International Journal of Pharmaceutics 472 (2014) 140–147
`
`Contents lists available at ScienceDirect
`
`International Journal of Pharmaceutics
`
`j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i j p h a r m
`
`Effect of crystal habits on the surface energy and cohesion of crystalline
`powders
`Umang V. Shah a, Dolapo Olusanmi b, Ajit S. Narang b, Munir A. Hussain b,
`John F. Gamble c, Michael J. Tobyn c, Jerry Y.Y. Heng a,*
`a Surfaces and Particle Engineering Laboratory (SPEL), Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7
`2AZ, UK
`b Bristol–Myers Squibb Pharmaceuticals, 1 Squibb Drive, New Brunswick, NJ 08903, USA
`c Bristol–Myers Squibb Pharmaceuticals, Reeds Lane, Moreton, Wirral CH46 1QW, UK
`
`A R T I C L E
`
`I N F O
`
`A B S T R A C T
`
`Article history:
`Received 4 April 2014
`Received in revised form 3 June 2014
`Accepted 7 June 2014
`Available online 10 June 2014
`
`Keywords:
`Crystal habit
`Surface energy heterogeneity
`Macroscopic single crystal
`Solvent polarity
`Crystal aspect ratio
`Cohesion
`
`The role of surface properties, influenced by particle processing, in particle–particle interactions (powder
`cohesion) is investigated in this study. Wetting behaviour of mefenamic acid was found to be anisotropic
`by sessile drop contact angle measurements on macroscopic (>1 cm) single crystals, with variations in
`contact angle of water from 56.3 to 92.0. This is attributed to variations in surface chemical
`functionality at specific facets, and confirmed using X-ray photoelectron spectroscopy (XPS). Using a
`finite dilution inverse gas chromatography (FD-IGC) approach, the surface energy heterogeneity of
`powders was determined. The surface energy profile of different mefenamic acid crystal habits was
`directly related to the relative exposure of different crystal facets. Cohesion, determined by a uniaxial
`compression test, was also found to relate to surface energy of the powders. By employing a surface
`modification (silanisation) approach, the contribution from crystal shape from surface area and surface
`
`“normalising” contribution from surface energy and surface area, needle
`energy was decoupled. By
`2.5 more cohesive compared to elongated plates or hexagonal
`shaped crystals were found to be
`
`cuboid shapes crystals.
`
`ã
`
` 2014 Elsevier B.V. All rights reserved.
`
`1. Introduction
`
`The role of the physicochemical properties of particulate
`flow
`pharmaceutical materials on their cohesion and powder
`properties has attracted extensive research interest in the past four
`al., 2004; Lam and
`decades (Feng et al., 2007; Kaerger et
`Nakagawa, 1994; Podczeck and Mia, 1996; Podczeck and Révész,
`1993; Ridgway and Morland, 1977). Understanding the role of
`physicochemical properties on cohesion and the development of
`strategies to control cohesion by tailoring the properties of
`pharmaceutical materials may be critically important for efficient
`and cost effective processing (Hou and Sun, 2008). The effect of
`flow and cohesion is referred
`particle shape and size on powder
`extensively in the literature (Jones et al., 2003). Moreland and
`first to report the effect of particle shape on bulk
`Ridgway were the
`
`* Corresponding author at: ACEX 417a, Department of Chemical Engineering,
`Imperial College London, South Kensington Campus, London SW7 2AZ, UK. Tel.: +44
`207 594 0784; fax: +44 207 594 5700.
`E-mail address: jerry.heng@imperial.ac.uk (J.Y.Y. Heng).
`
`http://dx.doi.org/10.1016/j.ijpharm.2014.06.014
`0378-5173/ã
` 2014 Elsevier B.V. All rights reserved.
`
`density (Ridgway and Morland, 1977). Podczeck and Mia reported
`the effect of particle size and shape on the Hausner ratio and angle
`of internal friction (Podczeck and Mia, 1996). They found particles
`with higher aspect ratio (needle shaped crystal), showed a higher
`angle of internal friction. Kaerger et al. investigated the effect of
`flow and compaction behaviour of
`particle shape and size on
`blends, reporting that blends containing spherical paracetamol
`particles (prepared using crystallisation by sonication) with
`flow properties compared
`microcrystalline cellulose had improved
`to micronised particles (Kaerger et al., 2004). Gamble et al.
`investigated the effect of different sub-populations, e.g. agglom-
`erates and primary particles, highlighting the effect of the presence
`flow properties of bulk primary
`of agglomerates in enhancing the
`particles (Gamble et al., 2011). Di Martino et al. studied the effect
`flow
`of different crystal shapes of ibuprofen on compression and
`properties, highlighting improved densification of the smooth coin
`type crystal habit compared to other crystals habits, which were
`attributed to the increase in powder bed porosity.
`Crystals of the same polymorphic form with different crystal
`shapes (habits) can be obtained by varying the relative growth
`rates of the different crystals facets. This in turn can be dependent
`
`Merck Exhibit 2250, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`U.V. Shah et al. / International Journal of Pharmaceutics 472 (2014) 140–147
`
`141
`
`on material intrinsic properties or be affected by properties of the
`inhibitors or additives
`crystallisation solvent, crystal growth
`(Berkovitch-Yellin, 1985; Bourne and Davey, 1976; Lovette et al.,
`2008). Crystal habits not only determine the main bulk properties
`flowability and
`of the crystalline material (e.g. bulk density,
`mechanical strength) but also alter the surface energy (Ho et al.,
`2012).
`Surface energy of crystalline pharmaceutical materials has been
`shown to be anisotropic (Heng et al., 2006a, 2007; Ho et al., 2008,
`2010). Facet specific surface energy of organic crystalline material
`was directly correlated with the chemical functional groups
`exposed on the crystal facet surfaces using contact angle and X-
`ray photoelectron spectroscopy (XPS) measurements. Considering
`the facet specific surface energy of a crystalline material, it is
`postulated that surface energetics of the crystalline powders
`depend on the relative surface energy contributions of the
`different crystal facets. Using D-mannitol as a model system, Ho
`et al. demonstrated that decreasing the aspect ratio of needle
`in a decreasing shift
`in the overall
`shaped crystals resulted
`contribution of the dispersive component of the surface energy.
`This was attributed to the increasing contribution of the facet
`showing lowest dispersive component of surface energy (facet
`(0 11)) (Ho et al., 2012). The ability to tailor crystal habits will result
`in the change in the relative contribution of different crystal facets.
`This change will result in a dissimilar surface energy of the
`impact on powder
`crystalline powders, which may have an
`cohesion.
`Although it is well known that particle–particle interactions,
`e.g. cohesion/adhesion, are governed by the surface, the effect of
`flow properties have not considered the
`crystal shapes on powder
`impact of the anisotropic surface properties of the crystals.
`Considering that particle shape, surface energy and surface area
`flow properties, it is essential to understand the
`can influence
`contribution of each of these factors.
`firstly study the effect of crystal habit on
`This study aims to
`surface energy and cohesion of crystalline pharmaceutical
`powders. Secondly, an approach to decouple the contributions
`of surface energy and particle shape is presented. Mefenamic Acid,
`is used as a model
`a non-steroidal, anti-inflammatory drug,
`compound. Macroscopic single crystals of mefenamic acid are
`grown and used for determining facet specific surface energies.
`is then correlated to the surface energy heterogeneity
`This
`measurements of crystalline powders of mefenamic acid crystal-
`lised in different crystal habits. Cohesion values of different shape
`mefenamic crystal were measured using a uniaxial compression
`test. Results were correlated with surface energy and crystal shape
`to elucidate their respective effect on cohesion. Further, the effect
`“normalised”
` with silanisation
`of surface energy on cohesion was
`of mefenamic acid, allowing de-coupling of the contribution of
`crystal shape on cohesion from that of surface energy and surface
`area.
`
`Loughborough, UK) were used as probe liquids for contact angle
`measurements. n-hexane (>99.0%), ethyl acetate (>99.5%) and
`dichloromethane (>99.5%) were obtained from VWR BDH Prolabo,
`Lutterworth, UK), whereas n-heptane (99.0), n-octane (99.0%),
`n-nonane (99.0%) and n-decane (99.0%) were obtained from
`Sigma–Aldrich, Dorset, UK and used without further purification as
`probe liquids in inverse phase gas chromatography.
`
`3. Methods
`
`3.1. Growth of mefenamic acid macroscopic single crystal
`
`Mefenamic acid seed crystals of a few millimetres in size were
`obtained from slow evaporation of a supersaturated solution of
`mefenamic acid in acetone or methanol at room temperature.
`Macroscopic single crystals were obtained by slow solvent
`evaporation from saturated methanol solution at room tempera-
`ture. A saturated solution of mefenamic acid in methanol was
`prepared under constant stirring. A single seed of mefenamic acid
`fibre and suspended in the
`crystal was tied with an aramid
`saturated solution, which is kept without stirring. Slow evapora-
`tion of the solvent was maintained resulting in the crystal growth.
`in methanol was
`The saturated solution of mefenamic acid
`periodically changed. Macroscopic single crystals of mefenamic
`acid obtained were dried under ambient conditions and used for
`further characterisation and contact angle measurements.
`
`3.2. Crystallisation of mefenamic acid from different solvent systems
`
`in seven different
`Saturated solutions of mefenamic acid
`organic solvents with varying polarity were prepared at 50 C. A
`single step cooling profile was adopted. Saturated solutions of
`mefenamic acid at 50 C were transferred to an incubator (Surface
`Measurement Systems Ltd. London, UK) maintained at 4 C.
`filtered through
`Mefenamic acid crystals obtained after 24 h were
`filter paper (Whatman, UK) and dried
`general-purpose laboratory
`under ambient conditions. Dried crystals were stored in a glass
`container and used for further characterisation.
`
`3.3. Silanisation of mefenamic acid powders
`
`Recrystallised mefenamic acid powders were silanised using a
`protocol reported in the literature (Al-Chalabi et al., 1990). In a
`typical process, 500 mg of mefenamic acid powder was added to a
`50 mL 5% (v/v) solution of dichlorodimethylsilane in cyclohexane.
`The mixture was refluxed at 80 C for 24 h. Then, the reaction
`filtered
`mixture is allowed to cool down to room temperature and
`filter paper (Whatman, UK)
`using general-purpose laboratory
`followed by drying in a vacuum oven at 80 C for 4 h. Post
`silanisation, the silanised mefenamic acid powders were stored in
`a glass vial at ambient conditions.
`
`2. Materials
`
`3.4. Single crystal X-ray diffraction for crystal facet indexing
`
`Mefenamic acid (2-(2, 3-dimethylphenyl) amino benzoic acid)
`(99.0% Sigma–Aldrich, Dorset, UK), acetone (>99.5%) and methanol
`(>99.5%) were obtained from VWR BDH Prolabo, Lutterworth, UK
`and used for growth of a macroscopic single crystal of mefenamic
`acid. Toluene (>99.5%), diethyl ether (>99.5%), ethyl acetate
`(>99.5%), dichloromethane (>99.0%), acetone (>99.5%), isopropyl
`alcohol (>99.5%) and methanol (>99.5%) were all received from
`VWR BDH Prolabo, Lutterworth, UK and used without further
`for crystallisation of different crystal habits of
`purification
`mefenamic acid. Deionised water, ethylene glycol (>99.0%, Sigma
`Aldrich, Dorset, UK), formamide (>99.5%, Acros Organics, Lough-
`(>99.0%, Acros Organics,
`borough, UK) and diiodomethane
`
`The indexing of the crystal faces was performed using an Oxford
`(Agilent Technologies,
`Diffraction Xcalibur 3E diffractometer
`Oxford, UK) equipped with ceramic XRD C-tech tube and Oxford
`Diffraction Sapphire detector. Single crystal X-ray diffraction data
`was obtained at 50 kV and 40 mA. Based on the single crystal X-ray
`facets were
`indexed using Agilent
`diffraction data crystal
`CrysAlisPro (Agilent Technologies, Oxford, UK) software system.
`
`3.5. Contact angle measurements of a macroscopic single crystal
`
`A Krüss Drop Shape Analyser DSA 10 (Krüss GmbH, Hamburg,
`Germany) was used for the static sessile drop contact angle
`
`Merck Exhibit 2250, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`142
`
`U.V. Shah et al. / International Journal of Pharmaceutics 472 (2014) 140–147
`
`measurement. Analytical grade DI water, diiodomethane, ethylene
`formamide were used as probe
`liquids and the
`glycol and
`measurements were carried out at ambient conditions. Properties
`of probe liquids can be found in existing literature (Heng et al.,
`fitted with a tangent method
`2006b). The shape of the droplet was
`to obtain the contact angle using the Drop Shape Analysis software
`(DSA version 1.0, Krüss GmbH, Hamburg, Germany). A minimum of
`three droplets on four different crystals facets of a macroscopic
`single crystal was measured. The measurements were repeated for
`two different single crystals.
`
`3.6. Surface energy analysis
`
`Surface energy analyser (SEA) (Surface Measurement Systems
`flame ionisation detector, was
`Ltd. London, UK), equipped with
`to characterise
`the surface energy heterogeneity
`for
`used
`1 g powders of different crystal habits
`mefenamic acid powders.
`were separately packed
`into pre-silanised
`iGC columns and
`conditioned with helium purge for 2 h at 30 C, followed by pulse
`injection measurements. Methane was used to determine column
`dead time and helium was used as a carrier gas. A series of
`dispersive n-alkane probes (hexane, heptane, octane, nonane and
`injected with an
`decane) at a range of concentrations were
`objective of obtaining target surface coverage (n/nm) ranging from
`0.7% to 10% to determine adsorption isotherms. The dispersive
`is then calculated using the Schultz method
`surface energy
`(Schultz et al., 1987). Mono-polar probes (dichloromethane and
`ethyl acetate) were injected at the same series of concentrations to
`determine non-dispersive interactions. Surface energy due to the
`non-dispersive interactions was calculated using the vOCG method
`reported in the literature (Das et al., 2010; Van Oss et al., 1988).
`Detailed method for determining surface energy heterogeneity can
`be referred from Ho and Heng (Ho and Heng, 2013).
`
`3.7. Surface area analysis
`
`Approximately 1 g of crystalline mefenamic samples was
`conditioned under helium purge at 40 C for at least 12 h, and
`is measured. A fully
`the mass of samples post-conditioning
`automated Micromeritics Tristar 3000 (Micromeritics, Norcross,
`USA) system was used for the measurement of isotherms at
`195.8 C. The surface area was calculated using the BET model
`(Brunauer et al., 1938) based on the linear region of the nitrogen
`adsorption isotherm (from p/p = 0.05–0.3) using the Micromer-
`itics analysis software (Micromeritics, Norcross, USA).
`
`3.8. Particle size and shape analysis
`
`Morphologi G3S particle characterisation system (Malvern
`Instruments Ltd. Malvern, UK) equipped with dry powder
`dispersion unit was used for particle shape and size analysis.
`Particles were dry dispersed using the dry sample dispersion unit
`on a sample glass plate mounted on an automatic stage. Particle
`imaging was conducted using a 20
` lens with the vertical z-
`stacking enabled to obtain information for the three dimensional-
`filtered using the analysis
`ity of the sample. Raw data were
`software (version 8.0) to remove partially imaged or overlapping
`particles on a sample by sample basis using a combination of
`filters. Details of the data
`convexity, solidity and particle width
`analysis method can be found elsewhere (Gamble et al., 2011).
`
`3.9. Scanning electron microscopy
`
`Mefenamic acid powders were stuck on the stubs using carbon
`adhesive tape and coated with gold. SEM images were obtained
`with table-top scanning electron microscope (Hitachi High-
`
`Technologies Europe GmbH, Krefeld, Germany) at an acceleration
`voltage of 15 kV. The SEM was equipped with a solid-state
`backscatter detector and operated in a standard imaging mode
`for obtaining the images.
`
`3.10. Polymorph identification
`
`for the mefenamic acid
`Powder X-ray diffraction spectra
`powders were measured using a PANalytical X’Pert Pro MPD
`(PANalytical B.V. Almelo, Netherlands). Measurements were
`performed at 40 mA and 40 kV. Diffraction data was collected
`using a CuKa
` X-ray source (1.541 Å) with nickel
`filter, a
`fixed
`10 mm mask, a soller slit of 0.04 rad, an antiscatter slit of 1/2 and a
`divergence slit of 1/4, over an angular range from 5 to 60 2u in
`continuous scan mode using step size of 0.08 2u and time per step
`of 35 s.
`
`3.11. Uniaxial compression test
`
`Powder cohesion was measured using a simple method adopted
`from the test originally developed for soil mechanics (ASTM D2166,
`BS1377: Part 7: 1990:7.2 and BS 1377: Part 2: 1990 7.3) (Head, 1994;
`Wang, 2013). Cylindrical compacts were prepared using 5 mm
`evacuable IR die (Specac Ltd. Slough, UK) at a minimum of three
`different compaction loads using an SMS texture analyser TA.XT2i
`(Stable Micro Systems Ltd. Godalming, UK). The compacts varied in
`height depending on the bulk density of the material; however, the
`least 2:1 was maintained for the compacts.
`L/D ratio of at
`Unconfined yield strengths of the compacts were measured using
`texture analyser. All the measurements were performed with a
`36 mm diameter cylindrical aluminium probe and a 5 kg load cell
`using a displacement compression mode with test speeds of
`0.02 mm/s. Yield stress was determined following the above
`mentioned method for compacts prepared at minimum of three
`different compaction loads.
`
`4. Results
`
`4.1. Anisotropic wettability of mefenamic acid crystals
`
`Three different polymorphs of mefenamic acid are reported. All
`three polymorphs are known to be triclinic with space group P1,
`however with different lattice parameters (SeethaLekshmi and
`Guru Row, 2012). Single crystals X-ray diffraction analysis revealed
`the crystals to be of form-I with crystal lattice parameters a = 14.6
`Å, b = 6.8 Å and c = 7.7 Å; a = 119.6, b = 103.9 and g = 91.3, which
`agrees well with the crystal structure for polymorphic form-I
`reported in the Cambridge crystal Structural Database (CSD) (Ref
`code: XYANAC). Single crystal X-ray diffraction was used to index
`crystals from crystallisation solvent methanol and found to have
`four major indexed facets (0 0 1), (11 2), (10 0) and (0 11).
`The contact angles of water, formamide, ethylene glycol and
`diiodomethane were measured on available facets of the form-I
`for a
`mefenamic acid crystals. Contact angles determined
`minimum of three droplets on two different crystals are reported
`in Table 1. Contact angle measurements demonstrated that,
`
`1
`Table
`Contact angle of four different probe liquids measured on four different facets of
`mefenamic acid single crystal.
`
`Facet
`
`Probe liquids
`
`Water
`
`
`
`
`
`59.5
`92.0
`56.3
`72.6
`
` 1.8
` 3.0
` 3.8
` 2.8
`
`(0 0 1)
`(11 2)
`(10 0)
`(0 11)
`
`Diiodomethane
`
`
`
`
`
`27.0
`19.3
`42.9
`11.5
`
` 2.1
` 3.1
` 2.0
` 2.6
`
`Formamide
`
`
`
`
`
`35.4
`50.2
`38.7
`34.9
`
` 3.2
` 4.1
` 1.6
` 2.1
`
`Ethylene glycol
`
`
`
`
`
`41.3
`36.1
`36.0
`49.3
`
` 2.2
` 2.4
` 1.6
` 4.2
`
`Merck Exhibit 2250, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`U.V. Shah et al. / International Journal of Pharmaceutics 472 (2014) 140–147
`
`143
`
`
`
`
`
`particularly for the water probe, sessile drop contact angle varies
`for crystal facets (0 0 1), (0 11) and (11 2), (10 0). Contact angles for
`water vary from 56 for (10 0 ) to 92 for (11 2). This confirms
`anisotropic wettability of the mefenamic acid crystal. The order of
`hydrophilicity for different facets, considering water contact angle
`
` (10 0) > (0 11) > (11 2).
`measurements, is as follows: (0 0 1)
`The differences in the contact angles can be explained by the
`variations in the local surface chemistry, differences in type and
`density of functional groups on different crystal facets. Based on
`crystallographic evaluation of mefenamic acid form-I, the phenyl
`ring with the carboxylic acid functional group and the imino bridge
`is found to be coplanar. The carbonyl functional group accepts a
`hydrogen bond from the amine and forms the imino bridge.
`to the carboxylic acid
`functional group’s
`Furthermore, due
`to self-associate, mefenamic acid molecules
`form
`tendency
`
`
`symmetric dimers and adjacent dimers are linked through CH
` interactions involving aromatic CH and alkylated phenyl ring
`p
`(SeethaLekshmi and Guru Row, 2012). Planes form different crystal
`facets, which slices through the crystal at different angles resulting
`in different proportions of functional end groups on different
`crystal facets. Crystallographic structures at the (0 0 1) and (11 2)
`from mefenamic acid polymorphic
`form-I
`facets, generated
`structure obtained from CSD using Mercury software, shows
`OH functional group exposed on the surface with H-bonding
`potential for facet (0 0 1), whereas negligible H-bonding potential
`OH functional groups, which
`was observed for facet (11 2). The
`are not involved in any hydrogen bonding in the crystal structure
`and available to act as electron donor (O) and acceptor (H) are
`considered to be available to interact with probe liquid for contact
`angle measurements.
`This analysis provides only an estimate of the potential
`contribution of hydroxyl as well as methyl functional groups on
`the surface. Investigating molecular orientation for individual
`crystal facets using Mercury software, facet (10 0) was found to
`have the highest concentration of available hydroxyl functional
`groups per unit area, whereas facet (11 2) has no hydroxyl
`functional group available. Considering the fact that (11 2) has no
`hydroxyl groups and has the presence of methyl groups, it is the
`most hydrophobic crystal facet. However, though facet (10 0) has
`the highest hydroxyl group density, methyl group density is
`equally high, which limits its hydrophilicity. Facet (0 0 1) has
`moderate density of hydroxyl groups and also contains equal
`density of carbonyl group (electron donor), both of which can
`participate in hydrogen bonding. For facet (0 0 1), the density of
`groups that can participate in hydrogen bonding is equal to that of
`
`facet (10 0) and absence of methyl functional group on facet (0 0 1)
`may result in higher hydrophilicity of the facet.
`Facet (10 0) has a higher concentration of hydroxyl groups and
`an equivalent concentration of methyl groups per unit area
`compared to (0 11). The higher hydrophilicity of facet (10 0)
`to
`facet
`(0 11) can be attributed
`to
`the higher
`compared
`concentration of the hydroxyl functional groups per unit cell area.
`first order estimation of contribu-
`However, analysis here provide
`tion from different functional groups, it does have limitations.
`Surface contributions of phenyl group as well as surface
`orientation of hydroxyl, carbonyl or methyl functional group is
`not considered in the analysis.
`Facet specific surface chemistry of mefenamic acid single
`crystals was characterised using X-ray photoelectron spectroscopy.
`The order of polarity determined by XPS, which is in agreement
`with the hydrophilicity order as calculated on the basis of sessile
`contact
`angle
`for water
`as
`a
`probe
`liquid
`drop
`((0 0 1) > (10 0) > (0 11) > (11 2)).
`
`4.2. Effect of crystal habit on surface energy
`
`4.2.1. Preparation of crystals with varying crystal habits
`It is well documented in the literature that crystal–solution
`interface determines many interfacial physical phenomena. In the
`current context, this relates to crystal growth and wetting
`(Berkovitch-Yellin, 1985; Bourne and Davey, 1976; Lovette et al.,
`2008). Crystal surface contribution at the solvent interface may be
`different from the bulk crystallographic structure, which depends
`on possible reconstruction and relaxation of surface. For solvents,
`atomic arrangement at the interface is postulated to be relatively
`to bulk
`liquid, considering
`the periodic
`ordered compared
`potential it experiences at the crystal surface (Bourne and Davey,
`1976; Davey et al., 1988). Despite of extensive research, a well
`interface
`characterised mechanism by which solution–crystal
`influences crystal growth is elusive (Singh and Banerjee, 2013).
`in the polar component of the Hansen
`Solvents varying
`from 1.4 MPa1/2 to 12.3 MPa1/2
`solubility parameter, ranging
`(which represents energy from dipolar
`intermolecular forces
`between molecules) (Hansen, 2007), were used for crystallisation
`of mefenamic acid to investigate effect of crystallisation solvent
`polarity on crystal habits. Crystals obtained from different solvent
`systems are shown in Fig. 1. It is evident from the scanning electron
`micrographs that mefenamic acid crystals grown from non-polar
`solvents such as toluene and diethyl ether result in needle shape
`crystals (Fig. 1(a)), mefenamic acid crystals grown from polar
`
`Fig.
`
`1. Mefenamic acid crystals obtained from (a) toluene, (b) DEE, (c) EA, (d) DCM, (e) acetone, (f) IPA and (g) methanol.
`
`Merck Exhibit 2250, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`U.V. Shah et al. / International Journal of Pharmaceutics 472 (2014) 140–147
`
`luene
`MA from To
`
`MA from DEE
`MA from EA
`MA from D
`CM
`
`MA from Acetone
`
`MA from IPA
`
`MA from
`Methanol
`
`MA-Form
`-I-R
`eference Pattern
`
`
`
`144
`
`Linear Intensity (a.u.)
`
`0
`
`10
`
`20
`
`40
`
`50
`
`30
`2θ (o)
`
`(a)
`(b)
`(c)
`(d)
`(e)
`(f)
`(g)
`(h)
`60
`
`aprotic solvents such as ethyl acetate, dichloromethane and
`acetone result in elongated plate shape crystals (Fig. 1(b–d)),
`whereas mefenamic acid grown from polar protic solvents such as
`isopropyl alcohol and methanol result in hexagonal cuboid shaped
`crystals (Fig. 1(e and f)). Quantitative analysis of the crystal
`elongation was conducted using image analysis of the crystals
`obtained from all seven different solvent systems using Malvern
`Morphologi G3S (see Section 4.2.2).
`Powder X-ray diffraction spectra were obtained for mefenamic
`acid crystals obtained from all seven solvent systems. X-ray
`diffraction patterns obtained were compared with the reference
`powder spectrum reported in CSD for mefenamic acid form-I (Ref
`code: XYANAC) (Fig. 2) and powder X-ray diffractograms for
`mefenamic acid crystal form-II and III by SeethaLekshmi and Guru
`Row (SeethaLekshmi and Guru Row, 2012). Crystals obtained from
`all seven solvent systems were identified to be of polymorphic
`form-I, which is known to be the most stable crystal polymorph.
`Analysis of the powder diffraction spectra from different crystal
`habits showed varying peak intensity, which depends on the
`amount of sample used, particle size, packing and sample
`In addition to the
`thickness (Pecharsky and Zavalij, 2005).
`12.6 2u was observed
`variations in peak intensity, a peak at
`for the mefenamic acid crystals obtained from all solvents except
`toluene, and this peak was more prominent for mefenamic acid
`
`4. Correlation between mefenamic acid elongation data (squares) obtained
`Fig.
`from solvents with varying polar component of the Hansen solubility parameter
`and dispersive component of surface energy (diamonds) (isostere at n/nm = 0.04).
`
`crystals obtained from methanol and acetone. Referring to the
`reference pattern for form-I mefenamic acid, a weak peak at
`12.66 2u can also be found. This peak is not observed in the
`powder X-ray diffraction patterns for mefenamic acid form-II or III.
`Although, no correlation between peak intensity and differences in
`crystal aspect ratio or relative surface area ratio of crystal facets
`was observed, it is important to note that all diffractograms
`matches with that of mefenamic acid form-I.
`
`4.2.2. Effect of crystal habit on surface energy
`The gd distributions of the mefenamic acid crystals with
`varying crystal habits are summarised in Fig. 3. An ascending trend
`in the gd profiles, whereas descending trend in gAB profiles was
`observed with descending crystal elongation (Figs. 4 and 5), i.e.
`considering two different extremes, gd for needle shape crystals,
`which were obtained from toluene, ranges from 42.1 mJ/m2 to
`39.9 mJ/m2, whereas for hexagonal cuboids shape crystals, gd
`ranges from 51.3 mJ/m2 to 46.4 mJ/m2 from the lower to higher
`fractional surface coverage (0.7–10%).
`Ascending gd profiles with descending crystal elongation can be
`attributed to increasing/decreasing relative contribution of differ-
`ent crystal facets with changing crystal habits. To investigate the
`effect of the relative contribution of different facets on gd profile,
`we consider contribution from two major facets namely (0 0 1) and
`(11 2) and how it varies with different crystallisation solvents.
`Mefenamic acid crystal facet (0 0 1) is demonstrated to have
`CQO and
`C6H5 functional groups exposed on the
`OH,
`surface which can result in strong hydrogen bonding interactions
`
`0
`
`2
`
`4
`
`6
`
`8
`
`10
`
`
`12
`
`
`14
`
`7
`
`6
`
`5
`
`4
`
`3
`
`2
`
`1
`
`0
`
` Toluene
`MA from
`
`MA from DEE
`MA from E
`A
`
`MA from DCM
`MA from
` Acetone
`
`MA from I
`PA
`
`MA from Methanol
`
`0
`
`0.01
`
`0.02
`
`0.07
`0.06
`0.05
`0.04
`0.03
`Fractional Surface Coverage (n/nm) (-)
`
`0.08
`
`0.09
`
`0.1
`
`57
`
`55
`
`53
`
`51
`
`49
`
`47
`
`45
`
`43
`
`41
`
`39
`
`37
`
`35
`
`Dispersive Surface Energy (γd) (mJ/m2)
`
`Fig. 3. Dispersive component of surface energy as a function of fractional surface
`coverage for mefenamic crystals obtained from seven different solvents (error bars
`represent standard deviation of three measurements).
`
`Fig. 5. Acid-base surface energy as a function of solvent polar component of the
`Hansen solubility parameter (isostere at n/nm = 0.04).
`
`Merck Exhibit 2250, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`U.V. Shah et al. / International Journal of Pharmaceutics 472 (2014) 140–147
`
`145
`
`8
`
`7
`
`6
`
`5
`
`4
`
`3
`
`2
`
`1
`
`Cohesion (kPa)
`
`between facet (0 0 1) and protic-polar solvents (Davey et al., 1988),
`CH3 and
`C6H5
`facet (11 2) has
`whereas the surface of
`functional end groups resulting in very weak polar interaction,
`which can result in strong interaction with the non-polar solvents
`e.g. DEE. With changing solvent polarity from the non-polar to
`polar-protic and polar-aprotic solvent, solvent interaction with
`crystal facet (11 2) will decrease and interaction with crystal facet
`(0 0 1) will increase. Bourne and Davey proposed that favourable
`interaction between the solvent with a particular crystal facet
`results in reduction of interfacial tensions, resulting in enhanced
`growth of the crystal facet.
`Bourne and Davey explained the growth of sucrose from
`aqueous solutions with this mechanism (Bourne and Davey, 1976).
`Considering the mechanism proposed by Bourne and Davey, with
`increasing solvent polarity, increase in solvent interaction with
`mefenamic acid crystal facet (0 0 1) can result in increased growth
`of that facet, resulting in lower morphological importance of facet
`(0 0 1) compared to facet (11 2). This can result in lowering the
`relative contribution of facet (0 0 1) compared to (11 2) and leading
`to a change in crystal habit from elongated needles to hexagonal
`cuboids. This is consistent with the experimental observations
`reported in Section 4.2.1. As the relative contribution of facet (0 0 1)
`decreases with increasing solvent polarity, crystals obtained with
`methanol will have a lower relative contribution compared to
`crystals obtained from toluene and reverse is true for the (11 2)
`facet.
`The shapes of crystals obtained from ethyl acetate, dichloro-
`methane and isopropyl alcohol are needles, elongated plates and
`hexagonal cuboids, while surface area of crystals obtained from all
`three solvents are 0.29 m2/g, 0.24 m2/g and 0.15 m2/g, respectively.
`gd at a fractional coverage (n/nm = 0.04) is shown as a function
`of solvent polarity in Fig. 4, and suggest that gd increases with
`increasing solvent polarity. Fig. 5 shows acid-base component of
`surface energy as a function of solvent polarity. Acid-base surface
`energy at a fractional surface coverage of 0.04 was found to vary
`with increasing solvent polarity. Although some scatter in the data
`can be observed, a weak trend of reducing acid-base surface energy
`with increasing solvent polarity was observed. Decrease in acid-
`base surface energy can be explained by lower relative contribu-
`tion from facet (0 0 1).
`In the current analysis, the gd distribution is dependent on
`elongation rather than particle size or indeed the BET surface area
`is normalised for surface coverage. This further
`as the data
`confirms that the relative exposure of different crystal facets plays
`a crucial role in overall surface energy of powder samples.
`
`0
`
`0
`
`12
`10
`8
`6
`4
`2
`
`Polar Componen