7760
J. Phys. Chem. 1995, 99, 7760-7765
Influence of Membrane Microstructure on the Diffusion Barrier of Supported Liquid Crystalline Membranes R. van den Berg,* D. Schulze, J. A. Bolt-Westerhoff, and F. de Jong Koninklijke/Shell Laboratorium, Amsterdam, Badhuisweg 3, 1031 CM Amsterdam, The Netherlands
D. N. Reinhoudt, D. Velinova, and L. Buitenhuis Technische Universiteit Twente, 7500 AE Enschede, The Netherlands Received: June 28, 1994; In Final Form: November 9, 1994@
Several analytical techniques have been used to characterize the self-diffusion behavior of crown ether based cation carriers inside membranes. With pulsed field gradient nuclear magnetic resonance a significant difference was found between the self-diffusion coefficient in the free solution and in the membrane pores. Furthermore, differences in the pore geometry of different brands of polypropylene membranes could be demonstrated. These results were verified with a combination of atomic force microscopy and confocal scanning laser microscopy, which showed characteristic differences in the morphology of the membranes. These morphological variations may well explain differences in the transport properties of the membranes when impregnated with liquid crystals.
Introduction Supported liquid membranes (SLMs) consist of an organic solution immobilized in a macroporous polymer sheet. A frequently used SLM consists of a polypropylene membrane filled with o-nitrophenyl n-octyl ether (NPOE (Figure 1)) as the solvent. Carriers dissolved in the NPOE may induce high transport rates and selectivities by virtue of their selective complexation. For example, macrocyclic carriers such as crown ethers have been used to transport cations between two aqueous reservoirs through SLMS.'-~ Valuable properties of SLMs are the small membrane volume compared to the large exchanging interface and their well-defined geometry. Liquid crystals in their isotropic phase can be incorporated into membranes, thereby adding the feature of temperature regulation: transport fluxes are expected to change drastically at the temperatures of phase transitions, from crystal (K) or smectic (S) to nematic (N) and isotropic (I). Thin films of liquid crystalline polymers show an extraordinary selectivity toward gases. It is speculated that the solubility in the anisotropic phase of the polymer is so low that transport takes place only through defects (domain boundaries) of the material: If this hypothesis is true, the permeability of such a film should increase when the fraction of material involved in domain boundaries increases. Such a situation can be mimicked by filling the pores of a microporous polypropylene membrane with a liquid crystal. Due to the small diameter of the pores (of around 100 nm), the fraction of the liquid crystal influenced by the polypropylene surface is expected to be very high. CELGARD and ACCUREL are both porous polypropylene membranes, whose permeabilities are about the same when impregnated with normal solvents. We will show in this paper, however, that when liquid crystals are confined in the pores, the permeability differs greatly. In order to prove our hypothesis that this difference is brought about by differences in morphology, we have used a number of analytical techniques such as pulsed-field gradient nuclear magnetic resonance (PFG-NMR), atomic force microscopy (AFM), and confocal scanning laser microscopy (CSLM). @
Abstract published in Advance ACS Abstracts, May 1, 1995.
NPOE
Lo& PSB18C6
M24
Figure 1. Structural formulas of the molecules studied in this study: (a) o-nitrophenyl n-octyl ether (NPOE); (b) polysiloxane with a covalently-boundbenzo-18-crown-6(PSB 18C6);(c) 4-cyano-4-(octy1oxy)biphenyl (M24); (d) n-butylbicyclohexanecarbonitrile(CCH4). TABLE 1: Phase Transition Enthalpies for CCH4 (kJ/mol) As Determined from DSC (Heating at 2 "C/min)
c-s bulk
13.0
4.83
ACCUREL CELGARD
12.2 0.21
4.66
N-I 0.63 0.67
3.11
0.21
S-N
PFG-NMR is a noninvasive analytical technique to study transport phenomena on a microscopic scale! It measures molecular displacements in a fixed spatial direction during a diffusion time on the order of tens to hundreds of milliseconds. Subsequently, self-diffusion coefficients can be calculated from these data. PFG-NMR experiments were performed on SLM systems of increasing complexity (using crown ethers as cation carriers) to study the mobility of each of the constituents as
0022-3654/95/2099-7760$09.00/0 0 1995 American Chemical Society
Diffusion Barrier of Liquid Crystalline Membranes
ta
J. Phys. Chem., Vol. 99, No. 19, 1995 7761
is limited by the wavelength of the exciting laser light used (jllaser = 488 nm). However, owing to a special (so-called confocal) setup of the optics, one can obtain in-focus images from within a bulk material.
/ I
Experimental Section
0.401
--
-
0.20
I
'
20
40
80
60
1
T("C)
4 ,
3
tb t2 3
1
-
0
T("C)
Figure 2. (a) Transport of sodium tetraphenylborate, NaB(C&)4, through NPOE confined in two different polypropylene membranes. (b) Transport of NaB(C6H5)4 through M24 liquid crystal confined in two different polypropylene membranes.
TABLE 2: Self-Diffusion Constants Determined by PFG-NMR (Measured Components in Bold Face)
D entry
solvent
carrier
support
orientation (10-lo mVs)
NPOE NPOE ACCUREL NPOE ACCUREL NPOE-dl7 PSB18c6 NPOE-dl7 PSBl8c-Kf NPOE-dl? PSBl8c6 ACCUREL NPOE-dl7 NPOE-dl7 CELGARD NPOE-dl7 CELGARD NPOE-dl7 PSB 1 8 ~ 6 NPOE-dl7 PSB 18~6-K+
1.4
I II I
I II
0.6 0.5
0.3 0.3 0.09 1.1 0.025 0.1 0.9 0.9
II and I indicate a direction respectively parallel with and perpendicular to the direction of the pores.
defined by the pore morphology. Using microscopic techniques, we have tried to image the pore structure of both membranes in order to explain the observed differences in transport properties. AFM and CSLM' are both relatively new microscopic techniques which have gained increasing popularity in recent years. In AFM a very sharp tip attached to a cantilever is scanned across a surface, and its deflection is monitored by determining the deflection of a laser beam off the end of the cantilever. In this way the forces between the tip and the surface are measured and can be displayed as a topographic image. Whereas this technique in principle allows microscopic information to be obtained on a molecular scale, the resolution of CSLM
Two commercial polypropylene membranes, CELGARD 2500 (thickness 25 pm, porosity 45%) and ACCUREL 1EPP (thickness 100 pm, porosity 65%), were obtained from Hoechst-Celanese and ENKA-MEMBRANES, respectively.The CSLM instrument used in this study was a LEICA Fluovert inverted microscope, operated in the fluorescent mode with a 40x/NA=1.3 oil-immersion objective. Image contrast was artificially increased by filling the pores of the membrane with a solution of rhodamine-6G in NPOE. All AFM images were recorded with a Nanoscope I1 from Digital Instruments in the contact mode in air using microfabricated cantilevers with a small spring constant (0.12 N/m). PFG-NMR experiments were performed on a Varian VXR 200 spectrometer at 200 MHz 'H resonance frequency using a Doty Scientific PFG-NMR probe. The self-diffusion coefficients were calibrated against benzene and glycerol standards.* Experiments using both the Hahn and stimulated echo versions of the pulse sequence were performed by increasing the gradient amplitude in ten steps at a fixed gradient time of 1 ms until an echo attenuation of at least 90% was obtained. For every system, a number of experiments with different diffusion times between 30 and 350 ms were performed which in no case gave significantlydifferent results outside the experimental error. The absolute accuracy of the data reported here is estimated to about 50%; repeatability and relative precision, however, are better than 15%. The polypropylene membranes were filled with NPOE as a solvent. To avoid signal overlap of carrier and solvent, the octyl chain of the NPOE was fully deuterated. Temperature was controlled at 25 "C. As carrier a polysiloxane with a covalently-bound benzo-18crown-6 moiety (PSB18C6, Figure 1) was used, which in some cases contained a K+-ion (indicated by PSB18C6-K+). The polysiloxane backbone increases the lipophilicity of the crown ether and thereby prevents leaking to the aqueous phase in cation transport experiments. The macroscopic transport occurs perpendicular to the membrane surface. For the average length of the pathway through the membrane, however, the diffusion in the membrane plane, Le. parallel to the surface, has to be taken into account as well. Since PFG-NMR measures the selfdiffusion in a certain direction that is given by the direction of the magnetic field gradient and coincides in our experiment with the sample tube axis, diffusion coefficients of NPOE could be measured in the membrane plane (i.e. orthogonal to the direction of the net flux in a transport experiment) and perpendicular to the membrane plane (Le. parallel to the flux direction). For the parallel configuration, the membrane was rolled up in the 5-mm sample tube; cut sheets of 5-mm diameter were stacked for the perpendicular case.
Results and Discussion The transport of sodium tetraphenyl borate, Na+B(C&)4-, from a 0.01 M aqueous source phase to a receiving phase containing water only, was measured using the experimental setup described previously.2 When NPOE was used as the solvent, the fluxes in CELGARD and ACCUREL were about the same at room temperature (Figure 2a), although the apparent activation energy for transport is slightly higher for the former. The fact that similar values for the flux are found for the two membranes is completely accidental: it depends on parameters
J. Phys. Chem., Vol. 99, No. 19, 1995 7763
Diffusion Barrier of Liquid Crystalline Membranes
Figure 4. Side views of both membranes (field-of-view: 49 x 49 pm) studied in this work obtained with the CSLM after filling the pores with a red dye in a direction perpendicular to the membrane surface. The CELGARD shows a much more ordered pore structure than the ACCUREL. The structures shown could be observed throughout both membranes.
like thickness, porosity, tortuosity, and diffusion coefficient. Transport of the same salt through either 4-cyano-4'-(octyloxy)biphenyl (M24, Figure 1) or n-butylbicyclohexanecarbonitrile (CCH4, Figure 1) gave different results: over the whole temperature range the fluxes in CELGARD were much higher than those in ACCUREL (Figure 2b), especially at temperatures below the smectic-nematic phase transition (at 67 "C). The different supports also affect the phase transition enthalpies in a very different way (see Table 1). Whereas the liquid crystal CCH4 confined in an ACCUREL membrane behaves very much like that in the bulk, in CELGARD the enthalpies for the C S and N I transitions are much smaller. Similar effects were reported for liquid crystals confined in the 0.2-pm pores of Anopore membra ne^.^ In Table 2 all data obtained by PFG-NMR for the different samples are presented. A number of conclusions can be drawn. In the first place the carrier-cation complex was found to diffuse at the same rate as the carrier alone, which can be seen from entry 4. Second, it is evident that the self-diffusion of both the solvent NPOE (entries 1-3 and 6-8) as well as the polysiloxane crown ether carrier (entries 4 and 5 ) is reduced inside the membranes as compared to that in the bulk. Self-
-
-
diffusion is therefore significantly hindered due to the influence of pore geometry. This finding is in disagreement with assumptions made earlier in the literature.1° Furthermore, the nearly identical results for the diffusion of NPOE parallel with and perpendicular to the ACCUREL membrane plane (entries 2 and 3) suggest a spongelike morphology of the latter with pores forming a three-dimensional network. A comparison with similar experiments performed on the CELGARD membrane, which is supposed to have much fewer channel interconnections, indeed showed a significant difference in the measured selfdiffusion coefficients in both directions (entries 6 and 7). This membrane therefore appears to have mainly one-dimensional pores. Finally, from entries 6 and 9 it can be seen that the addition of the carrier to the solvent hardly changes the coefficient of self-diffusion of the latter. Diffusion coefficients of PSB 18C6 in ACCUREL could be estimated from experiments in which cation transport was measured." Using a transport model for a 1:l crown ether cation complex and correcting for porosity (0.65) and tortuosity (2.1), these experiments yield a value of 0.24 x m2/s. Lag time experiments were performed by measuring the conductivity in the receiving phase after injection of cations
van den Berg et al.
7764 J. Phys. Chem., Vol. 99, No. 19, 1995
a
C
4
0
0
I
2
Figure 5. AFM images of the surface morphology of the CELGARD (a) and ACCUREL (b) membrane in a 3 x 3 p m image. The pores in the former appear to be elongated with dimensions of around I00 and 400 nm; those in the latter are much more random and clearly show a variation in pore size.
into the source phase. From the time the cations needed to cross m2/s was the membrane, a diffusion coefficient of 0.40 x obtained. In view of the model assumptions and the accuracies
of all methods, the agreement with the value of 0.13 x m2/s (value of 0.09 x m2/s found in Table 2, corrected for a D-isotope effect) determined by PFG-NMR is satisfactory.
Diffusion Barrier of Liquid Crystalline Membranes In order to verify the suggested difference in pore morphology, we have used CSLM on both membrane types. Parts a and b of Figure 3 are images obtained at three different magnifications at a depth of around 20 p m inside the two membranes studied, CELGARD and ACCUREL. Because the dye has filled the holes, a “negative” image contrast is obtained, in which the normally empty, and thus dark, pores are highlighted. The origin of the “superstructure” which is observed at lower magnification (Figure 3b) is unknown. Figure 4 is a side view (XZ-scan) of both membranes which shows that whereas the pores of the CELGARD are rather straight, relatively ordered, and perpendicular to the membrane surface, those of the ACCUREL are much more random in nature. These results on the CELGARD membrane are nicely corroborated by a transmission electron microscopy study published elsewhere.’* It should be noted, however, that the actual pore size, which in both cases (according to the manufacturer) is a few hundred nanometers, cannot be determined from these CSLM images. Therefore, we have tried to image the (surface) morphology of both membranes using AFM. In the normal AFM mode, the tip is in constant contact with the sample. In order to be able to study “soft” samples, one has to use very small forces on the order of a few nanoNewtons. For the samples studied here we have carefully checked whether repetitive scanning would damage the surface and have not detected any artifacts. In Figure 5a a 3 x 3 p m image of the CELGARD membrane surface is shown. It appears that horizontal bars form the basis for rows of thin, vertical pillars enclosing the actual pores. If a similar scan is performed on the ACCUREL membrane surface (Figure 5b), it becomes clear that the pores have widely varying sizes and are more randomly distributed. This is in accordance with the CSLM results. From the studies described above we conclude that in ACCUREL the liquid crystals have a “random” 3D-structure,
J. Phys. Chem., Vol. 99, No. 19, 1995 7765 as in bulk liquid crystalline material. In CELGARD on the other hand, the relatively straight and ordered pore structure leads to a so-called confined anisotropy, in which the liquid crystals are ordered by the pores.13 These morphological differences may explain the experimental observation that the confined liquid crystal in CELGARD shows good permeability, whereas that in ACCUREL is much smaller. The higher permeability of liquid crystals in very small pores resembles the observed increase in permeability of poly(viny1 chloride)/ liquid crystal composite membranes after alignment of the latter in an electric field.14
References and Notes (1) Izatt, R. M.; Pawlak, K.; Bradshaw, J. S.; Bruening, R. L. Chem. Rev. 1991, 91, 1721. (2) Nijenhuis, W. F.; Buitenhuis, E. G.; de Jong, F.; Sudholter, E. J. R.; Reinhoudt, D. N. J . Am. Chem. Soc. 1991, 113, 7963. (3) Wienk, M. M.; Stolwijk, T. B.; Sudholter, E. J. R.; Reinhoudt, D. N. J. Am. Chem. SOC.1990, 112, 797. (4) Chiou, J. S.; Paul, D. R. J . Polym. Sci., Polym. Phys. Ed. 1987, 25, 1699. (5) Stilbs, P. Prog. Nucl. Magn. Res. Spectrosc. 1987, 19, 1. (6) Binnig, G. Ulrramicroscopy 1992, 42-44, 7. (7) Brakenhoff, G. J. J . Microsc. 1979, 117, 233. (8) Datema, K. P.; Bolt-Westerhoff, J. A,; Nesbitt, G . J.; Maarssen, P. K.; Ylstra, W.; Tutunjian, P. N.; Vinegar, H.; K&ger, J. In Magnetic Resonance Spectroscopy; Bliimich, B., Kuhn, W., Eds.; VCH Verlagsgesellschaft: Weinheim, Germany, 1992. (9) Iannacchione, G. S.; Finotello, D. Phys. Rev. Lett. 1992,69, 2094. (10) Dozol, J. F.; Casas, J.; Sastre, A. Sep. Sci. Technol. 1993,28,2007. (1 1) Reichwein-Buitenhuis, L. Thesis, University of Twente, 1993. (12) Sararda, T.; Sawyer, L. C.; Ostler, M. I. J . Membr. Sci. 1983, 15, 97. (13) Crawford, G. P.; Stannarius, R.; Doane, J. W. Phys. Rev. A 1991, 44, 2558. (14) Kajiyama, T. Chem. Express 1991, 6, 719. JP9416275