Relating the tableting behavior of piroxicam polytypes to their crystal structures using energy-vector models

Accepted Manuscript Relating the Tableting Behavior of Piroxicam Polytypes to Their Crystal Structures Using Energy-Vector Models Pratik P. Upadhyay, Changquan C. Sun, Andrew D. Bond PII: DOI: Reference:

S0378-5173(18)30187-X https://doi.org/10.1016/j.ijpharm.2018.03.040 IJP 17383

To appear in:

International Journal of Pharmaceutics

Received Date: Revised Date: Accepted Date:

2 January 2018 19 March 2018 23 March 2018

Please cite this article as: P.P. Upadhyay, C.C. Sun, A.D. Bond, Relating the Tableting Behavior of Piroxicam Polytypes to Their Crystal Structures Using Energy-Vector Models, International Journal of Pharmaceutics (2018), doi: https://doi.org/10.1016/j.ijpharm.2018.03.040

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Relating the Tableting Behavior of Piroxicam Polytypes to Their

2

Crystal Structures Using Energy-Vector Models

3

Pratik P. Upadhyaya, Changquan C. Sunb* and Andrew D. Bonda,c*

4 5

a

6

b

7

c

8

UK

Department of Pharmacy, University of Copenhagen, 2100 Copenhagen Ø, Denmark Department of Pharmaceutics, University of Minnesota, Minneapolis, MN 55455, USA

Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge, CB2 1EW,

9 10

*Corresponding authors:

11

Dr. Andrew D. Bond

12

Department of Chemistry, University of Cambridge, Lensfield Road,

13

Cambridge, CB2 1EW, UK

14

Tel: +44 (0)1223 336352, +44 (0)1223 762015

15

E-mail: [email protected]

16

Prof. Changquan C. Sun

17

Department of Pharmaceutics, University of Minnesota, Minneapolis, MN 55455, USA

18

E-mail: [email protected]

1

19

Abstract. Piroxicam crystallises into two polytypes, 1 and 2 with crystal structures that

20

contain identical molecular layers but differ in the way that these layers are stacked. In spite of

21

having close structural similarity, the polytypes have significantly different powder tabletting

22

behaviour: 2 forms only weak tablets at low pressures accompanied by extensive capping and

23

lamination, which make it impossible to form intact tablets above 100 MPa, while 1 exhibits

24

superior tabletability over the investigated pressure range (up to 140 MPa). The potential

25

structural origin of the different behaviour is sought using energy-vector models, produced from

26

pairwise intermolecular interaction energies calculated using the PIXEL method. The analysis

27

reveals that the most stabilising intermolecular interactions define columns in both crystal

28

structures. In 2, a strongly stabilising interaction between inversion-related molecules links

29

these columns into a 2-D network, while no comparable interaction exists in 1. The higher

30

dimensionality of the energy-vector model in 2 may be one contributor to its inferior

31

tabletability. A consideration of probable slip planes in the structures identifies regions where the

32

benzothiazine groups of the molecules meet. The energy-vector models in this region are

33

geometrically similar for both structures, but the interactions are more stabilising in 2 compared

34

to 1. This feature may also contribute to the inferior tabletability of 2.

35

Keywords. Piroxicam; polytypes; compaction; tablet and tabletability; PIXEL.

36

Abbreviations

37

API, Active Pharmaceutical Ingredient; PXRD, Powder X-Ray Diffraction; HPLC, High

38

Pressure Liquid Chromatography; DSC, Differential Scanning Calorimetry; CSD, Cambridge

39

Structural Database.

2

40

1. Introduction

41

Mechanistic understanding of the deformation behaviour in organic crystals has long been a

42

topic of interest in the pharmaceutical field, primarily due to the need to predict the performance

43

of APIs in various processes such as milling and tabletting.(Datta and Grant, 2004; Sun, 2009)

44

Efforts have been made to develop an understanding of the relationship between deformation

45

behaviour and crystal structure by visualising structures, often with a focus on hydrogen bonds

46

(or other specific interactions) and by identifying probable slip planes. A robust understanding in

47

this area should eventually lead to better product design by imparting desired material properties

48

by means of crystal engineering.

49

Polymorphs, having the same chemical composition but different packing arrangements and/or

50

conformation of molecules in the crystal, are particularly useful to develop crystal structure-

51

property relationships.(Bag et al., 2012; Joiris et al., 1998; Khomane et al., 2012; Khomane et al.,

52

2013; Roberts and Rowe, 1996; Sun and Grant, 2001c; Upadhyay et al., 2013) For example,

53

Roberts et al. have analysed the relationship between crystal structure, bulk Young’s modulus

54

and compact strength for carbamazepine and sulfathiazole polymorphs.(Roberts and Rowe,

55

1996) It was found for carbamazepine that form III has a relatively open structure in comparison

56

to form I. Form III was correspondingly easy to deform and produced weaker tablets while form

57

I was difficult to deform but produced stronger compacts. For sulfathiazole, forms I and III

58

showed similar Young’s moduli despite different hydrogen bonding patterns, which was

59

attributed to a similar number of hydrogen bonds per molecule with similar strengths. Later,

60

Joiris et al showed that orthorhombic paracetamol has better tabletability owing to the presence

61

of clear slip planes that provide greater plasticity compared to the monoclinic form.(Joiris et al.,

62

1998) Similarly, Sun and Grant examined the deformation behaviour of sulfamerazine

3

63

polymorphs, where the better tabletability of form I compared to form II was attributed to easy

64

slip of the layered structure in form I, in contrast to the corrugated structure of form II.(Sun and

65

Grant, 2001c) For ranitidine hydrochloride, form I exhibited worse compressibility than the more

66

open structure of form II, but better tabletability.(Upadhyay et al., 2013) Similar observations

67

were reported for clopidogrel bisulphate polymorphs,(Khomane et al., 2012) and indomethacin

68

polymorphs.(Khomane et al., 2013) Thus, visual identification of apparent slip planes in crystals,

69

and probable associated plasticity, is not always reliable to rank tabletability.(Khomane and

70

Bansal, 2013)

71

The expectation that layered structures are generally easy to deform and exhibit good tabletting

72

properties has been used to improve the tabletting performance of crystals such as paracetamol

73

and methyl gallate, by synthesising co-crystals with a layered structure.(Chattoraj et al., 2010;

74

Karki et al., 2009) In other co-crystallization cases, for example piroxicam and saccharin,

75

tabletting properties deteriorated in the co-crystal because it did not display a layered

76

structure.(Chattoraj et al., 2014) Similar investigations have identified differences between the

77

tabletting properties of hydrates compared to anhydrous forms. For example, both para-

78

hydroxybenzoic acid monohydrate and anhydrate have a corrugated layer structure. The water of

79

crystallisation in the monohydrate facilitates sliding of corrugated layers and imparts plasticity,

80

which leads to improved tabletability compared to the anhydrous form.(Sun and Grant, 2004)

81

These studies are typical in that crystal structures are analysed by comparing intermolecular

82

interactions (usually hydrogen bonding) and identifying slip planes.

83

In the present paper, we consider the tabletting behaviour of piroxicam polytypes, and discuss

84

how this may be related to crystal structure using energy-vector models, produced from PIXEL

85

(Gavezzotti, 2011) calculations using the processPIXEL (Bond, 2014) program. The piroxicam

4

86

polytypes,(International Union of Crystallography) which contain identical 2-D layers but differ

87

in the way that these layers are stacked, provide an interesting opportunity to study the influence

88

of apparently subtle structural differences on the tabletting behaviour in this layered crystalline

89

system.

90

2. Experimental Section

91

2.1 Materials

92

Piroxicam was purchased from Chr. Olesen Pharmaceuticals A/S (Gentofte, Denmark). The

93

crystalline form was confirmed by PXRD to be form I. For crystallisation, HPLC grade methanol

94

and ethanol were purchased from Sigma-Aldrich.

95

2.2 Bulk crystallisation of the polytypes

96

Piroxicam polytype 1 was crystallised by crash cooling a hot solution of piroxicam (13 mg/mL)

97

in methanol. Polytype 2 was prepared by slow cooling of a hot solution of piroxicam (10

98

mg/mL) in ethanol. Both precipitated solutions were allowed to stand at room temperature

99

overnight before being vacuum filtered. Crystallised1 and 2 were dried in an oven at 50°C for

100

24 hours and the polymorphic form was identified by PXRD. The dried powders were used

101

directly for compression.

102

2.3 Powder X-Ray Diffraction (PXRD)

103

Powder X-ray diffraction (PXRD) data were collected on a Panalytical X’Pert Pro instrument

104

(Panalytical, Almelo, The Netherlands) equipped with a PIXcel detector using non-

105

monochromated CuK radiation (λ = 1.5418 Å). The sample was placed in a zero-background Si

106

holder and measured in reflection geometry with sample spinning.

5

107

2.4 Differential Scanning Calorimetry (DSC)

108

Thermal behaviour was characterised using a differential scanning calorimeter (Discovery DSC,

109

TA Instruments, New Castle, USA). Samples were accurately weighed in Tzero aluminium pans

110

and heated from 25 to 200°C with a heating rate of 10 °C/min under a flow of N2 gas at a rate of

111

50 mL/min. Data analyses, including melting point determination, were performed using the

112

Trios software.

113

2.5 Tablet preparation

114

Approximately 100 mg of powder was compressed under different pressures using a laboratory

115

press (Gamlen, Nottingham, UK). Flat-faced round tablets were prepared (6 mm diameter) at a

116

speed of 60 mm/min. Compression was applied using a 500 kg load cell. Prior to compression,

117

the punch and die were lubricated using 1% magnesium stearate in acetone.

118

2.6 Tablet characteristics

119

Compressibility, tabletability, and compactability are widely used to compare the tabletting

120

properties of powders.(Feng et al., 2007; Sun, 2006) Compressibility describes the ease of

121

powder volume reduction by compaction pressure, which is characterised by a plot of porosity

122

against applied pressure. Porosity of the compacts () can be calculated using equation (1):

123

 = 1 – (c / t)

(1)

124

where c is the density of the compact (calculated from tablet weight and volume) and t is the

125

true density of the material (as calculated from the crystal structure). The volume of the

126

cylindrical tablets was calculated from thickness and diameter measured by a digital calliper

127

(Mitutiyo Japan) accurate to 0.01 mm. A powder exhibits higher compressibility when tablet

6

128

porosity is lower at a given pressure. During this volume reduction, particles may undergo elastic

129

deformation, plastic deformation, and/or fragmentation, all of which influence the bonding area

130

among particles and, thus, the tablet tensile strength.(Osei-Yeboah et al., 2016) Heckel analysis

131

can be performed using compressibility data to obtain the mean yield pressure (Py), which is the

132

reciprocal of the slope of the linear portion of the plot of –ln() against compaction

133

pressure.(Heckel, 1961a, b; Sun, 2016) A material with lower Py is deemed to be more

134

plastic.(Sonnergaard, 1999)

135

Diametrical tablet breaking strength was measured using the same Gamlen press used to prepare

136

tablets, using a 50 kg load cell rather than 500 kg.

137

The tensile strength of the tablet, σ, was calculated from the crushing strength using equation (2):  = (2 F) / πd t

138

(2)

139

Where F is the observed breaking force (N), d is the diameter (mm), and t is the thickness of the

140

compact (mm). The units of  are N/mm2.

141

Tabletability is a measure of the ability of a powder to produce tablets of sufficient tensile

142

strength. It is characterised by a plot of tensile strength against compaction pressure. At a given

143

pressure, a powder that forms tablets with higher tensile strength is considered to have superior

144

tabletability. Finally, compactibility is the measure of the ability of a powder to form compacts

145

of sufficient strength by pore elimination. It is characterised by a plot of tensile strength against

146

porosity.

147

3. Results and discussions

148

3.1 Solid-state characterization

7

149

Piroxicam polytypes 1 and 2 crystallised as fine needles. Pawley fits to PXRD patterns of the

150

dried powders indicated phase purity, as described in a previous report (Upadhyay and Bond,

151

2015). Phase purity was further established by DSC, by which 1 showed a melting onset at

152

195.3°C while 2 showed a melting onset at 198.6°C, in agreement with the reported melting

153

onsets of the respective pure phases.(Upadhyay and Bond, 2015) Since both powders have the

154

same needle-like shape, they were gently crushed using a mortar and pestle to minimise the

155

impact of particle size on the powder compression.(Sun and Grant, 2001a, b) Tabletting data

156

were also collected for the unground powders to assess the effect of grinding on the tabletting

157

behaviour. The possibility for phase transformation under grinding was ruled out by re-

158

characterisation of the ground powders using both PXRD and DSC (Fig. 1).

159

3.2 Polytypic relationship and probable slip planes

160

Polytypism is a special case of polymorphism, where two polymorphs have similar 2-D layers

161

but differ in the way that these layers are arranged. The piroxicam polytypes 1 and 2

162

crystallise respectively

163

crystallographic information as indicated in Table 1. The chosen unit-cell settings correspond to

164

our previous crystallographic report on this system.(Upadhyay and Bond, 2015)

165

The consistent 2-D layers within the polytypes lie parallel to the bc planes. The O–H group of

166

each molecule (Scheme 1) is involved in an intramolecular hydrogen bond, and the N–H group

167

is shielded from formation of any hydrogen bond, so that there are no specific intermolecular

168

contacts that stand out in either structure. The difference between the polytypes is in the way that

169

the layers are arranged along the a axis (Figures 2 and 3). The structure of1 is non-

170

centrosymmetric and polar, where all molecules face in only one direction along the c axis, while

in the orthorhombic and monoclinic crystal systems, with

8

171

the structure of 2 is centrosymmetric with inversion centres between layers. There are two

172

different interlayer regions in each structure: (1) a region where the pyridyl ends of the

173

molecules meet; (2) a region where the benzothiazine ends of the molecules meet. The molecules

174

are clearly “interdigitated” at the pyridyl region (i.e. one layer protrudes into the other), but much

175

less so at the benzothiazine region (Fig. 2 and 3). Thus, the pyridyl region would appear to be

176

“interlocked”, and slip in these structures would be anticipated to occur primarily at the

177

benzothiazine regions.

178

3.3 Compaction behaviour of the polytypes

179

Tabletability is an important quality parameter because it is necessary to produce tablets with

180

sufficient tensile strength to withstand external shocks during handling and transportation. For

181

piroxicam, polytype 1 shows superior tabletability over 2 (Fig. 4). At 137 MPa, 2 showed

182

extensive capping and lamination upon ejection from the die (Fig. SI1), while 1 continued to

183

form strong intact tablets even at 137 MPa. The tabletability of the initially crystallised 1 is

184

slightly better than the ground powder (Fig. 4). However, grinding did not lead to any noticeable

185

change in tabletability for 2 (Fig. 4). Thus, differences in particle size do not explain the

186

different tabletability of the two polytypes. The different tabletability between the ground and

187

unground samples of 1 may be the result of some preferred orientation during compaction.

188

A similar trend is seen in the compactability, where for a given porosity, 1 has higher tensile

189

strength than 2 (Fig. 5a). The compressibility, on the other hand (Fig. 5b), is substantially

190

similar for the two polytypes, except for a minor difference at the lowest pressures below 60

191

MPa. The compression behaviour is also reflected on the Heckel plot (Fig. SI2) which yields

192

similar Py values: 106.38 MPa for 1 and 117.65 MPa for 2.

9

193

3.4 Structure-property correlation using energy-vector models

194

We have previously reported (Upadhyay and Bond, 2015) intermolecular interaction energies for

195

the 1 and 2 polytypes, calculated using the PIXEL method.(Gavezzotti, 2011) A summary table

196

is provided in the Supporting Information. The sum of all interaction energies (i.e. the total

197

cohesive energy in the crystal structure) is essentially identical for the two polytypes, but there

198

are differences in the way that the interactions are distributed. A convenient way to visualise the

199

situation is to use energy-vector models. These display pairwise intermolecular interactions as

200

vectors between the molecular centroids, scaled to reflect the relative interaction energies. The

201

most stabilising interaction in the structure is displayed as a complete line, and less stabilising

202

interactions are scaled proportionally so that gaps appear in the lines between molecules. The

203

concept and details are provided by Shishkin et al.,(Shishkin et al., 2012) who refer to the

204

resulting models informally as hedgehogs. A similar approach has been implemented by Turner

205

et al.,(Turner et al., 2015) who refer to energy frameworks in which the molecular centroids are

206

joined by complete lines and the interaction energies are differentiated by the line thickness.

207

Either visualisation technique can be applied using any quantitative calculation of the interaction

208

energy. Our energy-vector diagrams are calculated using the processPIXEL program (Bond,

209

2014) and visualised using Mercury.(Macrae et al., 2008) It is important to note that the

210

represented energies are between whole molecules, rather than highlighting specific group

211

interactions such as hydrogen bonds.

212

The complete energy-vector diagrams (Fig. 6) indicate a subtle difference in the effective

213

dimensionality of the structures. The energy-vector models resemble columns along the c axis,

214

with no substantial pairwise intermolecular interaction linking between adjacent columns along

215

the b axis. Thus, the models appear more like a “checkerboard” than a clearly layered network.

10

216

In 2, strongly stabilising interactions (marginally the most stabilising interactions in the

217

structure) exist between inversion-related molecules across the pyridyl region, thus defining a 2-

218

D network spanning across two layers of molecules in the 2 polytype. No similar interaction

219

exists in the 1 polytype. The different dimensionality of the energy frameworks is contrary to

220

the instinctive view of 2-D structural similarity in the polytypes, and the higher dimensionality of

221

2 may contribute to its inferior tabletability compared to 1.

222

In the earlier analysis, the benzothiazine regions were highlighted as the most likely principal

223

slip planes. It is clear from the gaps in both energy frameworks that interactions across this

224

region are relatively less stabilising. Considering all interactions across the benzothiazine region

225

(as listed in the Supporting Information), the total interaction energy is clearly less stabilising

226

than the other regions considered. This analysis is essentially equivalent to an attachment energy

227

calculation, and supports the identification of the benzothiazine region as the key area for slip.

228

Differences between the polytypes can be explored in more detail by constructing energy-vector

229

models using only intermolecular interactions across the benzothiazine region. These models

230

(Fig. 7) have basically identical shapes in the two polytypes, resembling a square grid in

231

projection onto the bc plane. In 2, however, the most stabilising interaction is –28.2 kJ/mol,

232

compared to –20.1 kJ/mol in 1. If the two diagrams are scaled to the common maximum

233

magnitude of –28.2 kJ/mol (as in Fig. 7), it is clear that the energy-vector model in the

234

benzothiazine region of 1 has the largest gaps. The presence of more stabilising interactions

235

across these likely slip planes is also consistent with the inferior tabletability of 2.

236

4. Conclusion

11

237

Bulk samples of piroxicam polytypes 1 and 2 show distinct tabletting behaviour in spite of the

238

substantial similarity in their crystal structures. In this system, visual inspection of the most

239

probable slip planes does not clearly explain why the two polytypes should have different

240

tabletting behaviour. Analysis of energy-vector models provides two potential insights: (1) the

241

most stabilising intermolecular interaction energies define a 2-D framework in 2, compared to

242

1-D columns in 1; (2) at the most probable slip planes, the geometrical arrangement of the

243

intermolecular interactions is comparable in the two structures, but the interactions are found to

244

be significantly more stabilising in 2. Both of these features are consistent with the superior

245

tabletability of 1 compared to 2. Although it cannot reasonably be claimed that the energy-

246

vector models provide a complete solution to a complex problem, this example indicates that

247

they can be useful to investigate structure-property relationships in the context of powder

248

compression.

249

12

250

Acknowledgements

251

This work was funded by the Danish Council for Independent Research | Natural Sciences

252

(DFF-1323-00122) and Department of Pharmacy, University of Copenhagen. The Lundbeck

253

Foundation (Denmark) is thanked for funding a Visiting Professorship to Prof. Sun (R143-2014-

254

25).

255

Supporting information.

256

Additional data on the Heckel analysis, PIXEL interactions and images of laminated tablets are

257

provided in supplementary information

258

References

259

Bag, P.P., Chen, M., Sun, C.C., Reddy, C.M., 2012. Direct correlation among crystal structure,

260

mechanical behaviour and tabletability in a trimorphic molecular compound. CrystEngComm 14,

261

3865-3867.

262

Bond, A.D., 2014. processPIXEL: a program to generate energy-vector models from

263

Gavezzotti's PIXEL calculations. J. Appl. Cryst. 47, 1777-1780.

264

Chattoraj, S., Shi, L., Chen, M., Alhalaweh, A., Velaga, S., Sun, C.C., 2014. Origin of

265

Deteriorated Crystal Plasticity and Compaction Properties of a 1:1 Cocrystal between Piroxicam

266

and Saccharin. Cryst. Growth Des. 14, 3864-3874.

267

Chattoraj, S., Shi, L., Sun, C.C., 2010. Understanding the relationship between crystal structure,

268

plasticity and compaction behaviour of theophylline, methyl gallate, and their 1 : 1 co-crystal.

269

CrystEngComm 12, 2466-2472.

270

International

271

http://reference.iucr.org/dictionary/Polytypism

Union

of

Crystallography,

Online

Dictionary

of

Crystallography.

13

272

Datta, S., Grant, D.J.W., 2004. Crystal structures of drugs: advances in determination, prediction

273

and engineering. Nat. Rev. Drug Discov. 3, 42-57.

274

Feng, Y., Grant, D.J.W., Sun, C.C., 2007. Influence of crystal structure on the tableting

275

properties of n‐alkyl hydroxybenzoate esters (parabens). J Pharm. Sci. 96, 3324-3333.

276

Gavezzotti, A., 2011. Efficient computer modeling of organic materials. The atom-atom,

277

Coulomb-London-Pauli

278

dispersion and repulsion energies. New J. Chem. 35, 1360-1368.

279

Heckel, R.W., 1961a. An Analysis of powder compaction phenomena. Transactions of the

280

Metallurgical Society of AIME 221, 1001-1008.

281

Heckel, R.W., 1961b. Density-Presure relationships in powder compaction. Transactions of the

282

Metallurgical Society of AIME 221, 671-675.

283

Joiris, E., Martino, P.D., Berneron, C., Guyot-Hermann, A.-M., Guyot, J.-C., 1998. Compression

284

Behavior of Orthorhombic Paracetamol. Pharm. Res. 15, 1122-1130.

285

Karki, S., Friščić, T., Fábián, L., Laity, P.R., Day, G.M., Jones, W., 2009. Improving Mechanical

286

Properties of Crystalline Solids by Cocrystal Formation: New Compressible Forms of

287

Paracetamol. Adv. Mater. 21, 3905-3909.

288

Khomane, K.S., Bansal, A.K., 2013. Weak Hydrogen Bonding Interactions Influence Slip

289

System Activity and Compaction Behavior of Pharmaceutical Powders. J. Pharm. Sci. 102,

290

4242-4245.

291

Khomane, K.S., More, P.K., Bansal, A.K., 2012. Counterintuitive Compaction behavior of

292

Clopidogrel Bisulfate Polymorphs. J. Pharm. Sci. 101, 2408-2416.

293

Khomane, K.S., More, P.K., Raghavendra, G., Bansal, A.K., 2013. Molecular Understanding of

294

the Compaction Behavior of Indomethacin Polymorphs. Mol. Pharm. 10, 631-639.

(AA-CLP)

model

for

intermolecular

electrostatic-polarization,

14

295

Macrae, C.F., Bruno, I.J., Chisholm, J.A., Edgington, P.R., McCabe, P., Pidcock, E., Rodriguez-

296

Monge, L., Taylor, R., van de Streek, J., Wood, P.A., 2008. Mercury CSD 2.0 - new features for

297

the visualization and investigation of crystal structures. J. Appl. Cryst. 41, 466-470.

298

Osei-Yeboah, F., Chang, S.-Y., Sun, C.C., 2016. A critical Examination of the Phenomenon of

299

Bonding Area - Bonding Strength Interplay in Powder Tableting. Pharm. Res. 33, 1126-1132.

300

Roberts, R.J., Rowe, R.C., 1996. Influence of polymorphism on the Young's modulus and yield

301

stress of carbmazepine, sulfathiazole and sulfanilamide. Int. J. Pharm. 129, 79-94.

302

Shishkin, O.V., Dyakonenko, V.V., Maleev, A.V., 2012. Supramolecular architecture of crystals

303

of fused hydrocarbons based on topology of intermolecular interactions. CrystEngComm 14,

304

1795-1804.

305

Sonnergaard, J.M., 1999. A critical evaluation of the Heckel equation. Int. J. Pharm. 193, 63-71.

306

Sun, C., Grant, D.J.W., 2001a. Effects of initial particle size on the tableting properties of l-

307

lysine monohydrochloride dihydrate powder. Int. J. Pharm. 215, 221-228.

308

Sun, C., Grant, D.J.W., 2001b. Influence of Crystal Shape on the Tableting Performance of L-

309

Lysine Monohydrochloride Dihydrate. J. Pharm. Sci. 90, 569-579.

310

Sun, C., Grant, D.J.W., 2001c. Influence of Crystal Structure on the Tableting Properties of

311

Sulfamerazine Polymorphs. Pharm. Res. 18, 274-280.

312

Sun, C., Grant, D.J.W., 2004. Improved Tableting Properties of p-Hydroxybenzoic Acid by

313

Water of Crystallization: A Molecular Insight. Pharm. Res. 21, 382-386.

314

Sun, C.C., 2006. A material-sparing method for simultaneous determination of true density and

315

powder compaction properties—Aspartame as an example. Int. J. Pharm. 326, 94-99.

316

Sun, C.C., 2009. Materials Science Tetrahedron—A Useful Tool for Pharmaceutical Research

317

and Development. J. Pharm. Sci. 98, 1671-1687.

15

318

Sun, C.C., 2016. Microstructure of Tablet—Pharmaceutical Significance, Assessment, and

319

Engineering. Pharm. Res., 1-11.

320

Turner, M.J., Thomas, S.P., Shi, M.W., Jayatilaka, D., Spackman, M.A., 2015. Energy

321

frameworks: insights into interaction anisotropy and the mechanical properties of molecular

322

crystals. Chem. Comm. 51, 3735-3738.

323

Upadhyay, P., Khomane, K.S., Kumar, L., Bansal, A.K., 2013. Relationship between crystal

324

structure and mechanical properties of ranitidine hydrochloride polymorphs. CrystEngComm 15,

325

3959-3964.

326

Upadhyay, P.P., Bond, A.D., 2015. Crystallization and disorder of the polytypic 1 and 2

327

polymorphs of piroxicam. CrystEngComm 17, 5266-5272.

328

16

329

Table 1. Crystallographic information for piroxicam polytypes 1 and 2. Space group Pbc21 is

330

a non-standard setting of space group type Pca21.

Polytype Space group

1

Pbc21

2

P21/c

Unit-cell dimensions (Å, °) a =17.3964(14), b = 11.7965(10), c = 6.9851(5)  =   90 a =17.577(3), b = 11.745(2), c = 6.8516(14)  =  90; 98.07(1)

CSD Refcode BIYSEH08

BIYSEH09

331

17

332

Figure Legends

333

Scheme 1. Molecular structure of piroxicam.

334

Fig. 1. Baseline characterization of the dried milled powders of piroxicam polytypes1 and 2:

335

(a) PXRD patterns and (b) DSC thermograms. The PXRD patterns of 1 and 2 are closely

336

comparable, but distinguishable by several characteristic lines. The DSC thermogram of 1

337

shows initial melting, followed by recrystallization and subsequent melting of form I.(Upadhyay

338

and Bond, 2015)

339

Fig. 2. Projection of the 1/2 structure along the c axis: the polytypes are indistinguishable in

340

this projection. Molecules lie in layers parallel to the bc plane (vertical in the diagram). The

341

pyridyl and benzothiazine interlayer regions are highlighted.

342

Fig. 3. Projection along the b axis for the 1 and 2 structures. 1 is non-centrosymmetric and

343

polar, with all molecules pointing in the same direction along the c axis. 2 is centrosymmetric.

344

Fig. 4. Tabletability plot for the piroxicam polytypes. Polytype 1 shows better tabletability than

345

2. A small difference is observed between the ground and unground powders for 1.

346

Fig. 5. (a) Compactability plot of piroxicam polytypes. (b) Compressibility plot.

347

Fig. 6. Energy-vector models viewed along the c axis (comparable to Fig. 2). Both structures

348

contain identical columns (along c) that adopt a “checkerboard pattern”. (a) In , the

349

interactions between columns are relatively less stabilising; (b) In , strongly stabilising

350

interactions link the columns into 2-D networks in the bc planes, spanning across the pyridyl

351

region.

18

352

Fig. 7. Views of the energy-vector models in the benzothiazine region for 1 (top) and 2

353

(bottom). Both models are scaled to the strongest interaction in 2 (–28.2 kJ/mol). The gaps in

354

the lines in the 1 structure show that the interactions in 1 are less stabilising than those in 2.

355

19

356 357

20

358 359

21

360 361

22

362 363

23

364 365

24

366 367

25

368 369

26

370

27

371

372 373

28

374

29