Powder Crystallography by NMR – Structure determination of microcrystalline powdered compounds without modification at natural abundance

.

Powder Crystallography by NMR – Structure determination of microcrystalline powdered compounds without modification at natural abundance

Characterization of the crystal structure of powdered molecular crystals is a considerable challenge, but one that enables a clearer understanding of the solid form, helping to inform manufacturing processes. This is even more challenging for pharmaceutical compounds that often exhibit polymorphic behavior {1}.
Quickly becoming the method of choice, solid-state nuclear magnetic resonance (NMR) characterizes powder forms at natural abundance whilst minimizing the risk of polymorphic change during the characterization process {2}. The sensitivity limitations of natural abundance solid-state NMR are overcome by focusing on the ¹H nucleus, with 100% natural abundance and a high gyromagnetic ratio.
A recent development has resulted in the establishment of a protocol that enables complete, ab initio, molecular structure determination in powders by NMR using ¹H-¹H spin diffusion, molecular modeling, periodic DFT calculations and chemical shifts {REF previous}.
This method studies the powder as is, without modification and at natural abundance, delivering structure determination of microcrystalline powdered compounds especially for those not suitable for other methods, or that can undergo changes in polymorphs during sample preparation.
Using the example of thymol {3}, a small drug molecule, the protocol is composed of two main steps illustrated in Fig.1.

Fig.1: Protocol for powder NMR crystallography (reproduced from ref 3)

First, the approximate powder crystal structure of thymol is determined by molecular modeling, in which the leading force is the agreement between experimental and calculated ¹H-¹H spin diffusion experiments. The resulting crystal structure is then refined using periodic DFT calculations {REF Castep}, and validated using comparison between measured chemical shifts and calculations.
Crystal structure determination of powdered thymol
The first essential step, the assignment of the NMR spectra (Fig. 2), is done for carbons using ¹³C-¹³C connectivities as determined from a refocused INADEQUATE {4}.

Fig.2 : INADEQUATE and INEPT experiments used to assign 13C and 1H resonances

Assignment of the ¹H resonances is then conducted with an INEPT {5} experiment,

showing the ¹H-¹³C connectivities. The large ¹H homonuclear dipolar couplings are reduced using advanced Combined Rotation And Multi-Pulse Spectroscopy (CRAMPS) to give access to high-resolution proton chemical shift information.

The pulse scheme shown in Fig. 3 is then used to obtain a series of two-dimensional ¹H-¹H spin-diffusion correlation spectra. Dipolar driven ¹H magnetization exchange occurs during τ_SD, and so each peak in the 2D spectrum contains information about proton internuclear distances. The dynamics of these magnetization exchanges can be followed by recording the evolution of the ¹H-¹H spin-diffusion spectrum and varying the spin-diffusion delay τ_SD. To this end the measured volume of each peak in the 2D spectrum is plotted as a function of τ_SD to obtain the series of build-up curves shown in Fig. 3.
Predicted build-up curves can be calculated for a given trial structure using a rate matrix description {6}. Fig. 3 shows the comparison between the experimental curves and the build-up curves simulated for the known single-crystal X-ray determined structure.
Structure determination is achieved by molecular modeling, restrained by experimental build-up curves to obtain an ensemble of potential thymol structures. The method is outlined in Fig. 4.

Fig.3: Pulse sequence for the 2D 1H-1H spin diffusion experiments and build-up curves. Experimental data are shown as circles and simulated build-up curves for the single-crystal X-ray structure are plotted as red lines. DUMBO-1{7} and eDUMBO-122{8} are proton homonuclear dipolar decoupling sequences (reproduced from ref 3).

Starting from 3000 crystal structures with random atomic coordinates, the 300 structures that are in best agreement with the experiment are selected and optimized. Finally, the lowest E_PSD crystal structures that are inside the estimated uncertainty are selected, resulting in 42 structures, with an all-atom standard deviation within the ensemble of 0.25 Å.
The second step of refinement in CASTEP produces the final structure that is validated independently by comparison of experimental chemical shifts and chemical shifts calculated using the GIPAW approach {REF GIPAW}. This final structure is identical to the single crystal X-ray determined structure.

References
{1} Brittain, H. G. J. Pharm. Sci. 2008, 97, 3611-3636.
{2} Harris, R. K. Analyst 2006, 131, 351-373.
{REF previous} (a) B. Elena and L. Emsley, J. Am. Chem. Soc., 2005, 127, 9140–9146. 10 (b) B. Elena, G. Pintacuda, N. Mifsud and L. Emsley, J. Am. Chem. Soc., 2006, 128, 9555–9560. 11 (c) C. J. Pickard, E. Salager, G. Pintacuda, B. Elena and L. Emsley, J. Am. Chem. Soc. 2007, 129, 8932-8933.
{3} Salager, E.; Stein, R. S.; Pickard, C. J.; Elena, B.; Emsley, L. Phys. Chem. Chem. Phys. 2009, 11, 2610-2621.
{REF Castep} S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. J. Probert, K. Refson and M. C. Payne, Z. Kristallogr., 2005, 220, 567–570.
{4} Lesage, A.; Bardet, M.; Emsley, L. J. Am. Chem. Soc. 1999, 121, 10987-10993.
{5} Elena, B.; Lesage, A.; Steuernagel, S.; Bockmann, A.; Emsley, L. J. Am. Chem. Soc. 2005, 127, 17296-17302.
{6} Elena, B.; Emsley, L. J. Am. Chem. Soc. 2005, 127, 9140-9146.
{7} D. Sakellariou, A. Lesage, P. Hodgkinson and L. Emsley, Chem. Phys. Lett., 2000, 319, 253–260.
{8} Elena, B.; de Paepe, G.; Emsley, L. Chem. Phys. Lett. 2004, 398, 532.
{REF GIPAW} (a)Pickard C.J. and Mauri F. Phys. Rev. B 2001, 63, 245101 (b) Yates J. R., Pickard C.J. and Mauri F., Physical Review B, 2007, 76, 024401

 ...........................