CRYSTALLINITY CHANGES AND PHASE TRANSITIONS OF SELECTED PHARMACEUTICAL SOLIDS WITH PROCESSING
by MARION W.Y. WONG B.Sc. (Pharm), The University of British Columbia, 1988
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF’ PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Faàulty of Pharmaceutical Sciences Division of Pharmaceutics and Biopharmaceutics
We accept this thesis as conforming to the required standard
THE UNIVERSITY OF BRITISH COLUMBIA November 1993 © Marion W.Y. Wong, 1993 ______
In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives, It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.
(Signature)
Department of
The University of British Columbia Vancouver, Canada
Date jt7’ /
DE-6 (2/88) 11
ABSTRACT
The solid state properties of drugs and pharmaceutical excipients can be significantly affected by processing (e.g. grinding, tabletting, heating and additive incorporation) and reflect structural changes within a solid. Such changes may involve alterations in both the chemical and physical nature of the crystal structure (e.g. hydrates), complete rearrangements of the same chemical components in three-dimensional space (e.g. polymorphs), or more subtle changes which involve neither the chemical composition nor the space lattice. These more subtle changes do not involve phase changes and are referred to as changes in the degree of crystallinity, X. Metronidazole (MTZ), acetylsalicylic acid (ASA), diphenylhydantoin (DPH) and chiorpromazine hydrochloride (CPZ) were selected to illustrate these various changes.
Many of the empirical methods which have been proposed for studying crystallinity were initially used to assess the X of MTZ before and after processing. Although a reduction and subsequent increase in X was indicated, the observed changes could not be adequately explained. The results were inconclusive and a more direct measure of X was necessary. X-ray powder diffractograms reflect the crystal structure, and when used in conjunction with the Rietveld structure refinement method, the processes which cause changes in X (i.e. crystallite size and lattice distortion) can be directly quantified.
Tabletting reduced the peak intensities of ground MTZ and this was accompanied by an increase in the full width at half maximum height (FWHM). Since the unit cell dimensions were not significantly altered, reductions in crystallite size were thought to be primarily responsible for the 111
reduction in the X of MTZ. This was confirmed using the Voigt profile function. Though the Gaussian component was slightly affected (indicating some lattice strain), it was the FWHM of the Lorentzian component of the diffractograms which showed dramatic increases with processing and
subsequent reductions with time at 25°, 54°, 700 and 10000. From the Lorentzian profile, a mean crystallite size for MTZ can be obtained. Tabletting the mechanically ground MTZ further reduced the mean crystallite size. With storage at elevated temperatures, a subsequent increase in crystallite size was observed, where the rate and extent of recovery was dependent on the storage temperature (i.e. recovery at 100°C was greater than recovery at 700, 54° or 25°C). Complete recovery was not observed. The extent to which the peak intensities of ASA were reduced with processing was similar to MTZ, but the underlying structural changes were different. Significant lattice distortion was observed with a 0.5% reduction in the b dimension on tabletting. No significant recovery was found on storage at elevated temperatures. Contrary to previous workers who suggested that the incorporation of DPH with 3-propanoyloxymethyl-5,5-diphenylhydantoin (PMDPH) caused significant “lattice disorder or disruption”, no significant changes in the lattice dimensions were detected. Analysis of bond lengths suggested that the incorporation of PMDPH into the crystal lattice was unlikely. CPZ illustrated a complete change in both the chemical and physical nature of the crystal lattice with processing. Wet granulation completely converted CPZ from a room temperature metastable form to a hemihydrate of the room temperature stable polymorph. Significant differences in the tablettability of each form were shown. iv
TABLE OF CONTENTS
Page
Abstract Table of Contents iv List of Tables viii List of Figures x List of Abbreviations and Symbols xiv Acknowledgements xix
I. INTRODUCTION 1 A. Solids 3 B. Crystallinity 6
C. Methods of Quantitating Crystallinity and Their Limitations 10 1. Density 10 2. Calorimetry 12 3. Nuclear Magnetic Resonance 14 4. Infrared Spectroscopy 15
5. Counting of Dislocation Etch Pits 15 6. Polarized-Light Microscopy 16
7. Water Adsorption 16 8. Kinetics 17 9. Powder X-ray Diffraction 17 V
Page
9.1. The Rietveld Structure Refinement Method 20 9.2. Application of the Rietveld Method to XRPD 21 9.2.1. The Structure Model 21 9.2.2. Data Collection 22 9.2.3. Profile Functions 24 9.3. Determination of Crystallite Size and Lattice Strain 32 D. Effect of Pharmaceutical Processing on Crystallinity 35 E. Trace Additives 40 F. Phase Changes of Pharmaceutical Solids 40 1. Polymorphism 41 1.1. Methods of Characterizing Polymorphs 41 1.2. Polymorphism and Pharmaceutical Processing 43 2. Solvation 43 II. EXPERIMENTAL 44 A. Materials 44 1. Chemicals 44 2. Solvents 45 3. Gases 45 B. Equipment 46 C. Methods 49 1. Suspension Density 49 2. Gas (Helium) Displacement Pycnometry 49 3. Specific Surface Area Measurements 49 4. Scanning Electron Microscopy 50
5. Solid-State Nuclear Magnetic Resonance 50 6. Differential Scanning Calorimetry 51 vi
Page
7. Thermal Microscopy 51 8. Relative Humidity-Composition Diagram 52 9. Solution Calorimetry 53 10. Solubility and Dissolution Rates 53 11. Gas Chromatography 55 12. Grinding 56 13. Tabletting 56 14. Tablet Strength Testing 56 15. X-ray Powder Diffraction 59 III. RESULTS AND DISCUSSION 63 A. Changes in Crystallinity with Pharmaceutical Processing 63 1. Determination of Crystallinity using traditional Methods 63 2. The Rietveld Structure Refinement Method 70 2.1. X-ray Powder Data and the Structural Model 72 2.2. Assessment of Crystallinity Changes 79 2.2.1. Grinding and Tabletting 79 2.2.2. Storage at Elevated Temperatures 86 2.2.3. Incorporation of Additives 96
B. Phase Transitions with Pharmaceutical Processing 100 1. Physical Characterization of Chiorpromazine HC1 and its 101 Granules 1.1. Scanning Electron Microscopy 101 1.2. Powder X-ray Diffraction 101 1.3. Thermal Analysis 105 1.4. Heat of Solution and True Density 108 1.5. Solubility and Dissolution Rate 111 vii
Page 1.6. Relative Humidity-Composition Studies 111 2. Tabletting 116 3. Tablet Strength Testing 122 IV. SUMMARY 126 V. REFERENCES 129 APPENDIX A 148 APPENDIX B 160 APPENDIX C 166 v-Ill
LIST OF TABLES
Page 1. Data collection and details of structure refinement for MTZ, ASA, 62 DPH and PMDPH doped DPH. 2. Comparison of refined cell dimensions of MTZ to literature values. 76 3. Comparison of refined cell dimensions of ASA to literature values. 77 4. Comparison of refined cell dimensions of DPH and DPH doped 78 with PMDPH to literature values for DPH.
5. Changes in the cell dimensions of ASA with processing. 83 6. Indexed X-ray diffraction pattern of DPH. 97 7. Thermal analysis of CPZ(II) and CPZ(I)-H. 107 8. A comparison of the heats of solution and true densities of CPZ(II), 110 CPZ(I), CPZ(I)-H’ and CPZ(I)-H. 9a. Analytical functions used to represent the diffraction profile. 152 9b. Variables used in profile functions. 153 10. Agreement indices for the Rietveld refinement. 156 11. Chemical structure, crystal system, space lattice, and space group 160 of MTZ.
12. Chemical structure, crystal system, space lattice, and space group 161 of ASA.
13. Chemical structure, crystal system, space lattice, and space group 162 of DPH. ix
Page 14. Chemical structure, crystal system, space lattice, and space group 163 of CPZ(I)-H. 15. Chemical structure, crystal system, space lattice, and space group 164 of CPZ(II). x
LIST OF FIGURES
PRg 1. Schematic representation in two dimensions of the structural 4 differences between a crystalline solid and a noncrystalline solid. 2. The relationship between crystalline and noncrystalline solids, 5 and liquids and gases. 3. The arrangement of crystallites within a mosaic crystal. 9 4. Deviation of the peak shape of XRPD from Gaussian behaviour. 26 5. Variations of peak width with Bragg angle. 29 6. Comparison of an asymmetric diffraction peak with a symmetric 31 and asymmetric corrected calculated profile. 7. Schematic diagram of the CT4O mechanical strength tester with 57 modifications. 8. Heat of solution of MTZ: (a) as received, (b) hand ground, and 65 stored at 54°C for (c) 90 and (d) 162 hours. 9. Changes in the intrinsic dissolution rate of MTZ tablets (270 MPa) 67 with storage at 25°C. 10. Heat of fusion of MTZ with grinding using a mechanical ball and 69 mill. 11. Diffractograms of MTZ (a) hand ground and (b) tabletted. 71 12. Diffractogram of MTZ shown with the calculated and difference 73 patterns. xi
Page 13. Diffractogram of ASA shown with the calculated and difference 74 patterns. 14. Diffractogram of DPH shown with the calculated and difference 75 patterns. 15. FWHM of MTZ (a) ground and (b) tabletted shown as a function of 80 20. Barium fluoride was used as the peak width standard.
16. Diffractograms of ASA (a) as received, (b) hand ground, and hand 82 ground and tabletted at (c) 270 and (d) 408 MPa. 17. FWHM of ASA (a) as received, (b) hand ground, and hand ground 84 and tabletted at (c) 270 and (d) 408 MPa shown as a function of 20.
18. FWHM of the Lorentzian component (FL) of MTZ hand ground and 87 tabletted.
19. FWHM of the Gaussian component (FG) of MTZ hand ground and 88 tabletted.
20. FWHM of the Lorentzian component (FL) of MTZ hand ground 89 with storage at 25°C. 21. FWHM of the Lorentzian component (FL) of MTZ hand ground 90 with storage at 54°C.
22. FWHM of the Lorentzian component (FL) of MTZ hand ground 91 with storage at 70°C. 23. FWHM of the Lorentzian component (FL) of MTZ hand ground 92 with storage at 100°C. 24. Changes in the crystallite size of ground MTZ with storage at 25°, 94 54°, 70° and 100°C. xii
Page
25. Changes in the b of ASA as received, hand ground, and tabletted 95 at 270 MPa. Changes in b with storage at 54°C are also shown after 3, 7, and 14 days. 26. Diffractogram of DPH, and DPH with fluorophiogopite. 98 27a. Scanning electron image of CPZ(II). 102 27b. Scanning electron image of CPZ(I)-H’. 103 28. X-ray diffractograms of(a) CPZ(II) and (b) CPZ(I)-H’. The 104 diffractograms of CPZ(I)-H and CPZ(I) are qualitatively the same as CPZ(I)-H’. 29. DSC thermograms of(a) CPZ(II), (b) CPZ(I)-H’ and (c) CPZ(I) 106 using hermetically sealed pans with pinhole. 30. Interconversions of CPZ. 109 31. Comparison of the dissolution rates of CPZ(I) and CPZ(II). 112 32. Relative humidity-composition profile of CPZ(I). 113 33. Changes in the lattice dimensions of CPZ(I) with the incorporation 115 ofwater. 34. X-ray diifractograms of CPZ(I)-H’ treated as follows: (a) dried 117 under vacuum with silica gel at 70°C; (b) hand ground and dried
as in (a); (c) compressed under 210 MPa (the top face of a tablet was scanned and the appearance of two new peaks at 5.3 and 9.5 20 are due to magnesium stearate and talc, respectively); and (d)
heated under vacuum with silica gel at 100°C. 35. Peak offset times of CPZ(II) and CPZ(I)-H’ with increasing 118 compression pressure. XIII
Page 36. Peak offset times of CPZ(I)-H’, CPZ(I)-H and CPZ(I) with 120 increasing compression pressure. 37. Elastic recoveries of CPZ(I), CPZ(II), CPZ(I)-H’ and CPZ(I)-H. 121 38. Force of failure of tablets of CPZ(I)-H’, CPZ(I)-H and CPZ(I). 123 39. Deformation of tablets of CPZ(I)-H’, CPZ(I)-H and CPZ(I). 124 40. Simple monocinic lattice. 165 41. Rectangular (orthorhombic) lattice. 165 42. 400 MHz 13C-NMR spectrum of CPZ(II). 167 43. 400 MHz 13C-NMR spectrum of CPZ(I)-H’. 168 xiv
LIST OF ABBREVIATIONS AND SYMBOLS
a dimension of the crystallographic unit cell along the x-axis A ampere A angstrom ASA 0-acetylsalicylic acid b dimension of the crystallographic unit cell along the y-axis BKPOS background position c dimension of the crystallographic unit cell along the z-axis degrees Celsius C equilibrium solubility heat capacity CPZ chiorpromazine hydrochloride CPZ(I) chiorpromazine hydrochloride, form I CPZ(I)-H’ partially hydrated chiorpromazine hydrochloride, form I CPZ(I)-H fully hydrated chiorpromazine hydrochloride, form I CPZ(II) chiorpromazine hydrochloride, form II Cu copper degree d Durbin-Watson d statistic DBW D.B.Wiles DSC differential scanning calorimetry esd estimated standard deviation xv
F Young’s modulus proportionality constant of Hooke’s Law at a given porosity FWHM full width at half height, Hk g gram pg microgram G Gaussian FWHM of the Gaussian peak F’WHM of the Lorentzian peak
AG Gibbs free energy
GC gas chromatography GofF Goodness-of-Fit GSAS Generalized Crystal Structure Analysis System mixing parameter h Planck constant H enthalpy AHf enthalpy of fusion
Elk full width at half maximum (height), FWHM AH enthalpy of solution Hz hertz IL Intermediate Lorentzian J joule k Boltzmann constant K kelvin kg kilogram
kob observed rate constant microlitre mL milliliter xvi
L Lorentzian LVDT linear variable differential transducer m metre m2 square metre urn micrometre mA milliampere mg milligram MHZ megahertz mm minute ml millilitre ML Modified Lorentzian MPa megapascal Mod-TCH Modified Thompson-Cox-Hastings pseudo-Voigt pV mol mole MTZ metronidazole N newton N’ number of steps NBS National Bureau of Standards Ni nickel NMR nuclear magnetic resonance Pa pascal Poly Edgeworth series ppm parts per million pV Pseudo-Voigt PVII Pearson VII R gas constant xvii
RB Bragg intensity R-factor Rexp expected R-factor RF structure amplitudes R-factor R pattern R-factor weighted R-pattern RH relative humidity rpm revolutions per minute 0 theta tris tris(hydroxymethyl)-aminomethane s second microsecond S entropy ASf entropy of fusion entropy of solution SA specific surface area SC solution calorimetry SEM scanning electron microscopy SRM standard reference material T counting time USGS United States Geological Society USP United States Pharmacopoeia UV ultraviolet V volt kV kilovolt VIS visible V Voigt X degree of crystallinity xviii
XRPD X-ray powder diffraction
Yib background intensity at step i Yic calculated intensity at step i yj0 observed intensity at step i Z number of molecules in a crystallographic unit cell xix
ACKNOWLEDGMENTS
It gives me great pleasure to acknowledge the guidance and support of my supervisor, Dr. Alan Mitchell. I am also very grateful to the members of my research committee, Dr. Frank Abbott, Dr. Helen Burt, Dr. Lee Groat and Dr. Mati Raudsepp, for their invaluable insight. Special thanks is extended to Dr. C. Fyfe for the solid-state NMR analysis of chiorpromazine hydrochloride and its granules, Dr. L. Groat for his assistance in the refinement of the unit-cell dimensions of chlorpromazine hydrochloride, Dr. M. Raudsepp for his assistance in the Rietveld analysis, and Dr. W. Riggs and Dr. A. Szeitz for the gas chromatographic analysis of ethanol. The technical assistance of Randall Oates and Sarvajna Dwivedi during the tabletting studies are gratefully acknowledged. I am also very grateful to my ‘extended family’, Ron Aoyama, Ibrahim El-bagory, John Jackson, Eva Law, and Chuck Winternitz, and my colleagues, Anthony Borel, Ahmad Doroudian, John Gordon, John Kim, Judit Orbay, Sue Panesar, George Tonn, Jing Wang and Matthew Wright, for their encouragement and helpful discussions. The financial support of the Medical Research Council of Canada, Merck Frosst Canada Inc., Stanley Pharmaceuticals and the University of British Columbia is gratefully acknowledged. Finally, I gratefully acknowledge my family and friends for their never failing love and continual prayers. 1
INTRODUCTION
Pharmaceutical processing can lead to marked changes in the solid state properties of drugs and excipients. These changes can include the formation of hydrates, which involve dramatic alterations in both the chemical and physical nature of the crystal structure, polymorphic changes, in which the chemical composition is the same but where the molecules occupy a different space lattice, and lastly, more subtle changes, which involve neither chemical changes nor changes in the space lattice. These latter changes, unlike the formation of hydrates and polymorphs, do not involve phase changes but can still have significant effects on solid state properties and will be referred to as changes in the degree of crystallinity. This thesis presents examples of pharmaceutical compounds which illustrate each of these changes. Particular emphasis is placed on the use of X-ray powder diffraction data to estimate crystallite size and the degree of lattice distortion as measures of structural order.
HYPOTHESES: 1. That pharmaceutical processing can lead to crystal disorder and to phase changes; 2. That, on storage, the crystal structure recovers to a more stable state. 2
OBJECTIVES: 1. To assess changes in the crystallinity and/or solid-solid phase changes of selected pharmaceutical solids as a result of processing; 2. To develop a direct quantitative method of assessing changes in crystal structure.
APPROACH: Phase changes which may arise as a result of pharmaceutical processing are readily identified by differential scanning calorimetry (DSC), solution calorimetry (SC) and X-ray powder diffraction (XRPD), but the quantitation of crystallinity remains a challenge in pharmaceutics. Most of the existing methods are indirect and infer changes in crystallinity from an empirical measurement. To understand changes in the structural order of the crystalline state, the effects of crystallite size were separated from the effects of lattice distortion by applying the Rietveld structure refinement method to XRPD data.
RESEARCH PLAN: The effects of grinding, tabletting and heating on the crystallinity of acetylsalicylic acid (ASA) and metronidazole (MTZ) were studied. Both solids exhibited extensive crystal growth post-compression, which suggested that they may be suitable model compounds for studying a decrease in crystallinity during processing, and a return to order during storage. The effect of additive incorporation on the crystallinity of phenytoin was studied since previous authors have suggested that significant lattice distortion occurs when additives are incorporated. Since pharmaceutical processing can also lead to phase changes, it was necessary to differentiate between changes in crystallinity and phase 3
changes. Chiorpromazine hydrochloride (CPZ) was studied as a model compound showing both polymorphism and solvation on processing.
A. Solids
Solids, by definition, can transmit shear waves or have a minimum viscosity of 1014 poise (1013 Nm2s) (Roy, 1970) and are crystalline or noncrystalline. The crystalline solid is demonstrated by long range three- dimensional order which occurs over a minimum distance of 30-50 A or six unit cells (Kiug and Alexander, 1974a); in a perfect crystal, atoms, ions or molecules are positioned accurately at points throughout an undistorted space lattice (Darwin, 1922; Ewald, 1958). All remaining solids are noncrystalline. Noncrystalline solids obtained from crystalline solids are amorphous, while noncrystalline solids obtained from liquids are glasses. The structural difference between a crystalline solid and a noncrystalline solid is illustrated in Figure 1. The transition from crystalline to noncrystalline occurs when free energy, in excess of that required to maintain a stable arrangement, is incorporated. This may result from a sudden change in a thermodynamic variable such as temperature or pressure. The relationship between crystalline and noncrystalline solids is illustrated in Figure 2 (Roy, 1970). 4
Figure 1. Schematic representation in two dimensions of the structural differences between a crystalline solid (left) and a noncrystalline solid (right). The composition of lattice is A2X3 where the solid circles are the atoms of A and the open circles are the atoms of X.
in glass. J. (Reproduced from Zachariasen, W.H., The atomic arrangement Am. Chem. Soc., 54 (1932) 3845-3846.) ______
5
Non—crystalline solids Metastable
phases
Glasses 2iiuorphized solids 1 + Energy Energy Energy Energy
in frozen added added left
in from by by by Excess shear radi— reac— melt of free ation tion same energy cmposi— tion by cooling
Stable Crystalline solids Liquids
phases
Figure 2. The relationship between crystalline and noncrystalline
solids, and liquids and gases are shown (after Roy, 1970).
(Reproduced from Suryanarayanan, R., Studies on the Crystallinity and Phase Transitions of Calcium Gluceptate. Ph.D. Thesis, the University of
British Columbia, 1985, p. 2.) 6
B. Crystallinity
Experimentally, the division between crystalline and noncrystalline solids is not clear; the term degree of crystallinity (Xe.) is used to describe a state of order intermediate between perfect crystals and noncrystalline solids. Crystallinity has been extensively studied in polymers which in many cases have been shown to exhibit crystalline and noncrystalline properties (Miller, 1966a). Two models have been used to relate the observed physicochemical properties of polymers to their structure: the fringed-micelle model and the folded-chain model (Chung and Scott, 1973).
The common premise in both models is that two distinct states coexist - namely, small perfectly crystalline regions embedded within an amorphous matrix, where the amorphous material is regarded as a distinct ‘state’. The degree of crystallinity is expressed as the weight fraction of crystalline component in a sample (Chung and Scott, 1973; Smith, 1989). The two-state model was adopted by the USP XIX (1975) and is still used to define the crystallinity of pharmaceutical solids (USPXXII/NF XVII, 1990). Solids are classified as either crystalline, noncrystalline or a mixture of the two forms. This definition implies that solids are either perfectly ordered (100% crystalline) or completely disordered (0% crystalline). Most solids, however, lie between these extremes. An alternative concept, the one-state model, was proposed by Suryanarayanan and Mitchell (1985). Unlike the two-state model, the one-state model does not assume a clear distinction between the crystalline and amorphous states. Rather, it gives X a value between 0% (amorphous) and 100% (perfectly crystalline) based upon the state of 7
structural order. An increasing concentration of structural imperfections increases the disorder and causes a decrease in X. Although the one-state model has improved our understanding of the crystallinity of pharmaceutical solids, few methods are available to distinguish between these two concepts. Most X calculations implicitly assume the two-state model (Suryanarayanan and Mitchell, 1985). The complex transition from crystalline to amorphous is also highly simplified and the structural changes which are responsible for this transition remain poorly understood. Solids contain different types and numbers of imperfections which cause regions of misfit and disorder (Boldyrev et al., 1979a). Most obvious is the interruption of periodicity at a crystal face. Other imperfections include surface irregularities and cracks. With respect to the properties of pharmaceutical solids, significant imperfections within the crystal include point defects (due to impurities and additives), dislocations, and grain boundaries. A direct correlation between the concentration of structural defects and X has not been established. Furthermore, even when present at maximum concentrations, crystal defects cannot be solely responsible for removing all long range order and causing a solid to become amorphous (Suryanarayanan, 1985). Crystallite size and lattice strain must also be considered when studying the crystallinity of a solid (Smith, 1989). The terms crystallite size and particle size are sometimes interchanged in the literature but they are not synonymous. A particle can be either a single crystal or an aggregate of crystals, while a crystallite is a single diffracting domain. Atomic planes within a crystal rarely traverse the entire crystal without appreciable distortion or 8
discontinuity. Most crystals are mosaic crystals composed of relatively perfect regions (mosaic blocks) separated by subgrain boundaries (arrays of dislocations) (Figure 3). These small-angle boundaries slightly misalign the mosaic blocks from the neighboring regions. Since the angular misalignment between blocks is only between 2’ to 30’ of arc (0.03° to 0.5°) (Ladd and Palmer, 1977), an aggregate of mosaic blocks can contribute to a single reflection and form a crystallite (refer also to Section I.C.9.). The periodicity within crystals is usually assessed using X-ray diffraction which measures crystallite size and not particle size. Crystalline solids are differentiated from noncrystalline solids by the size of their crystallites and the size limit used is dictated by the practical limits of this method (30-50 A). According to Roy (1970), noncrystalline solids contain crystallites of 10 to 100 A. Crystallites in this size range produce very broad X-ray diffraction peaks and are considered X-ray amorphous. In the present work, both crystallite size and lattice distortion were used to measure changes in X, where crystallite size has been discussed above and lattice distortion refers to changes in the dimensions of the unit cell within the mosaic domains. A decrease in crystallite size and/or an increase in lattice distortion as a result of pharmaceutical processing will lead to a decrease in X. 9
Figure 3. The mosaic structure of a crystal; the angular misalignment between blocks may vary from 2’ to about 30’ of arc (i.e. 0.03°
to 0.5°).
(Reproduced from Ladd, M.F.C. and Palmer, R.A., Structure Determination by X-ray Crystallography, Plenum Press, New York, 1977, p. 347.) 10
C. Methods of Quantitating Crystallinity and Their Limitations
The quantitation of X is an ongoing challenge in pharmaceutics. Many of the methods used originated from crystallinity studies with polymers. The principle methods include density, calorimetry, nuclear magnetic resonance (NMR), infrared spectrometry (IR), and X-ray powder diffraction (XRPD) (Chung and Scott, 1973). In most cases, an indirect measure of X is obtained by monitoring changes in specific properties affected by changes in X. Values of Xc are frequently based on regression analysis and rely heavily upon accurate measurements of perfectly crystalline and amorphous standards (Nakai et al., 1982) - neither of which exists. Therefore, absolute X values vary depending upon the choice of standards (Pikal et al., 1978). The definition oforder or crystallinity also differs depending on the method. Hence, the X values obtained using different methods have only empirical value and rarely agree.
A number of other methods are used, including polarized light microscopy, counting dislocation etch pits, water adsorption and kinetic studies. These techniques and their limitations will also be discussed.
1. Density
Density measurements are an alternative to estimating the state of order within a solid. The density of a crystalline solid is usually higher than its amorphous counterpart since interatomic distances are at their minimum (Suryanarayanan and Mitchell, 1985; Brown et al., 1990). A 11 decrease in lattice order or X increases the volume of a crystal which in turn decreases its density. Several techniques are available for determining the density of solids (Bauer and Lewin, 1972). Most recently, helium pycnometry has become popular in measuring the true density of pharmaceutical solids (with the purpose of assessing X) because it is simple to use, rapid, and nondestructive. Brown et al. (1990) used helium pycnometry to study changes in the X of ibuprofen when crystallized from acetonitrile at different cooling rates and Saleki-Gerhardt et al. (1992) used the same method to assess disorder in sucrose with mechanical milling. Measurement of true density using the suspension density (flotation) method, however, is unique in that samples having very small differences in density can be differentiated (Johnston and Hutchison, 1940), and the two models of crystallinity can be distinguished (Suryanarayanan and Mitchell, 1985). If the simple two-state model is valid, a partial crystalline sample would separate into two fractions when dispersed in the suspending liquid, due to the different densities of the crystalline and amorphous fractions. On the other hand, if the one-state model is valid, progressive changes in X will be accompanied by a gradual change in density.
The effect of temperature on the density of a solid is usually neglected when interpreting suspension density data and this is often unacceptable. Duncan-Hewitt and Grant (1986) found that as the incorporation of oleic acid into adipic acid or p-acetoxyacetanilide into acetaminophen increased, depending on the temperature used, thermal expansivity: (a) decreased with an accompanying increase in crystal density; 12
(b) simply decreased; or (c) was unaffected. The use of thermal expansivity as a more reliable indicator of X was proposed.
2. Calorimetry A thermodynamic definition of order is provided by calorimetric methods. The use of solution calorimetry (SC) is based on the observation that, for many solids, the energy of the amorphous form is higher than the energy of the crystalline form (Pikal et at., 1978; Vanderzee et at., 1981). If the energy difference between the amorphous and crystalline states is large, this method can provide a very accurate assessment of Xc. Care must be taken, though, in the interpretation of these data. Two processes are involved in the dissolution of a solid: the interaction between the solute and the solvent, and bond breaking. The heat measured is the sum of the heat of solvation (exothermic) and the heat absorbed to break up the crystal lattice (endothermic). Only the heat involved in the breaking of bonds is thought to reflect X, but separation of this process from the overall heat measured is difficult. The adsorption of atmospheric moisture (exothermic) during sample preparation presents another serious problem. With pharmaceutical processing (e.g. grinding and tabletting), the ability of solids to adsorb water increases as clean surfaces are created and existing crystal faces are activated. This, in turn, causes the heat involved in bond breaking to be underestimated. Differential scanning calorimetry (DSC) has also been used (York and Grant, 1985; Saleki Gerhardt et at., 1992). The heat of fusion is measured and non-crystalline (amorphous) compounds are characterized by the absence of a sharp 13
melting endotherm. A major disadvantage of this method is that heating can alter the concentration of defects by the process of annealing and may cause decomposition. Therefore, during DSC, the parameters being measured may be constantly changing. Thermodynamics is a powerful quantitative tool in the study of solids (Swalin, 1972). Specific models of a crystal need not be postulated and atomic details of the structure of a material are not necessary. Simply by applying the three laws of thermodynamics with standard mathematical techniques, many macroscopic properties can be obtained. Grant and coworkers extended the principles of thermodynamics to study the X of pharmaceutical solids and developed a number of approaches based on entropy (York and Grant, 1985; Grant and York, 1986a and 1986b; Vachon and Grant, 1987). The problems inherent in crystallinity scales were avoided. The first method quantified lattice disorder resulting from the incorporation of additives or impurities by using a dimensionless disruption index (d.i.). D.i. was defined as the rate of change of the difference between the entropy of the solid and the entropy of the liquid with respect to the ideal entropy of mixing of the components of the solid (York and Grant, 1985; Grant and York, 1986a). A more rigorous treatment of d.i. was presented later by Pikal and Grant (1987). The second more generally applicable method was the ‘entropy of processing (or imperfection)’, ASP, which assesses the contribution of lattice imperfections to the disorder of a solid by measuring the difference between the entropy of the sample and the entropy of the same amount of a reference material (Grant and York, 1986b). Measurements of either the heat of fusion and melting point or the heat of solution and dissolution rate were used to derive d.i., while determinations of the heat of fusion or heat 14
of solution with studies of solubility were used to calculate ASP. In a third approach, Vachon and Grant (1987) formulated the enthalpy-entropy compensation model to describe the complex interplay between pharmaceutical processing and inherent particle parameters. The concomitant energizing and disordering which occur imply an increase in enthalpy (AH) and entropy (AS) with the extent of such increases
dependent on the nature and intensity of the treatment. The small change (ö) in AH and AS can be related by the compensation principle,
o(AH)=f3.o(AS), where I is the proportionality constant and is termed the ‘compensation temperature’. Although these approaches are sound in theory, the power of thermodynamics becomes its disadvantage. Experimentally, the application of thermodynamics by itself is too general and simplistic (Swalin, 1972). No information is obtained about the detailed relationship among atoms or defects in crystals, and no mechanistic explanation is provided for the observed changes in X.
3. Nuclear Magnetic Resonance Nuclear magnetic resonance (NMR) has also been used to assess X but it measures motion rather than order. Slower-moving protons are used to represent the ‘rigid’ crystalline fraction while the faster-moving protons represent the ‘mobile’ amorphous fraction. Factors such as temperature, molecular weight, and the extent of molecular bonding in solids affect the amount of motion measured (Miller, 1966b). 15
4. Infrared Spectroscony According to Kossler (1967), some JR absorption bands of polymers may appear only when the materials exist in a crystalline state. However, there does not seem to be a predictable relationship between the X of a compound and its JR absorption behavior. Grant and Auburn (1965) observed sharp JR bands with anhydrous ampicillin while ampicillin monohydrate exhibited diffuse bands which were indicative of a low degree of order. Even when a relationship is established (Black and Lovering, 1977; Kamat et at., 1988), changes in X cannot be explained mechanistically.
5. Counting of Dislocation Etch Pits To gain insight into the crystal structure of single crystals, a physical measure of dislocation density was proposed (Burt and Mitchell (1981); Friesen et at. (1981)). Dislocations are sites of localized energy and two-dimensional nucleation is known to occur more rapidly at dislocations than anywhere else on the crystal surface. Treating a cleaved surface with etching solution reveals the sites of emergent dislocations and enables one to count them under a microscope. A direct estimate of the number of dislocations per unit area is obtained. Unfortunately, this method is restricted to large, well-formed crystals with a maximum of about 108 dislocations/cm2. Although the extent of X is dependent on the number and type of lattice imperfections, only dislocations can be quantified using this method. Furthermore, only dislocations on the cleaved surface are measured and a relationship between dislocation density and X has yet to be determined (discussed above). 16
6. Polarized-Light Microscopy Polarized-light microscopy is a qualitative method recognized by the USP XXII/NFXVII (1990) for determining whether or not a pharmaceutical solid is crystalline. Non-cubic crystalline solids are optically anisotropic and will exhibit birefringence (interference colors) and extinction when rotated between crossed-polarizers (Bunn, 1946; USP XXIIINFXVII, 1990). Unstrained amorphous solids, on the other hand, are optically isotropic and are extinct at all orientations between crossed- polarizers. Pikal et al. (1978), Oberholtzer and Brenner (1979), and Osawa et al. (1988) used this method in combination with X-ray powder diffraction to define whether a sample was crystalline or amorphous. Solids which did not show distinct peaks in the XRPD pattern nor birefringence when placed between crossed-polarizers were classified as amorphous.
7. Water Adsorption Huttenrauch (1978) proposed that solids of higher states of disorder possess higher free energies which in turn enhance vapor adsorption. The water adsorption isotherm of sodium chloride (Kontny et al., 1987), for example, confirmed that the extent of adsorption per unit area could be enhanced by hand-grinding. For cephalothin sodium (Pikal et al., 1978, Otsuka and Kaneniwa, 1990) and indomethacin (Imaizumi et aL, 1980), the amount of water adsorbed was linearly related to X. The hygroscopic behavior of cefazolin sodium was used by Osawa et al. (1988) to quantitate the crystallinity of their freeze-dried products. Saleki-Gerhardt et al. (1992) found that for sucrose, water vapor uptake measurements were effective in quantitating even low degrees of disorder. 17
8. Kinetics Changes in X are sometimes reflected in changes in chemical reactivity. Pikal et at. (1978) and Otsuka and Kaneniwa (1990) showed that the chemical stability of cephalothin sodium in the solid state was closely related to X; stability data were used successfully to evaluate X (Pikal et at., 1978). Since the crystalline sample remained stable at 50°C, only the solid-state decomposition of the amorphous counterpart was measured, thereby making this method particularly useful in quantitating crystallinity when a perfectly crystalline standard is not available. Decomposition kinetics was also used by Kitamura et at. (1989) to quantitate reductions in X with grinding when studying its effect on the chemical and color stability of cefixime trihydrate. The crystallinity values obtained were in agreement with values calculated using the internal standard method of XRPD (Otsuka and Kaneniwa, 1983).
9. X-ray Powder Diffraction X-ray powder diffraction (XRPD) is the method most widely used to determine the X of pharmaceutical solids (e.g. Black and Lovering, 1977; Nakai et at., 1977; Imaizumi et at., 1980; Morita and Hirota, 1982; Nakai et at., 1982; Suryanarayanan and Mitchell, 1985; Ryan, 1986; Kitamura et at., 1989; Ashizawa et at., 1990; Egawa et at., 1992; Saleki-Gerhardt et at., 1992) and provides a physical definition of order (Chung and Scott, 1973). The assessment of this structural order is one of the primary objectives of this study. Most of our understanding of the structure of crystalline materials has originated from single crystal X-ray and neutron diffraction studies 18
Most of our understanding of the structure of crystalline materials has originated from single crystal X-ray and neutron diffraction studies (Post and Bish, 1989). In many situations, however, the synthesis of crystals suitable for single-crystal studies is difficult (Stout and Jensen, 1989b) and, more importantly, in pharmaceutics, the analysis of finely crystalline or poorly ordered solids is required (e.g. the analysis of materials as received or after processing).
Traditionally, X-ray powder diffraction (XRPD) has been an important standard tool for the identification and characterization of pharmaceutical solids. Structural information can also be obtained from the position, intensity, and shape of the peaks in the diffractogram (Cullity, 1956a). Peak position provides information about the size and shape of the unit cell, peak intensity provides information about the type and position of the atoms within the unit cell, and peak shape and breadth provide information about the mean crystallite size or the crystallite size distribution, and the nature and extent of lattice imperfections (Kiug and Alexander, 1974c). Until recently, XRPD data were considered to be unsuitable for serious crystal structure studies, primarily because of the problems of peak overlap and the difficulties in measuring accurate Bragg intensities.
A basic requirement of XRPD is a clear separation between the amorphous halo and the crystalline pattern. According to Bragg’s law, the crystalline regions diffract X-rays to give sharp peaks while the amorphous regions scatter X-rays to produce a diffuse halo. The X-ray diffraction pattern of a poorly crystalline solid is a combination of both. Crystallinity indices are based on the ratio of the crystalline diffraction pattern to the amorphous scattering intensity and use either area 19 measurements or some function of peak heights and valleys (Smith, 1989). Absolute values are questionable but relative values of a series of related samples can be used to examine the progression of a reaction. However, little information is provided on the reason for the X changes observed (i.e. crystallite size and lattice distortion). As a solid becomes increasingly more crystalline, an increase in peak height is inherently assumed. However, the peak intensity of near perfect crystals may be less than expected because ofprimary extinction (Woolfson, 1970). When atoms in the crystal lattice diffract, there is no phase difference introduced by path differences from surrounding atoms.
The diffracted wave at any point is it radians behind the unscattered incident ray. Some rays of the incident beam are doubly diffracted when passing through the crystal. The unscattered radiation may be joined by radiation which has been doubly scattered and is out of phase, resulting in destructive interference. The intensity of the primary and diffracted beam is reduced and the total energy of diffraction is less than anticipated. Therefore, mosaic crystals are preferred for X-ray diffraction studies (refer to Section I.B.). Slight misalignments of the mosaic blocks reduce primary extinction, and as a crystal becomes increasingly imperfect, destructive interference becomes negligible. In an ideally imperfect crystal, another extinction process occurs and the intensity of the X-ray beam is attenuated while passing through amorphous material. This is known as secondary extinction. Some of the X-ray energy from the incident beam is removed and converted to thermal energy which contributes to the diffracted beam. Various methods for measuring crystallinity have been proposed (Matthews, Peiser & Richards, 1949; Ruland, 1961; Weidinger and 20
Hermans, 1961; Challa et at., 1962). Many empirical rules, correction factors, and/or abstract functions are required, making these methods inconvenient for routine analysis. Development of these methods was based on the two-state model of crystallinity, the application of which to pharmaceutical solids is questionable.
9.1. The Rietueld Structure Refinement Method The two methods most widely used to extract structural information from powder diffraction data are: the integrated-peak-intensity fitting method (Young et at., 1977) and the whole-pattern least-squares fitting or Rietveld structure refinement method (Rietveld, 1967 and 1969). In the integrated-peak-intensity method, the integrated intensities of individual Bragg reflections are measured, converted to structure factors, and used to solve or refine structures. This approach readily decomposes diffraction patterns with minimal peak overlap (i.e. relatively simple structures with high symmetry) into their constituent Bragg reflections. However, it quickly becomes inadequate for the patterns from more complex structures of lower symmetry or poorly crystalline materials because of the many Bragg reflections with severe peak overlap (Post and Bish, 1989). The Rietveld method (Rietveld, 1967 and 1969) does not use integrated peak intensities but takes each data point (20 step) as an observation. Thus, it is the preferred method for structural refinement. Structural parameters, background coefficients, and profile parameters are varied in a least-squares procedure to minimize differences between the calculated and observed patterns. Complex overlapping patterns produced by poorly crystalline and/or low symmetry materials can be studied (Post and Bish, 1989; Raudsepp et al., 1990). The amount of 21 information that is extracted from the pattern is also optimized. Very precise and accurate structure determinations are possible for crystals too small for single crystal studies (Post and Bish, 1989; Raudsepp et al., 1990). Twinning (regions where the lattice orientation is changed along a composite plane by homogeneous shear) is a serious problem with single crystal work but only promotes the random orientation of crystallites in the powder (Post and Bish, 1989).
9.2. Application of the Rietueld method to XRPD There are three basic requirements for Rietveld refinements: 9.2.1. a model that closely approximates the actual structure of the material studied; 9.2.2. accurate intensity data collected in a step-scan manner; and, 9.2.3. a model that accurately characterizes peak shape, peak width, and any systematic errors in peak position.
9.2.1. The Structure Model The requirement of the Rietveld method for a starting model that closely approximates the actual crystal structure is also its limitation (Post and Bish, 1989). In principle, structures can on1y be refined and not solved. Ideally, a powder pattern is calculated based upon the single crystal data of the material to be studied. In situations where single crystal information is not available, a partial model can be formulated from the known crystal structure or from a previous refinement of an isomorph or a 22 solid with structural similarities (McCusker et al., 1985; Post and Bish, 1989). Other alternatives include computer modeling procedures (e.g. distance-least-squares method (Meier and Villiger, 1969) or electrostatic energy minimization (e.g. Busing, 1981; Post and Burnham, 1986)) and high-resolution transmission electron microscopy (Bish and Post, 1988).
9.2.2. Data Collection Accurate step-scan intensities are essential for Rietveld refinement. Powder samples which contain randomly oriented crystallites are required so proper sample preparation is critical. Although both instrumental and sample effects influence the quality of the diffraction intensities obtained, preferred orientation is the most important factor to consider (Hubbard and Smith, 1976). Many crystals encountered in pharmaceutics are nonisometric (e.g. needles or plates) and preferred orientation is a serious problem. Inaccurate peak intensities result and the refined structures become inaccurate. Proper sample mounting techniques greatly reduce preferred orientation (Smith and Barrett, 1979). Grinding to very small particle sizes (<15 tm (Kiug and Alexander, 1974d); 0.1-10 tm (Cullity, 1956b) and loosely packing powders into a sample cavity (Kiug and Alexander, 1974e) have been recommended. Several other methods have been described in the literature and include side- or back-loading (Kiug and Alexander, 1974f), dilution with a second phase which is preferably non-diffracting (Otsuka and Kaneniwa, 1983), dusting onto glass fiber-filters (Davis, 1986), and forming spherulites by spray drying (Smith et al., 1979) or liquid phase agglomeration (spherical agglomeration) using small amounts of binder (Calvert and Sirianni, 1980). These techniques do not 23 always ensure random samples so data from several samples prepared by different methods must be compared. In all cases, the mounted sample should be ‘infinitely’ thick to X-rays and large enough to fully contain the X-ray beam at the lowest diffraction angle of interest. Mounting techniques which use grinding and spray-drying to reduce preferred orientation are not suitable for the present work since the method itself will reduce X. A wide range of step intervals and counting times have been used for data collection. Increasing the number of steps and the counting time improves data precisiOn but the experiment becomes time consuming. Since Bragg intensities and not step intensities are fundamental to structure analysis, long counting times are not always necessary for successful refinements. Decreasing the step interval is more efficient than increasing the counting time in improving precision. Hill and Madsen (1984 and 1986) recommend using counting times which produce a few thousand counts for the strongest peaks and step intervals 115-1/2 that of the minimum full-width at half maximum (FWHM). The reduction in peak intensities with increasing 20 should also be considered when formulating a strategy for data collection. This phenomenon is primarily due to the decrease in atomic scattering factors (Kiug and Alexander, 1974h) so longer counting times might be needed in collecting high-angle data. Data should be collected to the highest 20 angle possible. High angle data are especially important in facilitating the refinement of displacement factors. However, the overlap of reflections becomes increasingly more severe with increasing 20, and data collection at higher Bragg angles becomes limited by the practical number of reflections that 24
can be handled and/or resolved. The time required for data analysis increases substantially with the number of reflections and the amount of information that can be extracted decreases with peak overlap.
9.2.3. Profile Functions Accurate modeling of the diffraction profile (i.e. peak shape, peak width, and systematic errors in peak position) is essential. Profile parameters describing peak widths and shape are varied in a least- squares procedure with structure parameters (atom positions, displacement and occupancy factors), scale factor, unit-cell parameters, and background coefficients, until the calculated pattern approximates the observed pattern (Post and Bish, 1989). The observed intensity at a given step is the sum of the contributions from neighboring Bragg reflections and the background. Therefore, an accurate background description is also necessary. The background can arise from several factors including fluorescence and thermal diffuse scattering from the sample, detector noise, disordered and amorphous phases in the sample, incoherent scattering, and scattering of X-rays by air, diffractometer slits, and the sample holder. The most straightforward approach of background modeling is the linear interpolation between selected points which are removed from the Bragg peaks. While this method is adequate for relatively simple patterns with several well-spaced intervals, the inclusion of background coefficients as variables in the refinement is preferred for more complex patterns. Proper peak shape characterization facilitates background fit. The background can be fitted for even complex patterns when the peak shape is known, However, if a pattern is poorly resolved, systematic underestimations of 25 the standard deviations of other parameters occur, particularly with displacement factors. The background function used in this study is discussed in detail in Appendix A. The peaks in a powder diffraction pattern are the integrated intensities distributed by a peak shape function (Post and Bish, 1989). Accurate representation of the peak shape is important in extracting information about crystallite size and lattice distortion. Any inaccuracies in the fit of the profile shape also introduce large errors in the occupancy factors and displacement parameters. Peak intensities are determined by atomic positions and other structure parameters but can be affected by sample effects such as absorption, extinction, and preferred orientation. Kiug and Alexander (1974g) further identified six instrumental effects which are a function of: the geometry of the X-ray source, varying displacements of different portions of the flat specimen surface, axial divergence of the X-ray beam, specimen transparency, effects of receiving slits, and misalignment of the diffractometer. Consequently the peak shape becomes a complex convolution of sample and instrumental effects which vary depending upon the conditions of data collection and the characteristics of the sample. The peak shapes obtained with XRPD deviate from the near Gaussian peaks obtained using powder neutron diffraction (Figure 4) (Rietveld 1967 and 1969) thereby making the choice of profiles a problem. As a result, application of the Rietveld method to XRPD data was seriously delayed (Malmros and Thomas, 1977; Young et aL, 1977; Young, 1980). Complex functions are required to describe peak shape and several options are now available. In the DBW-9006PC, the Gaussian, Lorentzian
(Cauchy) , modified 1 Lorentzian, modified 2 Lorentzian, Edgeworth Series, 26
U, 4— C
D -- 0 U
0 123 124 125 126 127
Figure 4. Deviation of the peak shape of XRPD from Gaussian behavior is illustrated by comparing a measured diffraction peak (•) to a calculated Gaussian peak profile (—).
(Reproduced from Rietveld, H.M., A profile refinement method for nuclear and magnetic structures. J. Appi. Cryst., 2 (1969) 66.) 27
Voigt, pseudo-Voigt, modified Thompson-Cox-Hastings pseudo-Voigt, or Pearson VII can be used. Details are provided in Table 9a in Appendix A. The intermediate Lorentzian, modified Lorentzian, pseudo-Voigt and Pearson VII are most commonly used. Young and Wiles (1982) reported that the combined Gaussian and Lorentzian functions of the pseudo-Voigt and the Pearson VII performed consistently better. Of the two, the pseudo Voigt was recommended because of the simpler calculations and the refineable constants which can be used to interpret the degree of Lorentzian versus Gaussian component. Instrumental effects are described by the Gaussian function (David and Matthewman, 1985) and the sample related effects by the Lorentzian function. The pure Voigt is also a convolution of Gaussian and Lorentzian functions and is the mathematically correct way to combine the various phenomena to form the final peak shape (David, 1986). However, due primarily to its relative complexity, the Voigt function is commonly substituted by the pseudo Voigt. David (1986) developed a method of parameterizing the pseudo Voigt to allow for refinement of separate half-width functions for the Lorentzian and Gaussian components. This improved its functional similarity to that of the Voigt. Once the angular dependencies of the Lorentzian and Gaussian contributions are independently modeled, information on crystallite size and lattice strain can be extracted (Keisser et al., 1983; David and Matthewman, 1985; and Larson and Von Dreele, 1988). Madsen and Hill (1988) determined the crystallite size and strain properties for CaF2 using Si for instrument profile with the Rietveld refinement method and a Voigt profile function. The determination of crystallite size and lattice distortion is discussed below. 28
XRPD profiles can also be modeled by the learned peak-shape function developed by Baerlocher (1986). In this approach, a nonanalytical function, containing symmetric and asymmetric parts determined in numerical form (“learned”) from a single resolved peak in the pattern, is used (Hepp, 1981). The advantages of Baerlocker’s approach are that it is easily calculated, accounts for peak asymmetry, and describes virtually any peak shape (Baerlocher, 1986). The disadvantage is that a completely resolved reflection is the minimum
requirement - a condition that is rarely met. Improvements in the fit between the observed and calculated diffraction profiles occur when the pseudo-Voigt coefficient is allowed to vary as a function of 20 (Hill, 1984). If the angular dependence of peak
shape is neglected, overestimation of the displacement factors can result. With the Rietveld refinement method, peak widths are typically modeled as a quadratic function in tan 0 and describe the full-width at
half maximum height (FWHM), Hk, as a function of the diffraction angle (Caglioti et al., 1958):
H=Utan2O÷VtanO÷W (1)
where U, V, and W are refineable parameters dependent on the instrumental configuration and the profile shape function chosen.
The variation of peak width with Bragg angle is illustrated in Figure 5. However, the peak broadening which results from structural disorder cannot be explained using the peak-width expression of Caglioti et al. (1958). Thompson et at. (1987) modified the pseudo-Voigt profile function 29
I50•
(‘4
‘a
100
50
scattedcig angie (2w)
I 4 4 4 4 4 4 I 4 4 4 4 I 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Figure 5. Variations of peak width with Bragg angle. The measured halfwidths (•) are shown with the calculated curve of Caglioti et at. (1958) (—).
(Reproduced from Rietveld, H.M., A profile refinement method for nuclear and magnetic structures. J. AppL Cryst., 2 (1969) 67.) 30 so that information on Lorentzian strain and crystallite-size broadening could be extracted. Finally, any asymmetry in the profile shape should be corrected. Most of the commonly used profile shape functions are symmetrical about the peak and do not account for the significant asymmetry which can occur as a result of instrumental and sample effects. A comparison of an asymmetric diffraction peak with a symmetric- and asymmetric- corrected calculated profile is shown (Figure 6). The use of the semi-empirical asymmetry correction term primarily compensates for the asymmetry at low angles that is caused by the axial divergence of the X-ray beam. Integrated intensities are unaffected but the apparent peak positions are shifted. Using incident- and diffracted-beam Soller slits during data collection significantly reduces asymmetry on the low-angle side of peaks and improves resolution. Results might be improved if only the higher angle data are used in the refinement. In this work, the Rietveld structural refinement method was used to extract structural information from the powder diffraction data of selected pharmaceutical solids. Each data point was used as an observation and structural parameters (atomic positions, site occupancies, displacement parameters), background coefficients, and profile parameters were varied in a least-squares procedure. This minimized the difference between the complete observed and calculated diffraction patterns. Unit-cell parameters were refined, and Lorentzian strain and crystallite-size broadening were studied. 31
(I, C
0 C)
10 12 scatteriiig angle (209
Figure 6. Comparison of an asymmetric diffraction peak with a symmetric and asymmetric corrected calculated profile. Measured intensities (•) are shown with a symmetric Gaussian curve (---) and an asymmetric curve (—).
(Reproduced from Rietveld, H.M., A profile refinement method for nuclear and magnetic structures. J. Appi. Cryst., 2 (1969) 67.) 32
9.3. Determination of Crystallite Size and Lattice Strain The determination of crystallite size has been of great interest to materials science and has occupied the attention of crystallographers for many years (Howard and Snyder, 1985; Bonetto et al., 1990). A rapid approximation can be obtained using the width of a single diffraction line (Howard and Preston, 1989). The Scherrer equation is a well known example (Cullity, 1956c):
0.9?. (2) BcosOB
where B is the peak width.
The peak width is measured in radians at an intensity equal to half the maximum intensity. This method assumes that the diffraction peaks are triangular in shape, that peak broadening is solely a function of crystallite size (i.e., the sample is strain-free and instrumental effects have been deconvoluted), and that the crystals (e.g. 500 A thick) are sufficiently small such that peak broadening can be easily measured. These conditions are rarely met and other methods were developed for less ideally behaving materials. Rather than using just peak breadth, the interpretation of Warren and Averbach (1950) is based on peak shape analysis in attempts to obtain more information without making any a priori assumptions. The corrected shape is expressed as a cosine Fourier series where a set of A coefficients is determined (Warren-Averbach method):
P20 =KN(Acos22rnh3 +Bsin2rnh), (3) 33
where: A =(cos22thaZfl)AV, and
B,L = —(sin2th3Z,L)AV.
No appreciable shift in peak position is assumed, the sine terms are negligible, and the final equation becomes:
P20 =KNAcos2inh3, (4)
where: h3= 2a3(sinO)/A, and A =(cos2niZfl)AV.
A plot of the A coefficients versus n distinguishes between distortion and crystallite size broadening. Information on distortion is provided by the A coefficients and for both distortion and crystallite-size broadening, A0 is unity:
dA =_J1p(i)dt. (5)
The initial slope of A versus n is:
(ct4 /dn),0 = —N IN = —1/ N3, (6) where N3a is the average length.
The second derivative gives the length distribution. The crystallite size in direction a becomes:
(d2A,1)I(dn)= (1/N)p(n). (7) 34
A rapid assessment of the effects of crystallite size and lattice strain is provided by the Williamson-Hall method (Williamson and Hall, 1953):
fl•cosQl4AasinO (8) A A a A where: A=mean crystallite size normal to diffraction planes
Aa .. — =average lattice distortion normal to diffraction a planes f3=true breadth of the crystalline sample ?=X-ray wavelength O=diffraction angle
.cosO . sinO . A plot of against is linear, and the average strain value A A normal to the diffraction planes and the mean crystallite size normal to the diffraction planes are given by the slope and the reciprocal value of the
fJ•coso . . . intercept, respectively. Though originally introduced as a means A of separating size effects from strain broadening, this method has been superseded by other methods and is now used only as the first step in the ana1ysis of peak breadths (Ziegler, 1978; Langford and Louër, 1986). The inherent asymmetry of XRPD profiles has been one of the principle difficulties limiting this work. The development of the Rietveld refinement method enabled the angular dependencies of the Lorentzian and Gaussian contributions to be independently modeled. This allowed both crystallite size and lattice distortion information to be extracted simultaneously (de Keisser etal., 1983; David and Matthewman, 1985; and Larson and Von Dreele, 1988; Lutterotti and Scardi, 1990). Enzo et at. 35
(1988) and Benedetti et at. (1988) used a truncated exponential convolved with a pseudo-Voigt to model the instrumental function and a second pseudo-Voigt to model the specimen function. De Keijser et at. (1983) proposed two methods to measure size and strain simultaneously with the profile refinement method: direct analysis of breath parameters of the profile shape function, and analysis of the breath parameters of the pseudo-Voigt and the Pearson VII profile shape functions in terms of breath parameters of the Voigt function. The Voigt profile shape function (Table 9a of Appendix A) was preferred. Madsen and Hill (1988) determined the crystallite size and strain properties for CaF2 using Si for instrument profile with the Rietveld refinement method and a Voigt profile function. Modifications of this approach were used to obtain crystallite size and lattice distortion information from pharmaceutical solids.
D. Effect of Pharmaceutical Processing on Crystallinity
During the processing of pharmaceutical solids (such as crystallization, drying, milling, compression, heating and irradiation), defects within the crystal lattice may develop, migrate or disappear (Grant and York, 1986). XRPD, SC, and suspension density were used by Suryanarayanan and Mitchell (1985) to measure the reduction in X which resulted from grinding anhydrous calcium gluceptate. This marked reduction in X was accompanied by an increase in apparent water solubility. Significant changes in the thermal behavior of metoclopramide hydrochloride monohydrate resulted from grinding and compaction (Mitchell, 1985). It was suggested that the increased crystal disorder 36 induced by mechanical stress accelerated dehydration during differential scanning calorimetry and promoted the recrystallization of an anhydrous polymorph. The effect of processing on the X of cephalosporins has been extensively studied using XRPD. Otsuka and Kaneniwa (1983a) followed the reduction in X of cephalexin with grinding and examined the resulting increases in hygroscopicity, dehydration and decomposition (Otsuka and Kaneniwa, 1984). Egawa et at. (1992) found that with grinding, the apparent solubility and dissolution behavior increased as X was reduced. Amorphous cephalexin prepared by freeze-drying showed enhanced hygroscopicity and solubility (Otsuka and Kaneniwa, 1983b), more rapid dehydration (Otsuka and Kaneniwa, 1983c), and a reduced binding energy for water (Kaneniwa and Otsuka, 1984). On tabletting, cephalexin was converted to a less crystalline state; this conversion was accompanied by an increase in the rate of dehydration and decomposition (Kaneniwa et at., 1985). Similar results were obtained upon grinding cefazolin sodium (Osawa et at., 1988), cefixime trihydrate (Kitamura etal., 1989), and cephalothin sodium (Otsuka and Kaneniwa, 1990). Disordering within a solid can be reversed. Using SC, Vanderzee et at. (1981) demonstrated that perturbations in an organic system, tris(hydroxymethyl)-aminomethane (tris), resulting from milling can be subsequently removed upon storage at elevated temperatures. Significant amounts of energy (40 to 90 J/mol) were stored in the solid as a result of grinding in an agate mortar. Thermal annealing at 3500 to 360°K for 18 hours relieved the strains and returned the solid to its original energy state. 37
Mechanical stress due to milling and compression can be followed by subsequent recrystallization. The recrystallization of X-ray amorphous cellulose following grinding was reported by Hermans and Weidinger (1946a and b). Recent studies report a nucleation phenomena in amorphous sucrose following lyophilization (Scoik and Carstensen, 1990). Post-compressional recrystallization was first reported by Hess (1978) and established as a general phenomenon by Mitchell and Down (1984). Crystal growth on the surface and in the void spaces of compacts was observed with aspirin (ASA), anhydrous calcium gluceptate, ferrous sulfate monohydrate, metoclopramide hydrochloride monohydrate, methenamine, and sucrose after compaction on a single-punch tablet press using both unlubricated and lubricated dies. The materials examined were directly compressible and included examples of solids which undergo plastic deformation (aspirin and methenamine) and brittle fracture (metoclopramide hydrochloride and sucrose) upon consolidation. Surface recrystallization and alterations within the tablet matrix were observed following storage at both 0 and 43%RH (25°C). Crystal growth was observed both on the edge and faces of the compact, but was more pronounced on the edges where a greater frictional contact between the crystals and the die wall exists. The growth rate of aspirin crystals was particularly dramatic (Mitchell and Down, 1984). Crystals were seen within 15 minutes after compression and increased in size and definition with time. The nature and extent of crystal growth was independent of whether lubricant was used and exhibited a solid state Ostwald ripening effect where the development of larger crystals was accompanied by the disappearance of smaller ones. Miller et al. (1986) conducted a more extensive study to 38
examine the nature of crystals growing on the surface of ASA tablets. Crystalline acetylsalicylic acid was compressed using a hand-operated instrumented rotary tablet press. Compressional force, dwell time and compact weight were held constant and no excipients were used. After 5 hours at 30°C and 32%RH, the beginnings of step-like ridges were observed on the surface of aspirin tablets. These ridges grew more pronounced on storage and appeared to result from crystal growth along displaced slip planes. Hexagonal-, prism- and plate-shaped crystals also evolved. The undifferentiated mass of plates and discrete hexagons was similar to those obtained by Jamali and Mitchell (1973) after recrystallization from ethanol under unstirred conditions. Miller et al. (1986) postulated that prisms and plates were the result of differences in the development of the various crystal faces, and that post-compressional growth of these habits may depend on which faces of the original crystals were stressed during compaction. However, the habit obtained upon crystal growth following compaction was later found to be independent of the habit of the starting material (Mitchell, unpublished data). These findings coupled with observations that milling causes perturbations in tris (a reduction in X) and that storage at elevated temperatures results in a subsequent removal of the stored energy of crystal deformation (increase in X) (Vanderzee etal., 1981) suggested there was a link between the disordering of the lattice during pharmaceutical processing and the later restoration of order. The return of the solid back to a more ordered state following processing is one of the hypotheses of this research project. It is important to differentiate the post-compressional recrystallization discussed above, from crystal growth involving phase 39 changes (such as changes in the state of hydration or polymorphic changes). For example, carbamazepine exhibits crystal growth following tabletting but this was said to involve phase changes associated with hydration (Stahl, 1980), or polymorphism (DeCrosta and Eckhart, 1989), or recrystallization from solution in molten stearic acid, the tabletting lubricant (Matthews et at., 1989). At high humidities, Otsuka et at. (1978) measured the moisture sorption and volume expansion of amorphous f3- anhydrous lactose tablets and found that this was accompanied by an autocatalytic crystallization to form a-monohydrate lactose. The crystal growth of theophylline monohydrate from tablets containing anhydrous theophylline was shown to be mediated by the presence of water. This growth occurred when hygroscopic materials such as magnesium chloride or potassium acetate were placed in the formulation and the tablets were stored at high relative humidities (Ando et at., 1986). The recrystallization of amorphous penicillin (Matthews et al., 1966) and digoxin (Black and Lovering, 1978) were also mediated by water. Ando et at. (1985) observed needle-shaped crystals on the surface of a-lactose monohydrate and mannitol tablets when hygroscopic materials such as docusate sodium, magnesium chloride, and potassium acetate were added to the formulation. Tablets were stored at high relative humidities (59, 75 and 90%RH) and the extent of recrystallization was found to depend only on the hygroscopicity of the additive and the relative humidity of storage. Using scanning electron microscopy (SEM), Down and McMullen (1985) observed both interparticulate and surface recrystallization in sodium chloride compacts after storage at relative humidities (RH) of 33 to 94% (ambient temperatures). The observed growth appeared to depend upon atmospheric vapor pressure and the 40
crushing strength of these compacts was found to increase from 33%RH to the critical point of deliquescence (76%RH). This was correlated with the recrystallization observed.
E. Trace Additives
The incorporation of trace amounts of intermediates or degradation products during the chemical synthesis of drugs can affect the surface and bulk properties of the solids produced (York, 1983). A number of important pharmaceutical properties were reproducibly altered by incorporating traces of additives (0.45 mol%), with the observed changes attributed to the disordering and disruption of the lattice (Chow et at., 1984; Chow et al., 1985; Chow etaL, 1985a; Go and Grant, 1987; Chow and Hsia, 1991; and Gordon and Chow, 1992). The disordering was indirectly measured using DSC (Gordon and Chow, 1992), and in this study, a more direct method was used to assess the effect of the incorporation of 3-propanoyloxymethyl- 5 ,5-diphenylhydantoin (PMDPH) into 5 ,5-diphenylhydantoin (DPH) crystals.
F. Phase Changes of Pharmaceutical Solids
In the present study, changes in crystallinity are differentiated from phase changes due to polymorphism and solvation (or desolvation). -D-uring—phaimaceutica-I—processing, reductions in X only involve reductions in the periodic arrangement within a crystal, while phase 41
changes involve a complete rearrangement of the constituent atoms, ions or molecules (polymorphism), andlor a change in chemical composition (solvation).
1. Polymornhism
Polymorphism is the ability of a given compound to exist in more than one crystalline form, each chemically identical to the other and which differ only in their three-dimensional arrangement in space (Glasstone, 1946; Findlay, 1951; Leigh, 1990). At a given temperature and pressure (apart from triple points), there is only one stable form. All other forms are metastable.
Polymorphism is prevalent in pharmaceutical solids, particularly with steroids, sulfonamides, barbiturates, and antibiotics, but it is difficult to predict whether a compound will exhibit polymorphism (Kuhnert Brandstatter, 1971). Extensive literature reviews on the polymorphism of pharmaceutical compounds have been provided by Kuhnert-Brandstatter (1971), Haleblian and McCrone (1975), Borka and Haleblian (1990) and Borka (1991).
1.1. Methods of Characterizing Folymorphs
The significant lattice differences among the polymorphs of a given compound are manifested in the very different physical properties observed. Properties such as true density, chemical and physical stability, nuclear magnetic resonance, electrical and thermal conductivity, melting point and heat of fusion, heat of solution, solubility, dissolution rate, JR spectroscopy, and X-ray diffraction are affected (Carstensen, 1973c; Byrn, 1982a; Lindenbaum and McGraw, 1985). Several techniques are available 42
for characterizing the polymorphic behavior of pharmaceutical solids and these include X-ray powder diffraction (XRPD), JR spectroscopy, thermal analysis (thermal microscopy, differential thermal analysis (DTA), differential scanning calorimetry (DSC)), polarized-light (optical) microscopy, electron microscopy, densitometry, and the measurement of solubility and dissolution rate. XRPD and measurements of solubility are particularly important. Information about the crystal structure can be obtained from X-ray powder diffractograms. Since the three-dimensional arrangement of polymorphs differ, their diffractograms are significantly different. Solubility measurements, on the other hand, differentiate the stable form from the metastable forms. At a given temperature, the stable polymorphic form will have a lower solubility. When these two methods are used with thermal analysis, polymorphic systems can be fully characterized.
Polymorphs can be either enantiotropic or monotropic and these two systems are differentiated by the position of the transition point relative to the melting point of the solid. For enantiotropic polymorphs, the transition point lies below the melting point. Each polymorph has a range of stability and is capable of undergoing reversible solid-solid transformation to the other form. At this transition point, the vapor pressure of the two forms is identical. Monotropic polymorphs, on the other hand, have their transition point above the melting point of the stable form. At all temperatures up to the melting point of the stable form, only one form is stable and all other forms are metastable. The transformation from the metastable form to the stable form is irreversible. 43
1.2 Polymorphism and Pharmaceutical Processing Polymorphic transformations have been reported to occur in inorganic and organic substances under pressure (Bridgeman, 1956; Drickamer, 1967). Examples of pharmaceutical solids which undergo polymorphic transformations during grinding and/or tabletting include barbitone (Nogami et al., 1969; Summers et at., 1976), carbamazepine (Lefebvre and Guyot-Hermann, 1986), chioramphenicol (Kaneniwa and Otsuka, 1985; Otsuka and Kaneniwa, 1986 and 1989), chiorpropamide (Otsuka etal., 1989; Matsumoto etal., 1991), indomethacin (Otsuka etal., 1986), phenylbutazone (Ibrahim et at., 1977), and suiphathiazole (Summers et at., 1976).
2. Solvation Solvates are crystals that contain the solvent of crystallization in stoichiometric or nonstoichiometric amounts. When water is the solvent, these solvates are termed hydrates, and when lattice water is subsequently removed, an anhydrate is formed. One of three events can occur when the water of crystallization is removed from an hydrate: 1. the lattice collapses and recrystallizes to a lower hydrate or anhydrate; 2. the lattice collapses without recrystallization; or, 3. the lattice remains structurally identical to that of the hydrate but chemically different. Lattice collapse followed by recrystallization is the most common, while the maintenance of lattice integrity throughout the desolvation process is rare. 44
EXPERIMENTAL
A. Materials
1. Chemicals
0-Acetylsalicylic acid, BDH Chemicals Inc. (Lot 10927115851) Barium fluoride, high purity, BDH.
Calcium sulphate (Drierite®), Aldrich Chemical Company Inc. Chiorpromazine hydrochloride, Rhone Poulenc Pharma Inc. (Lot M-21673)
Chiorpromazine hydrochloride granules, Rhone Poulenc Pharma Inc. (Lot AM97)
5,5-Diphenyihydantoin, synthesized by Gordon and Chow (1992) 5 ,5-Diphenylhydantoin doped with 3-propanoyloxymethyl-5 ,5- diphenyihydantoin, synthesized by Gordon and Chow (1992) Fluorophiogopite mica (synthetic), NBS (SRM 675) Hydraulic oil, Shell Canada Products Ltd. (Tellus Oil 68) Hydrochloric acid, reagent grade, Caledon Laboratories Lithium fluoride extra pure, BDH Chemicals Inc. Magnesium stearate, Mallinckrodt Inc.
Metronidazole, Rhone Poulenc Pharma Inc. (Lot M-2 16 15) Nail polish clear base coat, Cutex 45
clear top coat, Max Factor Oil, certified index of refraction of 1.532, Cargille Laboratory Inc. Potassium chloride, analytical reagent, BDH Chemicals Inc. Silicon, NBS (SRM 640b) Stearic acid, BDH Chemicals Inc. Talc, Cyprus
Tris(hydroxymethyl)-aminomethane, Parr Instrument Co.
2. Solvents Absolute ethanol, StanChem Acetone, BDH Chemicals 1-Bromobutane, BDH Chemicals 1-Butanol, analytical grade, BDH Chemicals Carbon tetrachloride, BDH Chemicals Chloroform, GC grade, BDH Chemicals Dibromomethane, Aldrich Chemicals Ethanol, StanChem Ethylbromide, J.T. Baker Chemicals Methanol, GC grade, BDH Chemicals 2-Methylpropan-2-ol, analytical grade, BDH Chemicals 1-Octanol, analytical grade, BDH Chemicals Water, deionized via a Milli-RO Water System, Millipore Corp. Water, glass distilled
3. Gases Helium, Matheson Gas Products Canada 0.0322 and 0.1380% Krypton in helium, Matheson Gas Products 0.075% Krypton in helium, Linde Union Carbide 46
Nitrogen, ultra pure, Linde Union Carbide
B. Equipment
Apple II Plus personal computer with ADALAB analog-to-digital converter Balances, Mettler (Model AE 163) and Sartorius, and thermal controlled infrared moisture balance, Sartorius Borosilicate glass tubes, Kimax, with polytetrafluoroethylene-lined screw caps Caliper (electronic), NSK Corp. Centrifuge, Damon/IEC Division (Model HN-SII) CT4O tablet strength tester, Engineering Systems linear variable differential transducer, Sangamo DG5 Circulating water bath, Forma Scientific (Model 2376) Densitometer (digital), Paar (Model DMA 45) Differential interference contrast microscope, Nikon (Model R) Differential scanning calorimetry aluminum sample pans (standard and closed), PL Thermal Sciences differential scanning calorimeter, Du Pont Instruments (Model 910) thermal analyzer, Du Pont Instruments (Series 99) Dissolution-testing apparatus fraction collector, Vanderkamp (Model 10) programmable dissolution sequencer, VanKel Corp. (EDS-lO) 47
six-spindle dissolution tester, Vanderkamp (Series 600) temperature regulators for water baths, Julabo Co. (Model P) and Haake (E8) variable speed paddle stirrers, Heidolph Co. (Model RZR-2000) Fisher-Kendall mixer, Fisher Scientific Co. Gas chromatography gas chromatograph, Hewlett-Packard (Model 5840), with a flame ionization detector gas chromatograph terminal, Hewlett-Packard (Model 18850A) 25 m x 0.31 mm Tjltra-2 fused silica column with a bonded phase of 0.52 pm 5% phenyl-methylsilicone 0.1 mL Gastight syringe, Hamilton Co. Helium multipycnometer, Quantachrome Co. Hot plate/stirrer, Corning (Model PC-35 1) and Fisher (Thermomix) Hydraulic press (instrumented) hydraulic press, Fleck Brothers Ltd. (except for the hydraulic pump, J.S. Burns Corp.) linear variable differential transducer, Sangamo DG5 load cell, Sensotec RIIC D-01 0.5-inch flat-faced punch and die assembly of hardened steel (upper punch and adjustable arm, Tool Tech.) Incubator, Fisher Instrumented rotary tablet press, Manesty (Betapress) with IPT punches (0.5-inch flat-faced punches) Isoperibol solution calorimeter, Tronac (Model 458) Mechanical agate ball and mill, Fritsch (Pulverisette) 48
Ovens, Chicago Surgical and Electrical Co. and Fischer (Isotemp Model 350 and Model 500 series) Rotary stirrer, Fischer Scanning electron microscope, Hitachi, Model F-570 400 MHz MSL Solid-state NMR spectrophotometer, Bruker Sonicating water bath, Branson (Model 2200) Surface area analyzer, Quantachrome Corp. (Quantasorb Sorption System) Thermal microscopy camera, Nikon AFX-IIA gas-flow heating/freezing system, Fluid Inc. polarized light cross-nicols microscope, Nikon trendicator and thermocouple (calibrated), U.S.G.S., Fluid Inc., and monitored using a Doric 410A trendicator Thermometer (digital platinum resistance), Guildline (Model 9540) Ultraviolet/visible diode-array spectrophotometer, Hewlett Packard (Model 8452A), with Hewlett Packard 89530 MS-DOS UV/VIS computer software program Vacuum oven, National Appliance Co. WA measuring station temperature-controlled chamber, Rotronic Hygroskop BT temperature-humidity probe, Rotronic High resolution wide angle X-ray diffractometer, Rigaku (D/MAX 2MBX), with scintillation counter 49
C. Methods
1. Suspension Density
Mixtures of carbon tetrachloride with benzene (Blaton et at., 1979) or 1-bromobutane were used as suspending vehicles for determining the true density of MTZ. MTZ, as received, was dispersed in the various suspending mixtures and tightly sealed in borosilicate glass tubes with polytetrafluoroethylene-lined screwcaps. The tubes were then equilibrated in a jacketted beaker containing water. A mixture of 60% ethylene glycol in water was continually circulated from an external thermostatic water bath (± 0.01 °C) to the double-wall of the cell and then through the outer jacket surrounding the oscillating tube of a digital density meter. The temperature was gradually increased or decreased until the dispersed sample was suspended. A detailed description of the apparatus used, the equilibration process, the determination of density and the calibration procedure is given by Suryanarayanan and Mitchell (1985).
2. Gas (Helium) Displacement Pycnometry True densities were determined using a helium multipycnometer. 40-60g of MTZ and CPZ were accurately weighed directly into the large sample cell (149.59 cm3) with tapping. The reference cell (70.95 cm3) was pressurized to 15-17 psi with predried purified helium.
3. Specific Surface Area Measurements A surface area analyzer (Quantachrome Corporation, NY) interfaced to a strip chart recorder was used to determine the specific 50 surface area of CPZ by the multipoint BET method (Lowell, 1973). Gas concentrations of 0.0322%, 0.0750% and 0.1380% krypton (adsorbate) in helium (carrier) were used. Nitrogen was used as the calibration gas. 100 mg samples were accurately weighed in glass capillary sample cells. The heights of the desorption peaks were measured. Measurements were calibrated by injecting a volume (2-30 jiL) of nitrogen into the Kr/He stream with a gas-tight syringe equivalent to within ± 10% of the volume of krypton adsorbed. The peak heights and corresponding volumes of nitrogen were entered into a computer program (Orr, 1983) along with the ambient temperature and atmospheric pressure, allowing the surface area to be calculated by using the BET equation. Each measurement was performed in triplicate.
4. Scanning Electron Microscopy (SEM) The shape of solids was studied visually using scanning electron microscopy (SEM). Samples were sputter-coated with gold under vacuum in an argon atmosphere and examined in a scanning electron microscope using secondary electron imaging with an accelerating velocity of 20 kV.
5. Solid-State Nuclear Magnetic Resonance (NMR) solid-state NMR was performed using a 400 MHz MSL solid-state NMR spectrophotometer with 90° pulses on the proton of 7.5 ps. A contact time of 80 ps was used and up to 3728 scans were performed per sample at a recycle time of 20 seconds. Spectra were processed with a line broadening of 100 Hz. 51
6. Differential Scanning Calorimetry (DSC) Changes in the thermal behavior of selected solids with processing were studied using a 910 Differential Scanning Calorimeter module interfaced to a Du Pont Series 99 Thermal Analyzer, with the thermograms recorded on a chart recorder. 3-5 mg samples were accurately weighed directly into standard aluminum open pans, hermetically sealed pans, and sealed pans with 0.1-0.2 mm pinholes. The encapsulated sample was heated at rates varying between 2° and 20°C per minute under a purified nitrogen atmosphere. Immediately after each endotherm or exotherm, the temperature was held and the sample was reweighed before resuming the scan. The leading edge of each peak was used to determine the transition temperature. Hydrated samples were also cooled to below the freezing point of water using a cooling accessory with liquid nitrogen, and scanned under the conditions described above to differentiate between liquid water and molecular water. Weight loss on drying was confirmed using a thermal controlled infrared moisture balance. Samples were ground manually using an agate mortar and pestle. A minimum of three trials was performed for each set of experimental conditions.
7. Thermal Microscopy The thermal behavior exhibited during DSC analysis was confirmed visually using thermal microscopy. Samples were mounted in an oil-based liquid of a certified index of refraction and centered in the central chamber of a gas-flow heating/freezing system designed by the U.S. Geological Society (U.S.G.S.). Preheated atmospheric air was circulated uniformly and continuously above and beneath the sample. Vertical gradients were 52 negligible and horizontal gradients were low and easily calibrated. Heating was controlled using a calibrated trendicator and thermocouple and monitored with a Doric 410A trendicator. Samples were viewed at 480-fold magnification using a polarized light cross-nicols microscope (Nikon) with camera accessory.
8. Relative Humidity-Composition Diaram The relative humidity-composition phase diagram was constructed for the room temperature stable polymorph of chiorpromazine hydrochloride, CPZ(I). CPZ(I) was obtained by drying CPZ(I)-H’ at 70°C under vacuum, in the presence of silica gel, and stored at 25°C in desiccators over saturated salt solutions selected to give relative humidities (RH) of 15%, 31%, 43%, 52%, 66%, 76% and 90% (National Physical Laboratory, 1958). Once constant weight was established, the powders were sealed in the temperature-controlled chamber of a WA measuring station with a combined temperature-humidity probe (Mitchell, 1984). The %RH of the powders was measured at 25°C. Immediately after a constant humidity reading was recorded, the sample was removed from the chamber and a portion of the bulk was accurately weighed into a sealed pan with pin hole and scanned using DSC. The weight loss after the dehydration endotherm corresponds to the water content of the powder. After equilibration at the various RHs, samples of the CPZ(I) crystals were also subjected to XRPD. The room temperature metastable form, CPZ(II), did not exhibit significant weight changes when stored at the above humidities. No dehydration endotherm or weight loss was observed when scanned using DSC. Both Form I and Form II 53
deliquesced when stored under the ethanol-water saturated atmosphere of the granulation mixture.
9. Solution Calorimetry Heats of solution of MTZ and CPZ were measured using an isoperibol calorimeter. The bath temperature was maintained at 25.000 ±. 0.005°C. 50 mg samples were accurately weighed and encapsulated in sample cells designed by Winnike et at. (1988). All reactions were measured in a calibrated 50 mL capacity fully silvered vacuum flask at 25°C and a stirrer rotating at 600 rpm. Dissolution was rapid and complete with either glass distilled water or an aqueous solution of 0.1 M hydrochloric acid with 0.1% w/v sodium dodecyl sulphate as the solvent. Any temperature changes during the reaction were monitored by a thermistor bridge system. The calorimeter was interfaced to an Apple II Plus computer through an ADALAB analog-digital converter and accompanying signal amplifier to facilitate data collection and analysis. The chart recorder was calibrated with tris(hydroxymethyl) aminomethane and checked using potassium chloride. Heats of solution for potassium chloride were 17.00 ± 0.34 kJ/mol compared with a reported value of 17.22 kJ/mol (Lide, 1991). A minimum of five repetitions were performed.
10. Solubility and Dissolution Rates Equilibrium solubility and dissolution rates of CPZ were measured. Excess amounts of sample were placed in tightly closed glass containers containing various solvents (e.g. 1-octanol, 1-butanol, tertiary butanol, and a mixture of water in tertiary butanol (1:19)) and rotated continuously 54 while submersed in a water bath maintained at 25°C protected from light. Samples were drawn until equilibrium solubility was reached. Each extract was immediately centrifuged and the supernatant removed, diluted, and analyzed using a diode-array UV spectrophotometer. Each experiment was performed with a blank and standard curve (coefficient of determination between 0.97 and 0.99). Each run was done in triplicate. The dissolution behavior of intact flat-faced tablets of MTZ was determined using an automated six-spindle dissolution tester. A paddle stirring rate of 50 rpm was used and temperature regulators maintained the water bath at 25°C. Tablet sides were coated with clear top coat nail polish, and tablet bottoms were blanked off and fixed to the base of the dissolution flasks with clear base coat nail polish. 2.4 mL samples were withdrawn using an automated fraction collector at 0, 10, 15, 20, 25, 30, 45, 60, 90, and 120 minutes immediately following the addition of 900 mL of the dissolution medium (e.g. 0.1 M hydrochloric acid and 0.1% w/v sodium dodecyl sulphate solution). As above, these samples were analyzed using UV spectroscopy. A calibration curve was prepared for each experiment and typical r2 values of 0.9995 were obtained. Six replicates were performed for each sample. The mass dissolved was calculated from the concentration after correcting for the change in volume of the dissolution medium. Freshly degassed glass distilled water was used throughout. For CPZ which severely laminated and capped on compression, dissolution rates were examined using the compression die apparatus of Woodet at. (1965). The threaded inner die wall prevented layers of the laminated andlor capped tablet from flaking during the dissolution study. A constant surface area was maintained, the hydrodynamic conditions 55 were reproducible, and wetting problems due to adsorbed air and/or electrostatic charges on small particles were minimized. In each experiment, 250 mg of solid was accurately weighed into the die cavity and compressed using a manual hydraulic press at 113 MPa with a dwell time of 60 s. The assembly was then mounted vertically in a Fisher motor assembly and rotated at 100 rpm in a jacketed glass beaker containing 250 mL of either distilled water or a mixture of water and tertiary butanol. In studying phase changes of CPZ, aqueous buffered solutions were also used to decrease the dissolution rate and to differentiate between different forms. Experiments were performed at 25°C and 2 mL samples were taken at 0, 5, 10, 15, 20, 25, 30, 40, 50, and 60 s. Appropriate dilutions were made of the collected aliquots to yield a final absorbance reading between 0.2 and 0.8 absorbance units. Each sample was analyzed using TJV spectroscopy as above.
11. Gas Chromatography The ethanol content of chlorpromazine hydrochloride granules as received (CPZ-H’) was measured using a Hewlett-Packard 5840 gas chromatograph (GO) with a flame ionization detector (FID) interfaced to a Hewlett-Packard 18850A GO terminal. A 25 m x 0.31 mm Ultra-2 fused silica colunm was used with 0.52 iim 5% phenylmethylsilicone as the bonded phase. CPZ(I)-H’ was dissolved in chloroform and 2 pL were injected in the split mode using a split ratio of 3:1. The temperatures of the injection port, FID, and oven were 160°C, 250°C and 60°C respectively. Helium was used as the carrier gas and a constant flow rate of 1.0 mL/min was maintained. 56
12. Grinding The grinding of small amounts of powder was performed manually using an agate mortar and pestle while larger samples were ground in a mechanical ball and mill under controlled conditions.
13. Tabletting For studies of crystallinity, tablets of MTZ and ASA were compressed in a 1.29 cm die using an instrumented hydraulic press. A slow ramp rate was used to reach 270 and 408 MPa, with the pressure maintained for 10 s before decompression. Details of the instrumentation and analysis were given by Doroudian (1991). The effects of a phase change of CPZ on tablettability were studied under rurming conditions using an instrumented rotary tablet press. A detailed description of the tablet press is given elsewhere (Oates and Mitchell, 1989, 1990; Dwivedi etal., 1991, 1992). Compression profiles of the various forms of CPZ were obtained with 1.27 cm flat faced IPT punches and a turret time of 1.00 s. 0.5% w/w magnesium stearate (lubricant) and 0.5% w/w talc (glidant) were added to the samples to facilitate tabletting.
14. Tablet Strength Testing The diametral compression test was used to evaluate the mechanical strength of CPZ tablets stored for 24 hours at ambient conditions (23°C and 34%RH). A commercial tablet strength tester (CT4O) was modified to permit accurate measurements of tablet deformation simultaneously with measurements of force (Figure 7). A linear variable 57
crossbar
linear voltage displacement transducer (LVDT)
“fixecr platen reference surface
deflection under load
force movement of (N) upper platen
tablet deformation
Figure 7. Schematic diagram of the CT4O mechanical strength tester with modifications. 58 displacement transducer (LVDT) was added to measure displacement. The crosshead was replaced by a crosshead designed to carry both the upper platen and the LVDT. During the test, the crosshead descended and applied a load to a tablet placed on the lower platen. The LVDT measured the vertical distance travelled by the crosshead from zero force (the top of the tablet) to the force of failure (Ff), the maximum force applied to the tablet immediately prior to diametral failure. The top surface of the CT4O served as the fixed reference point. The force applied to the tablet was measured up to tablet failure by means of the load cell, whose analog output was calibrated against known weights placed on the lower platen. Lower platen deflection was determined by placing a steel ‘tablet’ between the upper and lower platens and then measuring the crosshead displacement as a function of applied force. Since steel was incompressible under the test conditions, the measured displacement was due entirely to deflection of the lower platen. The relationship between applied force and deflection was linear with a value of 1.05 x io cmJN. The force and displacement signals were collected on the computer using the analog-to-. digital converter. Software was written to correct the crosshead displacement for the load cell deflection and to analyze the data for tensile stress and the work of failure. Downward crosshead movement was set at 3.0 x 10 cm/s. The true speed was less than this since the lower platen deflection depended on the applied load. The average speed was calculated from the true change in distance between the upper and lower platens divided by the time to failure. No padding was used on the platens since this increased the errors in the measurement of diametral deformation. 59
15. X-ray Powder Diffraction (XRPD) XRPD was the primary tool in studying the X changes and phase transitions which occurred with MTZ, ASA, DPH and CPZ during and after processing. A high resolution wide-angle X-ray diffractometer was used with a scintillation counter. The D/max-B Bragg-Brentano goniometer was equipped with incident- and diffracted-beam Soller slits, 1/2° divergence and anti-scatter slits, and a 0.15 mm receiving slit. It was operated at a radius of 285 mm. For rapid qualitative analysis, diffractograms were obtained using the continuous scan mode at a scanning rate of 3.00-5.00°28 per minute; for elucidating structural information, a step scan mode was used with a step interval of 0.02 to 0.05°2 8 from 6.00 to 60.00°20 with a counting time at each step of 5 s. Samples which had been received with manufacturer’s certification or processed (i.e. ground, tabletted, heat-treated andlor doped) were back- loaded against a frosted glass slide into standard aluminum sample holders and irradiated with Ni-filtered CuK X-radiation (40 kV, 30 mA) using a take-off angle of 6°. Tabletted material was first scraped from the surface and edges of the tablet using a sieve of 420 i.m mesh size prior to loading. Scans were performed in triplicate and a variety of preparation techniques were compared to ensure that preferred orientation was not a problem. The frosted glass slide helped to promote a more random sample, ensured that a flat surface was obtained level with the top of the holder, and enabled consistent packing from sample to sample. Thus, specimen displacement and transparency errors were also minimized. Occasionally dilution of the sample with a second phase (e.g. lithium fluoride) and/or gentle hand-grinding was also necessary. To obtain accurate intensity 60
data for structure refinement, the already textured surface was finely serrated with a razor blade in a direction parallel to the path of the X-ray beam. This randomized the orientation of anisotropic crystals which will align during filling, while maintaining a relatively flat surface flush with the top of the sample holder. The Rietveld method (Rietveld, 1967 and 1969) was used to extract structural information from the diffraction pattern by implementing the DBW 3.2S program, version 9006PC (Wiles et al., 1988; Sakthivel and Young, 1991). Details of this package are summarized in Appendix A. Peaks were defined as pseudo-Voigts
(8)
where L is the Lorentzian component and G is the Gaussian component.
The percentage of Lorentzian contribution varies linearly as a function of
20 and is described by i’, the mixing parameter, which is refined using variables NA and NB
ij=NA+NB(20) (9)
A polynomial function was used to fit the background and the equation derived by Caglioti et al. (1958) was used to define the angular dependence of the full-width at half-maximum (F’WHM), Hk (Equation 1). Reported single-crystal data for MTZ, ASA, DPH and CPZ were used to provide the starting model. From there, the structures were refined in a least-squares method using the Newton-Raphson algorithm. Refinements 61 were performed in three stages. First, the scale factor, cell parameters, and zero-point were refined while atomic positions, site-occupancies, and isotropic temperature factors for individual atoms were fixed at values previously reported (or estimated based on knowledge of similar organic materials). A background model was selected based on inspection. Second, the cell dimensions were refined. The half-width parameters, W, U, and V were added to the refinement in that order and once convergence was obtained, parameters for peak shape (NA), peak asymmetry, peak shape (NB) and preferred orientation were included. The polynomial function in the background model was re-evaluated. Third, the remaining structural parameters (e.g. atomic positions, site occupancies, and isotropic displacement factors) were added and the background model was readjusted. Refinement of the preferred orientation parameters using the March-Dollase function had no effect on the final results. Final convergence was assumed when the shifts in the parameter refinement were less than 30% of their standard deviations. Information pertinent to data collection MTZ, ASA, DPH and PMDPH doped DPH is provided in Table 1. In addition to the Rietveld structure refinement method (without internal standard), cell dimensions of DPH and CPZ were refined with reference to SRM 675 (synthetic fluorophiogopite mica) and SRM 640b (silicon) using the method of Appleman and Evans (1973). 62
Table 1. Data Collection and Details of Structure Refinement for MTZ, ASA, DPH and PMDPH doped DPH.
MTZ1 ASA DPH PMDPH-DPH
20 scan range (°) 6-60 6-60 6-60 6-60 Step interval (°20) 0.05 0.05 0.05 0.05 Integration time/step (s) 5 5 5 5
Maximum step intensity 12288 11028 4394 4777 (counts)
No. of unique reflections 256 270 199 199 No. of structure parameters 41 44 61 61
No. of experimental 11 10 11 11 parameters
Scale factor x103 1.380 0.832 0.477 0.414 N-P2 1028 1026 1008 1008 R 11.10 10.20 11.69 8.48 R 14.36 13.22 14.93 11.10
RB 4.84 4.07 3.96 3.57 Durbin-Watson d-statistic 1.14 1.01 1.50 1.43
U 0.119 2.072 0.072 0.418 V 0.135 -0.331 0.114 0.316 W -0.006 0.021 0.027 0.027 fixed fixed fixed fixed
Y2 fixed fixed fixed fixed
1 MTZ crystals were too large and acicular. A random sample was not possible. Refinements could only be performed on ground and tabletted samples. The refinement of ground MTZ is reported here. 2 N-P is the no. of observations (steps) - no. of least-squares parameters 63
RESULTS AND DISCUSSION
A. Changes in Crystallinity due to Pharmaceutical Processing
1. Determination of Crystallinity Changes using traditional Methods The X of MTZ was studied before and after processing using suspension density, SC, DSC and XRPD. The density of metronidazole was measured by Blaton et al. (1979) using the suspension density method with carbon tetrachioride and benzene as the flotation vehicle. A density of 1.44 g/cm3 was reported. Attempts to reproduce their experiments in order to measure changes in X with processing were unsuccessful. Dissolution and recrystallization of MTZ was observed and the density values were not reliable beyond the second decimal place. Using varying mixtures of carbon tetrachloride in 1-bromobutane failed to eliminate recrystallization. The closest value to that reported by Blaton et al. (1979) was 1.45 ± 0.01 g/cm3 (n=5). However, because of recrystallization, the suspension density method is questionable for assessing changes in X. Helium pycnometry gave a value of 1.4441 ± 0.0003 g/cm3 (n=10) but the large sample requirements made this method unsuitable for our crystallinity determinations. The heat of solution obtained using SC is the sum of the heat of solvation (exothermic) and the heat absorbed to break up the crystal lattice (endothermic) (refer to Section I.C.2.). Changes in the heat required 64
to break bonds reflect changes in X. For many organic solids, the energy of the amorphous form is higher than that of the crystalline form (Pikal et at., 1978; Vanderzee et at., 1981), and therefore, less additional heat is required for bond breaking. With pharmaceutical processing, the sample
becomes less crystalline (reduction in X) and a decrease in the overall heat of solution is anticipated.
A reduction in the heat of solution was observed when MTZ was ground or tabletted and this was followed by a subsequent increase on
storage (Figure 8). The heat of solution of MTZ as received was 10.5 ± 0.2 kJ/mol (n=5), and when ground by hand with an agate mortar and pestle for 15 minutes, the heat of solution decreased to 9.7 ± 0.1 kJ/mol (n=5). The heat of solution of tabletted MTZ (compressed to 217 MPa) was identical to that of the hand ground material (9.7 ± 0.1 kJ/mol, n=5). Changes in the heat of solution of ground MTZ with time was studied at 54°C. After 90 hours, the heat of solution increased from 9.7 ± 0.1 kJ/mol to 10.1 ± 0.1 kJ/mol (n=5), and to 10.5 ± 0.2 kJ/mol (n=5) after storage for 162 hours. No further change in the heat of solution was observed after storage for 500 hours. A portion of the ground MTZ was also stored at room temperature in a dessicator containing calcium sulphate (Drierite®). After 500 hours of storage, the heat of solution of MTZ recovered to 10.3 ± 0.1 kJ/mol. The various treated samples were tested using single factor ANOVA at cz=0.05 and statistical significance was found (Fo.05(1),3,16 =27.41, P<0.0005). The most significant difference was observed between MTZ as received and ground.
Vanderzee et at. (1981) reported differences of 40 to 90 J/mol in tris(hydroxymethyl)-aminomethane (tris) and suggested that these differences were indicative of the significant amount of energy and 65
H (kJ/mol)
Figure 8. Heat of solution of MTZ: (a) as received, (b) hand ground, and stored at 54°C for (c) 90 and (d) 162 hours. The mean is provided with standard deviations (n=5). Statistically significant differences were shown using single factor
ANOVA at a=O.05 (Fo.05(l),3,16 =27.41, P<0.0005). 66 mechanical strain stored as the result of grinding. The energy difference between the native and ground sample was 800 JImol (i.e. 10 times that of tris) which suggests that substantial structural disordering had occurred (i.e., reductions in crystallite size, increases in the number of crystal defects (e.g. dislocations), andlor increases in lattice strain). Figure 8 also shows that the increase in energy due to processing was removed on storage. This can be attributed to the process of annealing. Annealing is a means by which mechanical strain is removed from processed solids; numerous dislocations are eliminated and others are rearranged into lower-energy configurations (Hayden et al., 1965; Vernon, 1975). This reduction in dislocation density occurs by the mutual annihilation of moving dislocations, and by the running out of dislocations into surfaces, grain boundaries and voids. Dislocations which cannot be removed tend to change their configuration; jogs may disappear and dislocation lines shorten, or similar dislocations may be arranged into subgrain boundaries. The reduction in dislocation density reduces misalignments between mosaic blocks and increases the number of mosaic blocks which contribute to a single reflection. An increase in the crystallite size (an increase in X) results (refer to Section I.B.). The more endothermic heat of solution observed with MTZ on storage indicated an overall increase in X, where the rate and extent of recovery was facilitated by temperature. Complete recovery was achieved after 162 hours at 54°, but 500 hours was required at room temperature. The dissolution rate of MTZ tablets was monitored with time to show whether or not mechanical energy stored within a crystal as a consequence of the compression process can affect this important physical property of a solid. Figure 9 shows the intrinsic dissolution rate of MTZ 67
1.9
1.8- Intrinsic dC/dt (gcm2min1) 1.7-
1.6-
0 10 20 30 Time (days)
Figure 9. Changes in the intrinsic dissolution rate of MTZ tablets (270 MPa) with storage at 25°C. Mean is shown with standard deviations (n=5). Differences were not statistically significant at a=0.05 using single factor ANOVA (Fo.05(l),5,24 =L70, 0. 10 (Fo.05( 1),5,24 = 1.70, 0. l0 (Fo.05(1),10,22 =1.59, 0.10 Heat of Fusion (lcJ/mol) IT tJ T C J 0 5 10 Time (hours) Figure 10. Heat of fusion of MTZ with grinding. Even after 12 hours of grinding with a mechanical ball and mill, no differences in the heats of fusion of MTZ were observed. Heats of fusion are shown with their mean and standard deviation (n=3). Statistical significance was not found using single factor ANOVA at cc=0.05 (Fo.05(l),1o,22 =1.59, 0.10 of the large acicular crystals. The ability to pack MTZ significantly improved with grinding and tabletting. Typical diffractograms of MTZ ground and tabletted are given in Figure 11. Tabletting the ground material reduced the intensity of the 12.25 20 peak (base peak) by 39-40%. Peak intensity ratios measured relative to an internal standard are commonly used to quantitate X (Imaizumi et at., 1980; Otsuka and Kaneniwa, 1983 and 1984; Kaneniwa et at., 1985; Suryanarayanan and Mitchell, 1985; Ryan, 1986; Kamat et al., 1988; Kitamura et at., 1989; and Ashizawa et at., 1990) (refer to Section I.C.9.). Only relative differences can be examined and no explanation is given as to the type of structural modifications involved which are responsible for the X changes observed. 2. The Rietveld Structure Refinement Method The Rietveld structure refinement method was originally developed to refine crystal structures from powder data when single-crystal work was not feasible, but a starting model which is based on published single crystal work is generally required (Rietveld, 1969). The high precision and accuracy of the method lies in its ability to determine the peak positions unambiguously, to correct peak intensities for preferred orientation, and to model peak shapes. This is exemplified by its use in mineralogy for the quantitation of individual components in a complex multiphase system (Raudsepp et at., 1990). Prior to the development of the Rietveld method, the ability to extract structural information from XRPD data was limited by the ability to accurately characterize the shape of the peaks. Although developments are still in progress, a large selection of profile shapes are now available which closely approximate the peak shape. It is from the peak shape that 71 2-THETA Figure 11. Diffractograms of MTZ (a) hand ground and (b) tabletted (270 MPa). The cliffractogram of MTZ as received could not be obtained since the large acicular crystals caused extensive preferred orientation which made proper sample packing impossible. 72 information about crystallite size and lattice distortion can be obtained. To date, application of the Rietveld method has been limited to inorganic solids and the analysis of peak profiles has been restricted to ideal inorganic materials (e.g. aluminum oxide (Hill and Madsen, 1986; Lutterotti and Scardi, 1990; Balasingh et al., 1991), calcium fluoride, lithium fluoride (Delhez et al., 1982), zinc oxide (Langford and Louër, 1986)). Unique features of the present research have been to extend the Rietveld method to organic crystals, and to adapt the analysis of crystallite size and lattice distortion of ideal inorganic solids to nonideal organic crystals. Assessments of X using XRPD need no longer be empirical. The processes that cause changes in X (i.e. crystallite size and lattice distortion) can now be directly quantified. 2.1. X-ray Powder Data and the Structural Model The Rietveld structure refinement method was .used to accurately characterize the profiles of MTZ (hand ground), ASA (as received) and DPH (as received). Typical diffractograms of MTZ (Figure 12), ASA (Figure 13) and DPH (Figure 14) are shown with their calculated and difference patterns. The difference patterns show that close agreement was achieved between the calculated and observed patterns. Therefore, preferred orientation was negligible and the refined structural model could be used to accurately describe the structure of the solid. Calculated cell parameters were in agreement with published single crystal work (Tables 2-4). Any minor discrepancies between the cell parameters obtained from single crystal studies and those calculated from powder data can be attributed to systematic errors or instrumental effects such as ____ 73 10.0K- U) 4J C :3 0 5.OK• O0K 1fL11i I 16 36 46 56 ,- 1 II II II III III 1111 111111 S 11111 II 11111111 RIIERSR Two—Theta (degrees) Figure 12. Observed, calculated and difference X-ray powder diffraction profiles for MTZ (hand ground). The observed data are indicated by dots, and the calculated profile is the continuous line overlying them. The short vertical lines below the pattern represent the positions of all possible Bragg reflections, and the lower curve is the value of sign (A)wA2 at each step, where A is the difference between the observed and calculated intensity and w is the weight applied during least- squares refinement. 74 6.0K -4-’ C D 0 0 ‘ 2.0K- O.OK 11 rut] lilt tltt ZiI -2.OK- I flU UI (U I III UI 11111 IRE (Elf IIfl UIE lI I- Two—Theta (degrees) Figure 13. Observed, calculated and difference X-ray powder diffraction profiles for ASA (as received). The observed data are indicated by dots, and the calculated profile is the continuous line overlying them. The short vertical lines below the pattern represent the positions of all possible Bragg reflections, and the lower curve is the value of sign (A)wA2 at each step, where A is the difference between the observed and calculated intensity and w is the weight applied during least-squares refinement. 75 4.0K C’) 4-, C D 0 U 2.0K ‘S 4-, (a C a) 0.0K —2.0K I I I I III I E III I RE I EERIE III lilt lIRE 1 lIE R ER 1 Two—Theta (degrees) F’igure 14. Observed, calculated and difference X-ray powder diffraction profiles for DPH (as received). The observed data are indicated by dots, and the calculated profile is the continuous line overlying them. The short vertical lines below the pattern represent the positions of all possible Bragg reflections, and the lower curve is the value of sign (A)wA2 at each step, where A is the difference between the observed and calculated intensity and w is the weight applied during least- squares refinement. 76 Table 2. Cell dimensions of MTZ with processing. Cell Dimensions1 a(A) b(A) c(A) 13(°) V(A3) MTZ2 7.034(2) 8.725(3) 12.818(3) 94.51(2) 784.2 MTZ ground34 7.051(6) 8.738(1) 12.839(2) 94.56(8) 788.5 MTZ tabletted3’5 7.055(6) 8.741(1) 12.841(2) 94.54(9) 789.8 1 cell dimensions are provided with their corresponding standard deviations in parentheses 2 from the single crystal work of Blaton et al. (1979) 3 calculated using the Rietveld method 4 ground for 30 minutes using a mechanical ball and mortar mill 5 tabletted at 270 MPa Note: The cell dimensions of MTZ as received could not be refined since the crystals were large and acicular, making the preparation of a proper sample impossible. Extensive preferred orientation could not be avoided. 77 Table 3. Cell dimensions of ASA refined using the Rietveld method. Cell Dimensions1 a(A) b(A) c(A) B(°) V(A3) ASA (Wheatley, 1964) 11.446(13) 6.596(6) 11.388(9) 95.55(3) 855.7 ASA (Kimetal., 1985) 11.430(1) 6.591(1) 11.395(2) 95.68(1) 854.2 ASA (Masaki etaL, 1991) 10.8 6.0 N/A 90 ASA as received2 11.449(3) 6.629(1) 11.429(2) 96.68(1) 861.5 1 cell dimensions are provided with their corresponding standard deviations in parentheses 2 cell dimensions refined using the Rietveld method 78 Table 4. Cell dimensions of DPH and DPH doped with PMDPH. Cell Dimensions1 a(A) b(A) c(A) V(A3) DPH2 6.230(1) 13.581(1) 15.532(2) 1314.2 DPH3 6.239(1) 13.608(2) 15.566(3) 1321.6 DPH4 6.213(8) 13.559(12) 15.550(17) 1310.0 DPH doped with 6.215(6) 13.537(10) 15.507(14) 1304.6 PMDPH4 1 cell dimensions provided with their corresponding standard deviations in parentheses 2 cell dimensions obtained from the single crystal work of Camerman and Camerman (1971) 3 cell dimensions calculated using the Rietveld method cell dimensions corrected by the addition of an internal standard. (The Rietveld-derived cell dimensions (DPH3) were 0.5-1.0% larger and this discrepancy is attributed to systematic errors which are usually of this magnitude (refer to Sections I.C.9.2.3. and III.A.2.1.). The cell dimensions of DPH doped with PMDPH also exhibited this discrepancy.) 79 the X-ray source geometry, displacement of the specimen surface, axial divergence of the X-ray beam, specimen transparency, effects of receiving slits, and diffractometer misalignment (refer to Section I.C.9.2.3.). 2.2. Assessment of Changes in Crystallinity The effects of grinding, compaction and heating on X were studied using MTZ and ASA as model compounds which illustrate the importance of examining both crystallite size and lattice distortion. The effect of additive incorporation on lattice distortion was studied using DPH. 2.2.1. Grinding and Tabletting Changes in the full-width at half-maximum (FWHM) of MTZ with processing is shown in Figure 15. Broadening of the peaks occur when the ground MTZ was tabletted at 270 MPa, suggesting changes in order, and more specifically, changes in crystallite size and/or lattice distortion. As an accurate diffr•actogram of native MTZ could not be obtained, changes in FWHM are shown in comparison to barium fluoride as the peak width standard. The single crystal data of Blaton et at. (1979) was used to refine the powder data of MTZ (Figure 12). Differences between the refined cell parameters of ground and tabletted MTZ (Table 2) were not statistically significant (differences were less than 3 standard deviations) and indicated that distortions of the lattice had not been extensive. The unit cell volumes of ground and tabletted MTZ were virtually identical. Buckling of the atomic planes within the crystal structure or disruption of structural continuity could also cause peak broadening without significantly changing the size of the unit cell. The level of precision to which cell parameters could be determined, however, would be affected 80 0.4 0.3 0.2 FWHM 0.1 0.0• I I I 0 10 20 30 40 2-ThETA Figure 15. FWHM of ground (u) and tabletted (0) MTZ are shown as a function of 20. Annealed barium fluoride (A) was used as the peak width standard. 81 (larger standard deviations would be obtained). The standard deviations of the cell dimensions of ground and tabletted MTZ were identical and indicated that these effects alone were not large enough to cause the peak broadening observed (Figure 15). This suggested that crystallite size may play an important role (crystallite size is defined in Section I.B. and discussed in further detail below, Section III.A.2.2.2.). Slight differences were observed following the grinding of ASA, and tabletting to a peak pressure of 270 MPa reduced the intensity of the base peak to the same extent as that observed with MTZ (i.e. 40%). Typical diffractograms for ASA as received, ground and tabletted are provided in Figure 16. Using the Rietveld method, diffractograms of ASA were fitted to a calculated diffractogram obtained from the single crystal data of Wheatley (1964) and Kim et al. (1985) (Figure 13). Structural parameters were refined. The cell parameters of native and ground ASA were unchanged and no differences were observed between the 270 and 408 MPa tablets. However, there was a reduction in the unit cell volume when tabletted and this was due to a 0.5% reduction in the b dimension of the unit cell, a statistically significant difference of over 3 standard deviations (Table 5). The larger standard deviations obtained for processed ASA compared to those of ASA as received indicated that buckling of the atomic planes and/or degradation of structural order had also occurred. The FWHM of ASA was increased with pharmaceutical processing. Figure 17 shows the full width of the diffraction peaks at half maximum (FWHM) between 0 to 40°20. The diffraction peaks of the native sample were narrow and following grinding, peak broadening was observed (FWHM increased). Tabletting further broadened all peaks and compared 82 a 2-ThETA Figure 16. Typical diffractograms of ASA (a) as received, (b) hand ground for 15 minutes in an agate mortar and pestle, and hand ground (h) and tabletted at (c) 270 MPa and (d) 408 MPa. 83 Table 5. Changes in the cell dimensions of ASA with processing. Cell Dimensions1 a(A) b(A) c(A) B(°) V(A3) ASA (as received) 11.449(3) 6.629(1) 11.429(2) 96.68(1) 861.5 Ground 11.449(3) 6.629(1) 11.406(2) 95.72(1) 861.4 Tabletted (270 MPa) 11.436(3) 6.590(2) 11.407(3) 95.72(1) 855.4 Tabletted (408 MPa) 11.444(3) 6.589(2) 11.417(3) 95.74(1) 856.6 1 cell dimensions were refined using the Rietveld method and are given with their standard deviations in parentheses. 84 0.5 0.4 0.3 FWfIM 0.2 0.1 0.0 0 10 20 30 40 2-ThETA Figure 17. Changes in the FWHM of ASA as received (a), hand ground (a), and tabletted at 270 MPa (0) and408 MPa (.). 85 to the native sample, a two-fold increase in FWHM was observed. No significant difference was seen between the peak intensities or F’WHM of tablets compressed at 270 and 408 MPa. Hence, even though the effect of pharmaceutical processing on the peak intensity of MTZ and ASA was the same, the underlying reasons for the reduction in X were very different. While crystallite size may be important for explaining the X changes of MTZ with processing, the peak broadening of ASA after grinding suggested that the reduction in X may be a result of a decrease in crystallite size and that further increases in FWHM with tabletting reflect both crystallite size reductions and lattice distortion. These observations are supported by the single crystal data of MTZ (Blaton etal., 1979) (Appendix B - Table 11) and ASA (Wheatley, 1964; Kim etal., 1985) (Appendix B - Table 12). Within the unit cell of MTZ, each molecule forms a hydrogen bond with a symmetrically related neighbor [O(12)-H(12)”N(3)’: O(12)-N(3Y=2.816(2) A, H(12)N(3)’=1.98(2) A,and LO(l2)-H(12)•N(3)’=l69(2) A] with the formation of two additional hydrogen bonds between the 0(8) atom of the nitro group and the neighboring H atoms {0(8)”H(10a)=2.47(2) A and O(8)”H(11a)=2.52(2) Al (Blaton et al., 1979) (Appendix B - Table 11). On the other hand, only two hydrogen bonds are formed between the molecules of ASA and involve the carboxylic oxygens [O(1)”O(2)’=2.649(2) A and H0(1)O(2=1.67(3) A] (Kimetal., 1985) (Appendix B - Table 12). The more extensive hydrogen bonding network of MTZ is exemplified by its higher melting point (159°C compared to the melting point of ASA, 135°C (Wheatley, 1964)) and since two portions of the molecule are secured, sufficient molecular movement to cause a noticeable change in the unit cell dimensions is unlikely. The energy of processing is more likely to cause an increase in the 86 concentration of crystal defects, which in turn, reduces crystallite size (as seen above in Section III.A.2.2.1. and discussed in further detail below, Section III.A.2.2.2.). Only the carboxylic oxygens of ASA are involved in hydrogen bonding and the molecules of ASA are dimerized along a center of symmetry. Shear along the ca and ab planes can occur during pressure, affecting crystallite size. The benzene rings are also planar and stacked flat one on top of the other along b axis (Wheatley, 1964). The length of the intermolecular van der Waal forces between the planes can be compressed and may explain the reduction in the b dimension with tabletting. The Voigt function was used to analyze the diffractograms of MTZ and separate crystallite-size effects from strain-broadening. The individual peak full-width at half-maximum of the total peak, Fp, shown above (Figure 15) can be separated into the Lorentzian component, FL, and the Gaussian component, FG, where the reduction in crystallite size with tabletting is reflected in an increase in FL (Figure 18). A small degree of lattice distortion (strain) is also measured (Figure 19). The lack of significant changes in the cell dimensions of MTZ with processing suggests that, unlike ASA, a more detailed analysis of FL is warranted. 2.2.2. Storage at Elevated Temperatures With pharmaceutical processing, the crystallite size of MTZ was affected, and on storage at elevated temperatures, an increase in L was observed (Figures 20-23). Figures 20-23 show a progressive reduction in FL, corresponding to the increase in crystallite size, at all storage temperatures. From FL, a hypothetical value of crystallite size can be calculated. Very equant crystallites of highly idealized behavior are assumed, and therefore, the grain sizes reported in the present study 87 0.075- LorentzianFWHM 0.07- 0.065 - 0.06- 0.055- o-05 -I I I I I I I 0 10 20 30 40 50 60 70 2-theta Figure 18. FWHM of the Lorentzian component (EL) of MTZ mechanically ground (D) and tabletted (<>). 88 0.4- Gaussian FWHM 0.3 - 0.2- 0.1- 0- I I 0 20 40 60 80 2-theta Figure 19. FWHM of the Gaussian component (FG) of MTZ mechanically ground (<>) and tabletted (a). 89 0.065- LorentzianFWHM 0.06- o 0 o::: oe 0 10 20 30 40 50 60 70 2-theta Figure 20. FWHM of the Lorentzian component (rL) of MTZ hand ground with storage at 25°C for 1(0), 2 (A), 7 (s), 10 (+)and 14(0) days. 90 0.06 0 LorentzianFWHM 0 0 0.055- 0 0 0 0 0 0 oJ 0 0.05- 00• 0.045- 0 10 20 30 40 50 60 70 2-theta Figure 21. FWHM of the Lorentzian component (rL) of MTZ hand ground with storage at 54°C for 1(o), 2 (ti), 3 (o), 4 (o), and 10 (,) days. 91 0.055- LorentzianFWHM 0.0525- 0.05• o, o0, 0.0475- 0 0 00 0.045- 0.0425- a i a a a 0 10 20 30 40 50 60 70 2-theta Figure 22. FWHM of the Lorentzian component (FL) of MTZ hand ground with storage at 70°C for 1(0), 3 (o), 4(o), and 10(e) days. 92 0.0525- LorentzianFWHM 0.05- 0 0.0475- 0 0 0.045- 0 0 QO 0 0.0425- 0.C)4— 0 10 20 30 40 50 60 70 2-theta Figure 23. FWHM of the Lorentzian component (FL) of MTZ hand ground with storage at 100°C for 1(0), 2 (tI), 4 (), and 7 () days. Days 10 and 14 were not different from day 7 and are not shown this plot. 93 should not be considered absolute since the existence of these ideal crystallites within the acicular organic crystals studied is unlikely. These values, however, are of importance as relative values in studying changes in X. The FL profiles of Figures 20-23 were used to calculate crystallite size, and to show changes in the crystallite size of MTZ with time (Figure 24). This can be attributed to annealing (discussed in Section III.A.1.), and recrystallization, which involves nucleation (the formation of stable strain-free areas with high-angle boundaries suitable for rapid growth) and growth (the expansion and eventual impingement of the stably nucleated grains) (Vernon, 1975). It should be emphasized that Figure 24 is provided for relative comparisons only. The increase in crystallite size was facilitated by elevated temperatures, but complete recovery of the tabletted material back to the ground starting material was never achieved. This incomplete recovery of order was contrary to the results obtained using SC which indicated that complete recovery of MTZ could be achieved on storage (Figure 8). Since the definition of order varies with the method (e.g. SC measures thermodynamic order while XRPD measures structural order), discrepancies among the X determined using different methods are not unusual (refer to Section 1.0.). The lattice distortion of ASA was monitored following tabletting (Figure 25). Complete recovery of the distortions in the lattice was not observed, and in fact, very little recovery occurred even at elevated temperatures. Shear along atomic planes in addition to the movement of molecules closer to one another may account for the largely irreversible distortion that occurred (refer to section III.A.2.2.1.). 94 2200 2100 Crystallite Size (A) 1700 0 5 10 15 Time (days) Figure 24. Changes in the crystallite size of ground MTZ with storage at 25° (ti), 54° (0), 70° ( 6.66 6.64 6.62- b(A) 1 a ef Figure 25. Changes in the b of ASA (a) as received, (b) hand ground, and (c) tabletted at 270 MPa. Changes in b with storage at54°C are aiso shown after (d) 3, (e) 7, and (f) 14 days. 96 2.2.3. Incorporation of Additives The mechanism by which additives are incorporated into pharmaceutical solids is not clear. For DPH doped with PMDPH, significant “lattice disorder and disruption” (19-times that expected from pure random mixing alone) was suggested by the disruption index of PMDPH (refer to Section I.C.2.) (Gordon and Chow, 1992). Very little X-ray work has been done on DPH and the first step was to accurately index the powder pattern. Published powder diffractograms to date were poorly indexed and exhibited severe preferred orientation, particularly along the (002) plane (Chakrabarti et al., 1978; Gong, 1982; Gordon, 1991). The indexed X-ray diffraction pattern of DPH is provided in Table 6. The Rietveld structure refinement method (without internal standard) was used in conjunction with the method of Appleman and Evans (1973) (with internal standard) to investigate the incorporation of 3- propanoyloymethy1-5,5-diphenylhydantoin (PMDPH) into 5,5 - diphenyihydantoin (phenytoin, DPH) (Figure 26). The sample packing required to overcome the severe preferred orientation of the acicular crystals compromised peak shape information and therefore accurate crystallite size information could not be obtained. Contrary to the findings of other workers, the suggested “considerable disorder and disruption” in the crystal lattice of DPH in the presence of PMDPH was not substantiated. Changes in cell dimensions were not observed suggesting that incorporation of PMDPH into the lattice of DPH was unlikely. These dimensions are compared in Table 4. The Rietveld cell was larger than the cell corrected by the addition of an internal standard, 97 Table 6 - Indexed X-ray powder pattern for DPH 2Ocalc (°) dcalc (A) JJ’calc “Jobs hkl 8.590 10.2855 12.8 17.7 011 11.326 7.8063 66.7 84.1 002 12.968 6.8213 28.4 33.7 020 15.254 5.8038 1.2 2.3 101 16.590 5.3393 100.0 100.0 111 17.264 5.1323 27.1 29.9 022 18.176 4.8768 36.2 42.4 102,013 19.316 4.5915 15.2 26.1 112 20.343 4.3620 55.0 42.7 121,031 22.406 3.9648 78.8 76.3 103,122 22.800 3.8971 13.5 20.7 004 23.183 3.8336 1.7 3.2 113 25.835 3.4458 20.4 10.8 123 26.039 3.4193 11.0 27.4 033 26.141 3.4061 19.8 28.9 040 26.949 3.3058 10.7 10.3 132,104 27.747 3.2125 18.4 20.4 114 29.317 3.0440 1.6 2.7 210,015 29.768 2.9989 5.7 5.9 133 30.025 2.9738 1.6 2.7 124 30.823 2.8986 3.1 2.8 202 31.532 2.8350 2.8 4.1 220,212 32.025 2.7925 1.3 1.7 221,142,105 33.459 2.6760 10.5 8.7 203,051,134 34.895 2.5691 1.7 3.5 035,044 35.334 2.5382 2.4 3.3 231 36.763 2.4427 0.9 1.8 232 37.246 2.41 22 3.6 5.5 053 37.482 2.3975 1.2 3.1 214,106 38.081 2.3612 1.3 1.8 144,116 39.679 2.2697 1.1 2.8 060 39.832 2.2613 1.3 2.7 126 41.405 2.1790 4.8 4.9 062,215 42.066 2.1462 1.1 2.0 234 42.755 2.1132 1.7 1.6 136,161,243 43.084 2.0979 2.5 3.2 154,225,107 43.819 2.0644 1.5 2.6 117,046 44.093 2.0522 3.9 3.8 162,250 45.402 1.9960 3.2 3.6 037,206,312 45.915 1.9749 0.9 1.6 216 46.266 1.9607 1.3 2.3 064 47.095 1.9281 3.0 3.9 322 47.612 1.9084 2.5 2.6 313,253,137 49.453 1.8416 1.0 1.6 118 49.874 1.8270 1.7 1.9 236,261,073 50.217 1.8153 1.0 1.3 314,254,207 98 a b 0 10 20 30 40 50 I I I I I 2-theta fluorophiogopite, Figure 26. Diffractogram of (a) DPH and (b) DPH with cell the internal standard used in the refinement of (1973). parameters using the method of Appleman and Evans standard. The spikes below denote the peaks of the internal 99 and any discrepancies were likely due to systematic errors, which are generally between 0.05 and 0.1% (refer to III.A.2.1.). Structural analysis indicated that the molecules of DPH were extensively hydrogen bonded within the crystal structure, involving the hydrogens bonded to the nitrogen atoms and the two carbonyl oxygen atoms (Camerman and Camerman, 1971) (Appendix B - Table 13). The strength of this interaction is demonstrated by the short H” 0 hydrogen bond distances of 1.92 A (H(1)•”•O(7’)) and 1.98 A (H(3) “0(6’)). This extensive network of hydrogen bonds bestows upon DPH a strong crystal structure which is reflected in its high melting point of 295°C (Philip et at., 1984). The absence of major structural changes suggest that the trace additive may be present in the DPH crystals in the form of a substitutional solid solution. Two types of substitutional solid solutions are possible for organic crystals, namely, true and interblock substitutional solid solutions (Kitaigorodsky, 1984). For the formation of true substitutional solid solution, the additive must at least be able to partially maintain the hydrogen bonding network of DPH. This may be feasible for PMDPH since it retains all the hydrogen bond acceptor sites and one of the two donor sites of DPH. The formation of an interblock solid solution, on the other hand, requires only that there be partial geometrical conformity between the impurity and host molecules, and this may be more likely. In such a solid solution the additive molecules either occupy defect sites within the crystals or the subgrain boundaries of mosaic blocks. Since these occupation sites are rarely the lattice positions, the presence of the additive is unlikely to alter the unit cell parameters of the host crystal. 100 B. Phase Transitions during Pharmaceutical Processing Chiorpromazine hydrochloride (C PZ( II)), a phenothiazine antipsychotic, makes poor tablets. Sticking to the die wall and picking by the punch faces occurred during the compaction of the pure unlubricated drug. Severe lamination and capping were observed at all compression pressures. A nonconventional wet granulation method employed by Rhone Poulenc significantly improves tablettability. In the granulation procedure, CPZ(II) (Appendix B - Table 15) is wetted with an ethanol:water mixture (80.5:22.9 v/v) and dried at 50°C. No binding agent is used. Changes in the X of CPZ(II) during wet granulation were initially thought to be responsible for the intact tablets formed, but Wong and Mitchell (1992) showed that a phase change had occurred. During the wet granulation process, CPZ(II) is completely converted to a stoichiometric hemihydrate, CPZ(I)-H (Appendix B - Table 14), and when dried at 50°C, some of the water of hydration is removed to form a partially dehydrated hydrate, CPZ(I)-H’. Complete dehydration produced a room temperature stable polymorph, CPZ(I). Polymorphs of CPZ have not been reported in the pharmaceutical literature. Dorignac-Calas and Marsau (1972) isolated three different crystal forms of chiorpromazine hydrochloride but only documented the single crystal X-ray data of the high temperature stable form. In this work, the phase change of CPZ(II) during wet granulation was characterized, and differences in the compaction properties of the various forms were assessed. 101 1. Physicochemical Characterization of Chlorpromazine HC1 and its Granules The morphology, X-ray diffractograms, thermal behavior and true densities of CPZ(II) and its granulated forms were significantly different. 1.1. Scanning Electron Microscopy SEM images of CPZ(II) and its granules are provided in Figures 27a and b, respectively. After wet granulation, the aggregates of large needle- shaped CPZ(II) crystals are converted to smaller composite irregularly shaped crystals. The step-like ridges on the crystal faces and the angular edges are replaced with an irregular pattern of indentations and rounded corners. 1.2. X-ray Powder Diffraction The X-ray powder diffractograms of CPZ(II) exhibited major peaks at 8.5, 15.7, 18.8, 22.3, 22.8 and 25.1 20, with minor peaks at 6.3, 10.0, 14.8, 16.0, 16.9, 20.3, 25.7 and 28.1 20 (Figure 28a). The diffractogram of CPZ(I) H’ was completely different. Major peaks of CPZ(I)-H’ were observed at 5.6, 11.3, 16.8, 19.2, 20.6, 22.6, 23.4, 27.1 and 28.2 20 (Figure 28b). Diffractograms of CPZ(I)-H, the fully hydrated granules, and CPZ(I), the dehydrated granules, were qualitatively the same as CPZ(I)-H’. It was apparent that wet granulation with the ethanol-water mixture led to a phase change where the new crystal lattice, CPZ(I)-H’, could take up or lose water molecules without a marked change in the lattice structure. This is discussed in further detail in Section III.B.1.6. below. The peaks of CPZ(II) were narrower and of greater intensity than those of CPZ(I)-H’, CPZ(I)-H and CPZ(I), indicating that CPZ(II) was more I 02 Figure 27a. Scanning electron images of’ CPZ(II) (magnification, X2000). 10:3 Figure 27b. Scanning electron images of CPZ(I)-Ht (magnification, X2000). 104 o b 5 10 15 20 25 30 35 2 THETA Figure 28. X-ray diffractograms of (a) CPZ(II) and (b) CPZ(I)-H’ (shown using the same intensity scale for comparison). The diffractograms of CPZ(I)-H and CPZ(I) are qualitatively the same as CPZ(I)-H’. 105 crystalline. This was confirmed by the sharper, more intense peaks of CPZ(II) obtained from preliminary solid-state NMR studies (Appendix C). 1.3. Thermal Analysis The DSC thermogram of CPZ(II) showed a single melting endotherm at 188-189°C (Figure. 29a), while the thermograms of CPZ(I)-H’ and CPZ(I)-H exhibited two additional endotherms prior to melting - a broad peak due to dehydration and vaporization, and a small endotherm at 132- 134°C due to a solid (CPZ(I)) to solid (CPZ(II)) transition (Figure 29b). These findings are summarized in Table 7. All transitions were verified using thermal microscopy. When heated at a temperature exceeding the dehydration endotherm (70°C), the weight lost from CPZ(I)-H’ and CPZ(I)-H corresponded to 1.90% and 2.47% w/w water, respectively.. The ethanol content of CPZ(I)-H’ was negligible (0.0 144±0.0004% wlv). When CPZ(I)-H’ was completely dried at 70°C under vacuum in the presence of silica gel, the dehydration endotherm disappeared but the second endotherm remained (Figure 29c). No further weight loss was detected after the second endotherm. Heating past the second endotherm followed by immediate cooling to room temperature and reheating produced a thermogram identical to that of CPZ(II) (Figure 29a). The phase change from CPZ(I) to CPZ(II) at elevated temperature was confirmed by XRPD. A diffractogram identical to that of CPZ(II) was obtained when CPZ(I)-H’ was heated in an oven at 150°C for 15 minutes. The solid-solid transition was confirmed using thermal microscopy and suggested enantiotropic polymorphism. By definition, the commercially available form, CPZ(II), is the high temperature stable form 106 a endo I I I I 50 100 150 200 Temperature (°C) Figure 29. DSC thermograms of (a) CPZ(II), (b) CPZ(I)-H’ and (c) CPZ(I) using hermetically sealed pans with pinhole. 107 Table 7. Thermal analysis of CPZ(II) and CPZ(I)-H. Pan Scan Rate Temperature’ Type C/min Peak Proposed Reaction Thermograms of CPZ(7!) A,C2 10 1 CPZ(1I)(s) --> CPZ 188-189 B2 10 1 CPZ(1I)(s) --> CPZ 184-185 Thermograms of CPZt’I)-H A 2-10 1 CPZ.H(s) > CPZ(1)(S) + l/2H2O(g) 2 CPZ(1)(s) > CPZ(II)(s) 132-134 189 3 CPZ(II)5 —> i) 58 B 10 1 CPZH(5) > CPZ(I)() + 112°(g) 2 CPZ(1)(5) --> CPZ(11)(5) 129 3 CPZ(ll) --> CPZ0 187 C 2-10 1 CPIH(s) .“> CPZ(I)( + ‘/2°(g) 50-55 2 CPZ(O(5) --> CPZ(U)(5) 134-137 3 CPZ(fl)5 --> CPZ 187-190 A,C 15,20 1 CPZ.H(s) > CPZ(I)(S) + ‘/2°(g) 58 2 CPZ(1)(s) --> CPZ(ll)(5) 134-135 3 CPZ(II)5 --> CPZ1 185-188 Thermograms ofCPZ(1 AC 10 2 CPZ(I)(5) --> CPZ(1I)(s) 135-137 3 CPZ(II)5 --> CPZ1 187-189 B 10 2 CPZ(I)(s) --> CPZ(II)(5) 134 3 CPZ(I1) --> CPZ(j) 186 ‘Transition temperature measured from theleading edge 2A=standard pan; B=sealed pan; C=sealed pan with pinhole 108 which exists between 134° and 189°C, and is metastable at room temperature. A similar high temperature stable form was reported by Dorignac-Calas and Marsau (1972), existing from 147° to 201°C (Appendix B - Table 15). No preparative details were given. The reverse transition, however, was not observed on cooling to below the transition temperature, and no evidence of conversion of CPZ(II) to CPZ(I) was found during storage under ambient conditions. When the granulation procedure was simulated on a microscope slide and observed under a light microscope, both the presence of liquid and CPZ(II) crystals were found to be necessary for the formation of CPZ(I)-H. It is apparent that while the conversion of CPZ(I) to CPZ(II) occurs at a specific transition temperature on heating, the conversion of CPZ(II) to CPZ(I) only occurs through a hydrate intermediate and not through a solid-solid phase change. CPZ(I), the low temperature stable form, is a dehydrated hydrate in which the hydrate lattice structure does not collapse and recrystallize when the water of crystallization is removed. This unusual phenomenon was also found with cephalexin and cephaloglycin (Pfeiffer et at., 1970), and calcium gluceptate (Suryanarayanan and Mitchell, 1986). Figure 30 summarizes the above. 1.4. Heats of Solution and True Density The heats of solution support the hypothesis that CPZ(II) is a metastable form of chlorpromazine hydrochloride at room temperature and that CPZ(I) is the stable form. The heat of solution for CPZ(I) is higher (more endothermic) than CPZ(II) (Table 8) . Their true densities also differ. When CPZ(I)-H’ is fully hydrated (CPZ(I)-H), its heat of solution and 109 -K’ 70 C RH>53% Conditions (vacuum, (25c) (20-26C, silica gel) 30-40%RH) wet granulation with ethanol :water b.C (vacuum,silica gel) C PZ (I) CPZ(II) cpz(r)-H 80.5:22.9 (by volume) RH>53% (25C) 132-134C Figure 30. Interconversions of CPZ. 110 Table 8. A comparison of the heats of solution and true densities of CPZ(II), CPZ(I), CPZ(I)-H’ and CPZ(I)-H. Heat of Solution True Density Material (kJ mol) (g cm3) CPZ(lI) 28.80 (098)a 1.312 (0.001) CPZ(1) 29.49 (0.24) 1.285 (0.003) CPZ(I)-H’ 34.89 (0.28) 1.299 (0.001) CPZ(I—H 35.89 (0.22) 1.304 (0.004) a mean ± standard deviation (n=5 for solution calorimetry; and, n=6 for true density measurements) 111 true density increase; and, when fuiiy dehydrated CPZ(I), its heat of solution and true density decrease (Table 8). 1.5. Solubility and Dissolution Rate Solubility and dissolution studies were performed on CPZ(I) and CPZ(II) in both aqueous and nonaqueous media but no differences were observed in the apparent solubilities of these two forms due to the rapid conversion of CPZ(II) to CPZ(I)-H. Rapid conversion also occurred during the dissolution studies using both buffered and unbuffered aqueous solutions, and a mixture of water and tertiary butanol. The dissolution rates of CPZ(I) and CPZ(II) were virtually identical (Figure 31). 1.6. Relative Humidity-Composition Studies A relative humidity (RH)-composition phase diagram was constructed to describe the hydration/dehydration behavior of CPZ(I) (Figure 32). A typical adsorptionldesorption isotherm was obtained at RHs between 8 and 53%, and at higher RHs, CPZ(I)-H (a stoichiometric hemihydrate) was formed. CPZ(II) did not form hydrates at any of the RH values tested. The lattice expansion of CPZ(I) with water incorporation was measured using XRPD. Cox et al. (1971) investigated a similar reversible expansion with cromolyn sodium when exposed to water vapor, but their method required the preparation of single crystals for single crystal diffractometry. Published single crystal data was used in this study. Klein and Conrad (1986) reported a detailed structural analysis of a hemihydrate of CPZ recrystallized from aqueous ethanol (Appendix B - 112 0.0030- Concentration 0.0025 - (mglmL) 0QD I] > C 0.0020- C 0.0015- 0.0010• 0 0.0005• 0.0000-hid 0 50 100 150 Time (s) Figure 31. Dissolution rates of CPZ(I) (C) and CPZ(II) (c>) in aqueous solution buffered at pH 11. 113 %RH 100• 80 60 desorption 40 absorption 20 0• 0 0.5 1 1.5 2 2.5 Water Composition (%) Figure 32. Relative humidity-composition profile of CPZ(I). (CPZ(I)-H’, as received, contained 1.90% w/w H20 and CPZ(I)-H contained 2.47% w/w H20 (0.5 mol H20/mol CPZ) 114 Table 14). The single crystal data was converted to a powder pattern using the LAZY PULVERIX program of Yvon et al. (1977). Information about the lattice parameters, the space-group symbol, and the coordinates and chemical symbols of atoms contained in one asymmetric unit were required. The necessary constants from the International Tables for X-ray Crystallography were stored in the program (e.g. scattering factor tables, anomalous dispersion correction terms and X-ray wavelengths). All symmetry information was derived and the multiplicities of special positions were automatically calculated. Using this method, a tabular listing of hkl, d spacing, 20 values, structure factors and intensities was obtained. The calculated powder pattern of Klein and Conrad (1986) was not only identical with that of CPZ(I)-H, but also CPZ(I)-H’ and CPZ(I). Having established that our powder data and the published single crystal data were in agreement, the published lattice parameters were used in the indexing and least-squares refinement method of Appleman et al. (1973). Slight shifts in the peak positions occurred when the amount of water in the lattice was varied. Powder data was collected for a number of CPZ(I) samples whose lattices contained varying amounts of water, and these diffractograms were indexed and refined to elucidate the corresponding changes in the dimensions of the unit cell. The three-dimensional changes in the monoclinic lattice as a function of water content are shown in Figure 33. While the unit cell gradually expands in the a and c direction as more water is incorporated, the longest length of the unit cell (i.e., along b) contracts. Upon formation of CPZ(I)-H (the hemihydrate) at 2.47% H20, expansion in all directions is observed, with the most dramatic increase along b. 115 Change in Lattice Dimensions (Angstroms) 0.2 0 0.1• cjJ 0- -0.3- 2.5 0.5 1 1.5 2 Water Content (%) FIgure 33. Changes in the lattice dimensions of CPZ(I) with the incorporation of water. CPZ(I).H is obtained at 2.47% wlw H20(dotted line). Dimensions in the a (ti), b (0) and c () directions are shown. Samples containing 1.85% w/w H20 were analyzed in triplicate, and the mean and range are provided. 116 Dorignac-Calas and Marsau (1972) isolated a form of chiorpromazine hydrochloride which was metastable at ambient temperatures (Appendix B - Table 15). The unit cell volume was one-half of that reported by Klein and Conrad (1986) for CPZ(I)-H. When the analysis of Yvon et at. (1977) was used to convert the single crystal data of Dorignac-Calas and Marsau (1972) to a powder pattern, the calculated diffractogram agreed with the powder pattern obtained for CPZ(II). XRPD studies also confirmed that the lattices of CPZ(I)-H’ and CPZ(I)-H remained intact after complete dehydration under vacuum in the presence of silica gel at 70°C (Figure 34a). However, at 100°C under vacuum (as above) for prolonged periods of time (i.e., >3 days), complete conversion of CPZ(I)-H’ to CPZ(II) occurred (Figure 34d). Polymorphic transformations have been reported to occur with certain organic substances and pharmaceutical solids when subjected to mechanical stress (Drickamer, 1967; Nogami et at., 1969; Summers et at., 1976; Ibrahim et at., 1977; Lefebvre and Guyot-Hermann, 1986; Otsuka and Kaneniwa, 1985, 1986 and 1989; Otsuka et at., 1986 and 1989; Matsumoto et at., 1991). However, hand grinding in an agate mortar (Figure 34b) or compression up to 210 MPa (Figure 34c) did not affect the crystal lattice of CPZ(I)-W. 2. Tabletting Materials which undergo extensive viscoplastic deformation during compression tend to form good tablets. In our Betapress analysis, peak offset time can be used as an indication of the extent of viscoplastic deformation (Oates and Mitchell, 1989; Dwivedi et at., 1991). The peak offset times of CPZ(II) were slightly shorter than for CPZ(I)-H’ (Figure 35). 117 0 b C d 5 10 15 20 25 30 35 2 THETA (a) dried Figure 34. X-ray difl±actograms of CPZ(I)-H’ treated as follows: under vacuum with silica gel at 70°C; (b) hand ground and of a dried as in (a); (c) compressed under 210 MPa (the top face at tablet was scanned and the appearance of two new peaks 5.3 and 9.5 20 are due to magnesium stearate and talc, gel at respectively); and (d) heated under vacuum with silica 100°C. Diffractograms (a)-(d) are shown using the same intensity scale for direct comparison. 118 Peak Offset Time (ms) 7 6 0 0 0 3. 0 o ° 0 2 0000 000 00 0 I o I I I 0 50 100 150 200 250 Peak Pressure (MPa) Figure 35. Peak offset times of CPZ(II) (0) and CPZ(I)41’ (0) with increasing compression pressures. 119 This difference is unlikely to account for the poor tablettability of CPZ(II) since, compared with other materials, its peak offset times are still relatively long and on this basis alone, CPZ(II) might be expected to form good tablets. Water has been shown to play a determinant role in the tablettability of pharmaceutical solids. Detailed investigations with dextrose (Armstrong and Patel, 1986), -cyc1odextrin (Giordano et al., 1990), maltodextrin (Li and Peck, 1990), and directly compressible dextrose-based diluents (Shukia and Price, 1991), for example, indicate that the extent of particle deformation and interparticulate friction, the strength of interparticulate bonds, and fragmentation during compression may be affected by water content. Crystal bridge formation as a result of dissolution and subsequent recrystallization may also be responsible. The water of hydration of chlorpromazine hydrochloride appears to play a role in the particle deformation mechanism during compression and in interparticulate bond formation. The length of the peak offset times show CPZ(I)-H deforming more during compression than CPZ(I). No significant differences however were observed between CPZ(I)-H (2.47% w/w H20) and CPZ(I)-H’ (1.90% w/w H20) (Figure 36). Successful tabletting also depends on the ability of the bonds within the tablet to withstand elastic recovery during decompression. The in-die tablet recovery was calculated and Young’s modulus, E, was estimated from recovery data (Dwivedi et at., 1992). Representative plots of the proportionality constant of Hooke’s Law at a given porosity, E (expressed in GPa), as a function of tablet porosity are shown in Figure 37. No differences were observed in the elastic 120 Peak Offset Time (ms) 0 6 00 5 Do® 0 ®oc4]®o 0 4 0 o 3 0 0 0 00 o0 0 a 1 Do 0 0 50 100 150 200 250 Peak Pressure (MPa) Figure 36. Peak offset times of CPZ(I)-H’ (CD), CPZ(I)-H (0.) and CPZ(I)(cI) with increasing compression pressures. 121 Ep (GPa) 14 12 0OD 0 10- 80 8 8.D0 0 o 6- 00 j 00 4 0 2 0 I 0 0.05 0.1 0.15 0.2 0.25 0.3 Tablet Porosfty Figure 37. The proportionality constant of Hooke’s Law at a given porosity, E, as a function of tablet porosity. CPZ(I) (ci); CPZ(II)(); CPZ(I)-H’ (0); and CPZ(I)-H (0). 122 recoveries of tablets of CPZ(I), CPZ(II), CPZ(I)-H’ and CPZ(I)-H during decompression. In previous work (Dwivedi et al., 1992), chlorpromazine hydrochloride lubricated with 0.5% magnesium stearate exhibited extensive recovery during decompression. An E value of 5.5 GPa was obtained, very similar to the microcrystalline celluloses (Avicel PH 102 and Emcocel, 5.8 and 6.1 GPa, respectively) which makes strong tablets, and ibuprofen (5.9 GPa) which makes very weak tablets. Tablets made with 0.5% magnesium stearate laminated and capped, but the addition of 0.5% w/w talc resulted in intact tablets. The E value was doubled and the extent of elastic recovery during decompression was reduced. 3. Tablet Strength Testing Tablet strength was measured using the diametral compression test. CPZ(II) did not form coherent tablets, and therefore, its force of failure, F and deformation could not be measured. The Ff values of tablets of CPZ(I)-H and CPZ(I)-H’ were similar to each other but consistently higher than CPZ(I) (Figure 38). Tablet deformation, on the other hand, was not different among the various forms of CPZ(I) (Figure 39). While the force of failure is indicative of the strength of the interparticulate bonds within a tablet, the extent of tablet deformation before failure was characteristic of the material. Tablet deformation was relatively independent of compression pressure and relative density despite the fact that the tablets were porous anisotropic bodies in which the stress was not uniformly distributed during compaction. The absence of differences in the deformation of CPZ(I), CPZ(I)..H’ and CPZ(I)-H was 123 Force of Failure (N) 200 0 150 0 0 0 0 0 0 100 0 0 0 0 0 50 0 0 D 0 0 0.7 0.75 0.8 0.85 0.9 0.95 Relative Density Figure 38. Force offailure of tablets of (JPZ(fl-H’ (0), CPZ(I)-H () and CPZ(I)(o). 124 Deformation (1OE-03 cm) 12 D .. D •10 CD C DC 0 8 0 0 Ll0 C 6 4. 2- I 0- 150 75 100 125 0 25 50 Force of Failure (N) Figure 39. Deformation of tablets of CPZ(I)-H’ (0), CPZ(I)-H () and CPZ(I) (C). 125 supported by XRPD, which showed that their diffractograms were qualitatively the same. Since these forms shared structural similarities, tablet deformation was expected to be similar. Differences in the Ff indicated that the completely dehydrated lattice, CPZ(I), formed much stronger interparticulate bonds on compression than CPZ(II), and during decompression, the tablets were able to withstand the stress of expansion. Water in the lattices of CPZ(I)-H’ and CPZ(I)-H led to a further increase in tablet strength, but this increase appeared to be independent of the water content (Figure 38). 126 SUMMARY A. Changes in Crystaffinity with Pharmaceutical Processing Standard methods used in quantitating crystallinity were modified for following changes in the X of organic solids. These findings suggested reductions in X with pharmaceutical processing and subsequent increases with storage but results were inconclusive. The reasons for the changes in X could not be explained. 2. The Rietveld structure refinement method was used successfully for the first time in studying the XRPD data of organic solids. 3. Modifications of the Rietveld method for the simultaneous measurement of crystallite size (relative) and lattice distortion were adapted from work on ideal inorganic crystals to the study of nonideal organic systems. MTZ and ASA were used as model compounds. Though the method was not definitive (e.g. crystallite size values were not absolute), the contribution of each phenomenon to the changes in X could be studied. 127 4. Both MTZ and ASA showed significant peak broadening and similar reductions in peak intensities with grinding and tabletting. Rietveld analysis revealed that while lattice distortion played an important role in the reduction of the X of ASA, the X changes of MTZ were more dependent on changes in crystallite size. 5. On storage, the b dimension of the unit cell of ASA remained unchanged, while the crystallite size of MTZ increased. The rate and extent of increase appeared dependent on the storage temperature. At any given time, larger crystallites were measured at the higher temperatures, and the size of crystallites stored at lower temperatures did not reach the sizes achieved for samples stored at higher temperatures. 6. The significant “lattice disruption or distortion” of doped DPH suggested by other workers was investigated using XRPD. Indexed X-ray powder diffraction data derived from Rietveld crystal structure refinements were reported. Using the Rietveld method (without internal standard) with the method of Appleman and Evans (1973) (with internal standard), accurate cell dimensions were obtained to study the incorporation PMDPH into DPH. Changes in cell dimensions were not observed suggesting that incorporation of PMDPH into the lattice of DPH was unlikely. Incorporation of PMDPH through the formation of a substitutional solid solution was proposed. 128 B. Phase Transitions with Pharmaceutical Processing (a) Differences in the tablettability of CPZ and its granules was found to result from a phase change rather than changes in X. (b) Detailed physicocheinical characterization revealed that complete conversion to a stoichiometric hemihydrate had occurred during wet granulation. (c) The stoichiometric hemihydrate could be fully dehydrated to the room temperature stable polymorph without the loss of lattice integrity. Structural refinement of X-ray powder diffraction data was used to measure the expansion of the lattice with water uptake. 2. (a) The improved tablettability of CPZ granules compared to CPZ could not be attributed to changes in viscoplastic deformation or elastic recovery. Differences in the extent of viscoplastic deformation during compression or in the elastic recovery of the compacts within the die during decompression were not significant. (b) Differences between the tabletting behavior of CPZ and its granules were related to differences in interparticulate bonding as a result of changes in the crystal lattice produced by wet granulation. 129 REFERENCES Aguiar, A.J., Krc, J., Jr., Kinkel, A.W. and Samyn, J.C., Effect of polymorphism on the absorption of chloramphenicol from chloramphenicol palmitate. J. Pharm. Sci., 56 (1967) 847-853. Aguiar, A.J. and Zelmer, J.E., Dissolution behaviour of polymorphs of chioramphenicol palmitate and mefenamic acid. J. Pharm. Sci., 58 (1969) 983-987. Albert, A. and Seijeant, E.P., The Determination of Ionization Constants: A Laboratory Manual, Third Edition, Chapman and Hall, London, 1984, p. 167. Ando, H., Watanabe, S., Ohwaki, T. and Miyake, Y., Crystallization of excipients in tablets. J. Pharm. Sci., 74 (1985a) 128-131. Ando, T., Kanaya, Y. and Asahina, K., Changes in physical characteristics of ethylaminobenzoate tablets during storage. Chem. Pharm. Bull., 33 (1985b) 3440-3446. Ando, H., Ohwaki, T., Ishii, M., Watanabe, S. and Miyake, Y., Crystallization of theophylline in tablets. mt. J. Pharm., 34 (1986) 153-156. Appleman, D.E. and Evans, H.T., Jr., Indexing and least-squares refinement of powder diffraction data. U.S. Geological Survey Computer Contribution 20. U.S. National Technical Information Service, 1973, Doc. PB2-16188. Armstrong, N.A. and Patel, A., The compressional properties of dextrose monohydrate and anhydrous dextrose of varying water contents. Drug Dev. md. Pharm., 12 (1986) 1885-1901. Ashizawa, K., Uchikawa, K., Hattori, T., Ishibashi, Y., Miyake, Y. and Sato, T., Estimation of the degree of crystallinity of crystalline E1040 by X ray diffraction and infrared photoacoustic methods. Yakugaku Zasshi, 110 (1990) 202-209. Baerlocher, C., Zeolite structure refinements using powder data. Zeolites, 6 (1986) 325-333. 130 Bauer, N. and Lewin, S.Z., Determination of density. In Weissberger, A. and Rossiter, B.W., Physical Methods of Chemistry, Volume I, Part IV, Wiley, New York, 1972, PP. 101-110. Benedetti, A., Fagherazzi, G., Enzo, S. and Battagliarin, M., A profile- fitting procedure for analysis of broadened X-ray diffraction peaks. II. Application and discussion of the methodology. J. Appl. Cryst., 21 (1988) 543-549. Black, D.B. and Lovering, E.G., Estimation of the degree of crystallinity in digoxin by X-ray and infrared methods. J. Pharm. Pharmacol. 29 (1977) 684-687. Black, D.B. and Lovering, E.G, The recrystallization of amorphous digoxin. J. Pharm. Pharmacol., 30 (1978) 380. Blaton, N.M., Peeters, O.M. and De Ranter, C.J., 2-(2-methyl-5-nitro-1- imidazoyl)ethanol (metronidazole). Acta Cryst., B35 (1979) 2465-2467. Bloss, F.B., Principles of Crystal Structures - An Introduction, Holt, Rhinehart and Winston, Inc., New York, 1971, Pp. 221-222, 283. Boldyrev, V.V., Bullens, M. and Delmon, B., The Control of the Reactivity of Solids - Studies in Surface Science and Catalysis, Volume 2, Elsevier, Amsterdam, 1979, (a) Pp. 23-24, and (b) P. 9. Bonetto, R.D., Viturro, H.R. and Alvarez, A.G., XTL-SIZE: a computer program for crystal-size-distribution calculation from X-ray diffraction line broadening. J. Appi. Cryst., 23 (1990) 136-137. Borka, L., Review of crystal polymorphism of substances in the European Pharmacopoeia. Pharm.ActaHelv., 66 (1991) 16-22. Borka, L. and Haleblian, J.K., Crystal polymorphism of pharmaceuticals. Acta Pharm. Jugosi., 40 (1990) 71-94. Bosley, D.E., X-ray determination of crystallinity in poly(ethylene terephthalate). J. Appi. Polymer Sci., 8 (1964) 152 1-1529. Brown, I.R., York, P. and Maclame, J.Q., The role of crystallisation conditions in the control of final product character. Pharm. Res., 7 (1990) S-107. Bunn, C.W., Chemical Crystallography, Oxford University Press, Oxford, 1946, pp. 67 and 93. Burt, H.M. and Mitchell, A.G., Crystal defects and dissolution. hit. J. Pharm., 9 (1981) 137-152. 131 Busing, W.R., WMIN, a computer program to model molecules and crystals in terms of potential energy functions. U.S. National Technical Information Service, (1981) ORNL-5747. Caglioti, G., Paoletti, A. and Ricci, F.P., Choice of collimators for a crystal spectrometer for neutron diffraction. Nuci. Instruin., 3 (1958) 223-228. Calvert, L.D. and Sirianni, A.F., A technique for controlling preferred orientation in powder diffraction samples. J. Appi. Cryst., 13 (1980) 462. Camerman, A. and Camerman, N., The stereochemical basis of anticonvulsant drug action. I. The crystal and molecular structure of diphenyihydantoin, a noncentrosymmetric structure solved by centric symbolic addition. Acta Cryst., B27 (1971) 2205-2211. Carstensen, J.T., Theory ofPharmaceutical Systems, Volume II, Academic Press, New York, 1973, (a) pp. 130-132, (b) pp. 128-130. Challa, G., Hermans, P.H. and Weidinger, A., On the determination of the crystalline fraction in isotactic polystyrene from X-ray diffraction, Makromol. Chem., 56 (1962) 169-178. Chan, H.K. and Doelker, E., Polymorphic transformation of some drugs under compression. Drug Dev. md. Pharm., 11 (1985) 315-332. Chi, S.-L., Hoag, S.G. and Yanchick, V.A., Electrolytic complexing of glucoheptonate and technetium-99m. J. Nucl. Med., 19 (1978) 520-524. Chow, A..H.L., Chow, P.K.K., Zhongshan, W. and Grant, D.J.W., Modification of acetaminophen crystals: influence of growth in aqueous solutions containing p-acetoxyacetanilide on crystal properties. mt. J. Pharm., 24 (1985) 239-258. Chow, A.H.L. and Hsia, C.K., Modification of phenytoin crystals: influence of 3-acetoxymethyl-5,5-diphenylhydatoin on solution-phase crystallization and related crystal properties. mt. J. Pharm., 75 (1991) 219-230. Chow, K.Y., Go, J., Mehdizadeh, M. and Grant, D.J.W., Modification of adipic acid crystals: influence of growth in the presence of fatty acid additives on crystal properties. mt. J. Pharm., 20 (1984) 3-24. Chow, K.Y., Go, J., Zhongshan, W., Mehdizadeh, M. and Grant, D.J.W., Modification of adipic acid crystals. II. Influence of growth in the presence of oleic acid on crystal properties. mt. J. Pharm., 25 (1985) 41-55. 132 Chung, F.H.and Scott, R.W., A new approach to the determination of crystaflinity of polymers by X-ray diffraction. J. Appi. Cryst., 6 (1973) 225-230. Cohen, M.U., The elimination of systematic errors in powder photographs. Z. Kristallogr., 94 (1936a) 288-298. Cohen, M.U., The calculation of precise lattice constants from X-ray powder photographs. Z. Kristallogr., 94 (1936b) 306-3 10. Cottrell, A.H., Dislocations and Plastic Flow in Crystals, Oxford University Press, London, 1953, pp. 39-41. Cox, J.S.G., Woodard, G.D. and McCrone, W.C., Solid-state chemistry of cromolyn sodium (disodium cromoglycate). J. Pharm. Sci., 60 (1971) 1458-1465. CRC Handbook of Chemistry and Physics, Lide, D.R. (ed.), 72nd Edition, CRC Press, Florida, 1991, p. 5-101. Cullity, B.D., Elements of X-ray Diffraction, Addison-Wesley Publishing Company, Reading, 1956, (a) pp. 297-323, (b) pp. 261-262, (c) 98-99. Darwin, C.G., The reflexion of x-rays from imperfect crystals. Phil. Mag., 43 (1922) 800-829. David, W.I.F. and Matthewman, J.C., Profile refinement of powder diffraction patterns using the Voigt function. J. Appi. Cryst., 18 (1985) 461-466. DeCrosta, M.A. and Eckhart, C.G., Utilization of X-ray powder diffraction and diffuse reflectance Infrared Fourier Transform Spectroscopy in assessing polymorphic changes in compressed solid dosage formulations. Pharm. Res., 6 (1989) S-i. de Keijser, Th.H., Langford, J.I., Mittemeijer, E.J. and Vogels, A.B.P., Use of the Voigt function in a single-line method for the analysis of X-ray diffraction line broadening. J. AppI. Cryst., 15 (1982) 308-3 14. de Keijser, Th.H.,Mittemeijer, E.J. and Rozendaal, H.C.F., The determination of crystallite-size and lattice-strain parameters in conjuction with the profile-refinement method for the determination of crystal structures. J. Appi. Cryst., 16 (1983) 309-316. Delhez, R., de Keijser, Th.H. and Mittemeijer, E.J., Determination of crystallite size and lattice distortions through X-ray diffraction line profile analysis. Fresenius Z. Anal. Chem., 312 (1982) 1-16. 133 Dorignac-Calas, M.R. and Marsau, P., Structure cristalline du chiorhydrate de chloro-3-(dimethylamino-3’propyl)- lO-phenothiazine. C. R. Acad. Sc. Paris, 274 (1972) 1806-1809. Doroudian, A., Determination of Compaction Parameters ofPharmaceutical Powders with an Instrumented Hydraulic Press, M.Sc. Thesis, University of British Columbia, 1991, PP. 19-25. Down, G.R.B. and McMullen, J.N., The effect of interparticulate friction and moisture on the crushing strength of sodium chloride compacts. Powder Tech., 42 (1985) 169-174. Drickamer, H.G., Pi electron systems at high pressure. Science, 156 (1967) 1183-1189. Duncan-Hewitt, W.C.. and Grant, D.J.W., True density and thermal expansivity of pharmaceutical solids: comparison of methods and assessment of crystallinity. mt. J. Pharm. , 28 (1986) 75-84. Dwivedi, S.K., Oates, R.J. and Mitchell, A.G., Peak offset times as an indication of stress relaxation during tableting on a rotary tablet press. J. Pharm. Pharmacol., 43 (1991) 673-678. Dwivedi, S.K., Oates, R.J. and Mitchell, A.G., Estimation of elastic recovery, work of decompression and Young’s Modulus using a rotary tablet press. J. Pharm. Pharmacol., 44 (1992) 459-464. Egawa, H., Maeda, S., Yonemochi, E., Oguchi, T., Yamamoto, K. and Nakai, Y., Solubility parameter and dissolution behavior of cefalexin powders with different crystallinity. Chem. Pharm. Bull., 40 (1992) 819-820. Elliott, A., Infra-red Spectra and Structure of Organic Long-Chain Polymers, St. Martin’s Press, New York, 1969, p. 111. Enzo, S., Fagherazzi, G., Benedetti, A. and Polizzi, S., A profile-fitting procedure for analysis of broadened X-ray diffraction peaks. I. Methodology. J. Appl. Cryst., 21 (1988) 536-542. Ewald, P.P., Group velocity and phase velocity in x-ray crystal optics. Acta Cryst., 11 (1958) 888-891. Findlay, A., The Phase Rule and Its Applications, Dover Publications, New York, 1951, pp.35-54. Florence, A.T. and Sable, E.G., Changes in crystallinity and solubility on comminution of digoxin and observations on spironolactone and oestradiol. J. Pharm. Pharmacol., 28 (1976) 637-642. 134 Florence, A.T., Sable, E.G. and Stenlake, J.B., The effect of particle size reduction on digoxin crystal properties. J. Pharm. Pharmacol., 26 (1974) 479-480. Frank, S.G., Inclusion compounds. J. Pharm. Sci., 64 (1975) 1585-1604. Friedel, J.,Dislocations, Pergamon, New York, 1964, p. 279. Friesen, M., Burt, H.M. and Mitchell, A.G., Crystal dislocations and dissolution. J. Pharm. Pharmacol., 33 (1981) 22P. Giordano, F., Gazzaniga, A., Bettinetti, G.P., and La Manna, A., The influence of water content on the binding capacity of f3-cyclodextrin. mt. J. Pharm., 62 (1990) 153-156. Glasstone, S., Physical Chemistry, Van Nostrand, New York, 1946, pp.463- 473. Go, J. and Grant, D.J.W., Locating the impurity in doped crystals using isotopic double labelling and a column “flow-through” dissolution cell: adipic acid doped with oleic acid. mt. J. Pharm., 36 (1987) 17-28. Gong, P., 5,5-cliphenyihydantoin (phenytoin). JCPDS Grant-in-Aid Report, Polytechnic Institute of New York, Brooklyn, New York, 1982. Gordon, J.D., Modification of Phenytoin Crystals: Influence of 3- propanoyloxymethyl-5, 5-diphenyihydantoin on Solution-phase Crystallization and Related Crystal Properties. M.Sc. Thesis, University of British Columbia, 1991, p. 67. Gordon, J.D. and Chow, A.H.L., Modification of phenytoin crystals. II. Influence of 3-propanoyloxymethyl-5,5-diphenylhydantoin on solution- phase crystallization and related crystal properties. mt. J. Pharm., 79 (1992) 171-181. Grant, N.H. and Album, H.E., Peptide hydrates: influence of the hydration state on the propertes of a-aminobenzylpenicillin. Nature, 207 (1965) 645-646. Grant, D.J.W. and York, P., A disruption index for quantifying the solid state disorder induced by additives or impurities. II. Evaluation from heat of solution. mt. J. Pharm., 28 (1986a) 103-112. Grant, D.J.W. and York, P., Entropy of processing: a new quantity for comparing the solid state disorder of pharmaceutical materials. mt. J. Pharm., 30 (1986b) 161-180. 135 Haleblian, J.K., Characterization of habits and crystalline modification of solids and their pharmaceutical applications. J. Pharm. Sci., 64 (1975) 1269-1288. Haleblian, J. and McCrone, W., Pharmaceutical applications of polymorphism. J. Pharm. Sci., 58 (1969) 911-929. Hayden, H.W., Moffatt, W.G. and Wulif, J., The Structure and Properties of Materials, Volume III (Mechanical Behavior), John Wiley and Sons, New York, 1965, p. 87. Herman, J., Remon, J.P., Visavarungroj, N., Schwartz, J.B. and Klinger, G.H., Formation of theophylline monohydrate during the pelletisation of microcrystalline cellulose-anhydrous theophylline blends. mt. J. Pharm., 42 (1988) 15-18. Herman, P.H. and Weidinger, A., Recrystallization of amorphous cellulose. J. Am. Chem. Soc., 68 (1946a) 1138, 68 (1946b) 2547-2552. Herman, P.H. and Weidinger, A., Quantitative X-ray investigations on the crystallinity of cellulose fibers. A background analysis. J. Appl. Phys., 19 (1948) 491-506. Herman, P.H. and Weidinger, A., Crystallinity of precipitated cellulose. J. Polymer Sci., 5 (1950) 565-568. Hess, H., Tablets under the microscope. Pharm. Tech., 11 (1987) 54-69. Hill, R. J., Expanded use of the Rietveld method in studies of phase abundance in multiphase mixtures. Powd. Diff, 6 (1991) 74-77. Hill; R.J. and Fischer, R.X., Profile agreement indices in Rietveld and pattern-fitting analysis. J. Appl. Cryst., 23 (1990) 462-468. Hill, R.J. and Flack, H.D., The use of the Durbin-Watson d statistic in Rietveld analysis. J. Appi. Cryst., 20 (1987) 356-361. Hill, R.J. and Madsen, I.C., The effect of profile step counting time on the determination of crystal structure parameters by X-ray Rietveld analysis. J. Appi. Cryst., 17 (1984) 297-306. Hill, R.J. and Madsen, I.C., The effect of profile step width on the determination of crystal structure parameters and estimated standard deviations by X-ray Rietveld analysis. J. Appl. Cryst., 19 (1986) 10-18. Hill, R.J. and Madsen, I.C., Data collection strategies for constant wavelength Rietveld analysis. Powd. Diff, 2 (1987) 146-161. 136 Holt, D.B., Assessment of crystalline perfection. In Pamplin, B.R. (ed), Crystal Growth, Second Edition, Pergamon, Oxford, 1980, P. 569. Howard, S.A. and Preston, K.D., Profile fitting of powder diffraction patterns. In Bish, D. L. and Post, J. E. (eds), Reviews in Mineralogy Volume 20: Modern Powder Diffraction, Mineralogical Society of America, 1989, pp. 240-244. Howard, S. A. and Snyder, R. L., Simultaneous crystallite size, strain and structure analysis from X-ray powder diffraction patterns. Material Science Research; Symposium on Advances in Materials Research, 19 (1985) 57-71. Hubbard, C.R. and Smith, D.K., Experimental and calculated standards for quantitative analysis by powder diffraction. Adv. X-ray Anal., 20 (1976) 27-39. Huttenrauch, R., Molekulargalenik als Grirndlage moderner Arzneiformung (Molecular pharmaceutics as a basis of modern pharmaceutical technology). Acta Pharm. Tech., APV Informationsdienst, Suppi. 6 (1978) 55-127. Hüttenrauch, R. and Fricke, S., The influence of lattice defects on the course and rate of drying of particulate solids, Part 45: Molecular Pharmaceutics. mt. J. Pharm. Technol. Prod. Manuf , 2 (1981) 35-37. Huttenrauch, R. and Keiner, I., How crystalline are “microcrystalline” celluloses? Pharmazie, 31 (1976) 183-187. Hüttenrauch, R. and Keiner, I., Influence of lattice defects upon mixing processes. Powder Tech., 22 (1979) 289-290. Ibrahim, 0., Pisano, F. and Bruno, A., Polymorphism of phenylbutazone: properties and compressional behavior of crystals. J. Pharm. Sci., 66 (1977) 669-673. Imaizumi, H., Nambu, N. and Nagai, T., Stability and several physical properties of amorphous and crystalline forms of indomethacin. Chem. Pharm. Bull., 28 (1980) 2565-2569. International Tables for X-ray Crystallography, Volume IV, Kynoch Press, Birmingham, 1974. Jamali, F. and Mitchell, A.G., The recrystallization and dissolution of acetylsalicyclic acid. ActaPharm. Suec., 10 (1973) 343-352. Johnston, H.K. and Hutchison, C.A., Efficiency of the electrolytic separation of lithium isotopes. J. Chem. Phys., 8 (1940) 869-877. 137 Kamat, MS., Osawa, T., DeAngelis, R.J., Koyama, Y. and DeLuca, P.P., Estimation of the degree of crystallinity of cefazolin sodium by X-ray and infrared methods. Pharm.Res., 5 (1988) 426-429. Kaneniwa, N. and Otsuka, M., The interaction between water and cephalexin in the crystalline and noncrystalline states. Chem. Pharm. Bull., 32 (1984) 4551-4559. Kaneniwa, N., Imagawa, K. and Otsuka, M., Effect of tableting on the degree of crystallinity and on the dehydration and decomposition points of cephalexin crystalline powder. Chem. Pharm. Bull., 33 (1985) 802- 809. Kaneniwa, N. and Otsuka, M., Effect of grinding on the transformations of polymorphs of cKloramphenicol palmitate. Chem. Pharm. Bull., 33 (1985) 1660-1668. Ke, B., Differential thermal analysis. In the Encyclopedia ofPolymer Science and Technology, Volume V, Wiley, New York, 1966, p. 41. Kim, Y., Machida, K., Taga, T. and Osaki, K., Structure redetermination and packing analysis of aspirin crystal. Chem. Pharrn. Bull., 33 (1985) 2641-2647. Kitaigorodsky, A.I., Molecular Crystals and Molecules, Academic Press, New York, 973, pp08±±& Kitamura, S., Miyamae, A., Koda, S. and Morimoto, Y., Effect of grinding on the solid-state stability of cefixime trihydrate. mt. J. Pharm., 56 (1989) 125-134. Klein, C.L. and Conrad III, J.M., Structure of a new crystallographic form of chiorpromazine hydrochloride hemihydrate. Acta Cryst., C42 (1986) 1083-1085. Kiug, H.P. and Alexander, L.E., X-ray Diffraction Procedures for Polycrystalline and Amorphous Materials, Second Edition, Wiley, New York, 1974, (a) pp. 1-4, (b) pp. 138-139, (c) pp. 566-567, 618, (d) pp. 202- 206, (e) pp. 372-373, (f) p. 369, (g) pp. 290-297 and (h) pp. 148-152. Kontny, M.J., Grandolfl, G.P. and Zografi, G., Water vapor sorption of water-soluble substances: studies of crystalline solids below their critical relative humidities. Pharm. Res., 4 (1987) 104-112. Kuhnert-Brandstatter, M., Thermomicroscopy in the Analysis of Pharmaceuticals, Pergamon Press, Oxford, 1971, pp. 34-42. Ladd, M.F.C. and Palmer, R.A., Structure Determination by X-ray Crystallography, Plenum Press, New York, 1977, p. 347. 138 Lame, E., Tuominen, V., Ilivessalo, P. and Kahela, P., Formation of dihydrate from carbamazepine anhydrate in aqueous conditions. mt. J. Pharm., 20 (1984) 307-314. Langford, J. I., A rapid method for analysing the breadths of diffraction and spectral lines using the Voigt function. J. App 1. Cryst., 11 (1978) 10-14. Langford, J.I., Accuracy of crystallite size and strain determined from the integral breadth of powder diffraction lines. In Block, S. and Hubbard, C.R. (eds.),Accuracy in Powder Diffraction, Nati. Bur. Stand. Spec. Pub. No. 567, National Bureau of Standards, Washington, 1980, pp. 255-269. Langford, J.I., Applications of total pattern fitting to a study of crystallite size and strain in zinc oxide powder. Powd. Duff, 1 (1986) 211-221. Larson, A.C. and von Dreele, R.B., GSAS, Generalized Crystal Structure Analysis System, Los Alamos National Laboratory, Los Alamos, NM, 1987,Report LAUR-86-748. Lefebvre, C., Guyot-Hermann, A.M., Draguet-Brughmans, M., Bouche, R. and Guyot, J.C., Polymorphic transitions of carbamazepine during grinding and compression. Drug Dev. md. Pharm., 12 (1986) 1913- 1927. Leigh, G.J. (ed.), Nomenclature of Inorganic Chemistry, Blackwell Scientific Publications, Oxford, 1990, pp. 40, 46 and 79-80. Li, L.C. and Peck, G.E., The effect of moisture content on the compression properties of maltodextrins. J. Pharm. Pharmacol., 42 (1990) 272-275. Lindenbaum, S. and McGraw, S.E., The identification and characterization of polymorphism in drug solids by solution calorimetry. Pharm. Manuf , 2 (1985) 26-30. Lowell, S., Continuous flow krypton adsorption for low surface area measurements. Anal. Chem., 45 (1973) 1576-1577. Lutterotti, L. and Scardi, P., Simultaneous structure and size-strain refinement by the Rietveld method. J. Appl. Cryst., 23 (1990) 246-252. Madsen, I. C. and Hill, R. J., Effect of divergence and receiving slit dimensions on peak profile parameters in Rietveld analysis of X-ray diffractometer data. J. Appi. Cryst., 21 (1988) 398-405. Malmros, G. and Thomas, J.O., Least-squares structure refinement based on profile analysis of powder film intensity data measured on an automatic microdensitometer. J. Appi. Cryst., 10 (1977) 7-11. 139 Mastropaolo, D., Camerman, A. and Camerman, N., Hydrogen bonding interaction of diphenylhydantoin and 9-ethyladenine. Molec. Pharmac., 23 (1983) 273-277. Mathews, A.G., Schram, C.J. and Minty, D., Crystallization of penicillin in the solid phase. Nature, 5052 (1966) 959. Matsumoto, T., Kaneniwa, N., Higuchi, S. and Otsuka, M., Effects of temperature and pressure during compression on polymorphic transformation and crushing strength of chiorpropamide tablets. J. Pharm. Pharmacol., 43 (1991) 74-78. Matthews, G.P., Lowther, N. and Shott, M.J., Crystallization of carbamazepine in tablets stored at elevated temperatures. mt. J. Pharm., 50 (1989) 111-115. Matthews, J.L., Peiser, H.S. and Richards, R.B., The X-ray measurement of the amorphous content of polythene samples. Acta Cryst., 2 (1949) 85- 90. Mayer, I. and Schneider, S., Metronidazole. JCPDS Grant-in-Aid Report, Hebrew University, Jerusalem, Israel, 1986. McCrone, W.C., Fusion Methods in Chemical Microscopy, Interscience Publishers, New York, 1957, (a) 192-197, (b) 134-142. McCrone, W.C. and Cheng, P.T., Grain growth in octachloropropane. J. Appi. Phys., 20 (1949) 230-231. McCusker, L.B., Baerlocher, C. and Nawaz, R., Rietveld refinement of the crystal structure of the new zeolite mineral gobbinsite. Z. Krist., 171 (1985) 281-289. Meares, P., Polymer Structure and Bulk Properties, Van Nostrand, London,1965, p. 127. Meier, W.M. and Villiger, H., Die Methode der Abstandsverfeinerung zur B estimmung der Atomkoordinaten idealisierter Gerüststrukturen. Z. Cryst., 129 (1969) 411-423. Miller, R.L., Crystallinity. In the Encyclopedia of Polymer Science and Technology, Volume IV, Wiley, New York, 1966, (a) pp. 449-45 3, (b) pp. 490-492. Mitchell, A.G., The preparation and characterization of ferrous sulphate hydrates. J. Pharm. Pharmacol., 36 (1984) 506-510. 140 Mitchell, A:G., Polymorhism in metoclopramide hydrochloride and metoclopramide. J. Pharm. Pharmacol., 37 (1985) 60 1-604. Mitchell, A.G., Current concepts in crystallinity presented at the symposium on the Solid State Phenomena Important in Dosage Form Design at the Second National Meeting and Exposition of the American Association of Pharmaceutical Scientists, Boston (June 8, 1987). Mitchell, A.G. and Down, G.R.B., Recrystallization after powder compaction. mt. J. Pharm., 22 (1984) 337-344. Mitchell, A.G. and Suryanarayanan, R., Solubility of Calcium Gluceptate and the Preparation of Calcium Gluceptate Injection USP. J. Parenter. Sci. Tech., 40 (1986) 48-51. Morita, M. and Hirota, S., Estimation of the degree of crystallinity and the disorder parameter in a drug by an X-ray diffraction method with a single sample. Chem. Pharm. Bull., 30 (1982) 3288-3296. Nakagawa, H., Takahashi, Y. and Sugimoto, I., The effects of grinding and drying on the solid state stability of sodium prasterone sulfate. Chem. Pharm. Bull., 30 (1982) 242, 248. Nakai, Y., Fukouka, E., Nakajima, S. and Hasegawa, J., Crystallinity and physical characteristics of microcrystalline cellulose. Chem. Pharm. Bull., 25 (1977) 96-101. Nakai, Y., Fukouka, E., Nakajima, S. and Morita, M., Physicochemical properties of crystalline lactose. Estimation of the degree of crystallinity and the disorder parameter by an X-ray diffraction method. Chem. Pharm. Bull., 30 (1982) 1811-1818. National Physical Laboratory (Great Britain), Notes on Applied Science No. 4: Measurement ofHumidity. London, HMSO, 1958. Nogami, H., Nagai, T., Fukuoka, E. and Yotsuyanagi, T., Dissolution kinetics of bartital polymorphs. Chem. Pharm. Bull., 17 (1969) 23-31. Nuffield, E.W.,X-ray Diffraction Methods, John Wiley and Sons, New York, 1966, pp. 147-149. Nyqvist, H. and Wadsten, T., Change in surface area of zimeldine dihydrochloride hydrate on storage. Acta Pharm. Suec., 21 (1984) 235- 244. Oates, R.J. and Mitchell, A.G., Calculation of punch displacement and work of powder compaction on a rotary tablet press. J. Pharm. Pharmacol., 41 (1989) 517-523. 141 Oates, R.J. and Mitchell, A.G., Comparison of calculated and experimentally determined punch displacement on a rotary tablet press using both Manesty and IPT punches. J. Pharm. Pharmacol., 42 (1990) 388-396. Oates, R.J., Wong, M.W.Y. and Mitchell, A.G., Determination of tablet deformation using a modified commercial tablet strength tester. mt. J. Pharm. Tech. Prod. Mfr., Fall Edition (1989). Oberholtzer, E.R. and Brenner, G.S., Cefoxitin sodium: solution and solid- state chemical stability studies. J. Pharin. Sci., 68 (1979) 863-866. Orr, J.M., Surface area calculations using the Quantasorb: a computer program, Faculty of Pharmaceutical Sciences, The University of British Columbia, 1983. Osawa, T., Kamat, M.S. and Deluca, P.P., Hygroscopicity of cefazolin sodium: application to evaluate the crystallinity of freeze-dried products. Pharm. Res., 5 (1988) 42 1-425. Otsuka, M. and Kaneniwa, N., Effect of grinding on the degree of crystallinity of cephalexin powder. Chem. Pharm. Bull., 31 (1983a) 4489-4495. Otsuka, M. and Kaneniwa, N., Effect of grinding on the physicochemical properties of cephalexin powder. Chern. Pharm. Bull., 32 (1983b) 107 1- 1079. Otsuka, M. and Kaneniwa, N., Hygroscopicity and solubility of noncrystalline cephalexin. Chem. Pharm. Bull., 31 (1983c) 230-236. Otsuka, M. and Kaneniwa, N., Dehydration of cephalexin hydrates. Chem. Pharm. Bull., 31 (1983d) 1021-1029. Otsuka, M. and Kaneniwa, N., Effect of seed crystals on solid-state transformation of polymorphs of chioramphenicol palmitate during grinding. J. Pharm. Sci., 75 (1986) 506-511. Otsuka, M. and Kaneniwa, N., Effect of grinding on the crystallinity and chemical stability in the solid state of cephalothin sodium. mt. J. Pharm., 62 (1990) 65-73. Otsuka, M., Matsumoto, T. and Kaneniwa, N., Effect of environmental temperature on polymorphic solid-state transformation of indomethacin during grinding. Chem. Pharm. Bull., 34 (1986) 1784- 1793. Otsuka, M., Matsumoto, T. and Kaneniwa, N., Effects of the mechanical energy of multi-tableting compression on the polymorphic 142 transformations of chiorpropamide. J. Pharm. Pharmacol., 41 (1989) 665-669. Otsuka, A., Wakimoto, T. and Takeda, A., Moisture sorption and volume expansion of amorphous lactose tablets. Chem. Pharm. Bull., 26 (1978) 967-971. Petkov, V., RAD, a program for analysis of X-ray diffraction data from amorphous materials for personal computers. J. AppI. Cryst., 22 (1989) 387-389. Pfeiffer, R.R., Yang, K.S. and Tucker, M.A., Crystal pseudopolymorphism of cephaloglycin and cephalexin. J. Pharm. Sci., 59 (1970), 1809-18 14. Philip, J., Holcomb, I.J., and Fusari, S.A., Phenytoin. In Florey, K. (ed.), Analytical Profiles of Drug Substances, Volume 13, Academic Press, Orlando, Florida, 1984, pp. 417-445. Pikal, M.J. and Grant, D.J.W., A theoretical treatment of changes in energy and entropy of solids caused by additives or impurities in solid solution. mt. J. Pharm., 39 (1987) 243-253. Pikal, M.J., Lukes, A.L., Lang, J.E. and Gaines, K., Quantitative crystallinity determinations for beta-lactam antibiotics by solution calorimetry: correlations with stability. J. Pharm. Sci., 67 (1978) 767- 773. Post, J.E. and Bish, D.L., Rietveld refinement of the todorokite structure. Amer. Mineral., 73 (1988) 861-869. Post, J.E. and Bish, D.L., Rietveld refinement of crystal structures using powder X-ray diffraction data. In Bish, D.L. and Post, J.E. (eds), Reviews in Mineralogy Volume 20: Modern Powder Diffraction, Mineralogical Society of America, Washington, 1989, pp. 277-308. Post, J.E. and Bish, D.L., Rietveld refinement of the coronadite structure. Amer. Mineral., 74 (1989b) 913-917. Post, J.E. and Burnham, C.W., Ionic modeling’of mineral structures and energies in the electron gas approximation: Ti02 polymorphs, quartz, forsterite, diopside. Amer. Mineral., 71 (1986) 142-150. Raudsepp, M., Hawthorne, F.C. and Turnock, A.C., Evaluation of the Rietveld method for the characterization of fine-grained products of mineral synthesis: the diopside-hedenbergite join. Canadian Mineralogist, 28 (1990) 93-109. Read, W.T., Jr., Dislocations in Crystals, McGraw-Hill, New York, 1953, pp. 13-24. 143 Richardson, J. W., Jr., Pluth, J. J. and Smith, J. V., Rietveld profile analysis of calcined AIPO4-11 using pulsed neutron powder diffraction. Acta Cryst., B44 (1988) 367-373. Rietveld, H.M., Line profiles of neutron powder diffraction peaks for structure refinement. Acta Cryst., 22 (1967) 151-152. Rietveld, H.M., A profile refinement method for nuclear and magnetic structures. J. Appl. Cryst., 2 (1969) 65-71. Roy, R., Classification of noncrystalline solids. J. Non-cryst. Solids, 3 (1970) 33-40. Ruland, W., X-ray determinaton of crystallinity and diffuse order scattering. Acta Cryst., 14 (1961) 1180-1185. Sakthivel, A. and Young, R.A., User’s Guide to Programs DBWS-9006 and DBWS-9006PC for Rietveld Analysis ofX-ray and Neutron Powder Diffraction Patterns, School of Physics, Georgia Institute of Technology, Atlanta, Georgia, (September) 1991. Saleki-Gerhardt, A., Ahineck, C. and Zografi, G., Assessment of disorder in crystalline solids, Pharm. Res., 9 (1992) S-120. Shell, J.W., X-ray nd crystallographic applications in pharmaceutical research. II. Quantitative X-ray diffraction. J. Pharm. Sci., 52 (1963) 24-29. Sherwood, J.N., Self-diffusion in organic solids. In Anderson, J.S., Roberts, M.W. and Stone, F.S. (eds), Proceedings of the Seventh International Symposium of the Reactivity ofSolids, Bristol, 1972, pp. 252-261. Shukia, A.J. and Price, J.C., Effect of moisture content on compression properties of two dextrose-based directly compressible diluents. Pharm. Res., 8 (1991) 336-340. Smith, D.K., Computer analysis of diffraction data, in the Reviews in Mineralogy Volume 20: Modern Powder Diffraction, Mineralogical Society of America, Washington, D.C., 1989, p. 183-216. Smith, D.K. and Barrett, C.S., Special handling problems in X-ray diffractometry. Adu. X-rayAnal., 22 (1979) 1-12. Smith, D.K. and Gorter, S., Powder diffraction program information 1990 program list. J. AppI. Cryst., 24 (1991) 369-402. 144 Smith, S.T., Snyder, R.L. and Brownell, W.E., Minimization of preferred orientation in powders by spray drying. Adv. X-rayAnal., 22 (1979) 77- 87. Sonneveld, E.J. and Visser, J.W., Automatic collection of powder data from photographs. J. Appl. Cryst., 8 (1975) 1-7. Stahl, P.H., The problems of drug interactions with excipients. In Breimer, D.D. (ed), Towards Better Safety of Drugs and Pharmaceutical Products, Elsevier, Amsterdam, 1980, pp. 265-280. Stanley-Wood, N.G. and Johansson, M.E., Measurement of interparticle voidage and particle contact area in compacts by nitrogen adsorption. Drug Dev. md. Pharm., 4 (1978) 69-94. Statton, W.O., An X-ray crystallinity index method with application to poly(ethylene terephthalate), J. Appl. Polymer Sci., 7 (1963) 803-8 15. Stout, G.H. and Jensen, L.H., X-ray Structure Determination - A Practical Guide, Second Edition, John Wiley and Sons, N.Y., 1989, (a) pp. 78-79, (b) pp. 74-78. Summers, M.P., Enever, R.P. and Carless, J.E., The influence of crystal form on the radial stress transmission characteristics of pharmaceutical materials. J. Pharm. Pharmacol., 28 (1976) 89-99. Suortti, P., Ahtee, M. and Unonius, L., Voigt function fit of X-ray and neutron powder diffraction profiles. J. Appi. Cryst., 12 (1979) 365-369. Suryanarayanan, R., Studies on the Crystallinity and Phase Transitions of Calcium Gluceptate, Ph.D. thesis, University of British Columbia, 1985, pp. 3-8. Suryanarayanan, R. and Mitchell, A.G., Evaluation of two concepts of crystallinity using calcium gluceptate as a model compound. mt. J. Pharm., 24 (1985) 1-17. Suryanarayanan, R. and Mitchell, A.G., Phase transitions of calcium gluceptate. mt. J. Pharm., 32 (1986) 213-221. Swalin, R.A., Thermodynamics of solids, John Wiley and Sons, New York, 1972, p. 1. Tomlinson, E., Enthalpy-entropy compensation analysis of pharmaceutical, biochemical and biological systems. mt. J. Pharm., 13 (1983) 115-144. 145 Toraya, H., Whole-powder-pattern fitting without reference to a structural model: application to X-ray powder diffractometer data. J. App I. Cryst., 19 (1986) 440-447. Toraya, H., Yoshimura, M. and Somiya, S., A computer program for the deconvolution of X-ray diffraction profiles with the composite of Pearson type VII functions. J. Appi. Cryst., 16 (1983) 653-657. Train, D., An investigation into the compaction of powders. J. Pharm. Pharmacol., 8 (1956) 745-761. The United States Pharmacopeia XXII and the National Formulary XVII, U.S. Pharmacopeial Convention Inc., Rockville, MD, 1990, pp. 1577, 1621-1623. Vachon, M.G. and Grant, D.J.W., Enthalpy-entropy compensation in pharmaceutical solids. mt. J. Pharm., 40 (1987) 1-14. Van Scoik, K.G. and Carstensen, J.T., Nucleation phenomena in amorphous sucrose systems. mt. J. Pharm., 58 (1990), 185-196. Vanderzee, C.E., Waugh, D.H. and Haas, N.C., The influence of grinding stress and thermal annealing on the enthalpy of solution of tris(hydroxymethyl)aminomethane and its use as a test substance in solution calorimetry. J. Chem. Thermodynamics, 13 (1981) 1-12. Varia, S.A., Schuller, S., Sloan, K.B. and Stella, V.J., Phenytoin prodrugs III: water-soluble prodrugs for oral and/or parentera use. J. Pharm. Sci., 73 (1984) 1068-1073. Vernon, R.H., Metamorphic Processes - Reactions and Microstructure Development, John Wiley and Sons, New York, 1975, p.166-167. Wakelyn, N.T. and Young, P.R., Crystallinity index of poly(ethylene terephthalate) by X-ray diffractometry and differential scanning calorimetry, J. Appi. Polymer Sci., 10 (1966) 1421-1438. Wall, G.M., Pharmaceutical applications of drug crystal studies. Pharm. Manuf, February (1986) 33-42. Warren, B.E., Averbach, B.L., The effect of cold-work distortion on X-Ray patterns. J. Appi. Phys., 21 (1950) 595-599. Wheatley, P.J., The crystal and molecular structure of aspirin. J. Chem. Soc., suppl. 1163 (1964) 6036-6048. Weidinger, A. and Hermans, P.H., On the determination of the crystalline fraction of isotactic polypropylene from X-ray diffraction, Makromol. Chem., 50 (1961) 98-115. 146 Wiedemann, K.E., Unnam, J. and Clark, R.K., Computer program for deconvoluting powder diffraction spectra. Powd. Diff, 2 (1987) 137-145. Wiles, D.B., Sakthivel, A. and Young, R.A., DBW 3.2S program for Rietveld analysis ofX-ray and neutron powder diffraction patterns (version 8804), 1988. Wiles, D.B. and Young, R.A., A new computer program for Rietveld analysis of X-ray powder diffraction patterns. J. Appi. Cryst., 14 (1981) 149-151 Will, G., Parrish, W. and Huang, T.C., Crystal-structure refinement by profile fitting and least-squares analysis of powder diffraction data. J. Appi. Cryst., 16 (1983) 611-622. Winnike, R.A., Wurster, D.E. and Guillory, J.K., A solid sampling device for use in batch solution calorimetry. Thermochimica Acta, 124 (1988) 99- 108. Wong, M.W.Y. and Mitchell, A.G., Physicochemical characterization of a phase change produced during the wet granulation of chiorpromazine hydrochloride and its effects on tableting. mt. J. Pharm., 88 (1992) 261- 273. Wood, J.H., Syarto, J.E. and Letterman, H., Improved holder for intrinsic dissolution rate studies. J. Pharm. Sci., 54 (1965) 1068. Woolfson, M.M., An introduction to X-ray crystallography, Cambridge University Press, London, 1970, pp. 182-189 Yamaguchi, T., Nishimura, M., Okamoto, R., Takeuchi, T. and Yamamoto, K., Glass formation of 4”-O-(4-methoxyphenyl)acetyltylosin and physicochemical stability of the amorphous solid. mt. J. Pharm., 85 (1992) 87-96. York, P., Solid state properties of powders in the formulation and processing of solid dosage forms. mt. J. Pharm., 14 (1983) 1-28. York, P. and Grant, D.J.W., A disruption index for quantifying the solid state disorder induced by additives or impurities. I. Definition and evaluation from heat of fusion. mt. J. Pharm., 25 (1985) 57-72. Young, R.A., Structural analysis from X-ray powder diffraction patterns with the Rietveld method. Symposium on Accuracy in Powder Diffraction, National Bureau of Standards, Gaithersburg, MD, (1980) 143-163. 147 Young, R.A., Mackie, P.E. and von Dreele, R.B., Application of the pattern- fitting structure-refinement method to X-ray powder diffractometer patterns. J. Appi. Cryst., 10 (1977) 262-269. Young, R.A., Prince, E. and Sparks, R.A., Suggested guidelines for the publication of Rietveld analyses and pattern decomposition studies. J. Appl. Cryst., 15 (1982) 357-359. Young, R.A. and Wiles, D.B., Application of the Rietveld method for structure refinement with powder diffraction data. Advances in X-ray Analysis, 24 (1981) 1-23. Young, R.A. and Wiles, D.B., Profile shape functions Rietveld refinements. J. Appi. Cryst., 15 (1982) 430-438. Yvon, K., Jeitschko, W. and Parthe, E., LAZY PUL VERIX, a computer program, for calculating X-ray and neutron diffraction powder patterns. J. Appi. Cryst., 10 (1977) 73-74. Ziegler, G., Structural and morphological investigations of ceramic powders and compacts. Pwdr. Met. mt., 10 (1978) 70-73. 148 APPENDIX A DESCRIPTION OF THE RIETVELD STRUCTURE ANALYSIS COMPUTER PROGRAMS The original structure refinement program by Rietveld(1967, 1969) has been extensively modified. Several computer packages are now available for the analysis of X-ray powder diffraction data (Young, 1980; Young and Wiles, 1981; Bish and Post, 1989; and, Smith and Gorter, 1991). Among these versions, DBW (D.B. Wiles; Wiles and Young, 1981) and GSAS (Generalized Crystal Structure Analysis System; Larson and von Dreele, 1987) are the most widely distributed (Bish and Post, 1989; Smith and Gorter, 1991). DBW (9006 series) enables the simultaneous refinement of up to eight phases and runs on either a mainframe or a personal computer (Sakthivel and Young, 1991). GSAS automatically calculates bond angles and distances but requires a VMS (Virtual rnemory System) operating system on a VAX ( irtual ddress e)jtension) computer (Larson and von Dreele, 1987; Bish and Post, 1989). In this project, the DBWS-9006PC program of Sakthivel and Young (1991) was used to perform Rietveld (1967 and 1969) analysis on X-ray powder diffraction data collected with a 0-20 diffractometer in the step-scan mode. No preparatory program is required (i.e. single-pass operation), and 149 direct applicability with all space groups and with all atoms is built-in by incorporating the required scattering factors listed in the International Tables for X-ray Crystallography (1974) as coefficients of an exponential series. This generates the X-ray scattering factors and necessary anomalous scattering corrections (Wiles and Young, 1981). A least-squares procedure is used to refine structure parameters (atomic positions, site occupancies, and displacement parameters) with the various instrumental parameters, in order to minimize differences between the observed and calculated diffraction patterns (i.e. the residual, 91) = iv, — r (A-i) where: y is the observed intensity, and yj is the calculated intensity at step i. The weight assigned to each step, wt, is determined by = Wi (A-2) yjo Yib where Yib is the background intensity at step i During the refinement, Yic is calculated as the sum of contributions from neighboring Bragg reflections and the background 150 =ALkIF4(2Oj—2OK)FK+yth (A-3) where: A is a scale factor, LKincludes the Lorentz, polarization and multiplicity factors for the Kth Bragg reflection (K=h,k,l), FK is the structure factor, PK is the preferred orientation function, and • is the reflection profile function which approximates both instrumental and sample effects. The ratio of the intensities of the two a wavelengths is accounted for in the calculation of F j. and therefore only one scale factor is required. Two options are available when refining Pj Toraya ‘s modification of the original Rietveld function: G2 ÷(1_G2)e1 (A-4) and the March-Dollase function: 3 (A-5) G1) ) where: G1 and G2 are refinable parameters, and cq(is the acute angle between the scattering vector and the presumed cylindrical symmetry axis normal to the crystallite. 151 As shown in equation 3, the calculation of y1 is very dependent on the background and, therefore, an accurate description of Yib is essential. The background intensity at each step is modelled from either the refinable background function: - lm y11, = Bm (A-6) LBKPOS or an operator-supplied table of background intensities, or the linear interpolation between operator-selected points in the diffiactogram. The accurate characterization of peak shape enables information about crystallite size and heterogeneous strain to be extracted. Unfortunately, the X-ray powder diffraction pattern is a complex convolution of several sample and instrumental effects. The peak shape deviates significantly from Gaussian and is very difficult to model mathematically. More complex functions are required. Several alternatives have been proposed and those available in the DBWS-9006PC package include the Lorentzian, Modified (mod 1) Lorentzian, Intermediate (mod 2) Lorentzian, Edgeworth Series, Voigt, pseudo-Voigt, modified Thompson-Cox-Hastings pseudo-Voigt, and Pearson VII. Further details on these profile functions are provided in Table 9aandb). The angle dependency of the peak shape is also modelled and enables the refinement to be valid over a wide range of diffraction angles. Profile breath, expressed as the full-width at half-maximum height (FWHM) or Hjç ______ 152 Table 9a - Analytical functions used to represent the diffraction profile1 Name Symbol Function2 C0(201—20K)2 Gaussian G e - H )2 0 11 Lorentzian L (2 — 20K + Ti-2 .wHK [ K j Modified Lorentzian 12 (Mod 1 Lorentzian) ML 2..j (20, — 20K)2 XHK[+C2 H j Intermediate Lorentzian (Mod2Lorentzian) IL (20,_20K)2{2 ‘/ 1i÷c 2HKL ‘“‘-K j Edgeworth Series3 (2o — 20K (Polynomial) Poly ( ) P, Q ) Voigt3 V A2fL(x)G((20—20K)—x)dx Pseudo-Voigt pV Modified Thompson-Cox- Mod-TCH Hastings pseudo- pV Voigt - (20._20K)21m PearsonVil p’\Tjj C4 F1 (2m_1) HK[ 1 References: Young and Wiles (1981); Smith (1989); and, Sakthivel and Young (1991). 2 detailed definitions of the function variables provided in Table 9b 3 not available with the DBWS-9006PC package 153 Table 9b - Variables used in profile functions* Variable Definition A1 A2 normalization factors C0 41n2 C1 4 C2 4(J_i C3 4(2_i 1 1 Cd 2Y(2rn_1)2 )(m-O.5)J 0.2 (r +2. 69269 TFL +2. 42843 FI’ + 4.47 163 rr \+O.O7842I’Ij, + ) (Utan2O+VtanO+W+ cosO) X tan 0+ cos 0 H (u tan2 0K + V tan °K + w) NA+NB(2q) i.36603q-O.47719q2+O.1116q3 NB NC m NA+j+()2 P polynomial with even exponents only Q polynomial with even exponents only * References: Young and Wiles (1981); Smith (1989); and, Sakthivel and Young (1991). t Gfu1l-width-at-half-maximum (FWHM) H=FWHM (Cagliotti et al., 1958) § NA and NB are refinable § NA, NB and NC are refinable 154 can be described as a function of the diffraction angle by the relationship (Cagliotti et al., 1958): 14 =Utan2O÷VtanO+W (A-7) where U, V, and W are refinable constants for the X-ray pattern (dependent on the instrumental configuration and choice of profile shape function). The Newton-Raphson algorithm is the least-squares procedure used and the normal matrix elements are formally given by: Mjk _2w{[Y0 —Y] (A-8) — but can be approximated by omitting the term [y, - The parameter is refinable. The standard deviation for thejth adjusted parameter, 5j, is calculated as: 1 = °j (A-9) where: M is the diagonal element in the inverse of the normal matrix, N is the number of observations, P is the number of parameters refined, and C is the number of constraints imposed. 155 During the refinement, agreement between the observed and calculated pattern is constantly monitored by visual inspection and the use of profile agreement indices. Visual inspection of the intensity versus 20 plots of the observed, calculated, and difference patterns is most informative in assessing goodness of fit (Young and Prince, 1982; Baerlocher, 1986) and provides information about the source of discrepancies (e.g. an improperly fit background or peak shape irregularities). Profile agreement indices, on the other hand, supply a numerical measure of fit and enable quantitative criteria of fit to be established (Hill and Fischer, 1990). These indices are usually defined in terms of R-factors (Table 10). The Bragg index (RB) provides a good measure of the validity of the crystal structure model and uses the ‘observed’ Bragg intensity (‘Jobs’) which is calculated by allocating the observed intensities, yj,3, to Bragg intensities, ‘Jobs’, on the basis of the calculated intensities, ‘calc This procedure is described by Rietveld (1969). Both the weighted profile residual (R) and the goodness-of-fit (GofF) indicate if the refinement is converging smoothly. Statistically, is considered the most important for following the progression of a refinement because its numerator is the quantity being minimized. Rexp, the expected R-factor, is the minimum value possible. As the refinement progresses, should approach Rexp and GofF should become unity. If the GofF is greater than unity, either the weights used are inappropriate, or the structural model or peak representation is incorrect. 156. Table 10 - Agreement indices for the Rietveld refinement1 Name Symbol Equation 100 — Pattern R-factor2 R 31k I yw Weighted Pattern R- , 100 (y — y factor3 w1(y,)2 100 101,8 Bragg intensity R-factor4 I — ‘calcl ‘obs Expected R-factor5 Rexp [(N - P + C)l w(y)2 loot j Structure Amplitudes 100 — Wcaic R-factor F 2 II1obs’ I II 2 = Goodness-of Fit GofF Y — y)2 (R (N1’) kflexp 2 Durbin-Watson statistic N ,‘ ( ‘ I — Yic) — Yio-i — Yk_1 (d-statistic)6 d ‘o j 1 References: Post and Bish (1989); and, Sakthivel and Young (1991). 2 y and Yic are the observed and calculated intensities, respectively, at step i 3 w is the weight assigned each step intensity. ‘Jobs’ and ‘caic are the observed and calculated intensities, respectively, for Bragg reflection K. “obs’ is not actually observed but determined by assuming that the ‘observed’ intensity is in the same proportion as its calculated intensity. 5 N is the number of data points, P is the number of parameters refined and C is the number of constraints. 6 cs = (Hill and Flack, 1987) 157 Good agreement is assumed if is between 10%-20%. RB , on the other hand, is biased towards the calculated model and a RB value greater than 10% indicates large model fit errors. The reliability of the estimated standard deviations (esd, 2) calculated by the Rietveld refinement model has been questioned. The esd erroneously decrease as the number of observations increase, thereby causing serial correlation and an increased GofF. The reliability of esd can be determined by the Durbin-Watson d statistic (or d statistic), the numerical representation of the degree of serial correlation observed between points of the observed and calculated profiles. The d statistic is a sensitive measure of the progress of the refinement and is the most reliable error because it remains discriminating when other agreement indices fail (Hill and Flack, 1987). The d statistic can be either weighted or unweighted. An unweighted d statistic between 1.5 and 2.5 is obtained when essentially no serial correlation exists. Positive serial correlation is indicated by an unweighted d statistic less than 2.0; negative serial correlation occurs with an unweighted d statistic greater than 2.0. Limits for the weighted d statistic h.ave yet to be determined. In the least-squares refinement the following parameters can be adjusted simultaneously: 1. Lattice; 2. Atom position (x,y,z); 3. Atom site occupancy; 4. Atom thermal vibrational (isotropic or anisotropic); 5. Profile (U,V,W, and asymmetry); 6. Preferred orientation; 158 7. Background function; 8. 20-zero correction; 9. Overall scale (one for each phase); 10. Overall isotropic thermal (B). The input information required is as follows: 1. Initial values of all variable parameters; 2. Step-scan data in equal increments in 20; 3. 20 limits and excluded regions in the data; 4. Wavelength data; 5. Background specifications; 6. Space-group symbol; 7. Chemical symbol and valance of each atom; 8. Number of phases; 9. Profile function choice; 10. Profile cut-off (in units of HK); 11. Preferred orientation vector for each phase or preferred orientation function (i.e. Rietveld-Toraya or March Dollase); 12. Constraints; 13. Termination control; 14. Relaxation factors for the shifts (separately specified for four different groups of parameters); 15. Output control flags. The output includes: 159 1. The refinement conditions and subject so that a given run can be reconstructed unambiguously; 2. Adjustable-parameter final values, last shift and standard deviations; 3. R,, R, RB, R,, GofF and d-statistic values. And the following printouts are available: 1. Observed and calculated intensities; 2. Line-printer plot; 3. I F2 and R-Bragg, with or without I FKI obs, I F I caic and R-F; 4. Correlation matrix; 5. Reflection list for each phase; 6. Corrected data list, with w values; 7. Merged relection list; 8. Symmetry operators list; 9. Off-line plot (e.g. Calcomp or Verastec); 10. Stacked summary of cycle-by-cycle values or summary of only last-cycle parameters. 160 APPENDIX B CHEMICAL STRUCTURES AND SINGLE CRYSTAL INFORMATION Table 11 - Chemical structure, crystal system, space lattice, and space group of MTZ (Blaton et al., 1979). Chemical Structure CH2 OH CH3 3-D Structure Space Lattice monoclinic Space Group P2 1/c 161 Table 12 - Chemical structure, crystal system, space lattice, and space group of ASA (Wheatley, 1964; Kim et al., 1985). Chemical Structure COOH OCOCH3 3-D Structure Space Lattice monoclinic Space Group P2 i/c 162 Table 13 - Chemical structure, crystal system, space lattice, and space group of DPH (Camerman and Camerman, 1971). Chemical Structure 3-D Structure not available Space Lattice orthorhombic Space Group Pn2 ia 163 Table 14 - Chemical structure, crystal system, space lattice, and space group of CPZ(I)-H (Klein and Conrad, 1986). Chemical Structure ( cr)2 • H20 3-D Structure Space Lattice monoclinic Space Group P2 i/c 164 Table 15 - Chemical structure, crystal system, space lattice, and space group of CPZ(II) (Dorignac-Calas and Marsau, 1972). Chemical Structure CI- L 3-D Structure C. Space Lattice monoclinic Space Group P2 i/c 165 Figure 40. Simple monoclinic lattice. (Reproduced from Cullity, B.D., Elements ofX-ray Diffraction, Second Edition, Addison-Wesley Publishing Company, Reading, MA, 1978, p. 36.) Figure 41. Rectangular (orthorhombic) lattice. (Reproduced from Cullity, B.D., Elements ofX-ray Diffraction, Second Edition, Addison-Wesley Publishing Company, Reading, MA, 1978, p. 36.) 166 APPENDIX C PRELIMINARY SOLID-STATE NMR SPECTRA FOR CHLORPROMAZINE HYDROCHLORIDE (CPZ(II)) AND ITS GRANULES (CPZ(I)-H) 167 RUPI PPG CPCYCL. PC 18-9-91 DOTE SF 1BO.6l 01 2924.SOS St 4296 TO 1224 SW 58000.220 HZ/PT 24.l4 RD 30 NE 1 115 3228 TO 273 11W 111.8 FW 60200 02 8343.022 tIP 61100 00 20.0005 01 7.52011 02 80.800U 03 1.00011 04 1.000U LB 120.800 OB 0.11 NC 1 CX 20.02 Cr 10.00 SR -1768.27 120 PPM Figure 42. 400 MHz 13C-NMRspectrum of CPZ(II). __ 168 9999 99 199 91 PHRRNC15. N&2 Ru: RUE” PPQ: LPCEtL. PC ORTE 21-Y-Y1 Sr wu.Oi’l 01 2924.8U5 SI 4U96 ro i124 SW S2uu.Sc4 Hz/Pr 24.414 EEC NE I NS 2’i2 rs V5 OW 111.8 FU 622811 02 6547.11112 OP 6H 00 Ii 02 58.811115 01 7.588U 02 613.IE11EEU 05 1.RIEEEU 04 1.81311U LEE 1814.828 58 2.14 NC 14 cx zurn CY U4.142 58 —1768.2’ - 213a 1148 Figure 43. 400 MHz 13C-NMR spectrum of CPZ(I)-H’.