A solid-state NMR method to determine domain sizes in multi-component polymer formulations

Journal of Magnetic Resonance 261 (2015) 43–48

Contents lists available at ScienceDirect

Journal of Magnetic Resonance journal homepage: www.elsevier.com/locate/jmr

A solid-state NMR method to determine domain sizes in multi-component polymer formulations Judith Schlagnitweit a, Mingxue Tang a,1, Maria Baias a,2, Sara Richardson b, Staffan Schantz b, Lyndon Emsley a,c,⇑ a b c

Université de Lyon, Institut de Science Analytiques, Centre de RMN à très hauts champs (CNRS/ENS Lyon/UCB Lyon1), Villeurbanne, France AstraZeneca R&D Pharmaceutical Development, Mölndal, Sweden Institut des Sciences et Ingénierie Chimiques, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland

a r t i c l e

i n f o

Article history: Received 13 May 2015 Revised 25 September 2015 Available online 14 October 2015 Keywords: Solid-state NMR Domain sizes Cellulose Spin diffusion

a b s t r a c t Polymer domain sizes are related to many of the physical properties of polymers. Here we present a solidstate NMR experiment that is capable of measuring domain sizes in multi-component mixtures. The method combines selective excitation of carbon magnetization to isolate a specific component with proton spin diffusion to report on domain size. We demonstrate the method in the context of controlled release formulations, which represents one of today’s challenges in pharmaceutical science. We show that we can measure domain sizes of interest in the different components of industrial pharmaceutical formulations at natural isotopic abundance containing various (modified) cellulose derivatives, such as microcrystalline cellulose matrixes that are film-coated with a mixture of ethyl cellulose (EC) and hydroxypropyl cellulose (HPC). Ó 2015 Published by Elsevier Inc.

1. Introduction Polymer domain sizes determine many of their macroscopic physical properties. For example, cellulose and cellulose based polymers (i.e. ether and ester derivatives) are used in almost all pharmaceutical formulations as stabilizers, binders, fillers, disintegrants, pore formers or gel matrixes [1]. For example, ethyl cellulose (EC) and hydroxypropyl cellulose (HPC) are used as binders or for film-coating [2,3]. To better understand the physical properties of multicomponent polymers it is important to elucidate the phase morphology, notably by determining the degree of crystallinity and the domain sizes in the mixture. Specifically for EC/HPC film coatings in controlled release formulations, the drug release behavior is governed primarily by the phase structure of the water-soluble phase (HPC) [4]. This structure is defined by the number of domains, their size and their connectivity [5,6] within the insoluble EC matrix. Once the formulation enters the gastrointestinal tract, ⇑ Corresponding author at: Institut des Sciences et Ingénierie Chimiques, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. E-mail address: [email protected] (L. Emsley). 1 Current Address: CNRS, CEMHTI(UPR3079), Université d’Orleans, 1D avenue de la recherché scientifique, 45071 Orléans Cedex 2, France. 2 Current Address: New York University Abu Dhabi, P.O. Box 129118, Abu Dhabi, United Arab Emirates. http://dx.doi.org/10.1016/j.jmr.2015.09.014 1090-7807/Ó 2015 Published by Elsevier Inc.

the water-soluble HPC domains of the sample dissolve, leaving a network of channels. Water then dissolves the active pharmaceutical ingredient (API), resulting in a diffusion process where the drug is released over a certain amount of time [4]. Obviously, finding better and more accurate ways to control the release of drugs is a key target in pharmaceutical science. There has been a great interest in developing controlled release formulations to maintain a desired drug concentration level for long periods of time without reaching a drug concentration above which the drug produces undesirable side effects and below which the drug is not therapeutically effective. Well-designed film coatings can provide a reduction in the dosage regimen as well as minimize the dose dumping effect. It is self-evident that many different factors, such as the ratio between EC and HPC [4], the viscosity, molecular weight [7] as well as the spray-coating conditions [8] may strongly influence the domain structure of the different components, and with it the release properties of the film. Determination of the structure and sizes of the domains in film coatings on actual pharmaceutical samples (pellets) could lead to the possibility to better control the coating structure, finely tune the functionality, and better control the drug release. Currently the approaches used to determine domain sizes in EC/HPC films are scanning electron microscopy (SEM) [7], Confocal laser scanning microscopy (CLSM) [8] or Small-angle neutron

44

J. Schlagnitweit et al. / Journal of Magnetic Resonance 261 (2015) 43–48

scattering (SANS) [9]. Here, we show that solid-state nuclear magnetic resonance (NMR) can be an advantageous alternative, especially to directly analyse the film coating of the pellets produced in the pharmaceutical process. The method developed here can also be applied to other areas where multi-component polymer mixtures are important. NMR has always been one of the most successful methods for studying disordered solids, and in favorable cases can be used to provide complete structural characterisation. In particular, nuclear spin-diffusion has proven to be a powerful method for the characterisation of semi-crystalline polymer morphology. It was shown that one-dimensional proton detected [10–12], as well as twodimensional carbon detected experiments [13] provide detailed information on various aspects of heterogeneity, both structural and dynamical, regarding the molecular mobility of the different components in the investigated system. To perform a spindiffusion experiment a spatially inhomogeneous non-equilibrium distribution of z-magnetization should be created, where different types of domains are polarized differently. The return to equilibrium driven by spin diffusion is monitored during a mixing time, and the dynamics of this process can be interpreted in terms of models of the domain-size and -structure [14]. Different procedures to select proton magnetization from particular domains have been proposed in the literature including filters based on differences in the dipolar couplings between domains, such as the double quantum filters [15] or dipolar filters [11,16], which make use of the differences in the decay times of the transverse proton magnetization. These sequences are usually applied to two component systems where the components exhibit significant differences in their molecularlevel mobility. Proton T1 relaxation times are generally too long to be used for selection since they are averaged by spin diffusion. However, differences in spin lattice relaxation in the rotating frame have also been the basis for filter methods [14]. In some cases selection can also be achieved due to proton chemical shift differences under CRAMPS conditions [17]. Following the mixing period magnetization is detected for different spin diffusion (mixing) times, either on protons directly or, to improve resolution, on 13C [10]. However, the applicability of these experiments to multicomponent mixtures at natural abundance, such as the filmcoated pellets, is very limited due to two principle difficulties: (i) The structure of the film coating, which only makes up a small fraction of the pellet, has to be investigated specifically. The sample does not only consist of the film but also of multiple other components such as the core material (for example microcrystalline cellulose) and the API, which will all give signals in the NMR experiment. These are usually unresolved in proton solid-state NMR experiments, and therefore experiments to efficiently specifically select the information from the film component need to be developed. The experiments described above use either chemical shift filters to select chemical different components, or mobility filters to select between mobile and rigid domains. So far there is no scheme to select the rigid or mobile component of only one chemical species, in the presence of both rigid and mobile components of other species. (ii) The domain sizes in these systems are expected to be relatively large, i.e. in the range of a few hundred nanometres up to micrometres, which leads to difficulties due to relaxation and low sensitivity using spin diffusion experiments. In this report we develop a new solid state NMR method, specifically a carbon-edited mobility-filtered 1H spin-diffusion scheme, that extends the pioneering experiments mentioned above, and which is capable of measuring domain sizes in the different components of polymer mixtures.

2. Theory Spin diffusion, first described by Bloembergen, is a process of nuclear magnetization exchange driven by energy-conserving flip–flop transitions in a dense network of dipolar coupled spins [18]. Here we focus on proton spin diffusion (PSD) which is expected to be fast under the moderate magic angle spinning conditions (5 kHz), used in this work [19]. The time dependence of spin diffusion can be related to inter-nuclear distances, and therefore it has been used to study many structural problems in the past, ranging from domain sizes in polymers and disorder in glassy materials to high-resolution crystal structure determination of small molecules and proteins. In order to obtain information on distances or domain sizes from PSD data, it is necessary to analyse the time dependence of the exchange process. Spin diffusion can be described phenomenologically using Fick’s second law of diffusion [14]:

@Pðr; tÞ ¼ DDPðr; tÞ @t

ð1Þ

with P(r, t) being the polarization (z-magnetization) as a function of position r and diffusion time t and D being the spin diffusion coefficient. This coefficient can be described as

D ¼ Xa2

ð2Þ

where the transition rate O is proportional to the rate of the dipolar coupling and a is the distance between spins. For polymers spin diffusion coefficients are usually about 1 nm2/ms (1015 m2/s) [14]. 3. Materials and methods 3.1. Materials Sprayed EC/HPC free films made of water-insoluble polymer EC and water-soluble HPC (ratio 70:30) were provided by AstraZeneca. The samples were prepared under different manufacturing conditions and therefore show different domain-structures, as verified by CSLM on fluorescein-labelled films in Ref. [8]. In this work we investigate two of these samples with different domain sizes using NMR. One of the films, which we will henceforth call sample 1, is expected to have large domains (lm size), see Fig. 2B in Ref. [8], while the other film, sample 2 has smaller domains (200 nm) as can be seen in Fig. 2D in Ref. [8] (these two figures are reproduced for convenience in the supplementary information). In addition we investigate film-coated pellets. This sample of most interest (sample 3) is not a pure film, but consists of pellets with an MCC core (and in some cases also a layer of API), which is spray-coated with EC/HPC. The size of the domains within this film-layer has so far previously only been characterized by SANS [9]. While it is not straightforward to obtain information by CLSM in this case, the newly developed NMR sequence is equally applicable to the film samples, as well as to the pellets directly coming from the pharmaceutical process. 3.2. Method Based on classical one-dimensional proton detected spin diffusion experiments [10–12] we designed a new carbon-edited mobility-filtered proton spin-diffusion experiment. By exploiting differences in the chemical shifts in the high-resolution carbon spectrum, this allows us to monitor proton spin diffusion from rigid EC domains to HPC domains in our sample. The pulse scheme for the new experiment is shown in Fig. 1, together with a series of spectra illustrating the mechanism of action.

J. Schlagnitweit et al. / Journal of Magnetic Resonance 261 (2015) 43–48

As for other experiments of similar type, the newly developed pulse sequence consists of a selection/filter block (Fig. 1, pulse sequence up until point (c)), followed by a spin diffusion part (see Fig. 1, from c to d). In this case we combine selective excitation for carbon chemical shift editing with a proton mobility filter. The selection process starts with a cross-polarization step from 1 H to 13C to enhance sensitivity (point (a) in Fig. 1) followed by a z-filter and a selective excitation step on carbon-13, consisting in a non-selective 90° ‘‘flip-back” pulse followed by a selective 90° pulse applied at the chemical shift of interest. This allows us to only select, for example, magnetization present in the (rigid and mobile) EC domains of the sample (Fig. 1, panel (b)). It should be noted that with this experiment one is limited to samples where at least one of the different components has at least one carbon signal which does not overlap, which can be used for selective excitation. However, it might be possible to use other filter elements prior to selective excitation to obtain non-overlapping signals in the carbon spectrum albeit at the cost of further reducing the sensitivity of the experiment. In the case of, for example, a controlled release pellet sample with API, whose 13C NMR signals overlap with those of EC, a T1-filter element (saturation pulse train and short recovery) could be inserted at the beginning of the sequence, since APIs are usually crystalline leading to longer spin–lattice relaxation times than cellulose. After the selective excitation step, the EC magnetization is cross-polarized back to the protons using a short contact time (point (c) in Fig. 1). This acts as a mobility filter, as it will be primarily effective for rigid components with strong dipolar couplings. This results in only proton magnetization of rigid EC domains being present after this step (Fig. 1, panel (c)). We found that using the back-CP step as a mobility filter is the most efficient way to select rigid EC magnetization. However, if necessary, an additional mobility filter element such as a double quantum filter [15,25] could be inserted following the back-CP block. It has to be noted that the signal intensities in the resulting proton spectra are about a factor 5000 lower compared to a spectrum obtained directly after a 90° proton pulse. This factor is a result of going through natural abundance carbon-13, selective excitation of magnetization of only CH3 groups within the EC molecule of the sample, as well as the efficiency of CP and back-CP steps and mobility filters. Optimization of CP, selective excitation and especially back-CP steps in combination with mobility filtering is absolutely crucial to efficiently select the magnetization of rigid EC. The selected magnetization is then aligned back along the z direction by a 90° proton pulse, and the magnetization is allowed to propagate to surrounding spins by spin diffusion through the dense network of dipolar coupled nuclear spins for a time s after which the proton signal is recorded following a final 90° pulse (Fig. 1, panel (d)). The experiment is then repeated for a series of different spin diffusion times to monitor the transfer from rigid EC domains to mobile HPC domains. Fig. 2 shows spectra obtained in this way for an EC/HPC free film sample (described above) for six different spin diffusion times. The spectra obtained for each spin diffusion time are deconvolved into four lines (detailed description in SI) using Matlab. Fig. 3 shows the resulting integrals of the different rigid and mobile components as a function of the spin diffusion delay for free film sample 2. (Spin diffusion curves for all investigated samples are shown in the SI). It can be seen that the relative integrals equilibrate within the first few hundred ls between the purple, turquois and orange lines (as opposed to a slower build-up of the narrow blue line). This indicates that the magnetization is spreading from the selectively excited CH3 groups within EC. Since we are mostly interested in the size of the HPC domains in the sample, we focus on the build-up of magnetization in these domains. The fact that HPC is

45

Fig. 1. Pulse sequence for the carbon-edited mobility-filtered proton spin-diffusion pulse sequence. Filled rectangles represent 90° pulses. White polygons represent cross polarization (CP), for the (first) forward CP step the amplitude of the 1H rf field was ramped during the contact time (1.5 ms) to improve efficiency [20]. The (second) back-CP duration was set to 100 ls to increase the selectivity for the rigid EC domains. The shaped pulse for selective excitation on carbon is represented by a Gaussian shape. Experimentally an EBURP [21] pulse shape or a Gaussian cascade [22,23] with 500 Hz excitation bandwidth were used. SPINAL64 decoupling [24] was applied on protons during the selective pulse (shown in gray). The delay between scans was set to 6s. The panels show experimental spectra (obtained on a 500 MHz Bruker Avance III solid-state NMR spectrometer equipped with a 4 mm double-resonance MAS probe at 295 K and 5 kHz MAS) of the pellets sample. They show the magnetization present after CP to 13C (panel a, 128 scans), after the selective excitation step (b, 128 scans), after back CP to 1H (panel c, 1024 scans), and after spin diffusion s (panel d, 3072 scans).

46

J. Schlagnitweit et al. / Journal of Magnetic Resonance 261 (2015) 43–48

Fig. 2. Carbon-edited mobility-filtered 1H spin-diffusion spectrum of film sample 2 obtained after different spin diffusion times using the pulse sequence shown in Fig. 1 (experimental details as described in Fig. 1, in this case 2272 scans (leading to a total measurement time of 3h50’) were acquired for each spectrum). The experimental spectra are deconvolved into four resonances (Gaussian lines of different width, with and without MAS sideband pattern [26]) representing different components of the sample: purple broad line – rigid components, turquoise broad line – CH3-groups, narrow orange line – mobile CH3-groups, narrow blue line – water molecules distributed in the water-soluble HPC components of the sample. The red line is the sum of the four different lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

assuming that these water molecules which are picked up by the sample from the air, are equally distributed within the HPC domains. (See supplementary information for more details). 3.3. Modelling/Fit As described above, to model spin diffusion a set of differential equations are used describing a classical one-dimensional diffusion process, with P(r, t) being the polarization as a function of distance r from the centre and diffusion time t.

@Pðr; tÞ ¼ DDPðr; tÞ; @t

ð4Þ

with DHPC, DEC being spin diffusion coefficients and dHPC, dEC domains sizes for EC and HPC domains: Fig. 3. Integrals resulting from the deconvolution of the experimental spindiffusion spectra of the different rigid and mobile components of sample 2 as a function of spin diffusion delay s. (As in Fig. 2 the colors represent the different components of the sample: purple line – rigid components, turquoise line – CH3groups, orange line – mobile CH3-groups, blue line – water molecules distributed in the water-soluble HPC components of the sample.) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

D ¼ DEC D ¼ DHPC

for

06r<

for

1 1 dEC 6 r 6 ðdHPC þ dEC Þ 2 2

ð5Þ

The initial conditions and boundary conditions are respectively:

Pðr; 0Þ ¼ 1 for water-soluble, in contrast to EC, allows us to monitor the build-up of magnetization of the water molecules which are distributed in the water-soluble HPC components of the sample, i.e. the slow increase of intensity of the narrow blue line. In this case we are probing the build up of this narrow line to obtain the domain size,

1 dEC 2

Pðr; 0Þ ¼ 0 for @P


1 2

1 dEC 2 1 6 r 6 ðdHPC þ dEC Þ 2

06r< 1 dEC 2 

ðdHPC þ dEC Þ; t ¼0 @t

ð6Þ

ð7Þ

47

J. Schlagnitweit et al. / Journal of Magnetic Resonance 261 (2015) 43–48

volume of the EC and HPC fractions are linked. Here the volume ratio of EC and HPC is 70/30. The domain size for HPC can then be calculated for a model of spherical HPC domains surrounded by EC by satisfying the condition that the volume ratio between EC and HPC is 70/30. We use Matlab to search for numerical solutions for this differential equation, fitting the experimental data of magnetization build-up on water molecules in the HPC phase, and obtaining the domain size. (The program to simulate the spin diffusion curves is available in the supplementary information). The resulting fit for one of the samples is shown in Fig. 4. (See supplementary information for all other samples).

4. Results and discussion

Fig. 4. Relative integrals of mobile HPC components (build-up, blue filled circles) and the sum of the other lines (blue filled circles, decreasing rel. integral) obtained from deconvolution of experimental data (film sample 2) versus spin diffusion time. The red lines represent relative integrals obtained from the fit including estimated errors indicated by the light red area using numerical modelling of the diffusion process as described in the text. (The diffusion constant was set to D = 0.5  105 Å2 s1 = 0.5 nm2 ms1). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Initially, at time 0 the magnetization is assumed to be 1 in one of the two domains and 0 in the other domain. The boundary condition corresponds to no polarization diffusing out of the system. To obtain relative signal intensities the polarization is integrated over r:

Z

1d 2 EC

SEC ðtÞ ¼

Pðr; tÞdr

0

Z SHPC ðtÞ ¼

1ðd þdEC Þ 2 HPC

ð8Þ Pðr; tÞdr

1d 2 EC

The model represents magnetization starting in the EC domains and then diffusing to the mobile part (water molecules bound to HPC). Note that here we use the simplest model for diffusion (onedimensional diffusion between domains). More sophisticated modelling could certainly be used, depending on any known nature of the samples, such as three-dimensional diffusion models, or including distributions of domain sizes, but since the sample here is probably quite heterogeneous, we choose the use a simple model as a demonstration. More complex models will change slightly the domain sizes obtained, but should not change the nature or trends on the results. Since in this experiment we are not able to directly detect the polarization of HPC but only water molecules distributed in the HPC domain, the integral of the narrow signal is only a fraction of the HPC polarization as obtained by the fit. This means that the fit directly leads to a domain size for the source domains (EC). To obtain an absolute value for the domain sizes also for HPC we introduce an additional relation. In a system with constant EC/HPC ratio the equilibrium polarization of EC/HPC, i.e. the

Using the above experiment and fitting procedure we obtain the domain sizes in Table 1 for the free film samples and the coated pellet samples. Even though the NMR experiment suffers from low sensitivity, and the changes we are detecting to monitor the spin diffusion process are small, these results clearly show that the EC/HPC film which is spray coated on the pellets has similar domain sizes to free film sample 2. The domain size for the coated pellet is in good agreement with the results obtained by SANS [9]. Experiments and fits carried out on free film sample 1 indicate that the average domain sizes are significantly bigger. This provides strong evidence that there are lm sized pores present in the distribution of domain sizes. This is in agreement with the CLSM images published earlier by AstraZeneca [8]. It should be noted that the NMR method only gives an average domain size, while the CSLM cross sections of the free films show that especially for films prepared at higher temperature (leading to bigger domains) the distribution of domain sizes is broader, ranging from multiple micrometres to smaller, i.e. tenths of nanometres, towards the surface of the film [8]. While the NMR experiment clearly allows us to compare film coatings with free film samples, the determination of the domain size depends on the selected spin diffusion model. The simple spherical model with fixed EC/HPC volume ratio could certainly be improved, for example by modelling minimal surface structures [27]. The experiment introduced here allows us to monitor spin diffusion from EC to HPC domains in the mixture, and is directly applicable to films in pharmaceutical formulations i.e. pellets. Therefore it is possible to verify if a film-coating has a similar structure, i.e. similar average domain sizes to a free film resulting from a spraying setup that was developed to mimic the film coating process as used for pellets [8]. This can support the transferability of properties determined for free films by other analytical tests, which can only be, or are much easier to be, carried out on free films, such as permeability tests or mechanical property measurements [7,28]. The experiment could find wider applications to determine domain sizes in mixtures wherever the domains of interest show different carbon-13 chemical shifts.

Table 1 Domain sizes determined for different EC/HPC samples.

dEC (nm) dHPC (nm)

EC/HPC free film (sample 1) (nm)

EC/HPC free film (sample 2) (nm)

EC/HPC coated pellets (nm)

145 ± 10 294 ± 20

76 ± 8 154 ± 16

86 ± 10 156 ± 20

48

J. Schlagnitweit et al. / Journal of Magnetic Resonance 261 (2015) 43–48

Acknowledgments J. S. acknowledges support from a Schrödinger Fellowship (J3377-N28) from the Austrian Science Fund (FWF). We would like to thank Alexandre Zagdoun, Andrew J. Pell and Cory Widdifield (all CRMN Lyon) for many useful discussions as well as Mariagrazia Marucci and Johan Hjärtstam (both AZ) for providing the samples. We also want to thank Mariagrazia Marucci, Christian von Corswant, J. Kohlbrecher and Ulf Olsson for sharing data from the SANS measurements prior to publication. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jmr.2015.09.014. References [1] J. Shokri, K. Adibkia, Application of Cellulose and Cellulose Derivatives in Pharmaceutical Industries, InTech, 2013. [2] R. Bodmeier, Tableting of coated pellets, Eur. J. Pharm. Biopharm. 43 (1997) 1–8. [3] F. Siepmann, J. Siepmann, M. Walther, R.J. MacRae, R. Bodmeier, Polymer blends for controlled release coatings, J. Controlled Release 125 (2008) 1–15. [4] M. Marucci, J. Hjärtstam, G. Ragnarsson, F. Iselau, A. Axelsson, Coated formulations: new insights into the release mechanism and changes in the film properties with a novel release cell, J. Controlled Release 136 (2009) 206– 212. [5] R.A. Siegel, Modeling of Drug Release from Porous Polymers, VCH Publishers Inc, New York, 1989. [6] S. Narisawa, H. Yoshino, Y. Hirakawa, K. Noda, Porosity-controlled ethylcellulose films coating. II. Spontaneous porous film formation in the spraying process and its solute permeability, Int. J. Pharm. 104 (1994) 95–106. [7] H. Andersson, J. Hjärtstam, M. Stading, C. von Corswant, A. Larsson, Effects of molecular weight on permeability and microstructure of mixed ethylhydroxypropyl-cellulose films, Eur. J. Pharm. Sci. 48 (2013) 240–248. [8] M. Marucci, J. Arnehed, A. Jarke, H. Matic, M. Nicholas, C. Boissier, C. von Corswant, Effect of the manufacturing conditions on the structure and permeability of polymer films intended for coating undergoing phase separation, Eur. J. Pharm. Biopharm. 83 (2013) 301–306. [9] M. Marucci, C.V. Corswant, J. Kohlbrecher, U. Olsson, in preparation (2015). [10] J. Clauss, K. Schmidt-Rohr, H.W. Spiess, Determination of domain sizes in heterogeneous polymers by solid-state NMR, Acta Polym. 44 (1993) 1–17.

[11] D.E. Demco, A. Johansson, J. Tegenfeldt, Proton spin diffusion for spatial heterogeneity and morphology investigations of polymers, Solid State Nucl. Magn. Reson. 4 (1995) 13–38. [12] D.L. VanderHart, G.B. McFadden, Some perspectives on the interpretation of proton NMR spin diffusion data in terms of polymer morphologies, Solid State Nucl. Magn. Reson. 7 (1996) 45–66. [13] K. Schmidt-Rohr, J. Clauss, H.W. Spiess, Correlation of structure, mobility, and morphological information in heterogeneous polymer materials by twodimensional wideline-separation NMR spectroscopy, Macromolecules 25 (1992) 3273–3277. [14] K. Schmidt-Rohr, H.W. Spiess, Multidimensional Solid-State NMR and Polymers, Academic Press, London, 1994. [15] R. Graf, D.E. Demco, J. Gottwald, S. Hafner, H.W. Spiess, Dipolar couplings and internuclear distances by double-quantum nuclear magnetic resonance spectroscopy of solids, J. Chem. Phys. 106 (1997) 885–895. [16] M. Goldman, L. Shen, Spin–spin relaxation in LaF3, Phys. Rev. 144 (1966) 321– 331. [17] K. Schmidt-Rohr, J. Clauss, B. Blümich, H.W. Spiess, Miscibility of polymer blends investigated by 1H spin diffusion and 13C NMR detection, Magn. Reson. Chem. 28 (1990) S3–S9. [18] N. Bloembergen, On the interaction of nuclear spins in a crystalline lattice, Physica 15 (1949) 386–426. [19] L. Emsley, Spin diffusion for NMR crystallography, in: R.K. Harris, D.M. Grant (Eds.), Encyclopedia of NMR, Wiley, Chichester, 2009. [20] O.B. Peersen, X.L. Wu, I. Kustanovich, S.O. Smith, Variable-amplitude crosspolarization MAS NMR, J. Magn. Res., Ser. A 104 (1993) 334–339. [21] H. Geen, R. Freeman, Band-selective radiofrequency pulses, J. Magn. Res. 93 (1991) 93–141. [22] L. Emsley, G. Bodenhausen, Self-refocusing effect of 270° Gaussian pulses. Applications to selective two-dimensional exchange spectroscopy, J. Magn. Res. 82 (1989) 211–221. [23] L. Emsley, G. Bodenhausen, Gaussian pulse cascades: new analytical functions for rectangular selective inversion and in-phase excitation in NMR, Chem. Phys. Lett. 165 (1990) 469–476. [24] B.M. Fung, A.K. Khitrin, K. Ermolaev, An improved broadband decoupling sequence for liquid crystals and solids, J. Magn. Res. 142 (2000) 97–101. [25] A. Buda, D.E. Demco, M. Bertmer, B. Blümich, B. Reining, H. Keul, H. Höcker, Domain sizes in heterogeneous polymers by spin diffusion using singlequantum and double-quantum dipolar filters, Solid State Nucl. Magn. Reson. 24 (2003) 39–67. [26] S. Hediger, A. Lesage, L. Emsley, A new NMR method for the study of local mobility in solids and application to hydration of biopolymers in plant cell walls, Macromolecules 35 (2002) 5078–5084. [27] D.M. Anderson, H. Wennerstroem, Self-diffusion in bicontinuous cubic phases, L3 phases, and microemulsions, J. Phys. Chem. 94 (1990) 8683–8694. [28] J. Hjärtstam, T. Hjertberg, Studies of the water permeability and mechanical properties of a film made of and ethyl cellulose–ethanol–water ternary mixture, J. Appl. Polym. Sci. 74 (1999) 2056–2062.