investigations on solid–solid phase transformation of...

Investigations on Solid–Solid Phase Transformationof 5-Methyl-2-[(4-Methyl-2-Nitrophenyl)Amino]-3-Thiophenecarbonitrile

HUI LI,1 JOSEPH G. STOWELL,1 XIAORONG HE,2 KENNETH R. MORRIS,1 STEPHEN R. BYRN1

1Department of Industrial and Physical Pharmacy, 575 Stadium Mall Drive, Purdue University,West Lafayette, Indiana 47907-2091

2GlaxoSmithKline, Building 2.3055, Five Moore Drive, RTP, North Carolina 27709

Received 7 March 2006; revised 23 August 2006; accepted 31 October 2006

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20937

ABSTRACT: Solid–solid transformation of 5-methyl-2-[(4-methyl-2-nitrophenyl)a-mino]-3-thiophenecarbonitrile from the dark-red to the red form was investigated. Bycontrolled crystallization, the dark-red form was prepared and the crystals weresieved into fractions: coarse (>250 mm), medium (125–177 mm), and fine (<88 mm).The transformation rate order (fastest to slowest) of the different fractions iscoarse>medium>fine. However, milling accelerates the transformation, that is,smaller particles generated by milling transforms faster. Furthermore, ethanol vaporannealing slows both the transformation of the coarse and medium fractions, especiallythe latter. Therefore, the mechanism of transformation is not directly related to thecrystal-size and most likely related to the amount and activity of the defects in thecrystals. The three-dimensional (3-D) Avrami–Erofe’ev model, know as ‘‘randomnucleation and growth’’ model, fits the kinetics of coarse fraction best. Higher relativehumidity accelerates the transformation dramatically even though the compound ishighly-hydrophobic. With minimal hydrogen bonding interaction involved, it appearseven small amounts of water can serve as a nucleation catalyst by binding to the crystalsurface, especially at defect sites, thus increasing the molecular mobility of these sites,promoting the transformation to the second phase and thereby increasing thetransformation rate. � 2007 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm

Sci 96:1079–1089, 2007

Keywords: crystallization; polymorphism; transformation; particle size; annealing;nucleation; crystal growth; relative humidity; milling; crystal defects

INTRODUCTION

It is estimated that about 40% of the compoundslisted in the United States Pharmacopeia exhibitpolymorphism. Polymorphic transformation is ofa great concern for the pharmaceutical industry

because it may affect the dissolution rate of drugproduct, drug delivery and bioavailability, anddrug physical and chemical stability.1–4 Further-more, it may also affect the mechanical propertiesof pharmaceutical dosage form.

To prevent polymorphic transformation, exten-sive studies need to be carried out. Previousliterature has shown that a number of conditions,which may be encountered during processing (e.g.,temperature, relative humidity, seeding, particlesize, additives, excipients, and solvents suchas water), may promote transformation.5,6 In

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 5, MAY 2007 1079

Hui Li’s present address is ICOS Corporation, 22021 20thAve SE Bothell, WA 98021Correspondence to: Hui Li (Telephone: 425-489-4799; Fax:

425-486-0300; E-mail: [email protected])

Journal of Pharmaceutical Sciences, Vol. 96, 1079–1089 (2007)� 2007 Wiley-Liss, Inc. and the American Pharmacists Association

addition, transformation can occur because of aparticular manufacturing unit operation (e.g.,grinding, milling, wet granulation, drying, com-pression, tableting, and storage).7–12 A great dealof interest is focused on the kinetics of the phasetransitions involved. Thorough characterization ofthe kinetics can enhance the fundamental under-standing of the underlying processes, assist in therational design and development of materials andprocesses, and address uncertainties regardingthe stability of the particular phases. In the 1980’s,Ohnishi et al.13–15 performed extensive kineticstudies on the isothermal transitions of some drugmolecules at high temperatures; transformationrates were obtained by fitting the data (time vs.percent transformed) to several kinetic modelsfrom which the Arrhenius activation energieswere calculated. Sheridan and Anwar16 usedtime-resolved energy-dispersive X-ray diffractionto study the solid-state phase transformation ofsulfanilamide from the b-form to the g-form.However, the mechanism of polymorphic transfor-mations, especially solid–solid transformations,needs to be further understood.

During the past 15 years our laboratory hascarried out extensive studies of ROY and ROYderivatives.11–19 These compounds provide an excel-lent system for the study of polymorphism sincedifferent polymorphs exhibit different colors. Thus,transformation between the different polymorphscan be studied visually or photographically. For thisstudy, we used a ROY derivative, 5-methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbo-nitrile (40-Me) as the model compound. 40-Me hasfour known polymorphs: dark-red, red, orange, andlight red. Only the dark-red crystal structure issolved and no appropriate single crystals for theother three polymorphs were identified for structureelucidation. Thermodynamic relationships of thesefour polymorphs were constructed by an energy–temperature diagram.20 An enantiotropic relation-ship exists between the dark-red and red formswhere the dark-red form is more stable form atambient temperatures (equilibrium transition tem-perature Ttr equals to 558C).

Byrn et al.3 provided an extensive review ofpolymorphic interconversions. These solid–solidphase transformations were typically inducedby heat or stress. These transformations wereaffected by relative humidity and defects. Nostudies of the effects of particle size were reported.This paper reports an extensive study of thethermal (858C) solid–solid phase transformationof the dark-red form to the red form. Thisphase transformation proved particularly complexand is affected by a range of factors includingannealing, and relative humidity. Most interest-ing are the observations reported in this paperthat: (1) relative humidity accelerates the reactioneven though 40-Me is very hydrophobic andexhibits very little moisture sorption; and (2) largeparticles transform faster than small particles forthe dark-red material prepared from a crystal-lization process. These results illustrate the com-plex combination of factors that control thesesolid–solid processes and bring us closer to acomplete understanding of solid–solid phasetransformations.

MATERIALS AND METHODS

Preparation of the Dark-Red, Red, and AmorphousForms and Sieving into Three Fractions

Model compound 40-Me was synthesized by Heet al.20 A controlled crystallization method wasused to produce enough material for this study.The dark-red form was prepared in ethanol (1-Lscale) with a 0.38C/min cooling rate, 200 rpmstirring rate, a supersaturation ratio of 3 at 208C,and the addition of 1% seeds at �438C. The dark-red crystals were collected by filtration and driedunder vacuum for several hours. The resultingcrystals were fractionated by size with a Tyler Ro-Tap1 sieve shaker (Sepor, Inc., Wilmington, CA)for 5 min; three fractions were collected forexperimentation: coarse (>250 mm), medium(125–177 mm), and fine (<88 mm). The red formwas prepared from dark-red crystals through asolid–solid transformation at 1058C for 1 h. 40-Meamorphous form was prepared by quenched meltmethod using liquid nitrogen. Around 200 mg 40-Me was melted on a X-ray holder and immersedinto liquid nitrogen. The resulting solids weretested on X-ray instrument immediately withscan rate 128/min because it is not stable and re-crystallized in 5–10 min at ambient temperature.A few single crystals of the dark-red form wereobtained by slow evaporation from toluene.

1080 LI ET AL.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 5, MAY 2007 DOI 10.1002/jps

Powder X-Ray Diffraction

Powder X-ray diffraction (PXRD) was used toquantify the transformation. The PXRD patternsof the fractionated samples were obtained usinga Siemens D500 Diffraktometer-Kristalloflexequipped with a Kevex Psi Peltier cooled silicon[Si(Li)] detector. Cu Ka radiation (20 mA, 40 kV),with a 17–198 scan range, a 0.68/min scan rate,and a 0.028 step size was used for quantification.Samples were packed into the well of aluminumholders produced in-house. Quantitation methodis detailed in the Result section.

Transformation Study at VariousRelative Humidities

Samples in glassine sachets (made in-house fromweigh paper) were placed into constant humiditychambers: 0% (P2O5), 58% (saturated NaNO3),and 95% (saturated K2SO4). The transformationstudies were conducted at 858C.

Vapor Annealing

Samples of the course and medium fractions weresuspended in containers above a layer of ethanoland sealed to saturate the headspace with ethanolvapor; the samples were annealed for 24 days atambient temperature. The samples were thendried in vacuo for 2 days. No residual solvent wasseen on thermo-gravimetric analysis (TGA). Theeffect of vapor annealing on phase transformationwas investigated at 908C, 0% RH.

Milling of the Dark-Red Form

Samples of the dark-red form were milled for setdurations (1, 5, 15, 20, and 30 min) using a Wig-L-Bug1 3100A or 3100B model mill (CrescentDental Manufacturing Co., Chicago, IL). Trans-formation kinetics after milling for 1 and 5 min at858C were investigated using PXRD.

Surface-Area Measurement by BET Method

Surface area was measured on a Micromeritics1

Surface Analyzer 2010 (Micromeritics InstrumentCorp., Norcross, GA). The samples were dried in a408C oven overnight and degassed under vacuumfor 2 days before analysis. The system was purgedwith helium gas. Surface area was measuredusing nitrogen gas and calculated using at leastfive data points.

Moisture Sorption

An SGA-100 symmetric vapor-sorption analyzer(VTI Corp., Hialeah, FL) was used. Crystallinedark-red form was pre-dried before performingthe relative humidity steps (15% RH increase perstep) at 258C. No phase changes were observedbefore and after the experiment.

Karl–Fischer Titration

The moisture content of solid samples wasdetermined in an Accumet1 Model 100 Titrator(Fischer Scientific, Pittsburg, PA). Samples wereweighed and added directly to the titration cell.

RESULTS AND DISCUSSIONS

Particle-Size Distribution of the Dark-RedForm after Controlled Crystallization

A controlled crystallization process was needed toproduce sufficient quantities of pure dark-red 40-Me ROY. Several initial experiments establishedthat seeds were required to control the crystal-lization. Reproducible yields and particle-sizedistributions from batch to batch were obtainedby using 1% dark-red seeds having a narrow sizerange (44–74 mm). Around 25 1-L scale batcheswere combined and fractionated using appropri-ate sieves; the particle-size distribution of theresulting material is shown in Figure 1. Morethan 62% of particles were coarse (>250 mm); 22%were medium (125–177 mm), and 7% were fine(<88 mm). For most of the following studies, thecoarse fraction was used. This fraction is poly-crystalline according to scanning electron micro-scopy imaging (not shown). Medium fraction ispolycrystalline as well and the fine fractions are

Figure 1. Particle-size distribution of the dark-redform collected from a controlled crystallization.

SOLID-STATE PHASE TRANSFORMATION 1081

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 5, MAY 2007

fragments, possibly generated from secondarycrystallization. For comparison, samples fromthe coarse, medium, and fine fractions, as wellas unfractionated material, were studied.

Powder X-Ray Diffraction (PXRD) Patterns andMethod of Analysis of the Dark-Red and Red Forms

Figure 2 shows the PXRDs of the dark-red and redforms, illustrating that these diffraction patternsare distinct. All fractions of the dark-red formshow PXRD patterns with flat baseline, indicatinghigh crystallinity. No obvious amorphous ‘‘halo’’was seen for any of the fractions. The peaksindicated by arrows at 15.68 2y (dark-red) and16.28 2y (red) were used to quantify the solid-statephase transformation from the dark-red to the redform. Standard curves were constructed for everyfraction, respectively, by plotting peak areaversus percentage of the dark-red or red formsand good linear relationship was observed(R2> 0.99; standard deviation within� 2%).Preferred orientation was minimized by using anarrow crystal-size range.

Solid-State Transformation of the Sieved Fractions(Different Particle Size Fractions)

Phase transformation studies of the threefractions and unfractionated material were inves-tigated at 858C, 0% RH (see Fig. 3). No residualsolvent (ethanol) was present as determined by

thermal gravimetric analysis (data not shown),indicating that this to be a solid–solid transfor-mation. The results, duplicated in replicateexperiments, show that the larger the particlesize, the faster the transformation from the dark-red to the red form. These results are contrary tosuggestions that smaller particles will transformat a faster rate. These results are consistent withanecdotal information on some phase transforma-tion but to our knowledge have not been reportedpreviously.

The data were fit to several kinetic models.14

The three-dimensioinal (3-D) Avrami–Erofe’evmodel, known as ‘‘random nucleation and growth’’model, was found to be the best fit (higher R2) forthe coarse and medium fractions as well as theunfractionated material (see Table 1).

To investigate the relationship betweensurface area and rate of phase transformations,the surface area of the coarse, medium, andfine fractions was measured using BET method.The crystals in the coarse fraction are aggregates(polycrystals) with a surface area 0.5423�0.0074 m2/g; the crystals in the medium and finefractions have surface areas of 0.4902� 0.0043 m2/g and 1.1974� 0.0108 m2/g, respectively. Thesurface area of the coarse fraction and mediumfractions are similar with the course fractionhaving a slightly higher value. This may be dueto difference in surface roughness (higher rough-ness for course fraction). However, significantdifferences in surface roughness between the large

Figure 2. Powder X-ray diffraction patterns of the dark-red (bottom) and red (top) forms.

1082 LI ET AL.


and medium particles could not be verified bySEM. The fine fraction has a much larger surfacearea (about 2�) than both the course and mediumfractions. The fine fraction has a rate of transfor-mation much slower than that of the course andmedium fractions which are more similar.

Annealing Studies

To further probe the role of defects in the processthe effect of annealing using ethanol vapor wasinvestigated. The phase transformation was

retarded after annealing (Fig. 4). Ninety degreeCelsius was used in this study to expedite thephase transformation. In addition, this tempera-ture condition, which is 128 higher than ethanolboiling point 788C, would remove any residualethanol content in the samples (if there is any).Particles for both coarse fraction and mediumfraction were analyzed using microscopybefore and after annealing. Particle size remainedun-changed for both fractions. This suggests thatthe particles are not fracturing during the phasetransformation. Ethanol vapor annealing ishypothesized to ‘‘relax’’ defects, especially thoseon crystal surface. Thus, the hypothesis thatthe reaction is defect mediated is supportedby the fact that the reaction rate is slowedby annealing. Additionally, the milling studydescribed below is also consistent with thehypothesis that the reaction is defect mediated.Annealing appears to retard the transformationrate of the medium fraction more than it does thecoarse fraction even though these two fractionshave similar surface areas. It is most probablethat the energy of the defects, which is technicallyhard to measure, for the coarse fraction is greaterthan that for the medium fraction, and thesehigher-energy defects sites are more difficult to beannealed.

Figure 3. Solid-state phase transformation kineticsof different fractions of the dark-red form at 858C (fromtop to bottom: coarse, unfractionated, medium, and fine;standard error within �2%).

Table 1. Kinetics Equations Fitting of Different Fractions of the Dark-Red Form Collected From a ControlledCrystallization (highest R2 shown in Bold)

Kinetic Equations ‘Mechanism’

Unfractionated(min�1) >250 mm (min�1) 125–177 mm (min�1)

k� 103 R2 k� 103 R2 k� 103 R2

a¼ kt zero-order 1.1 0.9503 1.6 0.9727 0.6 0.98731� (1� a)1/2¼ kt phase boundary,

cylindrical0.7 0.9185 1 0.4 0.4 0.9512

1� (1� a)1/3¼ kt phase boundary,spherical

0.5 0.9049 0.7 0.9423 0.3 0.9313

�ln(1� a)¼ kt random nucleation, 1-D 1.9 0.8733 2.6 0.9196 1.3 0.8815[�ln(1� a)]1/2¼ kt random nucleation,

2-D (Avrami–Erofe’ev)1.5 0.971 2.2 0.9893 0.8 0.9865

[�ln(1� a)]1/3¼ kt random nucleation,3-D (Avrami–Erofe’ev)

1.2 0.987 1.8 0.9952 0.6 0.9953

a2¼ kt 1-D diffusion 0.8 0.8145 1.1 0.867 0.5 0.8675(1� a) ln(1� a)þ a¼ kt 2-D diffusion 0.5 0.7754 0.7 0.8312 0.4 0.812[1� (1� a)1/3]2¼ kt 3-D diffusion

(Jander)0.2 0.7261 0.2 0.784 0.1 0.7397

(1 �2/3a) – (1 – a)2/3¼ kt 3-D diffusion(Ginsting–Brounshtein)

0.1 0.759 0.2 0.8158 0.1 0.7874

Note: a refers to the fraction of the produced phase (red form).The rate constant is the slope of the lease-squares fit, expressed as ‘k’.



Transformation after Milling

Milling is a common particle-size reductiontechnique used during drug product processing;while at the same time, milling generatescrystal defects and amorphous material in somecases.7–8 The transformation rate from the dark-red form to the red form increases with anincrease in milling time up to 5 min (Fig. 5). Theparticles were observed under optical microscope,showing that particle decreased to severalmicrons after 5 min milling. The resultsalso suggest that after milling smaller particlestransform faster. It should be noted that thelaser diffraction method, instead of sieve analysis,has to be used to measure particle size aftermilling.

Along with ethanol-vapor annealing study, thismilling study demonstrates again that the phasetransformation of the dark-red form to the redform is a defect-related process.

We note that milling-time is crucial. Whenmilled for longer than 20 min, the dark-red form

partially transformed to the red form right aftermilling, even before the samples were putinto 858C oven, where the phase transformationstudies were conducted. This is probably becauseduring milling at some sample sites the localtemperature is higher than the transition tem-perature from the dark-red form to the red form,which is determined to be 558C, although aftermilling the temperature measured is warm(�408C). The remainder of the dark-red formtransformed to the orange form overnight at roomtemperature, presumably because the orange formis the most stable form at ambient temperature(258C).

Speculations on Solid-State TransformationMechanism of the Dark-Red Form to the Red Form

Three solid-state polymorphic transformationmechanisms have been proposed in the litera-ture.21–23 The first mechanism is a ‘‘continuous’’model in which transformation takes place bymeans of a progressive reorientation and/ortranslation of the molecules in the crystals.Therefore, a portion of the original phase informa-tion is preserved in the final phase. The secondmechanism is a ‘‘nucleation and growth’’ modelin which transformations start at a particularsite (generally at crystal defects) and continueby a progressive propagation of the reactioninterface. The third mechanism is an ‘‘amorphousmediation’’ model in which the initial phaseundergoes a random transformation to produceamorphous material that subsequently undergoescrystallization.

The ‘‘continuous’’ model typically involves a‘‘single crystal to single crystal transformation’’.

Figure 4. Transformation of the dark-red form coarse fraction (left) and mediumfraction (right) at 908C before and after ethanol-vapor annealing (standard error within�2%).

Figure 5. Transformation of the dark-red form(coarse fraction) at 858C, 0% RH before and after millingfor 1 and 5 min (standard error within �2%).

1084 LI ET AL.


This is not the case for the 40-Me transformationof the dark-red form, which involves poly-crystals.

The possibility that the phase transformationmechanism involves the amorphous phasewas tested by monitoring the crystallization offreshly prepared amorphous material at ambienttemperature.

The possibility that the phase transformation ofthe DR form to the R form takes place through theamorphous phase was ruled out by the following:(1) amorphous 40-Me was prepared by thequenched-cooled melt method and solution NMRand HPLC afterwards showed no chemical degra-dation. In a very short time frame (5–10 min), theamorphous form re-crystallized into a deep-redform (named ‘‘DRa’’) that exhibits a PXRD patterndifferent from the four known forms including thered form, indicating that this is a new phase(Fig. 6). This observation excludes the possibilitythat the red form is produced by an amorphous-mediated process; (2) no amorphous material couldbe detected after milling or any other process.

The annealing and milling results indicate thatthe mechanism of the transformation from thedark-red form to the red form involves crystaldefects as discussed above. Therefore, the secondmechanism, the ‘‘nucleation and growth’’ mechan-ism is the best explanation for this transformationin the solid-state. We hypothesize that the coarsefraction transforms faster than fine fractionbecause it contains more defects. A reasonablerationalization of why larger crystals contain moredefects during crystallization is that the largercrystals are grown from nuclei/crystals developed

under higher supersaturation. Under these condi-tions crystals grow rapidly and are more likely toincorporate imperfections into the crystals. Thesmaller crystals, on the other hand, are the resultof secondary nucleation and grow under greatlyreduced supersaturation. Furthermore, Ostwaldripening of the small crystals reduces the imper-fections on its surface. It may not be a coincidencethat the 3-D Avrami–Erofe’ev equation, know as‘‘random nucleation and growth’’ model, fits thekinetics of coarse fraction better than all the othermodels, such as diffusion models (Table 1).

Images represented in Figure 7 show thattransformation seems to occur by spontaneousnucleation of the red form followed by subsequentgrowth of the red phase. The addition of red seedsdecreases the nucleation induction time of the redphase (data not shown).

Crystal energies (e.g., internal energy, heatcapacity, enthalpy, entropy, and free energy)change with increase in the density of latticedefects. The change in free energy corresponds to achange in the thermodynamic activity.24–28 Anincrease in free energy corresponds to an increasein the reactivity of the crystals because the defectsact as high-energy sites and hence influence thechemical stability of the crystals.29–32 For thetransformation of 5-methyl-2-[(4-methyl-2nitro-phenyl)amino]-3-thiophenecarbonitrile of thedark-red form to the red form, it is suggested thata collection of particles will have a distribution ofenergies. The nucleation sites, being sites of latticedefects and imperfections, are characterized bythis distribution of activation energies. These sites

Figure 6. PXRD of the five polymorphs and amorphous form of 5-methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbonitrile. From bottom to top: red, dark-red, lightred, orange, deep red (crystallized from the amorphous phase), and amorphous.



will be randomly distributed throughout thesample population. Consequently, only thosecrystallites that contain nuclei for which theprobability of activation is significant will trans-form at any given temperature; all others willremain untransformed. This implies that thereactivity and the minimum temperature are‘‘pre-coded’’ at nucleation sites, which are appro-priate lattice defects.

Our studies show that phase transformations ofthe course fraction go to completion at tempera-tures�858C. If the temperature is decreased below858C, the heterogeneity of particles becomesevident and the phase transformation is notcomplete because the activation energy is belowthe minimum energy required for all particles toreact at that temperature. In fact, about 100 mgcourse fraction of the dark-red form was heated at658C and only a few particles transformed to thered form (based on the color change).

Although quantitation of crystal defects inelectronic materials has been studied relativelyextensively, little work has been pursued inthe pharmaceutical field.33 Several studies inthe literature are presented briefly in this para-graph. Defect structures in mesophases of

pharmaceuticals were investigated using electronand light microscopy.34 Mitchell correlated thedissolution rate with the density of crystaldefects in KClO4 by microscopy.35 X-ray topogra-phy methods have been used to determinedefect structure.36 Calorimetric methods allowestimation of the point-defect concentration orconformational disorder.37,38 For impurity- oradditive-induced lattice defects, York and Grant39

introduced a term, disruption index (d.i.), toevaluate the lattice imperfection.

For the 5-methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbonitrile system in thispresent study, it is not possible to quantify thedefects directly because the dark-red form is poly-crystalline although all the experiments indicatethat the solid-state phase transformation is adefect-related process. Only a few single crystalsof the dark-red form were collected by slowevaporation from toluene solution. This quantityof crystals is not adequate for quantitative study.An alternative crystallization method is neededto produce enough single crystals of the pure dark-red form with a reasonable size distribution.

Relative Humidity Effect (Coarse Fraction)

These studies showed that the greater therelative humidity, the greater the transformationrate at 858C (see Fig. 8), even though 40-Me ishighly-hydrophobic. This result is consistentwith literature reports that residual water orrelative humidity can affect chemical andphysical stability of drugs in the solid-state.40

Typically, higher relative humidities can give riseto greater chemical instability as well as physicalinstability (e.g., rate of crystallization from theamorphous state, drug-excipient interaction, andformation of hydrates). Numerous studies ofhydration and dehydration kinetics with regards

Figure 7. Images demonstrating the transformationof dark-red form (course fraction) to the red from at1058C.

Figure 8. Effect of relative humidity on the transfor-mation kinetics of the dark-red form (course fraction) at858C.

1086 LI ET AL.


to changes in relative humidity can be found inthe literature. However, rarely has it beenreported that relative humidity will affect thephysical stability of a hydrophobic compound (nohydration or dehydration involved with a changein relative humidity). The only example, known tous, is the transformation of the phenylbutazoneforms a! d and b! d. In this system increases inthe relative humidity, resulted in an increasedtransformation rate.41 Although the compound inthe study reported herein is very hydrophobicwith almost zero aqueous solubility, relativehumidity has a significant effect on its solid-statetransformation.

The water content in the samples can beconsidered as ‘‘trace’’ amount, for example, only0.07% of water was absorbed at 95%. Both Karl–Fischer titration and vapor-sorption data (seeFig. 9) demonstrate that there is only a slightincrease in the water adsorbed on the crystallinesurface with an increase in the relative humidity.The formation of hydrogen bonds between watermolecules and diarylamine, nitrile, or nitro groupsis unlikely since these relatively-hydrophilicgroups are not exposed to crystal surface.42 There-fore, it appears that water, even a trace amount,can act as a catalyst to promote the nucleation ofthe second phase by increasing the molecularmobility, especially at defect sites, and henceaccelerate the transformation.

It is of interest to compare the effects of watervapor and ethanol vapor on the phase transforma-tion. Pretreatment of a sample with ethanol vaporat ambient temperature resulted in a decrease indefect sites and a slower rate of reaction. On theother hand, exposure of DR to water vapor whileheating at 858C resulted in an accelerated rate oftransformation. We hypothesize that molecular

mobility is increased by both processes. For therelative humidity effect, the increase of mobilityaccelerates the formation of the red form andtherefore increase phase transformation rate ofthe dark-red form to the red form at 858C(above the transition temperature). Converselyfor ethanol vapor annealing process, which iscarried out at ambient temperature, theincrease of mobility causes the relaxation ofmolecule at defect sites and therefore the rate ofpolymorphic transformation tested afterwards at908C (above the transition temperature) wasreduced. One would expect that ethanol interactsmore with 40-Me hydrophobic surface moiety thanwater molecule does. This may explain why 40-Mecan be annealed by ethanol even at ambienttemperature.

CONCLUSIONS

This study shows that the solid–solid phasetransformation of 40-Me DR to R is defect driven.This hypothesis is supported by: (1) millingstudies; and (2) ethanol vapor annealing studies.Additionally, the phase transformation rates ofdifferent size crystals are significantly different,with the larger crystals transforming faster.

For the solid-state phase transformationkinetics of the 5-methyl-2-[(4-methyl-2-nitrophe-nyl)amino]-3-thiophenecarbonitrile dark-red formto the red form, a 3-D nucleation and growth modelgives the best fit. Relative humidity also acceler-ates the transformation, even though the com-pound is hydrophobic, indicating that even aslight amount of water can act as a nucleationcatalyst by increasing the molecular mobility ofthe molecules.

ACKNOWLEDGMENTS

The authors thank the Purdue–Michigan–Wisconsin Program for the Study of the Chemicaland Physical Stability of Solid Pharmaceuticals aswell as the Purdue Research Foundation forfunding.

REFERENCES

1. Haleblian J, McCrone W. 1969. Pharmaceuticalapplications of polymorphism. J Pharm Sci 58:911–929.

Figure 9. Vapor-sorption curve of the dark-red form(course fraction) with a relative humidity step increaseof 15% at 258C.



2. Haleblian J. 1975. Characterization of habitsand crystalline modification of solids andtheir pharmaceutical applications. J Pharm Sci64:1269–1288.

3. Byrn SR, Pfeiffer RR, Stowell JG. 2000. Chapter 13:Solid-State Chemistry of Drugs. 2nd ed. WestLafayette, IN: SSCI, Inc.

4. 1992. Bioavailability problems from changes inpolymorphism form. Pharm J 10: 479.

5. Randall CS, Dinenno BK, Schultz RK, Dayter L,Konieczny M, Wunder SL. 1995. Solid-state trans-formation of a leukotriene antagonist. Int J Pharm120:235–245.

6. Otsuka M, Kaneniwa N. 1986. Effect of seedcrystals on solid-state transformation of poly-morphs of chloramphenicol palmitate during grind-ing. J Pharm Sci 75:506–511.

7. Gubskaya AV, Chishko KA, Lisnyak YV, BlagoyYP. 1995. Effect of cryogrinding on physico-chemi-cal properties of drugs. Cortisone acetate: Particlessizes and polymorphic transition. Drug Dev IndPharm 21:1965–1974.

8. Otsuka M, Otsuka K, Kaneniwa N. 1994. Relationbetween polymorphic transformation pathway dur-ing grinding and the physicochemical properties ofbulk powders for pharmaceutical preparations.Drug Dev Ind Pharm 20:1649–1660.

9. Wong M, Mitchell A. 1992. Physicochemical char-acterization of a phase change produced during thewet granulation of chlorpromazine hydrochlorideand its effects on tableting. Int J Pharm 88:261–273.

10. Chan HK, Doelker E. 1985. Polymorphic transfor-mation of some drugs under compression. Drug DevInd Pharm 11:315–332.

11. Otsuka M, Matsuda Y. 1993. Effects of environ-mental temperature and compression energy onpolymorphic transformation during tabletting.Drug Dev Ind Pharm 19:2241–2269.

12. Bauer-Brandl A. 1996. Polymorphic transitions ofcimetidine during manufacture of solid dosageforms. Int J Pharm 140:195–206.

13. Yokoyama T, Ohnishi N, Umeda T, Kuroda T, KitaY, Kuroda K, Matsuda Y. 1986. Kinetic study of theisothermal transition of benoxaprofen polymorphsin the solid state at high temperatures. ChemPharm Bull 34:917–921.

14. Umeda T, Ohnishi N, Yokoyama T, Kuroda T, KitaY, Kusoda K, Tatsumi E, Matsuda Y. 1985. Akinetic study on the isothermal transition ofpolymorphic forms of tolbutamide and mefenamicacid in the solid state at high temperatures. ChemPharm Bull 33:2073–2078.

15. Ohnishi N, Yokoyama T, Kiyohara Y, Kita Y,Kuroda K. 1987. Kinetic study on the isothermaltransition of bromovalerylurea polymorphs in thesolid state at high temperature. Chem Pharm Bull35:1207–1213.

16. Sheridan AK, Anwar J. 1996. Kinetics of the solid-state phase transformation of form b to g ofsulfanilamide using time-resolved energy-disper-sive X-ray diffraction. Chem Mater 8:1042–1051.

17. Chen S, Guzei LA, Yu L. 2005. New polymorphs ofROY and new record for coexisting polymorphs ofsolvated structures. J Am Chem Soc 127:9881–9885.

18. Yu L, Stephenson GA, Mitchell CA, Bunnell CA,Snorek SV, Bowyer JJ, Borchardt TB, Stowell JG,Byrn SR. 2000. Thermochemistry and confroma-tional polymorphism of ahexamorphic crystal sys-tem. J Am Chem Soc 122: 585–591.

19. Borchardt TB, Stowell JG, Byrn SR. 1998.The crystallization of 5-methyl-2-[(2-nitrophenyl)a-mino]-3-thiophenecarbonitrile in the presence ofstructurally similar compounds. Mol Cryst LiqCryst 313: 271–276.

20. He X, Griesser UJ, Stowell JG, Borchart TB, ByrnSR. 2001. Conformational color polymorphismand control of crystallization of 5-methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarboni-trile. J Pharm Sci 90:371–388.

21. Galway AK, Laverty GM. 1990. Recent insightsinto dehydration reactions of crystalline solids.J Chim Phys-Chim Biol 87:1207–1221.

22. Petit S, Coquerel G. 1996. Mechanism of severalsolid-solid transformations between dehydratedand anhydrous copper (II) 8-hydroxyquinolinates.Proposition for a unified model for the dehydrationof molecular crystals. Chem Mater 8:2247–2258.

23. Mnyukh YV. 1979. Molecular mechanism of poly-morphic transitions. Mol Cryst Liq Cryst 52:163–200.

24. Chan HK, Grant DJW. 1989. Influence of compac-tion on the intrinsic dissolution rate of modifiedacetaminophen and adipic acid crystals. IntJ Pharm 57:117–124.

25. Chow AHL, Chow PKK, Zhongshan W, Grant DJW.1985b. Modification of acetaminophen crystals:Influence of growth in aqueous solutions containingp-acetoxyacetanilide on crystal properties. IntJ Pharm 25:239–258.

26. Chow KY, Go J, Mehdizadeh M, Grant DJW.1984. Modification of adipic acid crystals:Influence of growth in the presence of fatty acidadditives on crystal properties. Int J Pharm 20:3–24.

27. Chow KY, Go J, Wang Z, Mehdizadeh M, GrantDJW. 1985a. Modification of adipic acid crystals II:Influence of growth in the presence of oleic acidadditives on crystal properties. Int J Pharm 25:41–55.

28. Duddu SP, Fung FKY, Grant DJW. 1994. Effect ofthe opposite enantiomer on the physicochemicalproperties of (�)-ephedrinium 2-naphthalenesulfo-nate crystals. Int J Pharm 94:171–179.

1088 LI ET AL.


29. Thomas JM, Renshaw GD. 1967. Lattice imperfec-tions in organic solids. Trans Farad Soc 63:1922–1928.

30. Begg ID, Halfpenny PJ, Hooper RM, Narang RS,Roberts KJ, Sherwood JN. 1983. X-ray topographicinvestigations of solid state reactions. I. Changes insurface and bulk structure during excipient ther-mal decomposition in sodium chlorate monocrys-tals. Proc R Soc Lond A386:431–442.

31. Pikal MJ, Lukes AL, Lang JE, Gaines K. 1978.Quantitative determination for b-lactam antibio-tics by solution calorimetry: Correlation withstability. J Pharm Sci 67:767–772.

32. Waltersson JO, Lundgren P. 1985. The effect ofmechanical comminution on drug stability. ActaPharm Suec 22:291–300.

33. Exner HE, Rettenmayr M. 1997. Applications ofimage analysis for the quantitative characteriza-tion of microstructural defects in materials. InstPhys Conf Ser No 160:237–244.

34. Rades T, Muller-Goymann CC. 1997. Electron andlight microscopial investigation of defect structuresin mesophases of pharmaceutical substances. Col-loid Polym Sci 275:1169–1178.

35. Burt HM, Mitchell AG. 1981. Crystal defects anddissolution. Int J Pharm 9:137–152.

36. Ester GR, Price R, Halfpenny PJ. 1999. Therelationship between crystal growth and defectstructure: A study of potassium hydrogen phtha-late using X-ray topography and atomic forcemicroscopy. J Pharm D: Appl Phys 32:A128–A132.

37. Uvarov NF, Hairetdinov EF. 1988. The use ofcalorimetric data for estimation of the point defectconcentration in solids. J Thermal Anal 33:279–284.

38. Wunderlich B. 1990. The detection of conforma-tional disorder by thermal analysis. ThermochimActa 162:59–67.

39. Grant DJW, York P. 1986. A disruption index forquantifying the solid state disorder induced byadditives or impurities. II. Evaluation from heat ofsolution. Inter J Pharm 28:103–112.

40. Kontny MJ, Zografi G. 1995. Sorption of water bysolids. Drugs Pharm Sci 70:387–418.

41. Matsuda Y, Tatsumi E, Chiba E, Miwa Y. 1984.Kinetic study of the polymorphic transformationsof phenylbutazone. J Pharm Sci 73:1453–1460.

42. Morris KR, Schlam RF, Cao W, Short MS. 2000.Determination of average crystallite shape by X-ray diffraction and computational methods.J Pharm Sci 89:1432–1442.



investigations on solid–solid phase transformation of...

Documents