Pore size distributions in polyelectrolyte multilayers determined by nuclear magnetic resonance cryoporometry

Transcription

1 THE JOURNAL OF CHEMICAL PHYSICS 126, Pore size distributions in polyelectrolyte multilayers determined by nuclear magnetic resonance cryoporometry Fabián Vaca Chávez a and Monika Schönhoff b Institute of Physical Chemistry, University of Münster, Corrensstrasse 30, D Münster, Germany Received 27 November 2006; accepted 16 January 2007; published online 14 March 2007 Polyelectrolyte multilayers PEMs are thin films, which are assembled one molecular layer at a time, by alternatingly adsorbing polycations and polyanions making use of their attractive electrostatic interaction. Since the porosity of PEMs is one of the properties of major interest, in the current work the first pore size distribution of PEMs in samples consisting of silica particles coated with poly allylamine hydrochloride and poly sodium 4-styrenesulfonate is presented. To this end, the nuclear magnetic resonance NMR cryoporometry technique was applied. The proton NMR signal of liquid water is analyzed assuming a log normal distribution of motional correlation times. From the results, it is possible to determine the size of water sites in the layers to around 1 nm. In addition, a slight variation with the number of layers is found. The average pore size agrees with cutoff sizes found in permeation experiments American Institute of Physics. DOI: / I. INTRODUCTION Since the possibility to adsorb layers of oppositely charged polymers with molecular-level control over the resultant multilayer structure was shown by Hong and Decher, 1 building polyelectrolyte multilayers PEMs has triggered a huge interest due to their application potential. Polyelectrolyte multilayers are thin films, which are assembled one molecular layer at a time by taking advantage of predominantly electrostatic attractive interactions between the constituents. Prominent applications are in the area of biomaterials with particular applications in drug delivery, biosensors, and also for technical applications in optics and chemical technology. In the area of drug delivery, polyelectrolyte multilayers coated onto drug microparticles have been shown to prolong the release time of the drug. 2 Permeability and porosity of polyelectrolyte multilayers are properties of major interest, as they are crucial for applications in encapsulation, controlled release, and membrane separation. 3 5 The permeability of the multilayer towards drug molecules could be controlled through changes in the number of layers, the polymers employed, and the ph and salt conditions used during multilayer assembly. 6 8 In consequence, the understanding of the internal molecular properties of polyelectrolyte multilayers are of huge interest nowadays as described in several reviews In particular, the distribution of water sites in the layered polyelectrolytes is an active current topic of research. For instance, the water mobility in PEMs was monitored by spin relaxation, where, for example, a pronounced odd-even effect of water mobility in dependence on the number of layers was found. 13 A number of reflectivity studies have dealt with swelling properties and water content in a Author to whom correspondence should be addressed. Electronic mail: fvchavez@uni-muenster.de b Electronic mail: schoenho@uni-muenster.de PEMs Furthermore, Jin et al. as well as Liu and Bruening have investigated pore sizes by permeation experiments of different probe molecules through freestanding PEM membranes. A cutoff of the permeability is found for probe sizes on the order of nanometers. 4,5 On the other hand, nuclear magnetic resonance NMR cryoporometry is an established technique to determine pore size distributions in different porous materials It makes use of the fact that small crystals of a liquid in pores melt at a lower temperature than the bulk liquid. In this method the porous material, saturated with a suitable liquid, is cooled until all liquid is frozen. Then, the sample is heated, and the crystals formed within smaller pores will melt first 25 obtaining a NMR signal that increases with increasing temperature. In contrast to differential scanning calorimetry, one does not detect the actual melting as manifested by heat transfer but instead measures the fraction of water in the liquid state in dependence on temperature. NMR can be used to quantify the liquid fraction, using the large discontinuity in spin-spin relaxation time, T 2, at the freezing point to discriminate solid from liquid. This measurement is performed by the spin-echo pulse sequence with appropriate echo times. The spin-echo NMR experiment detects the NMR signal from the nuclei in mobile molecules because their magnetization does not relax to zero under the selected echo time. On the other hand, the NMR signal from the nuclei in immobile molecules is effectively canceled because the transverse spin relaxation rate of those nuclei is large and thereby their magnetization relaxes to zero during the echo time. NMR cryoporometry has been successfully applied to various porous systems such as silica, 19,20,24,25 mesoporous MCM-41, 22,27 and even complex materials such as cement. 28 The pore size distributions obtained there agree very well with those obtained by other techniques such as N 2 sorption. 19,20,24, /2007/ /104705/7/$ , American Institute of Physics

2 F. Vaca Chávez and M. Schönhoff J. Chem. Phys. 126, In our approach, we consider polyelectrolyte multilayers as a porous material, where the hydration water can freeze in dependence on the size of water pores. Applying the NMR cryoporometry technique we are able to extract the pore size distribution in the PEMs. To our knowledge, so far no other technique has been applied to polyelectrolyte multilayers, which can provide pore size distributions. II. MATERIAL AND METHODS A. Materials and sample preparation Polymer materials poly allylamine hydrochloride PAH, M w = g mol 1, poly sodium 4-styrenesulfonate PSS, M w = g mol 1, and poly diallyl dimethyl ammonium chloride PDADMAC, M w = g mol 1, are purchased from Aldrich. PAH and PDADMAC are used without further purification, while PSS is dialyzed M w cutoff of g mol 1 before use. Negatively charged silica particles diameter 450 nm are purchase from Polysciences Inc. The water used in all experiments is prepared in a Milli-Q water purification system and has a resistivity higher than 18.2 M cm. NaCl AR grade is obtained from Merck. The polymers are dissolved in 0.5M solution of sodium chloride NaCl without any adjustment or change of ph. The polyelectrolyte multilayer assembly is accomplished by adsorption from solutions containing 0.2 wt % of polyelectrolyte. This concentration is far above that required for saturation of the surface. Adsorption and subsequent washing cycles are carried out with the method of centrifugation, a procedure described earlier. 10 The concentration of the SiO 2 particle dispersions is kept about 2 wt % to reduce bridging coagulation to a minimum while coating. The positively charged polyelectrolyte PDAD- MAC is allowed to adsorb to the negatively charged silica particles for 15 min under stirring and 5 min in ultrasound. The procedure is repeated with PSS and PAH to obtain alternating layers, until a total of the desired number of layers is deposited. PDADMAC is used only as a first layer, since this polyelectrolyte forms more stable layers. Successful deposition of the PEMs is monitored measuring the sign of the surface determined by the potential after each deposition step using the Zetasizer 4 Malvern Instruments. In addition, the particle diameter is determined by dynamic light scattering with a Zetaziser 300HSA Malvern Instruments. In the current study, we show the results of five different samples: bare silica particles, coated particles with seven layers PDADMAC- PSS/PAH 3, eight layers PDADMAC - PSS/PAH 3 -PAH, nine layers PDADMAC- PSS /PAH 4, and ten layers PDADMAC- PSS/PAH 4 -PAH, respectively. The samples are described by n, the number of layers. After the coating procedure, water is removed by freeze-drying obtaining a fine powder. Starting from a dry powder of coated or uncoated silica particles, respectively, water is added until a water fraction of 70 wt % is reached. After sample preparation the potential and particle size are again controlled to confirm well defined particles. B. NMR measurements The 1 H measurements are carried out at 9.4 T on a Bruker DMX 400 spectrometer. The sample temperature is regulated and stabilized to within ±0.5 K by means of a BVT 3000 Bruker temperature-control unit using a flow of cold nitrogen. Calibration curves are obtained for the temperature control unit using ethanol which freezes at 159 K. During the calibration procedure the temperature is monitored by inserting a digital Pt100 thermometer GMH 3710 in a 5 mm NMR tube into the probe. Each sample is first cooled to around 190 K and remains there for 1 h until the liquid is frozen to prevent any complication by undercooling. Then the samples are warmed slowly with a rate of 6 K/h. At the temperature values chosen for measurements, the sample is first equilibrated for 15 min, then the NMR signal of the nonfrozen water was measured. This procedure was repeated at increasing temperatures until all the ice was melted. The experiments are performed using the 1 H spin-echo pulse sequence, where =500 s. This value ensures that only the signal from nonfrozen water will be measured. At each temperature the echo signal corresponding to the liquid water is measured while the decay of the magnetization corresponding to the nonfrozen water will be negligible. At each temperature the 90 pulse is readjusted between 13 and 14 s and the probe is retuned to compensate for frequency shifts of the electronic oscillator. III. THEORETICAL BACKGROUND According to the Gibbs-Thompson equation a crystal of linear dimension a will melt at a temperature T a lower than the bulk material melting point T o, by an amount T a. Assuming a spherical shape, T a is inversely proportional to a T a = T o T a = k with k = 4 T m a H. 1 Here, k depends on the properties of the liquid and liquid/solid interface. is the surface energy of the solidliquid interface, is the density of the frozen water, and H is the bulk enthalpy of fusion. A polydisperse size distribution of the pores leads to a polydisperse distribution of correlation times, as the dynamics in each pore depend on its size. Following the model introduced by Overloop and Van Gerven, 29 the measured liquid state NMR signal can be analyzed assuming a log normal distribution of motional correlation times of the molecules, P, P d = 1 exp Z2 B B 2, Z =ln *, 2 where B characterizes the width of the distribution and * is the center of the distribution which represents an average of correlation times of water. * decreases when the temperature increases. In this model the temperature dependence of * is assumed Arrhenius-type. The measured intensity, I T, at a certain temperature is

3 Pore size distribution determined by NMR cryoporometry J. Chem. Phys. 126, FIG H NMR Hahn echo intensity vs inverse temperature for water in dispersions of particles coated with ten layers. The inset shows the range in which the free water is frozen and the NMR signal is due to water liquid inside the pores. FIG H NMR Hahn echo intensity vs inverse temperature of liquid water in bare silica particles and polyelectrolyte multilayer dispersions. The solid lines represent nonlinear fits to the experimental data; the parameter values are listed in Tables I and II. I = 0 c P d, where C is a critical correlation time. For water molecules with C the NMR signal is not detected due to their short T 2. If more than one pore size mode is present, then the total intensity as a function of the temperature can be written as a superposition of N modes 22 N I T = I 0i 1 + erf 1 i=1 2 i 1 1 T Ci T, where I 0i, T Ci, and i RB i / 2 Hi are the intensity, the transition temperature, and a parameter describing the width of the melting temperature distribution curve of mode i, respectively. erf x is the error function defined as erf x = 2 x exp u 2 du. For a sharp transition, i.e., a monodisperse pore size distribution, i approaches zero. IV. RESULTS Figure 1 shows the proton NMR spin-echo intensity of the liquid water as a function of the inverse of temperature for coated silica particles containing ten layers in the range of 190 K T 273 K. An abrupt increase in intensity at the melting point of bulk water, T o, is observed. The intensities are normalized using the fact that at T o all pore water and free water melt. The inset shows an expanded view for the range of T 273 K in which the free water is still frozen and the signal is only due to water inside the pores. The total amount of water inside the pores is three orders of magnitude smaller than the amount of free water, which agrees with the volume fraction in the sample occupied by the layers. 3 The intensities measured at an absolute temperature T are corrected according to Curie s law, it means, by multiplication by the factor T/T o. This accounts for the temperature dependence of the occupation of the spin levels under the assumption of a linearized Boltzmann distribution. In Fig. 2 the proton NMR spin-echo intensity of liquid water in bare silica particles and polyelectrolyte multilayer dispersions is shown as a function of the inverse of temperature. The smooth increase in the signal intensity with rising temperature is the result of the gradual melting of the frozen liquid in the pores. These pores can be present on the silica particle surface and/or in the PEMs. Uncoated silica particle dispersion is used as a reference in order to detect any potential contributions arising from the particles. The nonzero signal observed for the uncoated silica particles demonstrates a certain porosity of the template. Then, data of coated particles have to be corrected by this contribution to obtain the porous properties of multilayers. The total liquid intensity below T o in PEMs is much larger in coated dispersions; thus, there is a substantial contribution from pores in PEMs. No major differences in the shape of the curves are found between different samples, neither in dependence on the number of deposited layers nor in dependence on the sign of the terminating layer. TABLE I. Numerical values of the parameters I 0i, T Ci, and i obtained by fitting Eq. 3 to the intensity data of bare silica particle dispersion. The errors represent fitting errors only. Parameter Bare silica particles I ± K 1 8.7± /T C1 K ±0.02 I ± K 1 2.2± /T C2 K ±0.02

4 F. Vaca Chávez and M. Schönhoff J. Chem. Phys. 126, TABLE II. Numerical values of the parameters I 0i, T Ci, and i for coated particles obtained by fitting Eq. 3 to the experimental data. The errors represent fitting errors only. Parameter Seven layers Eight layers Nine layers Ten layers I ± ± ± ±0.003 I ± ± ± ±0.003 I ± ± ± ± K 1 3.6± ± ± ± /T C3 K ± ± ± ±0.02 A. Uncoated silica particles To describe the I T curve for uncoated silica, two modes N=2 in Eq. 3 are necessary to fit the experimental data by using nonlinear least-squares data fitting. Numerical values of the relative intensity I 0i, the width i, and the inverse of the transition temperature 1000/T Ci of the ith type of pore are presented in Table I. Two depression temperatures are found indicating that the pores arising from the silica particles can be described by a bimodal distribution. The values of I 01 and I 02, which are similar, indicate similar contributions to the I T curve. B. Coated particles In order to describe the results from the coated particles, three modes are necessary N=3 in Eq. 3. The third mode occurs in addition to the pores in the silica and can be attributed to pores in the PEMs. For the fit with N=3, we fixed the parameters T C1, 1, T C2, and 2 to the values found for the two modes in uncoated silica. Furthermore, the ratio between I 01 and I 02 is fixed. Thus, four adjustable parameters are remaining to fit each I T curve for coated particles. The obtained parameters are listed in Table II and the solid lines in Fig. 2 represent the fit according to these values. The reasonable agreement between experiment and theory, demonstrated in Fig. 2, indicates that the assumption of the model is well justified. However, the real test of the model is not just the quality of the fit but the values of the parameters derived from it. In Fig. 3 the different contributions to the total measured I T curve arising from the water present in the silica particle pores and from water in the multilayers are illustrated. The solid line is the sum of these two contributions. From the FIG H NMR Hahn echo intensity of liquid water vs inverse temperature for coated particle dispersions. Spheres: Experimental data points. Solid lines: nonlinear fits to the experimental data. Dashed lines: contributions from the pores in the silica particles ---- and from the pores in the PEMs.

5 Pore size distribution determined by NMR cryoporometry J. Chem. Phys. 126, FIG. 4. Pore size distribution curves derived from Eq. 4 corresponding to seven, eight, nine, and ten layers. Solid line: pore size distribution extracted from Eq. 4. Dashed lines: pore size distribution in silica particles ---- and in PEMs. relative intensities I 0i we observe that the signal from water in the silica pores is smaller compared with the one from water in the multilayers. The melting temperatures T C3 are very similar to one of the pore modes of the silica particles mode 2 ; however, since the contribution from the PEMs is dominating the total signal, the procedure of subtraction of the silica modes will not lead a major error in separating the modes. All samples of coated particles can thus be described by one single pore size mode occurring in the multilayers. The average pore size is slightly dependent on the number of layers. The decreasing trend of 1000/T C3 with increasing number of layers, n, implies a slight increase of the pore size. The width of the distribution, however, remains constant. C. Pore size distributions An analytical function which describes the pore size distribution is di/da vs a, where a is the pore size. This function is a simple sum of Gaussians centered at 1/T Ci and is obtained from Eqs. 1 and 3, N di 2k da = I0i ato exp 1 k 2 i=1 i 2 i 2 1 a 2. T Ci at o k 4 The value of k corresponding to water in Eq. 1 was obtained experimentally for different porous materials, for instance, K m in zeolite, Km in silica, 31 and K m in silica gel. 32 Assuming k= K m as a value for the entire system silica particles and PEMs, we obtain the pore size distributions shown in Fig. 4. Using the parameters listed in Tables I and II, different modes of pore sizes can be extracted. For the silica particles the pore sizes are around 1.1 and 1.8 nm, corresponding to the two depression temperatures found for this sample. For the PEM samples the size distribution has a maximum at about 1 nm and this mode can be attributed to pores present in the layers. In spite of the depression temperatures T C3 are slightly different, 227 K for seven layers and 234 K for ten layers, the maximum of the distributions from Eq. 4 are 1 nm for seven and eight layers and 1.1 nm for nine and ten layers. However, from inserting the center of the distribution Eq. 3, T C3 into Eq. 1, we found center pore sizes around 1.2, 1.3, 1.4, and 1.5 nm for n=7, 8, 9, and 10, respectively. The difference between maximum values and center values can be attributed to the width and the asymmetry of the pore size distribution. V. DISCUSSION The I T curves obtained for coated and uncoated silica particles are well represented by a sum of error functions

6 F. Vaca Chávez and M. Schönhoff J. Chem. Phys. 126, arising for the assumed log normal distribution of correlation times. The corresponding melting point distribution curves show three transition points that can be attributed to the depressed melting point of water in silica pores two modes and one corresponding to water in the multilayers. For all investigated PEMs samples, no liquid state signal at very low temperatures T 190 K is detected. Thus, a nonfreezing layer of surface water with a thickness of one to three water layers, which had been observed by Overloop and Van Gerven in silica, 29 is not detected here. The width 3 of the mode describing the porosity in the PEMs is the same, within the experimental errors, for all samples indicating that the distribution of pore size is an intrinsic property of PEMs which do not depend on the number of layers. This parameter is about a factor of 4 larger than the width obtained in a well defined porous size material such as MCM-41. 3,27 Bearing in mind that PEMs are a disordered material, where, for example, chains of subsequent layers are interpenetrating, this width of the distribution is fairly small. The even-odd effect, which was observed for water dynamics in particles coated with PSS/PAH, 12 had evidenced a water mobility depending strongly on the sign of the terminating layer. Such an odd-even effect variation is not observed in the pore size. It is possible that the different temperatures employed in either experiment are essential; i.e., water mobility at room temperature is not comparable to water mobility very close to the freezing point in nanopores. Despite that our system does not have a controlled pore size, from the experiments it is possible to determine the size of water sites between the layers to around 1 nm. With increasing n, the center of the distribution is shifted to slightly larger sizes. This could be a consequence of the first layers, which are close to the silica surface, being slightly more closed packed. It was shown before that the outermost few layers in a multilayer arrangement are less densely packed than the internal ones. 33 In addition, the hydration of the first layer adsorbed on silica was differing from the fifth layer, which bound a much larger amount of water. 12 It appears thus reasonable that with increasing layer number the average pore sizes are increasing. In order to compare our results with the ones obtained by other authors, we can say that recently Liu and Bruening 5 performed permeation experiments in samples consisting of seven bilayer PSS/PAH membranes and their results suggest that these films have pores with diameter of nm. In parallel, Jin et al. 4 investigated the size and charge selective transport of aromatic compounds across 60 bilayer PSS/PAH PEM films and the estimated mean pore size was 0.67±0.04 nm. It is interesting to note the agreement with our mean sizes, since the methods monitor porosity in completely different ways, while our experiment is an average over all pores, for permeation only paths interconnecting sufficiently large pores are relevant. The agreement can be understood as proof of a very homogeneous nanostructure of polyelectrolyte multilayers. We note that Eq. 1 holds for spherical and cylindrical pores. Since the true shape of the three-dimensional pores in PEMs is not known, it is an approximation, and the parameter k employed for calibration might differ due to the true pore shape. However, the agreement with permeation experiments is evidence of the quality of the approximation as spherical or cylindrical pores. VI. CONCLUSIONS We have presented an application of the nuclear magnetic resonance cryoporometry technique to determine the pore size distribution in polyelectrolyte multilayers by employing silica particles coated with seven, eight, nine, and ten layers of polyelectrolytes PAH/PSS. The proton NMR signal of nonfrozen water is analyzed assuming a log normal distribution of motional correlation times of the water molecules. From the results, the size of water sites between the PEMs layers is around 1 nm. With increasing numbers of layers, the maximum of pore size distribution is slightly shifting to larger values. This implies a less dense packing of layers which have a larger distance to the substrate. This value is consistent with permeation experiments carried out by other authors in samples consisting of planar layers of PSS/PAH membranes in which they suggest that these films have pores with diameter of nm. The present results therefore serve for understanding the permeability and porosity of polyelectrolyte multilayers which are properties of major interest, as they are crucial for applications in encapsulation, controlled release, and membrane separation. ACKNOWLEDGMENTS This work is funded by the Deutsche Forschungsgemeinschaft DFG within the German-French Network Complex Fluids: From 3 to 2 Dimensions, Proposal Nos. Scho 636/ 3-1 and Scho 636/ G. Decher, J. D. Hong, and J. Schmitt, Thin Solid Films 210, X. P. Qiu, S. Leporatti, E. Donath, and H. Möhwald, Langmuir 17, M. C. Berg, L. Zhai, R. E. Cohen, and M. F. Rubner, Biomacromolecules 7, A. Jin, A. Toutianoush, and B. Tieke, Appl. Surf. Sci. 246, X. Liu and M. L. Bruening, Chem. Mater. 16, A. A. Antipov, G. B. Sukhorukov, and H. Möhwald, Langmuir 19, G. Ibarz, L. Dähne, E. Donath, and H. Möhwald, Macromol. Rapid Commun. 23, C. Y. Gao, S. Leporatti, S. Moya, E. Donath, and H. Möhwald, Chem.- Eur. J. 9, G. Decher, Science 277, M. Schönhoff, Curr. Opin. Colloid Interface Sci. 8, M. Schönhoff, J. Phys.: Condens. Matter 15, R B. Schwarz and M. Schönhoff, Colloids Surf., A 198, B. Schwarz and M. Schönhoff, Langmuir 18, M. McCormick, R. N. Smith, R. Graf, C. J. Barrett, L. Reven, and H. W. Spiess, Macromolecules 36, R. N. Smith, M. McCormick, C. J. Barrett, L. Reven, and H. W. Spiess, Macromolecules 37, M. Lösche, J. Schmitt, G. Decher, W. G. Bouwman, and K. Kjaer, Macromolecules 31, R. Steitz, V. Leiner, R. Siebrecht, and R. von Klitzing, Colloids Surf., A 163, D. Carrière, R. Krastev, and M. Schönhoff, Langmuir 20, D. W. Aksnes, K. Førland, and L. Kimtys, Phys. Chem. Chem. Phys. 3,

7 Pore size distribution determined by NMR cryoporometry J. Chem. Phys. 126, D. W. Aksnes and L. Kimtys, Solid State Nucl. Magn. Reson. 25, J. C. Dore, J. B. W. Webber, and J. H. Strange, Colloids Surf., A 241, D. Akporiaye, E. W. Hansen, R. Schmidt, and M. Stöcher, J. Phys. Chem. 98, J. Rault, R. Neffati, and P. Judeinstein, Eur. Phys. J. B 36, R. M. E. Valckenborg, L. Pel, and K. Kopinga, J. Phys. D 35, J. H. Strange and J. B. W. Webber, Meas. Sci. Technol. 8, C. L. Jackson and G. B. McKenna, J. Chem. Phys. 93, R. Schmidth, E. W. Hansen, M. Stocker, D. Akporiaye, and O. H. Ellestad, J. Am. Chem. Soc. 117, J. Boguszyńska and J. Tritt-Goc, Z. Naturforsch., A: Phys. Sci. 59, K. Overloop and L. Van Gerven, J. Magn. Reson., Ser. A 101, E. Hansen, M. Stocker, and R. Schmidt, J. Phys. Chem. 100, J. B. W. Weber, J. H. Strange, and J. C. Dore, Magn. Reson. Imaging 19, K. Ishikiriyama, M. Todoki, and K. Motomura, J. Colloid Interface Sci. 171, R. v. Klitzing and H. Möhwald, Macromolecules 29,