Merck Exhibit 2173, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`Merck Exhibit 2173, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`/
`/
`CRYSTALLIZATION IN THE PHARMACEUTICAL AND
`BIOPROCESSING INDUSTRIES
`D J. Kirwan and C J. Orella
`
`11.1. THE ROLE OF CRYSTALLIZATION IN BIOPROCESSES
`
`The application of crystallization in the pharmaceutical industry
`directly parallels crystallization in other industries. There is a need
`to control particle size distribution through control of crystal
`growth versus nucleation and to control the purification achieved
`through crystallization. The latter requirement often requires dif(cid:173)
`ferent approaches and operating conditions than those for opti(cid:173)
`mizing yield. Precipitation is commonly used in the pharmaceutical
`industry. [In this chapter, we shall use the term precipitation to
`mean the creation of a solid phase (crystalline or not) by the
`addition of an agent. Precipitation historically often referred to
`reactions resulting in the formation of a sohd phase.] What is
`unique in the pharmaceutical industry is the chemical complexity
`of the entities that are crystallized. This complexity and chemical
`diversity impacts the thermodynamics (solubility and crystal struc(cid:173)
`ture) as well as the kinetics of crystallization. Several examples of
`the chemical diversity of products crystallized in the pharmaceut(cid:173)
`ical industry are shown in Figure 11.1. In this chapter we will focus
`on low molecular weight pharmaceutical compounds while protein
`crystallization is discussed in Chapter 12.
`The majority of these pharmaceutical compounds are between
`100 and 1000 daltons, and exhibit a great diversity of functional
`groups ranging from ionic moieties to very lipophilic or hydro(cid:173)
`phobic groups. Thus, their interactions with one another, with
`solvents or anti-solvents, and with co-solutes and impurities in
`solution are very diverse. The sohd phases (including polymorphs
`and various solvates) formed by such molecules are very poorly
`understood. The transformation rate between such solid phases
`may be kinetically limited. Therefore, sohds of differing character(cid:173)
`istics or even immiscible hquid phases, "oiling out," might be
`obtained from different modes of crystallization. This complexity
`is compounded by the limited experimental studies on such com(cid:173)
`pounds from which generalizations can be made.
`Owing to the final use of such compounds, strict control is
`required on their purity, crystal form and morphology, and par(cid:173)
`ticle size distribution (PSD). All of these characteristics are governed
`by the crystallization process. Obviously, control of purity is of
`great importance for products intended for human medicinal use
`in order to minimize exposure to anything other than the intended
`therapeutic agent. But, equally important is the PSD, which can
`dramatically impact the in-vivo dissolution of a drug, especially
`one that is hydrophobic and has limited solubility in aqueous
`solutions. This "bioavailabiUty" can be strongly hindered if many
`larger particles are present, and is enhanced by the presence of
`predominantly smaller particles. Less obvious is that a change in
`crystal morphology (shape) or crystal structure can impact the
`bioavailabihty. In addition, these same properties of the crystals
`can play a dramatic role in the stabihty of the product; and,
`therefore, its purity at time of use. Whereas small particle size is
`generally good for bioavailability, it is a disadvantage to crystals
`
`subject to oxidation during processing or storage because of the
`greater surface area per crystal mass.
`Crystallization also is employed as an intermediate purifica(cid:173)
`tion step in many processes because of good separation factors per
`stage and its effectiveness at low temperatures for thermally labile
`compounds. Particle size and habit are important in these steps
`as well because of their effect on filtration or centrifugation rates.
`The influence of crystallization conditions on morphology and
`PSD, and, therefore, on the "de-hquoring" characteristics is often
`overlooked when laboratory work is conducted. However, this
`becomes much more important in pilot- or full-scale manufactur(cid:173)
`ing where slow "de-Uquoring" can result in low productivity and
`reduced stability.
`There are several common problems encountered in the use of
`crystalHzation in the pharmaceutical industry; (1) the control of
`supersaturation (and PSD) in a batch crystallizer; (2) the effective
`use of seed; (3) efficient measurement of solubilities in multiple
`solvent systems to maximize purification and yield; and (4) identi(cid:173)
`fication and retention of the most stable polymorphic form.
`As stated above, control of the crystalHzation or precipitation
`process is essential to obtain crystals of biochemical compounds
`having appropriate properties. The phenomena, techniques, and
`analysis discussed in many of the previous chapters: solubility
`and supersaturadon, nucleation and growth kinetics, populadon
`balance methods, batch and continuous crystalUzers, and factors
`governing crystal purity, habit and morphology are all relevant to
`the discussion of the crystallization of pharmaceuticals. We shall
`analyze the crystallization/precipitation of biomolecules in terms
`of these concepts.
`
`11.2. SOLUBILITY AND THE CREATION
`OF SUPERSATURATION
`
`Crystallization obviously requires the creation of a condition
`where the equilibrium solubility value is below that of the concen(cid:173)
`tration of solute in the solution. Both growth and nucleation rates
`are dependent upon the departure of the solution conditions from
`equilibrium values. Further, the single-step yield achieved in a
`given crystallization is directly related to the equilibrium amount
`of solute remaining in solution, after nucleation and growth have
`relieved supersaturation. Crystallization can be accomplished by
`reducing the temperature as in a cooling crystallizer, by removing
`solvent as in an evaporative crystalHzer, or by altering the compos(cid:173)
`ition of the solution by the addition of acid, base, miscible anti-
`solvents, or salts as in a precipitating crystallizer. These techniques
`also can be used in concert to accomplish the desired solubility
`reduction. While solubihty is a thermodynamic variable not influ(cid:173)
`enced by the mode of crystallizing; in precipitation crystallization,
`the mode of addition in a batch crystallizer can be used to create
`very different transient conditions of solubility and supersaturation.
`
`249
`
`Merck Exhibit 2173, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`250 CRYSTALLIZATION IN THE PHARMACEUTICAL AND BIOPROCESSING INDUSTRIES
`
`HgN""
`
`coo-
`
`C
`
`H
`
`coo-
`
`coo-
`
`H
`
`HgN"*"
`
`C
`
`H
`
`HgN"*"
`
`C
`
`H
`
`CH.
`
`CH3CONH2
`
`Glycine
`
`Alanine
`
`Asparagine
`
`coo-
`
`HgN""
`
`C
`
`H
`
`coo-
`
`CH.
`
`HgN^
`
`C
`
`H
`
`Phenylalanine
`
`COO
`
`CHgCHgCHgCONH
`
`' — N,
`
`Cephalosporin
`
`CO2H
`
`coo-
`
`Protein
`
`Figure 11.1 Molecular structures of various biochemicals.
`
`That is, adding nonsolvent to the solute in solution is very different
`from the "reverse addition" of solution to nonsolvent and both are
`different from a rapid in-Hne blending to the final proportions.
`
`11.2.1. TEMPERATURE EFFECTS ON SOLUBILITY
`
`Most solutes, whether biological in nature or not, exhibit increas(cid:173)
`ing solubility with increasing temperature. For example. Figure
`11.2 shows the solubilities of citric acid and glutamic acid in water
`as a function of temperature. There can be complicating factors in
`the solubility behavior related to the actual form of the solute in
`the crystal phase and in solution as evidenced by the various forms
`of glutamic acid. For the case of citric acid a monohydrate exists
`below about 37 °C and the anhydrous form at higher temperatures.
`Although it can be useful to raise the temperature to increase the
`amount of material in solution for low molecular weight solutes
`prior to a crystallization step, such a practice may not be accept(cid:173)
`able for temperature sensitive materials subject to thermal degrad(cid:173)
`
`ation. The operational temperature range is limited, therefore, by
`the freezing point of the solvent (solution) at low temperatures and
`by thermal degradation of the solute at higher temperatures. Low
`temperatures generally minimize solubility and favor stability but
`slow kinetics. As with citric acid anhydrous forms can only exist
`above some temperature. The particular form may greatly affect
`kinetics as well as soHd stabiHty and morphology.
`
`11.2.2. pH EFFECTS ON SOLUBILITY
`
`The acidic and basic salts of glutamic acid exhibit very different
`solubiUty behavior as can be seen in Figure 11.2b. The solubility of
`amphoteric compounds such as amino acids or antibiotics are
`strongly pH dependent because the predominant form existing in
`solution changes with the hydrogen ion concentration. Acid/base
`solution equiUbria for a compound having one acidic and one
`basic group could be represented as
`
`Merck Exhibit 2173, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`11.2. SOLUBILITY AND THE CREATION OF SUPERSATURATION 251
`
`800
`
`"1
`
`\
`
`1
`
`r
`
`NaGlutamate
`
`Giutamate.HCI
`
`20
`40
`60
`80
`100
`Temperature T (""C)
`
`120
`
`20
`
`40
`
`60
`
`80
`
`100
`
`120
`
`Temperature T
`
`(""C)
`
`Glutamic Acid
`
`0.90
`
`CO
`CO
`^ 0.60
`
`0.50
`
`Figure 11.2 Temperature dependence of the aqueous solubility of (a) citric acid; and (b) glutamic acid. (Data from
`Samejima 1972.)
`
`K^
`
`K2
`
`H2A+ = HA^ + H+ = A- + 2H+
`
`(11.1)
`
`The crystal-solution equilibrium relates to that between a particu(cid:173)
`lar form of the solute in solution and the same form in the crystal(cid:173)
`line phase. The isoelectric (zwitterionic) form usually exists in the
`crystalline phase over most of the pH range. Such compounds
`usually exhibit the lowest apparent solubility at their isoelectric
`pH since at other pH's some fraction of the solute exists in solution
`in the acid or base form. However, at very high acid or base
`concentrations, a sah, e.g., H2A+Cr, could be the crystalhne
`form.
`Figure 11.3 shows the solubilities (defined as total solute dis(cid:173)
`solved) of some /^-lactam antibiotics as a function of pH. All of
`these compounds exhibit a minimum in their solubihty, C*o, at the
`isoelectric point of the compound. At a pH significantly removed
`from
`the isoelectric point, the total (apparent) solubility is
`increased. The U-shape of the apparent solubility versus pH curve
`can be described by taking into account the acid-base equilibria of
`the antibiotic solute and assuming that the solute in the crystal is in
`the zwitterionic (isoelectric) form.
`
`C* = C * o ( l +^ + ^
`a^+
`Ki
`
`(11.2)
`
`In Eq. (11.2) flH^ is the hydrogen ion activity.
`Further complications arise when the concentrations of salts
`formed at low or high pH exceed their solubility limit. Such an
`instance is shown in Figure 11.4 for L-isoleucine. At pH values
`above 2, the zwitterion is the dominant species in soludon and in
`
`the crystalhne phase. Upon addition of acid the solubihty reaches a
`maximum at pH = 1 corresponding to the formation of the chlor(cid:173)
`ide salt. When the chloride ion is further increased whether from
`HCl or a chloride salt, precipitation of the chloride salt of leucine
`occurs (common ion effect.)
`
`11.2.3. REDUCTION OF SOLUBILITY WITH
`ANTI-SOLVENTS
`
`The use of miscible anti-solvent liquids to precipitate low molecu(cid:173)
`lar weight compounds is quite common in pharmaceutical process(cid:173)
`ing as it often can rapidly create higher supersaturations as
`compared to cooHng or evaporation. For example, aliphatic alco(cid:173)
`hols such as isopropanol can be added to aqueous solutions of
`amino acids to reduce their solubility by orders-of-magnitude. Of
`course, this results in dilution of the stream. For compounds
`soluble in organic hquids, water or an alkane (heptane) may be
`the precipitating agent. The precipitating agent usually can be
`recovered by distillation for re-use in these processes.
`In Figure 11.5 are shown the solubihties of various amino
`acids at their isoelectric point as a function of isopropanol con(cid:173)
`centration (Orella and Kirwan 1989). Simple theories suggest that
`the effect of the alcohol is to reduce the dielectric constant and
`thereby reduce solubility (Kirkwood 1936). In fact, the solubihty
`behavior is much more complex and significantly influenced by the
`nature of the amino acid side chain. For example, compare the
`behavior of phenylalanine (hydrophobic side chain ) and aspara-
`gine (hydrophilic side chain) in Figure 11.5. Both hydrophobic and
`polar interactions between the mixed solvent and the various
`groups present on the molecule can occur. Correlation of the
`solubihty behavior in these soludons using Margules, NRTL, or
`Wilson activity coefficient equations can be successful (Orella and
`
`Merck Exhibit 2173, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2 5 2 CRYSTALLIZATION IN THE PHARMACEUTICAL AND BIOPROCESSING INDUSTRIES
`
`A^.5*- A- + H ^ ^^ A"+ 2H^
`
`product" (Chapter 1). In ideal solutions, where activities can be
`represented by concentrations, the solubility product is
`
`H
`
`R""—c
`
`• >:
`
`^
`o o
`
`a OH
`
`XOOH
`
`= Amoxicillin
`
`Ho
`
`e-flf-
`
`1.0
`
`0.6
`0.4
`
`0.2
`
`^
`
`0.1
`
`0.06
`S
`"O 0.04
`CO
`
`0.02
`
`0.01
`
`0.006
`
`pH
`
`Figure 11.3 Aqueous solubiUty of some antibiotics as a function of
`pH. (Data from Tsujima et al. 1978.)
`
`Kirwan 1991); however predictive models await a more fundamen(cid:173)
`tal understanding of the solution thermodynamics of complex
`organic/charged species in mixed organic-aqueous solvents.
`
`11.2.4. EFFECTS OF SALTS ON SOLUBILITY
`
`Since many biological molecules have groups capable of ioniza(cid:173)
`tion, they may exist as singly- or multiply-charged species or as
`neutral molecules with strong dipoles. As such, they are particu(cid:173)
`larly susceptible to other ions in solution through effects on ionic
`strength as well as to specific interactions with counter-ions. As
`one example, the solubility of glutamic acid in the presence of
`different anions is shown in Figure 11.6. These data show that
`the pH and the counter-ion both can have a significant impact on
`the solubility of the glutamate salt. In addition, the presence of an
`inorganic sodium salt such as sodium chloride will also impact the
`solubiUty. This type of behavior is most simply characterized by
`the recognition of ionic equilibria involving the solute and com(cid:173)
`mon ions. This leads to an equation for the equiUbrium constant,
`known for a product with limited solubility as the "solubiUty
`
`^sp = [Na+][Glu-]
`
`(11.3)
`
`Thus, adding sodium chloride to a solution of sodium glutamate
`drives sodium glutamate out of solution, provided the added salt
`remains in solution. However, as noted in Chapter 1 more complex
`behavior may be observed, where adding a common ion may either
`increase or decrease the solubility.
`The solubiUty of low molecular weight biological solutes
`depends on a variety of thermodynamic variables, which can be
`manipulated for an effective precipitation or crystallization pro(cid:173)
`cess. Hence, solubiUty relationships should be understood
`for effective design and optimization of a recovery process. For
`example, a typical crystaUization step might involve adjustment of
`pH to near the isoelectric point, the lowering of temperature to
`reduce solubility and minimize degradation, and the use of anti-sol(cid:173)
`vents, specific ions, or evaporation to effect precipitation or crys(cid:173)
`tallization. It should be apparent from the above discussion that
`many unanswered questions remain on the relationship of the
`solubility of a biological molecule to the relevant thermodynamic
`variables necessitating an experimental investigation during process
`research and development.
`Given the limited data base from which solubility correlations
`can be drawn, it is essential to measure the solubility directly for
`the system of interest during process development. Since process
`conditions often favor operation with high concentrations of
`solute, such systems are often thermodynamically nonideal. It is
`necessary to measure the solubiUty in the solvent system(s) of
`interest in order to optimize the yield and the purity. To accom(cid:173)
`plish the latter relies upon the abiUty to measure the solubiUty of
`the key impurities as weU as the product of interest. This requires
`the availabiUty of both the key impurities and product; however,
`the impurities often are not available as isolated soUds. In that
`case, the solubiUty of impurities must be deduced from the purity
`profile of mother Uquors taken from crystallizations of the actual
`process stream. It is often simplest and always fastest to measure
`the solubiUty and carry out crystalUzations in a single-solvent system.
`However, working in multiple-solvent systems increases the likeli(cid:173)
`hood of improving the yield, the separation factor, and the prospects
`of observing more of the possible crystal forms that may exist.
`Once the solubility is sufficiently defined, then the different
`operating modes should be considered. As mentioned earlier, there
`are three customary modes of carrying out a batch crystallization
`that uses a precipitating agent. The first mode is a "normal"
`addition of the crystallizing agent to the solution. The second is a
`"reverse" addition in which the solution is added to the precipit(cid:173)
`ating agent. The third is a semi-continuous addition to blend a
`constant ratio of the solute solution and the crystallizing agent into
`a separate tank. These three modes define very different environ(cid:173)
`ments at which the crystallization occurs. In the normal mode, the
`concentration of solute is high throughout much of the crystal(cid:173)
`lization, and the ratio of agent to solution is steadily increased to
`the final ratio. In the reverse mode, the concentration of solute is
`much lower, and the ratio of agent to solution is steadily decreased
`to the final ratio. FinaUy, the third mode is represented by a single
`ratio of agent to solution, and the concentration of solute drops
`from the concentration achieved by simply blending along a ver(cid:173)
`tical Une to the solubiUty at the constant ratio of agent to solution.
`This last mode is often referred to as "semi-continuous" crystal-
`Uzation. The different solvent environments can have an influence
`on the crystal morphology (Section 11.3.2), and the difference in
`solute concentration can be important if competing processes con(cid:173)
`sume the solute (e.g., degradation).
`
`Merck Exhibit 2173, Page 6
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`11.3. CONTROL OF PARTICLE SIZE AND MORPHOLOGY 253
`
`—I
`
`1
`
`1
`
`T=25,0'*C
`
`I A=Total He in Solution"
`
`0
`
`5
`
`10
`
`15
`
`20
`
`30
`
`25
`gHCi
`HCI Concentration (
`100g Solution
`
`Figure 11.4 The Solubility of L-isoleucine as function of pH or HCI concentration. (Reprinted
`with permission from Zumstein, R.C., and Rousseau, R.W., Ind. Eng. Chem. Res., 28, 1989,
`pp. 1226-1231. © 1989, American Chemical Society.)
`
`=5 35
`
`JD
`o
`CO
`CD
`>
`
`0)
`DC
`
`.001
`
`.0001
`0.0
`
`0.8
`0.6
`0.4
`0.2
`Mole Fraction Alcohol in Solution
`
`1.0
`
`Figure 11.5 The solubility of various amino acids in aqueous
`L-propanol
`solutions.
`(Reprinted with permission of
`the
`American Institute of Chemical Engineers from Orella, C.J., and
`Kirwan, D.J., Biotech. Progress 5, pp. 89-93 (1989). © 1989. All
`rights reserved.)
`
`Figure 11.6 SolubiUties of various metallic salts of L-glutamic
`acid. (Reprinted with permission of the publisher from Yamada, K.
`et al., Microbial Production of Amino Acids, J. Wiley, New York,
`1972, p. 234.)
`
`11.3. CONTROL OF PARTICLE SIZE AND MORPHOLOGY
`
`Pharmaceutical products are usually subject to quite restrictive
`particle size and shape specifications. For final products, small
`(micron-size) particles with a narrow size distribution are desirable
`because of their short and uniform dissolution time and conse(cid:173)
`quent good bio-availabiUty. Similarly, particle shape influences
`
`Merck Exhibit 2173, Page 7
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`2 5 4 CRYSTALLIZATION IN THE PHARMACEUTICAL AND BIOPROCESSING INDUSTRIES
`
`bio-availability as well as breakage and flow characteristics of the
`powder. Usually the simplest and preferred means of characteriz(cid:173)
`ing the particle size distribution is with a particle size analyzer (See
`Chapter 4) supplemented by microscopic examination. However,
`particle size is sometimes simply characterized by an adsorption
`area per gram measurement without a direct measurement of
`particle size and habit. Image analysis can be used to more fully
`characterize a distribution of particles. It has been common in the
`pharmaceutical industry to mill powders to a final desired size.
`This has a number of disadvantages: excessive local temperatures
`may degrade bio-active material and effective containment of fine
`powders of highly potent drugs can be costly to implement. For
`these reasons crystallization processes that can directly produce
`fine crystals of controlled small sizes are an area of active research
`and development. (See Section 11.5.2.)
`As noted earlier, when crystallization is employed as a crude
`fractionation step earlier in the purification train, larger particles
`are desired to facilitate centrifugation or filtration of the slurries.
`Control of particle size during these crystallizadons, therefore, is of
`great importance due to its influence on subsequent handling steps.
`The basic principles of particle size control for the crystal(cid:173)
`lization of pharmaceuticals are generally the same as those dis(cid:173)
`cussed in earlier chapters, that is, control of nucleation and growth
`rates with attention paid to agglomeration and breakage effects.
`The populadon balance approach permits incorporadon of all of
`these factors. Our discussion here, therefore, focuses on these
`phenomena for biological solutes. Some complicating factors in
`crystallizing pharmaceutical molecules are: (1) the complexity of
`the molecules involved; (2) the need for very fine particles (high
`supersaturation conditions); and (3) the common use of batch
`crystallizers. The use of batch processes inherently results in crys(cid:173)
`tallization under time-varying conditions and often under high
`(initial) supersaturations. The vessels may have poor mixing as
`well which causes spatial variations in supersaturation resulting
`in broader size distributions and variable quahty.
`
`11.3.1. CRYSTAL GROWTH KINETICS
`
`Single-crystal and muld-particle intrinsic growth rates for bio(cid:173)
`chemical compounds have only been measured in a relatively few
`instances. In general, similar techniques are used as are employed
`for inorganic compounds: observation of the change of size of
`single crystals growing under a microscope, measurement of the
`change in mass of single crystals over time; measurement of the
`PSD in an MSMPR crystallizer, or measurement of the de-super-
`saturation rate or PSD time evolution during batch crystallization.
`Of course, the influence of solution mass transfer on the rate must
`be experimentally eliminated or taken into account in interpreting
`the results.
`Growth kinetics from solution often are represented by a
`power law model in supersaturation.
`
`G = ki[C-C*Y
`
`(11.4)
`
`For intrinsic growth kinetics C is the interfacial solution concen(cid:173)
`tration of solute. The rate law may be expressed in terms of relative
`supersaturadon, a.
`
`G = k2oP
`
`(11.5)
`
`The reladve supersaturadon, a = {C - C*)/C*, is propordonal to
`the difference of the solute chemical potential between solution
`and crystal, RT\n{C/C*), for small values of (C - €*)/€* under
`the assumption of an ideal solution. For large supersaturation
`
`ranges, as commonly occur in precipitation, the chemical potential
`difference is more appropriately used as the driving force.
`
`G = k3[RT\n(CIC*)Y
`
`(11.6)
`
`For nonideal solutions, the concentration ratio in Eq. 11.6 is
`replaced with the ratio of thermodynamic activities. Since, for
`biological compounds, the equiUbrium solubility, C*, is affected
`by a number of thermodynamic factors (e.g., anti-solvent concen(cid:173)
`tration, ionic strength, pH) studies of their effect on growth rate
`can be used to test the relative usefulness of Eq. 11.4 versus Eq.
`11.5 or 11.6.
`Orella (1990) made single crystal intrinsic growth rate meas(cid:173)
`urements for a number of amino acids at their isoelectric points
`growing in aqueous and aqueous-propanol mixtures. Figure 11.7
`presents some of his results showing power law model kinetics for
`L-asparagine growth from water and from 1- and 2-propanol
`aqueous mixtures. The high apparent kinetic order of nearly 3
`for these systems suggests that the mechanism may involve two-
`dimensional nucleation (see Chapter 2). Comparison of Figures
`11.7a-b illustrates the advantage of using the relative supersatura(cid:173)
`tion as the driving force in the rate expression. The presence of
`alcohol reduces C* by well over an order-of-magnitude in these
`cases. The use of relative supersaturation appears to provide a
`means of properly distinguishing the effect of variables on the
`kinetic coefficient and on the driving force. For this case, the
`alcohols appear to have little effect on the kinetic coefficient
`(Figure 11.7a). If the data were treated as in Figure 11.7b, one
`might (incorrectly) conclude that growth kinetics in alcohol mix(cid:173)
`tures are more rapid than in water. This formuladon using DDr
`the chemical potential expression at large supersaturations) should
`be useful in growth rate modeling of biological compounds where
`many thermodynamic variables may affect the equilibrium solubi(cid:173)
`lity and growth rate.
`There have been a number studies of the growth kinetics
`of biochemicals in suspension crystallization. Rodriguez-Horneda
`et al. (1986) used an MSMPR crystallizer to obtain the growth rate
`kinetics for the drug, phenytoin, as a funcdon of pH. They found
`the kinetics to be size-independent and to increase at lower pH. An
`MSMPR crystallizer study by Harano and Yamamato (1982)
`determined the growth rate of glutamic acid from the measured
`size distribution.
`A series of studies by Orella (1990), Mahajan et al. (1991),
`Mahajan and Kirwan (1994), and Deshpande (1998) compared the
`growth kinetics of asparagine monohydrate in mixed propanol-
`water solutions as measured for single crystals, in batch and
`MSMPR crystallizers, and with the use of a rapid (grid) mixer at
`very high supersaturations. Figure 11.8 demonstrates good agree(cid:173)
`ment among these measurements considering that they were made
`by different investigators; that the suspension crystallizer environ(cid:173)
`ment was quite different from that for single crystal growth, i.e.,
`the presence of collisions; and that the crystal sizes were different
`in each of these systems. Single crystals were 1.5-3 mm, crystals in
`the MSMPR crystallizer were 20-300 |im, crystals in the batch
`vessel were 20—80|am and those from the rapid mixing device
`were 3-20 ^im. The results suggest a shght size dependence of the
`growth kinetics.
`A primary conclusion from the above results is that the stand(cid:173)
`ard methods for determining intrinsic crystal growth rates are well-
`suited for biochemical compounds. At present there are limited
`data on which to generalize with respect to growth mechanisms
`and the effects of various thermodynamic variables. Indeed, the
`large number of variables that may affect solubility and kinetics in
`these systems suggests that extensive and careful laboratory studies
`
`Merck Exhibit 2173, Page 8
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`11.3. CONTROL OF PARTICLE SIZE AND MORPHOLOGY 2 55
`
`2x10"
`
`10"'
`
`1 O Solvent: Water
`n Solvent: 50 w%1-PrOH
`A Solvent: 30 w% a-PrOH-
`
`1
`
`H
`
`O Solvent: Water
`D Solvent: 50 w%1-PrOH
`A Solvent: 30 w% 2-PrOH
`
`s/r
`§5
`
`k
`
`10 1-3
`
`e
`,
`
`8
`
`10 -4L_
`
`o /
`
`H
`
`a
`
`E o^
`O
`0)
`to
`DC
`x:
`%
`2
`O
`k.
`CO
`CD
`c
`
`1
`1
`1
`1
`1
`10-^
`800
`400
`200
`100
`50
`Chemical Potential Driving Force
`RTInCj/C*, (J/mole)
`
`.01
`
`6.0
`1.0
`0.1
`Concentration Supersaturation
`Cj-C*, (g/kg solvent)
`
`Figure 11.7 Single-crystal growth kinetics of L-asparagine in aqueous and aqueous-propanol solvents at 25 °C (Data
`from Orella, 1990.)
`
`should be conducted to understand and predict larger scale per(cid:173)
`formance.
`
`11.3.2. EFFECTS OF ADDITIVES, SOLVENTS, AND
`IMPURITIES
`
`Chapter 3 of this text describes in detail the mechanisms and
`effects of various components of a solution on the resulting crystal
`size, morphology (habit), growth rate, and purity. Here we simply
`
`100
`
`c 1
`E
`i
`I
`
`10
`
`1
`
`0.1
`
`•
`o
`•
`V
`
`Single Crystal (Orella, 1990)
`Batch (Mahajanetal., 1991)
`Grid Mixer (Mahajan and Kirwan, 1994)
`MSMPR (Deshpande, 1998)
`
`•
`
`T
`
`Vv
`
`•
`
`•o
`o
`
`•
`
`0.01
`
`0.01
`
`•
`
`0.1
`
`ln(Cactua|/C*)
`
`Figure 11.8 Comparison of growth kinetics of L-asparagine as
`single crystals, in batch and MSMPR crystallizers and from a
`rapid mixing device. (Adapted from the data of Orella (1990),
`Mahajan et al. (1991), Mahajan and Kirwan (1994) and
`Deshpande (1998).)
`
`relate some observations made on biochemical crystals. We con(cid:173)
`sider the effects both of low concentrations of additives or impur(cid:173)
`ities, particularly similar compounds commonly found in the
`process stream of a particular product, and of solvent (anti-sol(cid:173)
`vent) present at relatively high concentrations.
`
`Additives, Black et al. (1986) reported that small amounts of
`glutamic acid (mole fraction <10~^) reduced the growth rate of
`asparagine from aqueous solution by a factor of 10. Harano and
`Yamamoto (1982) observed that small amounts of four other
`amino acids reduced the growth rate of glutamic acid in an
`MSMPR crystallizer as well as having marked affects on crystal
`morphology. Differential kinetics (adsorption) on different crystal-
`lographic faces may explain the effect of additives or impurities on
`crystal habit or morphology. Since biochemical molecules contain
`a variety of chemical groups (Figure 11.1); it is clear that different
`chemical groups will exist on different crystal faces and will pro(cid:173)
`vide, therefore, a variety of possible interactions with solution
`components.
`There has been a series of systematic studies of the effect
`of impurities and tailor-made additives on the shape of crys(cid:173)
`tals grown from solution (Addadi et al. 1982; Berkovitch-YeUin
`1985; Weissbuch et al. 1999.) They used amino acids and amino
`acid derivatives as model solutes. By combining habit observations
`with crystal structure and molecular orientation, they were
`able to rationalize a number of effects, including resolution of
`steroisomers. Hopefully, further studies in this area will provide
`a rational basis for the many impurity and additive effects
`found during crystallization of biochemicals.
`
`Solvent Effects. Changes of solvent, as in the use of alcohols
`as anti-solvents for precipitation from aqueous solution, may
`also influence particle shape through altering growth kinetics on
`
`Merck Exhibit 2173, Page 9
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`

`

`256 CRYSTALLIZATION IN THE PHARMACEUTICAL AND BIOPROCESSING INDUSTRIES
`
`different crystal faces. Solvent effects on a number of organic
`compounds, such as resorcinol, hexamethylene diamine and succi(cid:173)
`nic acid, grown from solution were interpreted in terms of either
`a surface roughness effect or specific solvent adsorption on parti(cid:173)
`cular faces (Davey 1986; Bourne and Davey 1976; Davey et al.
`1982).
`Orella (1990) and Mahajan et al. (1991) did not observe any
`effect of solvent composition on the growth rate and habit of
`asparagine crystals grown from aqueous propanol solutions of
`varying composition. However, Orella (1990) found significant
`effects on alanine crystal growth and habit (see Figure 11.9.) As
`the propanol concentration increased from 0 to 35% by weight,
`the aspect ratio of the crystals changed from approximately 1 to
`well over 50! In fact, the growth rates in 35% propanol as com(cid:173)
`pared to water were reduced in both directions but b