Recent progress of continuous crystallization

Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Contents lists available at ScienceDirect

Journal of Industrial and Engineering Chemistry journal homepage: www.elsevier.com/locate/jiec

Review

Recent progress of continuous crystallization Ting Wanga,b , Haijiao Lua,b , Jingkang Wanga,b , Yan Xiaoa,b , Yanan Zhoua,b , Ying Baoa,b , Hongxun Haoa,b,* a National Engineering Research Center of Industry Crystallization Technology, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China b Co-Innovation Center of Chemical Science and Engineering, Tianjin 300072, China

A R T I C L E I N F O

Article history: Received 24 January 2017 Received in revised form 27 March 2017 Accepted 5 June 2017 Available online 15 June 2017 Keywords: Continuous crystallization Control strategies Predictive models Novel crystallizers

A B S T R A C T

Continuous crystallization has always been a hot topic in industrial crystallization. Many efforts have been made to improve the continuous crystallization, either by designing novel continuous crystallizers or by proposing improved design and operation of conventional continuous crystallizers. Some new models for continuous crystallization processes have also been proposed and tested in recent years. In this work, the development of continuous crystallization in recent years, including novel crystallizers, control strategies, models and some assistive technologies, is summarized. Promising as it is, continuous crystallization is still not as universal as batch crystallization due to the existence of the drawbacks, such as blockage and encrustation. Therefore, further efforts are needed before wider application of continuous crystallization. © 2017 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical models for continuous crystallization . . . . . . . . . . . . . . . . . . . . . . . . Population balance equation (PBE) model . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational fluid dynamics (CFD) and discrete element method (DEM) Control strategies of continuous crystallization process . . . . . . . . . . . . . . . . . . . Model-free control strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Model-based control strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hybrid control strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuous crystallizers and their applications . . . . . . . . . . . . . . . . . . . . . . . . . . MSMPR crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-stage MSMPR crystallizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multi-stage MSMPR crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plug-plow crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuous oscillatory baffled crystallizer (COBC) . . . . . . . . . . . . . . . . . . . . Continuous laminar shear crystallizers and continuous Couette–Taylor (CT) Continuous microfluidic crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluidized bed crystallizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Forced circulation crystallizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Draft tube (DT) crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Falling film crystallizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

........... ........... ........... ........... ........... ........... ........... ........... ........... ........... ........... ........... ........... ........... crystallizers ........... ........... ........... ........... ...........

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

15 15 15 16 16 16 17 17 18 18 18 19 20 21 21 22 22 23 23 24

* Corresponding author at: National Engineering Research Center of Industry Crystallization Technology, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China. Fax: +86 22 27374971. E-mail addresses: [email protected], [email protected] (H. Hao). http://dx.doi.org/10.1016/j.jiec.2017.06.009 1226-086X/© 2017 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Applications of new technologies in continuous Process analytical technologies (PAT) . . . . . Ultrasonic technique . . . . . . . . . . . . . . . . . . Coupled with membrane distillation . . . . . Conclusions and future scopes . . . . . . . . . . . . . Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

crystallization ............ ............ ............ ............ ............ ............

. . . . . .

. . . . . . .

. . . . . . .

Introduction Crystallization is one of the most important separation and purification processes in chemical engineering industries, especially in pharmaceutical industries. It is believed that approximately 90% of the active pharmaceutical ingredients (APIs) are organic crystals and crystallization process is one of key unite operations which determine the products’ final qualities [1]. In crystallization industry, batch crystallization has been the most frequently used technique for many years. However, batch crystallization has some drawbacks, such as variation of product quality, low capacity, high requirements of human intervention, high facility cost, etc. [2]. In contrast, continuous process provides many advantages, such as consistence of the product quality, improved yield and capacity, and lower facility cost and space requirement [3,4]. Hence, it is attracting the interests of more and more researchers in recent years, especially in the field of pharmaceutical crystallization. After Randolph and Larson [5] introduced the concept of population balance to crystallization, continuous crystallization process could be more precisely controlled than ever. Though promising, continuous crystallization is still not as universal as batch crystallization in industries. Problems, such as blockage and encrustation, need to be solved by some cost-effective solutions before wider application of continuous crystallization [6]. Besides, mixed suspension and mixed product removal (MSMPR) crystallizer, one of the typical continuous crystallizers, often causes cyclical oscillations in the crystal size distribution (CSD), which is also challenging for continuous crystallization. In this work, the progresseses of continuous crystallization in recent years, especially in pharmaceutical area, are summarized to give a brief review on the state of art and future research directions of continuous crystallization. Theoretical models of continuous crystallization are firstly introduced. Then, continuous crystallizers and their applications are summarized. Furthermore, new emerged assistive technologies for continuous crystallization are also outlined and discussed. Theoretical models for continuous crystallization Population balance equation (PBE) model Population balance equation (PBE) models are the most common models for the simulation of continuous crystallization. If a PBE model is applied to describe the crystal size distribution in a stirred tank, the model can be specially expressed as [7]:

@n @ðGnÞ dðlogV Þ n Q ¼ B  D  Sk k k þ þn dt @t @L V

ð1Þ

where n is the population of crystals, which is relative to time and particle size, G is the linear growth rate of crystal, L is the size of the crystal and V is the volume of the solution. B and D represents the birth and death rates of the crystal due to aggregation and breakage, respectively. Qk is the flow rate of influent and effluent streams.

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

15

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

25 25 26 27 27 27 27

Based on the following assumptions: (1) the grow rate is independent of the crystal size; (2) there are no seeds in feed flow; (3) the whole system maintains a steady state; (4) the solution in the tank is well mixed, which means that, the crystal population and size distribution are the same anywhere inside the tank and the condition of the withdraw flow is the same with the suspension in the tank; (5) the crystal size distribution curve in the system is continuous; (6) aggregation and breakage are ignored; (7) the mean resident time can be described by t = V/Qk, B = D = 0, t = V/Qk,ninlet = 0, and the right-hand side of Eq. (1) can be changed as:

Sk

nk Q k n ¼ V t

ð2Þ

When the system reaches a steady state, n and V would not change with time, then

@n dðlogV Þ ¼0 ¼n dt @t

ð3Þ

Finally, the model can be simplified as: dðGnÞ n þ ¼0 dL t

ð4Þ

If the crystal growth is assumed to be independent of particle size, Eq. (4) can be further simplified into Eq. (5). Gdn n þ ¼0 dL t Therefore, it can be solved by integration as:   L n ¼ n0 exp  Gt

ð5Þ

ð6Þ

Eq. (6) is the final form of the basic PBE model, in which n0 is the population of nucleus. On the basis of Eq. (6), the fluctuation of the CSD in the stirred tank during the process can be monitored and thus operating conditions can be adjusted in time to control the process. Since 1980s, the PBE models have been used for crystallization process modeling [8], such as the simulation of CSD, the nucleation process [9–11], the design of continuous crystallizers [12,13], the optimization of operating conditions [14,15], and the avoidance of unwanted problems such like fouling [16], etc. Because the conditions in the crystallizers are not always as ideal as assumptions, the neglect of some variables, such as breakage, aggregation and growth rate dispersion, might result in inaccuracy of the models. On the other hand, too many variables will lead to a highly-complicated simulation, which is also not desirable. Fortunately, due to the assistance of high-performance computers, new algorithms have been proposed [17–22], and models concerning factors such like aggregation and breakage could be solved effectively [23–26]. Fevotte and Fevotte [27] studied the effects of industrial impurities on the crystallization of citric acid using the PBE model. The impacts of the unsteady-state behavior of the absorption of impurities on the yield and CSD of the product were simulated and the results were used for the design of the model-based control strategies.

16

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Basically, one-dimensional PBE uses one characteristic length to define the crystal size. This would be applicable for spherical or cubic crystals or particles which are of the same shape. But if the shapes of the crystals inside the crystallizer are different, a multidimensional PBE model need to be implemented in order to capture the dynamics of the process [28]. It has been reported that a two-dimensional PBE model was successfully used to simulate the size-independent growth of hydroquinone in MSMPR crystallizer. The simulation results demonstrated the low-frequency oscillatory behaviors in the case of insufficient secondary nucleation [29]. On the basis of the multi-dimensional PBE models, researches on CSD, the agglomeration and breakage [30], and the comparison of accuracy validation of different methods [31] have been reported in recent years.

computational cost would be needed in comparison with any other models [47]. In addition, turbulence model is also a kind of model to simulate the flow field. Crystallizers such like the fluidized bed and baffled stirred bank, where there is always turbulence flow, often need to use turbulence models to simulate the turbulence flow inside crystallizers to get a better understanding of the CSD in crystallizers [48–52]. Meanwhile, there are also reports on simulation of the fouling effects [53] as well as the production of nanocrystals [54]. In particular, a kind of turbulence model called large eddy simulation (LES) turbulence model can help to simulate the melt crystallization very well [55–57].

Computational fluid dynamics (CFD) and discrete element method (DEM)

In a crystallization process, a control system is always essential to maintain the product quality, to suppress the influence of process disturbances and to economically optimize the process performance. Because the main properties of crystals — CSD and crystal shape are the consequence of crystal nucleation and growth, manipulating variables such as temperature profiles, solution concentration and the addition of additives or impurities to control the nucleation and growth of crystals are crucial to the success of a control strategy. In most industrial crystallization processes, feedback controls such like proportional–integral– derivative (PID) control and cascade control are the most common systems to follow simple operating policies mentioned above [58]. Some model predictive control strategies and automatic control systems have also shown promise in recent years. Generally, three kinds of control systems can be applied: model-free control, model-based control and hybrid control.

Computational fluid dynamics can be used to solve and analyze fluid flow problems by numerical analysis and algorithms. It is helpful to obtain solutions for single or multi-phase flow analysis, temperature and chemical composition distribution and so on. For the analysis of crystallization process, CFD is of great importance when the crystallization is at a large scale because the solution in large crystallizers is typically not wellmixed. Conditions such as concentration and temperature might vary a lot with different positions inside the vessel. In this case, PBE alone will not be enough to describe the CSD in the crystallizer [32]. To solve this problem, PBE is often combined with CFD to give more reliable results. There are many reports about the PBE-CFD models, such as the simulation of CSD [33], the scale-up of a crystallizer [34] and the optimization design of the crystallizers [35]. In some studies, CFD-PBE models could be further combined with a third model to achieve more precise analysis of continuous crystallization [36,37]. Compartmental models are one of the effective CFD models to simulate the flow conditions in the crystallizers. In compartmental models, the crystallizer is divided into several parts and each part is simulated separately [38]. Particularly, segregated feed models (SFM) divide the crystallizer into three parts: two feed zones and a bulk zone. SFM has been used for the investigation of semi-batch crystal precipitation [39]. Zauner and Jones [40] used the SFM to predict the effects of mixing on crystallization. It was predicted that it would take about ten resident times to reach the steady state and the overall nucleation rate was dependent on the mixing conditions. What highlighted this study was that the predictions were verified by both continuous and batch crystallization processes of calcium oxalate, making SFM potential for real applications. Discrete element method is another kind of simulation which is common in pharmaceutical industry. In a DEM simulation, Newton’s equations of motion in all three coordinate directions are evaluated for all the particles in the system, ensuring accuracy of particle movement trajectories [41]. In recent years, DEM or the combination of DEM with other models has been widely applied to improve the understanding of pharmaceutical process, such as the particulate processes [42,43], the coating processes [44], crystallization [45] and so on. Grof et al. [46] combined DEM-PBE model with experiments to determine the breakage kernel and the daughter distribution functions (breakage functions) of needleshaped crystals during the crystallization. Quantitative agreement between simulation and experiment was obtained, suggesting that the method was reliable and could significantly reduce the experimental effort. However, it is impractical to use DEM in applications such as sensitivity analysis or optimization which require numerous calls to the process model, because much higher

Control strategies of continuous crystallization process

Model-free control strategies Model-free control strategies are based on the direct use of PAT-based measurements and are widely used because of their simplicity. Consistent properties of the products can be obtained by the application of a model-free feedback controller, and it has been applied in the control of CSD [59], crystal shape [60], polymorph [6], etc. Sen et al. [61] developed an efficient control system for the crystallization of APIs, which was based on a PID controller consisted of both single and cascade control loops. It was demonstrated that a closed-loop control was much better than an open-loop one under the condition of random disturbances. With the rapid development of automatic technology, control systems are moving toward the direction of intelligence and automation in recent years. For example, Powell et al. [62] applied a so-called automated intelligent decision support (IDS) framework to monitor the continuous crystallization and determined the steady state of paracetamol. Yang et al. [63] studied the automated direct nucleation control (ADNC) in their studies, which was a focused beam reflectance measurement (FBRM)-based feedback control approach in the continuous cooling crystallization in a MSMPR crystallizer, including singlestage and multi-stage. This control system could achieve a quick startup and a good control of CSD and provide a highly automated and effective disturbance suppression. Furthermore, a novel control system, named as wet milling-based automated direct nucleation control (WMADNC), has been put forward [64]. This system was implemented upstream to provide closed-loop controlled primary kinetics as well as seed crystals, and downstream to provide closed-loop controlled secondary nucleation kinetics and reduce crystal sizes in a continuous cooling crystallization process. With this close-loop control approach, the process critical quality attributes (CQA) can be well controlled. Moreover, based on the success of pilot-scale tests,

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

automated control strategies have been tested in the industrial continuous crystallization [65]. Model-based control strategies Model-based control strategies have made a significant progress since they were proposed, especially in the last decades [66]. Model-based control strategies can not only give an increased understanding of process control, but also provide a theoretically optimal operating condition, which helps to minimize the demand of experiments [58]. PBE models are commonly used in the design of model-based control strategies. However, sometimes PBEs might not applicable in the practical controllers because of the infinite-dimensional nature of PBE models. Therefore, nonlinear reduced-order-models have been derived using reduced order modeling technique (ROM) and applied in the control of CSD in continuous process [67]. Geyyer et al. [68] tried to stabilize the crystallization process in the presence of uncertainties and external disturbances using control strategy based on reducedorder model. Two configurations of continuous crystallization processes –– with and without fines dissolution loop, were described by a nonlinear model with distributed parameters. And finally, a H1-loop-shaping controller design was performed using reduced-order models. Similarly, Gámez-García et al. [69] designed two practical robust control strategies, modeling error compensation and integral high order sliding model control respectively, to suppress the influence caused by nonlinear oscillation. For large molecules such like proteins, model-based control strategies are also applicable [60,70]. With the assistance of simulation software, more precise results can be obtained [71]. It is worth mentioning that a combination of PBE model with feedback control has been proposed in the process control. Majumder and Nagy [72] used a bivariate PBE model to predict the shape distribution of potassium dihydrogen phosphate in the presence of crystal growth modifiers (CGMs), in which a combination of PBE and crystal impurity model was proposed. Furthermore, a hybrid batch-continuous crystallization control was set up for the process. Su et al. [73] developed a general and rigorous mathematical modeling framework to realize the transformation from batch to continuous process. A C-control which focused on the variation of concentration was applied to ensure that the start-up procedures and on-line control conditions fell within the design-space of the original batch operation, and its ability against uncertainties was demonstrated. Model predictive control (MPC) is a promising control strategy in continuous crystallization [74,75]. It is based on the usage of models, such as PBE, to simulate the variations of the process CQA. PID control framework needs a change of control structure when a relatively large operating region is essential. However, the MPC, which provides fast closed loop response and large attainable region, has a better performance in view of CSD and yield [76]. Yang and Nagy [74] applied a nonlinear model predictive control to a two-stage MSMPR crystallizer based on

17

PBE. This control system considered both yield and CSD as objectives and the process was controlled by adjusting the temperatures and anti-solvent dosing rate in cooling and antisolvent crystallization. MPC has also been successfully applied in plant scale. Mesbah et al. [77] investigated the plant-wide end-toend continuous pharmaceutical manufacturing process with nearly 8000 state variables in the chemical synthesis and the crystallization steps. The closed-loop simulation results indicated that the plant-wide MPC was able to facilitate flexible process operation and could give an effective regulation of quality attributes of tablets. However, it should be noted that the accuracy of the model could have a significant effect on the performance of the MPC. Furthermore, crystal shape (crystal morphology) could remarkably influence the final quality of the product. Changes in crystal shape can be caused by varieties of factors, such as temperature, supersaturation, solvent, additives, impurities and so on [78]. Since the process of forming crystal shape is kinetically controlled rather than thermodynamically controlled, the growth mechanism is the key to shape control [79]. Generally, image analysis and modelbased control are the two main ways to investigate and control the crystal shape. However, image analysis is not suitable for quantitative determination of shape distribution [66]. For model-based control, kinetic Monte Carlo (kMC) models are one of the effective methods to simulate crystal shape. It was found that the kMC simulation can be used to predict the crystal growth dynamics at the operating conditions where experimental data are not available. Because kMC models are not readily available in a closed form, a population balance model is presented and the method of moments is applied to derive a reduced-order ordinary differential equation (ODE) model [70]. Motivated by this, Kwon et al. [60] used a kMC to model the nucleation, growth and dissolution process in a continuous crystallization process with fines trap, where the interplay of inflow/outflow in the crystallizer was included in the mass and energy balance equations. A model predictive controller was designed, which could allow for crystals with desired shape even under undesired effects such as process disturbances and measurement noise, and it has been successfully applied in a plug-flow crystallizer as well [80]. The advantages and disadvantages of model-free controls and model-based controls are compared in Table 1. Hybrid control strategies In consideration of the advantages of both model-free and model-based control approaches, the combination of these two strategies is a new direction which is worth studying in the future. Though this control strategy was not common in continuous crystallization, some researchers have already made a promising start in this area in recent years. For instance, Sen et al. [81] compared the performance of a MPC-PID scheme with a PID-only control scheme and demonstrated that the hybrid control scheme could significantly improve the efficiency of the control system.

Table 1 The comparison between model-free control and model-based control. Control strategies

Advantages

Disadvantages

Model-free control

1. Simple to apply in progress 2. Can avoid the errors related to the choice of a model

1. May need a lot of trials before application in practical 2. Cannot make an adjustment in time if there is a change in the system

Model-based control

1. Can provide optimal operation conditions with fewer experiments for identification 2. Increase the understanding of process

1. May need much time and effort for the model development in some complex situations

18

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Continuous crystallizers and their applications Basically, there are two types of continuous crystallizers: MSMPR crystallizers and the plug-flow (PF) crystallizers. MSMPR crystallizers feature their well internal mixing as well as a relatively long residence time, while PF crystallizers have a much shorter residence time and can produce products with smaller size than MSMPR crystallizers. On the basis of them, some other continuous crystallizers have been developed. Continuous oscillatory baffled crystallizers (COBC) and microfluidic crystallizers are derived from PF crystallizers. Laminar shear crystallizers, Couette–Taylor crystallizers and fluidized bed crystallizers are more or less similar to the MSMPR crystallizers because of the long residence time. Details of continuous crystallizers mentioned above and some others are provided in the following sections. MSMPR crystallizers MSMPR crystallizer is one of the most common continuous crystallizers at present. Products with narrow CSD, high yield and purity, as well as stable polymorph, can be obtained by using MSMPR crystallizers [82–85]. Apart from common crystallization processes, MSMPR crystallizers are also suitable for processes such as continuous co-crystallization [86], chiral separation [87] and wastewater treatment [88,89]. Single-stage MSMPR crystallizer Generally, a single-stage MSMPR crystallizer has only one tank (Fig. 1) and is often used to study nucleation and CSD in continuous crystallization processes [90,91]. Gerard et al. [92] studied the impacts of calcium ions addition on the reaction crystallization of sodium bicarbonate in a single-stage MSMPR crystallizer. The nucleation and growth rates could be determined when the steady state was reached. It was indicated that the calcium based additives had a significant influence on both the morphology and CSD. Peng et al. [93] studied the CSD of calcium sulfate dehydrate crystals in a continuous reactive crystallizer. The process was operated at different temperatures and supersaturations to

Fig. 1. Schematic diagram of a single-stage MSMPR crystallizer. The feed solution continuously enters into the vessel and the crystal slurry is continuously withdrawn from the vessel.

estimate the growth and nucleation kinetics. It was found that the simulated CSD and supersaturation were consistent with experimental results between 25  C and 60  C. In addition, some efforts have been made to improve the performance of single-stage MSMPR crystallizers, such as recycling part for mother liquor [94] to improve the yield and periodically purging nitrogen to avoid fouling [16]. However, common MSMPR crystallizers may not be able to achieve an ideal mixing state in some anti-solvent crystallization and reaction crystallization processes. Therefore, some modifications have been suggested for specific crystallization process. For example, a novel impinging jet (Fig. 2) was specially designed for reaction crystallization process. Two liquid

Fig. 2. 1L novel continuous crystallizer with impinging jet mixer: the schematic diagram (1 — the sodium lactate solution feeding; 2 — the acid cefuroxime solution feeding; 3 — digital stirrer; 4 — the 1L tank reactor with a jacket; 5 — the tubular reactor; 6 — the novel impinging jet mixer) Two liquid streams (1 and 2) in the form of narrow, coplanar jets at high velocities impinge upon each other inside a small mixing zone to achieve a good mixing as well as high supersaturation. Reprinted from Ref. [95] with permission. Copyright© 2016 Elsevier Ltd.

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

19

Fig. 3. Schematic diagram of a multistage MSMPR crystallizer. The former MSMPR crystallizer’s product goes directly into the next MSMPR crystallizer.

streams in the form of narrow, coplanar jets at high velocities impinged upon each other inside a small mixing zone to achieve a good mixing state, which could produce products with uniform CSD and superior crystallinity [95]. Multi-stage MSMPR crystallizers Multi-stage MSMPR crystallizers are a cascade of two or more MSMPR crystallizers, in which the former one’s outlet line connects directly to the next one as a feed line (Fig. 3). In comparison with single-stage MSMPR, multi-stage MSMPR can generate product with narrower CSD, use energy more efficiently, relieve fouling [96], give a greater throughput [97], increase mean crystal size [98,99] and improve crystal purity [100]. Studies of multi-stage MSMPR crystallizers have been reported in recent years [15,101]. Vetter et al. [102] applied the attainable region approach to identify attainable regions in a diagram of mean product particle size vs. total process residence time, in cascades of MSMPR crystallizers, plug-flow crystallizers and batch crystallizers. Three different experiments, including cooling crystallization, anti-solvent crystallization, and the combination of them, were carried out. The boundaries of these attainable regions could be found numerically by solving appropriate optimization problems. It was found that MSMPR crystallizers with low number of cascades could produce crystals with larger particle size. By adjusting operating variables according to proper models, multistage MSMPR crystallizers could produce products with different mean sizes and improve the yield as well as purity [103,104]. Similar to single-stage MSMPR crystallizers, the yield of multistage MSMPR crystallizers can be improved by combining with a

recycle system. A two-stage MSMPR crystallizer with solids recycle was reported to improve the yield of cyclosporine to 79.8%. And the yield could even be increased to 83.9% (nearly the same with that of batch process (86%)) by using a four-stage MSMPR crystallizer [105]. What is more, multi-stage MSMPR crystallizers can be decoupled into different steps during the crystallization processes, resulting in a more flexible control. Motivated by this, Peña and Nagy [106] designed a novel two-stage MSMPR crystallizer for continuous spherical crystallization. The structure of the two-stage MSMPR crystallizer is shown in Fig. 4. The first stage was designated for the nucleation and growth while the second one was designated for agglomeration. Decoupling the nucleation and growth from the agglomeration could help to satisfy the requirements of both manufacturing process and biopharmaceutics and offer more degrees of freedom in the control of each mechanism. Similarly, Zhang et al. [101] applied a two-stage MSMPR crystallizer with cooling process in the first stage and a combination of cooling and anti-solvent in the second stage to improve the properties of the final crystals. Multi-stage MSMPR crystallizers were also applied to so-called preferential crystallization, especially chiral separation. Take Vetter’s research [107] for example, two continuous crystallizers were coupled to exchange their clear liquid phases, each of which was connected to a suspension mill in charge of in situ seed generation through particle breakage. The system was proved to have the ability of recovery from the sudden appearance of undesired “seeds” of the counter-enantiomer in the crystallizers. This type of crystallizers has demonstrated its ability in preferential crystallization of enantiomers and been widely used in recent years [108,109]. The

Fig. 4. Schematic representation of a two-stage MSMPR crystallizer for continuous spherical crystallization. The first one is designated for nucleation and growth while the second one is designated for agglomeration. Reprinted with permission from Ref. [106]. Copyright© 2015 American Chemical Society.

20

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

applications of continuous preferential crystallization to enantiopure chemicals separation have been discussed in detail in a related review [110]. Unfortunately, fouling is also a serious issue in multi-stage MSMPR crystallizers. Moreover, the secondary nucleation, caused by the breakage of large crystals, intensifies the problem. To avoid this, Cui et al.[111] designed a new type of pressure-driven flow crystallizer (PDFC) on the basis of traditional MSMPR design, which could prevent the suspension from passing through pumps or valves. Moreover, it was applicable to a variety of crystallization methods. It should be mentioned that the number of the stages of multi-stage MSMPR crystallizers might be limited considering the space requirements and the cost of the processes. Therefore, it is crucial to improve the efficiency of every stage of multi-stage MSMPR crystallizers. Plug-plow crystallizers Plug-flow (PF) crystallizers are a kind of tube-like crystallizers (see Fig. 5). Studies such as the kinetic identification, experimental validation [112], and in-situ monitoring and characterization of PF crystallizers [113] have been well demonstrated. PF crystallizers with recycle systems have also been introduced [114,115]. It was found that they could effectively improve the yield of the product and be scaled up for industrial applications. In PF crystallizers, as the supersaturation is often achieved in a short time, fine crystals are inevitable due to the intense nucleation in PF crystallizers. To surmount this problem, a slight temperature rise for dissolving fine particles has been suggested. Majumder and Nagy [116] attempted to obtain the optimal temperature profile in their study, in which the crystallizer was consisted of multiple segments so that the temperature of each segment could be controlled individually. The results suggested that the sizedependent growth and dissolution kinetics determined the success of dissolution steps without compromising the final size of the crystalline products. Fouling is prone to happen in plug-plow crystallizers as the particles can attach to the shells of tube and block the tube in some conditions [117]. Majumder and Nagy [118] applied an encrust formation model coupled with the PBE model to predict the encrust thickness, concentration, temperature profiles and the CSD of a plug-flow crystallizer. It turned out that the influence of blockage on product CSD was serious because of the synthetic effects of the reduction of driving force and residence time. A strategy was proposed to solve the problem by the injection of pure solvent. Besides, a heating and cooling cycle has also been put forward to solve the fouling problem without affecting crystal mean size [119]. In recent years, some novel PF crystallizers have been proposed, such as the multi-segmented, multi-addition plug-flow crystallizers (MSMA-PFC) [12,120]. The crystallizer can be divided into many segments by the addition of anti-solvent, so each of them makes an independent crystallization process (Fig. 6). In addition, there is another kind of crystallizer similar to MSMA-PFC called

Fig. 5. Schematic diagram of one type of plug-flow crystallizer. Two feed flows are mixed in the mixer and then crystallization occurs in the tubular section. Reprinted with permission from Ref. [112]. Copyright© 2015 Elsevier Ltd.

slug-flow crystallizer [121–123]. Generally, the solution in a slugflow crystallizer could be segmented by an immiscible fluid. The size and the number of the slugs could be controlled by the amount and the frequency of immiscible fluid added to the crystallizer, and each slug mixes very well. However, this method will lead to an extra separation process. To solve this, Jiang et al. [124] designed a slug-flow crystallizer with liquid and gas introduced into one end of the tube to spontaneously generate slugs of liquid and gas (see Fig. 7). The slugs remained stable during the crystallization process, and it had been demonstrated to be able to produce large and uniform L-asparagine monohydrate crystals in less than 5 min. Moreover, there will be no extra separation process, saving a lot of time. With pure solvent frequently added between the air bubbles in the system, crust can also be relived. This kind of design has also been utilized in the continuous crystallization of proteins, such as the crystallization of enzyme lysozyme [125] and hen-egg-white lysozyme [58].

Fig. 6. Schematic diagram of one type of MSMA-PFC. The crystallizer is divided into many parts, and the anti-solvent is added between the segments. Reprinted with permission from Ref. [12]. Copyright© 2014 American Chemical Society.

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

21

Fig. 7. Schematic diagram of slug-flow cooling crystallization. The liquid (both hot and cold) and gas are introduced into one end of the tube to spontaneously generate slugs of liquid and gas, which could remain stable during the process. Reprinted with permission from Ref. [124]. Copyright© 2014 American Chemical Society.

Continuous oscillatory baffled crystallizer (COBC) Continuous oscillatory baffled crystallizer is a relatively new type of continuous crystallizer. A COBC consists of a series of tubes, with some baffles periodically arranged inside (Fig. 8). The vibration of the baffles or the movement of a piston or bellows produces repeating cycles of vortices, which can create strong radial motions, giving more uniform mixing along the column than PF crystallizers which depend on the net flow to mix the solution [2,126]. In recent years, significant progress has been made on the nucleation mechanisms in COBC [127,128]. Brown et al. [129] used COBC to investigate the kinetics of the crystallization of adipic acid. Their results showed that PBE-based method could maintain a minimal deviation (1.2%) and the other two methods gave increasing deviations with faster cooling rates (19.9% and 18.6% for average absolute relative error, respectively). Callahan and Ni [130] studied the causes of different nucleation behaviors of

Fig. 8. Schematic diagram of one type of continuous oscillatory baffled crystallizer. Series of plug-flow crystallizers are connected head-to-tail, with baffles inside the tubes to generate the vibration. Reprinted with permission from Ref. [133]. Copyright© 2014, Royal Society of Chemistry.

cooling crystallization in batch crystallizers and COBC under the same experimental conditions. The mixing mechanism was found to be the key factor, confirming the important influence of fluid mechanical environment on the nucleation and growth. Besides, some other critical conditions such as seeding [131] and impurity [132], have also been studied. It is worth to mention that Zhao et al. [133] have successfully used COBC in the design of a novel co-crystal and the scaling up of the co-crystallization process. First, a small-scale experiment was conducted, in which a proper 1:1 co-crystal of a-lipoic acid with nicotinamide was successfully screened. Then, the process was scaled up in a COBC and products with a purity of 99% was then obtained. Continuous laminar shear crystallizers and continuous Couette–Taylor (CT) crystallizers Continuous laminar shear crystallizers consist of four main sections [134]: feed unit, shearing mechanism, cooling system and power unit (Fig. 9). Two concentric cylinders compose the shearing mechanism, of which the inner cylinder is stationary with a cooling system and the outer cylinder rotates to produce a shear force. The gap between the two cylinders is designed for the entrance of feed materials. Many studies on the characters of nanostructure of fats have been carried out in continuous laminar shear crystallizers in recent years, such as the thermal and mechanical properties [135], the oil migration [136], the crystalline alignments [137], the polymorph [134] and so on. As one of the unique parameters of the continuous laminar shear crystallizers, laminar shear rate has an apparent effect on the properties of crystals. With the shear rate increasing, thicker crystals would be obtained with more evident crystalline orientation [138]. However, reports about the application and the scaling up of laminar shear crystallizers are still rare. Continuous Couette–Taylor (CT) crystallizers are similar to the continuous laminar shear crystallizers. CT crystallizers (see Fig. 10) also have two concentric cylinders, but the inner one can rotate freely. When the inner cylinder rotates at a low rate, a laminar Couette fluid motion is induced in the coaxial cylinders. Then, with the rotation speed increasing beyond a certain value which is called the critical Taylor number, this flow changes into radial vortex combined with a small axial dispersion, which is called as Couette–Taylor vortex [139]. Due to this unique flow, this vortex will have a significant effect on the crystallization processes, and the hydrodynamic flow regime can be simply and accurately adjusted from a laminar flow to a turbulent Taylor vortex just by changing the rotation speed of the inner cylinder [140]. As Couette–Taylor vortex is still a relatively new concept, researches on CT crystallizers have been limited in polymorph

22

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Fig. 9. (a) The detailed schematic drawing, showing the inner and outer tubes, water jackets with cooling water inlets and outlets. The inner cylinder is stationary with a cooling system and the outer cylinder rotates to produce a shear force, and solution flows and crystalizes between the cylinders. (b) The manufactured continuous laminar shear crystallizer, connected to the pump, the electromotor, and the controller. Reprinted with permission from Ref. [134]. Copyright© 2008 Elsevier Ltd.

transition [141], agglomeration [142,143], inhibition of flocculation [144] and CSD control [145]. Especially, Nguyen et al. have done a series of studies in CT crystallizers using guanosine 5-monophosphate (GMP) as model API, including the phase transformation [139], kinetic model [146], and optimization on structure of CT crystallizers [147]. In their studies, a multiple feed mode was applied and was found to be able to successfully promote the phase transformation of GMP. Another novel modified CT crystallizer has also been proposed [145]. In this crystallizer, the inner concentric cylinder was hotter than the outer one. This heating-cooling cycle gave a large crystal size and narrow CSD by dissolving the fines in the heat side and recrystallization in the cold side.

microfluidic crystallizers can be combined with ultrasound [155] and achieve automated screening in nanoliter-scale crystallization to an extent [156]. As most of microfluidic crystallizers’ tubes are made of poly (dimethylsiloxane), they are only applicable for aqueous system. To solve this problem, a PEEK/Teflon microfluidic device has been proposed to study the nucleation in both aqueous and organic solvents [157]. However, microfluidic crystallizers may not be suitable for systems in which the solvent concentration is too high, because too many crystals will block the thin tube during the crystallization process. State of the art of continuous microfluidic crystallization system has been concluded in literature [158].

Continuous microfluidic crystallizers

Fluidized bed crystallizer

In recent years, crystallization in microfluidic system is becoming more and more popular in continuous crystallization due to its low consumption and tight control of supersaturation (see Fig. 11). Moreover, just like the slug-flow crystallizers, if droplets are segregated by immiscible fluid or gas, each one of them can be regarded as an independent crystallizer with volume in small range. [148,149]. These characteristics demonstrate its great potential for systematic study on fundamental crystallization process and fast screening of crystallization conditions [150]. Many studies on microfluidic crystallizers have been reported, including the nucleation process [151,152], the spherical crystallization from water-in-oil emulsions [153] and so on. A microfluidic continuous seeded crystallizer can act as a screening platform for process parameter evaluation and process optimization. And it was demonstrated that it could be used to study the growth kinetics for three different forms of glycine [154]. In addition, the

Fluidized beds (see Fig. 12) are popular devices which have been applied in numerous fields, such as synthesis, drying, granulation and so on [159–162]. There have been many reports about their applications in recovering useful substances from wastewater [163–167]. In recent years, studies have also been focused on the establishment of models [168,169] and controlling of CSD of products [170]. Fluidized beds are commonly used in preferential crystallization. For instance, Binev [171] coupled two fluidized beds to study the continuous preferential crystallization to separate isomers in two different systems. It was found that the CSD of the products could reach a steady state with products’ purities higher than 97%. Though rare, there are still some studies about the scale-up of fluidized bed crystallizers [172,173], often assisted by the simulation of CFD. Besides, studies on combining the crystallization with drying in fluidized bed crystallizers have also been reported [174,175].

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

23

Fig. 10. Schematic diagram of continuous drowning-out crystallization of guanosine 5-monophosphate (GMP) using Couette–Taylor Crystallizer. (a) Taylor vortices in Couette–Taylor crystallizer (1, stationary outer cylinder; 2, Taylor vortices; 3, rotating inner cylinder) and (b) experimental apparatus of multiple feeding mode system for drowning-out crystallization (1, motor; 2, pump; 3, methanol; 4, feed GMP solution; 5, damper). The inner cylinder rotates at a high speed to generate the Couette–Taylor vortex. Reprinted with permission from Ref. [139]. Copyright© 2010, American Chemical Society.

Forced circulation crystallizer Forced circulation crystallizers are commercially available in continuous crystallization industry. The proposal of them could date back to 1970s in a patent [176]. Fig. 13 shows the basic structure of forced circulation crystallizers, which comprises three parts: a chamber intended to contain the solution, a circulating pump to force the mother liquor back to the chamber, and a heater. Forced circulation crystallizers have the advantages of flexible design, stable operation and high flexibility [177], making them favorable in continuous crystallization industry. In recent years, the most common application of forced circulation crystallizers may be water treatment, especially water desalination. Basically, the main function of forced circulation crystallizer is to produce salt and distilled water from the waste water [178]. There have been some reports concerning the modeling of forced circulation crystallizers to help to optimize operating conditions in water desalination [179,180]. In addition, they have also been applied in the basal aquifer water treatment in pilot scale [181]. However, to our best knowledge, there are barely reports about the application of forced circulation crystallizers in other fields except one concerning the continuous crystallization of lactose monohydrate [182] for the last decade. In recent years, some modifications of the design of forced circulation crystallizers

have been proposed, such as a two concentrically arranged chambers to give a more flexible control in CSD [183] and a design of large deposit accumulation volume located at the bottom of the crystallizer to relieve the fouling [184]. Draft tube (DT) crystallizers DT crystallizers have been reported decades of years ago [186]. As shown in Fig. 14, it is a tank with a draft tube connected to the bottom. This design gives a specific flow field in the crystallizer, which has an effect on the properties of the product [187]. Some DT crystallizers also have baffles to make a better flow field [188–190]. It is said that a DT crystallizer requires a much lower agitation speed to achieve the inner circulation, so the power consumption can be reduced [191]. In DT crystallizers, the nuclei are mainly produced due to the collisions between crystals and the pump or impeller, and reports are always focus on the variation of the kinetics of nucleation and growth of the crystals [192,193]. As the hydrodynamics is very sensitive to the structure of the crystallizers, changes in such as the bottom shape [194] as well as the impeller shape [35] will greatly affect the properties of products. DT crystallizers usually use impellers to achieve the uniform mixing of solution. However, the movement of impellers will cause mechanical contacts with crystals, leading to attrition and

24

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Fig. 11. Schematic diagram of a kind of microfluidic crystallizer: (a) photograph of microfluidic chip and housing; (b) sketch of microfluidic chip. An injector pump push small amount of solution into the microtubes slowly, in which crystals will be generated from drops of solution. Reprinted with permission from Ref. [159]. Copyright© 2011, American Chemical Society.

breakage. To relieve attrition and breakage, airlift crystallizers are proposed on the base of DT crystallizers (see Fig. 15). Air is purged from the bottom to act as the function of an impeller. The shear force is relatively low so that fewer crystals will get broken [195]. Though promising, airlift crystallizers are still not frequently reported, with only a few reported in recent years [196,197]. Falling film crystallizers Falling film crystallizers are popular in continuous melt crystallization in recent years. A falling film melt crystallization process is very attractive for its enhancement in mass transfer and separation effect [198]. The structure of a falling film melt crystallizer is illustrated by Fig. 16. Two tubes, one inside the other, are positioned in a tank with jacket. There are two cooling lines in the crystallizer: one passing through the jacket and the other one going through the space between the two tubes from top. Melt flows along the surface of the external tube and then is cooled down to crystallize. Jiang et al. have carried out a series of studies

Fig. 12. Schematic diagram of a continuous fluidized bed crystallizer. (A) Heat exchanger and (B) fluidized bed. Solution is added into the fluidized bed continuously.

Fig. 13. Schematic diagram of a forced circulation crystallizer. The pump was used to transport the mother liquor back to the tank again, with the heat exchanger adjusting the supersaturation. Reprinted from Ref. [185] with permission. Copyright© 2000 Elsevier Ltd.

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

25

Fig. 14. Schematic diagram of a DT crystallizer. The draft tube in the tank gives a different fluid field within the tank crystallizers. Reprinted from Ref. [187] with permission. Copyright© 2012 Elsevier Ltd.

on falling film melt crystallizers, including the model development, the simulation of dynamic sweating and the process optimization [199,200]. Falling film crystallizers have been applied in industrial purification and crystallization of many products [201–203]. Applications of new technologies in continuous crystallization In recent years, some assistive techniques, such as the process analytical technologies (PAT), ultrasonic technique, have been applied to help control or manipulate the continuous crystallization processes. Here, some representative technologies are introduced. Process analytical technologies (PAT) PAT tools have been used to investigate crystallization process since they first emerged [204]. Continuous crystallization processes benefit a lot from PAT tools as they can give a real-time feedback of the variations of the parameters to manage a robust control during the whole process. In recent years, more emphases have been put on this field due to the spread of PAT, aiming at a better understanding of crystallization process as well as a more robust control [205,206]. Basically, PAT tools are used to monitor the concentration and the CSD of the system, facilitate the control of the process and help to reveal the mechanisms of nucleation and growth [207]. Apart from conventional solvent-based continuous crystallization, PAT tools are also useful in solvent-free continuous co-crystallization [208]. The crystallization process is often monitored by near infrared reflectance at different stages to detect the formation of co-crystals. Moradiya [209] designed a solvent-free continuous cocrystallization process, in which PAT tools such as near infrared reflectance and spatial filter velocimetry probes were used for inline process control. In addition, some new operation strategies of the continuous crystallizers have also been proposed with the assistance of PAT [210].

Fig. 15. Schematic diagram of an 18-L air-lift crystallizer. TT = temperature transmitter, dPT = differential pressure transmitter. Reprinted from Ref. [197] with permission. Copyright© 2014 American Chemical Society.

However, it is common that more than a single PAT tool is applied in pharmaceutical industry to monitor physical and chemical phenomena during the processes, resulting in data with high dimensionality. Therefore, multivariate data analysis is essential for a better control of the multivariate relationships between these variables [211], and it has been applied in the polymorph control with Raman [212] and infrared spectroscopy [213]. Ferguson et al.[214] compared the CSD of benzoic acid from plug flow, MSMPR and the equivalent fed batch crystallizers respectively, with the help of FBRM and attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR). It turned out that the PAT tools could be used to monitor and characterize the crystallizations processes, indicating their potential in reducing process development time and improving process understanding.

26

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

the original 3D of crystal shapes from the measured 2D projections of potassium dihydrogen phosphate (KDP) crystals, and the results showed good agreement with the experimental values. However, the technique required projection to be bounded by convex polygon with six or eight vertices. Therefore, further studies need to be carried out to improve the applicability of this technology. Besides, image-based analysis can also be used to analyze the CSD [216–218], monitor fouling [219] and evaluate the growth rate [220]. Since the issuance of the FDA guidance of 2004, the Food and Drug Administration (FDA) has been encouraging the application of PAT tools. In this file, PAT is considered as “a system for designing, analyzing, and controlling manufacturing through timely measurements of critical quality and performance attributes of raw and in-process materials and processes to ensure final product quality” [221]. Appropriate use of PAT tools will be of great help in process control and optimization, and a combination of different PAT tools may be more useful for ensuring the products’ quality. Furthermore, FDA also emphasizes the necessity of risk analysis and multivariate methodologies. Therefore, a PAT method, which considers a scientific, risk-based approach, is recommended. The emergence and application of PAT has promoted the understanding of crystallization process and the improvement of product quality. However, the application of PAT tools is still limited because of the high cost [220]. More efforts are needed to improve the accuracy of PAT tools and reduce cost. Ultrasonic technique

Fig. 16. Schematic diagram of a falling film melt crystallizer. Melt flows along the external tube and then crystallizes. Reprinted from Ref. [201] with permission. Copyright© 2013 Elsevier Ltd.

As mentioned above, image-based analysis facilities are effective tools for analyzing shape evolution of the crystals during the growth. Borne et al. [215] presented a technique to reconstruct

Currently, ultrasound is mainly used to control the CSD of products, both in the traditional MSMPR crystallizers [222] and in the novel slug-flow crystallizers [223]. Jiang et al. [224] also designed indirect ultrasonication-assisted nucleation process to study the seed generation rate. However, the cost of ultrasound is relatively high and its potential effects on other properties such as morphology and polymorph transformation are still not well known in continuous crystallization process.

Fig. 17. Schematic diagram of continuous membrane crystallizer. The solution is distillated and then crystallized in A and the condensate will flow to D. (A) Membrane crystallizer, (B) pump, (C) feed tank, and (D) permeate liquid tank.

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

Coupled with membrane distillation Membrane distillation (MD) is a thermally driven separation process in which the liquid evaporates and the vapor passes through the hydrophobic membrane to achieve the separation (see Fig. 17). Coupling MD with crystallization contributes to the generation of supersaturation in a very short time. Meanwhile, the solvent can be recycled by cooling down the vapor. Membrane distillation crystallization has been applied in the desalination of sea water [225]. In recent years, membrane distillation has also been coupled with continuous crystallization. Chen et al. [226] studied the continuous membrane distillation crystallization process of a high concentration feed solution. The optimal operating parameters were determined by an orthogonal fractional factorial experiment design. The method could produce high quality water and obtain sodium chloride crystals with a relatively narrow CSD. Though continuous membrane distillation crystallization has been successfully applied in the recycle of concentrated salt solution, to our best knowledge, no studies on their application in continuous crystallization of API have been reported up to now. However, it is worth mentioning that this technique may serve as an alternative way for conventional continuous crystallization of some substance whose solubility is not sensitive to temperature or dilute feed solution. Conclusions and future scopes In this paper, continuous crystallization, including the categories of continuous crystallizers and their applications, the control strategies and their related models, and some assistive technologies for continuous crystallization, are summarized. Through decades of years of studies, significant progress, such as better control of CSD and larger yield, polymorph and morphology, has been achieved in the field of continuous crystallization. Some novel continuous crystallizers have been proposed and successfully applied in the crystallization of many compounds. Unfortunately, for some industries, such as pharmaceutical industry, continuous crystallization is still rarely applied although many famous pharmaceuticals companies such as Novartis [227] have shown great interest in continuous crystallization and have made significant progress. To promote a wider application of continuous crystallization in pharmaceutical industry, more significant efforts should be made in various of aspects. Specially, adaptability to smaller product volumes, better compatibility with other process equipment, better design to prevent blockage of transfer lines and addressing regulatory issues (especially for pharmaceutical products) regarding switch from batch to continuous processing, etc., are crucial factors worthy studying in future research [228]. Although continuous crystallization is not so widespread as batch processes at present, it is believed that progress on continuous crystallization will surely extend its application to a great extent due to the apparent advantages of continuous process. Acknowledgement This research is financially supported by National Key Research and Development Program of China (No. 2016YFB0600504). References [1] A.J. Alvarez, A.S. Myerson, Cryst. Growth Des. 10 (2010) 2219. [2] S. Lawton, G. Steele, P. Shering, L. Zhao, I. Laird, X. Ni, Org. Process Res. Dev. 13 (2009) 1357. [3] S.D. Schaber, D.I. Gerogiorgis, R. Ramachandran, J.M. Evans, P.I. Barton, B.L. Trout, Ind. Eng. Chem. Res. 50 (2011) 10083. [4] I.R. Baxendale, R.D. Braatz, B.K. Hodnett, K.F. Jensen, M.D. Johnson, P. Sharratt, J.P. Sherlock, A.J. Florence, J. Pharm. Sci. U. S. 104 (2015) 781.

27

[5] M.A. Larson, A.D. Randolph, Theory of Particulate Processes: Analysis and Techniques of Continuous Crystallization, Academic Press, New York, 1971. [6] K.A. Powell, G. Bartolini, K.E. Wittering, A.N. Saleemi, C.C. Wilson, C.D. Rielly, Z.K. Nagy, Cryst. Growth Des. 15 (2015) 4821. [7] P. Marchal, R. David, J.P. Klein, J. Villermaux, Chem. Eng. Sci. 43 (1988) 59. [8] R. Franck, R. David, J. Villermaux, J.P. Klein, Chem. Eng. Sci. 43 (1988) 69. [9] Y. Cui, J.J. Jaramillo, T. Stelzer, A.S. Myerson, Org. Process Res. Dev. 19 (2014) 1101. [10] R.J. Eder, S. Radl, E. Schmitt, S. Innerhofer, M. Maier, H. Gruber-Woelfler, J.G. Khinast, Cryst. Growth Des. 10 (2010) 2247. [11] M.O. Besenhard, R. Hohl, A. Hodzic, R. Eder, J.G. Khinast, Cryst. Res. Technol. 49 (2014) 92. [12] B.J. Ridder, A. Majumder, Z.K. Nagy, Ind. Eng. Chem. Res. 53 (2014) 4387. [13] Q. Su, C.D. Rielly, Z.K. Nagy, American Control Conference (ACC) (2015) 4276. [14] M.L. Rasche, M. Jiang, R.D. Braatz, Comput. Chem. Eng. 95 (2016) 240. [15] A.J. Alvarez, A. Singh, A.S. Myerson, Cryst. Growth Des. 11 (2011) 4392. [16] G. Hou, G. Power, M. Barrett, B. Glennon, G. Morris, Y. Zhao, Cryst. Growth Des. 14 (2014) 1782. [17] A. Majumder, V. Kariwala, S. Ansumali, A. Rajendran, Chem. Eng. Sci. 70 (2012) 121. [18] A. Majumder, V. Kariwala, S. Ansumali, A. Rajendran, Chem. Eng. Sci. 69 (2012) 316. [19] A. Majumder, V. Kariwala, S. Ansumali, A. Rajendran, Chem. Eng. Sci. 65 (2010) 3928. [20] B. Szilágyi, Z.K. Nagy, Comput. Chem. Eng. 91 (2016) 167. [21] R. Gunawan, I. Fusman, R.D. Braatz, AIChE J. 50 (2004) 2738. [22] S. Qamar, A. Ashfaq, G. Warnecke, I. Angelov, M.P. Elsner, A. SeidelMorgenstern, Comput. Chem. Eng. 31 (2007) 1296. [23] J. Cheng, C. Yang, Z. Mao, Chem. Eng. Sci. 68 (2012) 469. [24] C.B.B. Costa, M.R.W. Maciel, R. Maciel Filho, Comput. Chem. Eng. 31 (2007) 206. [25] S.M. Nowee, A. Abbas, J.A. Romagnoli, Chem. Eng. Process. Process Intensif. 46 (2007) 1096. [26] R. Ramachandran, C.D. Immanuel, F. Stepanek, J.D. Litster, F.J. Doyle, Chem. Eng. Res. Des. 87 (2009) 598. [27] F. Févotte, G. Févotte, Chem. Eng. Sci. 65 (2010) 3191. [28] M. Sen, A. Chaudhury, R. Singh, R. Ramachandran, Am. J. Mod. Chem. Eng. 1 (2014) 13. [29] F. Puel, G. Févotte, J.P. Klein, Chem. Eng. Sci. 58 (2003) 3715. [30] Y. Qian, G. Lu, Y. Sun, X. Song, J. Yu, CrystEngComm 17 (2015) 9394. [31] R. Ramachandran, P.I. Barton, Chem. Eng. Sci. 65 (2010) 4884. [32] W. Wantha, A.E. Flood, Chem. Eng. Commun. 195 (2008) 1345. [33] C. Shu, X.K. Wang, Z.L. Sha, Adv. Mater. Res. (2011) 1177. [34] H. Wei, Chem. Eng. Res. Des. 88 (2010) 1377. [35] X. Song, M. Zhang, J. Wang, P. Li, J. Yu, Ind. Eng. Chem. Res. 49 (2010) 10297. [36] X.Y. Woo, R.B. Tan, R.D. Braatz, Cryst. Growth Des. 9 (2008) 156. [37] X.Y. Woo, R.B. Tan, P.S. Chow, R.D. Braatz, Cryst. Growth Des. 6 (2006) 1291. [38] M. Liiri, H. Hatakka, J. Kallas, J. Aittamaa, V. Alopaeus, Chem. Eng. Res. Des. 88 (2010) 1297. [39] M. Ståhl, Å.C. Rasmuson, Chem. Eng. Sci. 64 (2009) 1559. [40] R. Zauner, A.G. Jones, Chem. Eng. Sci. 57 (2002) 821. [41] J. Rantanen, J. Khinast, J. Pharm. Sci. U. S. 104 (2015) 3612. [42] B. Remy, T.M. Canty, J.G. Khinast, B.J. Glasser, Chem. Eng. Sci. 65 (2010) 4557. [43] D. Barrasso, A. Tamrakar, R. Ramachandran, Chem. Eng. Sci. 119 (2014) 319. [44] M. Kodam, J. Curtis, B. Hancock, C. Wassgren, Chem. Eng. Sci. 69 (2012) 587. [45] B. Ashraf Ali, M. Börner, M. Peglow, G. Janiga, A. Seidel-Morgenstern, D. The’venin, Cryst. Growth Des. 15 (2014) 145. pánek, Int. J. Pharm. 407 (2011) [46] Z. Grof, C.M. Schoellhammer, P. Rajniak, F. Šte 12. [47] A. Rogers, M. Ierapetritou, Comput. Chem. Eng. 81 (2015) 32. [48] G.R. Kasat, A.R. Khopkar, V.V. Ranade, A.B. Pandit, Chem. Eng. Sci. 63 (2008) 3877. [49] M.H. Vakili, M.N. Esfahany, Chem. Eng. Sci. 64 (2009) 351. [50] H. Singh, D.F. Fletcher, J.J. Nijdam, Chem. Eng. Sci. 66 (2011) 5976. [51] J. Cheng, C. Yang, Z. Mao, C. Zhao, Ind. Eng. Chem. Res. 48 (2009) 6992. [52] D.O. Grady, M. Barrett, E. Casey, B. Glennon, Chem. Eng. Res. Des. 85 (2007) 945. [53] F. Albert, W. Augustin, S. Scholl, Chem. Eng. Sci. 66 (2011) 499. [54] L. Metzger, M. Kind, The influence of mixing on fast precipitation processes-A coupled CFD-PBE approach using the Direct Quadrature Method of Moment (DQMOM), Proc. 11th International Conference on CFD in the Minerals and Process Industries (2015). [55] L. Liu, X. Liu, Y. Wang, Int. J. Heat Mass Transf. 55 (2012) 53. [56] X. Liu, L. Liu, Z. Li, Y. Wang, J. Cryst. Growth 360 (2012) 38. [57] A. Krauze, N. Jekabsons, A. Muižnieks, A. Sabanskis, U. Lacis, J. Cryst. Growth 312 (2010) 3225. [58] Z.K. Nagy, G. Fevotte, H. Kramer, L.L. Simon, Chem. Eng. Res. Des. 91 (2013) 1903. [59] M.O. Besenhard, P. Neugebauer, C. Ho, J.G. Khinast, Cryst. Growth Des. 15 (2015) 1683. [60] J.S. Kwon, M. Nayhouse, P.D. Christofides, G. Orkoulas, Chem. Eng. Sci. 107 (2014) 47. [61] M. Sen, R. Singh, R. Ramachandran, J. Pharm. Innov. 9 (2014) 65. [62] K.A. Powell, A.N. Saleemi, C.D. Rielly, Z.K. Nagy, Org. Process Res. Dev. 20 (2016) 626. [63] Y. Yang, L. Song, Z.K. Nagy, Cryst. Growth Des. 15 (2015) 5839.

28

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29

[64] Y. Yang, L. Song, Y. Zhang, Z.K. Nagy, Ind. Eng. Chem. Res. 55 (2016) 4987. [65] R. Lakerveld, B. Benyahia, P.L. Heider, H. Zhang, A. Wolfe, C.J. Testa, S. Ogden, D.R. Hersey, S. Mascia, J.M. Evans, Org. Process Res. Dev. 19 (2014) 1088. [66] Z.K. Nagy, R.D. Braatz, Annu. Rev. Chem. Biomol. 3 (2012) 55. [67] T. Chiu, P.D. Christofides, AIChE J. 45 (1999) 1279. [68] R. Geyyer, A. Kienle, S. Palis, IFAC-PapersOnLine 48 (2015) 598. [69] V. Gámez-García, E. Bolaños-Reynoso, O. Velazquez-Camilo, H. Puebla, J. Math. Syst. Sci. 2 (2012). [70] J.S. Kwon, M. Nayhouse, G. Orkoulas, P.D. Christofides, Ind. Eng. Chem. Res. 53 (2014) 15538. [71] N. Moldoványi, J. Abonyi, Chem. Biochem. Eng. Q. 23 (2009) 195. [72] A. Majumder, Z.K. Nagy, Chem. Eng. Sci. 101 (2013) 593. [73] Q. Su, Z.K. Nagy, C.D. Rielly, Chem. Eng. Process. Process Intensif. 89 (2015) 41. [74] Y. Yang, Z.K. Nagy, American Control Conference (ACC) (2015) 4282. [75] Y. Yang, Z.K. Nagy, Chem. Eng. Sci. 127 (2015) 362. [76] C. Damour, M. Benne, B. Grondin-Perez, J. Chabriat, J. Food Eng. 99 (2010) 225. [77] A. Mesbah, J.A. Paulson, R. Lakerveld, R.D. Braatz, American Control Conference (ACC) (2015) 4301. [78] J. Kundin, C. Yürüdü, J. Ulrich, H. Emmerich, Eur. Phys. J. B Condens. Matter Complex Syst. 70 (2009) 403. [79] C. Borchert, K. Sundmacher, Chem. Eng. Sci. 84 (2012) 85. [80] J.S. Kwon, M. Nayhouse, G. Orkoulas, P.D. Christofides, Chem. Eng. Sci. 119 (2014) 30. [81] M. Sen, R. Singh, R. Ramachandran, Processes 2 (2014) 392. [82] T.C. Lai, J. Cornevin, S. Ferguson, N. Li, B.L. Trout, A.S. Myerson, Cryst. Growth Des. 15 (2015) 3374. [83] M. Schoenitz, S. Joseph, A. Nitz, H. Bunjes, S. Scholl, Eur. J. Pharm. Biopharm. 86 (2014) 324. [84] T.C. Lai, S. Ferguson, L. Palmer, B.L. Trout, A.S. Myerson, Org. Process Res. Dev. 18 (2014) 1382. [85] T.C. Farmer, C.L. Carpenter, M.F. Doherty, AIChE J. 62 (2016) 3505. [86] T. Lee, H.R. Chen, H.Y. Lin, H.L. Lee, Cryst. Growth Des. 12 (2012) 5897. [87] S. Qamar, M.P. Elsner, I. Hussain, A. Seidel-Morgenstern, Chem. Eng. Sci. 71 (2012) 5. [88] N. Hutnik, A. Kozik, A. Mazienczuk, K. Piotrowski, B. Wierzbowska, A. Matynia, Water Res. 47 (2013) 3635. [89] A. Kozik, N. Hutnik, K. Piotrowski, A. Mazienczuk, A. Matynia, Adv. Chem. Eng. Sci. 3 (2013) 20. [90] X. Sun, Y. Sun, J. Yu, J. Cryst. Growth 419 (2015) 94. [91] A.M. Kolbach-Mandel, J.G. Kleinman, J.A. Wesson, Urolithiasis 43 (2015) 397. [92] A. Gerard, H. Muhr, E. Plasari, D. Jacob, C. Lefaucheur, Powder Technol. 255 (2014) 134. [93] Y. Peng, Z. Zhu, R.D. Braatz, A.S. Myerson, Ind. Eng. Chem. Res. 54 (2015) 7914. [94] S.Y. Wong, A.P. Tatusko, B.L. Trout, A.S. Myerson, Cryst. Growth Des. 12 (2012) 5701. [95] W.J. Liu, C.Y. Ma, X.Z. Wang, Procedia Eng. 102 (2015) 499. [96] J.W. Mullin, Crystallization, 3rd edn, Butterworth-Heinmann, Oxford, England, 1993. [97] M.D. Johnson, S.A. May, J.R. Calvin, J. Remacle, J.R. Stout, W.D. Diseroad, N. Zaborenko, B.D. Haeberle, W. Sun, M.T. Miller, Org. Process Res. Dev. 16 (2012) 1017. [98] L.D. Shiau, K.A. Berglund, Ind. Eng. Chem. Res. 26 (1987) 2515. [99] N.S. Tavare, J. Garside, M.A. Larson, Chem. Eng. Commun. 47 (1986) 185. [100] J. Nyvlt, Int. Chem. Eng. 19 (1979) 547. [101] H. Zhang, J. Quon, A.J. Alvarez, J. Evans, A.S. Myerson, B. Trout, Org. Process Res. Dev. 16 (2012) 915. [102] T. Vetter, C.L. Burcham, M.F. Doherty, Chem. Eng. Sci. 106 (2014) 167. [103] G. Power, G. Hou, V.K. Kamaraju, G. Morris, Y. Zhao, B. Glennon, Chem. Eng. Sci. 133 (2015) 125. [104] J.L. Quon, H. Zhang, A. Alvarez, J. Evans, A.S. Myerson, B.L. Trout, Cryst. Growth Des. 12 (2012) 3036. [105] J. Li, B.L. Trout, A.S. Myerson, Org. Process Res. Dev. 20 (2015) 510. [106] R. Peña, Z.K. Nagy, Cryst. Growth Des. 15 (2015) 4225. [107] T. Vetter, C.L. Burcham, M.F. Doherty, AIChE J. 61 (2015) 2810. [108] J.H. Chaaban, K. Dam-Johansen, T. Skovby, S. Kiil, Org. Process Res. Dev. 17 (2013) 1010. [109] K. Galan, M.J. Eicke, M.P. Elsner, H. Lorenz, A. Seidel-Morgenstern, Cryst. Growth Des. 15 (2015) 1808. [110] C. Rougeot, J.E. Hein, Org. Process Res. Dev. 19 (2015) 1809. [111] Y. Cui, M.O. Mahony, J.J. Jaramillo, T. Stelzer, A.S. Myerson, Org. Process Res. Dev. 20 (2016) 1276. [112] Y. Zhao, V.K. Kamaraju, G. Hou, G. Power, P. Donnellan, B. Glennon, Chem. Eng. Sci. 133 (2015) 106. [113] S. Ferguson, G. Morris, H. Hao, M. Barrett, B. Glennon, Chem. Eng. Sci. 77 (2012) 105. [114] N. Yazdanpanah, A. Myerson, B. Trout, Ind. Eng. Chem. Res. 55 (2016) 5019. [115] G. Cogoni, B.P. de Souza, P.J. Frawley, Chem. Eng. Sci. 138 (2015) 592. [116] A. Majumder, Z.K. Nagy, AIChE J. 59 (2013) 4582. [117] N. Yazdanpanah, S.T. Ferguson, A.S. Myerson, B.L. Trout, Cryst. Growth Des. 16 (2015) 285. [118] A. Majumder, Z.K. Nagy, Cryst. Growth Des. 15 (2015) 1129. [119] A. Koswara, Z.K. Nagy, IFAC-PapersOnLine 48 (2015) 193. [120] Q. Su, B. Benyahia, Z.K. Nagy, C.D. Rielly, Org. Process Res. Dev. 19 (2015) 1859. [121] M.N. Kashid, I. Gerlach, S. Goetz, J. Franzke, J.F. Acker, F. Platte, D.W. Agar, S. Turek, Ind. Eng. Chem. Res. 44 (2005) 5003.

[122] C.J. Gerdts, V. Tereshko, M.K. Yadav, I. Dementieva, F. Collart, A. Joachimiak, R. C. Stevens, P. Kuhn, A. Kossiakoff, R.F. Ismagilov, Angew. Chem. Int. Ed. 45 (2006) 8156. [123] S. Guillemet-Fritsch, M. Aoun-Habbache, J. Sarrias, A. Rousset, N. Jongen, M. Donnet, P. Bowen, J. Lemaître, Solid State Ion. 171 (2004) 135. [124] M. Jiang, Z. Zhu, E. Jimenez, C.D. Papageorgiou, J. Waetzig, A. Hardy, M. Langston, R.D. Braatz, Cryst. Growth Des. 14 (2014) 851. [125] P. Neugebauer, J.G. Khinast, Cryst. Growth Des. 15 (2015) 1089. [126] T. McGlone, N.E. Briggs, C.A. Clark, C.J. Brown, J. Sefcik, A.J. Florence, Org. Process Res. Dev. 19 (2015) 1186. [127] C.J. Callahan, X. Ni, CrystEngComm 16 (2014) 690. [128] C.J. Callahan, X.W. Ni, Can. J. Chem. Eng. 92 (2014) 1920. [129] C.J. Brown, Y.C. Lee, Z.K. Nagy, X. Ni, CrystEngComm 16 (2014) 8008. [130] C.J. Callahan, X. Ni, Cryst. Growth Des. 12 (2012) 2525. [131] N.E. Briggs, U. Schacht, V. Raval, T. McGlone, J. Sefcik, A.J. Florence, Org. Process Res. Dev. 19 (2015) 1903. [132] H. McLachlan, X. Ni, J. Cryst. Growth 442 (2016) 81. [133] L. Zhao, V. Raval, N.E. Briggs, R.M. Bhardwaj, T. McGlone, I.D. Oswald, A.J. Florence, CrystEngComm 16 (2014) 5769. [134] F. Maleky, A.G. Marangoni, J. Food Eng. 89 (2008) 399. [135] F. Maleky, A. Marangoni, Cryst. Growth Des. 11 (2011) 2429. [136] F. Maleky, A. Marangoni, Soft Matter 7 (2011) 6012. [137] F. Maleky, A.K. Smith, A. Marangoni, Cryst. Growth Des. 11 (2011) 2335. [138] G. Mazzanti, M. Li, A.G. Marangoni, S.H. Idziak, Cryst. Growth Des. 11 (2011) 4544. [139] A. Nguyen, J. Kim, S. Chang, W. Kim, Ind. Eng. Chem. Res. 49 (2010) 4865. [140] W. Kim, J. Chem. Eng. Jpn. 47 (2014) 115. [141] S. Lee, A. Choi, W. Kim, A.S. Myerson, Cryst. Growth Des. 11 (2011) 5019. [142] Q. Mayra, W. Kim, Cryst. Growth Des. 15 (2015) 1726. [143] J. Kim, S. Chang, J.H. Chang, W. Kim, Colloid Surf. A Physicochem. Eng. Asp. 384 (2011) 31. [144] S. Lee, C. Lee, W. Kim, J. Cryst. Growth 373 (2013) 32. [145] A. Nguyen, J. Kim, S. Chang, W. Kim, Ind. Eng. Chem. Res. 50 (2011) 3483. [146] A. Nguyen, Y.L. Joo, W. Kim, Cryst. Growth Des. 12 (2012) 2780. [147] Z. Wu, S. Seok, D.H. Kim, W. Kim, Cryst. Growth Des. 15 (2015) 5675. [148] E. Kamio, Y. Seike, H. Yoshizawa, H. Matsuyama, T. Ono, Ind. Eng. Chem. Res. 50 (2011) 6915. [149] J. Aubin, M. Ferrando, V. Jiricny, Chem. Eng. Sci. 65 (2010) 2065. [150] S. Marre, K.F. Jensen, Chem. Soc. Rev. 39 (2010) 1183. [151] S. Teychené, B. Biscans, Chem. Eng. Sci. 77 (2012) 242. [152] J. Lu, J.D. Litster, Z.K. Nagy, Cryst. Growth Des. 15 (2015) 3645. [153] A.I. Toldy, A.Z.M. Badruddoza, L. Zheng, T.A. Hatton, R. Gunawan, R. Rajagopalan, S.A. Khan, Cryst. Growth Des. 12 (2012) 3977. [154] M. Sultana, K.F. Jensen, Cryst. Growth Des. 12 (2012) 6260. [155] D. Rossi, R. Jamshidi, N. Saffari, S. Kuhn, A. Gavriilidis, L. Mazzei, Cryst. Growth Des. 15 (2015) 5519. [156] Y. Zhu, L. Zhu, R. Guo, H. Cui, S. Ye, Q. Fang, Sci. Rep. U. K. 4 (2014) 5046. [157] M. Ildefonso, N. Candoni, S. Veesler, Org. Process Res. Dev. 16 (2012) 556. [158] P. Moschou, M.H. de Croon, J. van der Schaaf, J.C. Schouten, Rev. Chem. Eng. 30 (2014) 127. [159] S. Teychené, B. Biscans, Cryst. Growth Des. 11 (2011) 4810. [160] R. Sivakumar, R. Saravanan, A.E. Perumal, S. Iniyan, Renew. Sustain. Energy Rev. 61 (2016) 280. [161] A.A. Gagnon, Struvite Recovery From Source-Separated Urine Utilizing Fluidized Bed Technology, Virginia Tech, 2016, 2017. [162] M.A. Izquierdo-Barrientos, C. Sobrino, J.A. Almendros-Ibáñez, C. Barreneche, N. Ellis, L.F. Cabeza, Appl. Energy 181 (2016) 310. [163] K. Jiang, K. Zhou, Y. Yang, H. Du, J. Environ. Sci. 25 (2013) 1331. [164] Y. Chen, J.R. Davis, C.H. Nguyen, J.C. Baygents, J. Farrell, Environ. Sci. Technol. 50 (2016) 5900. [165] Y. Shih, H. Chang, Y. Huang, J. Taiwan Inst. Chem. Eng. 62 (2016) 177. [166] C. Chen, Y. Shih, Y. Huang, Chem. Eng. J. 279 (2015) 120. [167] C. Su, L.D. Dulfo, M.L.P. Dalida, M. Lu, Sep. Purif. Technol. 125 (2014) 90. [168] M. Mangold, L. Feng, D. Khlopov, S. Palis, P. Benner, D. Binev, A. SeidelMorgenstern, J. Comput. Appl. Math. 289 (2015) 253. [169] L. Shiau, Y. Lu, C. Lin, J.D. Ward, Taiwan Inst. Chem. Eng. 50 (2015) 76. [170] D. Binev, A. Seidel-Morgenstern, H. Lorenz, Chem. Eng. Sci. 133 (2015) 116. [171] D. Binev, A. Seidel-Morgenstern, H. Lorenz, Cryst. Growth Des. 16 (2016) 1409. [172] Z. Sha, C. Xiong, Q. Chen, Chem. Eng. Technol. 36 (2013) 1307. [173] M. Al-Rashed, J. Wójcik, R. Plewik, P. Synowiec, A. Kus, Chem. Eng. Process. Process Intensif. 63 (2013) 7. [174] N. Yazdanpanah, T.A. Langrish, Dry. Technol. 29 (2011) 1046. [175] N. Yazdanpanah, T.A. Langrish, Dairy Sci. Technol. 91 (2011) 323. [176] W. Clark, M. Japs, M.J. Burk, Google Patents (2013). [177] S.K. Bermingham, A.M. Neumann, J.P. Muusze, H.J. Kramer, P.J. Verheijen, Part. Part. Syst. Charact. 15 (1998) 56. [178] F. Farahbod, D. Mowla, M.J. Nasr, M. Soltanieh, Sol. Energy 97 (2013) 138. [179] F. Farahbod, D. Mowla, M.J. Nasr, M. Soltanieh, Desalination 285 (2012) 352. [180] H. Guo, H.M. Ali, A. Hassanzadeh, Appl. Therm. Eng. 108 (2016) 486. [181] K. Loganathan, P. Chelme-Ayala, M.G. El-Din, J. Environ. Manag. 165 (2016) 213. [182] S. Agrawal, A.T. Paterson, J. McLeod, J. Jones, J. Bronlund, J. Food Eng. 154 (2016). [183] M. Malfand, Google Patents (2013). [184] T.C. Frank, J.W. Moore, P.A. Larsen, P.D. Patil, R.M. Whittingslow, P.A. Gillis, Google Patents (2012).

T. Wang et al. / Journal of Industrial and Engineering Chemistry 54 (2017) 14–29 [185] H. Kramer, J.W. Dijkstra, P. Verheijen, G.M. Van Rosmalen, Powder Technol. 108 (2000) 185. [186] A.T. Cheng, Google Patents (1995). [187] P.M. Synowiec, A. Małysiak, J. Wójcik, Chem. Eng. Sci. 77 (2012) 78. [188] H.J. Pant, Appl. Radiat. Isot. 53 (2000) 999. [189] K. Xu, C. Wang, X. Wang, Y. Qian, Chemosphere 88 (2012) 219. [190] S.Y. Wong, R.K. Bund, R.K. Connelly, R.W. Hartel, Cryst. Growth Des. 10 (2010) 2620. [191] F.J. Wang, Computer Fluid Dynamics Analysis—Principles and Applications of CFD, Tsinghua University Press, Beijing, China, 2004. [192] B. Wierzbowska, K. Piotrowski, J. Koralewska, N. Hutnik, A. Matynia, Korean J. Chem. Eng. 26 (2009) 175. [193] J. Koralewska, K. Piotrowski, B. Wierzbowska, A. Matynia, Chin. J. Chem. Eng. 17 (2009) 330. [194] H. Pan, J. Li, Y. Jin, B. Yang, X. Li, Int. J. Chem. Eng. 2016 (2016). [195] A. Soare, R. Lakerveld, J. Van Royen, G. Zocchi, A.I. Stankiewicz, H.J. Kramer, Ind. Eng. Chem. Res. 51 (2012) 10895. [196] A. Soare, S.A. Pe’rez Escobar, A.I. Stankiewicz, M. Rodriguez Pascual, H.J. Kramer, Ind. Eng. Chem. Res. 52 (2013) 12212. [197] R. Lakerveld, J.J. Van Krochten, H.J. Kramer, Cryst. Growth Des. 14 (2014) 3264. [198] X. Jiang, B. Hou, G. He, J. Wang, Chem. Eng. Sci. 84 (2012) 120. [199] X. Jiang, B. Hou, G. He, J. Wang, Chem. Eng. Sci. 91 (2013) 111. [200] X. Jiang, W. Xiao, G. He, Chem. Eng. Sci. 117 (2014) 198. [201] M.L.P. Mostefa, H. Muhr, E. Plasari, M. Fauconet, Powder Technol. 255 (2014) 98. [202] A. König, M. Stepanski, A. Kuszlik, P. Keil, C. Weller, Chem. Eng. Res. Des. 86 (2008) 775. [203] Y. Sun, X. Song, M. Jin, W. Jin, J. Yu, Ind. Eng. Chem. Res. 52 (2013) 17598. [204] L.L. Simon, H. Pataki, G. Marosi, F. Meemken, K. Hungerbühler, A. Baiker, S. Tummala, B. Glennon, M. Kuentz, G. Steele, Org. Process Res. Dev. 19 (2015) 3. [205] C. Schaefer, C. Lecomte, D. Clicq, A. Merschaert, E. Norrant, F. Fotiadu, J. Pharm. Biomed. 83 (2013) 194. [206] H. Zhang, R. Lakerveld, P.L. Heider, M. Tao, M. Su, C.J. Testa, A.N.D. Antonio, P.I. Barton, R.D. Braatz, B.L. Trout, Cryst. Growth Des. 14 (2014) 2148. [207] A. Hertrampf, H. Müller, J.C. Menezes, T. Herdling, J. Pharm. Biomed. 114 (2015) 208.

29

[208] A.L. Kelly, T. Gough, R.S. Dhumal, S.A. Halsey, A. Paradkar, Int. J. Pharm. 426 (2012) 15. [209] H.G. Moradiya, M.T. Islam, N. Scoutaris, S.A. Halsey, B.Z. Chowdhry, D. Douroumis, Cryst. Growth Des. 16 (2016) 3425. [210] K.A. Powell, A.N. Saleemi, C.D. Rielly, Z.K. Nagy, Chem. Eng. Process. Process Intensif. 97 (2015) 195. [211] T. Rajalahti, O.M. Kvalheim, Int. J. Pharm. 417 (2011) 280. [212] D.E. Braun, S.G. Maas, N. Zencirci, C. Langes, N.A. Urbanetz, U.J. Griesser, Int. J. Pharm. 385 (2010) 29. [213] K. Pöllänen, A. Häkkinen, S. Reinikainen, J. Rantanen, M. Karjalainen, M. Louhi-Kultanen, L. Nyström, J. Pharm. Biomed. 38 (2005) 275. [214] S. Ferguson, G. Morris, H. Hao, M. Barrett, B. Glennon, Chem. Eng. Sci. 104 (2013) 44. [215] S. Le Borne, H. Eisenschmidt, K. Sundmacher, Chem. Eng. Sci. 139 (2016) 61. [216] C.J. Brown, X. Ni, Cryst. Growth Des. 11 (2011) 3994. [217] C.J. Brown, X. Ni, CrystEngComm 14 (2012) 2944. [218] S. Schorsch, D.R. Ochsenbein, T. Vetter, M. Morari, M. Mazzotti, Chem. Eng. Sci. 105 (2014) 155. [219] C. Tachtatzis, R. Sheridan, C. Michie, R.C. Atkinson, A. Cleary, J. Dziewierz, I. Andonovic, N.E. Briggs, A.J. Florence, J. Sefcik, Chem. Eng. Sci. 133 (2015) 82. [220] C.J. Brown, X. Ni, Cryst. Growth Des. 11 (2011) 719. [221] Guidance for industry: PAT—A framework for innovative pharmaceutical development, manufacturing, and quality assurance, F.D.A. Administration, DHHS, Rockville, MD (2004). [222] M. Furuta, K. Mukai, D. Cork, K. Mae, Chem. Eng. Process. Process Intensif. 102 (2016) 210. [223] R.J. Eder, S. Schrank, M.O. Besenhard, E. Roblegg, H. Gruber-Woelfler, J.G. Khinast, Cryst. Growth Des. 12 (2012) 4733. [224] M. Jiang, C.D. Papageorgiou, J. Waetzig, A. Hardy, M. Langston, R.D. Braatz, Cryst. Growth Des. 15 (2015) 2486. [225] X. Ji, E. Curcio, S. Al Obaidani, G. Di Profio, E. Fontananova, E. Drioli, Sep. Purif. Technol. 71 (2010) 76. [226] G. Chen, Y. Lu, W.B. Krantz, R. Wang, A.G. Fane, J. Membr. Sci. 450 (2014) 1. [227] N. Variankaval, A.S. Cote, M.F. Doherty, AIChE J. 54 (2008) 1682. [228] J. Chen, B. Sarma, J.M. Evans, A.S. Myerson, Cryst. Growth Des. 11 (2011) 887.