Investigation of swelling and diffusion in polymers by 1H NMR imaging: LCP networks and hydrogels

Journal of Molecular Structure 554 (2000) 69–79 www.elsevier.nl/locate/molstruc

Investigation of swelling and diffusion in polymers by 1H NMR imaging: LCP networks and hydrogels M. Kno¨rgen a, K.-F. Arndt b,*, S. Richter b, D. Kuckling b, H. Schneider a a

Department of Physics, Martin-Luther-Universita¨t Halle-Wittenberg, Friedemann-Bach-Platz 6, D-06108 Halle, Germany b Department of Chemistry, Technische Universita¨t Dresden, D-01062 Dresden, Germany Received 29 November 1999; accepted 23 December 1999

Abstract The kinetics of diffusion in polymers ranges from simple Fickian diffusion to higher-order diffusion, such as Case II diffusion. The conventional method for determining the characteristics of the diffusion of solvents into polymer matrices is by measuring the mass uptake of the polymer while the solvent penetrates the matrix. However, since such measurements perform observations at a macroscopic level, little information has been obtained relating to the properties of the solvent inside the polymer matrix and the mechanisms of the processes that control the diffusion. Nuclear magnetic resonance (NMR) imaging (MRI) is used to monitor the transport of solvents into solid systems in real-time. The method provides a one- or moredimensional image of the density and the mobility of the solvent in a material. Furthermore, the mobility of network chains itself changes due to the softening influence of the solvent. The former (imaging of the solvent) can be used for a quantitative measurement of the diffusion coefficient whereas the observation of the network gives information about changes in the network properties (mobility, decrystallization) during the swelling process. The potential applicability of the NMR imaging techniques to polymer gel systems is demonstrated on the diffusion of organic solvents in polymeric materials like natural rubber, nematic-like networks, and hydogels. 䉷 2000 Elsevier Science B.V. All rights reserved. Keywords: Diffusion; Kinetics of swelling; Aged rubber; Nematic-like network; Chain mobility; Hydrogels

1. Introduction Due to a close interaction between the solvent and the polymer matrix the kinetics of diffusion in polymers ranges from simple Fickian (Case I) to higher diffusion, such as Case II or non-Fickian diffusion [1,2]. Contrary to some conventional methods for the investigation of the diffusion process of solvents into polymer matrices (determination of the mass uptake, time-lag method and others) nuclear magnetic resonance imaging (MRI) provides information relating to the nature of the diffusion process in a * Corresponding author.

polymer matrix on a spatially resolved level with a lateral resolution of about 50–100 mm. In this paper we give some examples both of Case I diffusion in rubber-like material as well as abnormal diffusion observed in hydrogels at longer diffusion times and also an example of anisotropy diffusion observed in hydrogels at longer diffusion times as well as an example of anisotropy diffusion processes found in nematic LC-copolymers. The main feature of all these processes is the fact that only a spatially resolved measurement provides knowledge of the abnormal diffusion processes. For this, the method of NMR imaging has achieved success in the last decade [3,4]. For instance, these processes are

0022-2860/00/$ - see front matter 䉷 2000 Elsevier Science B.V. All rights reserved. PII: S0022-286 0(00)00560-3

70

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

Fig. 1. The principle of NMR imaging: the spatial resolution in r is obtained by a magnetic field gradient HG ; which shifts the resonance frequency v of the spectrum. The non-averaged dipolar interaction in solid-like samples broadens the signal. The NMR spectrum is the convolution of spatial distribution and point spread function.

influenced by inhomogeneities like surface layers (i.e. benzene in aged rubber [5]), polymer cracks, which allow a fast solvent uptake in relation to the matrix [1], or by a change of the type of diffusion in some regions of the sample (i.e. water in poly(N-isopropyl acrylamide) [6]), and by others. In the past few years a lot of research work has been done in application of the NMR imaging technique on the investigation of swollen networks, e.g. the image analysis of stress– strain in poly(methacrylic acid)/water [7], the analysis of the shrinkage by application of an electric field to a polyelectrolyte gel [8,9], and the self-diffusion of solvent and probe polymer [10,11] in a polymer gel. However, some experimental difficulties have prevented a routine application of the technique

until now. The reason is the large diversity of methods to obtain contrast in MRI. For example, the contrast may arise from the solvent concentration itself, which is influenced by diffusion or convection. In the equilibrium state the concentration of the solvent in a polymer network (degree of swelling) depends on the molecular weight of the network chain. So it seems to be possible to measure the distribution of cross-linking density inside the polymer sample, in a qualitative manner at least. To specify the parameter (e.g. T2, T1, chemical shift) for observation, a so-called filter sequence can be set before the imaging sequence is started. By this one can obtain images mainly showing the spin density of hydrogen atoms, the local mobility of a hydrogen atom (T2-weighted by a multi-echo experiment), or a picture reflecting the diffusion and convection by using a simple Hahn echo. On the other hand, the contrast can also arise from the polymer matrix itself. In this case, the diffusion process can be detected indirectly by measuring the changes in the polymer mobility due to the softening effects of the solvent. Furthermore, the change of composition of a mixed swelling agent can be followed and therefore the diffusion of a liquid into the swelling agent inside the gel. For all these cases, an example is given in this work.

2. Experimental 2.1. NMR imaging of diffusion The basic principles of imaging, first introduced by P.C. Lauterbur in 1973 [12], are the relation of a spatial coordinate (voxel) to a resonance frequency (see Refs. [13,14] for a more detailed description). This can be done by a magnetic field gradient HG (Fig. 1), which shifts the frequency by

v1 ˆ g …H0 ⫹ HG †;

Fig. 2. One-dimensional imaging pulse sequence used for a T2weighted (in dependence on the echo time t ) image. For other parameters (T2, T1, T1r ,…) a so-called “filter sequence” can be preset. Top: rf-pulse program on acquisition; and bottom: gradient pulse.

where v 1 is the resonance frequency of the signal, g the gyromagnetic ratio (in our case, g H is the gyromagnetic ratio for protons), H0 the static magnetic field, and HG the gradient field. The Hahn echo pulse sequence for a one-dimensional imaging measurement used here is shown in Fig. 2. The magnetization in the x–y plane, created

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

71

of about 100 mm [3]. For more liquid-like material (e.g. organic tissues) this can be reduced to 20 mm [3]. The measurements were performed on a self-made NMR microscope (actively shielded gradient coils up to 5 mT/cm) attached to a VARIAN unity 200 spectrometer. Fourier imaging with slice selection (about 500 mm slice thickness) processed the 2D images. 2.2. Magnetization decay, the T2-parameter Fig. 3. The NMR probe for the diffusion experiment. The rubber piece (thickness about 3 mm) is aged on both (top and bottom) sides, but is covered by the solvent only on the top.

by a 90⬚ radio frequency (rf) pulse, rotates in the laboratory system. Due to mainly dipolar interactions it dephases (loss of correlation) within a few ms (liquids) or ms (solids). The second pulse of 180⬚ refocuses the magnetization to the echo signal at 2t , which is acquired during the acquisition (AQ) period. A Fourier transformation transforms the signal into the frequency space, which corresponds to the local dimension. For the image construction in the two(2D) or three-dimensional (3D) case, different mathematical procedures (Fourier imaging, back-projection imaging) are used in dependence on the experimental situation (see Ref. [9] for a more detailed description). As shown in Fig. 1, the resolution is not only determined by the gradient power but also strongly influenced by the dipolar line width (⬎2 kHz) and the chemical shift distribution (for a rubber-like material, e.g. natural rubber at 60⬚C, ⬍1 kHz at v0 ˆ 200 MHz†: This leads to a resolution

To obtain some information about the nature of a material, “material parameter imaging” [15] was used. The echo time is changed during multiple measurement to obtain a decay of the magnetization in time, which is only influenced by the dipolar interactions in the case of a Hahn echo. However, for liquids a pure exponential decay (stochastic interaction due to Brownian motion) can be observed and the determination of a T2-parameter (T2 ˆ relaxation time of the transversal magnetization decay) is straightforward. Of course, if diffusion and/or convection occurs, they dominate the echo decay, if a gradient is used. (In principle, it is possible to separate the influence of diffusion also in NMR imaging by special pulse sequences [14]). On the other hand, the Hahn echo decay of more solid-like substances like rubber networks (or the polymer matrices) is non-exponential and cannot be described by a single parameter T2. The relaxation time T2 gives only a very rough estimation of the molecular mobility. It reflects the solid-like (shorter T2) or liquid-like (longer T2) nature of polymer networks. Using a multi-component fitprocedure for the first Gaussian-like part of the echo decay delivers some values (e.g. anisotropy parameter as a measure of the cross-linking density, part of intercross-linked chains), which give more information on the polymer matrix than a simple one-exponential fit. However, for imaging of a diffusion experiment in real-time a parameter-selective measurement can be too time consuming (as in our case) and a simple T2weighted experiment (Fig. 2) is preferred. 3. Results and discussion

Fig. 4. 1D image of the diffusion in aged rubber shown by the T2parameter of the network. The proton-free benzene-d6 gives no signal. The vertical axis of each picture is the time. Aging times at 170⬚C: (1) unaged rubber; (2) 55 min; (3) 100 min; and (4) 15 h.

3.1. Diffusion of organic solvent in aged natural rubber The thermal aging course of rubber samples

72

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

Fig. 5. Chemical structure of the LC-main-chain pre-polymer (above) and sketch of the cross-linking process (below).

exhibits a typical NMR relaxation behavior [16] shown by the change of the relaxation time T2. An aging front is formed by strong aging condition (high aging temperature in air and long aging time) [5]. To investigate the influence of a layer of aged rubber on diffusion (as an example for a macroscopic property), we measured diffusion profiles in a realtime procedure: the aged rubber pieces (carbonblack filled NR, aging at 170⬚C in air for different times, cf. Fig. 4) were placed in a tube and the solvent (benzene-d6) was poured on (Fig. 3). The pictures in Fig. 4 show the profiles (abcissa) and the time-course (ordinate) of the diffusion for the samples aged at different times. The beginning (2 min after the contact of the rubber with the solvent) is at the top of each picture, and the end of the measurement (after 30 min) is at the bottom of each picture. In view of the structural changes of the rubber due to the aging process, the images can be explained as follows [15]: at shorter aging times (55, 100 min) an aged surface layer characterized by a longer T2 (lower

cross-linking density) facilitates the diffusion and swelling. Cross-linking processes, which are caused by radical polymerization in the presence of oxygen, take place at very long aging times. A higher crosslinking density prevents swelling, but not the diffusion of a violet network structure (polymer breaks). 3.2. Anisotropic diffusion in nematic-like networks (cross-linked liquid crystalline main-chain polymer) The investigated nematic-like network was synthesized from a pre-polymer, which consists of statistically distributed segments of methylhydroquinone, sebacic acid, 4-hydroxybenzoic acid and itaconic acid. The cross-linking reaction [17]—induced and steered by the temperature—proceeds via the double bond of the itaconic acid segments [18,19] in a radical mechanism (Fig. 5). The anisotropy of this material (PES-MHC) was induced by mechanical stress during cross-linking. The fixed anisotropy of the structure leads to several

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

73

Fig. 6. Expansion ratio (parallel k and perpendicular ⬜ to the orientation of the network chain) vs. swelling time (T ˆ 25⬚C; swelling agent 1,1,2,2-tetrachloroethane).

non-typical network properties, because of which the system represents a model of an oriented network. For further details see Ref. [19]. The mechanical (stress– strain behavior, storage and loss modulus) and thermal (low-temperature thermal conductivity) properties depend on the chain orientation. The stress– strain measurements parallel to the direction of chain orientation above the glass transition temperature show a pronounced yield point and remind one of the stress–strain behavior of thermosetting resins above Tg. The stress–strain curves can be described with a model, which combines elasticity, yield phenomena and plasticity [20]. The molecular weight of the network chains was estimated with compression

Fig. 7. Experimental setup for the investigation of the swelling process in dependence on the time (nematic-like network PESMHC).

measurements to Mc ⬃ 5 × 10 4 g/mol. The volume degree of swelling in 1,1,2,2-tetrachloroethane was about 7. It is known that the swelling behavior reflects very strongly the anisotropy of a polymer. The expansion ratio perpendicular to the orientation direction is more than 1.5 times the expansion ratio parallel to the chain orientation at longer swelling times (⬎300 min) (Fig. 6). The swelling of this network is a two-step process: first, the penetration of the swelling agent into the fibrous network structure (with a low increase of the degree of swelling, first diffusion coefficient D1 ˆ 8:8 × 10⫺8 cm2 =s†; and second, the ten-fold quicker process of expanding of the network chains, generally called as the “free swelling process” with D2 ˆ 4:8 × 10⫺7 cm2 =s: The diffusion coefficients were determined from the time-dependence of the degree of swelling by a method described in Ref. [21]. By NMR imaging of the diffusion of CD2Cl2 in PES-MHC, we focus our attention on the time-dependence of the position of the diffusion front and to the concentration of the swelling agent in dependence on the chain orientation. The experiments were performed at room temperature during the first stage of swelling (0–60 min). The experimental setup used is shown in Fig. 7. The samples (3 mm × 10 mm) have a thickness of

74

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

Fig. 8. Diffusion of per-deuterated dichloromethane in the nematic-like network (PES-MHC) parallel and perpendicular to the chain orientation. The bright vertical strip (1) at 40 min is an area of hindered diffusion (swelling) due to a support device at the top of the polymer sample. The strongly swollen areas are dark. Adhesion effects influence the solvent uptake.

1 mm in the dry state. About 1 mm of the sample was immersed in the solvent. The resolution of the T2weighted images (Fourier imaging at long, but fixed, echo time) is better than 100 mm. Due to the perdeuterated solvent, only the protons of the polymer are able to give an NMR signal. The signal intensity depends on the swelling degree. The unswollen part behaves like a rigid lattice with very short magnetization decays (short T2) and does not give any or gives only a weak NMR signal. Only the part of the sample

volume that is already softened by the swelling process gives strong signals (Fig. 8). The anisotropy of the diffusion process is shown by the following facts: • The diffusion in the direction parallel to the chain orientation is emphasized by the fibrous structure of the polymer. This structure also diminishes the diffusion process in the perpendicular direction. • As a result of the fibrous structure of the polymer,

Fig. 9. Degree of swelling of PNIPAAm gel in water–alcohol mixtures at 21⬚C: (B) methanol, (X) ethanol, (O) 1-propanol, (P) 2-propanol, and (V) acetone.

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

Fig. 10. Scheme of the formation of a macronetwork at temperature above TVPT.

the diffusion in the direction perpendicular to chain orientation is stepwise.

3.3. Diffusion of a methanol/water mixed solvent in water-swollen gels (PNIPAAm) Poly(N-isopropylacrylamide) (PNIPAAm) gels show some very interesting properties, which offer some technical developments for the future (e.g. artificial muscle, temperature-sensitive actuators). It is known that the PNIPAAm/water solution has a lower critical solution temperature (LCST). In a mixture of organic solvents and water, the LCST shifts to lower values in dependence on the composition [22]. The cross-linked polymer gels undergo analogous collapse transitions in aqueous solvents. The degree of swelling depends on the temperature (transition from the low-temperature swollen phase to

75

the high-temperature shrunken phase at TVPT around 33⬚C in pure water [23]), and the composition of a mixed swelling agent [24]. Fig. 9 shows the swelling degree of PNIPAAm gel in mixtures of water with organic solvents at constant temperature (25⬚C). When the methanol concentration exceeds 20 vol.%, the gel shows a broad crossover region from the swollen to the shrunken state. The results for water–ethanol, –propanol, –acetone, mixed solvents are qualitatively similar to that for water–methanol mixture. Possible applications of these gels are chemomechanical valves [25], where the flow of a liquid is controlled by a gel actuator consisting of spherical gel particles. The shrinking of gel particles starts at T ⬎ TVPT (constant composition) or if the concentration (constant temperature) of the organic solvent inside the water-swollen gel is higher than a typical value, given by the swelling-composition characteristic (Fig. 9). It is necessary to determine the time that is required for this process in dependence on the alcohol concentration, for instance. Furthermore, it is important to clarify the dynamics of water molecules in polymer gels from a macroscopic point of view through microscopic information. The shrinking process induced by a change of the properties of the swelling agent shows a two-step behavior: In the first step, the elastic modulus of the gel increases very rapidly. The soft polymer gel

Fig. 11. 2D Fourier images of PNIPAAm gel (BIS 4) swollen in H2O in dependence on temperature (- - - contour of the sample).

76

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

Fig. 12. 2D Fourier imaging of PNIPAAm gel (BIS 4) swollen in D2O at different temperatures during a heating period.

changes to a more solid-like structure due to the formation of a macronetwork consisting of bundles of network chains [26] (Fig. 10). The degree of swelling is constant. However, after longer times the shrinking process takes place. The sample volume decreases with a typical time-constant determined by the cooperative diffusion coefficient of the polymer chains. Tanaka et al. [6] could observe by pulse-field gradient spin echo (PGSE) NMR that the diffusion constant (and also T2 measured by a multi-pulse sequence according to Carr–Purcel–Meiboom–Gill) of the water shows a rapid decrease at TVPT. Furthermore, they could show by NMR imaging the formation of a surface layer during the shrinkage process in highly crosslinked gels. We performed NMR imaging measurements to prove some of the statements given above. The samples were prepared by free radical polymerization of NIPAAm in water with different content (1– 10 mol%) of N,N 0 -methylenebisacrylamide (BIS) as the cross-linking agent. A solution of NIPAAm, BIS (overall monomer concentration: 0.53 mol/l) and potassium peroxodisulfate as the initiator (3 × 10 ⫺3 mol/mol monomer) were cooled to 0⬚C and purged with nitrogen for 15 min. The N,N,N 0 ,N 0 -tetramethylethylenediamine (TIMEDA) as the accelerator (3 × 10 ⫺3 mol/mol monomer) was added and the solution was immediately transferred into glass tubes. After 17 h reaction time at 20⬚C, the

gels were separated and washed with water for one week. The degree of swelling Q (defined by the ratio of the mass of swollen polymer gel to that of dried polymer) in the low-temperature swollen state (T ⬍ TVPT) depends on the BIS content. At a higher content of the cross-linker (BIS ⬎ 4 mol%) the samples show a porous structure. In Fig. 11, NMR imaging shows the shrinkage of a cylindrically shaped gel in water. The image contrast in the two-dimensional Fourier images with slice selection arises from the T2-weighted (diffusion!) spin density signal of the solvent. The gel was stepwise heated to a constant temperature and then measured. The best contrast of the gel in relation to the water is observed in the image after 18 min at TVPT due to the large difference in the mobility of the water molecules in the gel and in the surrounding volume. The shape and the cross-section of the gel are nearly the same in all pictures. (The bright surrounding areas in the figures at 22 and 28⬚C are caused by a hindrance of convection of water nearby contact surfaces between gel and NMR tube). The beginning of the shrinkage as the second step of the process is even later. This is in agreement with integral (non-imaging) NMR measurements [6], which show a rapid decrease of T2 a long time before shrinkage takes place. However, some inhomogeneities at higher temperatures (black areas at 40⬚C) can be seen. The shrinking process starts and the T2 of water changes. At 50⬚C the contrast decreases due to a decrease of T2 of the immobilized water (and to a smaller extent of the gel). The image of the polymer matrix swollen in perdeuterated water gives a further hint that the shrinking is a two-step process (Fig. 12). Contrary to Fig. 11, the image contrast arises from the signals of the polymer matrix. The brighter the areas are, the longer is T2 and the higher is the mobility of the network chains. At TVPT the T2-weighted signals of the matrix vanish rapidly due to the change of the network to a more solid-like (rigid lattice) material. In both figures the slice thickness was about 500 mm. In Fig. 11, the diameter of the sample was smaller than the inner NMR tube diameter (4 mm); in Fig. 12, the diameter of the sample was equal to that of the tube. With the help of MRI, we also investigated the change in the swelling behavior by using a mixture of D2O/CH3-OD (partially per-deuterated methanol)

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

77

Fig. 13. One-dimensional diffusion experiment (60 vol.% CH3OD in D2O). Left: experimental setup; and right: MRI in dependence on time (10 equidistant time-steps from 5 min (1) to 2600 min (10)).

as swelling agent. The CH3-OD concentration varies from pure water (mixture of D2O/H2O) to 80 vol.%. Fig. 13 shows the experimental setup. The time-dependence of the overall magnetization (concentration of CH3OD in the whole sample) in dependence on the square root of time is shown in Fig. 14. Up pto 300 min at 21⬚C the diffusion process follows a t -law (Case I diffusion) for all concentrations (including pure water). From this it follows that in the first time-period of the diffusion of methanol in the sample, no change of the network structure occurs, which would be able to hinder the diffusion. In Fig. 15, the increase of magnetization of a thin sample layer (thickness about 60 mm) at 2.78 mm distance to the surface as a measure of the CH3-OD concentration for several experimental starting conditions is shown. The curves are represented in dependence on the square root of time to emphasize the different parts of the diffusion process: • The time-lag (Q ) depends slightly on the methanol

Fig. 14. Overall magnetization (in arbitrary p units) of the whole sample volume (BIS 4) as a function of t …T ˆ 21⬚C; 60 vol.% CH p3OD). At short times (up to 300 min), the magnetization follows a t -law (Case I diffusion).

Fig. 15. Time-dependence of the solvent absorption (here shown as a normalized NMR signal of a thin layer (⬇0.5 mm) inside the gel. The curves are separated by a stepwise offset.) at a distance of 2.78 mm to the solvent–sample surface of a PNIPAAm gel. BIS 4: (1) D2O/H2O, (2) 20 vol.% CH3OD, (3) 40 vol.% CH3OD, (4) 60 vol.% CH3OD, (5) 80 vol.% CH3OD; BIS 10: (6) 40 vol.% CH3OD). The first line marks the end of the time-lag and the start of Case I diffusion (if recognizable).

concentration with a maximum at 40 vol.%. This reflects the shrinking minimum at this concentration (see above). A rough estimation of the diffusion constant is possible using the equation D ˆ l 2 =…6Q† (l ˆ diffusion length). The calculated values are nearly equal to the self-diffusion constant of free water …D water ˆ 2:3 × 10⫺3 mm2 =s [27]). For the 40 vol.% methanol solution, the value deviates slightly to higher values. • The end of the nearly linear part is more or less pronounced by effects of the finite volume of the sample. This is well observable at high (curve 5) and low (1,2) methanol concentrations due to a faster diffusion process. On the other hand, the curves for medium methanol concentrations (3,4) and for the highly cross-linked sample (6) shows a weaker (and more delayed) diffusion. This is also in good agreement with the swelling measurements shown in Fig. 9, which confirms a restricted swelling process at middle (30–60 wt.%) methanol concentration. • In the case of higher methanol concentrations or of a higher cross-linked sample, no sharp onset of the diffusion after the time-lag can be seen. In other words, there is a weak but observable diffusion also for very short times and no bigger change after the

78

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79

4. Conclusions

Fig. 16. Fourier images (slice selection, T2-weighted) of a vertical plane of shrunken PNIPAAm gels (3 days deswelling in D2O/ CH3OD (40 vol.%)). Left: BIS 10—(1) unshrunken volume; (2) shrunken surface layer; and (3) solvent. Right: (4) shrunken gel BIS 4.

time-lag period. One possible explanation could be a big difference between the individual mobility of some methanol molecules (or clusters) at the diffusion front and the overall diffusion process, which is hindered by the onset of the network change. • In the case of the highly cross-linked sample BIS 10, an additional reason for this could be the porous structure of this gel. Another feature of the diffusion of methanol in a PNIPAAm gel is the formation of surface layers, which delay or prevent further diffusion of the solvent. This was already observed [6] for highly cross-linked gels, but as an effect of the volume transition temperature in pure water. Surface layer formation in temperature-sensitive gels occurs if the temperature is suddenly increased above TVPT. In Fig. 16, the layer formation is shown for methanol diffusion at room temperature. At the beginning of the diffusion process, the volumes of both samples were nearly the same. The experiment shows that a surface layer (2) is observable at the samples BIS 10. The center of the sample (1) is swollen also after some days. In the case of BIS 4 the whole sample (4) is in the shrunken state and no surface layer is observable. An explanation of the different behavior of both networks is their structure. The response of the gel to the solvent change is faster for the more porous sample BIS 10. The polymer chains in an outer shell collapse immediately and a dense surface layer is formed, which prevents further shrinking.

The aim of this paper is to present some examples of MRI investigations of diffusion processes. The NMR imaging technique is a powerful tool for spatially resolved investigation of diffusion processes in polymeric networks in real-time conditions; however, it is far from being a “routine application” until now. In dependence on the material (rubber-like networks, polymer glasses, water gels) the matrix or the solvent can be observed by different NMR parameters (T2, T1, qM2,…), which have to be chosen carefully for the different purposes. In relation to denser materials (rubber, plastics) the imaging of diffusion in gels is complicated by selfdiffusion (and convection) effects, which dominate the magnetization decay. For the gel materials shown here, a Case I diffusion at the beginning of the diffusion process is stated in a first approximation. Because of the influence of the composition of the swelling agent on the chain collapse and the degree of swelling for longer diffusion times and organic solvent, the time-dependence deviates from Case I diffusion (two-step model of the shrinking process). The spatially resolved solvent uptake shows deviations already in the Case I regime in relation to pure water at different methanol concentrations. This is a hint that changes in the network are induced by the solvent also in the first period of Case I diffusion. MRI could confirm the otherwise observed features of the swelling/shrinking behavior at methanol concentrations of about 40 vol.%. The analysis of the concentration increase in thin layers at a fixed distance to the surface and the observation of a surface layer of hindered diffusion for highly cross-linked samples by 2D Fourier imaging can explain some of these diffusion features. Acknowledgements The German Research Fund (DFG) is gratefully acknowledged for its financial support of our work. References [1] P.Y. Ghi, D.J. Hill, D. Maillet, A.K. Whittaker, Polymer 38 (1997) 3985.

M. Kno¨rgen et al. / Journal of Molecular Structure 554 (2000) 69–79 [2] J. Crank, The Mathematics of Diffusion, 2nd ed., Clarendon Press, Oxford, 1975. [3] P. Blu¨mler, B. Blu¨mich, Rubber Chem. Technol. 70 (1997) 468. [4] M. Valtier, P. Tekely, L. Kie´ne, D. Canet, Macromolecules 28 (1995) 4075. [5] M. Kno¨rgen, U. Heuert, H. Menge, H. Schneider, Angew. Makromol. Chem. 261/262 (1998) 123. [6] N. Tanaka, S. Matsukawa, H. Kurosu, I. Ando, Polymer 39 (1998) 4703. [7] H. Yasunaga, H. Kurosu, I. Ando, Macromolecules 25 (1992) 6505. [8] T. Shibuya, H. Yasunaga, H. Kurosu, I. Ando, Macromolecules 28 (1995) 4377. [9] Y. Hotta, T. Shibuya, H. Yasunaga, H. Kurosu, I. Ando, Polym. Gels Networks 6 (1998) 1. [10] S. Matsukawa, I. Ando, Macromolecules 29 (1996) 7136. [11] S. Matsukawa, I. Ando, Macromolecules 30 (1997) 8310. [12] P.C. Lauterbur, Nature 242 (1973) 190. [13] P. Mansfield, J. Phys. C 10 (1977) L55. [14] P.T. Callaghan, Principles of Nuclear Magnetic Resonance Microscopy, Oxford, 1991. [15] M. Kno¨rgen, U. Heuert, H. Schneider, P. Barth, W. Kuhn, Polym. Bull. 38 (1997) 101.

79

[16] M. Kno¨rgen, U. Heuert, H. Schneider, in: P. Blu¨mler, B. Blu¨mlich, R. Botto, E. Fukushima (Eds.), Spatially Resolved Magnetic Resonance, Wiley, New York, 1999, p. 211. [17] J. Pionteck, S. Richter, S. Zschoche, K. Sahre, K.-F. Arndt, Acta Polym. 49 (1998) 192. [18] S. Richter, R. Ha¨ßler, K.-F. Arndt, Macromol. Chem. Phys. 200 (1999) 474. [19] S. Richter, PhD thesis, TU Dresden, 1999. [20] S. Richter, G. Heinrich, K. Schro¨ter, K.-F. Arndt, Macromol. Chem. Phys. 201 (2000). [21] P. Chiarelli, C. Domenici, G. Genuini, J. Mat. Sci.: Mater. Med. 4 (1993) 5. [22] H.G. Schild, M. Muthukumar, D.A. Tirrell, Macromolecules 24 (1991) 948. [23] Y. Hirokawa, T. Tanaka, J. Phys. Chem. 81 (1984) 6379. [24] S. Hirotsu, J. Phys. Soc. Jpn 56 (1987) 233. [25] K.-F. Arndt, A. Richter, D. Kuckling, Polym. Adv. Techn. (2000) in press. [26] M. Shibayama, M. Moritomo, S. Nomura, Macromolecules 27 (1994) 5060. [27] Z.J. Derlacki, A.J. Easteal, A.V.J. Edge, L.A. Woolf, Z. Roksandic, J. Phys. Chem. 89 (1985) 5318.