- Email: [email protected]
Transmission Electron Microscopy of Pharmaceutical Materials
Transmission Electron Microscopy of Pharmaceutical Materials MARK D. EDDLESTON,1 ERICA G. BITHELL,2 WILLIAM JONES1 1
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
2
Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2 3QZ, United Kingdom Received 12 February 2010; revised 16 April 2010; accepted 19 April 2010 Published online 2 June 2010 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.22220 ABSTRACT: Transmission electron microscopy (TEM) and its facility for electron diffraction has long been a key technique in materials science. Its use for characterization of pharmaceutical samples has, however, been very limited, largely due to the difficulties associated with the preparation of appropriately thin samples, as well as issues with sample damage caused by the electron beam. In this overview, we describe straightforward approaches for overcoming these issues which have enabled us to characterize a variety of pharmaceutical compounds, including theophylline, paracetamol and aspirin, and also pharmaceutical salts and cocrystals. A range of relevant information about these compounds is derived including morphology, polymorph identification, mapping of crystal habit to crystal structure and crystal defect characterization. With theophylline, we identify crystals of ‘‘impurity’’ polymorphic phases in samples that appear from powder X-ray diffraction to be monophasic, and observe that crystal growth behavior of samples prepared from nitromethane is significantly different to that of samples prepared from methanol. The existence of imperfections, such as dislocations, is also established and these are shown to be likely sites at which fracturing occurs when the crystals are stressed. The results demonstrate that various issues associated with pharmaceutical form development might usefully be addressed using TEM. ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:4072–4083, 2010
Keywords: microscopy; polymorphism; solid state; cocrystals; crystal defects; crystallography; crystals; materials science; morphology; nanoparticles
INTRODUCTION Characterization of crystal form is an important issue in pharmaceutical materials science.1,2 The existence of polymorphs, for example, is considered to be a key concern,3–5 as is the stability of a chosen form to various processing (and storage) conditions, for example, milling and tableting.6 The recent emergence of pharmaceutical cocrystals as alternatives to salts and amorphous forms is recognized,7 but raises questions during development of the importance that will need to be attached to the purity of the phases produced, for example, components of the target cocrystal present as ‘‘impurity’’ phases or the possible generation of small amounts of a different cocrystal stoichiometry beyond the usual detection limits of routine analytical methods such as powder X-ray diffraction (PXRD).
Correspondence to: William Jones (Telephone: þ44-1223336468; Fax: þ44-1223-762829; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 4072–4083 (2010) ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association
4072
Transmission electron microscopy (TEM) has long been used in materials science as a powerful analytical tool.8–13 Its application to pharmaceutical materials, however, has been very limited. Some examples include imaging crystals of the compounds dipyridamole14 and taxol,15 and defects in liquid crystals of fenoprofen,16 and an electron diffraction study on roxifiban,17 but the full potential of the technique has yet to be fully appreciated. There are two major reasons for the underdevelopment of TEM in this field. The first relates to sample preparation: because of the strong interaction of the electron beam with the sample18 it is required that the specimens be very thin (500 nm even for light, organic compounds).19 Methods have been developed, however, for organic molecular crystals and as we demonstrate these can be used to prepare appropriate pharmaceutical materials. Even when this is not possible (i.e., when the analysis is of a specific material such as milled crystals), we demonstrate that useful information can still be obtained. The second deterrent is the inherent susceptibility of an organic material to electron beam damage.19 For certain types of analysis, such as a full defect
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
TEM OF PHARMACEUTICAL MATERIALS
characterization, this will certainly limit the application of TEM. Nevertheless, we demonstrate that sufficient data can be obtained for identification of crystalline imperfections. In this overview, we illustrate various instances in which TEM has provided useful information on crystal habit, polymorph identification, defect content and the analysis of ground samples obtained using the recently developed approach of liquid assisted grinding.20,21 We use theophylline, paracetamol and ranitidine hydrochloride as examples along with a 2:1 caffeine/oxalic acid cocrystal. Given that the sample is held under high vacuum during analysis, TEM is not likely to be generally suitable for hydrated or solvated materials.
EXPERIMENTAL All chemicals were purchased from Sigma–Aldrich and used as received. Solution cooling, solvent evaporation, crystallization from the melt, crystal growth on a water surface and grinding were used to prepare samples for TEM analysis. Solution cooling samples were prepared by cooling solutions to below the saturation temperature to induce precipitation. The resulting crystals were then pipetted onto a TEM sample support grid. Solution evaporation specimens were prepared by evaporative precipitation of crystals directly onto TEM grids. Melt crystallization samples were heated until liquid and then spread thinly over a TEM grid. Crystallization on a water surface was performed by dissolving compounds in a water immiscible solvent such as p-xylene. A few drops of these solutions were pipetted onto a water surface and allowed to evaporate slowly. Grinding experiments were conducted on approximately 200 mg of sample in 10 mL stainless-steel containers with two stainless-steel balls of 7 mm diameter. The grinding was carried out in a Retsch MM200 mixer mill, operating at a frequency of 30 Hz. For liquid-assisted grinding experiments, 20 mL of nitromethane was also added into the grinding container. Less aggressive grinding conditions were used for some samples. In these cases, crystals were lightly crushed between two glass slides. Transmission electron microscopy characterization was performed at room temperature in a Philips CM30 instrument operating at 300 kV (unless stated) and data were collected on photographic films, which were scanned in order to generate digital images. A double tilt sample holder was used and the samples supported on holey-carbon films on 400 mesh copper grids. Electron diffraction patterns were indexed by comparison with known crystal structures. The positions of reflections in experimental diffraction patterns were measured, converted to d-spacings and DOI 10.1002/jps
4073
matched to calculated values for known structures, giving both the composition of the sample and the zone axis of the diffraction pattern. The experimental diffraction pattern was then compared with a simulated diffraction pattern of the given zone axis to ensure a match using CrystalMaker SingleCrystal v1.3 software from crystal structures published in the Cambridge Structural Database (CSD). In practice, some diffraction patterns, from high index zone axes, could not be unambiguously indexed. If a known crystal form had no reported crystal structure, obtained electron diffraction patterns were compared with a PXRD trace of the form by plotting reflections on a 2u scale. Scanning electron microscopy (SEM) images were obtained with a JEOL JSM-5510LV instrument. Samples were prepared on a sticky carbon sample mount placed on a brass SEM stub and sputter coated with platinum to reduce charging during analysis. Polarized light microscopy (PLM) was performed on a Leica DM1000 instrument with a polarising filter. Powder X-ray diffraction analysis was performed on a Philips X’Pert Diffractometer with Cu Ka radiation at ˚ and data collected between 3 a wavelength of 1.5406 A and 508 2u at ambient temperature.
RESULTS AND DISCUSSION Figure 1 displays electron diffraction patterns obtained for a sample of theophylline prepared by cooling a solution of theophylline in nitromethane. The patterns show loss of sample crystallinity caused by exposure to the electron beam. The direction of view, however, remained down the same zone axis of the crystal throughout the duration of the experiment, demonstrating that the electron beam induced little sample movement or tilting. Though beam damage is unavoidable during TEM analysis, it was found that the rate could be reduced sufficiently to allow specimens to be characterized. This was achieved primarily by reducing the flux of electrons through the sample during analysis. A liquid nitrogen cooled sample holder was used to further increase the stability of particularly beam sensitive compounds such as aspirin. Figure 2 (theophylline) illustrates the clear advantages that the high magnifications available when imaging with TEM, and the ability to generate diffraction data, give this technique over optical microscopy and SEM. The small crystallites are visible in the optical micrograph, but no detailed information can be obtained. In the SEM image the presence of small, overlapping, triangular crystals is clear, but no structural or compositional information is possible. The TEM image also shows the triangular morphology of the theophylline crystals. In addition, JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4074
EDDLESTON, BITHELL, AND JONES
Figure 1. (a–c) <100> zone axis diffraction pattern of form II of theophylline after approximately 2, 20, and 40 min, respectively, showing loss of crystallinity over time.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
dark lines running across the crystals are bend contours and signify regions where Bragg diffraction conditions are met. Disruption to bend contours as they cross the sample reveals the presence of defects running in the [010] crystal direction. Many of the bend contours cross, in the region marked with a circle, to form a bend contour pole. The associated zone axis electron diffraction pattern, taken from the circled region, was used to determine the crystal phase.19 The pattern can be indexed on the basis of the form II orthorhombic unit cell reported by Ebisuzaki et al.22 (CSD ref. BAPLOT01, Pna21, ˚ , b ¼ 3.8302 A ˚ , c ¼ 8.5010 A ˚ ) and correa ¼ 24.612 A sponds to a view down the <100> axis of the crystal (the indexing of selected reflections is shown, as is a simulated diffraction pattern of the <100> zone axis). Single diffraction patterns are often enough to uniquely identify the phase, but if required, a series of diffraction patterns at different sample tilts could be obtained as further confirmation. Furthermore, by moving around the specimen and collecting a series of diffraction patterns from different crystals it is possible to assess the phase purity of the sample. The identification of different polymorphs of paracetamol is shown in Figure 3. Paracetamol has been reported to have three polymorphic phases,23 of which form I is the most stable,24 though there has also been interest in form II due to its superior compaction properties.25 Samples of forms I and II were prepared from the melt using procedures described by Di Martino et al.26 Diffraction patterns from these two samples were indexed against the unit cells of both form I and form II reported by Haisa et al.27 (CSD ref. ˚ , b ¼ 9.40 A ˚ , c ¼ 7.10 A ˚, HXACAN01, P21/a, a ¼ 12.93 A b ¼ 115.98) and Nichols et al.28 (CSD ref. HXACAN08, ˚ , b ¼ 17.7773 A ˚ , c ¼ 7.212 A ˚ ), Pbca, a ¼ 17.1657 A respectively. The diffraction pattern from the sample of form I matched the <010> zone axis of this form, and was not a match for any of the zone axes of form II of paracetamol. The simulated electron diffraction pattern of the <010> zone axis is shown for comparison. Likewise, the diffraction pattern from the sample of form II corresponded to a view down the <001> zone axis and did not match any of the zone axes of form I. Again, the simulated electron diffraction pattern of this zone axis is shown for comparison. As shown in Figure 2, the area of sample used to generate an electron diffraction pattern is typically less than 10 mm2 in our experiments. Polymorph identification by TEM therefore requires just a single micro-crystal, and in favorable cases can be successful on the nanometre scale, making it a significantly more sensitive technique than PXRD. In Figure 4 the use of diffraction pattern indexing for the identification of the crystalline phase of a pharmaceutical salt (ranitidine hydrochloride) and a DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
4075
Figure 2. (a) Polarized light microscopy image of theophylline crystals prepared on a copper sample support grid. (b) Scanning electron microscopy image of triangular theophylline crystals. (c) TEM image of overlapping triangular crystals of form II of theophylline. The dark lines running across the crystals are bend contours. (d) Electron diffraction pattern from the circled region in image (c). (e) Simulated electron diffraction pattern of the <100> zone axis of form II of theophylline (CSD ref. BAPLOT01).22 DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4076
EDDLESTON, BITHELL, AND JONES
Figure 3. Selected area diffraction of two polymorphs of paracetamol. (a) <010> zone axis electron diffraction pattern of form I of paracetamol. (b) Simulated electron diffraction pattern of the <010> zone axis of form I of paracetamol (CSD ref. HXACAN01).27 (c) <001> zone axis electron diffraction pattern of form II of paracetamol. The 010 reflection should be systematically absent, but is seen in the diffraction pattern due to multiple scattering. This effect is commonly observed with electron diffraction, unlike X-ray diffraction, due to the stronger interaction of the electrons with specimens. (d) Simulated electron diffraction pattern of the <001> zone axis of form II of paracetamol (CSD ref. HXACAN08).28
cocrystal (2:1 caffeine/oxalic acid) is shown. Both specimens were prepared by careful crushing of large crystals between two glass slides. The ranitidine hydrochloride diffraction pattern was indexed on the basis of the form II unit cell reported by Mirmehrabi ˚, et al.29 (CSD ref TADZAZ03, P21/n, a ¼ 7.208 A ˚ , c ¼ 18.807 A ˚ , b ¼ 95.068) and correb ¼ 12.979 A sponds to a view down the <10-1> axis of the crystal. This diffraction pattern did not match any zone axis of form I of ranitidine hydrochloride or of forms I or II of ranitidine free base. The diffraction pattern of the 2:1 cocrystal of caffeine and oxalic acid was indexed on the basis of the unit cell reported by Trask JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
˚, et al.30 (CSD ref GANXUP, P21/c, a ¼ 4.41430 A ˚ , c ¼ 15.9119 A ˚ , b ¼ 96.48508) and correb ¼ 14.7701 A sponds to a view down the <110> axis of the crystal. This diffraction pattern did not match any zone axis of forms I and II of caffeine or of the alpha or beta polymorphs of oxalic acid. The detection of phases additional to those expected is also an important feature of TEM. The diffraction pattern in Figure 5, obtained for a sample of theophylline grown by cooling an ethyl acetate solution, illustrates the detection of a previously unidentified phase of theophylline. Despite repeated attempts to index the pattern on the basis of form II of DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
4077
Figure 5. Diffraction pattern, from a sample thought to be solely form II of theophylline, which could not be indexed to a zone axis of this theophylline polymorph. Several samples of theophylline have been analyzed to date, but this diffraction pattern has been observed just once.
compound.33,34 The diffraction pattern from the unknown phase could then be indexed against these simulated crystal structures to look for matches. Figures 6 and 7 show an example of linking the obtained diffraction and imaging information from TEM. The PXRD trace of a sample of theophylline crystallized from methanol, which appears to be monophasic form II, is shown alongside a simulated trace of form II of theophylline. During TEM analysis of this sample, many crystals indexable as form II
Figure 4. Selected area diffraction of a salt and a cocrystal confirming the identity of the phases present. (a) <10-1> zone axis electron diffraction pattern of form II of ranitidine hydrochloride. (b) <110> zone axis electron diffraction pattern of a 2:1 caffeine/oxalic acid cocrystal.
theophylline, no successful indexing was possible. Likewise, comparison of the diffraction pattern reflections with PXRD peak positions of other reported polymorphic forms of theophylline, for which there is no reported crystal structure (form I, and a form recently reported by Roy et al.),31,32 did not give a match. The two possible explanations for this result are the presence of an unknown impurity phase or the existence of a previously unknown polymorph. Obtaining such patterns may suggest that further polymorph screening experiments are necessary. For example, crystal structure prediction could be used to search for possible low energy crystal forms of the DOI 10.1002/jps
Figure 6. PXRD trace of a sample of theophylline. (Top) Simulated pattern of form II of theophylline (CSD ref. BAPLOT01).22 (Bottom) Theophylline crystallized from methanol. The difference in peak intensities between the experimental and simulated patterns is believed to result from preferred orientation in the experimental sample. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4078
EDDLESTON, BITHELL, AND JONES
Figure 7. TEM characterization of the same sample of theophylline as analyzed by PXRD in Figure 6. (a) TEM image of lath-shaped crystals of theophylline form II. (b) TEM image of a crystal which appeared to be significantly darker, and therefore thicker, than other crystals of a similar size. The inset is an electron diffraction pattern from this crystal, in the correct relative orientation. The diffraction pattern was a match for the polymorph of theophylline reported by Roy et al.32
were observed (Fig. 7). However, the imaging mode also revealed a small number of crystals that were morphologically different to the bulk sample. The associated diffraction pattern recorded from one of these crystals was a match for the theophylline polymorph reported by Roy et al.32 This result demonstrates how TEM can identify phases that JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
Figure 8. (a) TEM image of a triangular crystal of theophylline prepared by cooling a nitromethane solution. The corresponding <100> zone axis electron diffraction pattern of theophylline form II is shown as an inset. The direction of most rapid growth for this triangular crystal was the [001] direction. (b) TEM image showing a lathshaped crystal of theophylline form II that was crystallized from methanol. The asterisk marks a region of the crystal where bend contours are distorted as they pass across defects running in the [010] direction. The corresponding <110> zone axis diffraction pattern is included as an inset. The ring of diffraction spots is due to reflections from small crystallites which grew on the amorphous carbon film support during sample preparation. The direction of most rapid growth for this crystal was the [010] direction.
DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
are undetectable by XRPD. It also suggests that TEM could be an important tool for investigating samples which contain a mixture of phases. A further advantage of the combination of images and diffraction patterns is the ability to map crystal morphology to crystal structure. For example, images
4079
and corresponding diffraction patterns of crystals of theophylline prepared from nitromethane and methanol (Fig. 8) demonstrate that the direction of most rapid crystal growth of crystals of theophylline prepared from nitromethane is different to that of crystals prepared from methanol. The triangular
Figure 9. (a) TEM image showing a small region of a triangular crystal of theophylline with dislocations running in the [010] direction. The pair of bend contours marked with asterisks are unaffected as they cross the dislocations, indicating that the corresponding crystal planes are not bent by the defect. The corresponding diffraction pattern, of the <100> zone axis, is included as an inset. (b) TEM bright field image of paracetamol form I crystallized from a melt. The discontinuous nature of the bend contours as they run across the sample may indicate that micro-twinning has occurred in the crystals during growth. (c and d) Crystals of theophylline that originally crystallized with a triangular habit, have fractured along defects that run in the [010] direction to give smaller, trapezoid-shaped crystals. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4080
EDDLESTON, BITHELL, AND JONES
Figure 10. Preparation methods for TEM samples. (a) TEM bright field image of a crystal of aspirin prepared by rapid evaporation of an acetonitrile solution. The sample was cooled to 1788C during analysis to reduce beam damage. (b) Needle-shaped crystals of caffeine form I prepared by evaporation of a chloroform solution. Several of the crystals show signs of beam damage. The crystal marked with an asterisk is tubular.44 (c) TEM bright field image of a large thin-plate crystal of p-terphenyl, prepared by growth on a water surface, showing an extended network of defects. (d) A sample of a 1:1 theophylline/L-malic acid cocrystal prepared by liquid assisted grinding. The crystallites have habits ranging from thin plate to needle (250 kV). (e) Polycrystalline diffraction pattern from the same sample. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
crystals from nitromethane grew most rapidly in the [001] direction, whereas those from methanol grew fastest in the [010] direction. TEM is particularly useful for morphological examination of nanosized crystals. The existence of crystal defects has been established for theophylline and paracetamol. Series of parallel imperfections running in the [010] crystallographic direction, as shown in Figure 9, have been observed in many crystals of form II of theophylline. The nature of the interaction of bend contours with these defects suggests that they are dislocations. Samples of paracetamol form I grown from the melt show possible evidence of micro-twinning. The influence that crystal defects have on sample behavior was also apparent for theophylline. It was observed that triangular crystals of theophylline spontaneously changed shape over time. It appears that the crystals fractured in the [010] crystallographic direction, along the observed dislocations, giving smaller, trapezoid-shaped crystals. The fracturing could also be induced mechanically. It is known that defects are sites in crystals at which transformations, such as polymorphic transitions and hydrate formation, are initiated,35–41 and that processes routinely used in the manufacture of pharmaceutical products such as milling and tableting introduce defects.42,43 The ability to observe and characterize crystal defects, as afforded by TEM, could therefore significantly improve the understanding and control of the solid-state behavior of pharmaceutical compounds. Figure 10 compares the nature of the results which can be obtained from specimens prepared by different methods. The amount of information that can be obtained for samples prepared by solution cooling and solution evaporation is determined by crystal thickness. With large or block-shaped crystals the electron beam is scattered so completely that detailed imaging is not achievable. However, phase identification is still possible as electron diffraction patterns can be acquired from edges or corners of crystals where they are thinner. Flake or plate crystal habits, as exhibited by aspirin crystallized by evaporation from acetonitrile, can have large regions with suitable thickness for TEM analysis. Lath or needle-shaped crystals are often thin enough in at least one dimension for imaging as shown for needle-shaped crystals of form I of caffeine. It is evident that the crystal of caffeine marked with an asterisk in the Figure is tubular.44 However, the area of view for each crystal is small, making defects hard to detect. Suitable crystal morphologies for TEM can be targeted through choice of crystallization solvent and/or addition of small amounts of impurities.45–47. Crystallization of the organic compound p-terphenyl on a water surface, following a method reported by DOI 10.1002/jps
4081
Jones et al.,48 generated thin foils with thicknesses of <500 nm and areas of up to 1 cm2, making them ideal for TEM analysis. Thin foils could not be obtained for the pharmaceutical compounds used in this study, but it is believed that this method will prove to be an optimal way of preparing TEM specimens of other pharmaceutical materials. Melt crystallization offers the advantage that material can be thinly spread over a TEM sample grid while liquid, resulting in thin specimens, as shown for paracetamol (Fig. 9). Ball mill grinding was found to reliably generate crystals of sub-micron size, suitably thin to transmit electrons, for which crystal habit information could be obtained. However, phase identification for these samples proved difficult (but not impossible) as diffraction patterns usually contained contributions from several crystallites. The pictured crystallites of a 1:1 theophylline/L-malic acid cocrystal,49 prepared by liquid assisted grinding, have particle sizes ranging from 50 to 1000 nm and morphologies ranging from needle to thin plate. This result demonstrates the suitability of TEM for the characterization of pharmaceutical nanomaterials. TEM is already routinely used for imaging liposomal, micellular and polymeric nanoparticles containing encapsulated pharmaceutical compounds.14,50–52
CONCLUDING REMARKS Sample preparation and beam damage issues, widely considered to prohibit TEM characterization of pharmaceutical samples, can be sufficiently overcome with simple approaches to enable information about a range of pharmaceutical materials to be obtained. Furthermore, identification of defects in samples of theophylline, as described here, could not be achieved with any other analytical technique currently used in pharmaceutical analysis. In the wider materials context, TEM has moved somewhat away from bright/dark field diffractioncontrast based techniques towards approaches which yield three dimensional data and/or are directly interpretable (electron tomography, aberration-corrected high resolution, scanned probe methods).10,11 At the present time, the majority of pharmaceutical samples are not amenable to these techniques because of the requirement for high signal-noise ratios, and thus high total electron doses. We note here that the long-established diffraction contrast techniques13 still have an important role when the limitations on the analysis come not from the instrumentation but from the sample itself. We have demonstrated that TEM analysis could be advantageous in the pharmaceutically important areas of solid-phase identification and patent infrinJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4082
EDDLESTON, BITHELL, AND JONES
gement, and has the potential to provide a greater understanding of defects, and related reactivity, in pharmaceutical crystals.
ACKNOWLEDGMENTS M.D.E. and W.J. thank the EPSRC for funding. E.G.B. is grateful for support in the form of a Daphne Jackson Fellowship funded by Lucy Cavendish College, Cambridge, the Thriplow Charitable Trust and the Isaac Newton Trust and is currently supported by the ERC. We thank Andrew M.C. Cassidy for supplying a sample of the 2:1 caffeine/oxalic acid cocrystal.
REFERENCES 1. Brittain HG, editor. 1995. Physical characterization of pharmaceutical solids. New York: Marcel Dekker. 2. Vippagunta SR, Brittain HG, Grant DJ. 2001. Crystalline solids. Adv Drug Deliv Rev 48:3–26. 3. Haleblian J, McCrone W. 1969. Pharmaceutical applications of polymorphism. J Pharm Sci 58:911–929. 4. Singhal D, Curatolo W. 2004. Drug polymorphism and dosage form design: A practical perspective. Adv Drug Deliv Rev 56: 335–347. 5. Bauer J, Spanton S, Henry R, Quick J, Dziki W, Porter W, Morris J. 2001. Ritonavir: An extraordinary example of conformational polymorphism. Pharm Res 18:859–866. 6. Brittain HG, Fiese EF. 1999. Effects of pharmaceutical processing on drug polymorphs and solvates. In: Brittain HG, editor. Polymorphism in pharmaceutical sciences. New York: Marcel Dekker. pp. 331–361. 7. Fleischman SG, Kuduva SS, McMahon JA, Moulton B, Bailey Walsh RD, Rodriguez-Hornedo N, Zaworotko MJ. 2003. Crystal engineering of the composition of pharmaceutical phases: Multiple-component crystalline solids involving carbamazepine. Cryst Growth Des 3:909–919. 8. Hirsch PB, Howie A, Nicholson RB, Pashley DW, Whelan MJ. 1965. Electron microscopy of thin crystals. London: Butterworths. 9. McLaren AC. 1991. Transmission electron microscopy of minerals and rocks. 2nd edition. Cambridge: Cambridge University Press. 10. Thomas JM, Midgley PA. 2004. High-resolution transmission electron microscopy: The ultimate nanoanalytical technique. Chem Commun (Cambridge United Kingdom) 11:1253–1267. 11. Midgley PA, Dunin-Borkowski RE. 2009. Electron tomography and holography in materials science. Nat Mater 8:271–280. 12. Armigliato A. 1979. Analysis of crystal defects by transmission electron microscopy. Mater Chem 4:453–471. 13. Hirsch P, Cockayne D, Spence J, Whelan M. 2006. 50 years of TEM of dislocations. Past, present and future. Philos Mag 86: 4519–4528. 14. Tang Y, Liu Shiyong Y, Armes Steven P, Billingham Norman C. 2003. Solubilization and controlled release of a hydrophobic drug using novel micelle-forming ABC triblock copolymers. Biomacromolecules 4:1636–1645. 15. Liu Y, Yang Z, Du J, Yao X, Zheng X, Lei R, Liu J, Hu H, Li H. 2008. Interaction of Taxol with intravenous immunoglobulin: An inhibition of Taxol from crystallizing in aqueous solution. Int Immunopharmacol 8:390–400. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
16. Rades T, Mueller-Goymann CC. 1997. Electron and light microscopical investigation of defect structures in mesophases of pharmaceutical substances. Colloid Polym Sci 275:1169–1178. 17. Li ZG, Harlow RL, Foris CM, Li H, Ma P, Vickery RD, Maurin MB, Toby BH. 2002. New applications of electron diffraction in the pharmaceutical industry: Polymorph determination by using a combination of electron diffraction and synchrotron X-ray powder diffraction techniques. Microsc and Microanal 8:134–138. 18. Reimer L, Kohl H. 2008. Transmission electron microscopy: Physics of image formation. 5th edition. New York: Springer. 19. Jones W, Thomas JM. 1979. Applications of electron microscopy to organic solid-state chemistry. Progr Solid State Chem 12:101–124. 20. Trask AV, Jones W. 2005. Crystal engineering of organic cocrystals by the solid-state grinding approach. Top Curr Chem 254:41–70. 21. Friscic T, Jones W. 2009. Recent advances in understanding the mechanism of cocrystal formation via grinding. Cryst Growth Des 9:1621–1637. 22. Ebisuzaki Y, Boyle PD, Smith JA. 1997. Methylxanthines. I. Anhydrous theophylline. Acta Crystallogr C C53:777–779. 23. Perrin M-A, Neumann MA, Elmaleh H, Zaske L. 2009. Crystal structure determination of the elusive paracetamol Form III. Chem Commun: 3181–3183. 24. Perlovich GL, Volkova TV, Bauer-Brandl A. 2007. Polymorphism of paracetamol. Relative stability of the monoclinic and orthorhombic phase revisited by sublimation and solution calorimetry. J Therm Anal Calorim 89:767–774. 25. Di Martino P, Guyot-Hermann AM, Conflant P, Drache M, Guyot JC. 1996. A new pure paracetamol for direct compression: The orthorhombic form. Int J Pharm 128:1–8. 26. Di Martino P, Conflant P, Drache M, Huvenne JP, GuyotHermann AM. 1997. Preparation and physical characterization of forms II and III of paracetamol. J Therm Anal 48:447–458. 27. Haisa M, Kashino S, Kawai R, Maeda H. 1976. The monoclinic form of p-hydroxyacetanilide. Acta Crystallogr B B32:1283– 1285. 28. Nichols G, Frampton CS. 1998. Physicochemical characterization of the orthorhombic polymorph of paracetamol crystallized from solution. J Pharm Sci 87:684–693. 29. Mirmehrabi M, Rohani S, Murthy KSK, Radatus B. 2004. Characterization of tautomeric forms of ranitidine hydrochloride: Thermal analysis, solid-state NMR, X-ray. J Cryst Growth 260:517–526. 30. Trask AV, Motherwell WDS, Jones W. 2005. Pharmaceutical cocrystallization: Engineering a remedy for caffeine hydration. Cryst Growth Des 5:1013–1021. 31. Suzuki E, Shimomura K, Sekiguchi K. 1989. Thermochemical study of theophylline and its hydrate. Chem Pharm Bull 37: 493–497. 32. Roy C, Vega-Gonzalez A, Subra-Paternault P. 2007. Theophylline formulation by supercritical antisolvents. Int J Pharm 343:79–89. 33. Cruz Cabeza AJ, Day GM, Motherwell WDS, Jones W. 2006. Prediction and observation of isostructurality induced by solvent incorporation in multicomponent crystals. J Am Chem Soc 128:14466–14467. 34. Price SL. 2004. The computational prediction of pharmaceutical crystal structures and polymorphism. Adv Drug Deliv Rev 56:301–319. 35. Thomas JM. 1974. Topography and topology in solid state chemistry. Philosophical Transactions of the Royal Society of London, Series A: Mathematical. Phys Eng Sci 277:251–287. 36. Bhadeshia HKDH. 2001. Martensitic transformation. In: Buschow K, Cahn RW, Flemings MC, Lischner B, Kramer EJ, Mahajan S, editors. Encyclopedia of materials science: Science and technology. Pergamon Press: Elsevier Science. pp. 5203–5206. DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
37. Jones W, Thomas JM, Williams JO. 1975. Electron and optical microscopic studies of a stress-induced phase transition in 1,8dichloro-10-methylanthracene. Philos Mag 32:1–11. 38. Mnyukh YV. 1976. Polymorphic transitions in crystals: Nucleation. J Cryst Growth 32:371–377. 39. Thomas JM, Williams JO. 1971. Dislocations and the reactivity of organic solids. Progr Solid State Chem 6:119–154. 40. Thomas JM, Renshaw GD. 1969. Role of dislocations in the dehydration of nickel sulfate hexahydrate. II. Topographical studies by use of optical microscopy. J Chem Soc A: 2753–2755. 41. Prasad KVR, Ristic RI, Sheen DB, Sherwood JN. 2002. Dissolution kinetics of paracetamol single crystals. Int J Pharm 238:29–41. 42. Koivisto M, Heinaenen P, Tanninen VP, Lehto V-P. 2006. Depth profiling of compression-induced disorders and polymorphic transition on tablet surfaces with grazing incidence X-ray diffraction. Pharm Res 23:813–820. 43. Byard SJ, Jackson SL, Smail A, Bauer M, Apperley DC. 2005. Studies on the crystallinity of a pharmaceutical development drug substance. J Pharm Sci 94:1321–1335. 44. Eddleston MD, Jones W. 2010. Formation of tubular crystals of pharmaceutical compounds. Cryst Growth Des 10:365–370. 45. Lee AY, Myerson AS. 2006. Particle engineering: Fundamentals of particle formation and crystal growth. MRS Bull 31:881– 886.
DOI 10.1002/jps
4083
46. Lee T, Kuo CS, Chen YH. 2006. Solubility, polymorphism, crystallinity, and crystal habit of acetaminophen and ibuprofen: By initial solvent screening. Pharm Technol 30:72–92. 47. Berkovitch-Yellin Z, Addadi L, Idelson M, Lahav M, Leiserowitz L. 1982. Controlled modification of crystal habit via ‘‘tailormade’’ impurities: Application to benzamide. Angew Chem Int Ed Engl 21:1336–1345. 48. Jones W, Thomas JM, Williams JO. 1975. Electron microscopic studies of extended defects in organic molecular crystals. I. pTerphenyl. J Chem Soc, Faraday Trans 2 71:138–145. 49. Friscic T, Fabian L, Burley JC, Reid DG, Duer MJ, Jones W. 2008. Exploring the relationship between cocrystal stability and symmetry: Is Wallach’s rule applicable to multi-component solids? Chem Commun 14:1644–1646. 50. Almgren M, Edwards K, Karlsson G. 2000. Cryo transmission electron microscopy of liposomes and related structures. Colloid Surface A 174:3–21. 51. Kim I-S, Kim S-H. 2003. Development of polymeric nanoparticulate drug delivery systems: Evaluation of nanoparticles based on biotinylated poly(ethylene glycol) with sugar moiety. Int J Pharm 257:195–203. 52. Prabaharan M, Grailer JJ, Pilla S, Steeber DA, Gong S. 2009. Gold nanoparticles with a monolayer of doxorubicin-conjugated amphiphilic block copolymer for tumor-targeted drug delivery. Biomaterials 30:6065–6075.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
2
Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2 3QZ, United Kingdom Received 12 February 2010; revised 16 April 2010; accepted 19 April 2010 Published online 2 June 2010 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.22220 ABSTRACT: Transmission electron microscopy (TEM) and its facility for electron diffraction has long been a key technique in materials science. Its use for characterization of pharmaceutical samples has, however, been very limited, largely due to the difficulties associated with the preparation of appropriately thin samples, as well as issues with sample damage caused by the electron beam. In this overview, we describe straightforward approaches for overcoming these issues which have enabled us to characterize a variety of pharmaceutical compounds, including theophylline, paracetamol and aspirin, and also pharmaceutical salts and cocrystals. A range of relevant information about these compounds is derived including morphology, polymorph identification, mapping of crystal habit to crystal structure and crystal defect characterization. With theophylline, we identify crystals of ‘‘impurity’’ polymorphic phases in samples that appear from powder X-ray diffraction to be monophasic, and observe that crystal growth behavior of samples prepared from nitromethane is significantly different to that of samples prepared from methanol. The existence of imperfections, such as dislocations, is also established and these are shown to be likely sites at which fracturing occurs when the crystals are stressed. The results demonstrate that various issues associated with pharmaceutical form development might usefully be addressed using TEM. ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:4072–4083, 2010
Keywords: microscopy; polymorphism; solid state; cocrystals; crystal defects; crystallography; crystals; materials science; morphology; nanoparticles
INTRODUCTION Characterization of crystal form is an important issue in pharmaceutical materials science.1,2 The existence of polymorphs, for example, is considered to be a key concern,3–5 as is the stability of a chosen form to various processing (and storage) conditions, for example, milling and tableting.6 The recent emergence of pharmaceutical cocrystals as alternatives to salts and amorphous forms is recognized,7 but raises questions during development of the importance that will need to be attached to the purity of the phases produced, for example, components of the target cocrystal present as ‘‘impurity’’ phases or the possible generation of small amounts of a different cocrystal stoichiometry beyond the usual detection limits of routine analytical methods such as powder X-ray diffraction (PXRD).
Correspondence to: William Jones (Telephone: þ44-1223336468; Fax: þ44-1223-762829; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 4072–4083 (2010) ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association
4072
Transmission electron microscopy (TEM) has long been used in materials science as a powerful analytical tool.8–13 Its application to pharmaceutical materials, however, has been very limited. Some examples include imaging crystals of the compounds dipyridamole14 and taxol,15 and defects in liquid crystals of fenoprofen,16 and an electron diffraction study on roxifiban,17 but the full potential of the technique has yet to be fully appreciated. There are two major reasons for the underdevelopment of TEM in this field. The first relates to sample preparation: because of the strong interaction of the electron beam with the sample18 it is required that the specimens be very thin (500 nm even for light, organic compounds).19 Methods have been developed, however, for organic molecular crystals and as we demonstrate these can be used to prepare appropriate pharmaceutical materials. Even when this is not possible (i.e., when the analysis is of a specific material such as milled crystals), we demonstrate that useful information can still be obtained. The second deterrent is the inherent susceptibility of an organic material to electron beam damage.19 For certain types of analysis, such as a full defect
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
TEM OF PHARMACEUTICAL MATERIALS
characterization, this will certainly limit the application of TEM. Nevertheless, we demonstrate that sufficient data can be obtained for identification of crystalline imperfections. In this overview, we illustrate various instances in which TEM has provided useful information on crystal habit, polymorph identification, defect content and the analysis of ground samples obtained using the recently developed approach of liquid assisted grinding.20,21 We use theophylline, paracetamol and ranitidine hydrochloride as examples along with a 2:1 caffeine/oxalic acid cocrystal. Given that the sample is held under high vacuum during analysis, TEM is not likely to be generally suitable for hydrated or solvated materials.
EXPERIMENTAL All chemicals were purchased from Sigma–Aldrich and used as received. Solution cooling, solvent evaporation, crystallization from the melt, crystal growth on a water surface and grinding were used to prepare samples for TEM analysis. Solution cooling samples were prepared by cooling solutions to below the saturation temperature to induce precipitation. The resulting crystals were then pipetted onto a TEM sample support grid. Solution evaporation specimens were prepared by evaporative precipitation of crystals directly onto TEM grids. Melt crystallization samples were heated until liquid and then spread thinly over a TEM grid. Crystallization on a water surface was performed by dissolving compounds in a water immiscible solvent such as p-xylene. A few drops of these solutions were pipetted onto a water surface and allowed to evaporate slowly. Grinding experiments were conducted on approximately 200 mg of sample in 10 mL stainless-steel containers with two stainless-steel balls of 7 mm diameter. The grinding was carried out in a Retsch MM200 mixer mill, operating at a frequency of 30 Hz. For liquid-assisted grinding experiments, 20 mL of nitromethane was also added into the grinding container. Less aggressive grinding conditions were used for some samples. In these cases, crystals were lightly crushed between two glass slides. Transmission electron microscopy characterization was performed at room temperature in a Philips CM30 instrument operating at 300 kV (unless stated) and data were collected on photographic films, which were scanned in order to generate digital images. A double tilt sample holder was used and the samples supported on holey-carbon films on 400 mesh copper grids. Electron diffraction patterns were indexed by comparison with known crystal structures. The positions of reflections in experimental diffraction patterns were measured, converted to d-spacings and DOI 10.1002/jps
4073
matched to calculated values for known structures, giving both the composition of the sample and the zone axis of the diffraction pattern. The experimental diffraction pattern was then compared with a simulated diffraction pattern of the given zone axis to ensure a match using CrystalMaker SingleCrystal v1.3 software from crystal structures published in the Cambridge Structural Database (CSD). In practice, some diffraction patterns, from high index zone axes, could not be unambiguously indexed. If a known crystal form had no reported crystal structure, obtained electron diffraction patterns were compared with a PXRD trace of the form by plotting reflections on a 2u scale. Scanning electron microscopy (SEM) images were obtained with a JEOL JSM-5510LV instrument. Samples were prepared on a sticky carbon sample mount placed on a brass SEM stub and sputter coated with platinum to reduce charging during analysis. Polarized light microscopy (PLM) was performed on a Leica DM1000 instrument with a polarising filter. Powder X-ray diffraction analysis was performed on a Philips X’Pert Diffractometer with Cu Ka radiation at ˚ and data collected between 3 a wavelength of 1.5406 A and 508 2u at ambient temperature.
RESULTS AND DISCUSSION Figure 1 displays electron diffraction patterns obtained for a sample of theophylline prepared by cooling a solution of theophylline in nitromethane. The patterns show loss of sample crystallinity caused by exposure to the electron beam. The direction of view, however, remained down the same zone axis of the crystal throughout the duration of the experiment, demonstrating that the electron beam induced little sample movement or tilting. Though beam damage is unavoidable during TEM analysis, it was found that the rate could be reduced sufficiently to allow specimens to be characterized. This was achieved primarily by reducing the flux of electrons through the sample during analysis. A liquid nitrogen cooled sample holder was used to further increase the stability of particularly beam sensitive compounds such as aspirin. Figure 2 (theophylline) illustrates the clear advantages that the high magnifications available when imaging with TEM, and the ability to generate diffraction data, give this technique over optical microscopy and SEM. The small crystallites are visible in the optical micrograph, but no detailed information can be obtained. In the SEM image the presence of small, overlapping, triangular crystals is clear, but no structural or compositional information is possible. The TEM image also shows the triangular morphology of the theophylline crystals. In addition, JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4074
EDDLESTON, BITHELL, AND JONES
Figure 1. (a–c) <100> zone axis diffraction pattern of form II of theophylline after approximately 2, 20, and 40 min, respectively, showing loss of crystallinity over time.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
dark lines running across the crystals are bend contours and signify regions where Bragg diffraction conditions are met. Disruption to bend contours as they cross the sample reveals the presence of defects running in the [010] crystal direction. Many of the bend contours cross, in the region marked with a circle, to form a bend contour pole. The associated zone axis electron diffraction pattern, taken from the circled region, was used to determine the crystal phase.19 The pattern can be indexed on the basis of the form II orthorhombic unit cell reported by Ebisuzaki et al.22 (CSD ref. BAPLOT01, Pna21, ˚ , b ¼ 3.8302 A ˚ , c ¼ 8.5010 A ˚ ) and correa ¼ 24.612 A sponds to a view down the <100> axis of the crystal (the indexing of selected reflections is shown, as is a simulated diffraction pattern of the <100> zone axis). Single diffraction patterns are often enough to uniquely identify the phase, but if required, a series of diffraction patterns at different sample tilts could be obtained as further confirmation. Furthermore, by moving around the specimen and collecting a series of diffraction patterns from different crystals it is possible to assess the phase purity of the sample. The identification of different polymorphs of paracetamol is shown in Figure 3. Paracetamol has been reported to have three polymorphic phases,23 of which form I is the most stable,24 though there has also been interest in form II due to its superior compaction properties.25 Samples of forms I and II were prepared from the melt using procedures described by Di Martino et al.26 Diffraction patterns from these two samples were indexed against the unit cells of both form I and form II reported by Haisa et al.27 (CSD ref. ˚ , b ¼ 9.40 A ˚ , c ¼ 7.10 A ˚, HXACAN01, P21/a, a ¼ 12.93 A b ¼ 115.98) and Nichols et al.28 (CSD ref. HXACAN08, ˚ , b ¼ 17.7773 A ˚ , c ¼ 7.212 A ˚ ), Pbca, a ¼ 17.1657 A respectively. The diffraction pattern from the sample of form I matched the <010> zone axis of this form, and was not a match for any of the zone axes of form II of paracetamol. The simulated electron diffraction pattern of the <010> zone axis is shown for comparison. Likewise, the diffraction pattern from the sample of form II corresponded to a view down the <001> zone axis and did not match any of the zone axes of form I. Again, the simulated electron diffraction pattern of this zone axis is shown for comparison. As shown in Figure 2, the area of sample used to generate an electron diffraction pattern is typically less than 10 mm2 in our experiments. Polymorph identification by TEM therefore requires just a single micro-crystal, and in favorable cases can be successful on the nanometre scale, making it a significantly more sensitive technique than PXRD. In Figure 4 the use of diffraction pattern indexing for the identification of the crystalline phase of a pharmaceutical salt (ranitidine hydrochloride) and a DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
4075
Figure 2. (a) Polarized light microscopy image of theophylline crystals prepared on a copper sample support grid. (b) Scanning electron microscopy image of triangular theophylline crystals. (c) TEM image of overlapping triangular crystals of form II of theophylline. The dark lines running across the crystals are bend contours. (d) Electron diffraction pattern from the circled region in image (c). (e) Simulated electron diffraction pattern of the <100> zone axis of form II of theophylline (CSD ref. BAPLOT01).22 DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4076
EDDLESTON, BITHELL, AND JONES
Figure 3. Selected area diffraction of two polymorphs of paracetamol. (a) <010> zone axis electron diffraction pattern of form I of paracetamol. (b) Simulated electron diffraction pattern of the <010> zone axis of form I of paracetamol (CSD ref. HXACAN01).27 (c) <001> zone axis electron diffraction pattern of form II of paracetamol. The 010 reflection should be systematically absent, but is seen in the diffraction pattern due to multiple scattering. This effect is commonly observed with electron diffraction, unlike X-ray diffraction, due to the stronger interaction of the electrons with specimens. (d) Simulated electron diffraction pattern of the <001> zone axis of form II of paracetamol (CSD ref. HXACAN08).28
cocrystal (2:1 caffeine/oxalic acid) is shown. Both specimens were prepared by careful crushing of large crystals between two glass slides. The ranitidine hydrochloride diffraction pattern was indexed on the basis of the form II unit cell reported by Mirmehrabi ˚, et al.29 (CSD ref TADZAZ03, P21/n, a ¼ 7.208 A ˚ , c ¼ 18.807 A ˚ , b ¼ 95.068) and correb ¼ 12.979 A sponds to a view down the <10-1> axis of the crystal. This diffraction pattern did not match any zone axis of form I of ranitidine hydrochloride or of forms I or II of ranitidine free base. The diffraction pattern of the 2:1 cocrystal of caffeine and oxalic acid was indexed on the basis of the unit cell reported by Trask JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
˚, et al.30 (CSD ref GANXUP, P21/c, a ¼ 4.41430 A ˚ , c ¼ 15.9119 A ˚ , b ¼ 96.48508) and correb ¼ 14.7701 A sponds to a view down the <110> axis of the crystal. This diffraction pattern did not match any zone axis of forms I and II of caffeine or of the alpha or beta polymorphs of oxalic acid. The detection of phases additional to those expected is also an important feature of TEM. The diffraction pattern in Figure 5, obtained for a sample of theophylline grown by cooling an ethyl acetate solution, illustrates the detection of a previously unidentified phase of theophylline. Despite repeated attempts to index the pattern on the basis of form II of DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
4077
Figure 5. Diffraction pattern, from a sample thought to be solely form II of theophylline, which could not be indexed to a zone axis of this theophylline polymorph. Several samples of theophylline have been analyzed to date, but this diffraction pattern has been observed just once.
compound.33,34 The diffraction pattern from the unknown phase could then be indexed against these simulated crystal structures to look for matches. Figures 6 and 7 show an example of linking the obtained diffraction and imaging information from TEM. The PXRD trace of a sample of theophylline crystallized from methanol, which appears to be monophasic form II, is shown alongside a simulated trace of form II of theophylline. During TEM analysis of this sample, many crystals indexable as form II
Figure 4. Selected area diffraction of a salt and a cocrystal confirming the identity of the phases present. (a) <10-1> zone axis electron diffraction pattern of form II of ranitidine hydrochloride. (b) <110> zone axis electron diffraction pattern of a 2:1 caffeine/oxalic acid cocrystal.
theophylline, no successful indexing was possible. Likewise, comparison of the diffraction pattern reflections with PXRD peak positions of other reported polymorphic forms of theophylline, for which there is no reported crystal structure (form I, and a form recently reported by Roy et al.),31,32 did not give a match. The two possible explanations for this result are the presence of an unknown impurity phase or the existence of a previously unknown polymorph. Obtaining such patterns may suggest that further polymorph screening experiments are necessary. For example, crystal structure prediction could be used to search for possible low energy crystal forms of the DOI 10.1002/jps
Figure 6. PXRD trace of a sample of theophylline. (Top) Simulated pattern of form II of theophylline (CSD ref. BAPLOT01).22 (Bottom) Theophylline crystallized from methanol. The difference in peak intensities between the experimental and simulated patterns is believed to result from preferred orientation in the experimental sample. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4078
EDDLESTON, BITHELL, AND JONES
Figure 7. TEM characterization of the same sample of theophylline as analyzed by PXRD in Figure 6. (a) TEM image of lath-shaped crystals of theophylline form II. (b) TEM image of a crystal which appeared to be significantly darker, and therefore thicker, than other crystals of a similar size. The inset is an electron diffraction pattern from this crystal, in the correct relative orientation. The diffraction pattern was a match for the polymorph of theophylline reported by Roy et al.32
were observed (Fig. 7). However, the imaging mode also revealed a small number of crystals that were morphologically different to the bulk sample. The associated diffraction pattern recorded from one of these crystals was a match for the theophylline polymorph reported by Roy et al.32 This result demonstrates how TEM can identify phases that JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
Figure 8. (a) TEM image of a triangular crystal of theophylline prepared by cooling a nitromethane solution. The corresponding <100> zone axis electron diffraction pattern of theophylline form II is shown as an inset. The direction of most rapid growth for this triangular crystal was the [001] direction. (b) TEM image showing a lathshaped crystal of theophylline form II that was crystallized from methanol. The asterisk marks a region of the crystal where bend contours are distorted as they pass across defects running in the [010] direction. The corresponding <110> zone axis diffraction pattern is included as an inset. The ring of diffraction spots is due to reflections from small crystallites which grew on the amorphous carbon film support during sample preparation. The direction of most rapid growth for this crystal was the [010] direction.
DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
are undetectable by XRPD. It also suggests that TEM could be an important tool for investigating samples which contain a mixture of phases. A further advantage of the combination of images and diffraction patterns is the ability to map crystal morphology to crystal structure. For example, images
4079
and corresponding diffraction patterns of crystals of theophylline prepared from nitromethane and methanol (Fig. 8) demonstrate that the direction of most rapid crystal growth of crystals of theophylline prepared from nitromethane is different to that of crystals prepared from methanol. The triangular
Figure 9. (a) TEM image showing a small region of a triangular crystal of theophylline with dislocations running in the [010] direction. The pair of bend contours marked with asterisks are unaffected as they cross the dislocations, indicating that the corresponding crystal planes are not bent by the defect. The corresponding diffraction pattern, of the <100> zone axis, is included as an inset. (b) TEM bright field image of paracetamol form I crystallized from a melt. The discontinuous nature of the bend contours as they run across the sample may indicate that micro-twinning has occurred in the crystals during growth. (c and d) Crystals of theophylline that originally crystallized with a triangular habit, have fractured along defects that run in the [010] direction to give smaller, trapezoid-shaped crystals. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4080
EDDLESTON, BITHELL, AND JONES
Figure 10. Preparation methods for TEM samples. (a) TEM bright field image of a crystal of aspirin prepared by rapid evaporation of an acetonitrile solution. The sample was cooled to 1788C during analysis to reduce beam damage. (b) Needle-shaped crystals of caffeine form I prepared by evaporation of a chloroform solution. Several of the crystals show signs of beam damage. The crystal marked with an asterisk is tubular.44 (c) TEM bright field image of a large thin-plate crystal of p-terphenyl, prepared by growth on a water surface, showing an extended network of defects. (d) A sample of a 1:1 theophylline/L-malic acid cocrystal prepared by liquid assisted grinding. The crystallites have habits ranging from thin plate to needle (250 kV). (e) Polycrystalline diffraction pattern from the same sample. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
crystals from nitromethane grew most rapidly in the [001] direction, whereas those from methanol grew fastest in the [010] direction. TEM is particularly useful for morphological examination of nanosized crystals. The existence of crystal defects has been established for theophylline and paracetamol. Series of parallel imperfections running in the [010] crystallographic direction, as shown in Figure 9, have been observed in many crystals of form II of theophylline. The nature of the interaction of bend contours with these defects suggests that they are dislocations. Samples of paracetamol form I grown from the melt show possible evidence of micro-twinning. The influence that crystal defects have on sample behavior was also apparent for theophylline. It was observed that triangular crystals of theophylline spontaneously changed shape over time. It appears that the crystals fractured in the [010] crystallographic direction, along the observed dislocations, giving smaller, trapezoid-shaped crystals. The fracturing could also be induced mechanically. It is known that defects are sites in crystals at which transformations, such as polymorphic transitions and hydrate formation, are initiated,35–41 and that processes routinely used in the manufacture of pharmaceutical products such as milling and tableting introduce defects.42,43 The ability to observe and characterize crystal defects, as afforded by TEM, could therefore significantly improve the understanding and control of the solid-state behavior of pharmaceutical compounds. Figure 10 compares the nature of the results which can be obtained from specimens prepared by different methods. The amount of information that can be obtained for samples prepared by solution cooling and solution evaporation is determined by crystal thickness. With large or block-shaped crystals the electron beam is scattered so completely that detailed imaging is not achievable. However, phase identification is still possible as electron diffraction patterns can be acquired from edges or corners of crystals where they are thinner. Flake or plate crystal habits, as exhibited by aspirin crystallized by evaporation from acetonitrile, can have large regions with suitable thickness for TEM analysis. Lath or needle-shaped crystals are often thin enough in at least one dimension for imaging as shown for needle-shaped crystals of form I of caffeine. It is evident that the crystal of caffeine marked with an asterisk in the Figure is tubular.44 However, the area of view for each crystal is small, making defects hard to detect. Suitable crystal morphologies for TEM can be targeted through choice of crystallization solvent and/or addition of small amounts of impurities.45–47. Crystallization of the organic compound p-terphenyl on a water surface, following a method reported by DOI 10.1002/jps
4081
Jones et al.,48 generated thin foils with thicknesses of <500 nm and areas of up to 1 cm2, making them ideal for TEM analysis. Thin foils could not be obtained for the pharmaceutical compounds used in this study, but it is believed that this method will prove to be an optimal way of preparing TEM specimens of other pharmaceutical materials. Melt crystallization offers the advantage that material can be thinly spread over a TEM sample grid while liquid, resulting in thin specimens, as shown for paracetamol (Fig. 9). Ball mill grinding was found to reliably generate crystals of sub-micron size, suitably thin to transmit electrons, for which crystal habit information could be obtained. However, phase identification for these samples proved difficult (but not impossible) as diffraction patterns usually contained contributions from several crystallites. The pictured crystallites of a 1:1 theophylline/L-malic acid cocrystal,49 prepared by liquid assisted grinding, have particle sizes ranging from 50 to 1000 nm and morphologies ranging from needle to thin plate. This result demonstrates the suitability of TEM for the characterization of pharmaceutical nanomaterials. TEM is already routinely used for imaging liposomal, micellular and polymeric nanoparticles containing encapsulated pharmaceutical compounds.14,50–52
CONCLUDING REMARKS Sample preparation and beam damage issues, widely considered to prohibit TEM characterization of pharmaceutical samples, can be sufficiently overcome with simple approaches to enable information about a range of pharmaceutical materials to be obtained. Furthermore, identification of defects in samples of theophylline, as described here, could not be achieved with any other analytical technique currently used in pharmaceutical analysis. In the wider materials context, TEM has moved somewhat away from bright/dark field diffractioncontrast based techniques towards approaches which yield three dimensional data and/or are directly interpretable (electron tomography, aberration-corrected high resolution, scanned probe methods).10,11 At the present time, the majority of pharmaceutical samples are not amenable to these techniques because of the requirement for high signal-noise ratios, and thus high total electron doses. We note here that the long-established diffraction contrast techniques13 still have an important role when the limitations on the analysis come not from the instrumentation but from the sample itself. We have demonstrated that TEM analysis could be advantageous in the pharmaceutically important areas of solid-phase identification and patent infrinJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4082
EDDLESTON, BITHELL, AND JONES
gement, and has the potential to provide a greater understanding of defects, and related reactivity, in pharmaceutical crystals.
ACKNOWLEDGMENTS M.D.E. and W.J. thank the EPSRC for funding. E.G.B. is grateful for support in the form of a Daphne Jackson Fellowship funded by Lucy Cavendish College, Cambridge, the Thriplow Charitable Trust and the Isaac Newton Trust and is currently supported by the ERC. We thank Andrew M.C. Cassidy for supplying a sample of the 2:1 caffeine/oxalic acid cocrystal.
REFERENCES 1. Brittain HG, editor. 1995. Physical characterization of pharmaceutical solids. New York: Marcel Dekker. 2. Vippagunta SR, Brittain HG, Grant DJ. 2001. Crystalline solids. Adv Drug Deliv Rev 48:3–26. 3. Haleblian J, McCrone W. 1969. Pharmaceutical applications of polymorphism. J Pharm Sci 58:911–929. 4. Singhal D, Curatolo W. 2004. Drug polymorphism and dosage form design: A practical perspective. Adv Drug Deliv Rev 56: 335–347. 5. Bauer J, Spanton S, Henry R, Quick J, Dziki W, Porter W, Morris J. 2001. Ritonavir: An extraordinary example of conformational polymorphism. Pharm Res 18:859–866. 6. Brittain HG, Fiese EF. 1999. Effects of pharmaceutical processing on drug polymorphs and solvates. In: Brittain HG, editor. Polymorphism in pharmaceutical sciences. New York: Marcel Dekker. pp. 331–361. 7. Fleischman SG, Kuduva SS, McMahon JA, Moulton B, Bailey Walsh RD, Rodriguez-Hornedo N, Zaworotko MJ. 2003. Crystal engineering of the composition of pharmaceutical phases: Multiple-component crystalline solids involving carbamazepine. Cryst Growth Des 3:909–919. 8. Hirsch PB, Howie A, Nicholson RB, Pashley DW, Whelan MJ. 1965. Electron microscopy of thin crystals. London: Butterworths. 9. McLaren AC. 1991. Transmission electron microscopy of minerals and rocks. 2nd edition. Cambridge: Cambridge University Press. 10. Thomas JM, Midgley PA. 2004. High-resolution transmission electron microscopy: The ultimate nanoanalytical technique. Chem Commun (Cambridge United Kingdom) 11:1253–1267. 11. Midgley PA, Dunin-Borkowski RE. 2009. Electron tomography and holography in materials science. Nat Mater 8:271–280. 12. Armigliato A. 1979. Analysis of crystal defects by transmission electron microscopy. Mater Chem 4:453–471. 13. Hirsch P, Cockayne D, Spence J, Whelan M. 2006. 50 years of TEM of dislocations. Past, present and future. Philos Mag 86: 4519–4528. 14. Tang Y, Liu Shiyong Y, Armes Steven P, Billingham Norman C. 2003. Solubilization and controlled release of a hydrophobic drug using novel micelle-forming ABC triblock copolymers. Biomacromolecules 4:1636–1645. 15. Liu Y, Yang Z, Du J, Yao X, Zheng X, Lei R, Liu J, Hu H, Li H. 2008. Interaction of Taxol with intravenous immunoglobulin: An inhibition of Taxol from crystallizing in aqueous solution. Int Immunopharmacol 8:390–400. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
16. Rades T, Mueller-Goymann CC. 1997. Electron and light microscopical investigation of defect structures in mesophases of pharmaceutical substances. Colloid Polym Sci 275:1169–1178. 17. Li ZG, Harlow RL, Foris CM, Li H, Ma P, Vickery RD, Maurin MB, Toby BH. 2002. New applications of electron diffraction in the pharmaceutical industry: Polymorph determination by using a combination of electron diffraction and synchrotron X-ray powder diffraction techniques. Microsc and Microanal 8:134–138. 18. Reimer L, Kohl H. 2008. Transmission electron microscopy: Physics of image formation. 5th edition. New York: Springer. 19. Jones W, Thomas JM. 1979. Applications of electron microscopy to organic solid-state chemistry. Progr Solid State Chem 12:101–124. 20. Trask AV, Jones W. 2005. Crystal engineering of organic cocrystals by the solid-state grinding approach. Top Curr Chem 254:41–70. 21. Friscic T, Jones W. 2009. Recent advances in understanding the mechanism of cocrystal formation via grinding. Cryst Growth Des 9:1621–1637. 22. Ebisuzaki Y, Boyle PD, Smith JA. 1997. Methylxanthines. I. Anhydrous theophylline. Acta Crystallogr C C53:777–779. 23. Perrin M-A, Neumann MA, Elmaleh H, Zaske L. 2009. Crystal structure determination of the elusive paracetamol Form III. Chem Commun: 3181–3183. 24. Perlovich GL, Volkova TV, Bauer-Brandl A. 2007. Polymorphism of paracetamol. Relative stability of the monoclinic and orthorhombic phase revisited by sublimation and solution calorimetry. J Therm Anal Calorim 89:767–774. 25. Di Martino P, Guyot-Hermann AM, Conflant P, Drache M, Guyot JC. 1996. A new pure paracetamol for direct compression: The orthorhombic form. Int J Pharm 128:1–8. 26. Di Martino P, Conflant P, Drache M, Huvenne JP, GuyotHermann AM. 1997. Preparation and physical characterization of forms II and III of paracetamol. J Therm Anal 48:447–458. 27. Haisa M, Kashino S, Kawai R, Maeda H. 1976. The monoclinic form of p-hydroxyacetanilide. Acta Crystallogr B B32:1283– 1285. 28. Nichols G, Frampton CS. 1998. Physicochemical characterization of the orthorhombic polymorph of paracetamol crystallized from solution. J Pharm Sci 87:684–693. 29. Mirmehrabi M, Rohani S, Murthy KSK, Radatus B. 2004. Characterization of tautomeric forms of ranitidine hydrochloride: Thermal analysis, solid-state NMR, X-ray. J Cryst Growth 260:517–526. 30. Trask AV, Motherwell WDS, Jones W. 2005. Pharmaceutical cocrystallization: Engineering a remedy for caffeine hydration. Cryst Growth Des 5:1013–1021. 31. Suzuki E, Shimomura K, Sekiguchi K. 1989. Thermochemical study of theophylline and its hydrate. Chem Pharm Bull 37: 493–497. 32. Roy C, Vega-Gonzalez A, Subra-Paternault P. 2007. Theophylline formulation by supercritical antisolvents. Int J Pharm 343:79–89. 33. Cruz Cabeza AJ, Day GM, Motherwell WDS, Jones W. 2006. Prediction and observation of isostructurality induced by solvent incorporation in multicomponent crystals. J Am Chem Soc 128:14466–14467. 34. Price SL. 2004. The computational prediction of pharmaceutical crystal structures and polymorphism. Adv Drug Deliv Rev 56:301–319. 35. Thomas JM. 1974. Topography and topology in solid state chemistry. Philosophical Transactions of the Royal Society of London, Series A: Mathematical. Phys Eng Sci 277:251–287. 36. Bhadeshia HKDH. 2001. Martensitic transformation. In: Buschow K, Cahn RW, Flemings MC, Lischner B, Kramer EJ, Mahajan S, editors. Encyclopedia of materials science: Science and technology. Pergamon Press: Elsevier Science. pp. 5203–5206. DOI 10.1002/jps
TEM OF PHARMACEUTICAL MATERIALS
37. Jones W, Thomas JM, Williams JO. 1975. Electron and optical microscopic studies of a stress-induced phase transition in 1,8dichloro-10-methylanthracene. Philos Mag 32:1–11. 38. Mnyukh YV. 1976. Polymorphic transitions in crystals: Nucleation. J Cryst Growth 32:371–377. 39. Thomas JM, Williams JO. 1971. Dislocations and the reactivity of organic solids. Progr Solid State Chem 6:119–154. 40. Thomas JM, Renshaw GD. 1969. Role of dislocations in the dehydration of nickel sulfate hexahydrate. II. Topographical studies by use of optical microscopy. J Chem Soc A: 2753–2755. 41. Prasad KVR, Ristic RI, Sheen DB, Sherwood JN. 2002. Dissolution kinetics of paracetamol single crystals. Int J Pharm 238:29–41. 42. Koivisto M, Heinaenen P, Tanninen VP, Lehto V-P. 2006. Depth profiling of compression-induced disorders and polymorphic transition on tablet surfaces with grazing incidence X-ray diffraction. Pharm Res 23:813–820. 43. Byard SJ, Jackson SL, Smail A, Bauer M, Apperley DC. 2005. Studies on the crystallinity of a pharmaceutical development drug substance. J Pharm Sci 94:1321–1335. 44. Eddleston MD, Jones W. 2010. Formation of tubular crystals of pharmaceutical compounds. Cryst Growth Des 10:365–370. 45. Lee AY, Myerson AS. 2006. Particle engineering: Fundamentals of particle formation and crystal growth. MRS Bull 31:881– 886.
DOI 10.1002/jps
4083
46. Lee T, Kuo CS, Chen YH. 2006. Solubility, polymorphism, crystallinity, and crystal habit of acetaminophen and ibuprofen: By initial solvent screening. Pharm Technol 30:72–92. 47. Berkovitch-Yellin Z, Addadi L, Idelson M, Lahav M, Leiserowitz L. 1982. Controlled modification of crystal habit via ‘‘tailormade’’ impurities: Application to benzamide. Angew Chem Int Ed Engl 21:1336–1345. 48. Jones W, Thomas JM, Williams JO. 1975. Electron microscopic studies of extended defects in organic molecular crystals. I. pTerphenyl. J Chem Soc, Faraday Trans 2 71:138–145. 49. Friscic T, Fabian L, Burley JC, Reid DG, Duer MJ, Jones W. 2008. Exploring the relationship between cocrystal stability and symmetry: Is Wallach’s rule applicable to multi-component solids? Chem Commun 14:1644–1646. 50. Almgren M, Edwards K, Karlsson G. 2000. Cryo transmission electron microscopy of liposomes and related structures. Colloid Surface A 174:3–21. 51. Kim I-S, Kim S-H. 2003. Development of polymeric nanoparticulate drug delivery systems: Evaluation of nanoparticles based on biotinylated poly(ethylene glycol) with sugar moiety. Int J Pharm 257:195–203. 52. Prabaharan M, Grailer JJ, Pilla S, Steeber DA, Gong S. 2009. Gold nanoparticles with a monolayer of doxorubicin-conjugated amphiphilic block copolymer for tumor-targeted drug delivery. Biomaterials 30:6065–6075.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010