`
`Polymorphism in Process
`Development
`
`MIKEBAVIN
`
`Unexpected crystal structure transformations can lead to critical changes in chemical behaviour.
`Mike Bavin proposes that the occurrence of crystal polymorphs should be greeted as a commercial
`opportunity in the pharmaceutical industry, and draws on his experience at Smith Kline & French to
`demonstrate that results can be achieved quickly with some collaboration.
`
`C rystalline products are generally the easiest to isolate,
`
`purify, dry and, in a batch process, handle and fornm(cid:173)
`late. Despite these advantages, the process of crystalli(cid:173)
`sation is taken for granted by most chemists, and it takes a
`reaction vessel clogged with an unstirrable mass to provoke
`serious thought. Such an occurrence is likely to be due to the
`appearance of a previously unknown polymorph or solvate.
`The shapes of crystals are called habits: needles, cubes,
`plates. These can be altered by inhibiting growth on particular
`faces and forcing it onto others. Changes of this sort are
`unlikely to be caused by alteration of the impurity profile of an
`established product, and sudden changes of crystal habit are
`much more likely to be due to the appearance of a polymorph
`or solvate.
`Polymorphs have crystal lattices which differ in the ways in
`which the same molecule is bound in the unit cell. The differ(cid:173)
`ences may reflect different ways of packing molecules in the
`cell, or confonnational changes, which can be
`large.
`Hydrogen-bonding will be involved for most molecules of
`interest to the phannaceutical industry. It is self-evident that
`polymorphs will have different lattice energies, melting points,
`heats of fusion, solubilities and rates of solution. Differences
`can also be expected in density, hardness, etc. There are obvi(cid:173)
`ous implications for the pharmaceutical industry;
`Two types of polymorphs are distinguished, defined in text(cid:173)
`books by the use of pressure-temperature (PT) diagrams. For
`monotropic polymorphs, only one fonn is thennodynamically
`stable at all temperatures at which the solid exists. The other
`fonns are metastable and may transfonn into the stable one,
`particularly in the presence of a solvent. Despite the art of
`phannacists, it seems sensible and prudent to use the stable
`polymorph for preparing suspensions. The presence of water in
`the process of granulation for tabletting can lead to polymor(cid:173)
`phic transitions.
`Enantiotropic compounds have two or more polymorphs,
`each of which is thennodynamically stable over a limited tem(cid:173)
`perature range. The limits of these ranges are marked by transi(cid:173)
`tion temperatures, and by the melting point at the upper limit.
`In this case it is simplest to prepare suspensions using the poly-
`
`morph which is stable over the ambient temperature range.
`Obviously a transition temperature in the range 15-25'C can
`cause difficulties.
`
`Crystals and crystallisation
`The process of crystallisation and the factors governing the
`appearance of individual polymorphs are still poorly under(cid:173)
`stood. Crystals are in equilibrium only with a saturated solu(cid:173)
`tion but crystallisation requires supersaturation, so that the
`Phase Rule is not applicable. Which of several possible poly(cid:173)
`morphs is obtained seems to depend upon various factors: the
`temperature at which crystallisation occurs, the nature of the
`solvent (hydrophilic, hydrophobic) and the degree of supersat(cid:173)
`uration when crystallisation commenced. The use of seed crys(cid:173)
`tals can be helpful
`in obtaining a desired polymorph.
`Manufacturing processes seem to be worked out by trial and
`error aided by serendipity, and then adhered to rigidly. Even
`so, unwanted polymorphs are likely to be obtained occasional(cid:173)
`ly and have to be reworked.
`All decisions are based upon analytical data so that means
`are required for identifying polymorphs. They exist only in the
`solid state and infrared spectroscopy and X-ray diffraction are
`the most generally useful techniques. Nuclear magnetic reso(cid:173)
`nance (NMR) spectroscopy of solid samples is less readily
`available but can be valuable despite complications arising
`from having molecules with different confonnations in the unit
`cell. Quantitative analyses are time-consuming by X-ray pow(cid:173)
`der diffraction. They are carried out much more quickly by
`infrared spectroscopy particularly now that spectrometers have
`laser sources and dedicated computers with powerful software.
`A major requirement for quantitative analysis is to have sam(cid:173)
`ples of the pure polymorphs.
`It is essential to keep in mind that analyses are not only as
`good as the samples. Grinding is an energetic process and can
`bring about polymorphic changes. It may need to be limited to,
`say, 20 seconds, for reproducible results to be obtained. The
`self-orientation of crystal fragments can influence X-ray pow(cid:173)
`der diffractograms. Whether or not a sample is truly represen(cid:173)
`tative and how it is prepared require careful attention.
`
`Chemistry & Industry 21 August 1989 527
`
`Merck Exhibit 2169, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`Polymorphic transitions can occur during granulation for tabletting.
`
`Market opportunities
`Few compounds reach development and fewer still are ever
`marketed. In giving each development candidate the best
`chance of progressing, it seems better to search for polymor(cid:173)
`phs rather than to leave their appearance to time and chance
`with the consequent disruption. The methods used in attempt(cid:173)
`ing to obtain polymorphs include crystallisation from a range
`of solvents (polar and non-polar, hydrophilic and hydrophobic)
`at different temperatures, by chilling solutions rapidly, by
`adding a second solvent in which the solute is sparingly solu(cid:173)
`ble, by vigorously stirring excess solid with solvent, by heating
`excess solid with a high boiling solvent, by sublimation, and
`by very rapidly changing the pH of a solution to precipitate
`acidic or basic substances. Samples can then be prepared for
`evaluation and for use for reference. The identity and purity of
`such samples should always be checked, for example by solu(cid:173)
`tion NMR spectroscopy and high performance liquid chro(cid:173)
`matography.
`The discovery, whether accidental or intended, of poly(cid:173)
`morphs is unlikely to be greeted with enthusiasm by senior
`management, and the situation is better treated as an opportu(cid:173)
`nity rather than as a problem. Opportunities are likely to exist
`for increasing patent cover, for retaining a competitive edge
`through unpublished knowledge and in formulating pharma~
`ceutical products. A metastable polymorph can be used in cap(cid:173)
`sules or for tabletting, and the thermodynamically stable one
`for suspensions.
`The bringing together of chemists, pharmacists, toxicolo(cid:173)
`gists, biochemists, patent attorneys and colleagues from mar(cid:173)
`keting at an early stage in development can only be beneficial.
`The essential questions can then be posed so that the research
`
`528 Chemistry & Industry 21 August 1989
`
`is limited rather than open-ended. The following outlines ways
`and means of proceeding.
`The thermodynamically stable polymorph needs to be iden(cid:173)
`tified. If the compound is enantiotropic, there will be two or
`more stable polymorphs and transition temperatures as well.
`These can be identified by simple techniques, for example by
`stirring or shaking excess solid with solid at different tempera(cid:173)
`tures. The polymorphic form of the residual solid must be
`determined. These experiments can be scaled down to use a
`few crystals in a small dish of solvent on the stage of a hot
`stage microscope. If two or more polymorphs are available in
`gram quantities, then attempts should be made to determine
`their solubilities. This can be done using a shake-flask tech(cid:173)
`nique and a good thermostatted bath. Alternatively a filtered
`solution from a stirred suspension can be passed through the
`sample cell of an ultraviolet spectrometer. The plateau value of
`the absorbance is a measure of the solubility.
`Whichever method is used, it is essential that the solid
`residue consists of only one polymorph, otherwise the system
`cannot have reached equilibrium. This follows from applica(cid:173)
`tion of the Phase Rule. For this reason solubility measurements
`must be made at lea.st in duplicate. Perseverance may be neces(cid:173)
`sary due to polymorphic transitions but with luck as well, it
`should be possible to obtain solubility-temperature curves for
`the polymorphs. At any given temperature, the thermodynami(cid:173)
`cally stable polymorph will have the lowest solubility.*
`Transition temperatures will be marked by the intersections of
`the solubility curves.
`The differences in the physical properties of the polymorphs
`
`• This makes the reasonable assumption that the heat of solution is negative.
`
`Merck Exhibit 2169, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`~98°C
`
`-272"C
`
`-295°C
`
`NH·COCH 3
`
`Hydrate B
`
`Water slurry
`
`to pH6.0
`
`O
`
`l
`ij
`N e
`'-::,
`N
`Y ~ Hydrate A
`Water 1 slurry" B5"C
`Pseudomorph
`~ Polymorph l
`81
`Polymorph II
`
`occluded solvent can be established by solution NMR spec(cid:173)
`troscopy and by gas liquid chromatography (with mass spec(cid:173)
`trometry ifavailable).
`
`Hydrates and polymorphs of SK&F 94120
`The following is an example of what can be achieved quickly
`by collaboration.
`SK&F 94120 is weakly basic and this was made use of in its
`purification: a solution in alkali was filtered to remove impuri(cid:173)
`ties and the filtrate adjusted to pH 6. The precipitate (hydrate
`B) was finely divided and very difficult to collect and wash,
`but was prepared as a suspension for toxicological studies. In a
`short time, samples of the suspension recrystallised as a solid
`mass of long yellow needles, useless for the purpose for which
`they had been prepared. The new crystals were also a hydrate,
`A. Both hydrates A and B were characterised by DSC-TGA
`and infrared spectroscopy and X-ray powder diffraction** but
`water was lost too readily for their stoichiometry to be deter(cid:173)
`mined. We then showed that hydrate A could be obtained by
`slurrying B, simplifying the isolation of SK&F 94120 (see
`Figure).
`Hydrates B and A could be dried in an oven at 110-120°C to
`give anhydrous SK&F 94120 (polymorph I). DSC-TGA
`showed that after hydrate A had lost its water, a small
`exotherm occurred at 98°C. This was interpreted as rearrange(cid:173)
`ment of the pseudomorph t of hydrate A to polymorph I. The
`behaviour of SK&F 94120 close to its melting point is com(cid:173)
`plex. Three endotherms are usually observed in the range 272-
`295°C (Figure), suggesting polymorphic transitions prior to
`melting. Sublimation complicates the situation and reduces the
`usefulness of hot stage microscopy. However, we were able to
`obtain a second polymorph (II) by heating excess SK&F 94120
`with boiling ethyl benzoate. The samples always contained
`approximately 2 per cent of chemical impurities.
`Finally, efforts to determine the transition temperature
`hydrate A-polymorph I showed that polymorph I was obtained
`from aqueous suspensions above ~85°C, facilitating drying.
`Large quantities of hydrate A and polymorph I were prepared
`before the project was terminated. The other polymorphs indi(cid:173)
`cated by DSC were not investigated because they did not con(cid:173)
`tribute to our objectives.
`
`** Paul Brush was able to study the transfonnation of hydrate B to hydrate A in the
`X.-ray powder diffractometer, using an aqueous suspension sealed in the sample holder
`with cling-film. A heated cell holder was very useful for infrared spectroscopic
`studies.
`
`t The pseudomorph is equally-well described as an unstable polymorph.
`
`[
`i
`1<- Acknowledgement. The directors of SK&F Ltd are thanked
`[
`for permission to quote unpublished data. I am particularly
`indebted to my former colleagues for their cooperation and
`suggestions, and to W A M Duncan for his trust.
`
`This paper and the following article were presented at a meeting of the Fine
`Chemicals Group of the SCI, held in London on 29 November 1988.
`
`Figure Known relation of SK&F 94120 hydrates and polymorphs and DSC
`trace for a sample of SK&F 94120
`
`of organic compounds will probably not be large except for
`bulk density. Melting points may differ by just a few degrees
`and data from differential scanning calorimetry (DSC), differ(cid:173)
`ential thermal analysis (DTA) and hot stage microscopy are
`likely to be difficult to interpret. This is because the poly(cid:173)
`morphs will have been crystallised from different solvents and
`may have different impurity profiles. But the combined tech(cid:173)
`nique of DSC-thermogravimetric analysis (TGA) (DTA-TGA)
`is essential for studying solvates, which are as likely as poly(cid:173)
`morphs to appear during development. The identity of the
`
`Polymorph I shows melting and crystal change at 252°C (above);
`polymorph II at 250°C (below).
`
`Dr Bavin was formerly with Smith Kline & French. He is now at 2 Sol!ershot
`West, Letchworth, Hertfordshire SG6 JPX.
`
`Chemistry & Industry 21 August 1989
`
`529
`
`Merck Exhibit 2169, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`