Monitoring Cocrystal Formation via In Situ Solid-State NMR - The

Sep 15, 2014 - A detailed understanding of the mechanism of organic cocrystal formation remains elusive. Techniques that interrogate a reacting system...
0 downloads 0 Views 2MB Size
Letter pubs.acs.org/JPCL

Monitoring Cocrystal Formation via In Situ Solid-State NMR Venkata S. Mandala,§ Sarel J. Loewus,§ and Manish A. Mehta* Department of Chemistry and Biochemistry, Oberlin College, 119 Woodland Street, Oberlin, Ohio 44074, United States S Supporting Information *

ABSTRACT: A detailed understanding of the mechanism of organic cocrystal formation remains elusive. Techniques that interrogate a reacting system in situ are preferred, though experimentally challenging. We report here the results of a solid-state in situ NMR study of the spontaneous formation of a cocrystal between a pharmaceutical mimic (caffeine) and a coformer (malonic acid). Using 13C magic angle spinning NMR, we show that the formation of the cocrystal may be tracked in real time. We find no direct evidence for a short-lived, chemical shift-resolved amorphous solid intermediate. However, changes in the line width and line center of the malonic acid methylene resonance, in the course of the reaction, provide subtle clues to the mode of mass transfer that underlies cocrystal formation. SECTION: Kinetics and Dynamics

C

spectra of CA and MA, respectively. The caffeine spectrum shows broad features, which are known to arise from disorder in the crystal structure.16 In the cocrystal, made by ball milling CA and MA, all the carbon sites in CA and MA exhibit sharp resonances (Figure 1c). In order to identify any subtle changes to spectral features in the course of cocrystal formation, we collected a spectrum of CA and MA simultaneously packed in an MAS rotor, separated by a 1 mm spacer disc (Figure 1d), and we repeated it for CA, MA, and the cocrystal (Figure 1e). Additional details are given in the Supporting Information. Natural abundance, in situ 13C spectra for a spontaneously cocrystallizing CA-MA system are shown in Figure 2. In order to accumulate sufficient signal for this reacting system, each spectrum was signal averaged for 1 h. Slow relaxation times prevent faster sampling at natural abundance at this magnetic field strength. The three-dimensional surface plot shows that the signals of the reactants and products may be tracked simultaneously in real time, using the same cross-polarization conditions. The reactants and products show similar temporal behavior. The contour map in Figure 2 (lower left) shows that the signals from the reactants vanish smoothly and those corresponding to the products appear thus in the course of the experiment. More importantly, no new transient signals are seen on the time scale of signal averaging. The 13C resonances from caffeine in the cocrystal are sharp compared with the broad resonances in the pure compound (Figure 1), which serves as a diagnostic of sorts for reaction progress. The upfield and downfield regions essentially show the reactant and product resonances superimposed, much as in Figure 1e. At this level of spectral dispersion, spectral signatures of any species that is not the reactant or product is likely to lie in subtle changes to their line shapes. However, this does not rule out the possibility of a transient intermediate with a lifetime shorter than the 1 h used to accumulate signal for each time

ocrystallization of active pharmaceutical ingredients (API) with an excipient coformer is known to provide an alternative means to alter bulk macroscopic properties, such as solubility and bioavailability.1 Many such cocrystalline phases are accessible via mechanochemical synthesis, which has been shown to provide polymorph control, as well as a solvent-free, green synthetic route.2,3 A small number of systems have been reported to form cocrystals spontaneously, with outcomes similar to grinding.4,5 In either case, mechanistic details of cocrystal formation remain poorly understood. The central question surrounds the issue of mass transfer: how do molecules leave their pure phase to achieve stoichiometric mixing at the unit cell level in the resulting cocrystal? The answer is likely a combination of three leading hypotheses: mediation by vapor diffusion, formation of a liquid phase by a submerged eutectic, or by crystallization from an intermediate amorphous solid phase.6 Monitoring the course of a reaction in the solid state using atomic-level spectroscopies typically necessitates ex situ analysis of the reacting mixture, which carries its own challenges to preserve sample fidelity and relevance. Recent groundbreaking studies have shown that mechanochemical syntheses may be monitored in situ, in real time using powder X-ray diffraction (PXRD),7,8 Raman scattering9 and terahertz spectroscopy.10 Among other applications, in situ NMR has been used to investigate reactions in the solid state,11 as well as crystallization phenomena from solutions.12−14 However, given the incongruence between conditions needed for grinding and the mechanical stability required for magic angle spinning (MAS), it has not been realistically possible to employ solid-state NMR for similar in situ studies of mechanochemical processes. Here, we report the results of a solid-state NMR study, wherein we tracked the spontaneous formation of a cocrystal of an API mimic (caffeine, CA) and a coformer (malonic acid, MA) using in situ MAS NMR. The caffeine−malonic acid pair15 is one of several reported to form a cocrystal spontaneously, by simply mixing their powders.4,5,15 Figure 1a and 1b show solid-state 13C NMR MAS © XXXX American Chemical Society

Received: August 11, 2014 Accepted: September 15, 2014

3340

dx.doi.org/10.1021/jz501699h | J. Phys. Chem. Lett. 2014, 5, 3340−3344

The Journal of Physical Chemistry Letters

Letter

Figure 1. Solid-state 13C CP-MAS NMR spectrum of (a) caffeine, CA; (b) malonic acid, MA; (c) the 2:1 CA-MA cocrystal; (d) CA and MA simultaneously in a rotor, separated by a 1 mm Kel-F spacer; and (e) CA:cocrystal:MA simultaneously in a rotor. Each spectrum is at natural abundance, 12 kHz MAS and 600.381 MHz (1H), externally referenced to the downfield resonance of adamantane at 40.49 ppm (which corresponds to 0 ppm for 0.5% DSS in D2O). Signals in the region between 60 and 100 ppm are spinning side bands of the downfield carbons.

Figure 2. Chemical shift (13C) in situ MAS spectra of the reacting CA and MA. The 13C spectra, over 84 h, are shown as a surface plot (upper left); the downfield, midfield, and upfield regions are shown as contour maps (lower left). The same regions of five representative spectra are shown on the right. Each spectrum was collected by signal averaging for 1 h (90 scans, 40s recycle delay). The MAS rate was 12 kHz throughout.

It is natural to suspect that the MAS conditions themselves may have an adverse effect on the reacting mixture.17 The contents of the 4 mm rotor experience substantial rotational

slice. Greater sample volume in a larger diameter MAS rotor would reduce the time needed for signal averaging at natural abundance. 3341

dx.doi.org/10.1021/jz501699h | J. Phys. Chem. Lett. 2014, 5, 3340−3344

The Journal of Physical Chemistry Letters

Letter

Figure 3. MAS data for a spontaneously cocrystallizing in situ run using (2-13C; 99 atom %)-MA. Panel (a) shows the integrated signal from the CH2 resonance of MA in the product cocrystal, and (b) shows the same for the unreacted species. A plot of the inverse of the integrated signal in (b) is shown in (c). A 3D surface plot of the methylene region of MA is shown in (d).

integrated signal of MA in the product cocrystal is shown in Figure 3a, and a similar plot of the unreacted MA in Figure 3b. The latter shows there is rapid conversion to the cocrystal in the initial 10−15 h, followed by continued, slower conversion. While a log plot of the unreacted MA signal shows curvature, a plot of the inverse signal is nearly linear (Figure 3c). On first blush, a linear plot of the inverse signal of the reactant suggests a second order process; however, we hesitate to translate that into a single rate-limiting step, as the underlying mechanism in the condensed phases may be much more complex. For example, mixing plays an important role. Blagden and co-workers have reported that the spontaneous formation of the CA-MA cocrystal depends strongly on the grain size.4 In replicating their results, we have observed that the details of the mixing have a dramatic impact on the rate and extent of conversion. Our challenge was to mix the powders quickly in such a way as to allow the crystallites to come into maximum physical contact while maintaining grain integrity−that is, mixing without resorting to harsh milling conditions. The rotational vortexing method we used to mix the powders to initiate the cocrystallization reaction appears to lead to good, albeit not fully complete, mixing. The resulting mixture likely has small islands of physically segregated reactants in the MAS rotor. It is possible that more than one mode of mass transfer may be at work, wherein crystallites in physical contact react quickly, accounting for the rapid conversion early in the course of the reaction. The slower conversion seen at longer times could be due to a noncontact process, such as vapor diffusion in the sealed MAS rotor. Either, or both, could result in the phenomenological behavior we see. The fact that we tracked the reacting mixture,

acceleration, and subsequently high pressure, at 12 kHz MAS. That added pressure on the sample could alter the process under study. To test this, we subjected a reacting mixture to elevated pressure using a pellet press typically used to make KBr pellets for IR spectroscopy. A reaction was initiated by mixing the powders, then split into two portions: one left out at ambient conditions as a control and the other put into the press. The press was maintained at roughly 5000 bar for 5.5 h, which is greater than the pressure experienced by the reacting mixture under MAS.18 A PXRD pattern of the control and the pressed pellet showed that, if anything, the reacting mixture in press progressed a little more slowly than the control. No new phases, beyond the reactants and the cocrystal product, are seen at elevated pressure. (The PXRD patterns and experimental details are given in the Supporting Information.) We therefore conclude that the forces experienced by the reacting mixture under MAS are not altering the outcome in an unexpected way. In order to decrease the sampling time, we repeated the natural abundance in situ run using (2-13C)-labeled malonic acid. The enhanced signal from the methylene carbon of malonic acid allowed us to take a snapshot of the reacting mixture every 80s, versus 1 h for the natural abundance runs. On this time scale, we see no evidence for an intermediate that exhibits a different chemical shift. Instead, we see the pure malonic acid signal (42.84 ppm) gradually give way to the malonic acid signal in the cocrystal (45.75 ppm). Ball milling the almost-converted contents of the MAS rotor, for both the natural abundance and 13C-labeled runs, achieves complete conversion to the cocrystal, as confirmed by CP-MAS NMR and PXRD. The three-dimensional surface plot of signal versus time is shown in Figure 3d. The temporal profile of the 3342

dx.doi.org/10.1021/jz501699h | J. Phys. Chem. Lett. 2014, 5, 3340−3344

The Journal of Physical Chemistry Letters

Letter

Figure 4. Fwhm spectral line width (Γ, in hertz) and the line center (ν0, in ppm) of the methylene resonance of malonic acid in the cocrystal product (a, b) and the reactant (c, d), respectively. The values were determined using a Lorentzian fit to each line shape. Data taken from the (2-13C)-MA run, shown in Figure 3. Line widths and spectral centers are plotted on the same vertical scale. The data for the reactant (c, d) become noisy as the reaction approaches completion, as there is little remaining signal and correspondingly high relative error.

grains. These line shape data will provide valuable constraints on any future mechanistic models of the cocrystallization process. We have shown that solid-state NMR is a useful tool for the kinetic investigations of cocrystal formation. Previous in situ studies that employed PXRD report only phases that diffract, whereas solid-state NMR is sensitive to a wider variety of solid phases. Thus, solid-state NMR provides atomic-level information that is complementary to diffraction-based methods.12,19 Depending on the questions asked, the in situ NMR strategy may be tailored using the many sophisticated NMR experiments that have been developed, including single-scan 2D methods.20,21 For the CA−MA system investigated here, we do not find evidence of an amorphous intermediate solid phase, one that is distinguished by a set of chemical shifts different from the product and reactants. The time scale of sampling, spectral dispersion, and the sensitivity of our experiments place constraints on the transiency of such an intermediate and subtle changes to the line shapes of the malonic acid CH2 resonances provide clues regarding the underlying mechanism of mass transfer. The linearity of the plot in Figure 3c suggests second order behavior, although more experiments are needed to uncover the molecular details. These include investigation of the process from the perspective of the API mimic (CA), isolation of the effects of direct and nondirect crystallite contact, and utilizing information from the full chemical shift tensor using slow MAS. Traditional Arrhenius-type variable temperature experiments will likely provide much-needed detail of the barrier energetics, although for heterogeneous processes in the solid state, the interpretation of that data is far from straightforward.22 Our observations also underscore the need

undisturbed, has allowed us to detect both rapid and slow conversion. Had the reaction been tracked under milling conditions, the energetic grinding (and the concomitant mixing) would have overpowered the slower conversion seen at later times in our experiments. On further investigation, this could turn out to be one of the major advantages of solid-state NMR over other forms of spectroscopic in situ monitoring of solid-state heterogeneous reactions. The NMR data may be further mined for additional clues to the underlying processes. We undertook a line shape analysis of the data shown in Figure 3a and b. The methylene resonance line shapes of the unreacted MA and the cocrystal MA are described well by a single Lorentzian. The spectral centers (ν0) and fwhm line widths (Γ) of the two line shapes, over the course of the reaction, are shown in Figure 4. The MA 13CH2 line width in the product cocrystal (Figure 4a) steadily narrows in the course of the reaction from approximately 80 to 45 Hz, whereas the spectral center (Figure 4b) shows a slight (0.02 ppm) downfield shift. The line width of the same resonance in the unreacted MA is initially 40 Hz, which then decreases to almost 30 Hz, then increases over time to over 45 Hz (Figure 4c). The center of that resonance shows a monotonic shift upfield by nearly 0.1 ppm in the course of the reaction (Figure 4d). These are subtle, albeit reproducible effects. The line width data, in particular, suggest a combination of homogeneous and inhomogeneous line broadening that might be at work in the course of the reaction, which may reflect changes in the dihedral angles surrounding the reporter nucleus (13CH2 in MA) at different stages of transport along the reaction coordinate. The broadening may also be due to changes in the anisotropic bulk magnetic susceptibility of the malonic acid 3343

dx.doi.org/10.1021/jz501699h | J. Phys. Chem. Lett. 2014, 5, 3340−3344

The Journal of Physical Chemistry Letters

Letter

for high magnetic field strengths for the study of small molecule systems. A detailed understanding of molecular diffusion that leads to the spontaneous formation of cocrystal, one that is consistent with the spectral data reported here and elsewhere, remains elusive. As such, spontaneously cocrystallizing systems may prove to be a useful appurtenance of mechanochemical processes, the mechanisms of which are also poorly characterized at this time.



Milling Reactions by Raman Spectroscopy. Angew. Chem., Int. Ed. 2014, 126, 6307−6311. (10) Nguyen, K. L.; Frišcǐ ć, T.; Day, G. M.; Gladden, L. F.; Jones, W. Terahertz Time-Domain Spectroscopy and the Quantitative Monitoring of Mechanochemical Cocrystal Formation. Nat. Mater. 2007, 6, 206−209. (11) Haw, J. F.; Nicholas, J. B.; Xu, T.; Beck, L. W.; Ferguson, D. B. Physical Organic Chemistry of Solid Acids: Lessons From in Situ NMR and Theoretical Chemistry. Acc. Chem. Res. 1996, 29, 259−267. (12) Hughes, C. E.; Williams, P. A.; Peskett, T. R.; Harris, K. D. M. Exploiting in Situ Solid-State NMR for the Discovery of New Polymorphs During Crystallization Processes. J. Phys. Chem. Lett. 2012, 3, 3176−3181. (13) Hughes, C. E.; Harris, K. D. M. Direct Observation of a Transient Polymorph During Crystallization. Chem. Commun. 2010, 46, 4982−4984. (14) Hughes, C. E.; Harris, K. D. M. A Technique for in Situ Monitoring of Crystallization From Solution by Solid-State 13C CPMAS NMR Spectroscopy. J. Phys. Chem. A 2008, 112, 6808−6810. (15) Trask, A. V.; Motherwell, W. D. S.; Jones, W. Pharmaceutical Cocrystallization: Engineering a Remedy for Caffeine Hydration. Cryst. Growth Des. 2005, 5, 1013−1021. (16) Enright, G. D.; Terskikh, V. V.; Brouwer, D. H.; Ripmeester, J. A. The Structure of Two Anhydrous Polymorphs of Caffeine From Single-Crystal Diffraction and Ultrahigh-Field Solid-State 13C NMR Spectroscopy. Cryst. Growth Des. 2007, 7, 1406−1410. (17) Xu, M.; Harris, K. D. M. Altering the Polymorphic Product Distribution in a Solid-State Dehydration Process by Rapid Sample Rotation in a Solid-State NMR Probe. J. Am. Chem. Soc. 2005, 127, 10832−10833. (18) Asano, A.; Hori, S.; Kitamura, M.; Nakazawa, C. T.; Kurotsu, T. Influence of Magic Angle Spinning on T1H of SBR Studied by Solid State 1H NMR. Polymer J. 2012, 44, 706−712. (19) Vogt, F. G.; Clawson, J. S.; Strohmeier, M.; Edwards, A. J.; Pham, T. N.; Watson, S. A. Solid-State NMR Analysis of Organic Cocrystals and Complexes. Cryst. Growth Des. 2009, 9, 921−937. (20) Tal, A.; Frydman, L. Single-Scan Multidimensionl Magnetic Resonance. Prog. Nucl. Mag. Res. 2010, 57, 241−292. (21) Hughes, C. E.; Williams, P. A.; Harris, K. D. M. “CLASSIC NMR”: an in-Situ NMR Strategy for Mapping the Time-Evolution of Crystallization Processes by Combined Liquid-State and Solid-State Measurements. Angew. Chem., Int. Ed. 2014, 53, 8939−8943. (22) Khawam, A.; Flanagan, D. R. Basics and Applications of SolidState Kinetics: a Pharmaceutical Perspective. J. Pharm. Sci. 2006, 95, 472−498.

ASSOCIATED CONTENT

S Supporting Information *

Details of the sample preparation, NMR experiments, powder X-ray diffraction data, elevated pressure experiments, and multiply packed rotors. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions §

V.S.M. and S.J.L. contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Matthew J. Elrod (Oberlin College) for his assistance in the interpretation of kinetic data. M.A.M. gratefully acknowledges helpful conversations with Siddhartha Sarma (IISc, Bangalore, India) during the inception of this work. This work was supported by NSF-RUI grant CHE-1012813, NSFMRI grant DMR-0922588 for the acquisition of a powder X-ray diffractometer, and a Henry Dreyfus Teacher−Scholar grant (M.A.M.) from the Camille and Henry Dreyfus Foundation.



REFERENCES

(1) Steed, J. W. The Role of Co-Crystals in Pharmaceutical Design. Trends Pharmacol. Sci. 2013, 34, 185−193. (2) Braga, D.; Maini, L.; Grepioni, F. Mechanochemical Preparation of Co-Crystals. Chem. Soc. Rev. 2013, 42, 7638−7648. (3) Frišcǐ ć, T. Supramolecular Concepts and New Techniques in Mechanochemistry: Cocrystals, Cages, Rotaxanes, Open Metal− Organic Frameworks. Chem. Soc. Rev. 2012, 41, 3493−3510. (4) Ibrahim, A. Y.; Forbes, R. T.; Blagden, N. Spontaneous Crystal Growth of Co-Crystals: the Contribution of Particle Size Reduction and Convection Mixing of the Co-Formers. CrystEngComm 2011, 13, 1141−1152. (5) Maheshwari, C.; Jayasankar, A.; Khan, N. A.; Amidon, G. E.; Rodríguez-Hornedo, N. Factors That Influence the Spontaneous Formation of Pharmaceutical Cocrystals by Simply Mixing Solid Reactants. CrystEngComm 2009, 11, 493−500. (6) Frišcǐ ć, T.; Jones, W. Recent Advances in Understanding the Mechanism of Cocrystal Formation via Grinding. Cryst. Growth Des. 2009, 9, 1621−1637. (7) Halasz, I.; Puškarić, A.; Kimber, S. A. J.; Beldon, P. J.; Belenguer, A. M.; Adams, F.; Honkimäki, V.; Dinnebier, R. E.; Patel, B.; Jones, W.; Štrukil, V.; Frišcǐ ć, T. Real-Time in Situ Powder X-Ray Diffraction Monitoring of Mechanochemical Synthesis of Pharmaceutical Cocrystals. Angew. Chem., Int. Ed. 2013, 52, 11538−11541. (8) Frišcǐ ć, T.; Halasz, I.; Beldon, P. J.; Belenguer, A. M.; Adams, F.; Kimber, S. A. J.; Honkimäki, V.; Dinnebier, R. E. Real-Time and In Situ Monitoring of Mechanochemical Milling Reactions. Nat. Chem. 2012, 5, 66−73. (9) Gracin, D.; Štrukil, V.; Frišcǐ ć, T.; Halasz, I.; Užarević, K. Laboratory Real-Time and in Situ Monitoring of Mechanochemical 3344

dx.doi.org/10.1021/jz501699h | J. Phys. Chem. Lett. 2014, 5, 3340−3344