- Email: [email protected]
Effect of an applied electric field on liquid crystalline membranes: control of permeability
Journal of Membrane Elsevier
Science
Science,
Publishers
24 (1985)
B.V.,
83
83-96
Amsterdam
-
Printed
in The Netherlands
EFFECT OF AN APPLIED ELECTRIC FIELD ON LIQUID CRYSTALLINE MEMBRANES: CONTROL OF PERMEABILITY
R.K.
BHASKAR’,
R.V.
SPARER’
and K.J.
HIMMELSTEIN’~*~*
1Department
of Chemical and Petroleum Engineering, University of Kansas, Lawrence, KS 66045 (U.S.A.) ‘INTER, Research Corporation, Merck Sharp and Dohme Research Laboratories, 2201 W. 21 St., Lawrence, KS 66046 (U.S.A.) (Received
April
16, 1984;
accepted
in revised
form
April
3, 1985)
Summary This paper considers the development of liquid crystalline variable permeability membranes. The electrically induced changes in permeability of a liquid crystalline membrane composed of a 23% w/v solution of poly(r-benzyl-l-glutamate) (PBLG) in dichloromethane are described. PBLG exists in the cholesteric mesophase in solution in a suitable cu-helicogenic solvent. When an electric field is applied to a thin membrane comprised of this material, a transition to the nematic mesophase is triggered, and the permeability of the membrane for small organic permeants increases. This increase takes the form of a 50-60% increase in the steady state flux through the membrane.
Introduction
The use of controlled release technology for the prolonged administration of drugs by various routes of administration has recently been the subject of considerably study. However, almost all of the systems produced thus far have a common feature: the rate of release of drug is either constant or decays with time. There has been little work on inexpensive and simple systems which can deliver drugs on demand in response to an external signal. A delivery system which can deliver drugs in such a manner would allow treatment schedules which could be pulsative, periodic, or be in direct response to some signal generated by the disease state itself. One approach to the problem is to have a polymeric delivery system, either membrane/reservoir or matrix, whose permeability to the drug species can be altered by some external agency. Several methods to .control the permeability of membranes immersed in aqueous media have been reported in the literature. Briefly, the permeability of a membrane system may be enhanced by two classes of methods: modification of membrane structure and modification of membrane environment. Most published work on permeability control by means of *To
whom
correspondence
0376-7388/85/$03.30
should
o 1985
be addressed.
Elsevier
Science
Publishers
B.V.
84
modification of a membrane’s environment were related to boundary layer effects, i.e., the “unstirred” film immediately adjacent to the membrane surface (Lakshminarayanaiah [l] ). Among the factors examined were the relative abundance of protons in the boundary layer (Lobe1 and Caplan [2]), thickness of the boundary layer, temperature, etc. Cherkasov et al. [3] have reported an increase in the permeability of ultrafiltration membranes caused by a breakdown of water in the boundary layer, which in turn was caused by an applied electric field. Examples of alteration of permeability by modification of membrane structure have been rather more numerous - specifically, photochemical [4], magnetic [ 51, thermal [6, 71, and electrical [S] methods have been reported. Of these, only magnetic triggers have been demonstrated as potentially useful for controlled release applications. The problem reduces to the selection of a material such that an externally applied stimulus such as a magnetic field, temperature change or an electric field causes a change of phase or other large scale conformational transition, which in turn results in a change in the permeability of the membrane. This work deals with the use of a liquid crystalline polymer as a variable permeability membrane. Poly(r-benzyl-l-glutamate) (PBLG) is a well-characterized liquid crystalline polymer; the electrotropic (field induced) cholesteric to nematic phase transition of PBLG has been extensively investigated a.
Cholesteric
mesophase
b.
SIDE
4
Cholesteric
Nemotic
VIEW
mesophase
OPTIC
AX6
Axis
Fig. 1. Schematic representations of (a) cholesteric (b) nematic mesophase (from Ref. [lo] ).
mesophase
(Samulski,
1970),
and
85
[ 91. Figure 1 depicts the cholesteric and nematic mesophases of PBLG. Figure lb is reproduced from Ref. [lo] with permission from the publisher. If this material is contained in a thin film which then has an electric field applied to it, then a membranic drug delivery system may be constructed to deliver drugs in response to an electrical signal. This paper demonstrates that permeability changes can be induced in such a membrane. Experimental materials and methods Poly(y-benzyl-l-glutamate), molecular weight 210,000 and 28,000, was obtained from Sigma Chemicals Corporation, St. Louis, MO. and used without additional purification. Approximately 23% w/v solutions of the polymer in dichloromethane (Aldrich, spectrochem grade) were prepared and were allowed to age for 2-3 days at 10°C so as to form the cholesteric mesophase. When freshly made the solutions were colorless and highly viscous; after they were aged they were pale yellow and more mobile. Preliminary characterization of the liquid crystal was performed on a Perkin-Elmer differential scanning calorimeter and revealed an endothermic low-energy transition at 27°C. This was presumably the anisotropic to cholesteric transition of the polymer that has been reported by Guha Sridhar et al. [ 111. An infrared spectrum of the liquid crystalline material was also taken on a Nicolet 5DX Fourier transform infrared spectrometer; this spectrum matched the spectra of cholesteric PBLG that were reported by Kajiyama et al. [12], revealing that no significant degradation of the polymer occurred during aging. An Amicon Model RK-8010 ultrafiltration cell was used, modified so as to allow the confinement of liquid crystalline membranes and the application of an electric field. The final configuration of the cell is shown in Fig. 2. Among other advantages, the vertical arrangement allowed the use of motordriven stirrers for the donor compartment of the membrane, thereby minimising the effect of unstirred boundary layers. The liquid crystal was confined between two Amicon C-300 cellulose acetate microporous filters separated by a 1 mm thick silicone rubber O-ring, giving an effective membrane thickness of 1 mm. A microporous filter was placed on the bottom of the membrane holder and was held in place with a thin layer of Dow Corning Silastic- silicone adhesive, and the O-ring was placed above it. The liquid crystal was then poured into the cavity formed by the O-ring and the filter. The surface of the liquid crystal was covered immediately with the other filter, whose edges had been previously coated with adhesive. The membrane holder was then clamped into the cell body using a threaded aluminum retaining ring. The application of an electric field across the membrane was accomplished as follows: A thin layer of gold/palladium was deposited across the surface of the microporous filters confining the membrane. The coated surfaces were placed inward toward the liquid crystal and fine copper wires (41
86
S.W.G.) were bonded to the inner surface of the filters using Polysciences Colloidal Graphite conductive adhesive. The wires were color coded for identification and were connected to a B&K Precision Model 1800 DC regulated power supply in series with a microammeter. Variation of the strength of the electric field was done in two ways: (1) changing the applied voltage across the membrane, and (2) changing the thickness of the membrane. The latter was accomplished by the use of polycarbonate (rather than silicone rubber) O-rings. These rings were machined to thicknesses of 1.5 and 2.0 mm, respectively. Permeation data were collected for two different permeants: phthalyl alcohol (1,2-benzene dimethanol) and methylene blue (3,7-bis(dimethylamino phenazathionium) chloride). These permeants provide neutral and charged species in solution, respectively, The solubility coefficients of both these permeants in the liquid crystal were measured; these were found to be 6.2 and 5.7, respectively. Immediately after the membrane was assembled, the donor compartment was filled with 10 ml of a solution containing either 0.09981 mg/ml of methylene blue or 0.1436 mg/ml of phthalyl alcohol. Both stirrers and a stopwatch were started immediately. Aliquots of 1 ml were drawn from the receptor compartment (1000 ml) and replaced with 1 ml of distilled water. The absorbance of the samples was measured at 665 nm using a
Microporous
Electrode
2”-d Microporous
Electrode
Membrane
Holder
Threaded
Clamp
RinQ
I
Fig. 2. Experimental
apparatus.
87
Varian Associates Cary 219 spectrophotometer. Samples were initially drawn every fifteen minutes; later, sample collection was automated by continuous flow of the receptor phase through a sampling loop which included the receptor compartment, a peristaltic pump, and a UV/vis detector. The flow rate through the sampling loop was set at 100 ml/mm and the spectrophotometer was programmed to monitor the absorbance for 5 minutes at 15minute intervals. Although this arrangement introduced a small time lag (l-2 min) between the conditions in the receptor compartment and the conditions in the membrane, the lag was considered small in comparison to the time required for the transitions that were being studied. Steady state fluxes reported in this paper are the mean values of three sets of data. Permeation runs were carried out at 30.0, 35.0, and 37.0% with electric fields of 0 and 38 V for each run. The temperatures were maintained by a Cole-Parmer Versatherm electronic temperature controller. Stirrer speeds were set at the same level (200 rpm) for all the runs. Permeation runs were carried out at pH 7.0, 5.0, and 6.1 (unbuffered). The pH of both compartments was regulated using a stock HPOi-/H,POgbuffer solution. The temperature was maintained at 30°C and the voltages set at 0 and 38 V for all cases. Experiments were also performed in which the field was turned off 1.0, 2.0, and 3.5 hours after the start of the run, respectively. The collection of permeation data was continued after the field was turned off to ascertain the duration of the effects caused by the field. Finally, a set of data was collected in which the polarity of the electric field was reversed after 1.0,2.0, and 3.5 hours, respectively. The cholesteric pitch of the liquid crystal was measured by estimating the distance between successive retardation lines under the microscope. (The distance between any two successive lines is equal to one-half of the cholesteric pitch.) Cholesteric pitch values were determined for liquid crystals prepared from PBLG of molecular weights of both 310,000 and 28,000, and were found to be 32 urn and 17 pm, respectively. Results and discussion This section presents permeation data for the two permeants through the system, and a discussion of its significance. The effect of an applied electric field was examined for both a neutral permeant, phthalyl alcohol, and a cationic permeant, methylene blue. Referring to Fig. 3, the steady state flux was taken to be the slope of the linear portion of the absorbance versus concentration curve for the receptor compartment of the membrane system, and the diffusional lag time was given by the intercept of the linear portion projected on to the x-axis. For both permeants, the steady state fluxes were observed to increase and the diffusional lag times to decrease. As shown in Fig. 4, the steady state flux increased from 1.33 X 10W2 mg/cm’-min at 15 V to 2.84 X lob2 mg/cm’min at 35 V for methylene blue, with an abrupt change occurring around
88
31 V. Similarly, as shown in Fig. 5, the steady state flux for phthalyl alcohol increased from 4.3 X 10m2 mg/cm2-min at 15 V to 7.1 X 10d2 mg/cm2-min at 35 V with the abrupt change also occurring between 28 and 35 V. Diffusional lag times were seen to decrease from 66 minutes to 37 minutes for methylene blue, and from 43 minutes to 21 minutes for phthalyl alcohol,
0 Time,
Fig. 3. Appearance run.
of permeant
in the receptor
20.0 DC
Fig. 4. blue.
Steady
state
flux
minutes
Electric
as a function
compartment
during a typical
permeation
field for the permeant
methylene
30.0 field,
V
of DC electric
89
DC
Fig. 5. Steady alcohol.
state
flux
30 field,
20 Electric
as a function
of
DC
V
electric
field
for
the permeant
phthalyl
as shown in Tables 1 and 2, respectively. Again, the most abrupt changes occurred between 28 and 35 V for both permeating species. A simple Fickian model such as the one used by Mohadger [7] may be used to evaluate diffusivities in the system. In such a system, the steady state flux through the system is given by J = -PA C/l where J is the steady state flux, AC is the concentration difference between donor and receptor compartments, and 1 is the thickness of the membrane. P, the membrane permeability, is defined for the purposes of this work to TABLE Effect Field 0 15 31 35 38 45 50
I of electric
(V)
fields
on steady
state flux
Lag time (min)
S.S.F.
66.2 + 2.7 65.0 + 2.6 36.5 f 1.65 35.2 * 1.60 37.2 k 1.71 36.5 * 1.50 38.0 i 1.91
1.33 1.33 2.28 2.35 2.82 2.81 2.84
( lo2 mg/cm*-min) + + t i r f +
0.06 0.06 0.08 0.09 0.14 0.11 0.14
Permeant: Methylene blue. No. of observations = 3; values are + standard
error of the mean.
90
TABLE 2 Effect of electric fields on steady state. flux Field (V)
Diffusional lag time (min)
Steady-state flux ( lo2 mg/cm2-min)
0.0 15.0 30.0 33.0 35.0 38.0 40.0 50.0
43.2 41.5 41.1 23.8 21.4 21.9 23.4 21.4
4.27 4.21 5.01 7.15 7.21 7.36 7.28 7.26
f f + * f + * k
1.94 1.95 1.89 0.98 0.90 1.03 0.98 1.24
+ f * + i i i f
0.19 0.20 0.23 0.29 0.30 0.34 0.30 0.42
Permeant: Phthalyl alcohol. Donor compartment: 0.1436 mg/ml phthalyl alcohol. Receptor compartment: 1000 ml distilled water/PO:- buffer, pH = 7.0. Temperature: 30°C t 0.1%. Membrane: 1 mm cholesteric PBLG, mol. wt. 310,000. No. of observations = 3; values are * standard error of the mean.
be the product of the solubility coefficient, coefficient, Da,, :
s, and the average diffusion
P=SD, This model reveals an increase in the average diffusivity from 3.6 X lo5 cm’/sec at 0 V to 7.6 X lo5 cm*/sec at 35 V for methylene blue and from 7.2 to 9.5 X lo5 cm’/sec over the same voltage range for phthalyl alcohol. The observed changes in lag time between no field and the above critical voltage were also consistent with a change in the diffusivity. Diffusion coefficients were not calculated using the lag times since this method is generally not reliable for the absolute determination of diffusivities. In this system, it is seen from an examination of Fig. 2 that the nematic directors of the liquid crystalline membrane would lie perpendicular to the plane of the membrane itself, i.e., parallel to the direction of transport. A transition within the membrane from the cholesteric to the nematic mesophase would thus be expected to facilitate transport across the membrane. It is contended that the increase in the diffusion coefficient was due to the parallel alignment of the polymeric species in the membrane as a consequence of the formation of the nematic mesophase. It is speculated that this alignment restricts the freedom of a permeant molecule in all directions but one - the direction of the alignment of the polymeric species. A crude analogy to this situation would be pinball machine in which the channels for the balls were all parallel to the gradient of the board. It is easily visualized that a ball in such a machine would reach its destination much faster than in a machine where it would have to travel longer distances due to the random orientation of the channels.
91
In order to show that the permeability increases were not caused by nonphase transitional electrotropic effects or related boundary phenomena, permeation runs were performed in which the field was interrupted and in which the polarity of the DC electric field was reversed. Enhanced steady state fluxes (2.3-2.8 X lo-* mg/cm*-min) were maintained for up to 2.3 hours after the field was interrupted, as shown in Fig. 6. When the polarity of the electric field was reversed during the permeation run, very little effect on the steady state flux was noted once the enhanced fluxes were triggered. No statistical difference in the steady state flux could be seen. These effects were observed for both neutral and ionic permeants.
I
I
J
0 Fig.
30
6. Appearance
of methylene
60 Time.
90 m!?utes
blue in the receptor
120
compartment
150
after
field interrupt
at 90 min.
The concentration of PBLG in the membrane was varied so as to connect the liquid cyrstalline behavior of the polymer with the changes in permeability of the membrane. The normalized change in the steady state flux, p, was studied as a function of the concentration of PBLG in the membrane. fl was defined as S.S,F.31v
P=
- S.S.F.oV
S.S.F.ov
The subscripts are the voltages selected for the evaluation of p, (0 V and 31 V), and correspond to the pre- and post-transitional field strengths, respectively. The Flory critical concentration for PBLG in dichloromethane (i.e., the concentration at which the formation of the cholesteric mosophase is just complete) is around 23% w/v [ 131. The value of $ was seen to increase from 0.0793 at 15% w/v to 0.5284 at 25% w/v, as shown in Fig. 7. There appears to be a critical concentration below which no electric field induced
92 0.60
0.48 ;; : c -$ 0.36
-
:
.-E t
0.24
0.1 2
0.00
i ;J1, , , / , [ /
ti
m
: 0
I
1
20
10
Fig. ‘7. Fractional the membrane.
change,
8, in steady
40
30
Concentration,
w/v
state flux
%
as a function
of PBLG
concentration
in
permeability changes take place. Above this concentration, however, the changes in permeability are rather dramatic. This effect was observed for both neutral and ionic permeants. The existence of cholesteric and nematic mesophases at field strengths of 0 and 380 V/cm, respectively, was confirmed by observation through an optical microscope at a magnification of 150-200X. The time lag for the reversion of the nematic mesophase to the cholesteric was estimated by viewing the fractional area of the respective textures under the microscope, and was found to be 3.3 hr. This time lag matched the corresponding (3.4 hr) lag for the reversion of normal steady state fluxes within the membrane after field removal as discussed above. Photomicrographs of the cholesteric and the nematic mesophases are shown in Figs. 8 and 9, respectively, The cholesteric pitch of the liquid crystal was measured by direct visual observation (method of Duke and DuPre [14] ) for PBLG of molecular weight 28,000 and 310,000, respectively. Theoretical field strengths for the cholesteric-nematic transition were calculated by means of the deGennes relation [ 151 as follows: E, = (7?/2)
(47TK/e~)l’~
(l/p,)
where eA is the dielectric anisotropy of the medium, K is the twist elastic constant, and P,, is the cholesteric pitch of the liquid crystal. Table 3 shows the theoretical field strengths for the cholesteric-nematic transition in comparison to the actual field strengths required to trigger permeability enhancement. In both cases, agreement is within +15 V/cm. Tables 4 and 5 show the effect of temperature and pH on the steady state flux through the system. Lobe1 and Caplan [2] have shown that the presence
Fig.
PBLG.
9. Nematic mesophase of PBLG.
94
TABLE 3 DeGennes and experimental
critical field strengths
DeGennes field strength
Cholesteric pitch
Experimental
(V/cm) 17.0 32.0
field strength
W/cm)
593.0 380.0
610.0 368.0
These experimental values are derived from other standard error is given.
experimental
data. Therefore,
no
TABLE 4 Effect of temperature Temperature 30.0 35.0 37.0
(“C)
on steady state flux
S.S.F. (0 V)
S.S.F. (38 V)
(1.33 f 0.07) x 10-a (1.39 ? 0.06) x lo-’ (1.43 i 0.09) x 10-z
(2.82 i: 0.15) x lo-* (2.85 * 0.11) x lo-’ (2.93 + 0.17) x 1o-2
Permeant: Methylene blue. No. of observations = 3; values are + standard error of the mean. TABLE
5
Effect of pH on steady state flux PH
S.S.F. (0 V)
S.S.F. (38 V)
7.0 6.1 5.0
(1.33 (1.41 (1.52
(2.82 i 0.15) x 10-l (3.01 f 0.07) x lo-* (3.23 f 0.06) x 1O-2
+ 0.07) x lo-* + 0.04) x lo+ r 0.03) x lo-*
Permeant: Methylene blue. No, of observations = 3; values are + standard error of the mean.
of protons in the boundary layer can affect the transport properties of solvent type membranes. A decrease in the steady state flux with pH was also observed; however, the number of data points was too few to be conclusive of any trend. In any event, the observed decreases were considerably less than the observed field-induced changes. Consequently it was assumed that boundary layer effects of the type discussed by Lobe1 et al. were not induced by the electric field. The changes in steady state flux with temperature were consistent with the dependence of the diffusion coefficient on temperature, which typically follows an Arrhenius-type relationship. Conclusions The results presented above show that the enhancement of the permeability of the membrane due to an externally applied electric field is consistent
95
with a cholesteric-nematic phase transition within the membrane, and that non-phase transitional electrotropic effects and related boundary layer phenomena were absent as dominant factors. This assertion is based on the following reasons: (1) There was no reason to believe that non-phase transitional electrotropic effects would persist after the electric field had been turned off or when the polarity of the field had been reversed. (2) Other electrotropic effects would cause differences in permeation characteristics between neutral and ionic permeants. The differences that were observed were small enough to be ascribed entirely to permeabilities based on different solubilities and diffusion coefficients. (3) The abrupt changes in p, the normalized steady state flux, with the composition of the membrane offer additional evidence in favor of the hypothesis that a liquid crystalline phase transition was responsible for the electrically induced changes in the steady state flux. The concentration at which fl underwent the largest increase agreed to within 2% of the Flory critical concentration for PBLG in dichloromethane. (4) Theoretical critical field strengths were calculated from experimentally obtained values of the cholesteric pitch by means of the deGennes relation [ 141. The critical electric field strength for the cholesteric-nematic phase transition is directly proportional to the dielectric anisotropy and the twist elastic constant of the liquid crystal and inversely proportional to the cholesteric pitch. The theoretical field strengths so obtained agreed well with actual field strengths required for the permeability enhancement. (5) Finally, optical studies on the PBLG-dichloromethane system revealed that the onset of the nematic phase of PBLG coincided with the onset of the increase in steady state flux. Furthermore, the time lag for the resumption of normal steady state fluxes agreed well with the time lag for the reversion of the nematic mesophases in the optical samples to the cholesteric mesophases. Acknowledgments Technical assistance from the staff of INTER, Research Corporation, Lawrence, Kansas is gratefully acknowledged. This work was funded in part by a grant from INTER, Research Corporation, Merck Sharp and Dohme Research Laboratories. References 1 2
3
N. Lakshminarayanaiah, Transport Phenomena in Membranes, Academic Press, New York, N.Y., 1969, p. 129. C. Lobe1 and S.R. Caplan, Unstirred layer controlled transport of organic bases and drugs through solvent-type membranes and the influence of bound enzymes, J. Membrane Science, 6 (1980) 203-220. A.N. Cherkasov, V.A. Pasechnik, V.P. Zhemkov, V.V. Chechina and R.S. Alimardanov, Effect of direct current on the permeability of ultrafiltration membranes, Kolloid. Zh., 41( 3) (1979) 586-590.
96
8
9
10 11
12
13 14
15
W. Prins, J.C. Chuang and M. DeSorgo, Mechanochemistry of photochromic model membranes, J. Mechanochem. Cell Motility, 2 (1973) 105-112. R. Langer, J. Folkman and D.S.T. Hsieh, Magnetic modulation of release of macromolecules from polymers, Proc. Natl. Acad. Sci. U.S.A., 78(3) (1981) 1863-1867. C.E. Rogers, Structural factors, in: D.F. Paul and F.W. Harris (Eds.), Controlled Release Polymeric Formulations, Am. Chem. Sot., Washington, DC, 1976, p. 15. Y. Mohadger, A.V. Tobolsky and T.K. Kwei, Permeation of isomers of mandelic acid through isotropic and cholesteric fluid polypeptide membranes, Macromolecules, 7(6) (1974) 924-929. A.J. Grodzinsky and S.R. Eisenberg, Electric field control of membrane permeability, in: Proceedings of the International Conference on Biomedical Engineering, New York, 1980, pp. 237-241; also S.R. Eisenberg, S.M. Thesis, Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 1979. H. Finkelmann, The synthesis, structure, and properties of polymers with liquid crystalline side chains, in: A. Ciferri (Ed.), Polymeric Liquid Crystals, Academic Press, New York, N.Y., 1982, pp. 35-62. F.D. Saeva, Nematic liquid crystals, in: Liquid Crystals: The Fourth State of Matter, Marcel Dekker, New York, N.Y., 1979, p. 75. C. Guha Sridhar, W.A. Hines and E.T. Samulski, Polypeptide liquid crystals: Magnetic susceptibility, twist elastic constant, rotational viscosity coefficient, and polyr-benzyl-l-glutamate side chain configuration, J. Chem. Phys., 61(3) (1974) 947953. T. Kajiyama, A. Kumano and 0. Niwa, Photoinduced ion permeation through a ternary composite membrane composed of polymer/liquid crystal/azobenzenebridged crown ether, Chem. Lett., (1983) 1327-1330. P.J. Flory, Phase equilibria in solutions of rod-like particles, Proc. Roy. Sot., Ser. A, 234 (1956) 73. R.W. Duke and D.B. DuPre, Examination of the electrotropic cholesteric to nematic phase transition of liquid crystal poly(r-benzyl-l-glutamate) by laser light beating spectroscopy, Macromolecules, 3 (1974) 374-376. P.G. deGennes, Long range order and thermal fluctuations in liquid crystals, Mol. Cryst. Liq. Cryst., 7 (1969) 325-345.
Science
Science,
Publishers
24 (1985)
B.V.,
83
83-96
Amsterdam
-
Printed
in The Netherlands
EFFECT OF AN APPLIED ELECTRIC FIELD ON LIQUID CRYSTALLINE MEMBRANES: CONTROL OF PERMEABILITY
R.K.
BHASKAR’,
R.V.
SPARER’
and K.J.
HIMMELSTEIN’~*~*
1Department
of Chemical and Petroleum Engineering, University of Kansas, Lawrence, KS 66045 (U.S.A.) ‘INTER, Research Corporation, Merck Sharp and Dohme Research Laboratories, 2201 W. 21 St., Lawrence, KS 66046 (U.S.A.) (Received
April
16, 1984;
accepted
in revised
form
April
3, 1985)
Summary This paper considers the development of liquid crystalline variable permeability membranes. The electrically induced changes in permeability of a liquid crystalline membrane composed of a 23% w/v solution of poly(r-benzyl-l-glutamate) (PBLG) in dichloromethane are described. PBLG exists in the cholesteric mesophase in solution in a suitable cu-helicogenic solvent. When an electric field is applied to a thin membrane comprised of this material, a transition to the nematic mesophase is triggered, and the permeability of the membrane for small organic permeants increases. This increase takes the form of a 50-60% increase in the steady state flux through the membrane.
Introduction
The use of controlled release technology for the prolonged administration of drugs by various routes of administration has recently been the subject of considerably study. However, almost all of the systems produced thus far have a common feature: the rate of release of drug is either constant or decays with time. There has been little work on inexpensive and simple systems which can deliver drugs on demand in response to an external signal. A delivery system which can deliver drugs in such a manner would allow treatment schedules which could be pulsative, periodic, or be in direct response to some signal generated by the disease state itself. One approach to the problem is to have a polymeric delivery system, either membrane/reservoir or matrix, whose permeability to the drug species can be altered by some external agency. Several methods to .control the permeability of membranes immersed in aqueous media have been reported in the literature. Briefly, the permeability of a membrane system may be enhanced by two classes of methods: modification of membrane structure and modification of membrane environment. Most published work on permeability control by means of *To
whom
correspondence
0376-7388/85/$03.30
should
o 1985
be addressed.
Elsevier
Science
Publishers
B.V.
84
modification of a membrane’s environment were related to boundary layer effects, i.e., the “unstirred” film immediately adjacent to the membrane surface (Lakshminarayanaiah [l] ). Among the factors examined were the relative abundance of protons in the boundary layer (Lobe1 and Caplan [2]), thickness of the boundary layer, temperature, etc. Cherkasov et al. [3] have reported an increase in the permeability of ultrafiltration membranes caused by a breakdown of water in the boundary layer, which in turn was caused by an applied electric field. Examples of alteration of permeability by modification of membrane structure have been rather more numerous - specifically, photochemical [4], magnetic [ 51, thermal [6, 71, and electrical [S] methods have been reported. Of these, only magnetic triggers have been demonstrated as potentially useful for controlled release applications. The problem reduces to the selection of a material such that an externally applied stimulus such as a magnetic field, temperature change or an electric field causes a change of phase or other large scale conformational transition, which in turn results in a change in the permeability of the membrane. This work deals with the use of a liquid crystalline polymer as a variable permeability membrane. Poly(r-benzyl-l-glutamate) (PBLG) is a well-characterized liquid crystalline polymer; the electrotropic (field induced) cholesteric to nematic phase transition of PBLG has been extensively investigated a.
Cholesteric
mesophase
b.
SIDE
4
Cholesteric
Nemotic
VIEW
mesophase
OPTIC
AX6
Axis
Fig. 1. Schematic representations of (a) cholesteric (b) nematic mesophase (from Ref. [lo] ).
mesophase
(Samulski,
1970),
and
85
[ 91. Figure 1 depicts the cholesteric and nematic mesophases of PBLG. Figure lb is reproduced from Ref. [lo] with permission from the publisher. If this material is contained in a thin film which then has an electric field applied to it, then a membranic drug delivery system may be constructed to deliver drugs in response to an electrical signal. This paper demonstrates that permeability changes can be induced in such a membrane. Experimental materials and methods Poly(y-benzyl-l-glutamate), molecular weight 210,000 and 28,000, was obtained from Sigma Chemicals Corporation, St. Louis, MO. and used without additional purification. Approximately 23% w/v solutions of the polymer in dichloromethane (Aldrich, spectrochem grade) were prepared and were allowed to age for 2-3 days at 10°C so as to form the cholesteric mesophase. When freshly made the solutions were colorless and highly viscous; after they were aged they were pale yellow and more mobile. Preliminary characterization of the liquid crystal was performed on a Perkin-Elmer differential scanning calorimeter and revealed an endothermic low-energy transition at 27°C. This was presumably the anisotropic to cholesteric transition of the polymer that has been reported by Guha Sridhar et al. [ 111. An infrared spectrum of the liquid crystalline material was also taken on a Nicolet 5DX Fourier transform infrared spectrometer; this spectrum matched the spectra of cholesteric PBLG that were reported by Kajiyama et al. [12], revealing that no significant degradation of the polymer occurred during aging. An Amicon Model RK-8010 ultrafiltration cell was used, modified so as to allow the confinement of liquid crystalline membranes and the application of an electric field. The final configuration of the cell is shown in Fig. 2. Among other advantages, the vertical arrangement allowed the use of motordriven stirrers for the donor compartment of the membrane, thereby minimising the effect of unstirred boundary layers. The liquid crystal was confined between two Amicon C-300 cellulose acetate microporous filters separated by a 1 mm thick silicone rubber O-ring, giving an effective membrane thickness of 1 mm. A microporous filter was placed on the bottom of the membrane holder and was held in place with a thin layer of Dow Corning Silastic- silicone adhesive, and the O-ring was placed above it. The liquid crystal was then poured into the cavity formed by the O-ring and the filter. The surface of the liquid crystal was covered immediately with the other filter, whose edges had been previously coated with adhesive. The membrane holder was then clamped into the cell body using a threaded aluminum retaining ring. The application of an electric field across the membrane was accomplished as follows: A thin layer of gold/palladium was deposited across the surface of the microporous filters confining the membrane. The coated surfaces were placed inward toward the liquid crystal and fine copper wires (41
86
S.W.G.) were bonded to the inner surface of the filters using Polysciences Colloidal Graphite conductive adhesive. The wires were color coded for identification and were connected to a B&K Precision Model 1800 DC regulated power supply in series with a microammeter. Variation of the strength of the electric field was done in two ways: (1) changing the applied voltage across the membrane, and (2) changing the thickness of the membrane. The latter was accomplished by the use of polycarbonate (rather than silicone rubber) O-rings. These rings were machined to thicknesses of 1.5 and 2.0 mm, respectively. Permeation data were collected for two different permeants: phthalyl alcohol (1,2-benzene dimethanol) and methylene blue (3,7-bis(dimethylamino phenazathionium) chloride). These permeants provide neutral and charged species in solution, respectively, The solubility coefficients of both these permeants in the liquid crystal were measured; these were found to be 6.2 and 5.7, respectively. Immediately after the membrane was assembled, the donor compartment was filled with 10 ml of a solution containing either 0.09981 mg/ml of methylene blue or 0.1436 mg/ml of phthalyl alcohol. Both stirrers and a stopwatch were started immediately. Aliquots of 1 ml were drawn from the receptor compartment (1000 ml) and replaced with 1 ml of distilled water. The absorbance of the samples was measured at 665 nm using a
Microporous
Electrode
2”-d Microporous
Electrode
Membrane
Holder
Threaded
Clamp
RinQ
I
Fig. 2. Experimental
apparatus.
87
Varian Associates Cary 219 spectrophotometer. Samples were initially drawn every fifteen minutes; later, sample collection was automated by continuous flow of the receptor phase through a sampling loop which included the receptor compartment, a peristaltic pump, and a UV/vis detector. The flow rate through the sampling loop was set at 100 ml/mm and the spectrophotometer was programmed to monitor the absorbance for 5 minutes at 15minute intervals. Although this arrangement introduced a small time lag (l-2 min) between the conditions in the receptor compartment and the conditions in the membrane, the lag was considered small in comparison to the time required for the transitions that were being studied. Steady state fluxes reported in this paper are the mean values of three sets of data. Permeation runs were carried out at 30.0, 35.0, and 37.0% with electric fields of 0 and 38 V for each run. The temperatures were maintained by a Cole-Parmer Versatherm electronic temperature controller. Stirrer speeds were set at the same level (200 rpm) for all the runs. Permeation runs were carried out at pH 7.0, 5.0, and 6.1 (unbuffered). The pH of both compartments was regulated using a stock HPOi-/H,POgbuffer solution. The temperature was maintained at 30°C and the voltages set at 0 and 38 V for all cases. Experiments were also performed in which the field was turned off 1.0, 2.0, and 3.5 hours after the start of the run, respectively. The collection of permeation data was continued after the field was turned off to ascertain the duration of the effects caused by the field. Finally, a set of data was collected in which the polarity of the electric field was reversed after 1.0,2.0, and 3.5 hours, respectively. The cholesteric pitch of the liquid crystal was measured by estimating the distance between successive retardation lines under the microscope. (The distance between any two successive lines is equal to one-half of the cholesteric pitch.) Cholesteric pitch values were determined for liquid crystals prepared from PBLG of molecular weights of both 310,000 and 28,000, and were found to be 32 urn and 17 pm, respectively. Results and discussion This section presents permeation data for the two permeants through the system, and a discussion of its significance. The effect of an applied electric field was examined for both a neutral permeant, phthalyl alcohol, and a cationic permeant, methylene blue. Referring to Fig. 3, the steady state flux was taken to be the slope of the linear portion of the absorbance versus concentration curve for the receptor compartment of the membrane system, and the diffusional lag time was given by the intercept of the linear portion projected on to the x-axis. For both permeants, the steady state fluxes were observed to increase and the diffusional lag times to decrease. As shown in Fig. 4, the steady state flux increased from 1.33 X 10W2 mg/cm’-min at 15 V to 2.84 X lob2 mg/cm’min at 35 V for methylene blue, with an abrupt change occurring around
88
31 V. Similarly, as shown in Fig. 5, the steady state flux for phthalyl alcohol increased from 4.3 X 10m2 mg/cm2-min at 15 V to 7.1 X 10d2 mg/cm2-min at 35 V with the abrupt change also occurring between 28 and 35 V. Diffusional lag times were seen to decrease from 66 minutes to 37 minutes for methylene blue, and from 43 minutes to 21 minutes for phthalyl alcohol,
0 Time,
Fig. 3. Appearance run.
of permeant
in the receptor
20.0 DC
Fig. 4. blue.
Steady
state
flux
minutes
Electric
as a function
compartment
during a typical
permeation
field for the permeant
methylene
30.0 field,
V
of DC electric
89
DC
Fig. 5. Steady alcohol.
state
flux
30 field,
20 Electric
as a function
of
DC
V
electric
field
for
the permeant
phthalyl
as shown in Tables 1 and 2, respectively. Again, the most abrupt changes occurred between 28 and 35 V for both permeating species. A simple Fickian model such as the one used by Mohadger [7] may be used to evaluate diffusivities in the system. In such a system, the steady state flux through the system is given by J = -PA C/l where J is the steady state flux, AC is the concentration difference between donor and receptor compartments, and 1 is the thickness of the membrane. P, the membrane permeability, is defined for the purposes of this work to TABLE Effect Field 0 15 31 35 38 45 50
I of electric
(V)
fields
on steady
state flux
Lag time (min)
S.S.F.
66.2 + 2.7 65.0 + 2.6 36.5 f 1.65 35.2 * 1.60 37.2 k 1.71 36.5 * 1.50 38.0 i 1.91
1.33 1.33 2.28 2.35 2.82 2.81 2.84
( lo2 mg/cm*-min) + + t i r f +
0.06 0.06 0.08 0.09 0.14 0.11 0.14
Permeant: Methylene blue. No. of observations = 3; values are + standard
error of the mean.
90
TABLE 2 Effect of electric fields on steady state. flux Field (V)
Diffusional lag time (min)
Steady-state flux ( lo2 mg/cm2-min)
0.0 15.0 30.0 33.0 35.0 38.0 40.0 50.0
43.2 41.5 41.1 23.8 21.4 21.9 23.4 21.4
4.27 4.21 5.01 7.15 7.21 7.36 7.28 7.26
f f + * f + * k
1.94 1.95 1.89 0.98 0.90 1.03 0.98 1.24
+ f * + i i i f
0.19 0.20 0.23 0.29 0.30 0.34 0.30 0.42
Permeant: Phthalyl alcohol. Donor compartment: 0.1436 mg/ml phthalyl alcohol. Receptor compartment: 1000 ml distilled water/PO:- buffer, pH = 7.0. Temperature: 30°C t 0.1%. Membrane: 1 mm cholesteric PBLG, mol. wt. 310,000. No. of observations = 3; values are * standard error of the mean.
be the product of the solubility coefficient, coefficient, Da,, :
s, and the average diffusion
P=SD, This model reveals an increase in the average diffusivity from 3.6 X lo5 cm’/sec at 0 V to 7.6 X lo5 cm*/sec at 35 V for methylene blue and from 7.2 to 9.5 X lo5 cm’/sec over the same voltage range for phthalyl alcohol. The observed changes in lag time between no field and the above critical voltage were also consistent with a change in the diffusivity. Diffusion coefficients were not calculated using the lag times since this method is generally not reliable for the absolute determination of diffusivities. In this system, it is seen from an examination of Fig. 2 that the nematic directors of the liquid crystalline membrane would lie perpendicular to the plane of the membrane itself, i.e., parallel to the direction of transport. A transition within the membrane from the cholesteric to the nematic mesophase would thus be expected to facilitate transport across the membrane. It is contended that the increase in the diffusion coefficient was due to the parallel alignment of the polymeric species in the membrane as a consequence of the formation of the nematic mesophase. It is speculated that this alignment restricts the freedom of a permeant molecule in all directions but one - the direction of the alignment of the polymeric species. A crude analogy to this situation would be pinball machine in which the channels for the balls were all parallel to the gradient of the board. It is easily visualized that a ball in such a machine would reach its destination much faster than in a machine where it would have to travel longer distances due to the random orientation of the channels.
91
In order to show that the permeability increases were not caused by nonphase transitional electrotropic effects or related boundary phenomena, permeation runs were performed in which the field was interrupted and in which the polarity of the DC electric field was reversed. Enhanced steady state fluxes (2.3-2.8 X lo-* mg/cm*-min) were maintained for up to 2.3 hours after the field was interrupted, as shown in Fig. 6. When the polarity of the electric field was reversed during the permeation run, very little effect on the steady state flux was noted once the enhanced fluxes were triggered. No statistical difference in the steady state flux could be seen. These effects were observed for both neutral and ionic permeants.
I
I
J
0 Fig.
30
6. Appearance
of methylene
60 Time.
90 m!?utes
blue in the receptor
120
compartment
150
after
field interrupt
at 90 min.
The concentration of PBLG in the membrane was varied so as to connect the liquid cyrstalline behavior of the polymer with the changes in permeability of the membrane. The normalized change in the steady state flux, p, was studied as a function of the concentration of PBLG in the membrane. fl was defined as S.S,F.31v
P=
- S.S.F.oV
S.S.F.ov
The subscripts are the voltages selected for the evaluation of p, (0 V and 31 V), and correspond to the pre- and post-transitional field strengths, respectively. The Flory critical concentration for PBLG in dichloromethane (i.e., the concentration at which the formation of the cholesteric mosophase is just complete) is around 23% w/v [ 131. The value of $ was seen to increase from 0.0793 at 15% w/v to 0.5284 at 25% w/v, as shown in Fig. 7. There appears to be a critical concentration below which no electric field induced
92 0.60
0.48 ;; : c -$ 0.36
-
:
.-E t
0.24
0.1 2
0.00
i ;J1, , , / , [ /
ti
m
: 0
I
1
20
10
Fig. ‘7. Fractional the membrane.
change,
8, in steady
40
30
Concentration,
w/v
state flux
%
as a function
of PBLG
concentration
in
permeability changes take place. Above this concentration, however, the changes in permeability are rather dramatic. This effect was observed for both neutral and ionic permeants. The existence of cholesteric and nematic mesophases at field strengths of 0 and 380 V/cm, respectively, was confirmed by observation through an optical microscope at a magnification of 150-200X. The time lag for the reversion of the nematic mesophase to the cholesteric was estimated by viewing the fractional area of the respective textures under the microscope, and was found to be 3.3 hr. This time lag matched the corresponding (3.4 hr) lag for the reversion of normal steady state fluxes within the membrane after field removal as discussed above. Photomicrographs of the cholesteric and the nematic mesophases are shown in Figs. 8 and 9, respectively, The cholesteric pitch of the liquid crystal was measured by direct visual observation (method of Duke and DuPre [14] ) for PBLG of molecular weight 28,000 and 310,000, respectively. Theoretical field strengths for the cholesteric-nematic transition were calculated by means of the deGennes relation [ 151 as follows: E, = (7?/2)
(47TK/e~)l’~
(l/p,)
where eA is the dielectric anisotropy of the medium, K is the twist elastic constant, and P,, is the cholesteric pitch of the liquid crystal. Table 3 shows the theoretical field strengths for the cholesteric-nematic transition in comparison to the actual field strengths required to trigger permeability enhancement. In both cases, agreement is within +15 V/cm. Tables 4 and 5 show the effect of temperature and pH on the steady state flux through the system. Lobe1 and Caplan [2] have shown that the presence
Fig.
PBLG.
9. Nematic mesophase of PBLG.
94
TABLE 3 DeGennes and experimental
critical field strengths
DeGennes field strength
Cholesteric pitch
Experimental
(V/cm) 17.0 32.0
field strength
W/cm)
593.0 380.0
610.0 368.0
These experimental values are derived from other standard error is given.
experimental
data. Therefore,
no
TABLE 4 Effect of temperature Temperature 30.0 35.0 37.0
(“C)
on steady state flux
S.S.F. (0 V)
S.S.F. (38 V)
(1.33 f 0.07) x 10-a (1.39 ? 0.06) x lo-’ (1.43 i 0.09) x 10-z
(2.82 i: 0.15) x lo-* (2.85 * 0.11) x lo-’ (2.93 + 0.17) x 1o-2
Permeant: Methylene blue. No. of observations = 3; values are + standard error of the mean. TABLE
5
Effect of pH on steady state flux PH
S.S.F. (0 V)
S.S.F. (38 V)
7.0 6.1 5.0
(1.33 (1.41 (1.52
(2.82 i 0.15) x 10-l (3.01 f 0.07) x lo-* (3.23 f 0.06) x 1O-2
+ 0.07) x lo-* + 0.04) x lo+ r 0.03) x lo-*
Permeant: Methylene blue. No, of observations = 3; values are + standard error of the mean.
of protons in the boundary layer can affect the transport properties of solvent type membranes. A decrease in the steady state flux with pH was also observed; however, the number of data points was too few to be conclusive of any trend. In any event, the observed decreases were considerably less than the observed field-induced changes. Consequently it was assumed that boundary layer effects of the type discussed by Lobe1 et al. were not induced by the electric field. The changes in steady state flux with temperature were consistent with the dependence of the diffusion coefficient on temperature, which typically follows an Arrhenius-type relationship. Conclusions The results presented above show that the enhancement of the permeability of the membrane due to an externally applied electric field is consistent
95
with a cholesteric-nematic phase transition within the membrane, and that non-phase transitional electrotropic effects and related boundary layer phenomena were absent as dominant factors. This assertion is based on the following reasons: (1) There was no reason to believe that non-phase transitional electrotropic effects would persist after the electric field had been turned off or when the polarity of the field had been reversed. (2) Other electrotropic effects would cause differences in permeation characteristics between neutral and ionic permeants. The differences that were observed were small enough to be ascribed entirely to permeabilities based on different solubilities and diffusion coefficients. (3) The abrupt changes in p, the normalized steady state flux, with the composition of the membrane offer additional evidence in favor of the hypothesis that a liquid crystalline phase transition was responsible for the electrically induced changes in the steady state flux. The concentration at which fl underwent the largest increase agreed to within 2% of the Flory critical concentration for PBLG in dichloromethane. (4) Theoretical critical field strengths were calculated from experimentally obtained values of the cholesteric pitch by means of the deGennes relation [ 141. The critical electric field strength for the cholesteric-nematic phase transition is directly proportional to the dielectric anisotropy and the twist elastic constant of the liquid crystal and inversely proportional to the cholesteric pitch. The theoretical field strengths so obtained agreed well with actual field strengths required for the permeability enhancement. (5) Finally, optical studies on the PBLG-dichloromethane system revealed that the onset of the nematic phase of PBLG coincided with the onset of the increase in steady state flux. Furthermore, the time lag for the resumption of normal steady state fluxes agreed well with the time lag for the reversion of the nematic mesophases in the optical samples to the cholesteric mesophases. Acknowledgments Technical assistance from the staff of INTER, Research Corporation, Lawrence, Kansas is gratefully acknowledged. This work was funded in part by a grant from INTER, Research Corporation, Merck Sharp and Dohme Research Laboratories. References 1 2
3
N. Lakshminarayanaiah, Transport Phenomena in Membranes, Academic Press, New York, N.Y., 1969, p. 129. C. Lobe1 and S.R. Caplan, Unstirred layer controlled transport of organic bases and drugs through solvent-type membranes and the influence of bound enzymes, J. Membrane Science, 6 (1980) 203-220. A.N. Cherkasov, V.A. Pasechnik, V.P. Zhemkov, V.V. Chechina and R.S. Alimardanov, Effect of direct current on the permeability of ultrafiltration membranes, Kolloid. Zh., 41( 3) (1979) 586-590.
96
8
9
10 11
12
13 14
15
W. Prins, J.C. Chuang and M. DeSorgo, Mechanochemistry of photochromic model membranes, J. Mechanochem. Cell Motility, 2 (1973) 105-112. R. Langer, J. Folkman and D.S.T. Hsieh, Magnetic modulation of release of macromolecules from polymers, Proc. Natl. Acad. Sci. U.S.A., 78(3) (1981) 1863-1867. C.E. Rogers, Structural factors, in: D.F. Paul and F.W. Harris (Eds.), Controlled Release Polymeric Formulations, Am. Chem. Sot., Washington, DC, 1976, p. 15. Y. Mohadger, A.V. Tobolsky and T.K. Kwei, Permeation of isomers of mandelic acid through isotropic and cholesteric fluid polypeptide membranes, Macromolecules, 7(6) (1974) 924-929. A.J. Grodzinsky and S.R. Eisenberg, Electric field control of membrane permeability, in: Proceedings of the International Conference on Biomedical Engineering, New York, 1980, pp. 237-241; also S.R. Eisenberg, S.M. Thesis, Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 1979. H. Finkelmann, The synthesis, structure, and properties of polymers with liquid crystalline side chains, in: A. Ciferri (Ed.), Polymeric Liquid Crystals, Academic Press, New York, N.Y., 1982, pp. 35-62. F.D. Saeva, Nematic liquid crystals, in: Liquid Crystals: The Fourth State of Matter, Marcel Dekker, New York, N.Y., 1979, p. 75. C. Guha Sridhar, W.A. Hines and E.T. Samulski, Polypeptide liquid crystals: Magnetic susceptibility, twist elastic constant, rotational viscosity coefficient, and polyr-benzyl-l-glutamate side chain configuration, J. Chem. Phys., 61(3) (1974) 947953. T. Kajiyama, A. Kumano and 0. Niwa, Photoinduced ion permeation through a ternary composite membrane composed of polymer/liquid crystal/azobenzenebridged crown ether, Chem. Lett., (1983) 1327-1330. P.J. Flory, Phase equilibria in solutions of rod-like particles, Proc. Roy. Sot., Ser. A, 234 (1956) 73. R.W. Duke and D.B. DuPre, Examination of the electrotropic cholesteric to nematic phase transition of liquid crystal poly(r-benzyl-l-glutamate) by laser light beating spectroscopy, Macromolecules, 3 (1974) 374-376. P.G. deGennes, Long range order and thermal fluctuations in liquid crystals, Mol. Cryst. Liq. Cryst., 7 (1969) 325-345.