Polymorph Screening: Comparing a Semi-Automated Approach with a
`High Throughput Method
`
`Alejandro J. Alvarez, Aniruddh Singh, and Allan S. Myerson*
`
`Department of Chemical and Biological Engineering, Illinois Institute of Technology, Chicago,
`Illinois 60616
`
`Received April 14, 2009; Revised Manuscript Received June 15, 2009
`
`ABSTRACT: Polymorph screening studies of sulfathiazole, mefenamic acid, flufenamic acid, and ROY were carried out using
`a semi-automated apparatus. Cooling crystallization and slurry aging experiments were conducted with varying process
`conditions and a selection of 16 diverse solvents to find as many polymorphic forms as possible. Results yielded four out of five
`polymorphs of sulfathiazole, both polymorphs and a solvate of mefenamic acid, four out of the seven stable forms of ROY, as
`well as the two most commonly encountered polymorphs and a solvate of flufenamic acid. The results obtained in this study were
`compared with a novel high throughput method based on patterned substrates of self-assembled monolayers.17,32,38 It was
`shown that in the case of sulfathiazole and mefenamic acid the same number of polymorphs were obtained using the two
`approaches. In the case of ROY, the semi-automated approach was not able to produce three of the forms found using the
`patterned self-assembled monolayers (SAMs) method. These three forms were found in fewer than 1% of approximately 10 000
`experiments performed using the high throughput approach and thus will be very difficult to find in the 58 experiments
`performed using the semi-automated approach. Results of this study demonstrate that the simple semi-automated approach of
`∼60 experiments described in this work is suitable for early stage polymorph screening as it was able to reproduce effectively the
`diversity of polymorphs in model compounds.
`
`Introduction
`
`The ability of a compound to exist in more than one
`crystalline form is known as polymorphism. The phenomenon
`of a molecule existing in more than one solid-state structure is
`a result of differences in packing arrangement and/or mole-
`cular conformation.1 Different polymorphs of the same com-
`pound exhibit different physical and chemical properties. One
`example of a compound showing such behavior is ritonavir, a
`protease inhibitor, developed by Abbott Laboratories. The
`appearance of a less-soluble second polymorph of ritonavir
`resulted in the need to reformulate the drug two years after it
`was launched.2 In the case of acetaminophen, a well-known
`analgesic drug, form I of the compound lacks slip planes in its
`crystal structure, which make it unsuitable for direct compres-
`sion into tablets. On the other hand, form II of the compound
`has well-developed slip planes which give it processing advan-
`tages over form I.3
`The importance of discovering all polymorphs of an active
`pharmaceutical ingredient cannot be overstated. The late
`discovery of polymorphs can lead to a delay in the time to
`market for a drug. Once a drug is launched, discovery of new
`polymorphs can lead to patent protection issues. The U.S.
`Food and Drug Administration (FDA) also requires char-
`acterization of all possible polymorphs and identification of
`the stable form of a drug. Thus, polymorph screening is
`needed in the early stages of drug development.
`The discovery of polymorphs requires extensive experimen-
`tation. Typically, a variety of factors such as supersaturation,
`agitation rate, cooling rate, solvent composition, temperature,
`seed crystals, additives, impurities, etc. are varied as they are
`known to affect crystallization.4-7 Increasing the number
`
`*To whom correspondence should be addressed. Mail: Philip Danforth
`Armour Professor of Engineering, Department of Chemical and Biological
`Engineering, Illinois Institute of Technology, 10 W. 33rd St., Chicago,
`IL 60616. Phone: 312-567-3101. E-mail: myerson@iit.edu.
`
`of experiments leads to a higher possibility of identifying
`the majority of different polymorphs.8 In a high through-
`put polymorphism study on acetaminophen, Peterson et al.
`obtained form II in only 29 out of 7776 trials.9
`The use of technology to assist in parallel experimentation
`and polymorph screening is becoming increasingly common.
`Recently, Rubin et al. have presented a review of the emerging
`technologies supporting chemical process research and devel-
`opment and their impact on the pharmaceutical industry.10
`The use of automation to carry out experiments helps in
`reducing the time and labor required. As target drug materials
`are often available in limited quantities, methods that utilize
`minimal amount of material are particularly useful. Storey
`et al. presented an automated system for polymorph screening
`in combination with automated isolation of samples. High
`throughput powder X-ray diffraction (PXRD) was used to
`characterize the samples.11 Raman spectroscopy has also been
`used to characterize crystals obtained from high throughput
`experiments and is particularly useful when the characteriza-
`tion needs to be rapid.12 Recently, our group developed a
`small-scale automated solubility measurement apparatus,
`which offers substantial savings in material, time, and labor.13
`This apparatus can also be used for solvent screening before
`polymorph screening experiments are carried out.
`The crystal form produced from solution is the result of
`competing thermodynamic and kinetic factors that govern
`crystallization of polymorphs. The polymorph with lower free
`energy is the thermodynamic stable form, whereas the other
`polymorphs are known as metastable forms. According to
`Ostwald’s rule of stages, the metastable form is the first to
`crystallize, followed by transformation to the more stable
`form.14 This transformation proceeds in many cases through a
`dissolution-recrystallization mechanism. Under certain con-
`ditions, the transformation process can be hindered or sup-
`pressed, leading to the generation of a metastable polymorph
`as the final crystal form.
`
`r 2009 American Chemical Society
`
`Published on Web 07/14/2009
`
`pubs.acs.org/crystal
`
`DOI: 10.1021/cg900421v
`
`2009, Vol. 9
`4181–4188
`
`IPR2018-00126
`
`Page 1 of 8
`
`I-MAK 1016
`
`

`

`The main controlling factors in the crystallization of poly-
`morphs include temperature, supersaturation, and type of
`solvent, as well as the addition of seed crystals, stirring rate,
`and interfaces.15 It is well-known that in enantiotropic sys-
`tems, the thermodynamic stability order among polymorphs
`can be inverted by shifting temperature above and below the
`transition temperature.16 Moreover, the temperature can
`change the dissolution rate, and the kinetics of nucleation
`and growth of each polymorph retarding the appearance of
`certain polymorphs and promoting others. Also, it has been
`shown that a rapid generation of supersaturation provides
`crystals of different polymorphic forms when compared with
`those obtained with a slow increase in supersaturation.17 In
`the case of the effect of solvents, the interactions between
`solute and solvent molecules result in solute molecules assem-
`bling in particular conformation structure an/or packing
`mode.18
`There is, as yet, no failsafe method to predict the extent of
`polymorphism of a given compound. Hence, subjecting the
`active pharmaceutical ingredient (API) to a variety of crystal-
`lization conditions is the only method that can expose the
`diversity of its forms. High throughput polymorph screening
`methods allow researchers to carry out a large number of
`crystallization experiments while providing savings in time,
`material, and labor. Systems such as the fully automated
`crystallization platform CrystalMax, developed by Trans-
`form Pharmaceuticals Inc., are capable of carrying out more
`than 10 000 parallel crystallization experiments using <1 mg
`of the active pharmaceutical ingredient (API) per trial.12
`Symyx Technologies, Inc. has also developed high throughput
`systems which include solid dispensers and liquid handlers
`with complete automation, as well as informatic capabilities
`to support polymorph screening studies.19 However, the high
`cost of these systems makes them unaffordable for a number
`of research laboratories.
`In this work, we evaluated a simple and relatively inexpensive
`semi-automated method to carry out initial polymorph screens.
`We assessed the React Array RS12 from Barnstead Interna-
`tional as a platform for polymorph screening studies. We used
`the RS12 platform to evaluate the effect of initial temperature,
`cooling rate, and type of solvent on the crystallization of
`polymorphic forms of model APIs. Experiments on sulfathia-
`zole (64), mefenamic acid (66), acetaminophen (66), flufenamic
`acid (68), and ROY (58) were carried out and compared to
`a high throughput method developed in this laboratory17
`employing patterned self-assembled monoloayers.
`
`Experimental Section
`
`Materials. Sulfathiazole, 4-amino-N-(2,3-dihydro-2-thiazolyidene)-
`benzenesulfonamide, is an antibacterial drug. It possesses multiple solid
`forms and has been used as a model pharmaceutical compound in
`the study of polymorphism.20,21 Sulfathiazole has five known poly-
`morphs.22 The Cambridge Structural Database reference codes for the
`five forms are Suthaz, Suthaz01, Suthaz02, Suthaz04, and Suthaz05. It
`is also known to form over 100 solvates due to its multiple hydrogen
`bonding capabilities.23
`Mefenamic acid, 2-[(2,3-(dimethylphenyl)amino] benzoic acid, is
`a nonsteroidal anti-inflammatory, antipyretic, and analgesic agent
`used to release pain and inflammation. Mefenamic acid has two
`crystalline forms, form I and form II.24 Forms I and II are
`enantiotropically related with a transition temperature between 86
`to 87 °C. Form I is the stable form below this temperature while
`form II is stable above it.25
`Acetaminophen is an important analgesic and antipyretic drug. It
`is used worldwide in the manufacture of tablets and other dosage
`forms. It has three known polymorphs, forms I, II, and III. Form I is
`
`the thermodynamically stable form at room temperature while form
`III is very unstable. Form I is readily obtained from aqueous
`solution; however, obtaining form II from solution has proved
`difficult. Form II is readily obtained by melt crystallization after
`melting form I.3
`5-Methyl-2-[(2-nitrophenyl)amino]-3-thiophenecarbonitrile, com-
`monly known as ROY for its red, orange, and yellow crystals, is a
`precursor to the antipsychotic agent olanzapine.26 ROY is currently
`the most polymorphic system of known structures. ROY has 10
`known polymorphs, seven with solved structures (Y-yellow prism,
`YN-yellow needle, YT04-Y04 transformed, ON-orange needle, OP-
`orange plate, ORP-orange red plate, and R-red prism) and three
`whose structures have not been solved (Y04-yellow (2004), RPL-red
`plate, and R05-red (2005)). At room temperature, Y is the most stable
`form.27
`Flufenamic acid, 2-([3-(trifluoromethyl) phenyl] amino) benzoic
`acid, is a potent nonsteroidal drug with analgesic, anti-inflamma-
`tory, and antipyretic properties. It has been reported that FFA has
`at least eight polymorphs,28 although forms III and I are the most
`commonly encountered. Most of the other polymorphs can only be
`obtained by sublimation, fusion, or a boiling solvent method, and
`cannot be isolated easily. Form III is the stable form at room
`temperature, and forms III and I are enantiotropic, with a transition
`temperature of 42 °C.
`The pharmaceutical products sulfathiazole, mefenamic acid,
`acetaminophen, and flufenamic acid were purchased from Sigma
`Aldrich Chemicals and were used without further purification.
`ROY as forms R and Y was a gift from Eli Lilly & Company.
`Deionized water was obtained from a Barnstead Nanopure Infinity
`water purification system. N,N-Dimethylformamide (99.95%),
`dimethylsulfoxide (99.99%), ethanol (200 proof), and acetonitrile
`were supplied from Pharmco Products. N,N-Dimethylacetamide
`(99%), formamide (98%), and acetone (99.5%) were acquired from
`Sigma Aldrich Chemicals. 1,4-Dioxane (99%) and chloroform
`(99.8%) were obtained from Fisher Scientific. n-Propanol (99.9%)
`was purchased from Mallinckrodt. Benzonitrile (99%), methyl tert-
`butyl ether (99%), N-methyl pyrrolidone (99%), and o-tolunitrile
`(98%) were supplied from Acros Organics.
`Experimental Apparatus. A Barnstead ReactArray Workstation
`was used to perform crystallization and slurry aging experiments in
`the present work. The workstation integrates a Gilson 175SW liquid
`handler and syringe pump with reaction and reagent racks. The
`dual-syringe pump has two syringes with capacities of 500 μL and
`10 mL. The system has two RS12 reaction racks and each rack holds
`48 glass vials arranged in 12 rows of 4 vials each. The volume of the
`vials is ∼2 mL. Each row in a reaction rack can be given an
`independent temperature profile and the temperature range is -30
`to 150 °C. The maximum controlled heating/cooling rate is 5 °C /
`min while the minimum is 0.1 °C /min. Micro magnetic stirring bars
`can be used for stirring with a stirring speed range of 250-1200 rpm.
`There are two reagent racks in the system that can hold 6 (∼130 mL
`each) and 18 (∼37 mL each) reagent vials, respectively. The system
`is connected to a computer and can be controlled through the
`ReactArray control software.
`A Barnstead Clarity system was used for solubility measurement.
`The solubility measurement was carried out for solvent screening
`purposes before designing polymorph screening experiments. The
`system consists of a RS10 reaction block and a multi-IR unit
`connected to a computer and controlled by the RSPCclient soft-
`ware. Solubility data can be obtained from solution volumes as low
`as 1 mL. The RS10 block has 10 independently controlled cells with
`independent temperature zones and stirring rates. The temperature
`range is -30 to 150 °C. The maximum controlled heating/cooling
`rate is 5 °C /min while the minimum is 0.1 °C /min. The multi-IR unit
`consists of 10 IR turbidity probes. The software generates a plot of
`the IR value vs temperature, and a sharp increase in the IR value at a
`particular temperature indicates a solubility point.
`Crystals obtained were characterized using Raman spectroscopy.
`Raman spectra were obtained using a Raman Microprobe from
`Kaiser Optical Systems, Inc. The Raman microprobe was equipped
`with a 450-mW external cavity stabilized diode laser as the excita-
`tion source, operating at 785 nm. The unit consisted of a Leica
`optical light microscope, a motorized translational stage, and a
`CCD camera. Data were collected with HoloGRAMS version 4.0
`
`4182 Crystal Growth and Design, Vol. 9, No. 9, 2009
`
`Alvarez et al.
`
`IPR2018-00126
`
`Page 2 of 8
`
`I-MAK 1016
`
`

`

`Table 1. Initial Temperature for Cooling Crystallization Experiments
`initial temperature (°C)
`
`compound
`
`solvent(s)
`
`high
`
`intermediate
`
`low
`
`sulfathiazole
`
`90
`W, P
`48
`A, AC
`mefenamic acid BZN, DMF, T, DMA 145
`A, P, MTBE, AN
`50
`acetaminophen DMF, W, DO
`90
`E, P, A, AN
`50
`BZN, DMF, T, DMA
`90
`NMP, DO, AN
`75
`flufenamic acid BZN, DO, T, DMA,
`90
`A, P, CYC, AN
`60
`
`ROY
`
`65
`38
`90
`40
`60
`40
`60
`60
`60
`45
`
`30
`30
`30
`30
`30
`30
`30
`30
`30
`30
`
`and processed and analyzed using GRAMS (Thermo Electron
`Corporation).
`Procedure. The polymorph screening crystallizations were per-
`formed using the Barnstead ReactArray Workstation. Two
`approaches were explored to produce different solid forms of the
`pharmaceutical products: cooling crystallization and slurry aging.
`In cooling crystallization, a solution was cooled at a controlled rate
`to create supersaturation and promote the formation of crystal
`polymorphs. The crystals were immediately characterized using
`Raman spectroscopy to try to prevent their transformation to a
`more stable form. In the slurry aging experiments, particles were
`suspended in different solvents for a long equilibration period to
`allow polymorphic transformation.
`Cooling Crystallization Experiments. A total of 40 solutions of
`sulfathiazole (SZ) were prepared by placing a measured amount of
`solid in 2 mL glass vials. In order to estimate the initial concentra-
`tion of the solutions, preliminary solubility tests were carried out
`using the small-scale automated apparatus developed in our
`group13 to obtain solubility data as a function of temperature. A
`volume of 1.5 mL of solvent was automatically dispensed into each
`glass vial using the robot arm of the Barnstead System. The
`following solvents were used: water (W), n-propanol (P), acetone
`(A), and a mixture (3:2) of acetone/chloroform (AC). Each vial was
`heated to reach the initial temperature, as per the experimental
`design. Three different levels of initial temperature: high (HT),
`intermediate (IT), and low (LT) were explored as shown in Table 1.
`The heating rate was 5 °C/min. Stirring rate was constant during
`the experiment. Vials were maintained at the initial temperature
`for at least 30 min for complete dissolution. Then, solutions were
`cooled down to 10 °C at either a slow (1 °C/min) or fast (5 °C/min)
`cooling rate, according to the experimental design. Once the
`solutions had reached the final temperature, the reflux head was
`removed, and each vial was manually removed from the well plate.
`The crystals were harvested with a spatula and immediately
`analyzed with Raman Spectroscopy. The Raman spectra obtained
`were compared with standard reference spectra of the known
`polymorphs of the compound to identify the type of polymorph
`obtained.
`Mefenamic acid (MA) was crystallized with the same procedure
`as described above. A total of 48 solutions were prepared with 8
`different solvents: benzonitrile (BZN), N,N-dimethylformamide
`(DMF), o-tolunitrile (T), N,N-dimethylacetamide (DMA), acetone
`(A), n-propanol (P), methyl tert-butyl ether (MTBE), and acetoni-
`trile (AN). Forty-two experiments were conducted using acetami-
`nophen with seven different solvents: DMF, water (W), 1,4-dioxane
`(DO), ethanol (E), n-propanol (P), acetone (A), and acetonitrile
`(AN). A total of 42 experiments with ROY were conducted. BZN,
`DMF, o-tolunitrile (T), DMA, NMP, 1,4-dioxane (DO), and
`acetonitrile (AN) were the solvents used. Finally, 48 experiments
`were conducted using flufenamic acid with eight different solvents:
`benzonitrile (BZN), 1,4-dioxane (DO), o-tolunitrile (T), N,N-
`dimethylacetamide (DMA), acetone (A), n-propanol (P), cyclohexane
`(CYC), and acetonitrile (AN).
`Slurry Aging Experiments. A total of 24 suspensions of sulfathia-
`zole (SZ) were prepared by placing an excess of solid in 2 mL glass
`vials and adding 1.5 mL of the following solvents: DMSO, BZN,
`DMF, DMA, formamide (F), NMP, acetone (A), MTBE, n-pro-
`panol (P), water (W), ethanol (E), and cyclohexane (CH). Each
`experiment was duplicated.
`
`Table 2. Experimental Temperature for Slurry Aging Experiments
`
`compound
`
`solvent(s)
`
`sulfathiazole
`
`DMSO, BZN, DMF, DMA, F, NMP
`A, MTBE
`P, W, E, CH
`mefenamic acid BZN, DMF, T, DMA
`A, MTBE
`P, AN, W
`acetaminophen DMF, DMSO, F, NMP
`W, DO
`E, P
`A, MTBE
`CH, AN
`AN
`F, W, DMSO, P, BZN, T, DO
`flufenamic acid BZN, DO, T, DMA
`A, P, CYC, AN, W, M
`
`ROY
`
`temperature
`(°C)
`110
`45
`70
`110
`40
`70
`110
`85
`60
`45
`70
`70
`90
`90
`60
`
`Each vial was heated to reach the experimental temperature, as
`per the experimental design shown in Table 2. The heating rate was
`5 °C/min. Stirring rate was constant during the experiment. Vials
`were maintained overnight at the experimental temperature. At the
`end of the experiment, the reflux head was removed and each vial
`was manually removed from the well plate. The crystals were
`harvested with a spatula and immediately analyzed with Raman
`spectroscopy. The Raman spectra obtained were compared with
`standard reference spectra of the known polymorphs of the com-
`pound to identify the type of polymorph obtained. Eighteen experi-
`ments were conducted using mefenamic acid with nine different
`solvents: BZN, DMF, o-tolunitrile, DMA, acetone, n-propanol
`MTBE, acetonitrile, and water, and 24 experiments were conducted
`using the compound acetaminophen with 12 different solvents:
`DMF, DMSO, water, 1,4-dioxane, ethanol, n-propanol, acetone,
`MTBE, cyclohexane, acetonitrile, formamide, and NMP. Sixteen
`experiments were conducted using ROY with eight different sol-
`vents: acetonitrile, formamide, water, DMSO, n-propanol, BZN,
`o-tolunitrile, and 1,4-dioxane. Finally, 20 experiments were con-
`ducted using flufenamic acid with 10 solvents: BZN, 1,4-dioxane,
`o-tolunitrile, DMA, acetone, n-propanol cyclohexane, acetonitrile,
`water, and methanol.
`
`Results and Discussion
`
`Sulfathiazole. Cooling Crystallization Experiments. Four
`out of the five polymorphs of sulfathiazole were obtained in
`our experiments, as shown in Table 3. There is some confusion
`in the literature regarding the nomenclature for different
`polymorphs of sulfathiazole as noted by Blagden et al.29 and
`Apperley et al.21 We have used the notation of the Cambridge
`Structural Database reference codes in this report. The stabi-
`lity order of the polymorphs is Suthaz04>Suthaz02>Suthaz>
`Suthaz01>Suthaz05.22,29 Figure 1 shows the Raman spectra
`of four of the five polymorphs of sulfathiazole.
`Suthaz02 and Suthaz04 were obtained in cooling crystal-
`lization experiments with water as the solvent. When solu-
`tions were cooled from high temperature Suthaz04 crystals
`were obtained in both fast and slow cooling experiments.
`Fast cooling from intermediate temperature gave Suthaz02
`crystals in one vial and Suthaz04 crystals in the second vial.
`No crystals were obtained in fast cooling from low tempera-
`ture experiments while slow cooling from low temperature
`gave Suthaz02 crystals. Blagden et al. have previously
`reported that crystallization of sulfathiazole from water
`favors Suthaz04.30
`Suthaz01, Suthaz02, and Suthaz04 were obtained from
`n-propanol solutions. It has been reported in the literature
`that crystallization in n-propanol favors Suthaz01.30,31 How-
`ever,
`in our experiments three different polymorphs of
`sulfathiazole were obtained from n-propanol solutions.
`
`Article
`
`Crystal Growth and Design, Vol. 9, No. 9, 2009
`
`4183
`
`IPR2018-00126
`
`Page 3 of 8
`
`I-MAK 1016
`
`

`

`Table 3. Results Obtained from Sulfathiazole Cooling Crystallization Experiments
`
`initial temperature
`
`solvent
`
`5 °C/min
`
`water
`
`Suthaz04
`
`1 °C/min
`Suthaz04
`
`high
`
`n-propanol
`
`Suthaz02 (vial 1) and
`Suthaz04 (vial 2)
`SuthazþSuthaz02 mixture
`acetone
`(vial 1) and Suthaz02 (vial2)
`acetone/chloroform SuthazþSuthaz02
`mixture (vial 1) and
`Suthaz02 (vial 2)
`
`Suthaz02 (vial 1)
`and Suthaz04 (vial 2)
`SuthazþSuthaz02
`mixture
`SuthazþSuthaz02
`mixture
`
`intermediate
`5 °C/min
`Suthaz02 (vial 1)
`and Suthaz04 (vial 2)
`Suthaz01 (vial 1)
`and Suthaz04 (vial 2)
`SuthazþSuthaz02
`mixture
`SuthazþSuthaz02
`mixture
`
`low
`
`5 °C/min
`no crystals
`
`no crystals
`SuthazþSuthaz02
`mixture
`SuthazþSuthaz02
`mixture
`
`1 °C/min
`Suthaz02
`
`Suthaz02
`
`Suthaz02
`SuthazþSuthaz02
`mixture
`
`Table 4. Results Obtained from Sulfathiazole Slurry Aging Experiments
`
`solvent
`
`DMSO
`BZN
`DMF
`DMA
`formamide
`NMP
`acetone
`MTBE
`n-propanol
`water
`ethanol
`cyclohexane
`
`polymorph
`
`no crystals
`no crystals
`no crystals
`no crystals
`no crystals
`no crystals
`Suthaz02 (100%)
`Suthaz02 (100%)
`Suthaz02 (100%)
`Suthaz02 (100%)
`Suthaz02 (100%)
`Suthaz02 (100%)
`
`diversity particularly for compounds such as sulfathiazole.
`The use of automation in experimentation helps in carrying
`out a high number of experiments while providing savings in
`time and labor.
`Slurry Aging Experiments. Crystals were obtained in 50%
`of the experiments. In the remaining 50% of the experiments
`the solute was dissolved in the solvent. Because of the high
`solubility of sulfathiazole in some solvents it was not possible
`to form a slurry in the 2 mL reaction vials. No polymorphic
`transformation was observed in the slurry aging experi-
`ments, as shown in Table 4. The Raman spectra of all the
`crystals obtained matched that of form Suthaz02, which is
`the commercial form.
`Comparison of Results Obtained with a High-Throughput
`Approach. Recently, our group has developed patterned
`substrates of self-assembled monolayers (SAMs) which can
`be used to carry out a large number of independent crystal-
`lization trials with a minimal amount of material.17 We have
`previously used this method to perform polymorph screen-
`ing experiments with sulfathiazole.32 It is important to note
`here that the experiments carried out in each case were
`different; in the SAMs experiments evaporation of solvent
`was used to create supersaturation while cooling crystal-
`lization and slurry aging experiments were carried out in the
`present work. However, it is interesting to compare the
`results obtained from a polymorph screening perspective.
`When comparing the results obtained in our current
`experiments with the SAMs experiments we find that in the
`case of sulfathiazole the same four polymorphs (out of the
`five known forms) were obtained using the two approaches.
`Although the amount of material required per crystallization
`trial is low when using our present approach, it is even lower
`in the SAMs experiments, for example, in the sulfathiazole
`cooling crystallization experiments, the amount of sulfathia-
`zole required for each trial varied from 0.6 to 33.75 mg. In the
`case of SAMs, the material required per trial is often as low as
`0.01-0.02 mg. When studying the effect of solvent on the
`
`Figure 1. Raman spectra of sulfathiazole polymorphs.
`
`Suthaz02 and mixtures of Suthaz and Suthaz02 were
`obtained. It has been previously reported that only Suthaz01
`and Suthaz can be obtained from acetone, while Suthaz01,
`Suthaz04, and Suthaz can all be obtained from acetone/
`chloroform (3:2).31
`Effect of Solvent on the Polymorphic Outcome of Sulfathia-
`zole. Sulfathiazole has been previously used as a model
`compound to study the effect of solvent on crystallization of
`polymorphs.30,31 It has been reported that crystallization of
`sulfathiazole from n-propanol solutions favors Suthaz01.
`However, in a paper on solvates of sulfathiazole, Bingham
`et al. have noted that sulfadrugs crystallize erratically from
`solution, despite the contrary impression that might be gained
`from the literature.23 Lee et al. also reported that although the
`type of solvents employed can influence the crystallization
`outcome, sulfathiazole might not be an accurate example
`of this behavior.32 Hughes et al. have also noted the erratic
`crystallization of sulfathiazole, usually as mixtures of poly-
`morphs, from solution and how guaranteed recipes for pro-
`ducing single polymorphs are difficult to obtain.33
`Our results using the semi-automated polymorph screen-
`ing equipment also support the latter view as mixtures of
`polymorphs were frequently obtained and forms obtained
`from particular solvents were different than those previously
`reported. When carrying out experiments with n-propanol as
`the solvent, we were able to obtain forms Suthaz01,
`Suthaz02, and Suthaz04 contrary to previous reports.30,31
`In the case of water Suthaz04 and Suthaz05 have been
`reported to be the preferred forms; however, we obtained
`Suthaz02 and Suthaz04. Because of the stochastic nature
`of nucleation from solution extensive experimentation is
`needed to acquire a better understanding of solid form
`
`4184 Crystal Growth and Design, Vol. 9, No. 9, 2009
`
`Alvarez et al.
`
`IPR2018-00126
`
`Page 4 of 8
`
`I-MAK 1016
`
`

`

`Table 5. Mefenamic Acid Polymorphs Obtained in Cooling Crystalliza-
`tion Experiments
`
`initial temperature
`
`high
`intermediate
`low
`5 °C/min 1 °C/min 5 °C/min 1 °C/min 5 °C/min 1 °C/min
`I
`I
`I
`II
`I
`I
`II
`II
`II
`II
`II
`II
`I
`I
`I
`I
`I
`I
`I
`I
`I
`II
`I
`II
`II
`I
`I
`I
`I
`II
`I
`II
`II
`I
`I
`I
`I
`
`I
`II
`I
`I
`I
`
`I
`I
`II
`I
`
`solvent
`
`benzonitrile
`DMF
`o-tolunitrile
`DMA
`acetone
`n-propanol
`MTBE
`acetonitrile
`
`Figure 2. Raman spectra of mefenamic acid polymorphs.
`
`polymorphic outcome and to compare the results with those
`obtained while carrying out conventional crystallization
`experiments, the present approach is more suited. This is
`because in the present approach each trial is similar to a
`conventional crystallization experiment while the scale of the
`experiments is smaller to provide savings in material and
`some degree of automation is added to provide savings in
`time and labor. In the case of SAMs template nucleation
`takes place and factors such as the monolayer may affect the
`polymorph obtained in an experiment. The total number of
`islands (crystallization trials) tested for sulfathiazole using
`patterned SAMs was 4200.
`Mefenamic Acid. Cooling Crystallization Experiments.
`Crystals were obtained in 96% of the cooling crystallization
`experiments. Lee et al., reported three distinct characteristic
`peaks for the two polymorphic forms of mefenamic acid.32
`Raman characteristic peak positions are 623, 702, and
`-1 for Form I, and 631, 694, and 1573 cm
`-1 for
`1581 cm
`Form II. Figure 2 shows the Raman spectra of mefenamic
`acid polymorphs. Comparing the experimental Raman spec-
`tra obtained in each experiment against the characteristic
`peak position, Form I was identified in 54% of the experi-
`ments, and Form II in 19% as shown in Table 5.
`Both polymorphs were nucleated by cooling crystalliza-
`tion in methyl tert-butyl ether (MTBE). Form I was observed
`at a slow cooling rate from high and intermediate tempera-
`ture. At a fast cooling rate, Form II was obtained from high
`and intermediate temperature. The metastable Form II was
`also observed when crystallized from low temperature at fast
`and slow cooling rates.
`It is known that metastable solid forms are favored with
`the creation of high supersaturation. Lee et al. obtained the
`metastable β-glycine as a result of the high supersatura-
`tion generated on confined engineered surfaces.17 Kitamura
`observed that only the metastable B-form of L-hystidine
`crystallized by rapid cooling a mixed solvent water-ethanol
`solution with high ethanol fraction.34 The preferred appear-
`ance of the metastable form is observed when a less stable
`state can be reached faster because its kinetics is faster than
`the stable state. In our experiments, the appearance of the
`metastable form II of mefenamic acid with MTBE at a fast
`cooling rate at high, intermediate, or low temperature can be
`explained as the result of the high supersaturation that is
`generated from the rapid cooling.
`The metastable form II of mefenamic acid was obtained
`from acetone by cooling crystallization in four out of six
`
`Table 6. Mefenamic Acid Polymorphs obtained in Slurry Aging Experi-
`ments
`
`solvent(s)
`
`benzonitrile
`DMF
`o-tolunitrile
`DMA
`acetone
`propanol
`MTBE
`acetonitrile
`water
`
`polymorph
`
`I
`II
`I and II
`
`I
`I
`I
`I
`I
`
`experiments. These results can be explained as an experimental
`example of Ostwald’s rule, which suggests that metastable
`form is the first crystal form to crystallize, followed by solvent
`mediated transformation to the stable form. Moreover, the
`stable form I was observed when crystallized from high and
`intermediate temperature with a fast and slow cooling rate,
`respectively. The increase in the rate of transformation of
`form II into form I with increased temperature, reported by
`Osuka,35 may explain the appearance of the stable form I at
`high and intermediate temperature.
`Interaction between solvent and solute molecules can pro-
`mote the nucleation of a metastable form and inhibit the
`formation of the stable form. Blagden29 explained this phe-
`nomenon as a result of inhibiting nucleation and/or growth by
`adsorbing on the fastest growing faces of the crystal. The
`metastable Form II of mefenamic acid was obtained through
`cooling crystallization from DMF at all experimental condi-
`tions. These results agree with data obtained by Aguair,36
`Otsuka,35 and Cesur,37 who observed that crystallization
`of MA form II is induced by DMF. Together with the
`metastable form II, the presence of a solvate was observed,
`as indicated by additional vibrational bands in the Raman
`spectra. This solvate has been previously reported by Lee.28
`Metastable form II of mefenamic acid was also obtained
`with benzonitrile when crystallized from intermediate tem-
`perature at a slow cooling rate. This may be due to the
`stochastic nature of nucleation of different polymorphic
`forms. Other solvents were screened with mefenamic acid,
`including o-tolunitrile, N,N-dimethylacetamide, n-propanol,
`and acetonitrile. Under the experimental conditions explored
`with these solvents, form I was solely observed.
`Slurry Aging Experiments. Crystals were obtained in 89%
`of the slurry aging experiments, as shown in Table 6. Form I
`was obtained in most of the solvents, except for DMF and
`o-tolunitrile. In the case of DMF, it was previously observed
`in the cooling crystallization experiments that this solvent
`favored the formation of the metastable Form II.
`A mixture of forms I and II was observed in the slurry
`aging experiments with o-tolunitrile at 110 °C as evidenced
`
`Article
`
`Crystal Growth and Design, Vol. 9, No. 9, 2009
`
`4185
`
`IPR2018-00126
`
`Page 5 of 8
`
`I-MAK 1016
`
`

`

`Table 7. Results Obtained from ROY Cooling Crystalliz