Re­cent Ad­vances in En­gi­neer­ing As­pects of Phar­ma­ceu­ti­cal Crys­tal­liza­tion - Part 1

Chemical Industry Digest - - What’s In? - K.He­malatha, Dr S.Vedan­tam, Dr K.Ya­muna Rani - IICT Hy­der­abad.

This ar­ti­cle sum­marises re­cent de­vel­op­ments from an en­gi­neer­ing per­spec­tive in the key as­pects of the crys­tal­liza­tion process.


Crys­tal­liza­tion, the most im­por­tant sep­a­ra­tion and pu­rifi­ca­tion process in the phar­ma­ceu­ti­cal in­dus­try has con­trib­uted to sig­nif­i­cant im­prove­ments in build­ing ef­fi­cient man­u­fac­tur­ing prac­tices for pro­duc­tion of APIs. Past few decades have seen con­tin­u­ous rise in re­search and de­vel­op­ment ac­tiv­i­ties in both com­ing up with novel ap­proaches for deeper un­der­stand­ing of the process as well com­bin­ing ex­per­i­men­tal and mod­el­ling meth­ods for more ro­bust and uni­fied ap­proaches in the mon­i­tor­ing, con­trol as well as de­sign and scale-up of in­dus­trial crys­tal­liza­tion pro­cesses. This ar­ti­cle sum­ma­rizes re­cent de­vel­op­ments from an en­gi­neer­ing per­spec­tive in the key as­pects of crys­tal­liza­tion process.

The ar­ti­cle is in two parts, the first part, pub­lished here, cov­ers poly­mor­phism , co-crys­talli­sa­tion, ki­net­ics, nu­cle­ation, etc. The sec­ond part of the ar­ti­cle will be pub­lished in the forth­com­ing is­sue.

1. In­tro­duc­tion

The re­quire­ments of ther­a­peu­tic mar­ket place, such as con­sis­tent high qual­ity and bioavail­abil­ity of Ac­tive Phar­ma­ceu­ti­cal In­gre­di­ents (APIs) de­mand ef­fi­ciency of crys­tal­liza­tion process in or­der to meet de­mand­sup­ply bal­ance, batch-to-batch con­sis­tency and ro­bust­ness in prop­er­ties in all phases of drug de­vel­op­ment. As 70% of all solid prod­ucts and 90 % of APIs in­volve a crys­tal­liza­tion step (Gao et al., 2017), there are many tech­nol­ogy and eco­nomic driv­ers for con­trol and mon­i­tor­ing of crys­tal­liza­tion pro­cesses. Al­most 70 per­cent of new APIs be­ing pur­sued are poorly sol­u­ble in wa­ter (Kawa­bata et al., 2011). The dis­so­lu­tion and dis­in­te­gra­tion rates and con­se­quently the bioavail­abil­ity in the body, par­tic­u­larly for low-sol­u­bil­ity APIs, de­pend strongly on phys­i­cal prop­er­ties such as the size and shape of the crys­tals. The crys­tal size and shape dis­tri­bu­tion de­ter­mines the in­ter­fa­cial sur­face area and there­fore af­fects the dis­so­lu­tion rate. Dif­fer­ent crys­tal struc­tures of the same mol­e­cule, in other words ‘poly­morphs’ can ex­hibit dif­fer­ent sol­u­bil­i­ties. While one poly­morph dis­solves in the di­ges­tive sys­tem, oth­ers might not, which would ham­per the ther­a­peu­tic ef­fect of a drug. All of these phys­io­chem­i­cal prop­er­ties, i.e., crys­tal sizes, shapes, and solid forms, can and must be con­trolled via the fi­nal API crys­tal­liza­tion step (Var­i­ankaval et al., 2008). Fur­ther­more, the crys­tal size dis­tri­bu­tion and shape af­fect the pow­der’s flow abil­ity, seg­re­ga­tion phe­nom­ena and down­stream oper­a­tions like fil­tra­tion, blend­ing, gran­u­la­tion, cap­sule fill­ing, tablet­ting, and there­fore the en­tire drug de­vel­op­ment / man­u­fac­tur­ing process. In this con­text, the field of “crys­tal en­gi­neer­ing” is in­creas­ingly re­ceiv­ing at­ten­tion and the crys­tal­liza­tion process plays a key role in defin­ing the phys­io­chem­i­cal prop­er­ties of solid APIs and their fi­nal dosage forms. Added to these, strict reg­u­la­tory re­quire­ments re­lated to the vari­a­tion of qual­ity lead to high eco­nomic penalty of pro­duc­ing off-spec­i­fi­ca­tion prod­uct (USD 1-2 mil­lion / batch). These call for con­trol of crys­tal prop­er­ties fol­lowed by ef­fi­cient down­stream oper­a­tions (fil­tra­tion, dry­ing) and prod­uct ef­fec­tive­ness (tablet sta­bil­ity, bio-avail­abil­ity); qual­ity by de­sign, fast scale-up and prod­uct con­sis­tency. In this per­spec­tive, stand-alone as well as uni­fied ap­proaches pre­sented be­neath may be adopted to meet the con­straints and re­quire­ments of the crys­tal­liza­tion of APIs.

2.Ther­mo­dy­namic As­pects in Crys­tal­liza­tion 2.1 Sol­u­bil­ity

In so­lu­tion crys­tal­liza­tion, the equi­lib­rium be­tween solid solute and dis­solved solute in spe­cific sol­vent, termed as sol­u­bil­ity is di­rectly im­pacted by the tem­per­a­ture and en­tropy of fu­sion, equiv­a­lent to the dif­fer­ence of chem­i­cal po­ten­tials be­tween the solid and liq­uid of a par­tic­u­lar com­pound. In ad­di­tion, the sol­vent com­po­si­tion of mixed sol­vents, as well as other mi­nor com­po­nents or im­pu­ri­ties can af­fect the ac­tiv­ity co­ef­fi­cient.

In the sol­u­bil­ity curve shown in Fig­ure 1, if the curve is steep, i.e. the sub­stance ex­hibits a strong tem­per­a­ture de­pen­dence of sol­u­bil­ity (e.g. many salts and or­ganic sub­stances), then a cool­ing crys­tal­liza­tion might be suit­able. But if the metastable zone is wide (e.g. su­crose so­lu­tions), ad­di­tion of seed crys­tal might be nec­es­sary. This can be de­sir­able, par­tic­u­larly if a uni­formly sized prod­uct is re­quired. If on the other hand, the equi­lib­rium line is rel­a­tively flat (e.g. for aque­ous com­mon salt so­lu­tions), then an evap­o­ra­tive process might be­come nec­es­sary. If the yield from ei­ther of the pro­cesses is low, then a sec­ond sol­vent can be added to re­duce the ef­fec­tive­ness of the first and thus de­crease the resid­ual so­lu­tion con­cen­tra­tion (some­times called `drown­ing-out’ or `wa­ter­ing-out’). If the solute oc­curs as a con­se­quence of chem­i­cal re­ac­tion or ad­di­tion of a com­mon ion, and is rel­a­tively in­sol­u­ble, then pre­cip­i­ta­tion or `fast crys­tal­liza­tion’ oc­curs. Typ­i­cally, the driv­ing force for crys­tal­liza­tion, su­per­sat­u­ra­tion is gen­er­ated by anti-sol­vent ad­di­tion or cool­ing. In batch crys­tal­liza­tion, a crys­talline prod­uct with uni­form size

and shape is de­sir­able, as the crys­tal size dis­tri­bu­tion (CSD) de­ter­mines the ef­fi­ciency of var­i­ous down­stream pro­cesses. Con­trol of su­per­sat­u­ra­tion, by ad­just­ing the cool­ing or ad­di­tion of anti-sol­vent will de­ter­mine the fi­nal crys­tal prop­er­ties.

Sol­u­bil­ity mea­sure­ment and its pre­dic­tion have re­ceived a lot of at­ten­tion in re­cent years. Pre­dic­tive and cor­rel­a­tive mod­els, for ex­am­ple the NRTL-SAC model (Chen and Song., 2004) which is con­sid­ered to be the bench­mark, have been devel­oped and ap­plied to solve prac­ti­cal in­dus­trial ap­pli­ca­tions. For the de­sign of a crys­tal­liza­tion process, ini­tially the con­di­tions are cho­sen such that all of the API or in­ter­me­di­ate solid is dis­solved. At the end, the con­di­tions are cho­sen so that the ma­jor­ity of the batch is crys­tal­lized. Though it is de­sir­able that im­pu­ri­ties such as re­ac­tion byprod­ucts or other un­de­sired species re­main sol­u­ble in the mother liquor at this point, much less ef­fort is spent in the in­dus­try to quan­tify the sol­u­bil­ity of im­pu­ri­ties in sol­vents; one key rea­son be­ing the amount of im­pu­ri­ties iso­lated be­ing in­suf­fi­cient for the sol­u­bil­ity mea­sure­ment. How­ever, the diver­sity of im­pu­ri­ties within one com­pound can­not be ne­glected. In­stead, sim­ple rules based upon other in­for­ma­tion are ap­plied in guid­ing the se­lec­tion of sol­vents, for ex­am­ple the re­ten­tion times of im­pu­ri­ties from the liq­uid chro­matogram or the chem­i­cal struc­tures of im­pu­ri­ties if avail­able. It should be noted that along with the sol­u­bil­ity dif­fer­ence be­tween the de­sired com­pound and im­pu­ri­ties, other fac­tors such as solid so­lu­tion, ab­sorp­tion of im­pu­rity on the crys­tal sur­face also af­fect the re­jec­tion ef­fi­ciency as well (Tung et al., 2009).

2.2 Poly­mor­phism, Crys­tal Habit

The im­pact of poly­mor­phism (same molec­u­lar species, with dif­fer­ent crys­tal struc­tures) goes be­yond the level of crys­tal­liza­tion as it af­fects the down­stream for­mu­la­tion and drug bioavail­abil­ity. The search for pos­si­ble poly­morphs, the­o­ret­i­cally / em­pir­i­cally, or in a com­bi­na­tion of both, is a sub­ject un­der­go­ing ex­ten­sive re­search and de­vel­op­ment ef­forts. Un­til re­cently, this search has been mostly em­pir­i­cal. No par­tic­u­lar em­pir­i­cal meth­ods, such as cool­ing, an­ti­sol­vent, evap­o­ra­tion, or salt for­ma­tion, have been demon­strated to be su­pe­rior to other em­pir­i­cal meth­ods.

The fo­cus of the search for poly­morphs could vary at dif­fer­ent phases of drug de­vel­op­ment (Tung, 2013). Fur­ther, it is not un­heard of that a new crys­tal form of a drug can­di­date ap­pears in the late phase of drug de­vel­op­ment or af­ter the drug is ap­proved and dis­trib­uted in the mar­ket (Chem­burkar, 2000). Af­ter the is­sue with Ri­ton­avir in 1998 (18 months af­ter the new com­mer­cial prod­uct Ri­ton­avir was launched, a new sta­ble poly­morph (form II) was iden­ti­fied in sup­plies of the drug, which greatly re­duced Ri­ton­avir’s sol­u­bil­ity com­pared with the orig­i­nal crys­tal form, lead­ing to an oral bioavail­abil­ity prob­lem) served as a warn­ing to phar­ma­cists and crys­tal en­gi­neers, poly­mor­phism be­came in­creas­ingly im­por­tant in both fun­da­men­tal re­search and in­tel­lec­tual prop­erty rights. In ad­di­tion to its ef­fect on drug safety, poly­mor­phism is an im­por­tant fac­tor in the test­ing of generic drugs, a huge ex­pan­sion of which has oc­curred fol­low­ing the ex­pi­ra­tion of many patents of orig­i­nal drugs (Gao et al., 2017).

Poly­mor­phic crys­tals are known to have dif­fer­ent crys­tal struc­tures. They have dif­fer­ent X-ray dif­frac­tion pat­terns and Ra­man spectra. In ad­di­tion, they should have dif­fer­ent phys­i­cal at­tributes, for ex­am­ple mor­phol­ogy, solid den­sity, heat ca­pac­ity, heat of melt­ing, melt­ing point or de­com­po­si­tion tem­per­a­ture, sol­u­bil­ity, etc. The cor­re­la­tion be­tween crys­tal form and mor­phol­ogy us­ing the mi­cro­scope could of­fer po­ten­tial ad­van­tages over other es­tab­lished tech­niques, like Ra­man or pow­der X-ray dif­frac­tion meth­ods, such as sim­plic­ity of mea­sure­ment, a sig­nif­i­cant re­duc­tion of equip­ment and main­te­nance cost, and a bet­ter sen­si­tiv­ity in de­tect­ing crys­tals of un­de­sired forms within the ma­trix of crys­tals of de­sired form.

Crys­tal habit (ap­pear­ance of crys­tals), though does not re­flect the in­ter­nal struc­ture of crys­tals, has po­ten­tial im­pact on down­stream fil­tra­tion, dry­ing and for­mu­la­tion. From the shape of a crys­tal, it is pos­si­ble to in­fer the sur­face area and rel­a­tive growth rates of dif­fer­ent crys­tal sur­faces. For nee­dle-like crys­tals, the sur­face for crys­tal growth is pri­mar­ily on the two tips. The sur­face on the nee­dle has a much slower growth rate. For plate-like crys­tals, the sur­face for crys­tal growth is at the edges. The sur­face on the plate has a much slower growth rate. For rod-like or cube-like crys­tals, all crys­tal sur­faces grow at com­pa­ra­ble rates. In­fer­ring the rel­a­tive growth rates on dif­fer­ent crys­tal sur­faces has prac­ti­cal im­pact on the crys­tal­liza­tion process de­vel­op­ment. In gen­eral, for nee­dle-like crys­tals, more seed is needed since it has less sur­face area for crys­tal growth, de­spite the over­all sur­face area per unit mass of crys­tals be­ing high. On the other hand, for rod-like crys­tals, less seed is needed even though the over­all sur­face area per unit mass of crys­tals is low. Var­i­ous fac­tors im­pact­ing the mor­phol­ogy can be crys­tal struc­ture, sol­vents, ad­di­tives, im­pu­ri­ties, and su­per­sat­u­ra­tion or desu­per­sat­u­ra­tion rates dur­ing crys­tal growth and dis­so­lu­tion pe­ri­ods, etc. Ad­di­tion­ally, it is shown that mul­ti­ple heat/cool cy­cles cou­pling with wet mill-

ing at each cool cy­cle or with­out wet milling (Lovett et al., 2011) can be em­ployed to mod­ify the crys­tal mor­phol­ogy as well. This ap­proach has been prac­ticed over the years in the in­dus­try in mod­i­fy­ing the crys­tal mor­phol­ogy, with­out mod­i­fy­ing the sol­vents or crys­tal forms or charg­ing ad­di­tives dur­ing the crys­tal­liza­tion due to chem­i­cal pu­rifi­ca­tion or bioavail­abil­ity re­quire­ments. At present, in­dus­try has built a sig­nif­i­cant amount of know-how of this ap­proach (Tung, 2013).

Crys­tal shape and poly­morph in­flu­ence sol­u­bil­ity, dis­so­lu­tion rate (which in­flu­ence bioavail­abil­ity), com­press­ibil­ity (cru­cial for tablet­ting), and sta­bil­ity. The crys­tal enan­tiomorph is im­por­tant in the man­u­fac­ture of chi­ral ma­te­ri­als, which has be­come a $100 bil­lion in­dus­try in re­cent years. Ta­ble 1 in­di­cates the ef­fect on drug prod­uct for each spe­cific solid state prop­erty.

More­over, crys­tal size dis­tri­bu­tion, and shape have a ma­jor im­pact on the de­sign of the man­u­fac­tur­ing process since small crys­tals are dif­fi­cult to sep­a­rate from so­lu­tion, and nee­dle-like crys­tals or plate-like crys­tals can be dif­fi­cult to fil­ter and dry. These ef­fects have been rec­og­nized as the ma­jor batch-to-batch vari­a­tion is­sues lead­ing to in­con­sis­tency of the fi­nal tablet prop­er­ties (Shekunov and York, 2000). Hence, it is nec­es­sary to con­trol the qual­ity of the crys­tals dur­ing crys­tal­liza­tion process to crit­i­cally sat­isfy the per­for­mance cri­te­ria of the drug prod­uct. These solid state prop­er­ties are thus termed as crit­i­cal qual­ity at­tributes (i.e. pu­rity, crys­tal form, par­ti­cle size, spe­cific sur­face area, crys­tal mor­phol­ogy and habit etc.) that di­rectly or in­di­rectly de­ter­mine the qual­ity of the API or drug prod­uct.

2.3 Co-crys­tal­liza­tion

Co-crys­tal­liza­tion is a method to im­prove drug qual­ity. Phar­ma­ceu­ti­cal co-crys­tals are mul­ti­com­po­nent molec­u­lar sys­tems that are typ­i­cally formed through the hy­dro­gen bond­ing of a co-former mol­e­cule with the API (Pow­ell et al., 2015). The re­ac­tants are solids at am­bi­ent con­di­tions. Ex­cip­i­ents, amino acids, biomolecules, vi­ta­mins, min­er­als, and other APIs can be cho­sen as co-crys­tal for­m­ers (CCF). Salts and co-crys­tals are mul­ti­com­po­nent crys­tals and a con­tin­uum ex­ists link­ing cocrys­tals and salts based on the ex­tent of pro­ton trans­fer be­tween the com­po­nents. In re­cent

years, in or­der to pro­duce poorly sol­u­ble com­pounds, solid forms such as co-crys­tals and metastable poly­morphs are be­ing devel­oped and pro­cesses for their pro­duc­tion are be­ing devel­oped and scaled up as well (Chen et al., 2011). Af­ter the first com­mer­cial prod­uct’ En­tresto (No­var­tis)’, in 2015, there has been a surge of in­ter­est in the de­vel­op­ment of phar­ma­ceu­ti­cal cocrys­tals. Also, in­tel­lec­tual prop­erty (IP) op­por­tu­ni­ties have led to se­ri­ous ef­forts by both in­no­va­tor and generic com­pa­nies cre­ated to ex­plore, de­velop and patent unique crys­tal forms.

An ef­fec­tive way to an­tic­i­pate the ex­is­tence of mul­ti­ple forms of crys­tals, in­clud­ing sol­vates and hy­drates, is to pre-in­vest in ex­per­i­men­tal crys­tal form screen­ing, lab­o­ra­tory au­to­ma­tion and an­a­lyt­i­cal meth­ods. A sum­mary of var­i­ous re­cent tech­niques that used dif­fer­ent sys­tem­atic ap­proaches for screen­ing new poly­morphs has been re­ported (Gao et al., 2017). These in­clude de­vel­op­ment of or­ga­nized sol­vent data­base with prop­erty de­scrip­tors for poly­morph screen­ing and high-through­put crys­tal­liza­tion plat­forms such as Crys­talMax (Trans­Form Phar­ma­ceu­ti­cals, Inc.) and Crys­tal16™ (Avan­tium Tech­nolo­gies, Inc.) to screen the poly­morphs of a given API with high ef­fi­ciency. The for­ma­tion and screen­ing, trans­for­ma­tion of a sol­vate car­ries high costs and more time which re­mains a chal­lenge in sol­vate drug de­vel­op­ment. How­ever, de­vel­op­ment of new forms poses chal­lenges re­gard­ing the large-scale syn­the­sis and sta­bil­ity of these drugs in the pres­ence of ex­cip­i­ents which need im­me­di­ate at­ten­tion.

2.4 Sol­vent Se­lec­tion and De­sign

Sol­vent plays an im­por­tant role in the crys­tal­liza­tion process and its de­sign or se­lec­tion could some­times de­ter­mine the suc­cess or fail­ure of the oper­a­tion, es­pe­cially to meet the de­mands of pro­duc­ing phar­ma­ceu­ti­cal prod­ucts of high pu­rity, con­sis­tent qual­ity and high yield. The choice of the sol­vent is dic­tated by many pa­ram­e­ters such as sol­va­tion power of the sol­vent, the slope of the sol­u­bil­ity curve ver­sus tem­per­a­ture, boil­ing point, safety and tox­i­c­ity, cost, abil­ity to par­tic­i­pate in form­ing hy­dro­gen bond­ing as an ac­cep­tor or a donor, and vis­cos­ity. More­over, the use of dif­fer­ent sol­vents and pro­cess­ing con­di­tions dur­ing crys­tal­liza­tion also al­ters the crys­tal habit and mor­phol­ogy of the pu­ri­fied drug af­fect­ing the prod­uct char­ac­ter­is­tics such as dis­so­lu­tion pro­file, bioavail­abil­ity, flow abil­ity, and the ease with which the crys­tals are com­pressed into tablets. Winn et al. (2000) pre­sented a re­view on mod­el­ing of crys­tal shapes of or­ganic ma­te­ri­als grown from so­lu­tion. It has been ob­served that crys­tals grown from a mix­ture of sol­vents have dif­fer­ent char­ac­ter­is­tics than the crys­tal grown from pure sol­vent, and this ef­fect is sig­nif­i­cant if the solute has very dif­fer­ent sol­u­bil­ity in each sol­vent. Frank et al. (1999) re­viewed strate­gies for sol­vent se­lec­tion meth­ods used to es­ti­mate the sol­u­bil­ity of or­ganic solids in a wide va­ri­ety of sol­vents for var­i­ous types of crys­tal­liza­tion pro­cesses such as cool­ing crys­tal­liza­tion and drown­ing out crys­tal­liza­tion, where the task of sol­vent se­lec­tion is car­ried out from a list of good sol­vents from a data­base with sol­u­bil­ity be­ing the only cri­te­rion for se­lec­tion. This ap­proach may miss some bet­ter sol­vents or may re­quire too ex­ten­sive ex­per­i­men­ta­tion to ar­rive at the de­sired sol­vent. There­fore, com­puter aided molec­u­lar de­sign (CAMD) meth­ods for sol­vent de­sign have been attempted as ad­vance­ment in this di­rec­tion. Harini et al. (2013) have pro­vided a com­pre­hen­sive re­view on the method­olo­gies avail­able for the pre­dic­tion of prod­uct (sol­vent) prop­er­ties from the molec­u­lar struc­ture and prop­erty es­ti­ma­tion tech­niques, along with the var­i­ous op­ti­miza­tion ap­proaches and com­pu­ta­tional schemes to solve the sol­vent de­sign prob­lem for crys­tal­liza­tion and sol­vent ex­trac­tion in phar­ma­ceu­ti­cal in­dus­try. The prop­er­ties of in­ter­est for crys­tal­liza­tion process in a phar­ma­ceu­ti­cal in­dus­try in­clude melt­ing point, boil­ing point and di­elec­tric con­stant, sol­u­bil­ity (Hilde­brand sol­u­bil­ity pa­ram­e­ter, Hansen sol­u­bil­ity pa­ram­e­ter), vis­cos­ity, heat of fu­sion, crit­i­cal tem­per­a­ture, en­thalpy of va­por­iza­tion, flash point, po­lar­ity of or­ganic sol­vents, hy­dro­gen bond­ing propen­sity or in­ter­ac­tion pa­ram­e­ter, the oc­tanol−wa­ter par­ti­tion co­ef­fi­cient, the donor/ac­cep­tor num­bers, the sol­va­tochromic pa­ram­e­ters, and other en­vi­ron­men­tal­re­lated pa­ram­e­ters. It has been ob­served that hy­dro­gen bond­ing ten­dency can af­fect crys­tal mor­phol­ogy or habit. Ibupro­fen crys­tals crys­tal­lized from sol­vents with high hy­dro­gen-bond­ing abil­ity were plate-like, with a low as­pect ra­tio and large size. On the other hand, ibupro­fen crys­tal­lized from sol­vents with low hy­dro­gen bond­ing abil­ity were nee­dle-like crys­tals with high as­pect ra­tio (Karunanithi et al., 2007). It has been found and ver­i­fied that ibupro­fen crys­tals formed from 2-ethoxy ethyl ac­etate as the sol­vent are sig­nif­i­cantly larger and have a low as­pect ra­tio, when com­pared to crys­tals formed from the sol­vent n-hex­ane, by the com­bined ap­proach of ex­per­i­ments, data­base search, and CAMD.

3.Crys­tal­liza­tion Ki­net­ics

As su­per­sat­u­ra­tion is the driv­ing force for crys­tal nu­cle­ation and growth, ul­ti­mately dic­tat­ing fi­nal crys­tal size dis­tri­bu­tion, the re­la­tion­ship is de­fined by a

well known set of equa­tions out­lined by Nyvlt (1968). Fig­ure 2 re­lates su­per­sat­u­ra­tion to nu­cle­ation, growth and crys­tal size. The value of the growth or­der is typ­i­cally be­tween 1 and 2, while the value of nu­cle­ation or­der is typ­i­cally be­tween 5 and 10. At low su­per­sat­u­ra­tion, crys­tals can grow faster than they nu­cle­ate, re­sult­ing in larger crys­tal size dis­tri­bu­tion. How­ever, at higher su­per­sat­u­ra­tion, crys­tal nu­cle­ation dom­i­nates crys­tal growth, ul­ti­mately re­sult­ing in smaller crys­tals. Mod­ern tech­niques such as At­ten­u­ated to­tal re­flec­tion (ATR) - Fourier trans­form in­frared (FTIR) spec­troscopy al­low sol­u­bil­ity traces to be devel­oped quickly and eas­ily, and the pre­vail­ing level of su­per­sat­u­ra­tion to be mon­i­tored con­tin­u­ously through­out a crys­tal­liza­tion ex­per­i­ment. The key con­trib­u­tors in­flu­enc­ing the ki­net­ics of crys­tal­liza­tion (in other words, the size dis­tri­bu­tion) are Nu­cle­ation / Growth ki­net­ics (gov­erned by su­per­sat­u­ra­tion) and Seed.

3.1 Nu­cle­ation and Growth, Crys­tal Size Dis­tri­bu­tion

Su­per­sat­u­ra­tion af­fects both crys­tal growth and nu­cle­ation rates, which in turn im­pact the crys­tal size dis­tri­bu­tion. A higher level of nu­cle­ation leads to smaller crys­tals and vice versa. Also, a high de­gree of nu­cle­ation rate over crys­tal growth rate due to a high de­gree of su­per­sat­u­ra­tion can lead to poorer re­jec­tion of im­pu­ri­ties. Given these con­sid­er­a­tions, con­trol of su­per­sat­u­ra­tion, cou­pled with the uti­liza­tion of proper seed, to max­i­mize crys­tal growth and min­i­mize nu­cle­ation is gen­er­ally pre­ferred. It should be pointed out that, if a higher de­gree of nu­cle­ation is cre­ated by a higher level of mix­ing in­ten­sity in­stead of su­per­sat­u­ra­tion alone, it does not nec­es­sar­ily af­fect the prod­uct pu­rity. Com­mon nu­cle­ation sit­u­a­tions from so­lu­tion may be seen in Fig­ure 3.

Crys­tal growth, on the other hand, is the ad­di­tion of more solute mol­e­cules to the nu­cle­ation site or crys­tal lat­tice to evo­lu­tion of macro­scopic crys­tal form of de­fined size and shape. In other words, crys­tal size dis­tri­bu­tion and mor­pholo­gies pro­duced are a re­sult of the rel­a­tive rates of re­ac­tion of nu­cle­ation and crys­tal growth. Crys­tal growth is con­sid­ered to be a re­verse dis­so­lu­tion process and cer­tain dif­fu­sion the­o­ries con­sider that mat­ter is de­posited con­tin­u­ously on a crys­tal face at a rate pro­por­tional to the dif­fer­ence of con­cen­tra­tion be­tween the sur­face and the bulk so­lu­tion.

To gain good con­trol of su­per­sat­u­ra­tion, some quan­ti­ta­tive mea­sures of crys­tal growth and nu­cle­ation rate con­stants are needed. Due to the di­ver­si­fied na­ture of API (or in­ter­me­di­ate) crys­tals, nu­cle­ation and crys­tal growth rates can vary dras­ti­cally over sev­eral or­ders of mag­ni­tude. At the cur­rent time, it is not pos­si­ble to quan­ti­ta­tively pre­dict the crys­tal­liza­tion be­hav­ior on the ba­sis of the­o­ret­i­cal mod­els. To ad­dress this lim­i­ta­tion, a model-based ex­per­i­men­tal de­sign (MBED) method­ol­ogy for crys­tal­liza­tion was re­ported (Davey et al., 2002). How­ever, lim­i­ta­tions re­side in the va­lid­ity of the pop­u­la­tion bal­ance equa­tion pri­mar­ily used for pre­dict­ing the crys­tal size dis­tri­bu­tion. Sev­eral so­lu­tion meth­ods for solv­ing the Pop­u­la­tion Bal­ance Equa­tions, such as Meth­ods of Mo­ments, Method of Classes, Or­thog­o­nal col­lo­ca­tion meth­ods etc., re­ported have been fairly in prac­tice over the past two decades in un­der­stand­ing the crys­tal size dis­tri­bu­tions (Cameron et al., 2005). Sev­eral ap­proaches to un­der­stand crys­tal size dis­tri­bu­tion avail­able are sum­ma­rized in Ta­ble 3 (Also shown are the dif­fer­ent tech­niques / meth­ods avail­able for crys­tal­liza­tion stud­ies). An in­ter­est­ing in­ter­play be­tween ther­mo­dy­namic, ki­netic and molec­u­lar recog­ni­tion phe­nom­ena that gov­erns crys­tal­liza­tion is re­ported, which acts as a good aid in in­ter-link­ing these key as­pects to­wards mon­i­tor­ing and con­trol as far as ap­pli­ca­tion of crys­tal­liza­tion phe­nom­e­non in the

in­dus­try is con­cerned. The same is shown in Ta­ble 2. 3.2 Seed

Seed­ing a so­lu­tion with prod­uct crys­tals is a very well-es­tab­lished method to in­duce crys­tal­liza­tion. Seed­ing im­pacts and in­ter­feres with all as­pects of crys­tal­liza­tion in­clud­ing crys­tal size, size dis­tri­bu­tion, crys­tal form, nu­cle­ation & growth rates, yield and prod­uct pu­rity. Seed­ing is im­por­tant for sys­tems that are very dif­fi­cult to nu­cle­ate or tend to in­duce liq­uid-phase sep­a­ra­tion, be­cause seed­ing re­duces nu­cle­ation time that may be oth­er­wise too long from an eco­nomic per­spec­tive. Ad­di­tion of small seeds to a su­per­sat­u­rated so­lu­tion can in­crease the nu­cle­ation rate and also can se­lec­tively nu­cle­ate a de­sired crys­tal form (es­pe­cially a poly­morph), pro­vided the prod­uct seed crys­tals are avail­able. Seed­ing with a de­sired crys­tal form can also re­duce the risk of for­ma­tion of un­de­sired crys­tal forms. A de­tailed sum­mary on seed­ing ap­proaches for poly­morph con­trol has been pre­sented by Linas and Good­man (2008).

In en­gi­neer­ing per­spec­tive, the pur­pose of seed­ing a su­per­sat­u­rated so­lu­tion is to pro­vide start­ing sur­face area for crys­tal growth and avoid or re­duce nu­cle­ation as much as pos­si­ble. Seed load­ing (Weight % of seeds), seed size, time of seed ad­di­tion and ad­di­tion point of seed are the crit­i­cal quan­ti­ta­tive pa­ram­e­ters for a seed- ing pol­icy to at­tain crys­tal qual­ity at­tributes.

Seed­ing and Metastable zone: Most of the API’s are pro­duced through batch cool­ing crys­tal­liza­tion as they are very much tem­per­a­ture sen­si­tive. For seed­ing, it is nec­es­sary to un­der­stand the width of metastable zone (MSZ) and a range of op­er­at­ing con­di­tions over which a so­lu­tion can be su­per­sat­u­rated. The point in the MSZ where the seed is added (seed­ing tem­per­a­ture) also has a big im­pact on the rel­a­tive rates of nu­cle­ation and growth. Seed­ing close to sol­u­bil­ity curve (low su­per­sat­u­ra­tion) re­sults in slow nu­cle­ation and less num­ber of fine par­ti­cles. Whereas, close to MSZ (high su­per­sat­u­ra­tion) re­sults in high nu­cle­ation rates and thus, more num­ber of fine par­ti­cles. Like­wise, Seed­ing close to MSZ re­sults in less crys­tal growth, whereas close to sol­u­bil­ity curve pro­duces more num­ber of large par­ti­cles. Seed­ing tem­per­a­ture can be ma­nip­u­lated to have a nu­cle­ation or growth dom­i­nated process that can in turn in­flu­ence the prod­uct size or size dis­tri­bu­tion.

Seed load­ing and seed size: For a growth dom­i­nated crys­tal­liza­tion process, where nu­cle­ation is min­i­mized, fi­nal par­ti­cle size can be pre­dicted or con­trolled based on amount of seeds, seed size and amount of solid grown on the seed crys­tals. There has been a lot of dis­cus­sion in lit­er­a­ture about seed load­ing. Kub­ota et al. (2001) de­fined seed load­ing that sup­presses nu­cle­ation

ef­fec­tively as ‘crit­i­cal seed load­ing’ as a func­tion of seed size to achieve uni­modal dis­tri­bu­tion of fi­nal prod­uct crys­tals. There have been ev­i­dences about the ef­fect of con­cen­tra­tion of solute and tem­per­a­ture of seed­ing on the yield of crys­tal­liza­tion. A seed­ing strat­egy with a com­bi­na­tion of seed­ing tech­nique and ma­nip­u­lat­ing the pro­file of su­per­sat­u­ra­tion-gen­er­at­ing vari­ables in­creases the batch con­sis­tency. This has been demon­strated re­cently through rig­or­ous sim­u­la­tions us­ing mul­ti­ob­jec­tive op­ti­miza­tion ap­proach (He­malatha & Rani, 2017). In their work, op­ti­mal seed­ing pol­icy in terms of seed mass and seed size has been de­ter­mined for pro­duc­ing op­ti­mal crys­tal size and dis­tri­bu­tion of the fi­nal prod­uct. Seed prop­er­ties have been in­cluded as op­ti­mized vari­ables along with op­er­at­ing pro­file in the for­mu­la­tion of op­ti­miza­tion prob­lem. Var­i­ous com­bi­na­tions of seed mass and size pro­vide op­tions for choice of an op­er­at­ing tra­jec­tory (in this case tem­per­a­ture) to at­tain de­sired prod­uct prop­er­ties with ob­jec­tives be­ing max­i­miza­tion of mean crys­tal size, min­i­miza­tion of co­ef­fi­cient of vari­a­tion and min­i­miza­tion of nu­cle­ated crys­tals. Also it has been proven that ad­di­tion of seeds de­creases the su­per­sat­u­ra­tion and min­i­mizes nu­cle­ation. Sev­eral stud­ies re­port seed load­ing and size through mod­el­ling and op­ti­miza­tion (Chung et al., 2001; Choong and Smith, 2004; and Sarkar et al., 2005).

Seed­ing tech­niques- Ex­ter­nal seed­ing: There are dif­fer­ent ap­proaches for gen­er­a­tion of seed crys­tals, ei­ther dry or in slurry (wet) form. For the dry ap­proach, a batch of solid API (or in­ter­me­di­ate) is dry-milled. A por­tion of the dry milled API is re­tained as a fresh seed for the next batch while the bulk is sent for down­stream pro­cess­ing. Based on the prod­uct par­ti­cle spec­i­fi­ca­tion, dif­fer­ent equip­ment are used for dry milling (pin mill and jet mill). As the dry milling oper­a­tions con­sume less en­ergy for process de­vel­op­ment, they are used in the early phase of process de­vel­op­ment. How­ever, there are many dis­ad­van­tages as­so­ci­ated with it. Some crys­tals may get dis­solved or get struck to the sur­face of the mill which may in­ter­act with the metal sur­face. Some par­ti­cles may lose their crys­tallinity (Tung, 2013). In turn, ad­di­tion of dry seeds may in­duce un­wanted sol­vent en­trap­ment and ag­glom­er­a­tion, dur­ing crys­tal­liza­tion process as the dry seeds might form ag­gre­gates dur­ing stor­age. Dur­ing dry milling, the sur­face of the solids might get rup­tured and can form var­i­ous ac­tive and non-uni­form edges which form non-uni­form growth sites for the prod­uct crys­tal that causes prod­uct crys­tals to grow ir­reg­u­larly. To avoid this ‘Heal­ing’ process for the seed crys­tals can be done where the seed crys­tals are kept in a sat­u­rated slurry for many hours where the sharp edges get dis­solved due to Oswald ripen­ing. Another sim­ple pro­ce­dure em­ployed of­ten is ‘Isother­mal hold’ where the seed crys­tals are added to the su­per­sat­u­rated so­lu­tion and kept at ini­tial tem­per­a­ture for some time where dis­so­lu­tion takes place and crys­tal growth dis­lo­ca­tion can be avoided.

Adding the seed in slurry form (ei­ther in anti-sol­vent or mother liquor) may help the seeds dis­perse very well to their ac­tual crys­tal size. Hence, wet milling is a pre­ferred oper­a­tion for later phase in drug de­vel­op­ment. Wet seed­ing of­fers ad­van­tages of con­trolled seed crys­tal form, tun­able seed size and amount, and ro­bust­ness of scale up. Wet seed can be gen­er­ated through wet milling the API in a slurry that is sat­u­rated with the crys­tal­liz­ing solute. Based on the size re­quire­ment of the seeds, dif­fer­ent wet milling op­tions are avail­able. Me­dia mill is used for gen­er­at­ing sub­mi­crom­e­ter par­ti­cles, ul­tra­sound de­vice or ro­tor/sta­tor ho­mog­e­nizer for pro­duc­ing mi­crom­e­ter par­ti­cles.

Seed­ing tech­niques- In­ter­nal seed­ing: Some of the con­cerns in ex­ter­nal seed­ing can be ad­dressed by cre­at­ing the seed par­ti­cles in­ter­nally in the process which is known as in­situ seed gen­er­a­tion ap­proach. In this ap­proach, an API is dis­solved in a sol­vent mix­ture and fresh solid are gen­er­ated un­der con­di­tions of su­per­sat­u­ra­tion with con­trolled mix­ing. Rapid pre­cip­i­ta­tion us­ing im­ping­ing jet de­vice to gen­er­ate fine par­ti­cles also falls un­der this cat­e­gory.

Ad­vanced seed­ing tech­niques have been in­ves­ti­gated as pos­si­ble al­ter­na­tives to these tra­di­tional ap­proaches with ad­vances in par­ti­cle mea­sure­ment tech­niques like Fo­cussed beam re­flectance mea­sure­ment (FBRM). An in situ par­ti­cle seed­ing ap­proach, which con­sists of an au­to­mated closed loop feed­back con­trol tech­nique us­ing FBRM for un­seeded cool­ing crys­tal­liza­tions, to pro­duce seeds dur­ing the process for en­sur­ing con­sis­tent and re­peat­able crys­tal prod­uct qual­ity was re­ported (Chew et al., 2007). Im­prove­ment in the fi­nal prod­uct par­ti­cle dis¬tri­b­u­tion and con­sis­tency is ob­served for Glycine-wa­ter and Parac­eta­mol-wa­ter sys­tems when this feed­back con­trol ap­proach is used com­pared to con­ven­tional cool­ing crys­tal­liza­tion ap­proach. For this ap­proach, pre­dic­tion of model pa­ram­e­ters is nec­es­sary which may not be pos­si­ble or may be a te­dious job for some API’s. In such cases, ‘Di­rect nu­cle­ation con­trol’ (DNC) seems promis­ing as it is a model free con­trol ap­proach for in­situ seed gen­er­a­tion (Abu Bakar et al., 2009). DNC is a mod­i­fi­ca­tion of the feed­back con­trol ap­proach. In this ap­proach, the seed

count is mea­sured on­line through FBRM in the con­trol re­gion and the nu­cle­ation and dis­so­lu­tion is con­trolled through a feed­back con­trol strat­egy. The ad­van­tage of this tech­nique is that it does not use pre-de­ter­mined heat­ing or cool­ing pro­files. The tem­per­a­ture pro­files are gen­er­ated au­to­mat­i­cally and con­tin­u­ously dur­ing the crys­tal­liza­tion process in re­sponse to the num­ber of par­ti­cles gen­er­ated by the nu­cle­ation events. No prior knowl­edge of the model, process ki­net­ics or the metastable zone width is nec­es­sary as this ap­proach will au­to­mat­i­cally de­ter­mine the op­ti­mal op­er­at­ing pro­file by con­tin­u­ously de­tect­ing the metastable zone limit in real time us­ing a feed­back con­trol strat­egy. DNC has ap­pli­ca­tions in fines re­moval also apart from con­tin­u­ous seed gen­er­a­tion as this ap­proach re­duces ef­fects of break­age and pro­duces uni­form crys­tals.

These in situ ap­proaches ob­vi­ate gen­eral test­ing needed for stor­age and sta­bil­ity of the seeds. In ad­di­tion, it is very flex­i­ble over other seed gen­er­a­tion meth­ods as seed is gen­er­ated on de­mand which can avoid ex­cess seed amount. But, con­trol of crys­tal forms dur­ing seed gen­er­a­tion is another con­cern. The metastable forms, amor­phous solids, or oil droplets can form ini­tially. If the sta­ble crys­tal form is de­sired for the seed, an ad­di­tional aging with in­ten­sive mix­ing to fa­cil­i­tate the con­ver­sion of metastable crys­tal forms to the sta­ble crys­tal form is a good ap­proach. Ad­di­tion of dry/wet seed of the sta­ble crys­tal form prior to the gen­er­a­tion of in situ seed is another sim­ple op­tion.


Crys­tal­liza­tion is one of the key process steps for man­u­fac­tur­ing a large num­ber of prod­ucts in fine and spe­cialty chem­i­cals sec­tor (in­clud­ing API). Prod­uct pu­rity and other de­sired prod­uct at­tributes (like size dis­tri­bu­tion, habit, mor­phol­ogy and poly­morphs) are of­ten cru­cial. The ar­ti­cle will con­tinue into the sec­ond part to be pub­lished in the forth­com­ing is­sue. Ref­er­ences:

1. Abu Bakar, M. R.; Nagy, Z. K.; Saleemi, A. N.; Rielly, C. D., 2009, Cryst. Growth Des., 9(3), 1378-1384.

2. Cameron, I. T.; Wang, F. Y.; Im­manuel, C. D., Stepanek, F., Chem. Eng. Sci., 2005, 60(14) , 3723-3750.

3. Chen, C.; Song,Y., Ind. Eng. Chem. Res., 2004, 43, 8354−8362.

4. Chen, J.; Sarma, B.; Evans, J. M.; My­er­son, A. S, Cryst. Growth. & Des, 2011, 11(4), 887-895.

5. Chew, J. W.; Chow, P. S.; Tan, R. B. , Cryst. Growth Des., 2007, 7(8), 1416-1422.

6. Choong, K. L.; Smith, R., Chem. Eng. Sci., 2004, 59 (2), 313−327.

7. Chung, S. H.; Ma, D. L.; Braatz, R. D., Can. J. Chem. Eng., 1999, 77 (3), 590−596.

8. Davey, R. J.; Blag­den, N.; Righ­ini, S.; Ali­son, H.; Fer­rari, E. S. Phys. Chem. B 2002, 106, 1954−1959.

9. Doki, N.; Seki, H.; Takano, K.; Asa­tani, H.; Yokota, M.; Kub­ota, N, Cryst. Growth Des., 2004, 4(5), 949-953.

10. Frank, M.; Hö­land, W.; Rhein­berger, V.; J. non-crys­talline solids, 1999, 253 (1-3), 170-177.

11. Gao, Z.; Ro­hani, S.; Gong, J.; Wang, J., En­gi­neer­ing, 2017, 3(3), 343-353.

12. Harini, M.; Ad­hikari, J.; Rani, K. Y. , Ind. & Engg. Chem. Res., 2013, 52(21), 6869-6893.

13. He­malatha, K. , Rani, K.Y. , Ind. Eng. Chem. Res., 2017, 56, 6012–6021.

14. Karunanithi, A. T.; Ac­quah, C.; Ache­nie, L. E.; Sitham­baram, S.; Suib, S. L.; Gani, R., Chem. Eng. Sci., 2007, 62(12), 3276-3281.

15. Kawa­bata, Y.; Wada, K.; Nakatani, M.; Ya­mada, S.; Onoue, S., Int. J. Pharm., 2011, 420(1),1–10.

16. Lli­nas, A; Good­man J. M., Drug Dis­cov To­day, 2008, 13 (5–6), 198–210.

17. Lovett, M. A.; Mu­ra­tore, M.; Do­herty, M. AIChE J. 2011, 58, 1465−1474.

18. Nývlt, J. , J. Cryst. Growth,, 1968 , 3, 377-383.

19. Pow­ell, K.A.; Bar­tolini, G.; Wit­ter­ing, K. E.; Saleemi, A.N; Wil­son, C.C; Rielly, C. D, Cryst. Growth Des., 2015, 15(10), 4821–36.

20. Ro­driguez-Spong B.; Price, C.P.; Jayasankar A., Adv Drug De­liv Rev, 2004, 56, 241-274.

21. Sarkar, D.; Ro­hani, S.; Ju­tan, A. Chem. Eng. Sci., 2006, 61 (16), 5282−5295.

22. Shekunov, B.; York, P. J. Cryst. Growth, 2000, 211, 122-36.

23. Tung, H. H., Org. Proc. Res. & Dev., 2013, 17(3) 445-454.

24. Tung, H.-H.; Paul, E.; Mi­dler, M.; McCauley, J.; Wi­ley: Hobo­ken, NJ, 2009.

25. Var­i­ankaval, N.; Cote A. S.; AIChE J , 2008, 54 (7), 16821688.

26. Winn, D.; Do­herty, M. F., AIChE J, 2000, 46( 7), 13481367.

Fig­ure 1: Sol­u­bil­ity curve (Adapted from Jones, 2002)

Ta­ble 1: Solid-state prop­er­ties de­fined by crys­tal­liza­tion process and their re­la­tion­ship with spe­cific char­ac­ter­is­tics of drug sub­stances and drug prod­ucts (Adapted from Shekunov., & York., 2000)

Fig­ure 2: Re­la­tion­ship be­tween su­per­sat­u­ra­tion, nu­cle­ation, growth and crys­tal size (adapted from

Fig­ure 3: Com­mon nu­cle­ation sit­u­a­tions from so­lu­tion (Adapted from Crow­der et al., 2003)

Ta­ble 2: In­ter­play be­tween ther­mo­dy­namic, ki­netic and molec­u­lar recog­ni­tion phe­nom­ena in crys­tal­liza­tion (Adapted from Ro­driguez-Spong et al., 2004)

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.