Engineering 3 (2017) 343–353

Contents lists available at ScienceDirect

Engineering
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / e n g

Research
Green Chemical Engineering—Review

Recent Developments in the Crystallization Process: Toward

the Pharmaceutical Industry
Zhenguo Gao a,b, Sohrab Rohani a,*, Junbo Gong b, Jingkang Wang b
a
Department of Chemical and Biochemical Engineering, The University of Western Ontario, London, ON N6A 5B9, Canada
b
School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China

a r t i c l e i n f o a b s t r a c t

Article history: Crystallization is one of the oldest separation and purification unit operations, and has recently contributed
Received 17 April 2017 to significant improvements in producing higher-value products with specific properties and in building
Revised 14 May 2017 efficient manufacturing processes. In this paper, we review recent developments in crystal engineering and
Accepted 15 May 2017
crystallization process design and control in the pharmaceutical industry. We systematically summarize
Available online 8 June 2017
recent methods for understanding and developing new types of crystals such as co-crystals, polymorphs,
and solvates, and include several milestones such as the launch of the first co-crystal drug, Entresto (No-
Keywords: vartis), and the continuous manufacture of Orkambi (Vertex). Conventional batch and continuous processes,
Crystallization
which are becoming increasingly mature, are being coupled with various control strategies and the recently
Crystal engineering
developed crystallizers are thus adapting to the needs of the pharmaceutical industry. The development of
Polymorphism
Crystallization process design and control crystallization process design and control has led to the appearance of several new and innovative crystal-
Crystal size distribution lizer geometries for continuous operation and improved performance. This paper also reviews major recent
progress in the area of process analytical technology.
© 2017 THE AUTHORS. Published by Elsevier LTD on behalf of the Chinese Academy of Engineering and
Higher Education Press Limited Company. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction ance of a new polymorph during storage. The topic of maintaining

the stability of a solid-state drug in a dosage form has attracted
The considerable developments in the crystallization process increasingly significant attention in order to ensure product quality
in the pharmaceutical industry have been accelerated by several [5–7]. Different forms of solid state can lead to variations in product
high-profile cases over the past few decades. For example, thalid­ performance, such as a reduction of solubility and dissolution rates
omide was marketed as a sedative or hypnotic in the late 1950s or an increase in tablet hardness. Therefore, crystallization technol-
and early 1960s and was used by many pregnant women as an ogy, as a core technology, was selected as a means of controlling the
anti-nausea agent. However, while (R)-(+)-thalidomide served as a factors that impact solid-state phase transformations [8]. The US
sedative, its optical isomer (S)-(−)-thalidomide was tragically found Food and Drug Administration (FDA) and other regulatory agencies
to act as a teratogen, resulting in the malformation and death of have set strict standards to ensure the safety and stability of phar-
thousands of infants [1,2]. Another example occurred in 1998, 18 maceuticals. Further top-down supervision has put forward higher
months after the new commercial product ritonavir was launched. requirements for medicine production, and particularly for the
A new stable polymorph (form II) was identified in supplies of the crystallization process. Based on these practices and on advances in
drug [3], which greatly reduced ritonavir’s solubility compared with nucleation and growth theory at the molecular level [9–12], crystal-
the original crystal form, leading to an oral bioavailability problem lization is developing from an empirical science to an evidence- and
[4]. In another example in 2008, rotigotine (Neupro) was recalled in theory-based science.
the United States and in Europe because of the unexpected appear- Because the requirements for improving the efficiency and

* Corresponding author.
E-mail address: [email protected]

http://dx.doi.org/10.1016/J.ENG.2017.03.022
2095-8099/© 2017 THE AUTHORS. Published by Elsevier LTD on behalf of the Chinese Academy of Engineering and Higher Education Press Limited Company.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
344 Z. Gao et al. / Engineering 3 (2017) 343–353

properties of drugs are becoming more stringent, the pharmaceu- Schmidt [34] in 1971. Today, crystal engineering is a powerful tool
tical manufacturing sector is considering implementing process for designing pharmaceutical solids with desirable physicochemical
automation and launching continuous production facilities [13–15]. properties [35]. The diverse structures in pharmaceutical solids,
Precise control of batch processes and the design of continuous as highlighted by Cherukuvada and Nangia (Fig. 1) [36], provide
processes lead to more reliable products and to a higher production considerable maneuverability for optimizing product quality. Vari-
rate. Significant progress has been made in the control of crystal- ous intermolecular interactions and packing modes can be used at
lization processes, leading to improvements in different aspects of the molecular level in order to fine-tune the crystal structure with
crystalline product quality including the crystal size distribution desired physical and chemical properties [34,37]. “Fine-tuning”
(CSD), polymorphic form, morphology, purity, tap density, flowabil- includes introducing guest molecules to form multiple-component
ity, compactibility, solubility, and dissolution rate [16–19]. The de- crystals, screening the crystallization condition for different pack-
velopment of population balance models of crystallization systems ing arrangements and/or conformations, and promoting preferred
has provided a better understanding of the effects of major process crystal nucleation and growth via tailor-made additives and a solid-
variables, such as agglomeration, breakage, additives and impu­ liquid surface.
rities, and process control strategies, on the quality of the crystalline
material [20]. Two factors promote the research and application of 2.1. Polymorphism
crystallization process control: first, advances in the understanding
of the crystallization mechanism; and second, the advent of process After the issue with ritonavir in 1998 served as a warning to
analytical technology (PAT) [20–23]. pharmacists and crystal engineers [3], polymorphism became in-
In recent years, continuous crystallization has attracted increas- creasingly important in both fundamental research and intellectual
ing interest for crystal production. Mixed-suspension mixed-product property rights. In addition to its effect on drug safety, polymor­
removal (MSMPR) crystallizer is the most widely used type of con- phism is an important factor in the testing of generic drugs, a huge
tinuous crystallizer; it can be coupled with different control strat- expansion of which has occurred following the expiration of many
egies, including model-free and model-based approaches [24,25]. patents of original drugs.
The recently developed plug flow crystallizer (PFC), slug flow The question of how to screen the new polymorphs using a
crystallizer (SFC), microfluidic crystallizer, airlift crystallizer, and systematic approach rather than by chance has become an im-
impinging jet mixer crystallizer have shown promising results for portant one. Llinàs and Goodman [38] summarized the time scale
optimizing crystal qualities. The oscillatory baffled crystallizer (OBC) of different crystallization experiments. The rapid crystallization
also exhibits prospects for practical applications [26–29]. In addi- process is more likely to form metastable polymorphs (Fig. 2) [38].
tion, the coupling of other unit operations with the crystallization Mirmehrabi and Rohani [39] developed a method based on atomic
process and the incorporation of novel designs have enhanced the electronegativity for selecting a suitable solvent in the preparation
process efficiency [30–33]. of a desired polymorph. Hydrogen-bonding ability can be predicted
In this review paper, the factors that contribute to the develop- by calculating the partial charge distribution of solvent and solute
ment of the pharmaceutical crystallization process are grouped into molecules. A comprehensive database was explored by Allesø et al.
two categories: crystal engineering, and advanced solution crystalli- [40], containing 218 organic solvents and 24 property descriptors.
zation process design and control. Principal component analysis and self-organizing map analyses en-
able the convenient and rapid selection of diverse solvents. Besides
2. Crystal engineering the organized solvent database, high-throughput crystallization
platforms such as CrystalMax (TransForm Pharmaceuticals, Inc.) and
The concept of “crystal engineering” was first proposed by Crystal16™ (Avantium Technologies, Inc.) were developed to help

Fig. 1. Structural diversity of pharmaceutical solids. (Caption and figure reprinted with permission from Ref. [36])
 Z. Gao et al. / Engineering 3 (2017) 343–353 345

Fig. 2. Crystallization experiments showing the timescales that can be employed to favor stable or metastable polymorphs. (Caption and figure reprinted with permission from
Ref. [38])

screen the polymorphs of a given active pharmaceutical ingredient formulation and storage. Additives and some excipients have ex-
with high efficiency. Pfund and Matzger [41] created a high-density- hibited inhibiting effects on polymorphic transformation, indicating
format polymer-induced heteronucleation (PIHn) platform, in which a promising method of stabilizing the metastable crystal form.
288 distinct polymers act as crystallization directors for obtaining As the study of “polymorphism” suggests, an amorphous phase
novel solid forms. Advances in comprehensive polymorphic screen- can be treated as a special kind of polymorphic phase [48]. The de-
ing based on solvent selection and high-throughput platforms have velopment of an amorphous pharmaceutical lies in the competition
paved the way toward the ultimate goal of harvesting the desired between the advantages gained in the solubility and dissolution
polymorphs. The development of a multivariable control system rates, and the disadvantage of enhanced instability. Developing an
(i.e., controlling solvent, temperature, and supersaturation) coupled amorphous drug is an attractive method to improve the oral delivery
with high-throughput powder X-ray diffraction (PXRD) or Raman of poorly water-soluble drugs, as shown in Fig. 3 [49–52]. Numerous
detectors in an automated fashion is highly desirable for the phar- cases have been successfully developed to date, and the solubility
maceutical industry [41,42]. By comparing medicinal qualities based enhancement can be assessed using measured thermodynamic
on all possible solid forms, the best drug candidate can be selected quantities [53,54]. However, there is a challenge of instability due to
for further development. higher thermodynamic activity in the amorphous state. To inhibit
The transformation between different forms of pharmaceutical the inherent tendency of the amorphous phase to recrystallize, pro-
ingredients has attracted considerable attention in recent years as cessing and storage conditions as well as various newer polymers
researchers seek to produce or maintain the stability of a specific and excipients have been exploited. Amorphous solid dispersions
crystal form. Much research has been conducted to study the mech- can be stabilized by stabilizers, which can potentially modify the
anisms and conditions of transformation. Two kinds of mechanisms glass transition temperature or form non-covalent interactions, thus
of polymorphic transformation that affect the specific form pro- impacting the rate of crystallization [55]. As a single-phase blend,
duced and drug stability—solvent-mediated and solid-state—have co-amorphous formulations can be produced by using low-weight
been studied in the pharmaceutical industry. Driven by differences co-formers, which can significantly reduce the amount of stabilizers
in solubility, solvent-mediated transformation is generally divided compared with polymer and mesoporous silica [56,57].
into three steps: ① the dissolution of the unstable phase, ② nuclea-
tion, and ③ the growth of the stable form [43]. In the past few years, 2.2. Modification by guest molecules
modeling and in situ composite sensor arrays have helped research-
ers to understand mechanisms and optimize the conditions in vari- By introducing guest molecules, crystal engineering provides a
ous systems [44–46]. Temperature, stirring speed, solvent type, pH, number of routes to optimize the properties of active pharmaceuti-
seeding, and other variables have been emphasized by researchers, cal ingredients (APIs) and may also be used as a strategy to extend
with the nucleation rate or the growth rate of the new crystal form (or avoid) patent protection in the development of new drugs. Guest
usually being specified as the rate-determining step. Takeguchi et molecules such as salt-formers, co-formers, and solvates can occupy
al. [47] successfully obtained the desired high-purity polymorph the crystal lattice to remedy the deficiency of the original crystals
in the first step during a scale-up manufacturing process. Through without changing the chemical identity or biological activity of the
solvent- and temperature-screening experiments, higher tem- API. Properties of interest include the crystal size, shape, stability, and
perature and hydrogen-bond-donating solvents that promote the especially the aqueous solubility, because up to 90% of new API candi-
formation of hydrogen bonds were found to be preferable in the dates under development are poorly water soluble [57].
optimization and designing process. In solid-state transformation,
polymorphic transformation can happen during the formulation and 2.2.1. Salt
storage processes. Influential factors include drying, milling, granu- Salt formation is widely used with ionizable drugs, and over half
lation, and tabletting, as well as temperature and humidity chan- of the APIs approved by the FDA are pharmaceutical salts. Ionized
ges during the storage period. Extensive research on these factors APIs usually have greater solubility and dissolution rate, which
provides guidelines on the stability of the metastable form during are achieved by forming a corresponding salt or by modifying the
346 Z. Gao et al. / Engineering 3 (2017) 343–353

Fig. 3. The biopharmaceutics classification system (BCS), as defined by Amidon et al. [51], divided into four classes of solubility. Viable formulation options based on the BCS and
the proportion of marketed drugs versus pipeline drugs (insert column chart). (Adapted from Refs. [49,52])

solution pH. More and more applications of salt drugs have been de- ③ formulation and process scale-up [74]. Table 1 [74,76,77] lists
veloped in order to improve solubility and stability, as well as to ex- recently developed co-former selection methods and experimental
tend intellectual property protection [58–62]. The presence of a two- screening technologies. Lin et al. [75] developed a differential scan-
unit difference in pKa between APIs and acid/bases to ensure proton ning calorimetry Fourier-transform infrared (DSC-FTIR) technique,
transfer is a well-known screening principle. High-throughput which can realize a one-step screening and qualitative detection
technologies have been exploited for this purpose. Consideration procedure for co-crystal formation in real time. A process that in-
should also be given to safety and to the common ion effect. Typical volves theoretical prediction combined with a high-throughput
examples show the existence of common ions that may suppress and/or quick one-step method is being developed to accelerate
the dissolution of a salt and decrease its solubility. Thackaberry [63] co-crystal formation.
reviewed the non-clinical toxicity of counter ions in pharmaceutical Research on multidrug co-crystals has boosted the development
salts. Additional research to provide examples of typical counter of effective therapeutic hybrids [76]. As more co-crystals appear in
ions, especially commonly used anions/cations, would help to opti- the open literature and in patents publications, they present chal-
mize salt-formers and formulation methods in the future. lenges regarding the large-scale synthesis and stability of these
drugs in the presence of excipients.
2.2.2. Co-crystals
Starting with the first introduction of co-crystals into the 2.2.3. Solvates
pharmaceutical industry in the early 2000s, and through to the When a solvent molecule is crystallized with a host molecule in
launch of the first commercial product, Entresto (Novartis), in 2015, the same crystal lattice, whether stoichiometrically or non-stoichio-
great progress has been made in the development of pharmaceu­ metrically, a new solid phase called the solvate is formed. The
tical co-crystals, and hundreds of case studies have been published solvate usually exhibits different physicochemical properties than
[64–66]. We now know that co-crystals are single-phase crystalline the original solid phase. Hydrates are the most important kind of
solids that include two or more different molecules and/or ionic pharmaceutical solvate because of their nontoxic and stable prop-
compounds in a stoichiometric ratio [67]. The FDA recently released erties [78]. These properties are due to the excellent ability of water
a paper outlining standards and guidance for the development of molecules to form hydrogen bonds, in comparison with other or-
pharmaceutical co-crystals and providing specifications for how ganic solvents. A growing number of studies have focused on the
to develop and test co-crystal drugs [68]. Various co-crystal model solvent-molecule interactions that trigger solvate formation, and
compounds have demonstrated improved physicochemical perform- have tried to establish principles for solvate screening, preparation,
ance that could benefit the pharmaceutical industry. Improvements and storage. The main factors involved in solvate formation are con-
have been reported in solubility/dissolution rate, stability, bioavail- sidered to be the hydrogen bond acceptor and donor abilities and
ability, melting point, mechanical properties, permeability, and so the polarity of the solvent [79]. Two challenges remain in solvate
forth [69–73]. drug development: ① The screening and exploration of the forma-
In general, there are three main steps in the development of tion, transformation, and storage conditions of a solvate take a great
a new co-crystal drug, prior to final approval: ① the design and deal of time and carry a high cost [80]; and ② it is difficult to distin-
selection of the co-former and experimental screening; ② an guish between stoichiometric and non-stoichiometric solvates when
evaluation of the solid properties and preclinical performance; and the solvent content is similar. These challenges leave considerable
 Z. Gao et al. / Engineering 3 (2017) 343–353 347

Table 1
Co-former selection methods and experimental screening technologies for co-crystal preparation. (Summarized from Refs. [74,76,77])
Selection/screening methods Key notes
Co-former selection Supramolecular compatibility Based on Cambridge Structural Database or Hansen solubility parameter prediction
methods
Shape and polarity analysis Based on shape and polarity of co-former and API
Lattice energy calculation Based on lattice energy minimization methodology
Virtual co-crystal screening Based on molecular electrostatic potential surfaces
Conductor-like screening Fluid-phase thermodynamics theory conductor-like screening model
Experimental screening Solvent evaporation The most widely used, cost-efficient method
technologies
Solution co-crystallization Cooling, anti-solvent, slurry, ultrasound-assisted, and microwave-assisted crystallization
Mechanical grinding Neat solvent/polymer-assisted grinding
Supercritical fluid technology Co-crystallization with supercritical solvent
DSC-FTIR micro spectroscopy Simultaneous DSC-FTIR micro spectroscopic system
High-throughput technology Using in situ Raman microscope and a multi-well plate, high efficiency
Spray drying A promising method for large-scale co-crystal generation
High-shear granulation High-shear wet granulation

room for controversial issues in patents and in the development of ing undesired phenomena such as liquid-liquid phase separation,
solvate drugs [78]. jelly-like phase formation, and highly viscous systems.

2.3. Crystal structure prediction 3. Solution crystallization process control and design

Crystal structure prediction (CSP) methods can provide a micro- Prior to the 1990s, solution crystallization process control was
scopic perspective to supplement experimental studies on thermo- limited because of a lack of sufficiently accurate in situ sensors (i.e.,
dynamic stability and polymorphism, and to guide the course of ex- to monitor concentration, CSD, and polymorphic nature) and a lack
perimentation. Substantial progress has been made in the computed of understanding about the processing factors of crystallization [23].
crystal energy landscape over the past decade, combined with the re- Advances in crystallization process design and control have occurred
sults from industrial crystallization processes [81,82]. A series of blind over the last couple of decades, based on the development of real-time
tests of CSP that deal with flexible molecules have been performed monitoring and on emerging commercial software for the crystal-
since 1999, hosted by the Cambridge Crystallographic Data Center lization process, such as the gCRYSTAL®, DynoChem®, and COMSOL
(CCDC) [82]. The prevailing consensus is that the experimental results Multiphysics® software. Process-modeling software permits more
analysis and computational optimization are based on close packing, effective design and operation of crystallizers, and facilitates the
conformation preferences, and intermolecular interactions between optimization of seeding, operation profiling, and scaling up. For
API molecules and guest molecules. CSP can provide reliable guidance example, computational fluid dynamics (CFD) simulations help
in identifying the most stable form of conformation through the lattice researchers to understand the hydrodynamics and crystallization
energy gap. However, several bottlenecks hinder the progress of CSP kinetics that occur during the scale-up process, and can be used to
as an accurate method for solid-form screening: ① The vast number guide scale-up strategies. In addition, PAT enables better control and
of potential crystal structures poses a challenge for computational design of pharmaceutical processes; its goals of quality-by-design
work when dealing with crystal cell parameters and flexible torsion (QbD) and quality-by-control (QbC) help to improve efficiency and
angles; ② there are differences between the real crystal free energy regulate risk in the pharmaceutical industry.
and the calculated lattice energy at 0 K; and ③ there is a lack of kinetic
knowledge of crystal nucleation and growth, and of the combination 3.1. Process analytical technology
of thermodynamic and kinetic simulation [83].
The PAT concept was proposed by the FDA in 2004 to ensure final
2.4. Other developments in crystal engineering product quality by the timely monitoring, analyzing, and controlling
of process parameters. During the last decade, a wide usage of in
Researchers have continued to explore the theory and applica- situ monitoring technologies, including attenuated total reflectance
tions of crystal nucleation and morphology over the past decades. Fourier-transform infrared (ATR-FTIR) spectroscopy, focused beam
Two-step nucleation theory and crystal morphology prediction were reflectance measurement (FBRM), Raman spectroscopy, and particle
developed in recent years, and have been applied in various prac- vision measurement (PVM), has helped to improve data quality
tices [84,85]. Tailor-made or functional additives and nanoporous and agility, as well as process reliability and performance. ATR-FTIR
templates have been developed that permit the selective nucleation enables accurate monitoring and control of the solution concen-
and growth of a specific crystal form, thus significantly affecting the tration. Its accuracy and agility have been improved by purging
crystallization process and the product properties [86,87]. Diao et al. the background and using chemometrics techniques on multiple
[88,89] conducted a systemic study on the role of surface chemistry wavenumbers in order to correlate spectral intensity and solution
and the morphology of various polymeric substrates on heteroge­ concentration. The use of fiber optics has improved flexibility and
neous nucleation. Ultrasound- and microwave-assisted crystalliz- promoted wide application of the PAT concept in both academia
ation have also shown considerable prospects for intensifying the and the industry. The continuous tracking of particle count and size
nucleation and growth processes [90]. Recent studies reported on distributions by FBRM has made quantitative calculations available
the gel formation or jelly-like phase-mediated crystallization of for the development of accurate crystallization models. The newest
inefficient crystallization systems, which provides a new method of generation of FBRM has overcome the probe-fouling problem, which
producing crystals [91,92]. However, challenges still remain regard- required frequent cleaning during the monitoring period [93]. As a
348 Z. Gao et al. / Engineering 3 (2017) 343–353

well-developed technology, Raman spectroscopy has been used to comparison with the liquid-phase measurement of T/C control, DNC
identify differences in polymorphism and to test solvent-mediated directly measures the counts of solid particles and controls nuclea-
polymorphic transformation, solution concentration, and the poly- tion and dissolution through a feedback control strategy [102]. DNC
morphic ratio of the solid mixture [94]. PVM can track the real-time has shown promising applications in fines removal, continuous
visualization of particles in progress and can provide image-based seed preparation, mitigating the effects of breakage, and producing
particle trending, crystal growth, polymorphic transformation, ag- more uniform crystals. Yang et al. [25] first implemented the feed-
glomeration, and oiling out in crystallization processes. Combined back DNC method in single-stage and two-stage continuous MSMPR
with Raman spectroscopy and FBRM, image-based tracking can help crystallization processes; they achieved the desired CSD and were
to monitor and control crystal form, size, and shape [94]. Based on able to effectively suppress disturbances. The combination of DNC
these technologies, the in situ monitoring of the properties of both and T/C control methods has an advantage over DNC in the nuclea-
the liquid and solid phases has become possible. tion process and over T/C control in crystal growth, although such a
In recent years, attempts have been made to develop PAT-based combination inevitably increases the complexity of the system.
high-performance sensors as well as monitoring and control met­
h­o ds. Ref. [95], a multi-author review paper, clearly presents 3.2.2. Model-based control strategy
the state-of-the-art progress and current trends of PAT from a By combining crystallization process modeling and control with
multi-disciplinary perspective. The basic principles of different PAT optimization algorithms, model-based control strategies can be
sensors can be divided into imaging, spectroscopy, acoustic signals, developed to provide tight quality control in the presence of uncer-
and electronic signals. Spectroscopy is the most widely used PAT tainties and disturbances, while requiring fewer experiments. Fig. 4
in practice, and image-based monitoring and analysis technology [103] provides a comparison of the frameworks of the model-based
has shown potential applications. Simon et al. [96] introduced and model-free approaches. Model-based optimization can be ful-
endoscopy/stroboscopy-based technology, which is usually used in filled by means of a dynamic optimizer, by calculating the optimal
medical diagnosis, for use in the crystallization process; this tech- trajectory based on real-time sampled data.
nology can be used to acquire richer information such as particle Model-based optimization is subject to the constraints of the sys-
color, transparency, shape, and size. El Arnaout et al. [97] used an tem. To ensure consistency of the process model, especially for the
in-line imaging probe with an automated analysis algorithm to pro- model predictive control (MPC) strategy, a state observer/estimator
vide high-quality information on particle shape, size, and counts. can be used to estimate the internal states of the real crystallization
However, challenges still exist regarding image recognition (particle process, using a limited quantity of input and output data from the
overlapping) and mathematical modeling for particle feature analy- real system. The objective functions of model-based optimization
sis. Coupling ultrasonic velocity and attenuation, a new ultrasound include increasing the product yield and mean crystal size, or re-
technique has shown some success in simultaneously monitoring ducing batch/residence time and narrowing the width of the CSD.
solution concentration, particle size, and suspension density [98]. Attempts have been made to better understand process mechan-
Acoustic emission (AE) was developed as a non-contact in-line tech- isms, constraints and the effect of disturbances, process uncertain-
nology to track both the liquid and solid phases in suspension [99]. ties, and model/process mismatches. Mesbah et al. [104] developed
AE has the advantages of pre-nucleation acoustic signals and real- a model-based approach to control the growth rate below a con-
time crystal purity measuring during the crystallization process. strained value. The growth rate was correlated with the measured
PAT-based manufacturing has shown the ability to regulate pro- CSD through an extended Luenberger-type observer, and effective
cess variability and final product quality. However, PAT development control was implemented, thus improving yield and product quality.
faces certain challenges, including: ① the integration of signals from The solute concentration can also be controlled by using a phase
different PAT sensors, which would require chemometrics analy- diagram assisted by an analytical CSD estimator [105], or by using
sis and would affect control strategy; ② the development of high- the concentration control strategy in a hierarchical structure [106].
accuracy multicomponent monitoring systems for multicompon- Trifkovic et al. [107] proposed a novel way to directly calculate the
ent suspensions, in the presence of other species and impurities; nucleation and growth rates from the moments of particle popu-
③ high-quality solid-phase characterization and analysis, especial- lation density obtained by FBRM. By combining the crystallization
ly at high-suspension density; and ④ the application of multiple model with the population and mass balances, the optimal anti-sol-
sensors in the emerging continuous crystallization process. vent flow rate was obtained using both a single- and multi-objective
optimization algorithm. The nucleation rate was suppressed and the
3.2. Solution crystallization process control growth rate was minimized. Continuing this line of investigation,
Sheikhzadeh et al. [108] implemented a real-time optimization of
3.2.1. Model-free control strategy those objectives. MPC strategy has demonstrated advantages for
Model-free control strategy is carried out by means of a pre- multivariable control systems [109,110]. In future, the analysis of
defined temperature or solvent/anti-solvent ratio trajectory. It is multiple variables will have higher requirements; systematic an-
actually a feedback control strategy that is based on the difference alysis methods of statistical process control have shown potential
between the set points and the real-time measurements of factors application in this field [20].
such as concentration/supersaturation, particle counts, temperature
or concentration (T/C) control, direct nucleation control (DNC), or 3.3. Design of innovative solution crystallization processes
their combinations. Temperature control uses a predefined temper-
ature trajectory to control nucleation and growth processes. How- 3.3.1. Solution process design based on APIs
ever, this strategy can be seriously influenced by uncertainties in the Conventional crystallization technologies are mainly classified
operation optimal trajectory and by disturbances during the process as solution crystallization, melt crystallization, or reaction crys-
[100]. Concentration control was developed in the past decade and tallization. New types of crystallization processes have also been
has benefited from progress in real-time and accurate concentra- proposed, including membrane crystallization and supercritical
tion measurement (e.g., ATR-FTIR, ATR-UV/Vis). Chemometrics and fluid crystallization. The selection of a specific type of crystallization
calibration-free strategies for ATR-FTIR were successfully developed process mostly depends on the features of the model compound.
in order to construct the operating zone, which is usually confined For example, sodium chloride is prone to evaporation crystalliza-
between the solubility curve and the metastable limit [17,101]. In tion due to the limited sensitivity of solubility versus temperature;
 Z. Gao et al. / Engineering 3 (2017) 343–353 349

Fig. 4. Schematic representation of the model-based and model-free approaches for crystallization systems. (Caption and figure reprinted with permission from Ref. [103])

in contrast, APIs are more likely to require cooling crystallization, 3.3.2. New types of crystallizers
or a combination of cooling and anti-solvent crystallization, due to Crystallization operations can be classified into batch and con-
thermal sensitivity. Specific operational strategies have been de- tinuous processes, both of which have advantages and disadvantages.
veloped to optimize the process and product, such as seeding, fines Continuous crystallizers such as the Oslo-type crystallizer are suitable
removal, use of an optimal cooling rate or evaporation rate, and use for fragile crystals because they permit crystals to grow without in-
of a reactant/anti-solvent addition rate. However, various properties tense mechanical attrition. The details of other types of crystallizers,
of APIs can lead to difficulties during the process of crystallization, such as forced-circulation (FC) crystallizers, draft tube baffle (DTB)
such as oiling out, the gelation phenomenon, and so forth. crystallizers, and so forth, have been reviewed by Rohani [111] and
Oiling out is usually considered to be an undesirable phenome­ Paroli [116]. This section highlights several newly developed crystal-
non in the crystallization process [111]. De Albuquerque and Mazzotti lizer types and discusses their advantages and disadvantages.
[112] developed a robust process to avoid oiling out in a vanillin and The recently developed microfluidic crystallization technol-
water system. By using thermodynamic modeling and a phase dia- ogy has the advantages of being able to adapt to trace amounts of
gram, and designing the operational trajectory, the crystallization sample and good mass/heat transfer performance, and also shows
yield and crystal purity were maximized. It is interesting to note high efficiency and accuracy in experiments. A typical microfluidic
that Takasuga and Ooshima [113] designed oiling-out crystallization crystallizer is shown in Fig. 5(a) [117], with channels ranging from
in order to control crystal size, and resolved the problem of the low tens to hundreds of micrometers. Crystallization takes place in the
recovery of crystal product that occurred in a single-phase crystal- defined nanoliter volumes. This method permits high efficiency and
lization process. The crystal size can be controlled by changing the accuracy in screening crystallization conditions, measuring solubil-
oil droplet size, which can be affected by the agitational speed and ity, and measuring the kinetics of nucleation and growth [118–120].
composition. In fact, gelation can completely break off the crystal- It offers the potential to grow large single crystals and to study the
lization process. Yin et al. [114] reported the reason for gelation and mechanism of crystallization, while suffering from limited scale-
evaluated the polarity and Hanson solubility parameters for the up potential for industrial applications. In addition, using micro
gelation process. Bao et al. [91,92] developed a gel-mediated crys- mixing, the impinging jet crystallizer can effectively mix the solu-
tallization process for cefotaxime sodium and valnemulin hydrogen tion and anti-solvent or reactants together, and achieve a uniformly
tartrate that produced the desired crystals. highly supersaturated solution. This method has been shown to be
Even though crystal size and shape can be optimized by the pro- promising for producing small particles with narrow CSD [121,122].
cess control method, some crystal properties are still determined by Recently, Liu et al. [123] presented an impinging jet mixer-batch-
the crystal’s molecular structure. For example, the crystals of ceph- tubular crystallizer for reactive crystallization; this technology can
alosporins cannot usually grow larger than 100 μm. To avoid crystals run continuously and is easy to scale up. To narrow the CSD, micro-
with a small size and with needle- or flake-like shapes, spherical ag- wave-/ultrasound-assisted crystallizers and airlift crystallizers have
glomeration and spherical crystallization were used to successfully been designed to reduce fine crystals [31,90,124–126]. Microwaves
optimize the product properties of cephalosporins. Yang et al. [115] can quickly dissolve fine particles and reduce the heating cycles
realized a spherical growth strategy using gelatin as an induced that are needed to remove fine crystals in batch crystallization
polymer to overcome the disadvantages of the flake-like shape of [90]. Ultrasound can effectively trigger nucleation and narrow the
L-tryptophan. The introduction of an induced polymer and the opti- CSD and metastable zone width [124]. Thus, both microwave- and
mization of concentration and temperature greatly improved the ultrasound-assisted crystallization can effectively reduce batch time
bulk properties of L-tryptophan particles, such as particle size dis- and improve product quality. Instead of using an impeller in con-
tribution, bulk density, and flow property. Choosing process design ventional stirred crystallizers, or moving internal parts in Oslo-type
considerations that were specific to the features of the model com- crystallizers, an airlift crystallizer (Fig. 5(c)) [127] can effectively re-
pounds played an important role in improving the product quality duce crystal collisions using air mixing and can suppress secondary
and efficiency of the process. nucleation [31].
350 Z. Gao et al. / Engineering 3 (2017) 343–353

Compared with batch operation, continuous processes offer high combined Kenics static mixers with a multiple-points anti-solvent
production efficiency and can reduce product variability without strategy to optimize the CSD. Eder et al. [128] developed a Kenics
interruptions. The continuous manufacture of Orkambi (Vertex) and continuously seeded process for the production of APIs in a tubular
the conversion of the Darunavir (Janssen) manufacturing process crystallizer. An idea has been presented to combine an anti-solvent
from batch to continuous were approved by the FDA in July 2015 and tube with a Kenics static mixer in order to promote homogeneous
April 2016, respectively. PFCs facilitate the crystallization process mixing in a double-jacketed tubular crystallizer, as shown in Fig. 6.
along the length of the crystallizer, resulting in a time that is similar Unlike a PFC, an OBC, as shown in Fig. 5(b), uses periodically spaced
to that of a batch process. The back-mixing in a PFC is eliminated restrictions to produce oscillatory mixing. The advantages of the
or minimized, effectively decreasing agglomeration and secondary OBC include enhanced heat and mass transfers, reduced induction
nucleation and leading to a narrower CSD. Because of the lower time and residence time, and narrower metastable zone width and
mixing intensity, methods to induce nucleation must be considered. CSD [29,129]. Commercialized crystallizers from NiTech Solutions
Raphael and Rohani [26] used Kenics static mixers to promote mix- (Scotland) have contributed to propelling forward the application of
ing at the entrance of a tubular crystallizer. Alvarez and Myerson [27] the continuous OBC in industry.

Fig. 5. Schematic drawings of (a) a typical microfluidic crystallizer, (b) a continuous oscillatory baffled crystallizer, and (c) an internal circulation airlift crystallizer. (Parts (a) and (b)
are adapted from Refs. [117,129])

Fig. 6. Schematic diagram of a double-jacket tubular crystallizer with an anti-solvent tube and a Kenics static mixer inside to promote homogeneous mixing. T T: temperature
transmitter.
 Z. Gao et al. / Engineering 3 (2017) 343–353 351

4. Conclusions GA, et al. Microscopic evidence for liquid-liquid separation in supersaturated

CaCO3 solutions. Science 2013;341(6148):885–9.
[10] Kuhs M, Zeglinski J, Rasmuson ÅC. Influence of history of solution in crys-
This paper highlighted recent developments in the crystallization tal nucleation of fenoxycarb: Kinetics and mechanisms. Cryst Growth Des
of pharmaceuticals in industry, with a focus on crystal engineer- 2014;14(3):905–15.
[11] Ito F, Suzuki Y, Fujimori J, Sagawa T, Hara M, Seki T, et al. Direct visualization of
ing and on crystallization process design and control. Advances in
the two-step nucleation model by fluorescence color changes during evapora-
our current understanding of crystal engineering and in the design tive crystallization from solution. Sci Rep 2016;6:22918.
of novel crystallizers and crystallization processes have helped to [12] Srisanga S, Flood AE, Galbraith SC, Rugmai S, Soontaranon S, Ulrich J. Crystal
develop scientific methods for polymorphic screening and product growth rate dispersion versus size-dependent crystal growth: Appropriate
modeling for crystallization processes. Cryst Growth Des 2015;15(5):2330–6.
properties optimization. However, challenges remain: There is a [13] Mascia S, Heider PL, Zhang H, Lakerveld R, Benyahia B, Barton PI, et al. End-to-
need for high-throughput technologies for the in situ screening and end continuous manufacturing of pharmaceuticals: Integrated synthesis, puri-
testing of new crystals and the prediction of crystal structures. Oth- fication, and final dosage formation. Angew Chem Int Ed 2013;52(47):12359–
63.
er challenges involve dealing with oiling out and undesired gelation [14] Myerson AS, Krumme M, Nasr M, Thomas H, Braatz RD. Control systems engi-
phenomena. In crystal engineering, remaining challenges involve un- neering in continuous pharmaceutical manufacturing. May 20–21, 2014 Con-
derstanding and controlling the nucleation of desired forms and pol- tinuous Manufacturing Symposium. J Pharm Sci 2015;104(3):832–9.
[15] Adamo A, Beingessner RL, Behnam M, Chen J, Jamison TF, Jensen KF, et al.
ymorphic transformation through the use of solvents and additives. On-demand continuous-flow production of pharmaceuticals in a compact, re-
Although thousands of studies on co-crystals have been published, configurable system. Science 2016;352(6281):61–7.
further challenges remain in bringing co-crystals to market in a safe [16] Woo XY, Tan RB, Braatz RD. Precise tailoring of the crystal size distribution by
controlled growth and continuous seeding from impinging jet crystallizers.
and well-regulated way. In addition, the robustness of a product’s CrystEngComm 2011;13(6):2006–14.
quality and of the production of co-crystals is a bottleneck in prac- [17] Kee NC, Tan RB, Braatz RD. Selective crystallization of the metastable α-form
tical application. The flexibility and economics of batch crystallizers of L-glutamic acid using concentration feedback control. Cryst Growth Des
2009;9(7):3044–51.
can be enhanced through different control strategies, but continu- [18] Singh MR. Towards the control of crystal shape and morphology distributions
ous crystallizers are preferable due to their higher process efficiency in crystallizers [dissertation]. West Lafayette: Purde University; 2013.
and constant product quality. Differences in lab-scale and industrial- [19] Wang Y, Chen A. Crystallization-based separation of enantiomers. In: An-
drushko V, Andrushko N, editors Stereoselective synthesis of drugs and natural
scale crystallization pose scale-up challenges in areas such as hydro-
products, two volume set. 1st ed. Hoboken: John Wiley & Sons, Inc.; 2013. p.
dynamics, heat and mass transfer performance, and so forth. Scaling 1663–82.
up can lead to changes in nucleation, growth, breakage, and agglom- [20] Nagy ZK, Fevotte G, Kramer H, Simon LL. Recent advances in the monitor-
ing, modelling and control of crystallization systems. Chem Eng Res Des
eration, and will affect crystal qualities. Operational strategies such
2013;91(10):1903–22.
as seeding, use of a cooling rate, use of an anti-solvent addition rate, [21] US Food and Drug Administration. Pharmaceutical cGMPs for the 21st century—
and establishing a mode are vitally important during the scale-up A risk based approach. Final report. US Food and Drug Administration; 2004
process. Opportunities and challenges exist in the novel design of Sep.
[22] Chew W, Sharratt P. Trends in process analytical technology. Anal Methods
crystallization processes and crystallizers, which will help research- 2010;2(10):1412–38.
ers to achieve the objective of precise process control, constant [23] Nagy ZK, Braatz RD. Advances and new directions in crystallization control.
product quality, and robust and efficient process operation. Annu Rev Chem Biomol Eng 2012;3:55–75.
[24] Yang Y, Nagy ZK. Advanced control approaches for combined cooling/antisol-
vent crystallization in continuous mixed suspension mixed product removal
Acknowledgements cascade crystallizers. Chem Eng Sci 2015;127:362–73.
[25] Yang Y, Song L, Nagy ZK. Automated direct nucleation control in continuous
mixed suspension mixed product removal cooling crystallization. Cryst Growth
The authors acknowledge the financial support provided by the Des 2015;15(12):5839–48.
China Scholarship Council and the Natural Science and Engineering [26] Raphael M, Rohani S. Sunflower protein precipitation in a tubular precipitator.
Research Council (NSERC) of Canada. Can J Chem Eng 1999;77(3):540–54.
[27] Alvarez AJ, Myerson AS. Continuous plug flow crystallization of pharmaceutical
compounds. Cryst Growth Des 2010;10(5):2219–28.
Compliance with ethics guidelines [28] Brown CJ, Ni XW. Evaluation of growth kinetics of antisolvent crystallization of
paracetamol in an oscillatory baffled crystallizer utilizing video imaging. Cryst
Growth Des 2011;11(9):3994–4000.
Zhenguo Gao, Sohrab Rohani, Junbo Gong, and Jingkang Wang
[29] McGlone T, Briggs NE, Clark CA, Brown CJ, Sefcik J, Florence AJ. Oscillatory flow
declare that they have no conflict of interest or financial conflicts to reactors (OFRs) for continuous manufacturing and crystallization. Org Process
disclose. Res Dev 2015;19(9):1186–202.
[30] Jiang X, Lu D, Wu X, Ruan X, Fang J, He G. Membrane assisted cooling crystalli-
zation: Process model, nucleation, metastable zone, and crystal size distribu-
References tion. AIChE J 2016;62(3):829–41.
[31] Lakerveld R, van Krochten JJ, Kramer HJ. An air-lift crystallizer can suppress
[1] D’Amato RJ, Loughnan MS, Flynn E, Folkman J. Thalidomide is an inhibitor of secondary nucleation at a higher supersaturation compared to a stirred crystal-
angiogenesis. Proc Natl Acad Sci USA 1994;91(9):4082–5. lizer. Cryst Growth Des 2014;14(7):3264–75.
[2] Wnendt S, Finkam M, Winter W, Ossig J, Raabe G, Zwingenberger K. Enantiose- [32] Liu WJ, Ma CY, Wang XZ. Novel impinging jet and continuous crystallizer
lective inhibition of TNF-α release by thalidomide and thalidomide-analogues. design for rapid reactive crystallization of pharmaceuticals. Procedia Eng
Chirality 1996;8(5):390–6. 2015;102:499–507.
[3] Bauer JF, Saleki-Gerhardt A, Narayanan BA, Chemburkar SR, Patel KM, Spiwek [33] Yazdanpanah N, Ferguson ST, Myerson AS, Trout BL. Novel technique for filtra-
HO, et al., inventors; Abbott Laboratories, assignee. Polymorph of a pharmaceu- tion avoidance in continuous crystallization. Cryst Growth Des 2016;16(1):285–
tical. United States patent US 8193367 B2. 2012 Jun 5. 96.
[4] Bauer J, Spanton S, Henry R, Quick J, Dziki W, Porter W, et al. Ritonavir: [34] Schmidt GM. Photodimerization in the solid state. Pure Appl Chem 1971;27(4):
An extraordinary example of conformational polymorphism. Pharm Res 647–78.
2001;18(6):859–66. [35] Mahata G, Dey S, Chanda J. Crystal engineering: A powerful tool towards de-
[5] Huang LF, Tong WQ. Impact of solid state properties on developability assess- signing pharmaceutical solids with desirable physicochemical properties. Am J
ment of drug candidates. Adv Drug Deliv Rev 2004;56(3):321–34. Drug Dis 2014;1(1):1–9.
[6] Lee AY, Erdemir D, Myerson AS. Crystal polymorphism in chemical process de- [36] Cherukuvada S, Nangia A. Eutectics as improved pharmaceutical materials: De-
velopment. Annu Rev Chem Biomol Eng 2011;2:259–80. sign, properties and characterization. Chem Commun 2014;50(8):906–23.
[7] Sood J, Sapra B, Bhandari S, Jindal M, Tiwary AK. Understanding pharmaceutical [37] Desiraju GR. Crystal engineering: A holistic view. Angew Chem Int Ed
polymorphic transformations I: Influence of process variables and storage con- 2007;46(44):8342–56.
ditions. Ther Deliv 2014;5(10):1123–42. [38] Llinàs A, Goodman JM. Polymorph control: Past, present and future. Drug Dis-
[8] Wu JX, Xia D, van den Berg F, Amigo JM, Rades T, Yang M, et al. A novel image cov Today 2008;13(5–6):198–210.
analysis methodology for online monitoring of nucleation and crystal growth [39] Mirmehrabi M, Rohani S. An approach to solvent screening for crystalli-
during solid state phase transformations. Int J Pharm 2012;433(1–2):60–70. zation of polymorphic pharmaceuticals and fine chemicals. J Pharm Sci
[9] Wallace AF, Hedges LO, Fernandez-Martinez A, Raiteri P, Gale JD, Waychunas 2005;94(7):1560–76.
352 Z. Gao et al. / Engineering 3 (2017) 343–353

[40] Allesø M, van den Berg F, Cornett C, Jørgensen FS, Halling-Sørensen B, availability of apixaban via apixaban-oxalic acid cocrystal. Cryst Growth Des
de Diego HL, et al. Solvent diversity in polymorph screening. J Pharm Sci 2016;16(5):2923–30.
2008;97(6):2145–59. [70] Chattoraj S, Shi L, Chen M, Alhalaweh A, Velaga S, Sun CC. Origin of deteriorated
[41] Pfund LY, Matzger AJ. Towards exhaustive and automated high-throughput crystal plasticity and compaction properties of a 1:1 cocrystal between piroxi-
screening for crystalline polymorphs. ACS Comb Sci 2014;16(7):309–13. cam and saccharin. Cryst Growth Des 2014;14(8):3864–74.
[42] Storey R, Docherty R, Higginson P, Dallman C, Gilmore C, Barr G, et al. Automa- [71] Weyna DR, Cheney ML, Shan N, Hanna M, Zaworotko MJ, Sava V, et al. Improv-
tion of solid form screening procedures in the pharmaceutical industry—How ing solubility and pharmacokinetics of meloxicam via multiple-component
to avoid the bottlenecks. Crystallogr Rev 2004;10(1):45–56. crystal formation. Mol Pharm 2012;9(7):2094–102.
[43] Kralj D, Brečević L, Kontrec J. Vaterite growth and dissolution in aqueous solu- [72] Aakeröy CB, Forbes S, Desper J. Using cocrystals to systematically modulate
tion III. Kinetics of transformation. J Cryst Growth 1997;177(3–4):248–57. aqueous solubility and melting behavior of an anticancer drug. J Am Chem Soc
[44] Sheikholeslamzadeh E, Rohani S. Modeling and optimal control of solution 2009;131(47):17048–9.
mediated polymorphic transformation of L-glutamic acid. Ind Eng Chem Res [73] Steed JW. The role of co-crystals in pharmaceutical design. Trends Pharmacol
2013;52(7):2633–41. Sci 2013;34(3):185–93.
[45] Trifkovic M, Rohani S, Sheikhzadeh M. Kinetics estimation and polymorphic [74] Duggirala NK, Perry ML, Almarsson Ö, Zaworotko MJ. Pharmaceutical cocrys-
transformation modeling of buspirone hydrochloride. J Cryst Process Technol tals: Along the path to improved medicines. Chem Commun 2016;52(4):640–
2012;2(2):31–43. 55.
[46] Simone E, Saleemi AN, Nagy ZK. In situ monitoring of polymorphic transforma- [75] Wu TK, Lin SY, Lin HL, Huang YT. Simultaneous DSC-FTIR microspectroscopy
tions using a composite sensor array of Raman, NIR, and ATR-UV/vis spectros- used to screen and detect the co-crystal formation in real time. Bioorg Med
copy, FBRM, and PVM for an intelligent decision support system. Org Process Chem Lett 2011;21(10):3148–51.
Res Dev 2015;19(1):167–77. [76] Thipparaboina R, Kumar D, Chavan RB, Shastri NR. Multidrug co-crystals: To-
[47] Takeguchi K, Obitsu K, Hirasawa S, Orii R, Ieda S, Okada M, et al. Effect of wards the development of effective therapeutic hybrids. Drug Discov Today
temperature and solvent of solvent-mediated polymorph transformation on 2016;21(3):481–90.
ASP3026 polymorphs and scale-up. Org Process Res Dev 2016;20(5):970–6. [77] Fábián L. Cambridge structural database analysis of molecular complementari-
[48] US Food and Drug Administration. Guidance for industry: ANDAs: Pharmaceu- ty in cocrystals. Cryst Growth Des 2009;9(3):1436–43.
tical solid polymorphism: Chemistry, manufacturing, and controls information. [78] Hilfiker R, editor. Polymorphism: In the pharmaceutical industry. Hoboken:
Silver Spring: Center for Drug Evaluation and Research, US Food and Drug Ad- John Wiley & Sons, Inc.; 2006.
ministration; 2007 Jul. [79] Berziņš A, Skarbulis E, Rekis T, Actiņš A. On the formation of droperidol
[49] Kawabata Y, Wada K, Nakatani M, Yamada S, Onoue S. Formulation design for solvates: Characterization of structure and properties. Cryst Growth Des
poorly water-soluble drugs based on biopharmaceutics classification system: 2014;14(5):2654–64.
Basic approaches and practical applications. Int J Pharm 2011;420(1):1–10. [80] Ulrich J, Frohberg P. Problems, potentials and future of industrial crystalliza-
[50] Lindenberg M, Kopp S, Dressman JB. Classification of orally administered tion. Front Chem Sci Eng 2013;7(1):1–8.
drugs on the World Health Organization Model list of Essential Medicines ac- [81] Ismail SZ, Anderton CL, Copley RC, Price LS, Price SL. Evaluating a crystal energy
cording to the biopharmaceutics classification system. Eur J Pharm Biopharm landscape in the context of industrial polymorph screening. Cryst Growth Des
2004;58(2):265–78. 2013;13(6):2396–406.
[51] Amidon GL, Lennernäs H, Shah VP, Crison JR. A theoretical basis for a biophar- [82] Reilly AM, Cooper RI, Adjiman CS, Bhattacharya S, Boese AD, Brandenburg JG, et
maceutic drug classification: The correlation of in vitro drug product dissolu- al. Report on the sixth blind test of organic crystal structure prediction meth-
tion and in vivo bioavailability. Pharm Res 1995;12(3):413–20. ods. Acta Crystallogr B Struct Sci Cryst Eng Mater 2016;72(Pt 4):439–59.
[52] Lin SY. Molecular perspectives on solid-state phase transformation and chem- [83] Price SL, Braun DE, Reutzel-Edens SM. Can computed crystal energy landscapes
ical reactivity of drugs: Metoclopramide as an example. Drug Discov Today help understand pharmaceutical solids? Chem Commun 2016;52(44):7065–77.
2015;20(2):209–22. [84] Myerson AS, Trout BL. Chemistry. Nucleation from solution. Science 2013;341
[53] Almeida e Sousa L, Reutzel-Edens SM, Stephenson GA, Taylor LS. Assessment of (6148):855–6.
the amorphous “solubility” of a group of diverse drugs using new experimental [85] Dandekar P, Kuvadia ZB, Doherty MF. Engineering crystal morphology. Annu
and theoretical approaches. Mol Pharm 2015;12(2):484–95. Rev Mater Res 2013;43:359–86.
[54] Skrdla PJ, Floyd PD, Dell’orco PC. Practical estimation of amorphous solubility [86] Shtukenberg AG, Lee SS, Kahr B, Ward MD. Manipulating crystallization with
enhancement using thermoanalytical data: Determination of the amorphous/ molecular additives. Annu Rev Chem Biomol Eng 2014;5:77–96.
crystalline solubility ratio for pure indomethacin and felodipine. J Pharm Sci [87] Diao Y, Harada T, Myerson AS, Hatton TA, Trout BL. The role of nanopore shape
2016;105(9):2625–30. in surface-induced crystallization. Nat Mater 2011;10(11):867–71.
[55] Yu L. Amorphous pharmaceutical solids: Preparation, characterization and sta- [88] Diao Y, Myerson AS, Hatton TA, Trout BL. Surface design for controlled crystal-
bilization. Adv Drug Deliv Rev 2001;48(1):27–42. lization: The role of surface chemistry and nanoscale pores in heterogeneous
[56] Dengale SJ, Grohganz H, Rades T, Löbmann K. Recent advances in co-amorphous nucleation. Langmuir 2011;27(9):5324–34.
drug formulations. Adv Drug Deliv Rev 2016;100:116–25. [89] Diao Y, Whaley KE, Helgeson ME, Woldeyes MA, Doyle PS, Myerson AS, et
[57] Löbmann K, Grohganz H, Laitinen R, Strachan C, Rades T. Amino acids as al. Gel-induced selective crystallization of polymorphs. J Am Chem Soc
co-amorphous stabilizers for poorly water soluble drugs—Part 1: Preparation, 2012;134(1):673–84.
stability and dissolution enhancement. Eur J Pharm Biopharm 2013;85(3 Pt [90] Kacker R, Salvador PM, Sturm GS, Stefanidis GD, Lakerveld R, Nagy ZK, et al. Mi-
B):873–81. crowave assisted direct nucleation control for batch crystallization: Crystal size
[58] US Food and Drug Administration. Naming of drug products containing salt control with reduced batch time. Cryst Growth Des 2016;16(1):440–6.
drug substances; guidance for industry; availability. Silver Spring: Center for [91] Ouyang JB, Wang JK, Huang X, Gao Y, Bao Y, Wang YL, et al. Gel formation and
Drug Evaluation and Research, US Food and Drug Administration; 2015 Jun. phase transformation during the crystallization of valnemulin hydrogen tar-
[59] Fernández Casares A, Nap WM, Ten Figás G, Huizenga P, Groot R, Hoffmann trate. Ind Eng Chem Res 2014;53(43):16859–63.
M. An evaluation of salt screening methodologies. J Pharm Pharmacol [92] Gao ZG, Li L, Bao Y, Wang Z, Hao HX, Yin QX, et al. From jellylike phase to crys-
2015;67(6):812–22. tal: Effects of solvent on self-assembly of cefotaxime sodium. Ind Eng Chem
[60] Serajuddin AT. Salt formation to improve drug solubility. Adv Drug Deliv Rev Res 2016;55(11):3075–83.
2007;59(7):603–16. [93] Zhou G, Moment A, Cuff J, Schafer W, Orella C, Sirota E, et al. Process develop-
[61] Saal C, Becker A. Pharmaceutical salts: A summary on doses of salt formers ment and control with recent new FBRM, PVM, and IR. Org Process Res Dev
from the Orange Book. Eur J Pharm Sci 2013;49(4):614–23. 2015;19(1):227–35.
[62] Prohotsky DL, Zhao F. A survey of top 200 drugs—Inconsistent practice of drug [94] Simone E, Saleemi AN, Nagy ZK. Raman, UV, NIR, and Mid-IR spectroscopy with
strength expression for drugs containing salt forms. J Pharm Sci 2012;101(1):1– focused beam reflectance measurement in monitoring polymorphic transfor-
6. mations. Chem Eng Technol 2014;37(8):1305–13.
[63] Thackaberry EA. Non-clinical toxicological considerations for pharmaceutical [95] Simon LL, Pataki H, Marosi G, Meemken F, Hungerbühler K, Baiker A, et al. As-
salt selection. Expert Opin Drug Metab Toxicol 2012;8(11):1419–33. sessment of recent process analytical technology (PAT) trends: A multiauthor
[64] Bolla G, Nangia A. Pharmaceutical cocrystals: Walking the talk. Chem Commun review. Org Process Res Dev 2015;19(1):3–62.
2016;52(54):8342–60. [96] Simon LL, Merz T, Dubuis S, Lieb A, Hungerbuhler K. In-situ monitoring of phar-
[65] Remenar JF, Morissette SL, Peterson ML, Moulton B, MacPhee JM, Guzmán HR, et maceutical and specialty chemicals crystallization processes using endoscopy—
al. Crystal engineering of novel cocrystals of a triazole drug with 1,4-dicarboxylic Stroboscopy and multivariate image analysis. Chem Eng Res Des 2012;90(11):
acids. J Am Chem Soc 2003;125(28):8456–7. 1847–55.
[66] Wang JR, Yu Q, Dai W, Mei X. Drug-drug co-crystallization presents a [97] El Arnaout T, Cullen PJ, Sullivan C. A novel backlight fiber optical probe and
new opportunity for the development of stable vitamins. Chem Commun image algorithms for real time size-shape analysis during crystallization. Chem
2016;52(17):3572–5. Eng Sci 2016;149:42–50.
[67] Aitipamula S, Banerjee R, Bansal AK, Biradha K, Cheney ML, Choudhury AR, [98] Pertig D, Buchfink R, Petersen S, Stelzer T, Ulrich J. Inline analyzing of industrial
et al. Polymorphs, salts, and cocrystals: What’s in a name? Cryst Growth Des crystallization processes by an innovative ultrasonic probe technique. Chem
2012;12(5):2147–52. Eng Technol 2011;34(4):639–46.
[68] US Food and Drug Administration. Guidance for industry: Regulatory classifi- [99] Gherras N, Serris E, Févotte G. Monitoring industrial pharmaceutical crystalliza-
cation of pharmaceutical co-crystals. Silver Spring: Center for Drug Evaluation tion processes using acoustic emission in pure and impure media. Int J Pharm
and Research, US Food and Drug Administration; 2013 Apr. 2012;439(1–2):109–19.
[69] Chen Y, Li L, Yao J, Ma YY, Chen JM, Lu TB. Improving the solubility and bio- [100] Nagy ZK, Chew JW, Fujiwara M, Braatz RD. Comparative performance of
 Z. Gao et al. / Engineering 3 (2017) 343–353 353

concentration and temperature controlled batch crystallizations. J Process Con- Growth Des 2015;15(10):5124–32.
tr 2008;18(3–4):399–407. [116] Paroli F. Industrial crystallizers design and control. In: Chianese A, Kramer HJ,
[101] Duffy D, Barrett M, Glennon B. Novel, calibration-free strategies for super- editors Industrial crystallization process monitoring and control. Weinheim:
saturation control in antisolvent crystallization processes. Cryst Growth Des Wiley-VCH; 2012. p. 203–24.
2013;13(8):3321–32. [117] Sultana M, Jensen KF, inventors; Massachusetts Institute of Technology, assign-
[102] Abu Bakar MR, Nagy ZK, Saleemi AN, Rielly CD. The impact of direct nucleation ee. Systems and methods for microfluidic crystallization. United States patent
control on crystal size distribution in pharmaceutical crystallization processes. US 20100298602 A1. 2010 Nov 25.
Cryst Growth Des 2009;9(3):1378–84. [118] Teychené S, Biscans B. Crystal nucleation in a droplet based microfluidic crys-
[103] Nagy ZK, Fujiwara M, Braatz RD. Modelling and control of combined cooling tallizer. Chem Eng Sci 2012;77:242–8.
and antisolvent crystallization processes. J Process Contr 2008;18(9):856–64. [119] Ildefonso M, Candoni N, Veesler S. A cheap, easy microfluidic crystalliza-
[104] Mesbah A, Landlust J, Huesman AE, Kramer HJ, Jansens PJ, Van den Hof PM. A tion device ensuring universal solvent compatibility. Org Process Res Dev
model-based control framework for industrial batch crystallization processes. 2012;16(4):556–60.
Chem Eng Res Des 2010;88(9):1223–33. [120] Ildefonso M, Revalor E, Punniam P, Salmon JB, Candoni N, Veesler S. Nuclea-
[105] Aamir E, Rielly CD, Nagy ZK. Experimental evaluation of the targeted direct tion and polymorphism explored via an easy-to-use microfluidic tool. J Cryst
design of temperature trajectories for growth-dominated crystallization pro- Growth 2012;342(1):9–12.
cesses using an analytical crystal size distribution estimator. Ind Eng Chem Res [121] Liu WJ, Ma CY, Liu JJ, Zhang Y, Wang XZ. Analytical technology aided optimi-
2012;51(51):16677–87.
zation and scale-up of impinging jet mixer for reactive crystallization process.
[106] Nagy ZK. Model based robust control approach for batch crystallization product
AIChE J 2015;61(2):503–17.
design. Comput Chem Eng 2009;33(10):1685–91.
[122] Woo XY, Tan RB, Braatz RD. Modeling and computational fluid dynamics–popu-
[107] Trifkovic M, Sheikhzadeh M, Rohani S. Kinetics estimation and single and multi-
lation balance equation–micromixing simulation of impinging jet crystallizers.
objective optimization of a seeded, anti-solvent, isothermal batch crystallizer.
Cryst Growth Des 2009;9(1):156–64.
Ind Eng Chem Res 2008;47(5):1586–95.
[123] Liu WJ, Ma CY, Liu JJ, Zhang Y, Wang XZ. Continuous reactive crystallization of
[108] Sheikhzadeh M, Trifkovic M, Rohani S. Real-time optimal control of an anti-
pharmaceuticals using impinging jet mixers. AIChE J 2017;63(3):967−74.
solvent isothermal semi-batch crystallization process. Chem Eng Sci 2008;
[124] Kaur Bhangu S, Ashokkumar M, Lee J. Ultrasound assisted crystallization of
63(3):829–39.
paracetamol: Crystal size distribution and polymorph control. Cryst Growth
[109] Moldoványi N, Lakatos BG, Szeifert F. Model predictive control of MSMPR crys-
tallizers. J Cryst Growth 2005;275(1–2):e1349–54. Des 2016;16(4):1934–41.
[110] Chianese A, Kramer HJ, editors. Industrial crystallization process monitoring [125] Soare A, Lakerveld R, van Royen J, Zocchi G, Stankiewicz AI, Kramer HJ. Mini-
and control. Hoboken: John Wiley & Sons, Inc.; 2012 mization of attrition and breakage in an airlift crystallizer. Ind Eng Chem Res
[111] Lu J, Li YP, Wang J, Ren GB, Rohani S, Ching CB. Crystallization of an active phar- 2012;51(33):10895–909.
maceutical ingredient that oils out. Separ Purif Tech 2012;96:1–6. [126] Leonelli C, Mason TJ. Microwave and ultrasonic processing: Now a realistic op-
[112] De Albuquerque I, Mazzotti M. Crystallization process design using thermody- tion for industry. Chem Eng Process: Process Intensification. 2010;49(9):885–
namics to avoid oiling out in a mixture of vanillin and water. Cryst Growth Des 900.
2014;14(11):5617–25. [127] Soare A, Lakerveld R, van Royen J, Zocchi G, Stankiewicz AI, Kramer HJ. Mini-
[113] Takasuga M, Ooshima H. Control of crystal aspect ratio and size by changing mization of attrition and breakage in an airlift crystallizer. Ind Eng Chem Res
solvent composition in oiling out crystallization of an active pharmaceutical 2012;51(33):10895–909.
ingredient. Cryst Growth Des 2015;15(12):5834–8. [128] Eder RJ, Radl S, Schmitt E, Innerhofer S, Maier M, Gruber-Woelfler H, et al. Con-
[114] Yin YH, Gao ZG, Bao Y, Hou BH, Hao HX, Liu D, et al. Gelation phenomenon tinuously seeded, continuously operated tubular crystallizer for the production
during antisolvent crystallization of cefotaxime sodium. Ind Eng Chem Res of active pharmaceutical ingredients. Cryst Growth Des 2010;10(5):2247–57.
2013;53(3):1286–92. [ 129] Lawton S, Steele G, Shering P. Continuous crystallization of pharmaceuti-
[115] Yang JX, Wang YL, Hao HX, Xie C, Bao Y, Yin QX, et al. Spherulitic crystalliza- cals using a continuous oscillatory baffled crystallizer. Org Process Res Dev
tion of L-tryptophan: Characterization, growth kinetics, and mechanism. Cryst 2009;13(6):1357–63.