RECENT ADVANCES IN SOLID-STATE NMR

Robin K. Harris

Department of Chemistry,
University of Durham,
South Road,
Durham, DH1 3LE, U.K.

1 INTRODUCTION

Since the first successful use of the now-common combination of techniques (cross polarisation, magic-angle spinning ,and high-power proton decoupling) in 1976, NMR has transformed our ability to obtain structural and mobility information at the molecular level in a wide range of solids. Its value was first exploited for polymers and for zeolites but is now applied to all branches of chemistry and has spread to areas such as earth sciences, molecular biology – and food science. The recent expansion in applications has been sparked by new developments in techniques, especially in the use of a wide variety of specialised pulse sequences. Indeed, solid-state NMR now rivals, or even exceeds, solution-state NMR in its complexity. In principle, more information is available from NMR for solids than for solutions because anisotropic interactions (such as shielding, dipolar coupling and quadrupolar coupling) have more direct effects on spectra in the former case because there is less averaging from mobility at the molecular level.

The objective of this article is to introduce some of the important themes in modem solid-state NMR and to illustrate some applications, mostly using examples studied at Durham. Obviously, in a short article one has to be highly selective, and the choice of topics of importance is both personal and subjective. The article does not address food science matters directly, but the issues it raises do have relevant substantial consequences in this area. There is no doubt that the topics mentioned herein will find substantial applications in food science studies in the coming decade.

To start with, it is pertinent to point out the complementarity of diffraction techniques and solid-state NMR. The former rely on full three-dimensional long-range order for their optimum use, whereas NMR is sensitive to the local environment. The multinuclear aspect of magnetic resonance is very important: the method is isotope-specific, and there is never any problem of knowing which nucleus is being accessed. This means it can clearly distinguish between, say, nitrogen and oxygen, thus often providing information on disorder in crystalline lattices in cases where the placement of isoelectronic atoms may cause some difficulties for diffraction. Some simple crystallographic information, such as knowledge of the asymmetric unit, may be readily provided by NMR, which on occasion may give initial assistance to full crystal structure determination from diffraction experiments. Above all, NMR functions efficiently for microcrystalline "powder" samples. Although additional information may be obtained for single crystals by NMR this is largely of a rather esoteric nature. NMR spectra are sensitive to the form of crystalline solids and can therefore be readily used to study polymorphism and phase transitions. NMR is of substantial value for both amorphous and heterogeneous materials (which are, of course, ubiquitous in food materials), so that useful chemical and physical microstructure information can be obtained, particularly in the latter case because discriminating pulse sequences can be applied to obtain subspectra of components in different domains. NMR can provide (for instance via relaxation times) detailed information on mobility at the molecular level over a wide range of rates and, in favourable cases can distinguish static (spatial) from dynamic (temporal) disorder, giving thermodynamic data on chemical "exchange" processes. Finally, in many instances, NMR can produce direct data on interatomic distances (see the next section).

1.1 Dipolar Coupling: its Re-introduction

Unlike "scalar" coupling, which is mediated through bonds, dipolar coupling between two nuclear spins is direct ("through-space") and therefore in principle gives interatomic distances. The dipolar coupling constant between two heteronuclear spins A and X is given (in frequency units) by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where µ0 is the permeability constant, γA and γX are the magnetogyric ratios of A and X respectively, and rAX is the distance between the nuclei. The primary objective of most high-resolution NMR studies of solids is to remove the influence of dipolar coupling by magic-angle spinning and/or high-power heteronuclear decoupling. Remarkably, much effort is then expended to re-introduce the effects of dipolar coupling. The difference is that the recoupling required is selective, thus facilitating the measurement of a particular dipolar coupling constant free from the influence of other nuclei. A wide variety of pulse sequences have been invented to achieve recouling These fall into two classes:

homonuclear (such as rotational resonance, RFDR, HORROR and C7) and heteronuclear (for example REDOR, TEDOR and TRAPDOR)

Mostly they rely on zero quantum (flip-flop) or double quantum (flop-flop) coherence. This characterisation is not clear-cut, and Terao, Takegoshi and co-workers have described the rotational resonance in the tilted rotating frame (R2TR) method, which uses flip-flop and flop-flop mechanisms for homonuclear dipolar coupling measurement in different cases. They have discussedits application to a uniformly 13C, 15N isotope-labelled sample of glycylisoleucine. By a single series of five R2TR experiments on the one sample, they established all the relevant intramolecular internuclear distances and thus obtained the six dihedral angles which define the molecular conformation. This they describe as "the first solid-state NMR experiment for the determination of the complete three-dimensional molecular structure iin a powdered sample. In this case they were able to check their result against the structure obtained from single-crystal X-ray experiments, and the agreement is impressive. It may be noted, though, that determination of interatomic distances (via D) is not restricted to crystalline materials. It can be applied equally validly to amorphous systems, though obviously these may give rise to some distribution in distances.

However, in most cases, selective enrichment is important for NMR determinations of interatomic distances, in order for the system involved to be limited to a spin pair, though efforts are being made to account for multiple-spin interactions. We have, for example, examined the case of a host-guest system, namely fluorobenzene in t-butylcalix[4]arene, with observation of the methyl carbon resonances of the t-butyl groups. The involvement of 19F (100% natural abundance) in the system is helpful since it is not necessary to undertake any isotopic enrichment. The compound has a layer structure with alternate layers back-to-back, in which, as diffraction measurements eventually showed (Figure 1), a given t-butyl group has a fluorine atom in the adjacent layer slightly nearer than a fluorine in the same layer. Thus a REDOR experiment' involving the 13C, 19F system gives a plot which cannot be fitted by a calculation involving a single spin pair (Figure 2). However, involvement of the two fluorine atoms near to a t-butyl group results in a substantial improvement in the calculation. In such calculations, account needs to be taken of the mutual orientation of the two relevant 13C, 19F interatomic vectors.

1.2 Dipolar Coupling: The Problem of Mobility

The aim of recoupling experiments is ultimately to obtain geometry information from NMR independent of diffraction results. However, apart from the problem presented by multi-spin systems, any motional effects which are rapid on the NMR timescale need to be considered since only suitably averaged dipolar coupling constants will be derived. Thus, for the case discussed above, internal rotation about the C-CMe3 bond must be incorporated into the calculations, which gives (Figure 2) a less steep REDOR curve, (approaching more closely the experimental result). However, there are still discrepancies, which goes to illustrate the difficulties of this aspect of NMR.

In some cases definitive information can be obtained. For example, the system of long-chain α,ω-fluoroalkenes in urea host/guest compounds has been considered. These inclusion systems involve tunnel structures with the host and guest molecules incommensurate, making diffraction experiments problematic. However, 19F MAS NMR with high-power proton decoupling relates to a simple two-spin homonuclear system involving adjacent CH2F end groups of neighbouring molecules, ideal for recoupling (and other) strategies. The mutual conformations of these groups (which may be gauche-gauche, trans-gauche or trans-trans in principle) were unknown prior to NMR experiments. Motion of the guest molecules about the tunnel axis is expected to be facile. The MELODRAMA pulse sequence was used, giving the result shown in Figure 3. The fitting proceeds quite well in this case, especially in the crucial early part of the curve, and yields a motionally-averaged intermolecular (F,F) dipolar coupling constant, , of 995 Hz, which is reasonably consistent with other NMR experiments. Motion about the tunnel axis can be explicitly considered and the result is consistent with a simple average over the gauche-gauche and gauche-trans mutual conformations of the end-groups, with no contribution from any trans-trans form.

1.3 Mobility and Crystallographic Disorder

Solid-state NMR is a valuable complementary tool to diffraction experiments. One area in which the latter have a weakness is that of disorder in the crystal structure. Diffraction techniques cannot readily distinguish static (i.e. spatial) and dynamic (i.e. temporal) disorder. Since NMR can measure "exchange" rates over a wide range by means of different techniques, it is frequently capable of yielding such a distinction. An example is the 3:2 phenol-triphenylphosphine oxide (TPPO) supramolecular compound, which we have recently studied. The crystal structure shows that in the unit cell two phenol molecules are hydrogen-bonded to one TPPO molecule each, but a third phenol has "half" a hydrogen bond to each TPPO, i.e. it is disordered. In fact 31P CPMAS spectra show only one signal at ambient probe temperature, but two at 149 K (Figure 4). A classic coalescence phenomenon may be observed in the spectra between these two temperatures, showing that the disorder is dynamic. A combination of bandshape-fitting, T1ρ variation and selective polarization inversion experiments produces enough kinetic information to enable the thermodynamics of the barrier to be explored, yielding ΔH[double dagger] = 38 kJ mol-1 and ΔS[double dagger] = -23 kJ mol-1.

1.4 Mobility and Crystallography: Case 3

In pharmaceutical chemistry and related areas, the role of water in a solid is vitally important, especially since it is frequently very mobile. This can give difficulties even for crystalline materials, where diffraction methods may be insensitive to the presence of such species. A case in point is sildenafil citrate (the active component in Viagra), shown in an un-ionised form below (I).

Water is readily taken up or lost from this substance depending on the ambient humidity. Powder XRD shows no significant changes in unit cell size between samples equilibrated at 0 or 90% relative humidity, even though in the latter situation there are 0.78 mole equivalents of water associated with the solid. However, 13C CPMAS NMR experiments reveal (Figure 5) that there are distinctive changes in the spectra with water content. It is shown that, though the water molecules must be highly mobile, they are specifically associated with one of the citric acid carboxyl groups and with the propyl side-chain.

1.5 NMR Crystallography

The preceding examples illustrate that NMR is increasingly important as a contributor to crystallography. In fact it can be of considerable significance in determining complete crystallographic structures. This is particularly the case for the many examples where single-crystal diffraction methods are unsuitable. For such systems the Rietveld refinement method applied to powder diffraction data can yield results, but the major problem is the number of variables involved. It is therefore important to have good methods for arriving at trial structures which are relatively close to the true structures. The use of recoupling methods for determining molecular conformations in crystalline glycylisoleucine has already been mentioned above. It is also often feasible to pin-point the location of intermolecular hydrogen bonds, as in the case of polymorphs and pseudo-polymorphs of cortisone acetate. It is one of the primary aims of current NMR research to utilise NMR data as constraints in the mathematical algorithms of deriving structures from powder diffraction data. Already there are many examples of publications jointly reporting NMR and diffraction data, but the objective now is to integrate the two techniques more fully so as to properly utilise their combined power.