- Email: [email protected]
Characterization of Dihydrates Prepared from Carbamazepine Polymorphs
+
+
Characterization of Dihydrates Prepared from Carbamazepine Polymorphs LAURA E. MCMAHON*, PETER TIMMINS‡X, ADRIAN C. WILLIAMS*,
AND
PETER YORK*
Received March 6, 1996, from the *Postgraduate Studies in Pharmaceutical Technology, The School of Pharmacy, University of Bradford, Bradford, W. Yorkshire BD7 1DP, U.K., and ‡International Development Laboratories, Bristol-Myers Squibb Pharmaceutical Research Institute, Reeds Lane, Moreton, Merseyside L46 1QW, U.K. Final revised manuscript received June 10, 1996 . Accepted for publication July 11, 1996X. Abstract 0 To clarify earlier literature suggesting its existence in more than one solid form, carbamazepine dihydrate was prepared from individual anhydrous carbamazepine polymorphs I and III. The resultant materials were characterized by thermal and spectroscopic techniques, including variable-temperature Fourier transform (FT) Raman spectroscopy. On thermal dehydration, the carbamazepine dihydrates formed different anhydrous polymorphs depending on both starting material and dehydration conditions. Under conditions where liberated hydrate water cannot readily escape from around the sample, the dihydrate originating from polymorph III reverted to polymorph III, but under conditions of lower humidity/moisture, this dihydrate reverted to polymorph I. The dihydrate originating from polymorph I reverted to polymorph I irrespective of dehydration conditions. These observations could be due to a trace amount of nonhydrated original polymorph existing in the dihydrate phase that acts as a seed to regenerate the original polymorph under conditions where liberated hydrate water cannot readily escape from around the sample during dehydration. No evidence was generated that supported the existence of true polymorphs of carbamazepine dihydrate.
Introduction Hydrates are the most common solvates encountered in pharmaceutical compounds and can be stoichiometric (e.g., ampicillin trihydrate) or nonstoichiometric (e.g., sodium cromoglycate hydrates). In hydrates, water occupies definite positions in the crystal lattice, usually by forming hydrogen bonds with the anhydrate drug molecules. The solubility and dissolution behavior of anhydrous and hydrated forms are dissimilar; anhydrous forms usually have greater aqueous solubility and faster dissolution rates than the hydrated species, and this effect could be reflected upon absorption invivo.1 Upon dehydration, hydrates can convert to one of three different modifications: (a) the crystal lattice of the dehydrated form is virtually identical to that of the original hydrate, (b) the dehydrated form is poorly crystalline or amorphous, (c) the dehydrated form recrystallizes with a crystal lattice that is different from the original hydrate.2 Several compounds convert to hydrated forms on addition of water, including carbamazepine,3,4 theophylline,5 and mercaptopurine.6 Anhydrous carbamazepine will rapidly transform to a dihydrate when dispersed in water.3,4 Crystal growth is by the whisker mechanism,4 and conversion has been shown, by X-ray powder diffraction, to be 95% complete after 1 h.7 In addition to the dihydrate, two well-defined enantiotropic polymorphs of carbamazepine, forms I and III,8 have been described. Whereas most previous work has agreed on the properties of the dihydrate itself, several studies have indicated that different crystal modifications can be obtained on removal of hydrate water.3,4,9-14 The relationship between the dihydrate and two principal polymorphs of carbamazepine is sumX
Abstract published in Advance ACS Abstracts, September 1, 1996.
1064 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
Scheme 1sRelationship between carbamazepine forms III and I, and dihydrate (after Umeda et al.).15
marized in Scheme 1.11,15 Most researchers used form III to prepare the dihydrate, although Lefebvre et al.9 chose the metastable form I. The resultant dihydrate was shown, using thermal analysis, to convert to form I on water loss. Grinding of the dihydrate led to the introduction of a form III endotherm, but no explanation was offered. The existence of two crystalline modifications of the dihydrate has been proposed10 but never further examined. Dugue et al.12 also published evidence of differences in dihydrate dehydration properties, depending on local humidity conditions. In an open environment, water loss led to formation of form I, whereas in a sealed environment, dehydration led to the production of form III. The polymorphic form of the material used to prepare the dihydrate was not indicated. The present work examines properties of carbamazepine dihydrate prepared from carbamazepine of known polymorphic form.
Experimental Section MaterialssPolymorph III carbamazepine was USP grade material and obtained from Diamalt (Munich, Germany). Polymorph I carbamazepine was prepared by heating polymorph III at 170 °C for 2 h as described by Lefebvre et al.9 Polymorphic form was confirmed by differential scanning calorimetry (DSC) analysis and comparing the results with published data.16 Dihydrate was prepared separately from polymorph I or III by suspending anhydrate in distilled water and stirring for 24 h at room temperature.7 Hydrated carbamazepine was filtered under suction and dried at room temperature for 30 min on the filter. Water content was confirmed by thermogravimetric analysis (TGA) and Karl Fischer analysis. The hydrates were stored at room temperature and 55-60% relative humidity. These conditions assured that hydrate integrity was maintained.17 Thermal AnalysissTGA was run on a Perkin Elmer Series 7 instrument. Isothermal runs at 25 °C under a flow of dry nitrogen were used to measure the kinetics of water loss. DSC was run in crimped pans with pin-hole pierced lids on an indium-calibrated Perkin-Elmer Series 7 instrument. Heating rates in the range 5-40 °C min-1 were used. Hot stage microscopy (HSM) was undertaken with a Stanton Redcroft hot stage and a Labophot-2A microscope equipped with a water-cooled universal temperature controller. Video recording (JVC TM-1500PS recorder) allowed accurate evaluation of thermal events.
S0022-3549(96)00117-7 CCC: $12.00
© 1996, American Chemical Society and American Pharmaceutical Association
+
+
Table 1s Water Content of Dihydrates of Carbamazepinea Dihydrate
Karl Fischer (n ) 3)
TGA (n ) 2)b
III I
12.62 (± 0.23) 13.32 (± 0.08)
11.30 (± 0) 10.98 (± 0.12)
aResults are expressed as % (± standard error). b Run isothermally at 25 °C under dry nitrogen stream.
Spectroscopys13C-Solid-state nuclear magnetic resonance (SSNMR) analysis was undertaken on a Varian VXR300 instrument with magic angle spinning (MAS) and cross-polarization. A field of 7.05 Tesla and a resonance frequency for 13C at 74.5 MHz were employed. A Doty Scientific multinuclear MAS probe with a 7-mm diameter kelF-rotor at a maximum spin speed of 5.54 Hz was used. Chemical shifts were determined with reference to the external standard adamantane, with the CH2 line at 38.4 ppm relative to tetramethylsilane. Diffuse-reflectance infrared Fourier transform (DRIFT) spectroscopy was undertaken on a Mattson Instruments 6020 Galaxy Series spectrophotometer equipped with a deuterium triglycine sulphate detector. Samples were triturated with potassium bromide, thus avoiding mechanically induced polymorphic transitions possibly induced by extended grinding,9 and scanned 500 times over the range 4000-400 cm-1 at a resolution of 4 cm-1 to accumulate spectra. FT-Raman spectroscopy was accomplished with a Bruker FRA 106 module mounted on an IFS 66 optics bench and equipped with a 750 mW Nd:YAG laser operating at 1064 nm. Samples were held in a stainless steel cup and exposed to the focussed laser with a spot diameter of 0.1 mm. Standard FT-Raman spectra were collected from 200 scans at 4 cm-1. A liquid nitrogen-cooled germanium detector with an extended spectral band width that covered the range 503500 cm-1 was used. Band wave numbers were calibrated against internal laser frequencies, which provided vibrational band wave numbers correct to (1 cm-1. An FT-Raman kinetic heating study was undertaken in a novel environmental chamber. The heating chamber was adapted from a Graseby-Specac 19930 variable-temperature diffuse reflectance accessory with a removable glass window. The orientation of the accessory was modified to permit a 90° sampling geometry. This setup allowed collection of spectra in-situ during heating to analyze differences during and following dehydration. Temperature control was obtained with a Graseby-Specac 20120 series programmer. The temperature of the sample was monitored with a thermocouple connected to the sample port. Samples were heated at 5 °C min-1 over the range of 25-30 to 180 °C. Spectra were collected throughout the heating cycle with the OPUS programmable cycling mode set at zero delay between spectra. Forty scans at 4 cm-1 provided spectra of sufficient resolution. Heating studies were performed with the chamber window attached or unattached to simulate sealed or exposed conditions, respectively, as described for thermal analysis by Dugue et al.12 Crystallographic StudiessVariable-temperature X-ray powder diffraction (XRPD) studies were run on a Siemens D5000 diffractometer, with CuKR radiation of λ 1.541 Å. Powder samples were presented in the cavity of a stainless steel holder. An Anton Paar TTK2-HC controller was used to heat the sample at 10 °C min-1 in an Anton Paar 589775 sample holder from 30 to 200 °C. Data were collected at 10 °C intervals between 3° and 40° 2θ in a step size of 0.05° 2θ and at a count rate of 3 s/step.
Results and Discussion Dihydrates prepared from form III are referred to as ‘dihydrate III’ and those prepared from polymorph I as ‘dihydrate I’. Water contents as determined by Karl Fischer and TGA are presented in Table 1. These results compare favorably with the theoretical stoichiometric water content of 13.2% w/w. The TGA values were slightly lower than theoretical values, which may be due to a small water loss inside the apparatus at the initial temperature under the nitrogen flow. Rates of water loss from dihydrates III and I, when maintained isothermally at 25 °C in the TGA, were similar
Figure 1sTGA thermograms for samples of dihydrates I and III held at 25 °C.
Figure 2sDSC curves for dihydrates III and I.
(Figure 1). Because water is readily removed at 25 °C under the dry nitrogen flow during TGA, the water molecules have only weak associations within the dihydrate crystal lattice. Examination of the dehydrated materials from the TGA, employing FT-Raman spectroscopy, showed spectra consistent with carbamazepine form I,18 regardless of whether the dihydrate employed was prepared with form I or form III. The DSC curves showed water removal at ∼50-90 °C for dihydrates III and I, which is consistent with previously published data.12 In addition, on heating to higher temperatures, it was possible to discern subtle differences in the profiles of dihydrates III and I (Figure 2). Dihydrate III exhibited a small endotherm (1.04 J g-1) at ∼160 °C, which was followed by a melting endotherm at 190 °C. The enthalpy for this latter endotherm was lower than that seen for this endotherm when examining anhydrous forms III or I (dihydrate III, 94.0 J g-1 (n ) 2); polymorph III, 109.0 J g-1 (n ) 3); polymorph I, 110.3 J g-1 (n ) 2)]. This result suggests that under the chosen experimental conditions, dehydration leads to a material of lower crystallinity than the form from which it was prepared. This reduced crystallinity following water removal can be discerned in the changes in intensity and width of peaks in, for example, the 15-25° 2θ region in the variable-temperature XRPD plots over the range 30-170 °C, as shown in Figures 3a and 3b. The endotherm for dihydrate III at 160 °C was reproducible over a range of DSC heating rates (5, 10, 20, and 40 °C min-1). Given the temperature at which the endotherm for dihydrate III occurs, it is suggested that under DSC conditions this dihydrate transforms to form III material on dehydration and then undergoes a phase change to form I at ∼160 °C, similar to the thermal characteristics of anhydrous carbamazepine form
Journal of Pharmaceutical Sciences / 1065 Vol. 85, No. 10, October 1996
+
+
Figure 3sXRPD curves of (a) carbamazepine dihydrate III and (b) carbamazepine dihydrate I over the range 30−170 °C.
Figure 5sFTIR spectra of carbamazepine dihydrates I and III. Figure 4sFT-Raman spectra of carbamazepine polymorphs and materials obtained after heating dihydrate III or I: (a) heated carbamazepine dihydrate I; (b) heated carbamazepine dihydrate III; (c) carbamazepine I; and (d) carbamazepine III.
III. On the other hand, after water removal, dihydrate I exhibited only a form I melting endotherm, indicating that this dihydrate transforms to form I material in the DSC pan after water removal. Dihydrate samples heated to 110 °C in the DSC pan and subsequently examined by FT-Raman spectroscopy showed spectra consistent with carbamazepine form III obtained from dihydrate III and carbamazepine form I from dihydrate I.18 These spectra are further evidence that on dehydration in the DSC pan, dihydrates III and I transform to different forms (Figure 4). This behavior is similar in the case of dihydrate III to that reported for the dehydration of a carbamazepine dihydrate in elevated-pressure DSC, although behavior similar to dihydrate I was not observed in elevated-pressure DSC.14 Dehydration behavior for dihydrates III and I was also evaluated by HSM. At ∼50 °C, both dihydrates darkened in 1066 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
color; this change was associated with loss of water of crystallization. On further heating, minute crystals appeared to grow from the surface of the formerly dihydrate needles. No further changes were exhibited until the needles melted at 190 °C. Thus, no differences were thus seen between the two dihydrates with HSM. Major differences were not discernible on comparison of various spectra run on dihydrates III and I. The 13C-SSNMR spectra for both hydrates were identical, although the spectrum for dihydrate I was better resolved. This better resolution probably reflects a difference in crystallinity between the samples. The DRIFT spectra for both hydrates were very similar and consistent with those in the literature.19 There was a small additional peak in the spectrum of dihydrate I at 3486 cm-1 that is present in the spectrum of carbamazepine form I and associated with -NH2 stretching (Figure 5). Whereas this vibration might be unique to a true form I dihydrate, the corresponding vibration in the spectrum of dihydrate III may be masked by the prominent band for -OH stretching from water molecules. Alternatively, this vibration could arise
+
+
Figure 6sFT-Raman spectra of (a) carbamazepine dihydrate I and (b) carbamazepine dihydrate III recorded when heated in ‘exposed’ conditions (see text).
Figure 7sFT-Raman spectra of (a) carbamazepine dihydrate I and (b) carbamazepine dihydrate III recorded when heated in ‘sealed’ conditions (see text).
because of the presence of a small amount of anhydrous form I remaining unconverted in the isolated dihydrate. No other bands specific to anhydrous form I were identifiable in the DRIFT spectrum of dihydrate I. FT-Raman spectroscopy offered a better opportunity to discern differences between the dihydrates. Water is a weak Raman scatterer and hence provides minimal interference in the FT-Raman spectra, unlike in Fourier transform infrared (FTIR) spectra where masking of some bands that might have been diagnostic for a particular polymorphic form was observed. Assignment of the dihydrate to a particular anhydrous polymorphic conformation was not straightforward when comparing FT-Raman spectra because the dihydrate peak positions did not equate with either anhydrous form. The strongest similarities were with carbamazepine form III. The lattice vibration regions in the spectra (50-200 cm-1) were similar for both dihydrates, suggesting that both exist in a very similar lattice (Figure 6).
Conditions used for in-situ variable temperature FT-Raman spectroscopy were adjusted to simulate conditions occurring in the DSC pans. Heating studies performed with the chamber window attached (“sealed”; conditions analogous to the DSC pan) or unattached (“exposed”; conditions analogous to heating with TGA or HSM) were undertaken. Dihydrates III and I were subjected to heating in both environments. The lattice vibration regions of FT-Raman spectra obtained when dihydrates were heated in the “exposed” environment are shown in Figures 6a and 6b. Between ∼60-70 °C, the dihydrate spectra changed and, in both cases, displayed lattice vibration associated with carbamazepine form I that implies loss of hydrate water. No additional changes in these vibrations were observed during further temperature increases up to 170 °C. When the dihydrates were heated in the “sealed” environment, the spectra show that the dihydrates revert to the crystal form from which they were prepared. The lattice
Journal of Pharmaceutical Sciences / 1067 Vol. 85, No. 10, October 1996
+
+
Figure 8sXRPD curves for carbamazepine dihydrates (a) III and (b) I, and carbamazine polymorphs (c) III and (d) I.
vibration regions of the spectra obtained on heating dihydrate I are shown in Figure 7a. At ∼70 °C, the dihydrate had lost water and transformed to carbamazepine form I, and no change was seen on further heating. This result is similar to the observations made when both dihydrates were heated in the exposed conditions. The spectra obtained on heating dihydrate III under “sealed” conditions are shown in Figure 7b. Upon losing water at ∼60 °C, the lattice vibrations associated with form III carbamazepine are evident (notably, the ratio of the peaks at ∼110 and 130 cm-1), although the spectra do not indicate that complete reversion to form III has occurred (compare with Figure 4). As the temperature is raised, these vibrations transform to the form I lattice vibrations at ∼160 °C, which is the expected form III-to-I transition temperature. Variable temperature XRPD data further indicated the nature of transformations of the dihydrates on heating. The diffraction patterns of both dihydrates over the temperature range 30-170 °C are shown in Figures 3a and 3b. Reference patterns for polymorphs I and III are shown in Figure 8. The presence of form I carbamazepine diffraction peaks between 7 and 9.5° 2θ indicates that form I is produced from both dihydrates after dehydration under the conditions employed. However, the diffraction patterns are dissimilar in other regions, especially in the 13-16° 2θ region. Dihydrate I appears to dehydrate to form I material of low crystallinity, as indicated by the poorly resolved low intensity peaks. As the temperature increases, the bands become sharper as form I recrystallization occurs. Upon dehydration, dihydrate III also produces a modification of lower crystallinity. The presence of some form III anhydrous carbamazepine is suggested by positions of peaks in the 13-16° 2θ range. Thus these data support the DSC results that the two dihydrates can produce materials of differing crystalline modifications 1068 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
after dehydration. These differences are apparent when samples are heated in the pierced DSC pan or in the FT-Raman in the “sealed” environment, where the liberated hydrate water does not readily escape. Where the liberated hydrate water can readily escape, as in a TGA or HSM experiment, the differences are not detected. The dihydrates produced from form I or form III carbamazepine differ in dehydration properties under certain conditions, and hence might have alternative molecular organization; however, no major differences were detectable using the high resolution techniques described earlier. Whereas 13C SSNMR is one of the more sensitive techniques available to detect differences in the environments of functional groups of a molecule, no significant difference was identified between the spectra of the two dihydrates, suggesting molecular conformation and the association of water molecules with carbamazepine molecules are the same in both dihydrates. The XRPD patterns were also similar (Figure 8), implying that the unit cells are equivalent in the two dihydrates, although features of carbamazepine III were discernible in patterns collected during the dehydration of dihydrate III. Therefore, from the available analyses, it seems unlikely that distinct polymorphic dihydrates were formed on hydration of carbamazepine forms I and III. An explanation of the dehydration phenomena observed may be that small amounts of the anhydrous polymorphic form remain after formation of the dihydrate. Such low levels, below the detectional resolution limit of the techniques employed, act in some way during the dehydration process or immediately after it as a template to “seed” the reversion of the carbamazepine to its original polymorphic starting form under conditions where the liberated water is not readily removed. A “memory” in the dihydrate crystals is activated by the water-carbamazepine interaction allowed to occur in the pierced DSC pan or “sealed”
+
+
Improved understanding of the dehydration of carbamazepine dihydrates prepared from different carbamazepine polymorphs is offered in Scheme 2. Although the existence of more than one hydrate polymorph of the same pharmaceutical compound can occur (e.g., amiloride hydrochloride dihydrates20), this occurrence does not appear to be the case for carbamazepine dihydrate.
References and Notes
Scheme 2sProposed relationship between carbamazepine forms III and I and dihydrate.
FT-Raman cell. Where liberated water is readily removed during the dehydration process, as in TGA, HSM, and in the FT-Raman cell in the “exposed” mode, the “seed” is unable to drive the crystallization of the starting polymorphic form. The high temperature metastable form I is favored in these cases, as the poorly crystalline material formed on dehydration recrystallizes under the heating conditions employed. A carbamazepine-water interaction was proposed as an explanation for the behavior of a carbamazepine dihydrate in elevated pressure DSC (where water is not really eliminated from the immediate environment of the drug),14 which is similar to the behavior of carbamazepine dihydrate III in “sealed” FT-Raman conditions seen in the present work. However, the dehydration under pressure of dihydrate prepared from carbamazepine form I was not examined in this previous work. In an environment that allows the carbamazepine-water interaction, this dihydrate only reverts to its starting polymorph and not to polymorph III (the only change seen in previous pressure DSC work14) and also not to the form that is stable at the temperature of the sample immediately after dehydration.
1. Khankari, R. K.; Grant, D. J. W. Thermochimica Acta 1995, 248, 61-79. 2. Garner, W. E. In Chemistry of the Solid State; Garner, W. E., Ed.: Academic: New York, 1955; pp 213-231. 3. Kahela, P.; Aaltonen, R.; Lewing E.; Antilla, M.; Kristofferson, E. Int. J. Pharm. 1983, 14, 103-112. 4. Laine, E.; Tuominen, V.; Ilvessalo, P.; Kahela, P. Int. J. Pharm. 1984, 20, 307-314. 5. Shefter, E.; Higuchi, T. J. Pharm. Sci. 1963, 52, 781-791. 6. Niazi, S. J. Pharm. Sci. 1978, 67, 488-491. 7. Young, W. L. L.; Suryanarayanan, R. J. Pharm. Sci. 1991, 80, 496-500. 8. Kala, H.; Haack, U.; Pollandt, P.; Brezesinski G. Acta. Pharm. Technol. 1986, 32, 72-77. 9. Lefebvre, C.; Guyot-Hermann, A. M.; Draguet-Brughmans, M.; Bouche, R.; Guyot, J. C. Drug. Dev. Ind. Pharm. 1986, 12, 19131927. 10. Reck, G.; Dietz, G. Cryst. Res. Technol. 1986, 21, 1463-1468. 11. Kaneniwa, N.; Ichikawa, J.; Yamaguchi, T.; Hayashi, K.; Watari, N.; Sumi, M. Yakugaku Zasshi 1987, 107, 808-813. 12. Dugue, J.; Ceolin, R.; Rouland, J. C.; Lepage F. Pharm. Acta. Helv. 1991, 66, 307-310. 13. Lovrecich, M.; Orzincolo, O.; Rubessa, F. Acta Tech. Leg. Medic. 1994, 5, 29-48. 14. Han, J.; Suryanarayanan. R. Pharm. Res. 1995, 12 (Suppl.) , S-141. 15. Umeda, T.; Ohnishi, N.; Yokaoyama, T.; Kuroda, K.; Kuroda, T.; Tatsumi, E.; Matsuda, Y. Yakugaku Zasshi 1984, 104, 786792. 16. Behme, R. J.; Brooke D. J. Pharm. Sci. 1991, 80, 986-990. 17. Suryanarayanan, R. Pharm. Res. 1989, 6, 1017-1024. 18. McMahon. L. E.; Williams, A. C.; York, P.; Edwards, H. G. M.; Timmins, P. In Proceedings of the XIVth International Conference on Raman Spectroscopy; Yu, N.-T., Li, X.-Y., Eds.; John Wiley & Sons: Chichester, 1994; pp 908-909. 19. Borka, L.; Lonmo, A.; Wisnes, R. Pharm. Acta. Helv. 1992, 67, 231-233. 20. Jozwiakowski, M. J.; Williams, S. O.; Hathaway R. D. Int. J. Pharm. 1993, 91, 195-207.
JS960117E
Journal of Pharmaceutical Sciences / 1069 Vol. 85, No. 10, October 1996
+
Characterization of Dihydrates Prepared from Carbamazepine Polymorphs LAURA E. MCMAHON*, PETER TIMMINS‡X, ADRIAN C. WILLIAMS*,
AND
PETER YORK*
Received March 6, 1996, from the *Postgraduate Studies in Pharmaceutical Technology, The School of Pharmacy, University of Bradford, Bradford, W. Yorkshire BD7 1DP, U.K., and ‡International Development Laboratories, Bristol-Myers Squibb Pharmaceutical Research Institute, Reeds Lane, Moreton, Merseyside L46 1QW, U.K. Final revised manuscript received June 10, 1996 . Accepted for publication July 11, 1996X. Abstract 0 To clarify earlier literature suggesting its existence in more than one solid form, carbamazepine dihydrate was prepared from individual anhydrous carbamazepine polymorphs I and III. The resultant materials were characterized by thermal and spectroscopic techniques, including variable-temperature Fourier transform (FT) Raman spectroscopy. On thermal dehydration, the carbamazepine dihydrates formed different anhydrous polymorphs depending on both starting material and dehydration conditions. Under conditions where liberated hydrate water cannot readily escape from around the sample, the dihydrate originating from polymorph III reverted to polymorph III, but under conditions of lower humidity/moisture, this dihydrate reverted to polymorph I. The dihydrate originating from polymorph I reverted to polymorph I irrespective of dehydration conditions. These observations could be due to a trace amount of nonhydrated original polymorph existing in the dihydrate phase that acts as a seed to regenerate the original polymorph under conditions where liberated hydrate water cannot readily escape from around the sample during dehydration. No evidence was generated that supported the existence of true polymorphs of carbamazepine dihydrate.
Introduction Hydrates are the most common solvates encountered in pharmaceutical compounds and can be stoichiometric (e.g., ampicillin trihydrate) or nonstoichiometric (e.g., sodium cromoglycate hydrates). In hydrates, water occupies definite positions in the crystal lattice, usually by forming hydrogen bonds with the anhydrate drug molecules. The solubility and dissolution behavior of anhydrous and hydrated forms are dissimilar; anhydrous forms usually have greater aqueous solubility and faster dissolution rates than the hydrated species, and this effect could be reflected upon absorption invivo.1 Upon dehydration, hydrates can convert to one of three different modifications: (a) the crystal lattice of the dehydrated form is virtually identical to that of the original hydrate, (b) the dehydrated form is poorly crystalline or amorphous, (c) the dehydrated form recrystallizes with a crystal lattice that is different from the original hydrate.2 Several compounds convert to hydrated forms on addition of water, including carbamazepine,3,4 theophylline,5 and mercaptopurine.6 Anhydrous carbamazepine will rapidly transform to a dihydrate when dispersed in water.3,4 Crystal growth is by the whisker mechanism,4 and conversion has been shown, by X-ray powder diffraction, to be 95% complete after 1 h.7 In addition to the dihydrate, two well-defined enantiotropic polymorphs of carbamazepine, forms I and III,8 have been described. Whereas most previous work has agreed on the properties of the dihydrate itself, several studies have indicated that different crystal modifications can be obtained on removal of hydrate water.3,4,9-14 The relationship between the dihydrate and two principal polymorphs of carbamazepine is sumX
Abstract published in Advance ACS Abstracts, September 1, 1996.
1064 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
Scheme 1sRelationship between carbamazepine forms III and I, and dihydrate (after Umeda et al.).15
marized in Scheme 1.11,15 Most researchers used form III to prepare the dihydrate, although Lefebvre et al.9 chose the metastable form I. The resultant dihydrate was shown, using thermal analysis, to convert to form I on water loss. Grinding of the dihydrate led to the introduction of a form III endotherm, but no explanation was offered. The existence of two crystalline modifications of the dihydrate has been proposed10 but never further examined. Dugue et al.12 also published evidence of differences in dihydrate dehydration properties, depending on local humidity conditions. In an open environment, water loss led to formation of form I, whereas in a sealed environment, dehydration led to the production of form III. The polymorphic form of the material used to prepare the dihydrate was not indicated. The present work examines properties of carbamazepine dihydrate prepared from carbamazepine of known polymorphic form.
Experimental Section MaterialssPolymorph III carbamazepine was USP grade material and obtained from Diamalt (Munich, Germany). Polymorph I carbamazepine was prepared by heating polymorph III at 170 °C for 2 h as described by Lefebvre et al.9 Polymorphic form was confirmed by differential scanning calorimetry (DSC) analysis and comparing the results with published data.16 Dihydrate was prepared separately from polymorph I or III by suspending anhydrate in distilled water and stirring for 24 h at room temperature.7 Hydrated carbamazepine was filtered under suction and dried at room temperature for 30 min on the filter. Water content was confirmed by thermogravimetric analysis (TGA) and Karl Fischer analysis. The hydrates were stored at room temperature and 55-60% relative humidity. These conditions assured that hydrate integrity was maintained.17 Thermal AnalysissTGA was run on a Perkin Elmer Series 7 instrument. Isothermal runs at 25 °C under a flow of dry nitrogen were used to measure the kinetics of water loss. DSC was run in crimped pans with pin-hole pierced lids on an indium-calibrated Perkin-Elmer Series 7 instrument. Heating rates in the range 5-40 °C min-1 were used. Hot stage microscopy (HSM) was undertaken with a Stanton Redcroft hot stage and a Labophot-2A microscope equipped with a water-cooled universal temperature controller. Video recording (JVC TM-1500PS recorder) allowed accurate evaluation of thermal events.
S0022-3549(96)00117-7 CCC: $12.00
© 1996, American Chemical Society and American Pharmaceutical Association
+
+
Table 1s Water Content of Dihydrates of Carbamazepinea Dihydrate
Karl Fischer (n ) 3)
TGA (n ) 2)b
III I
12.62 (± 0.23) 13.32 (± 0.08)
11.30 (± 0) 10.98 (± 0.12)
aResults are expressed as % (± standard error). b Run isothermally at 25 °C under dry nitrogen stream.
Spectroscopys13C-Solid-state nuclear magnetic resonance (SSNMR) analysis was undertaken on a Varian VXR300 instrument with magic angle spinning (MAS) and cross-polarization. A field of 7.05 Tesla and a resonance frequency for 13C at 74.5 MHz were employed. A Doty Scientific multinuclear MAS probe with a 7-mm diameter kelF-rotor at a maximum spin speed of 5.54 Hz was used. Chemical shifts were determined with reference to the external standard adamantane, with the CH2 line at 38.4 ppm relative to tetramethylsilane. Diffuse-reflectance infrared Fourier transform (DRIFT) spectroscopy was undertaken on a Mattson Instruments 6020 Galaxy Series spectrophotometer equipped with a deuterium triglycine sulphate detector. Samples were triturated with potassium bromide, thus avoiding mechanically induced polymorphic transitions possibly induced by extended grinding,9 and scanned 500 times over the range 4000-400 cm-1 at a resolution of 4 cm-1 to accumulate spectra. FT-Raman spectroscopy was accomplished with a Bruker FRA 106 module mounted on an IFS 66 optics bench and equipped with a 750 mW Nd:YAG laser operating at 1064 nm. Samples were held in a stainless steel cup and exposed to the focussed laser with a spot diameter of 0.1 mm. Standard FT-Raman spectra were collected from 200 scans at 4 cm-1. A liquid nitrogen-cooled germanium detector with an extended spectral band width that covered the range 503500 cm-1 was used. Band wave numbers were calibrated against internal laser frequencies, which provided vibrational band wave numbers correct to (1 cm-1. An FT-Raman kinetic heating study was undertaken in a novel environmental chamber. The heating chamber was adapted from a Graseby-Specac 19930 variable-temperature diffuse reflectance accessory with a removable glass window. The orientation of the accessory was modified to permit a 90° sampling geometry. This setup allowed collection of spectra in-situ during heating to analyze differences during and following dehydration. Temperature control was obtained with a Graseby-Specac 20120 series programmer. The temperature of the sample was monitored with a thermocouple connected to the sample port. Samples were heated at 5 °C min-1 over the range of 25-30 to 180 °C. Spectra were collected throughout the heating cycle with the OPUS programmable cycling mode set at zero delay between spectra. Forty scans at 4 cm-1 provided spectra of sufficient resolution. Heating studies were performed with the chamber window attached or unattached to simulate sealed or exposed conditions, respectively, as described for thermal analysis by Dugue et al.12 Crystallographic StudiessVariable-temperature X-ray powder diffraction (XRPD) studies were run on a Siemens D5000 diffractometer, with CuKR radiation of λ 1.541 Å. Powder samples were presented in the cavity of a stainless steel holder. An Anton Paar TTK2-HC controller was used to heat the sample at 10 °C min-1 in an Anton Paar 589775 sample holder from 30 to 200 °C. Data were collected at 10 °C intervals between 3° and 40° 2θ in a step size of 0.05° 2θ and at a count rate of 3 s/step.
Results and Discussion Dihydrates prepared from form III are referred to as ‘dihydrate III’ and those prepared from polymorph I as ‘dihydrate I’. Water contents as determined by Karl Fischer and TGA are presented in Table 1. These results compare favorably with the theoretical stoichiometric water content of 13.2% w/w. The TGA values were slightly lower than theoretical values, which may be due to a small water loss inside the apparatus at the initial temperature under the nitrogen flow. Rates of water loss from dihydrates III and I, when maintained isothermally at 25 °C in the TGA, were similar
Figure 1sTGA thermograms for samples of dihydrates I and III held at 25 °C.
Figure 2sDSC curves for dihydrates III and I.
(Figure 1). Because water is readily removed at 25 °C under the dry nitrogen flow during TGA, the water molecules have only weak associations within the dihydrate crystal lattice. Examination of the dehydrated materials from the TGA, employing FT-Raman spectroscopy, showed spectra consistent with carbamazepine form I,18 regardless of whether the dihydrate employed was prepared with form I or form III. The DSC curves showed water removal at ∼50-90 °C for dihydrates III and I, which is consistent with previously published data.12 In addition, on heating to higher temperatures, it was possible to discern subtle differences in the profiles of dihydrates III and I (Figure 2). Dihydrate III exhibited a small endotherm (1.04 J g-1) at ∼160 °C, which was followed by a melting endotherm at 190 °C. The enthalpy for this latter endotherm was lower than that seen for this endotherm when examining anhydrous forms III or I (dihydrate III, 94.0 J g-1 (n ) 2); polymorph III, 109.0 J g-1 (n ) 3); polymorph I, 110.3 J g-1 (n ) 2)]. This result suggests that under the chosen experimental conditions, dehydration leads to a material of lower crystallinity than the form from which it was prepared. This reduced crystallinity following water removal can be discerned in the changes in intensity and width of peaks in, for example, the 15-25° 2θ region in the variable-temperature XRPD plots over the range 30-170 °C, as shown in Figures 3a and 3b. The endotherm for dihydrate III at 160 °C was reproducible over a range of DSC heating rates (5, 10, 20, and 40 °C min-1). Given the temperature at which the endotherm for dihydrate III occurs, it is suggested that under DSC conditions this dihydrate transforms to form III material on dehydration and then undergoes a phase change to form I at ∼160 °C, similar to the thermal characteristics of anhydrous carbamazepine form
Journal of Pharmaceutical Sciences / 1065 Vol. 85, No. 10, October 1996
+
+
Figure 3sXRPD curves of (a) carbamazepine dihydrate III and (b) carbamazepine dihydrate I over the range 30−170 °C.
Figure 5sFTIR spectra of carbamazepine dihydrates I and III. Figure 4sFT-Raman spectra of carbamazepine polymorphs and materials obtained after heating dihydrate III or I: (a) heated carbamazepine dihydrate I; (b) heated carbamazepine dihydrate III; (c) carbamazepine I; and (d) carbamazepine III.
III. On the other hand, after water removal, dihydrate I exhibited only a form I melting endotherm, indicating that this dihydrate transforms to form I material in the DSC pan after water removal. Dihydrate samples heated to 110 °C in the DSC pan and subsequently examined by FT-Raman spectroscopy showed spectra consistent with carbamazepine form III obtained from dihydrate III and carbamazepine form I from dihydrate I.18 These spectra are further evidence that on dehydration in the DSC pan, dihydrates III and I transform to different forms (Figure 4). This behavior is similar in the case of dihydrate III to that reported for the dehydration of a carbamazepine dihydrate in elevated-pressure DSC, although behavior similar to dihydrate I was not observed in elevated-pressure DSC.14 Dehydration behavior for dihydrates III and I was also evaluated by HSM. At ∼50 °C, both dihydrates darkened in 1066 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
color; this change was associated with loss of water of crystallization. On further heating, minute crystals appeared to grow from the surface of the formerly dihydrate needles. No further changes were exhibited until the needles melted at 190 °C. Thus, no differences were thus seen between the two dihydrates with HSM. Major differences were not discernible on comparison of various spectra run on dihydrates III and I. The 13C-SSNMR spectra for both hydrates were identical, although the spectrum for dihydrate I was better resolved. This better resolution probably reflects a difference in crystallinity between the samples. The DRIFT spectra for both hydrates were very similar and consistent with those in the literature.19 There was a small additional peak in the spectrum of dihydrate I at 3486 cm-1 that is present in the spectrum of carbamazepine form I and associated with -NH2 stretching (Figure 5). Whereas this vibration might be unique to a true form I dihydrate, the corresponding vibration in the spectrum of dihydrate III may be masked by the prominent band for -OH stretching from water molecules. Alternatively, this vibration could arise
+
+
Figure 6sFT-Raman spectra of (a) carbamazepine dihydrate I and (b) carbamazepine dihydrate III recorded when heated in ‘exposed’ conditions (see text).
Figure 7sFT-Raman spectra of (a) carbamazepine dihydrate I and (b) carbamazepine dihydrate III recorded when heated in ‘sealed’ conditions (see text).
because of the presence of a small amount of anhydrous form I remaining unconverted in the isolated dihydrate. No other bands specific to anhydrous form I were identifiable in the DRIFT spectrum of dihydrate I. FT-Raman spectroscopy offered a better opportunity to discern differences between the dihydrates. Water is a weak Raman scatterer and hence provides minimal interference in the FT-Raman spectra, unlike in Fourier transform infrared (FTIR) spectra where masking of some bands that might have been diagnostic for a particular polymorphic form was observed. Assignment of the dihydrate to a particular anhydrous polymorphic conformation was not straightforward when comparing FT-Raman spectra because the dihydrate peak positions did not equate with either anhydrous form. The strongest similarities were with carbamazepine form III. The lattice vibration regions in the spectra (50-200 cm-1) were similar for both dihydrates, suggesting that both exist in a very similar lattice (Figure 6).
Conditions used for in-situ variable temperature FT-Raman spectroscopy were adjusted to simulate conditions occurring in the DSC pans. Heating studies performed with the chamber window attached (“sealed”; conditions analogous to the DSC pan) or unattached (“exposed”; conditions analogous to heating with TGA or HSM) were undertaken. Dihydrates III and I were subjected to heating in both environments. The lattice vibration regions of FT-Raman spectra obtained when dihydrates were heated in the “exposed” environment are shown in Figures 6a and 6b. Between ∼60-70 °C, the dihydrate spectra changed and, in both cases, displayed lattice vibration associated with carbamazepine form I that implies loss of hydrate water. No additional changes in these vibrations were observed during further temperature increases up to 170 °C. When the dihydrates were heated in the “sealed” environment, the spectra show that the dihydrates revert to the crystal form from which they were prepared. The lattice
Journal of Pharmaceutical Sciences / 1067 Vol. 85, No. 10, October 1996
+
+
Figure 8sXRPD curves for carbamazepine dihydrates (a) III and (b) I, and carbamazine polymorphs (c) III and (d) I.
vibration regions of the spectra obtained on heating dihydrate I are shown in Figure 7a. At ∼70 °C, the dihydrate had lost water and transformed to carbamazepine form I, and no change was seen on further heating. This result is similar to the observations made when both dihydrates were heated in the exposed conditions. The spectra obtained on heating dihydrate III under “sealed” conditions are shown in Figure 7b. Upon losing water at ∼60 °C, the lattice vibrations associated with form III carbamazepine are evident (notably, the ratio of the peaks at ∼110 and 130 cm-1), although the spectra do not indicate that complete reversion to form III has occurred (compare with Figure 4). As the temperature is raised, these vibrations transform to the form I lattice vibrations at ∼160 °C, which is the expected form III-to-I transition temperature. Variable temperature XRPD data further indicated the nature of transformations of the dihydrates on heating. The diffraction patterns of both dihydrates over the temperature range 30-170 °C are shown in Figures 3a and 3b. Reference patterns for polymorphs I and III are shown in Figure 8. The presence of form I carbamazepine diffraction peaks between 7 and 9.5° 2θ indicates that form I is produced from both dihydrates after dehydration under the conditions employed. However, the diffraction patterns are dissimilar in other regions, especially in the 13-16° 2θ region. Dihydrate I appears to dehydrate to form I material of low crystallinity, as indicated by the poorly resolved low intensity peaks. As the temperature increases, the bands become sharper as form I recrystallization occurs. Upon dehydration, dihydrate III also produces a modification of lower crystallinity. The presence of some form III anhydrous carbamazepine is suggested by positions of peaks in the 13-16° 2θ range. Thus these data support the DSC results that the two dihydrates can produce materials of differing crystalline modifications 1068 / Journal of Pharmaceutical Sciences Vol. 85, No. 10, October 1996
after dehydration. These differences are apparent when samples are heated in the pierced DSC pan or in the FT-Raman in the “sealed” environment, where the liberated hydrate water does not readily escape. Where the liberated hydrate water can readily escape, as in a TGA or HSM experiment, the differences are not detected. The dihydrates produced from form I or form III carbamazepine differ in dehydration properties under certain conditions, and hence might have alternative molecular organization; however, no major differences were detectable using the high resolution techniques described earlier. Whereas 13C SSNMR is one of the more sensitive techniques available to detect differences in the environments of functional groups of a molecule, no significant difference was identified between the spectra of the two dihydrates, suggesting molecular conformation and the association of water molecules with carbamazepine molecules are the same in both dihydrates. The XRPD patterns were also similar (Figure 8), implying that the unit cells are equivalent in the two dihydrates, although features of carbamazepine III were discernible in patterns collected during the dehydration of dihydrate III. Therefore, from the available analyses, it seems unlikely that distinct polymorphic dihydrates were formed on hydration of carbamazepine forms I and III. An explanation of the dehydration phenomena observed may be that small amounts of the anhydrous polymorphic form remain after formation of the dihydrate. Such low levels, below the detectional resolution limit of the techniques employed, act in some way during the dehydration process or immediately after it as a template to “seed” the reversion of the carbamazepine to its original polymorphic starting form under conditions where the liberated water is not readily removed. A “memory” in the dihydrate crystals is activated by the water-carbamazepine interaction allowed to occur in the pierced DSC pan or “sealed”
+
+
Improved understanding of the dehydration of carbamazepine dihydrates prepared from different carbamazepine polymorphs is offered in Scheme 2. Although the existence of more than one hydrate polymorph of the same pharmaceutical compound can occur (e.g., amiloride hydrochloride dihydrates20), this occurrence does not appear to be the case for carbamazepine dihydrate.
References and Notes
Scheme 2sProposed relationship between carbamazepine forms III and I and dihydrate.
FT-Raman cell. Where liberated water is readily removed during the dehydration process, as in TGA, HSM, and in the FT-Raman cell in the “exposed” mode, the “seed” is unable to drive the crystallization of the starting polymorphic form. The high temperature metastable form I is favored in these cases, as the poorly crystalline material formed on dehydration recrystallizes under the heating conditions employed. A carbamazepine-water interaction was proposed as an explanation for the behavior of a carbamazepine dihydrate in elevated pressure DSC (where water is not really eliminated from the immediate environment of the drug),14 which is similar to the behavior of carbamazepine dihydrate III in “sealed” FT-Raman conditions seen in the present work. However, the dehydration under pressure of dihydrate prepared from carbamazepine form I was not examined in this previous work. In an environment that allows the carbamazepine-water interaction, this dihydrate only reverts to its starting polymorph and not to polymorph III (the only change seen in previous pressure DSC work14) and also not to the form that is stable at the temperature of the sample immediately after dehydration.
1. Khankari, R. K.; Grant, D. J. W. Thermochimica Acta 1995, 248, 61-79. 2. Garner, W. E. In Chemistry of the Solid State; Garner, W. E., Ed.: Academic: New York, 1955; pp 213-231. 3. Kahela, P.; Aaltonen, R.; Lewing E.; Antilla, M.; Kristofferson, E. Int. J. Pharm. 1983, 14, 103-112. 4. Laine, E.; Tuominen, V.; Ilvessalo, P.; Kahela, P. Int. J. Pharm. 1984, 20, 307-314. 5. Shefter, E.; Higuchi, T. J. Pharm. Sci. 1963, 52, 781-791. 6. Niazi, S. J. Pharm. Sci. 1978, 67, 488-491. 7. Young, W. L. L.; Suryanarayanan, R. J. Pharm. Sci. 1991, 80, 496-500. 8. Kala, H.; Haack, U.; Pollandt, P.; Brezesinski G. Acta. Pharm. Technol. 1986, 32, 72-77. 9. Lefebvre, C.; Guyot-Hermann, A. M.; Draguet-Brughmans, M.; Bouche, R.; Guyot, J. C. Drug. Dev. Ind. Pharm. 1986, 12, 19131927. 10. Reck, G.; Dietz, G. Cryst. Res. Technol. 1986, 21, 1463-1468. 11. Kaneniwa, N.; Ichikawa, J.; Yamaguchi, T.; Hayashi, K.; Watari, N.; Sumi, M. Yakugaku Zasshi 1987, 107, 808-813. 12. Dugue, J.; Ceolin, R.; Rouland, J. C.; Lepage F. Pharm. Acta. Helv. 1991, 66, 307-310. 13. Lovrecich, M.; Orzincolo, O.; Rubessa, F. Acta Tech. Leg. Medic. 1994, 5, 29-48. 14. Han, J.; Suryanarayanan. R. Pharm. Res. 1995, 12 (Suppl.) , S-141. 15. Umeda, T.; Ohnishi, N.; Yokaoyama, T.; Kuroda, K.; Kuroda, T.; Tatsumi, E.; Matsuda, Y. Yakugaku Zasshi 1984, 104, 786792. 16. Behme, R. J.; Brooke D. J. Pharm. Sci. 1991, 80, 986-990. 17. Suryanarayanan, R. Pharm. Res. 1989, 6, 1017-1024. 18. McMahon. L. E.; Williams, A. C.; York, P.; Edwards, H. G. M.; Timmins, P. In Proceedings of the XIVth International Conference on Raman Spectroscopy; Yu, N.-T., Li, X.-Y., Eds.; John Wiley & Sons: Chichester, 1994; pp 908-909. 19. Borka, L.; Lonmo, A.; Wisnes, R. Pharm. Acta. Helv. 1992, 67, 231-233. 20. Jozwiakowski, M. J.; Williams, S. O.; Hathaway R. D. Int. J. Pharm. 1993, 91, 195-207.
JS960117E
Journal of Pharmaceutical Sciences / 1069 Vol. 85, No. 10, October 1996