mulder MHV, Smolders CA, Strathmann H. J Membr Sci 1996;121:51. [11] Cheng LP, Dwan AH, Gryte CC. J Polym Sci: Polym Phys 1994;32:1183. [12] Cheng LP, Dwan AW, Gryte CC. J Polym Sci: Polym Phys 1995;33:223. [13] Thomas JL, Olzog M, Drake C, Shih CH, Gryte CC. Polymer 2002;43:4153. [14] Kho YW, Kalika DS, Knutson BL. Polymer 2001;42:6119. [15] Young TH, Huang JH, Chuang WY. Eur Polym J 2002;38:63. [16] Men Y, Rieger J. Eur Polym J 2004;40:2629. [17] Elzein T, Brogly M, Schultz J. Polymer 2002;43:4811. [18] Fornes TD, Paul DR. Polymer 2003;44:3945. [19] Vasanthan N, Salem DR. J Membr Sci 2000;38:516. [20] Thanki PN, Ramesh C, Singh RP. Polymer 2001;42: 535. [21] Park YS, Hatae T, Itoh H, Jang MY, Yamazaki Y. Electrochim Acta 2004;50:595.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367 [22] Pall DJ. US Patent 1982; 4,340,479. [23] Kohan MI. Nylon plastics handbook. New York: Hanser/ Gardner; 1995. p. 89. [24] Klein N, Marom G, Wachtel E. Polymer 1996;37:5493.
367
[25] Zhang GZ, Yoshida H, Kawai T. Thermochim Acta 2004;416:79. [26] Vasanthan N, Ruetsch SB, Salem DR. J Polym Sci: Polym Phys 2002;40:1940.
European Polymer Journal 42 (2006) 356–367
www.elsevier.com/locate/europolj
Fine structure and crystallinity of porous Nylon 66 membranes prepared by phase inversion in the water/formic acid/Nylon 66 system Dar-Jong Lin, Chi-Lin Chang, Chih-Kang Lee, Liao-Ping Cheng
*
Department of Chemical and Materials Engineering, Tamkang University, Taipei 25137, Taiwan, ROC Received 3 May 2005; received in revised form 4 July 2005; accepted 13 July 2005 Available online 24 August 2005
Abstract Microporous Nylon-66 polymeric membranes were prepared by immersion–precipitation from a ternary system, water/formic acid/Nylon 66. Depending upon the precipitation conditions, membranes with morphologies that reflect the sequence of liquid–liquid demixing (as characterized by cellular pores) and crystallization (as characterized by crystal particles) events during the course of precipitation were obtained. The details of the membrane morphologies were disclosed using a low voltage field emission scanning electron microscope (FESEM) at very high resolutions. In particular, nano-scale fine structures such as dendritic crystal elements, nano-pores, nano-grains, branching lamellae, etc., which were rarely presented in the membrane literature. Wide angle X-ray diffraction analyses indicated that Nylon66 crystallized into ÔaÕ structure in all prepared membranes. Crystallinities were determined by appropriate deconvolution of the diffraction peaks. The results indicated that membranes prepared by a well-dissolved casting dope had a somewhat higher crystallinity than those prepared by incipient dopes being in metastable states with respect to crystallization. This observation was confirmed by Fourier transform infrared spectroscopy and DSC thermo analyses. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Polyamide; Immersion–precipitation; Cellular pores; Spherulites; Crystallinity
1. Introduction Nylon 66, a polyamide derived from 1,6-hexamethylene diamine and adipic acid, is a semicrystalline polymer, which possesses good thermal stability and mechanical strength, and is considered to be an important engineering thermoplastic [1–3]. Porous polyamide membrane has been commercialized for many years
*
Corresponding author. E-mail address: [email protected] (L.-P. Cheng).
and is nowadays widely used in fine-separation processes [4,5]. Microporous membranes are often manufactured by the so-called immersion–precipitation process [6], in which a polymer solution is cast on a substrate and then immersed in a nonsolvent bath to induce polymer precipitation by means of crystallization and/or liquid– liquid demixing. Unlike a nonporous Nylon 66 film, which has a water contact angle of ca. 60°, a skinless microporous Nylon 66 membrane is water wettable; i.e., water drops can penetrate into the membrane matrix within a few seconds. This property is associated with the porous morphology of the membrane. It can be
0014-3057/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.eurpolymj.2005.07.007
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
produced by a fine-tuned immersion–precipitation process, which involves precipitation of an incipient casting dope in a soft coagulation bath [7,8]. Because the casting dope is metastable with respect to crystallization and the bath contains a large amount of solvent, crystallization dominates over liquid–liquid demixing to yield a structure consists of interlinked stick-like crystalline particles. The formation mechanism of crystallizable polyamide membranes has been investigated by a number of authors. On the basis of phase diagram and mass transfer modeling, Bulte et al. obtained the initial diffusion paths and concentration profiles for various immersion conditions for the preparation of Nylon 46 membranes. These calculated results correlated well with the membrane morphologies [9,10]. Cheng et al. took into consideration the effects of dopeÕs aggregation state (e.g., a supersaturated dope may be prepared which contains a high population of pre-nucleation embryo) and bath strength (e.g., a soft bath containing significant solvent may be used) on the relative level of liquid–liquid demixing and crystallization during the course of precipitation [8,11,12]. A mathematical model was developed to illustrate the concentration evolution of nonsolvent, solvent, and polymer within the membrane prior to phase separation. In conjunction with the phase diagram, these results help to understand the relation between morphology and immersion conditions. Recently, Thomas et al. used confocal microscope to observe scattering profiles of casting dopes during immersion–precipitation, and found that precipitation occurred initially at some distance from the membrane–bath interface, which then propagates both upward and downward through the membrane [13]. Techniques other than immersion– precipitation have also been used to prepare porous polyamide membranes. For example, Kho et al. introduced compressed carbon dioxide as an antisolvent to precipitate Nylon 6 from 2,2,2-trifluoroethanol solutions to form porous membranes. In this way, pressure became an additional effective variable for tailoring the membrane microstructure [14]. Despite that Nylon-66 membrane formation mechanism has been discussed previously [8,11,12,15], the issues concerning nano-scale fine structure of the membrane such as twisted lamellae, nano-grains, etc., and thermal behavior and crystallinity of the membrane
357
have never been presented. These characteristics were investigated using low voltage field emission scanning electron microscopy (LVFESEM, fine structure), differential scanning calorimetry (DSC, thermal behavior and crystallinity), Fourier transform infrared spectroscopy (FTIR, crystallinity) and wide angle X-ray diffraction (WAXD, crystallinity and crystal structure) [16–18]. In particular, a spectrum deconvolution technique based on Gauss and Lorentzian functions were employed to analyze both WAXS and IR spectra, and to determine the degree of crystallinity of the membrane. The results were found to be consistent with those reported in the literature [17,19,20].
2. Experimental 2.1. Materials Poly(hexamethylene adipamide) (Nylon 66, Zytel 101, Du Pont, measured intrinsic viscosity = 2.68 g/dl, Mv = 87,000 g/mole) was received in pellet form. Formic acid (Acros, 99%) was used as the solvent and distilled–deionized water was used as the nonsolvent for Nylon 66. All materials were used as received. 2.2. Membrane formation A dope solution was uniformly spread on a glass surface using a casting knife. It was then immersed at 25 °C into a coagulation bath composed of water and formic acid. After precipitation has completed (typically a few min), a white solid membrane was obtained. The nonsolvent and the residual solvent in the nascent membrane were removed by a sequence of washing steps. Finally, the formed membrane was held tightly in a press between two sheets of filter papers and then dried at 40 °C. The immersion conditions for various membranes are summarized in Table 1. Dope ‘‘A’’ was a homogeneous solution consisting only of polymer and solvent. Dope ‘‘B’’ contained a substantial amount of nonsolvent. It was in a supersaturated state with respect to crystallization (i.e., its composition is located below the crystallization line) and would gel upon standing at 25 °C for an extended period of time. However, prior
Table 1 Preparation conditions and properties of various membranes Code
AI AII BI BII
Dope composition (%) Water
FA
N-66
0 0 12 12
75 75 63 63
25 25 25 25
Bath FA content (%)
10 40 10 40
Wettability (s) Top
Bottom
300 300 264 62
300 2 15 19
Melting point (°C)
DH (J/g)
264.0 265.8 264.7 265.5
86.4 85.8 76.1 74.0
358
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
to gelation, the dope could be cast and immersed into a nonsolvent bath to form porous membranes. Bath ‘‘I’’ was a typical harsh nonsolvent for many polymers. A Nylon 66 dope precipitated very rapidly in this bath. Bath ‘‘II’’ was a very soft precipitant. The exchange of solvent and nonsolvent proceeded slowly in it, leading to a delayed phase separation situation. 2.3. Membrane characterization Several methods were adopted to characterize the formed membranes: (1) Morphologies of the membranes including top, bottom, and cross sectional views were observed using a low voltage FESEM (Leo 1530, GMBH) at high magnifications to disclose the nano-scale fine structure. The cross section of the membrane was obtained by fracturing the membrane in liquid nitrogen. To minimize artifacts caused by metal coating, only a very thin layer (ca. 0.5–1.5 nm) of Pt–Pd alloy was plated on the sample, using a sputter coater equipped with a Quartz Crystal Microbalance thickness controller. Since the conductive layer was extremely thin, low electron acceleration voltage (e.g., 2.5 kV) could be employed so as to obtain a stable image while reducing sample damage. (2) Differential Scanning Calorimeter (DSC, model 2010, TA Instrument Ltd., USA) was used to measure the melting temperature (Tm) and heat of fusion (DHf) of both Nylon 66 pellets and membranes. The instrument was first calibrated with an indium standard before running the tests. An appropriate amount, typically 5 mg, of a dried sample was sealed in an aluminum pan and placed in the heating chamber together with an empty reference pan. The temperature was raised from 25 °C to 290 °C at a constant rate (20 °C/min) in nitrogen atmosphere. The Tm and DHf of the sample were determined from the thermograms. (3) Wide angle X-ray diffraction (D8 Advanced, Bruker, Germany) experiments were carried out to determine the crystal structure and the crystallinity of the membranes. The wave length of the X-ray radiation ˚ (copper Ka line), the width of the slit was was 1.54 A 0.6 mm, and the scanning rate was 0.05°/s with a resolution of 0.01°. The crystallinity of the membrane sample was determined by decomposition of the diffraction peaks into contributions from amorphous halo (broad) and crystalline peaks (sharp) following a method described in the literature [21]. The curve fitting scheme incorporated Gaussian with Lorenztian functions in a mixed function, Mix(Gauss + Lorentz): MixðGauss þ LorentzÞ ¼ xðGaussÞ þ ð1 xÞðLorentzÞ x ¼ fraction of Gauss In the present work, x was taken to be 0.5. Such is justifiable by the fact that the crystallinity thus determined
matched reasonably well with the crystallinity data reported in the literature (30–45%) and those from FTIR and DSC methods in the present research [17,19,20]. The areas under the amorphous and crystalline regions were calculated by appropriate integrations and the area ratio of the crystalline component to the total diffraction peak defined the crystallinity [21]. (4) Infrared absorption spectra of various membranes were taken using an Fourier Transform Infrared Spectrophotometer (Nicolet spectrometer 550. USA). For each spectrum, at least 256 scans were performed to achieve an adequate signal-to-noise ratio. All IR spectra were carried out with a built-in Ominic software and curve fitting was also carried out with the GRAMS/ AITM software using the Mix(Gauss + Lorentz) function as in the deconvolution of XRD diffractograms. (5) Water wettability of the membranes was determined using the method described by Pall [22]. A drop of water (2 ll) was deposited on the top or bottom surface of the membrane and the time needed for the drop to be completely absorbed was recorded. (6) Pure water permeation fluxes of the prepared membranes were measured with parallel-flow type cell (effective area = 7.84 cm2) at different trans-membrane pressures over the range suitable for micro-filtration.
3. Results and discussion 3.1. Dope, bath, and phase equilibrium isotherms The phase diagram at 25 °C of the ternary system, water/formic acid/Nylon 66, is shown in Fig. 1 [11]. The crystallization line and liquid–liquid demixing line are indicated. An originally dissolved dope below the
FORMIC ACID 0.00
1.00
crystallization line 0.25
A 0.75
liquid-liquid demixing line
B 0.50
0.50
II 0.75
0.25
I 1.00 0.00
WATER
0.25
0.50
0.75
0.00 1.00
NYLON-66
Fig. 1. Ternary phase diagram (weight fraction) for water– formic acid–Nylon 66 system at 25 °C.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
359
crystallization line will eventually crystallize. The region of liquid–liquid demixing is defined by the binodal. A solution within this region will separate into two liquid phases in equilibrium. Since the binodal is below the crystallization line, both types of phase separations can occur for solutions within the binodal region. During the course of precipitation, these two-phase separation events compete to yield membranes with a wide variety of porous structures. In this report, membranes are prepared using two representative dopes and baths whose compositions cover the widest possible range for Nylon membrane formations. Their compositions are shown in Fig. 1 and Table 1 as dopes A, B and baths I, II. Other immersion conditions can be considered to be the intermediate cases of the present immersion conditions. 3.2. Morphology of Nylon 66 membrane Fig. 2 presents the cross sectional views of membrane ‘‘AI’’, which was prepared by immersion–precipitation of dope ‘‘A’’ from bath ‘‘I’’. At a low magnification, Fig. 2(a) shows that the cross section contains porous domains in the polymer matrix. The pores are largely of spherical shape (ca. 3–5 lm dia.) and they are separated with limited pore–pore interconnections. Such morphological features are typical of amorphous membranes precipitated from a harsh bath such as water. The pores are known to be produced by the liquid– liquid demixing process. However, for the precipitation of a Nylon 66 casting dope, both liquid–liquid demixing and crystallization can occur (referring to the phase diagram). Fig. 2 demonstrates an immersion case wherein liquid–liquid demixing predominates, and it fixes the overall porous structure prior to crystallization, which takes place in the gel surrounding the already grownup liquid pores. It appears that the long induction time associated with polymer crystallization (nucleation and growth) gives rise to this sequence of phase separation events. Some more detailed description of the evolution of the liquid phases that eventually leads to the cellular structure can be found in the literature [7,8,11,12]. Occurrence of crystallization during the precipitation process can be demonstrated by the XRD, DSC, and SEM techniques. High resolution SEM images are shown in Fig. 2(b) and (c), whereas the XRD and DSC analyses will be discussed in later sections. Fig. 2(b) presents the image of a few cellular pores that are formed just underneath the top surface. Unlike normal amorphous membranes whose pore walls are generally smooth, the pore surfaces are corrugated containing very small voids of several tens nm. A high magnification image of the region near the top surface (marked rectangle in Fig. 2(b)) is shown in Fig. 2(c). The crystals are highly branched. In addition to sticklike crystal elements, some sheaf-like crystals (e.g., arrow S in the figure) are observed with their ends splaying
Fig. 2. SEM photomicrographs of the cross section of the Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 5000X, (b) 20,000X, (c) 150,000X.
out and twisting to form granular entities (e.g., arrow G). Crystal lamellae can also be identified (e.g., arrow L) with thickness of ca. 6–10 nm.
360
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
The top surface of membrane ‘‘AI’’, as shown in Fig. 3(a), develops into a dense nonporous skin. The fact that the interfacial polymer concentration increases sharply soon after contact with a harsh bath (water) is
responsible for the formation of the dense skin. During precipitation, a stiff gel layer was formed at the top surface which precluded the possibility of nucleation of liquid micelles in this region. However, because this layer was supersaturated with respect to crystallization (in the area below the crystallization line), it crystallized to form a skin packed by large crystal grains (10–20 lm) [7,8]. A magnified view of three neighboring crystals is shown in Fig. 3(b). The nucleation center (marked ‘‘N’’) can clearly be identified, emanating from which crystal grew outwards following the habits of spherulite until they impinged on adjacent crystals with linear boundaries. Apparently, substantial branching of crystal elements have occurred during their growth. A very high magnification image of Fig. 3(b) near point ‘‘N’’ is shown in Fig. 3(c). Stick-like crystals (200–500 nm long) are observed with a little twisted appearance and each contained several branching points. Furthermore, the central part of each polygonal grain bulged up making the crystal looked like a ‘‘hill’’ in Figs. 2(b) and 3(b). The morphology of the bottom surface of membrane ‘‘AI’’ is presented in Fig. 4. It is composed of large spherulites (ca. 5 lm) in the shape of a truncated sphere. Cellular pores are not in evidence at this surface, which is consistent with the fact that the bottom surface stayed outside of the binodal for a substantial long time, during which crystals completed their growth to form fully developed spherulites. As a result of their growth against the glass plate, the spherulites are truncated and exhibited a flattened feature in Fig. 4(a). As shown in the magnified view in Fig. 4(b), the spherulites contact each other with linear boundaries. The branching feature of the crystal is again clearly demonstrated. The width of the lamella can be measured to be ca. 7 nm at a higher magnification. 3.3. The effect of formic acid content in the bath
Fig. 3. SEM photomicrographs of the top surface of Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 1500X, (b) 20,000X, (c) 200,000X.
When precipitation was carried out in a soft bath that contained a high solvent concentration, the rate of solvent–nonsolvent exchange across the membrane– bath interface could be significantly lowered. As a result, the diffusion path of the casting solution approached the binodal very slowly. In that case, crystallization may occur together with (or even earlier) than liquid–liquid demixing to yield a morphology bearing characteristics from both types of phase separation mechanisms. One such example is demonstrated in Fig. 5, for which the membrane was prepared by immersion of dope ‘‘A’’ in bath ‘‘II’’. Fig. 5(a) shows the cross sectional morphology, of which the upper half is composed of cellular pores and defected spherulites, and the lower half is dominated by large full spherulites (ca. 10 lm). A magnified view of the upper half region is shown in Fig. 5(b). Interestingly, liquid–liquid demixing left its imprint as large holes on the surfaces of the spherulites. It was
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
361
Fig. 4. SEM photomicrographs of the bottom surface of Nylon 66 membrane (AI) prepared by immersion of dope A in bath I. (a) 1500X, (b) 20,000X.
possible that liquid–liquid demixing occurred first to form micelles and soon after crystallization took place around the cell walls. On the other hand, crystallization might occur first and during spherulitic growth liquid– liquid demixing commenced nearby. The growing liquid micelles pushed aside the interfacial soft gel, which afterward merged into the growing spherulites. In either case, it can be certain that the two types of phase separations competed intensively during the precipitation process. The surfaces of the spherulites are very porous, as clearly manifested in the high magnification image in Fig. 5(c) (region near point ‘‘A’’ in Fig. 5 (b)). The pores (e.g., arrow ‘‘P’’) are very small (ca. 80–100 nm) and are thought to be produced by impurity rejection (solvent and nonsolvent) activities of growing crystals. From central towards the bottom of Fig. 5(c) shows a large piece of crystal element that demonstrates interesting features such as winding, bending, branching, and splaying that sometimes observed in an intermediate-stage spherulite.
Fig. 5. SEM photomicrographs of the cross section of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 1000X, (b) 10,000X, (c) 200,000X.
The top surface of membrane ‘‘AII’’ is shown in Fig. 6. Unlike the corrugated surface shown in Fig. 2(b), this surface is relatively flat. Although it has the appearance of a skin, it is not as dense and there
362
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
Fig. 6. SEM photomicrographs of the top surface of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 5000X, (b) 50,000X.
can be observed random crevices between crystal grains; some of them even become large holes (e.g., 3 lm). At a magnified view, Fig. 6(b) exhibits the fine structure of a few crystal grains. The polygonal shape, linear boundary, and dendritic lamellae are typical of a planar spherulite in polyamide films. This implies that the crystal nuclei were created and grown two-dimensionally in the gel layer just underneath the top surface where liquid–liquid phase separation was prohibited by the interfacial boundary conditions [7,8,12]. The sizes of the crystals on this top surface (ca. 3–5 lm) are considerably smaller than that on membrane ‘‘AI’’ (cf. Fig. 3). This is because the top gel layer that contacted the 40% formic acid bath was soft and thus more nuclei could be generated. The bottom surface of the membrane, as shown in Fig. 7, is composed of well-developed yet truncated spherulites, as in the bottom surface of membrane ÔAIÕ. Fig. 7(b) shows the image of a spheruliteÕs surface, form which it can be seen that ribbon-like lamellae are curled and branched with small voids distributed among them.
Fig. 7. SEM photomicrographs of the bottom surface of Nylon 66 membrane (AII) prepared by immersion of dope A in bath II. (a) 1500X, (b) 100,000X.
3.4. The effect of water content in the dope As an incipient casting solution, such as dope ‘‘B’’, was employed, crystallization dominated the precipitation process to yield a membrane packed by nearly equal-sized crystallites. Two such cases were studied, viz. membranes ‘‘BI’’ and ‘‘BII’’. The SEM photomicrographs of membrane ‘‘BI’’ are illustrated in Fig. 8. It can be seen that the cross section is very uniform without any large cellular pores. Fig. 8(b) shows the sheaf-like crystals (1.5 lm) consisting the cross section, which interlock to form a continuous network coexisting with the porous channels. The voids are interconnected to form numerous channels within the polymer matrix. Compared with the membrane prepared by dope ‘‘A’’ immersed in the same bath (cf. Fig. 2), it is clear that the relative level of crystallization vs. liquid–liquid phase separation depends strongly on the dope condition. Dope ‘‘A’’ was a true solution in which polymer chains
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
363
Fig. 8. SEM photomicrographs of membrane ‘‘BI’’ prepared by immersion of dope B in bath I. (a) Cross section 1000X, (b) cross section 150,000X, (c) top surface 100,000X, (d) bottom surface 50,000X.
were well solvated by formic acid through hydrogen bonding [23]. Conversely, dope ÔBÕ was in a metastable state with respect to crystallization in which the polymer coils were on the edge of collapsing. That is, they may have associated into an aggregated structure, and thus could crystallize upon slight concentration fluctuation rapidly after immersion of the dope into the bath. The top surface of membrane ÔBIÕ is shown in Fig. 8(c). It exhibits a skin-like structure that is derived from the top gel layer contacting the harsh bath, water. The boundaries between the polygonal spherulites are not as well-defined as those in membrane ÔAIÕ (Fig. 3(a) and (b)), and some of them became troughs (20–50 nm) between crystal grains. The size of the spherulites are quite small, ca. 20–30 times smaller than those in Fig. 3(a), implying that there was a much higher nucleation density on this top surface than on membrane ÔAIÕ, which is consistent with the dopeÕs aggregation state for making these membranes. The fine structure of the bottom surface is shown in Fig. 8(d).
The stick-like crystallites with their ends slightly branched, i.e., preliminary spherulites, are in evidence, in contrast to the large truncated full spherulites in membrane AI (Fig. 4). The pores here are very small (ca. 100–300 nm), which is associated with the flatten effect of the crystallites at the bottom surface. Membrane ÔBIIÕ was prepared by immersion of an incipient dope (‘‘B’’) in a soft bath (‘‘II’’). Fig. 9(a) shows the cross section of the membrane. It is as uniform as membrane ‘‘BI’’. The high magnification image of this cross section resembles Fig. 8(b) also consisting of sheaf-like spherulites, and thus is not shown here. The top surface of membrane ‘‘BII’’ is shown in Fig. 9(b). Unlike membrane ‘‘BI’’ shown in Fig. 8(c), it is very porous and the shape of the crystals resemble those in membraneÕs cross section. That is, membrane ‘‘BII’’ is skinless symmetric membrane. The water flux data of membrane ‘‘BII’’ are shown in Fig. 10 together with that of membrane ‘‘BI’’. The trans-membrane pressure was set to the range, 0.2–1.0 bar, for common micro-filtration operations. As is expected, water fluxes
364
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
face, its water flux is much lower than that of membrane ‘‘BII’’. With a uniform microporous structure and relatively high water flux, membrane ‘‘BII’’ can be applied to aqueous micro-filtration applications. 3.5. Crystallinity and thermal behavior Semicrystalline Nylon 66 is usually regarded as a two-phase system consisting of crystalline and amorphous regions. This is also the case for porous Nylon 66 membranes. Three different methods, XRD, FTIR, and DSC were adopted to examine the crystalline characters (crystal type, crystallinity, thermal behavior) of the membranes, in addition to the morphological studies shown previously. The XRD patterns of the four membranes (AI, AII, BI, BII) are shown in Fig. 11. For each pattern, two strong reflection peaks at 2h equal to 20.12° and 23.94° and a weak peak at 13.41° are observed, which correspond to the (1 0 0), (0 1 0, 1 1 0) doublet, and (0 0 2) planes of a-type Nylon 66 crystals oriented in a triclinic unit cell [23–25]. The diffraction pattern can be decomposed into a broad amorphous halo and sharp peaks from reflections of the crystalline regions by a curve fitting technique, as described in the experimental section. As an example, the decomposed results of membrane ÔBIIÕ is shown in Fig. 11 (the two solid curves under pattern ‘‘BII’’), from which the crystallinity was calculated to be 30.1%, a value close to that of Nylon 66 film or fiber reported in the literature [17,19,23,26]. The crystallinity data of the other membranes are listed in Table 2, as were obtained from the same decomposition procedure. It is interesting to find Fig. 9. SEM photomicrographs of membrane ‘‘BII’’ prepared by immersion of dope B in bath II. (a) Cross section 3000X, (b) top surface 100,000X.
(100)
Water flux (g/cm2 min)
0.5
BI BII
(010) (110)
0.4
0.3
(002)
AI
0.2
AII 0.1
BI 0.0 0.2
0.4
0.6
0.8
BII
1.0
∆P (bar)
Fig. 10. Pure water fluxes of membranes ‘‘BI’’ and ‘‘BII’’.
12
14
16
18
20
22
24
26
28
2 Theta (deg.)
of both membranes increase with increasing pressure. However, because membrane ‘‘BI’’ has a denser top sur-
Fig. 11. X-ray (Cu Ka, k = 0.154056 nm) diffraction spectrum of Nylon 66 membranes.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
that the crystallinities of the membranes prepared from dope ÔAÕ are higher than those from dope B (viz., membrane AII AI, BI BII). Given the fact that the morphologies of membranes ÔAIÕ and ÔAIIÕ were largely derived from liquid–liquid demixing, it can be certain that crystallization has occurred in the soft gel-phase after the fast liquid–liquid demixing process. Some evidences have been demonstrated earlier as dendritic crystal elements on the walls of the cellular pores. IR absorption spectra of the membranes were also used to determine the relative levels of crystalline and amorphous phases in the membranes [17,19,23,26]. For a-type Nylon-66 crystals, the relation between crys-
Table 2 Comparison of the crystallinity values obtained from three different methods Code
Degree of crystallinity (Xc) DSC
XRD
FTIR
AI AII BI BII
43.9 43.5 38.6 37.5
38.3 39.1 30.6 30.1
39.8 42.1 36.7 34.5
1200
1190
1180
1170
1160
1180cm-1
1210
Wavenumbers (cm-1) BI
1190
1180
Wavenumbers (cm-1)
1190
1180
1170
1160
BII
1199cm-1
Absorbance
Absorbance
1180cm-1
1200
1200
Wavenumbers (cm-1)
1199cm-1
1210
AII
1199cm-1
Absorbance
Absorbance
1180cm-1
1210
tallinity and dichroic ratio, R = A119/A1180, are available from the literature, where A119 and A1180 are the areas of the integrated absorbances at bands 1199 and 1180 cm1, respectively. The band at 1199 cm1 is a characteristic of the a-crystalline phase whereas the band at 1180 cm1 represents the amorphous component [23]. Curve fitting of the infrared spectra in the range between 1160 and 1215 cm1 for Nylon 66 membranes are shown in Fig. 12. The calculated crystallinities for various membranes are summarized in Table 2. Obviously, the FTIR results for membranes ÔAIÕ and ÔAIIÕ are close to those from XRD measurements; yet, for membranes ÔBIÕ and ÔBIIÕ, the FTIR results are somewhat larger than those from XRD. However, both techniques indicate that membranes ÔAIÕ and ÔAIIÕ have higher crystallinities than membranes ÔBIÕ and ÔBIIÕ. DSC thermal analyses of various membranes were performed and the results are shown in Fig. 13. The scanning rate was 20 °C/min. All thermograms exhibit a major melting peak typical of the melting of normal Nylon 66 crystal and a less significant peak or shoulder at a lower temperature representing the melting of small less stable crystalline units. The measured major melting peak temperatures (Tm), as summarized in Table 1, are all close to 265 °C, a value typical of Nylon 66 polymer
AI
1199cm-1
365
1170
1160
1180cm-1
1210
1200
1190
1180
1170
Wavenumbers (cm-1)
Fig. 12. Curve fitting of the infrared spectra in the range between 1160 and 1215 cm1.
1160
366
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367
Fig. 13. DSC thermograms of the Nylon 66 membranes.
determined by thermal analysis [17,23]. The heat of fusion (DHf) were determined and apparent crystallinities were calculated using DHf0 (197 J/g) of perfect crystalline Nylon 66 taken from literature [17]. Although the double peak in the thermograms suggest a high possibility of an irreversible melting process involving re-crystallization, the calculated crystallinities, as presented in Table 2, are largely in agreement with the results from XRD and FTIR measurements. This implies a low level re-crystallization during thermal analysis or the heat absorbed by re-crystallization was compensated by the heat released during melting of the formed crystals. This, however, require further investigation and is beyond the scope of the present work.
4. Conclusions In this article, nano-scale morphology and crystalline characters of Nylon 66 membranes prepared by immersion–precipitation of two representative dopes in both harsh and soft baths are discussed. During the precipitation process, both liquid–liquid demixing and crystallization occurred and these phase separation mechanisms competed to yield structures which reflected their level of influences. When a well-dissolved casting dope (i.e., dope ÔAÕ) was adopted, liquid–liquid demixing was important regardless of the bath type, and the formed membrane demonstrated a porous morphology contained many cellular pores. In this case, crystallization might either leave its imprint on the pore wall in terms of dendritic crystal elements, such as in membrane ‘‘AI’’, or develop into spherulites in specific regions of the membrane, such as the lower half of the cross section of membrane ‘‘AII’’. The fine structure of a full spherulite indicated a porous surface consisting of twisted
lamellae (<10 nm thickness). When an incipient dope containing a high population of pre-nucleation aggregates (e.g., dope ‘‘B’’) was employed, crystallization dominated the precipitation process, and a membrane with a uniform porous cross section was produced. The crystallites have a sheaf-like or stick-like appearance, being typical of early stage spherulites. If the bath was harsh, the formed membrane still had a skin (‘‘BI’’); yet, if the bath was soft, the formed membrane (‘‘BII’’) became skinless. This membrane was a promising candidate for micro-filtration applications. XRD and DSC analyses of the membranes showed that the crystals in the membranes were of a-type with a melting temperature of ca. 265 °C. The crystallinities of the formed membranes, as determined by FTIR, XRD, and DSC analyses, were over the range 30–45% consistent with the value of Nylon 66 polymer reported in the literature.
References [1] Tohgo K, Fukuhara D, Hadano A. Compos Sci Technol 2001;61:1005. [2] Albano C, Sciamanna R, Gonzalez R, Papa R, Navarro O. Eur Polym J 2001;37:851. [3] Murthy NS, Wang ZG, Akkapeddi MK, Hsiao BS. Polymer 2002;43:4905. [4] Persson A, Jonsson AS, Zacchi G. J Membr Sci 2003;223:11. [5] Castilho LR, Anspach FB, Deckwer WD. J Membr Sci 2002;207:253. [6] Mulder M. Basic principles of membrane technology. Dordrecht: Kluwer Academic Publisher; 1991. [7] Shih CH, Gryte CC, Cheng LP. J Appl Polym Sci 2005;96:944. [8] Cheng LP, Dwan AW, Gryte CC. J Polym Sci: Polym Phys 1995;33:211. [9] Bulte AMW, Mulder MHV, Smolders CA, Strathmann H. J Membr Sci 1996;121:37. [10] Bulte AMW, Mulder MHV, Smolders CA, Strathmann H. J Membr Sci 1996;121:51. [11] Cheng LP, Dwan AH, Gryte CC. J Polym Sci: Polym Phys 1994;32:1183. [12] Cheng LP, Dwan AW, Gryte CC. J Polym Sci: Polym Phys 1995;33:223. [13] Thomas JL, Olzog M, Drake C, Shih CH, Gryte CC. Polymer 2002;43:4153. [14] Kho YW, Kalika DS, Knutson BL. Polymer 2001;42:6119. [15] Young TH, Huang JH, Chuang WY. Eur Polym J 2002;38:63. [16] Men Y, Rieger J. Eur Polym J 2004;40:2629. [17] Elzein T, Brogly M, Schultz J. Polymer 2002;43:4811. [18] Fornes TD, Paul DR. Polymer 2003;44:3945. [19] Vasanthan N, Salem DR. J Membr Sci 2000;38:516. [20] Thanki PN, Ramesh C, Singh RP. Polymer 2001;42: 535. [21] Park YS, Hatae T, Itoh H, Jang MY, Yamazaki Y. Electrochim Acta 2004;50:595.
D.-J. Lin et al. / European Polymer Journal 42 (2006) 356–367 [22] Pall DJ. US Patent 1982; 4,340,479. [23] Kohan MI. Nylon plastics handbook. New York: Hanser/ Gardner; 1995. p. 89. [24] Klein N, Marom G, Wachtel E. Polymer 1996;37:5493.
367
[25] Zhang GZ, Yoshida H, Kawai T. Thermochim Acta 2004;416:79. [26] Vasanthan N, Ruetsch SB, Salem DR. J Polym Sci: Polym Phys 2002;40:1940.