Letter pubs.acs.org/ac
Chemical Mapping of Pharmaceutical Cocrystals Using Terahertz Spectroscopic Imaging Danielle M. Charron,† Katsuhiro Ajito,* Jae-Young Kim, and Yuko Ueno NTT Microsystem Integration Laboratories, NTT Corporation, 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 243-0198 Japan S Supporting Information *
ABSTRACT: Terahertz (THz) spectroscopic imaging is a promising technique for distinguishing pharmaceuticals of similar molecular composition but differing crystal structures. Physicochemical properties, for instance bioavailability, are manipulated by altering a drug’s crystal structure through methods such as cocrystallization. Cocrystals are molecular complexes having crystal structures different from those of their pure components. A technique for identifying the twodimensional distribution of these alternate forms is required. Here we present the first demonstration of THz spectroscopic imaging of cocrystals. THz spectra of caffeine−oxalic acid cocrystal measured at low temperature exhibit sharp peaks, enabling us to visualize the cocrystal distribution in nonuniform tablets. The cocrystal distribution was clearly identified using THz spectroscopic data, and the cocrystal concentration was calculated with 0.3−1.3% w/w error from the known total concentration. From this result, THz spectroscopy allows quantitative chemical mapping of cocrystals and offers researchers and drug developers a new analytical tool.
T
determinant role in dictating the physicochemical properties of the cocrystal such as solubility, solution stability, and processability.12 Physicochemical properties of crystalline materials are dependent not only on their molecular species but on their crystal structures. By adjusting synthesis conditions, cocrystals may take multiple polymorphic forms owing to their conformational flexibility.13 In the pharmaceutical industry, the purpose of crystallizing cocrystals and polymorphs is to improve drug properties in order to enhance bioavailability without altering the chemical nature, and thus bioactivity, of an API. Differences in crystal structure between cocrystallized and pure pharmaceuticals, therefore, affect the dosage response of a drug. Furthermore, chemical imaging methods are required for observing and evaluating the distribution of cocrystallized pharmaceuticals in tablets. Chemical mapping has been used to distinguish genuine and counterfeit drugs, as API distribution reflects tablet processing conditions.14 In addition, spatial nonuniformity affects drug release kinetics and therapeutic efficiency;15 therefore, chemical mapping is important in process monitoring and control. Techniques such as Raman microscopy and near-infrared spectroscopic imaging are advantageous for chemical mapping near the tablet surface; however, strong scattering precludes bulk imaging. THz radiation is lower in energy and so is capable of penetrating pharmaceutical tablets16,17 and biological
erahertz (THz) spectroscopy is a useful technique for chemical recognition based on both molecular and crystal structure fingerprinting. Occupying the part of the electromagnetic spectrum between microwaves and infrared radiation, the THz region corresponds to the frequency range between 0.1 and 10 THz (3−333 cm−1). Absorption of THz waves is dominated by weak intermolecular and intramolecular interactions within molecular networks, such as hydrogen bonding and van der Waals forces. Separation of intermolecular and intramolecular modes for assignment of absorption bands1−3 is both challenging and ambiguous, making it difficult to isolate molecular and crystal structures. It is useful, however, that these modes are highly collective, involving vibrations in multiple regions of a molecule.4 Thus, THz spectra are more strongly dependent on molecular arrangement and intermolecular interactions than on molecular composition. For this reason, closely related chemicals such as polymorphs can easily be distinguished by THz spectroscopy.5−8 Similarly, cocrystals can be differentiated from mixtures of their pure components. Cocrystals are solid molecular complexes of two or more neutral components bound together in stoichiometric amounts by noncovalent intermolecular interactions.9−11 Cocrystallization plays a similar role to polymorphism in pharmaceutical development. Pharmaceutical cocrystals contain an active pharmaceutical ingredient (API). In its pure form the API is itself crystalline, but after cocrystallization the API molecules are interspersed within the new crystal structure. This new structure is comprised of the API molecules and one or more accessory component molecules. These additional components are called coformers, and their selection plays a © 2013 American Chemical Society
Received: October 30, 2012 Accepted: January 28, 2013 Published: January 28, 2013 1980
dx.doi.org/10.1021/ac302852n | Anal. Chem. 2013, 85, 1980−1984
Analytical Chemistry
Letter
samples,18 enabling nondestructive characterization of the bulk chemical distribution. It has also been shown that quantitative analysis is possible with THz spectroscopy.19−21 Previous THz spectroscopic imaging analysis has focused on the comparison of absorption images at different THz frequencies to identify large regions of desired chemicals within nonuniform tablets6 and to distinguish homogeneous tablets comprised of different chemicals.7,22,23 Moreover, low temperature measurement has been shown to be advantageous for chemical mapping since measured THz spectra exhibit sharper peaks due to narrowing in the distribution of populated vibrational modes.11 Here we present the first demonstration of chemical mapping of cocrystal distributions. Imaging test tablets including nonuniform caffeine−oxalic acid cocrystal powders were prepared and investigated by THz spectroscopic imaging. The cocrystal of the API caffeine and the coformer oxalic acid was chosen, as the crystal structure is simple and known in pharmaceutical research.24 Spectral analysis was performed to quantitatively determine the distribution of caffeine−oxalic acid, identifying the cocrystal as separate from its pure components caffeine and oxalic acid. The total cocrystal concentration was determined with between 0.3 and 1.3% w/ w error from the known concentration.
pulse laser (Fusion, Femtolasers). Spectra of the 0, 5, 10, 15, and 20% cocrystal, 20% caffeine, and 20% oxalic acid tablets were collected at 120 K in a vacuum chamber and averaged over 32 acquisitions. All spectra were normalized for tablet thickness, and the pure polyethylene spectrum was subtracted from each sample spectrum. THz-Imaging. The THz-imaging setup is shown in Figure 1. It consists of a THz-TDS system using a 9 fs near-infrared
Figure 1. Diagram of the terahertz imaging system.
pulse laser (Integral Pro, Femtolasers), with the sample chamber mounted on a translational stage. A detailed description of the apparatus was reported elsewhere.7 A cryostat was also added to the sample chamber of the THzimaging system. Images with a step size of 200 μm and spatial resolution of about 600 μm at 1 THz were taken, rastering over a 12 mm × 12 mm scanning area. The two nonuniform tablets and the 0, 10, and 20% cocrystal tablets were imaged. All images were collected at the lowest preset temperature achievable with the THz-imaging system, 120 K under vacuum, and took approximately 7 h to complete. The THz-imaging temperature was confirmed by THz-TDS cocrystal peak position temperature calibration. Samples were set into a metal holder mounted to a quartz plate so that the metal formed a ring around each tablet. Pixels corresponding to the metal sample holder were removed from the image data. Each image was also normalized for tablet thickness. This and all further spectral analysis was completed using MATLAB (MathWorks).
■
EXPERIMENTAL SECTION Cocrystal Synthesis. Synthesis of caffeine−oxalic acid cocrystal was performed following a modified version of a previously reported solution precipitation technique.20 A 2:1 molar ratio of caffeine (Tokyo Chemical Industry Co., Ltd.) and anhydrous oxalic acid (Wako Pure Chemical Industries, Ltd.) dissolved in chloroform and methanol (Kanto Chemical Co., Inc.) was heated at 330 K for 20 min under reflux. The solution was allowed to cool to room temperature, then rotoevaporated for 15 min. A white, cotton-like crystal was extracted by vacuum filtration and then dried under low vacuum overnight. Powder X-ray Diffraction (PXRD). Cocrystallization was verified by PXRD using a Rigaku RU-300 rotating-anode generator (0.154 nm Cu source, 40 kV voltage, 200 mA filament emission). Powdered cocrystal and caffeine samples were scanned from 7 to 30° 2θ with a step size of 0.02° 2θ. Raman Spectroscopy. Cocrystallization was also confirmed by Raman spectrometry (inVia Reflex Raman microscope, Renishaw) using a 532 nm line laser-diode continuouswave laser (H39969, Renishaw) at 1 mW power. Spectra of powdered samples were collected at room temperature, with an objective lens (×100, NA 1.85) providing a spectral resolution of 1 cm−1. Tableting. Polyethylene was selected as a test tablet matrix material because of its low absorption and the absence of large peaks in its terahertz spectrum. The cocrystal was ground to powder and mixed with polyethylene (Aldrich) to dilute to 0, 5, 10, 15, and 20% cocrystal by weight. The homogeneous sample mixtures were compressed at 40 kg/cm2 for 2 min to form tablets 10 mm in diameter, 1.5−2.0 mm thick. Homogeneous tablets of 20% caffeine or oxalic acid in polyethylene were similarly prepared. Two nonuniform samples were made by arranging a mixture of cocrystal and polyethylene on one side of a tablet, with either pure polyethylene or a mixture of polyethylene, caffeine, and oxalic acid on the opposite side. THz-TDS. Cocrystallization was confirmed by THz timedomain spectroscopy (THz-TDS) using a commercial spectroscope (THz-TDS 2004, AISPEC) with a 13 fs near-infrared
■
RESULTS AND DISCUSSION THz-TDS Cocrystal Spectra. Cocrystallization was readily confirmed by THz-TDS. Caffeine−oxalic acid exhibits three strong absorption peaks (νi) at 1.35, 1.43, and 1.56 THz at around 120 K, whereas caffeine and oxalic acid exhibit no absorption peaks within the 1−2 THz range studied (Figure 2a). The oxalic acid spectrum compares well with a previously reported THz spectrum.25 Cocrystallization was verified by PXRD (Figure S-1 in the Supporting Information). The caffeine and cocrystal diffraction patterns are similar to reported measurements.26 These results are also supported by Raman spectroscopy (Figure S-2 in the Supporting Information). The cocrystal Raman spectrum exhibits a distinct new peak at 899 cm−1 that is not observed in either the pure caffeine or oxalic acid spectra. Caffeine and oxalic acid Raman spectra are similar to reported spectra.27,28 Figure 2b illustrates the changes in cocrystal spectral shape measured from 298 K down to 120 K. The result clearly indicates that terahertz spectra measured at low temperatures exhibit sharper peaks. A blue shift is also observed with decreasing temperature. 1981
dx.doi.org/10.1021/ac302852n | Anal. Chem. 2013, 85, 1980−1984
Analytical Chemistry
Letter
Figure 2. (a) Terahertz spectra of caffeine (dashed line), oxalic acid (dotted line), and their cocrystal (solid line) measured at 120 K. The concentration of each tableted chemical is 20%. (b) Temperature-dependent THz spectra of caffeine-oxalic acid cocrystal measured at 298 K (red), 220 K (green), 170 K (blue), and 120 K (black). (c) Concentration-dependent THz spectra of caffeine-oxalic acid cocrystal measured at 120 K. Tableted cocrystal concentrations are 5% (black), 10% (blue), 15% (green), and 20% (red).
THz-TDS was used to verify observed trends in THzimaging and any spectral analysis assumptions made. Each image pixel corresponds to only one spectral acquisition, whereas the THz-TDS spectra are averaged over multiple acquisitions, providing much higher quality spectra. For the 5, 10, 15, and 20% cocrystal THz spectra at 120 K (Figure 2c), the trend in optical density at ν3 is well approximated by a linear polynomial (Figure S-3 in the Supporting Information, y = ax + b, a = 0.026, b = 0.050). Attenuation of THz waves is therefore dependent on cocrystal concentration, satisfying the Beer− Lambert Law for absorbance. From this result it is expected that the THz-imaging system can be calibrated linearly and that THz spectroscopy may be used for quantitative analysis. THz Mapping System Performance. The most intense spectral peak ν3 will be used to identify the cocrystal because it offers the highest signal-to-noise ratio and is separate from ν1 and ν2. THz images of the 0, 10, and 20% cocrystal tablets (Figure 3) were used to calibrate the imaging system for cocrystal concentration. The ν3 peak transmittance was averaged across all pixels within the tablet region for each calibration standard image. Transmittance as a function of sample concentration provided a linear calibration (Figure S-4 in the Supporting Information, y = ax + b, a = −0.942 ± 0.109, b = 66.0 ± 1.20). Error is due primarily to nonuniformity in cocrystal concentration within the calibration standard images and due to fluctuation in THz pulse intensity. No shift in peak frequency was detected across the images, so it is unlikely that variance observed is due to a change in temperature. The 10% cocrystal sample was slightly damaged along the edge prior to imaging; however, the average transmission 54.5 cm−1 has a standard deviation of only 1.16 cm−1, comparable to the 20% cocrystal tablet. Nonuniform Cocrystal Mapping. Figure 4a shows visible light photographs of the two nonuniform sample tablets. Regions of concentrated chemicals are indicated in Figure 4b, though their distribution cannot be seen in the photographs or in the THz pulse intensity images (Figure 4c). THz spectral data was used to map these unknown regions. Cocrystal pixels within the two nonuniform tablets were identified by the presence of ν3. A frequency range is used for identification instead of the absolute peak frequency to account for high noise in the image spectra and local temperature fluctuations. Presence of this peak was determined within ν3 ± 0.5Δν3 THz. Within this range, the minimum transmittance is assumed to be the peak intensity only if the intensities at the two neighboring frequencies are comparable to each other. If this is not true, the minimum transmittance within the studied range
Figure 3. Chemical mapping of (a) 20%, (b) 10%, and (c) 0% caffeine−oxalic acid cocrystal calibration standard tablets using the THz-imaging system set to 120 K. Left, THz pulse amplitude image (dark blue corresponds to metal sample holder); right, transmittance at ν3 peak.
is assumed to be due to noise, and instead the absolute transmittance at ν3 is taken as the peak intensity. The cocrystal concentration (Figure 4d) for each pixel previously identified as containing cocrystal in the two nonuniform tablets was determined using the THz-imaging concentration calibration calculated previously. The absolute error in the calculated concentration due to variation in the homogeneous calibration standards is given in Figure 4e. Propagated error is proportional to the calculated concentration, not to the concentration-dependent signal-to-noise ratio of ν3. For the tablet containing a nonuniform distribution of caffeine−oxalic acid in polyethylene, the total cocrystal concentration was found to be 5.25% by weight, only 0.352% more than the known concentration. Similarly, for the tablet 1982
dx.doi.org/10.1021/ac302852n | Anal. Chem. 2013, 85, 1980−1984
Analytical Chemistry
Letter
spectrum database of cocrystals and other excipients will allow us to analyze each component quantitatively, even if several kinds of chemicals are mixed in the focal spot of the terahertz beam.29 To improve the spatial resolution beyond the optical diffraction limit, near-field THz imaging techniques with subwavelength resolution have been proposed.30 Recent progress in near-field imaging has demonstrated spatial resolution down to λ/1000 (150 nm at 2 THz)31,32 which may allow THz imaging of pharmaceutical crystal powders. In addition, the spatial resolution in the z-direction of our system is over 2 mm; therefore, no z-direction information is included in the THz mapping data. The use of confocal imaging techniques33,34 in the THz mapping system would provide zdirection chemical distribution in tablets. The use of frequency tunable THz lasers instead of THz pulse beams is a very important factor in reducing measurement time for practical applications by reducing the delay-line scanning time. Such suitable lasers for terahertz spectroscopic imaging systems are now under development.
■
CONCLUSIONS
■
ASSOCIATED CONTENT
We have shown the first report demonstrating THz chemical mapping of pharmaceutical cocrystals. Cocrystallization is a method used to manipulate the physicochemical properties of drugs. Accurate chemical mapping is important for drug design and development. THz spectra of caffeine-oxalic acid cocrystal measured at low temperature exhibit sharp peaks, enabling us to visualize the distribution of cocrystal in tablets. For the nonuniform tablets, the cocrystal caffeine−oxalic acid was successfully mapped based on its spectral information as separate from its pure precursor components, caffeine and oxalic acid. Quantitative analysis using a linear concentration calibration was also shown to be accurate for determining the total cocrystal concentration. This result broadens the prospective applications of THz spectroscopy in pharmaceutical chemical imaging.
Figure 4. Chemical mapping of caffeine−oxalic acid cocrystal in nonuniform tablets with the THz-imaging system set to 120 K. Left, cocrystal in polyethylene; right, cocrystal in polyethylene with caffeine and oxalic acid. (a) Photograph of imaging tablet, (b) diagram of regions of high sample concentration within tablet, (c) THz pulse amplitude image, (d) cocrystal fraction by weight at ν3 peak for each identified pixel, and (e) absolute error in calculated cocrystal concentration. Approximate tablet position is outlined by a dashed black line in parts d and e.
S Supporting Information *
X-ray diffraction patterns and Raman spectra of powdered samples and calibration plots for cocrystal concentration using THz-TDS and THz-imaging. This material is available free of charge via the Internet at http://pubs.acs.org.
containing cocrystal in caffeine, oxalic acid, and polyethylene, the total cocrystal concentration was calculated to be 8.61%, an overestimate of 1.24%. Overestimation may be due to error in the concentration calibration or to fluctuations in temperature and THz pulse intensity during imaging. In addition, at the interface between the tablet and the metal sample holder, THz radiation is scattered independent of frequency due to the change in density between the two materials. Therefore, transmittance near the edge of the imaged tablet may be lower than without this halo effect. For both nonuniform tablets, the cocrystal was successfully mapped based on its spectral information as separate from the diluent polyethylene and its pure precursor components, caffeine and oxalic acid. For more practical investigation of tablets in the pharmaceutical market, various kinds of tablet matrix materials should be tested. Furthermore, more complex distributions of cocrystals should be analyzed, including evaluation of smaller-scaled domains and differences between surface and inner regions of tablets. Assembling a THz
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +81-46-240-3565. Fax: +81-46-240-4041. E-mail: ajito.
[email protected]. Present Address †
University of Waterloo, 200 University Ave. W., Waterloo Ontario, N2L 3G1 Canada. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
We thank Ms. Maro Yamaguchi of WDB for her kind assistance with the THz measurements and Dr. Ho-Jin Song, Mr. Makoto Yaita, Dr. Naoya Kukutsu, and Mr. Osamu Kagami of NTT for their encouragement. 1983
dx.doi.org/10.1021/ac302852n | Anal. Chem. 2013, 85, 1980−1984
Analytical Chemistry
■
Letter
REFERENCES
(1) Ueno, Y.; Rungsawang, R.; Tomita, I.; Ajito, K. Chem. Lett. 2006, 35, 1128−1129. (2) Ueno, Y.; Ajito, K. Anal. Sci. 2007, 23, 803−807. (3) Williams, M.; True, A.; Izmaylov, A.; French, T.; Schroeck, K.; Schmuttenmaer, C. Phys. Chem. Chem. Phys. 2011, 13, 11719−11730. (4) Castro-Camus, E.; Johnston, M. Chem. Phys. Lett. 2008, 455, 289−292. (5) King, M.; Buchanan, W.; Korter, T. Anal. Chem. 2011, 83, 3786− 3792. (6) Ajito, K.; Ueno, Y.; Song, H.-J.; Tamechika, E.; Kukutsu, N. ECS Trans. 2011, 35, 157−165. (7) Ajito, K.; Ueno, Y.; Song, H.-J.; Tamechika, E.; Kukutsu, N. Mol. Cryst. Liq. Cryst. 2011, 538, 33−38. (8) Taday, P. F.; Bradley, I. V.; Arnone, D. D.; Pepper, M. J. Pharm. Sci. 2003, 92, 831−838. (9) Frišcǐ ć, T.; Jones, W. J. Pharm. Pharmacol. 2010, 62, 1547−1559. (10) Schultheiss, N.; Newman, A. Cryst. Growth Des. 2009, 9, 2950− 2967. (11) Shen, Y.; Upadhya, P.; Linfield, E.; Davies, A. Appl. Phys. Lett. 2003, 82, 2350−2352. (12) Qiao, N.; Li, M.; Schlindwein, W.; Malek, N.; Davies, A.; Trappitt, G. Int. J. Pharm. 2011, 419, 1−11. (13) Schultheiss, N.; Henck, J.-O. Role of co-crystals in the pharmaceutical development continuum. In Pharmaceutical Salts and Co-crystals; Wouters, J., Quéré, L., Eds.; Royal Society of Chemistry: Cambridge, U.K., 2012; pp 110−125. (14) Kwok, K.; Taylor, L. J. Pharm. Biomed. Anal. 2012, 66, 126−135. (15) Franch-Lage, F.; Amigo, J.; Skibsted, E.; Maspoch, S.; Coello, J. Int. J. Pharm. 2011, 411, 27−35. (16) Zeitler, J. A.; Shen, Y. C.; Baker, C.; Taday, P. F.; Pepper, M.; Rades, T. J. Pharm. Sci. 2007, 96, 330−340. (17) Russe, I. S.; Brock, D.; Knop, K.; Kleinebudde, P.; Zeitler, J. A. Mol. Pharm. 2012, 9, 3551−3559. (18) Oh, S.; Kang, J.; Maeng, I.; Suh, S.-J.; Huh, Y.-M.; Haam, S.; Son, J.-H. Opt. Exp. 2009, 17, 3469−3475. (19) Ueno, Y.; Rungsawang, R.; Tomita, I.; Ajito, K. Anal. Chem. 2006, 78, 5424−428. (20) Nguyen, K.; Frisccaron, T.; Day, G.; Gladden, L.; Jones, W. Nat. Mater. 2007, 6, 206−209. (21) Taday, P. F. Trans. R. Soc. London A 2004, 362, 351−364. (22) Watanabe, Y.; Kawase, K.; Ikari, T.; Ito, H.; Ishikawa, Y.; Minamide, H. Opt. Commun. 2004, 234, 125−129. (23) Kawase, K.; Ogawa, Y.; Watanabe, Y. Opt. Express 2003, 11, 2549−2554. (24) Trask, A.; Motherwell, W.; Jones, W. Cryst. Growth Des. 2005, 5, 1013−1021. (25) King, M.; Korter, T. J. Phys. Chem. A 2010, 114, 7127−7138. (26) Aher, S.; Dhumal, R.; Mahadik, K.; Ketolainen, J.; Paradkar, A. Pharm. Dev. Technol. 2013, 18, 55−60. (27) Hédoux, A.; Decroix, A.; Guinet, Y.; Paccou, L.; Derollez, P.; Descamps, M. J. Phys. Chem. 2011, 115, 5746−5753. (28) Jeziorowski, H.; Moser, B. Chem. Phys. Lett. 1985, 120, 41−44. (29) Ueno, Y.; Ajito, K.; Kukutsu, N.; Tamechika, E. Anal. Sci. 2011, 27, 351−356. (30) Hunsche, S.; Koch, M.; Brener, I.; Nuss, M. C. Opt. Commun. 1998, 150, 22−26. (31) Blanchard, F.; Doi, A.; Tanaka, T.; Hirori, H.; Tanaka, H.; Kadoya, Y.; Tanaka, K. Opt. Exp. 2011, 19, 8277−8284. (32) Chen, H.-T.; Kerstinga, R.; Cho, G. C. Appl. Phys. Lett. 2003, 83, 3009−3011. (33) Salhi, M. A.; P.a, I.; Koch, M. J. Infrared Millimeter Terahertz Waves 2010, 31, 358−366. (34) Cumis, U. S.; Xu, J.-H.; Masini, L.; Degl’Innocenti, R.; Pingue, P.; Beltram, F.; Tredicucci, A.; Vitiello, M. S.; Benedetti, P. A.; Beere, H. E.; Ritchie, D. A. Opt. Exp. 2012, 20, 21924−21931.
1984
dx.doi.org/10.1021/ac302852n | Anal. Chem. 2013, 85, 1980−1984
