Application of Wet Milling-Based Automated Direct Nucleation Control

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES

Article

Application of Wet Milling Based Automated Direct Nucleation Control in Continuous Cooling Crystallization Processes Yang Yang, Liangcheng Song, Yuqi Zhang, and Zoltan K Nagy Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.5b04956 • Publication Date (Web): 11 Apr 2016 Downloaded from http://pubs.acs.org on April 17, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Application of Wet Milling Based Automated Direct Nucleation Control in Continuous Cooling Crystallization Processes Yang Yanga, Liangcheng Songa,b, Yuqi Zhanga, Zoltan K. Nagy a,*

a

School of Chemical Engineering, Purdue University, West Lafayette, IN 47907, USA

b

School of Chemical Engineering & Technology, Harbin Institute of Technology, Harbin 150001,

P. R. China * Corresponding author. Email: [email protected]

Abstract A novel wet milling based automated direct nucleation control (WMADNC) concept is proposed and implemented in two continuous cooling crystallization processes in this work. In process 1, WMADNC is implemented upstream to provide closed-loop controlled primary nucleation kinetics and produce seed crystals in situ for continuous mixed suspension mixed product removal crystallizer (MSMPRC). In process 2, WMADNC is applied downstream to achieve closed-loop controlled secondary nucleation kinetics and reduce particle size for crystals in the MSMPRC through recycling. For both processes, a focused beam reflectance measurement (FBRM) based closed-loop control approach is successfully implemented to control the nucleation kinetics inside the wet mill therefore to control the final chord length distribution (CLD). In the proposed WMADNC approach, the process critical quality attributes (CQA) (e.g. particle number or size) can be directly controlled by the automated closed-loop control framework, which on one hand is aligned with the quality-by-design (QbD) concept, and on the other hand also applies the new quality-by-control (QbC) framework in continuous crystallization processes. The proposed 1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

WMADNC approach although was not able to improve startup dynamics, it can provide high quality control of CLD and automated and effective disturbance rejection for continuous cooling crystallization processes.

Keywords Automated direct nucleation control; Wet milling; Mixed suspension mixed product removal crystallizer; Crystal size distribution control

1. Introduction Batch crystallization is one of the most widely used unit operations in pharmaceutical, food and fine chemical industries to purify and produce particulate products. Critical quality attributes (CQA) in this process include solid purity, crystal size distribution (CSD), crystal morphology.1,2 CSD is important because it affects not only the efficiency of downstream operations such as filtration and drying, but also physical and chemical properties of the final product.1,3 Although large crystals are desired for the purpose of efficient downstream processing,3 small and uniform crystals are also of great interests in pharmaceutical industry to meet dissolution, tableting and bioavailability requirements and avoid downstream milling process.4 Significant amount of work have been reported in literature studying different approaches to achieve a target crystal size in batch crystallization processes. These approaches can be differentiated in two categories. The first category is by controlling and optimizing operating profiles (e.g. temperature or/and antisolvent addition profiles) within the batch crystallization process in either open-loop or closed-loop approach using process analytical technology (PAT) to estimate crystal size or concentration in real-time. 5

7

For example, automated direct nucleation control (ADNC) and supersaturation

2


Page 2 of 32

Page 3 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


control (SSC) are two widely used model-free (or direct design) PAT-based feedback control approaches to achieve optimal operating profiles in batch crystallization process.7, 8 Both approaches are useful and robust to avoid fine particles in batch crystallization. The second category is by integrating a second unit operation with batch crystallization process to achieve target crystal size. The most well-known examples are wet mill, T-mixer, Y-mixer and impinging jet mixer. Wet mill can be integrated with a batch or semi-batch crystallizer through recycling.9

12

In this configuration, either crystallization followed by wet milling, or simultaneous crystallization and wet milling can be implemented. It was observed that narrow and well controlled CSD could be obtained due to controlled secondary nucleation and breakage mechanisms in wet mill. T-mixer, Y-mixer and imping jet mixer can directly produce small crystals of narrow distribution.4,13

16

Woo

et al. proved that it could be used to provide optimal seed crystals for batch crystallizer or aging vessel to achieve target CSD.4 The aim of this work is to investigate the possibility and performance of combining approaches from both categories (i.e. PAT-based feedback control and wet milling) in order to achieve a high level of control of crystal size in crystallization processes. To the best of our knowledge this is the first time when a feedback control strategy is implemented in an integrated wet milling and crystallization process. The study is carried out in continuous crystallization, also motivated by the recent increased interest in adopting continuous manufacturing technologies, supported by the US Food and Drug Administration (FDA), because of the potential of increased efficiency and better product quality.17 The development of a robust continuous crystallization process that can provide superior crystal size control has significant industrial interest. Compared to batch crystallization that is operated close to equilibrium state, continuous crystallization is operated at steady-state, which leads to

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 32

more consistent product quality, higher process efficiency and smaller operation cost.18

26

Mixed

suspension mixed product removal crystallizer (MSMPRC) and plug flow crystallizer (PFC) are the two most widely studied continuous crystallizers. Compared to PFC, MSMPRC is more similar to batch crystallizer in terms of operation and control therefore it is used more often.24 The development of continuous MSMPRC dates back to half century ago when population and mass balances were solved to model CSD in MSMPRC.27 Since 1980s, both modelling and experimental works were carried out in literature to design and control continuous crystallization in MSMPRC to measure crystallization kinetics and/or achieve a target CSD.18,19,22,28

41

Mashingaidze et al.

developed a mathematical model for continuous MSMPRC as a dissolver to tailor CSD.39 Nishio et al. designed a small-scale MSMPRC for kinetics measurement.40 Chemaly et al. investigated continuous crystallization of calcium lactate in MSMPRC with a wide range of residence time, supersaturation and magma density covered to determine crystallization kinetic parameters.41 Hou et al. studied the effects of supersaturation, residence time and magma density on the steady-state CSD.30 Power et al. achieved maximum average crystal size and minimum width of size distribution by optimizing process temperatures and residence times for cooling crystallization in two-stage MSMPRC.32 Yang and Nagy maximized average crystal size for combined cooling/antisolvent crystallization in multi-stage MSMPRC.18 However, the works reported in literature mainly used open-loop approaches, which were neither automated nor robust. Yang et al. reported that very uniform crystals can be obtained by integrating a wet mill unit with continuous MSMPRC.42 The advantages of integrating wet milling (WM) into continuous crystallization, such as improved yield and narrowed size distribution, have been observed. However, there was no process control framework implemented in the study of this WM technique. In addition, Yang, et al. introduced a focused beam reflectance measurement (FBRM) based

4


Page 5 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


feedback control approach for cascade MSMPRC systems by extending the automated direct nucleation control (ADNC) approach developed originally for batch crystallization processes. 43 In this ADNC approach the crystal size is controlled by manipulating the MSMPRC jacket temperature and nucleation kinetics. However, one potential limitation of this ADNC technique is that the MSMPRC nucleation cannot be separated from growth therefore this nucleation control approach will always be also influenced by the growth kinetics in the MSMPRC. Therefore in this paper, we would like to introduce a novel approach called wet milling based automated direct nucleation control (WMADNC), which is a technique that combines the advantages of WM and ADNC. This approach is aligned with both the quality-by-design (QbD) and the new quality-by-control (QbC) concept in pharmaceutical manufacturing.43 It implements the continuous ADNC closed-loop approach in a continuous wet mill unit which behaves as a perfect nucleator. The idea is to use the total particle chord counts measured by FBRM in the crystallizer to manipulate the jacket temperature of the wet mill in closed-loop. By manipulating the temperature in the wet mill the supersaturation at which nucleation happens is controlled. Because of the very short residence time in the wet mill this approach controls the nucleation kinetics directly in the wet mill without being influenced by the growth kinetics. The proposed WMADNC concept is integrated with a continuous MSMPRC in both upstream and downstream configurations to achieve closed-loop controlled primary or secondary nucleation kinetics. This WMADNC approach should not only contain all the advantages of WM and ADNC, but also have better control performance on particle size due to decoupled nucleation and growth kinetics. Paracetamol, which is an active pharmaceutical ingredient (API), and ethanol are used as solute and solvent, respectively. Performances of WMADNC under startup, steady-state, set-point change and disturbance are studied for continuous cooling crystallization in the MSMPRC.

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2. Experimental section 2.1. Upstream application of WMADNC approach (process 1) The experimental setup for process 1 is shown in Figure 1(a). The feed vessel and MSMPRC were 1 liter round bottom jacketed vessel. A magic LAB rotor-stator wet mill unit (IKA) was installed between the feed vessel and MSMPRC. A medium and a fine generators were used in the wet mill. The angular speed of the wet mill was set to 10,000 rpm in all the experiments. Three temperature bathes (Huber) were connected to the jackets of feed vessel, MSMPRC and wet mill. Clear feed solution was transferred from the feed vessel directly into the wet mill through a peristaltic pump (Masterflex). Both wet mill and MSMPRC were equipped with thermocouples to measure the process temperatures. In addition, a FBRM ParticleTrack G400 probe (Mettler-Toledo) was installed in the MSMPRC to provide real-time information of particle chord counts and chord length distribution (CLD). The FBRM information was measured every 15 seconds and collected in the software iC FBRM 4.3, which communicated via an RS232 interface with the LABVIEW based Crystallization Monitoring and Control (CryMOCO) software that was installed on a supervisory computer.42,43 All temperature measurements were also collected in CryMOCO. Slurry from the MSMPRC was transferred out semi-continuously through a transfer unit that used vacuum and nitrogen gas.42,43 The feed solution was prepared from paracetamol (Alfa Aesar, purity > 98%) and ethanol (KOPTEC, purity > 99%) at a concentration of 0.26 g/g, which corresponds to solubility at 38 o

C.44 The feed flow rate was kept at 20 mL/min in all the experiments. Every 40 seconds around

13 mL slurry was transferred out from the MSMPRC through the transfer unit. The total volume including wet mill and MSMPRC was maintained at around 400 mL. Therefore the total residence time (RT) was 20 minutes. The jacket temperatures of feed vessel and MSMPRC were kept at 55

6


Page 6 of 32

Page 7 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


o

C and 25 oC, respectively, in all the experiments. The feed solution, final stage temperature and

residence times are the same for all the experiments, therefore the yields are comparable in all the experiments. The product slurry was recycled for preparing feed solutions in order to save API.32,42,43 Slurry samples were taken at steady-states and filtered immediately. The sample crystal images were taken using SMZ1500 optical microscope (Nikon). DXR Raman microscope (Thermo Scientific) was used for polymorph analysis with 532 nm laser wavelength and 8 mW laser power.

2.2. Downstream application of WMADNC approach (process 2) Figure 1(b) presents the experimental setup for process 2. Compared to process 1, the only difference was that the wet mill was connected to the MSMPRC through external recycling. In this case the wet mill worked as a downstream size reduction unit under high shear secondary nucleation and breakage mechanisms. All the process operating conditions (e.g. feed flow rate, operating volume, temperatures, etc.), wet mill configurations (e.g. generators, angular speed) and measurements settings used in process 2 were exactly the same as in process 1. In addition, the recycling flow rate through the wet mill in process 2 was 200 mL/min, which can be used to calculate the number of turnover per residence time (TOPRT):42

TOPRT =

qmill  qmill = V qMSMPRC

(1)

where  is the MSMPRC residence time, qmill and qMSMPRC are the recycling flow rate through wet mill and the feed flow rate through the MSMPRC, respectively. In this case a TOPRT = 10 was used. The physical meaning of this TOPRT value is that on average, the slurry was milled 10 times before it was washed out from the MSMPRC. The TOPRT is an important operating parameter in wet milling integrated continuous crystallization since it defines how frequently the 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

slurry is milled.42

Figure 1. Experimental setups for (a) process 1 and (b) process 2.

8


Page 8 of 32

Page 9 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2.3. WMADNC methodology in continuous cooling crystallization WMADNC is a FBRM based closed-loop (feedback) control approach that we propose to control primary nucleation in process 1, or secondary nucleation in process 2 directly in the wet mill therefore control the final CLD. Both primary and secondary nucleation kinetics are strong functions of supersaturation:45

Bprim  k prim c

n prim

(2)

Bsec  ksec cnsec m2

(3)

where Bprim and Bsec are primary and secondary nucleation rates, k prim and k sec are primary and secondary nucleation rate constants, nprim and nsec are primary and secondary nucleation orders,

c is absolute supersaturation, and m2 is the second moment of the CSD. Since the real-time particle total chord counts measurement from FBRM can reflect total particle number and nucleation kinetics in the system, it is used in the closed-loop control to automatically adjust the jacket temperature set-point of the wet mill therefore control supersaturation and primary or secondary nucleation rate directly inside the wet mill. Similar as the continuous ADNC algorithm,43 first of all an upper and a lower bounds for desired particle total chord counts setpoint must be defined in WMADNC (Figure 2). During the operation, if the measured total counts was found smaller than the lower bound, a pre-defined cooling rate would be actuated automatically by CryMOCO to cool down the wet mill jacket therefore increase supersaturation and hence nucleation rate in the wet mill, and vice versa. If the measured total counts fell between the upper and the lower bounds, then the wet mill jacket temperature would be kept constant by CryMOCO in order to maintain total counts and steady-state operation. In this work, heating and

9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 32

cooling rates were set as ± 0.1 oC/min in all the experiments that used WMADNC. The initial wet mill jacket temperature was set to 20 oC in all the experiments.

Figure 2. WMADNC methodology.

3. Results and discussion 3.1. Upstream application of the WMADNC approach to achieve closed-loop controlled primary nucleation (process 1) In process 1 (Figure 1(a)), the wet mill behaves as an upstream in situ seed generator under high shear primary nucleation mechanism since the feed solution is clear. It produces seed crystals for the MSMPRC, which acts as a growth vessel. WMADNC approach is used upstream to achieve closed-loop controlled primary nucleation kinetics. The dynamics of process 1 applied with WMADNC approach under startup, steady-state and disturbance were studied. 3.1.1. Startup and steady-state dynamics The startup and steady-state dynamics of process 1 with the WMADNC approach applied is plotted in Figure 3(a), which includes milling outlet temperature, milling jacket temperature set10


Page 11 of 32

point, FBRM total particle chord counts, WMADNC total counts set-point and FBRM square weighted mean chord length (SWMCL). The upper and lower bounds of the WMADNC total counts set-point are 18,500 and 16,000, respectively. The fast increase of total counts indicates large primary nucleation rate at the beginning of the process. Then once the total counts overshoots above the set-point, it can be seen that the WMADNC approach automatically heats up the wet mill jacket to reduce the primary nucleation rate. Finally two oscillations of total counts are observed during startup before it converges to the desired WMADNC set-point and steady-state operation. In this work, steady-state is assumed reached once the measured total counts are around the WMADNC set-point for 2 RT or longer.43 Figure 3(b) presents the comparison of the total counts dynamics for process 1 with WMADNC approach implemented and case without control when mill jacket and outlet temperatures are constant. It is clear that both experiments end up with two oscillations during startup and reach steady-state at a similar time. Therefore the WMADNC approach in this case does not show any improvement in the startup duration. Because the primary nucleation kinetics in wet mill is too fast and too sensitive to supersaturation therefore change in milling temperature is unfavorable to stabilize startup. This observation proves that closed-loop control for upstream wet mill which works as in situ nucleator is redundant during startup operation.

(a) 52 48

o

Temperature [ C]

44

36

120

20000 16000

40

140

24000

12000

32

100 80 60

28

8000

24

4000

20

0

0

20 0

1

2

3

4 5 6 No. of Residence Time (RT) [-]

11


7

8

9

40

Mean Chord Length [m]

Mill outlet temperature Mill jacket setpoint Total counts WMANDC set-point SWMCL

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



(b) 30000

Total Counts Without control Total Counts With WMADNC

25000

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 32

20000 15000 10000 5000 0

0

1

2

3

4 5 6 No. of Residence Time (RT) [-]

7

8

9

10

Figure 3.(a) Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and square weighted mean chord length (SWMCL) during startup and steady-state operation in process 1 with WMADNC approach. (b) Comparison of total particle chord counts dynamics in process 1 without control and with WMADNC approach.

3.1.2. Set-point change and control of CLD After steady-state 1 is reached in process 1 with WMADNC approach, the upper and lower bounds of WMADNC total counts set-point are reduced to 12,500 and 10,000. As shown in Figure 4, when the total counts set-point is reduced, the wet mill jacket temperature automatically increases to decrease supersaturation and hence nucleation rate in the wet mill. The process is able to reach the new total counts set-point and steady-state 2 very quickly. The SWMCL in steadystate 2 is much larger than in steady-state 1 due to the reduced nucleation rate and increased supersaturation level in the MSMPRC for crystal growth. Square weighted chord length distributions (CLDs) of these two steady-states shown in Figure 5 confirm the increase of average crystal size. It is noticed that many more large crystals and less fines can be obtained by decreasing the total counts set-point in process 1. Therefore this WMADNC approach provides an efficient and novel way to directly control CLD by adjusting the WMADNC total chord counts set-point. Additionally, it can be noticed that at the end of steady-state 1 zone in Figure 4, the total counts drops slightly below set-point lower bound and shows a downward trend. This extent of variation of FBRM data is very common in continuous MSMPRC experiments in literature21,30,42,43 because of unavoidable disturbances in continuous crystallization experiments, including variation of 12


Page 13 of 32

slurry volume due to semi-continuous transferring, non-ideal transferring due to not well-mixed slurry, among others. Therefore perfect steady-state operation and FBRM data is never achievable in real experiments, and only quasi steady-state is achievable. However, for simplification the term “steady-state” will be used in this work instead.43 Since the FBRM data has already been stabilized for more than 2 RT in steady-state 1, and total counts stopped decreasing right after steady-state 1 zone, it is reasonable to assume steady-state 1 is reached and maintained. In addition, in Figure 4 as soon as the total counts drops below the set-point lower bound, it can be seen that the controller actuates a cooling rate to counteract the disturbance, and very soon the total counts stopped decreasing. This shows the ability of WMADNC to counteract disturbance. More detailed study of WMADNC’s ability to reject disturbance will be introduced in the next section. Mill outlet temperature Mill jacket setpoint Total counts WMADNC set-point SWMCL

56 52

Steady-state 2

44

o

Temperature [ C]

Steady-state 1

40

16000

90

12000

80

8000

36 32

4000

28 24 20

100

7

8

9

10 11 No. of Residence Time [-]

12

13

14

70 60

0

50

-4000

40

Figure 4. Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and SWMCL in process 1 with WMADNC approach under set-point change.

13



48

20000

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



Steady-state 1 Steady-state 2

14 12

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 32

10 8 6 4 2 0

10

100 Chord Length [m]

Figure 5. Square weighted CLDs of steady-state 1 and steady-state 2 in process 1 with WMADNC approach.

3.1.3. Disturbance rejection Since the WMADNC is a closed-loop nucleation control approach, its performance to reject disturbance is also evaluated. According to Figure 6, at first, process 1 is at steady-state 2 with the WMADNC approach enabled. Then suddenly a step disturbance is introduced on purpose by changing the angular speed of the wet mill from 10,000 rpm to 6,000 rpm. Decrease of the wet mill angular speed can lead to two opposite effects. On one hand, the reduced shear rate decreases nucleation rate. On the other hand, the reduced angular speed decreases the energy introduced therefore decreases milling temperature and increases supersaturation and primary nucleation rate. It is observed that the total counts and SWMCL start to decrease and increase, respectively, after the step disturbance is introduced. Therefore in this case the first effect is dominating and primary nucleation rate is reduced. But once the measured total counts drop below the set-point lower bound, the disturbance is noticed by the WMADNC approach which then cools down the wet mill automatically to enhance supersaturation and nucleation rate in the wet mill. Finally both total

14


Page 15 of 32

counts and SWMCL are able to return to their original values although the disturbance still exists, and steady-state 3 is reached. The CLDs of steady-state 2 and steady-state 3 in this test (Figure 7) demonstrate that the disturbance is successfully rejected, and same CLD is obtained before and after this disturbance. 60

Mill@ Mill@ 10k rpm 6k rpm

56

o

Temperature [ C]

44

Steady-state 3

Steady-state 2

81 12000 10000

40 36

8000

32 28

84

14000

6000

24 16.0

16.5

17.0 17.5 No. of Residence Time (RT) [-]

18.0

18.5

78 75 72 69 66

Figure 6. Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and SWMCL in process 1 under disturbance with WMADNC approach.

12


Counts [#/s]

10 8 6 4 2 0

Chord Length [m]

Figure 7. Square weighted CLDs at steady-state 2 and steady-state 3 in process 1 with WMADNC implemented upstream.

The microscope images and Raman spectra of crystals obtained in steady-state 1, steady-state 2 and steady-state 3 during process 1 servo control (i.e. set-point change) and regulatory control

15



52 48

Mill outlet temperature Mill jacket setpoint Total counts WMADNC set-point SWMCL

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 32

(i.e. disturbance change) tests are presented in Figure 8 and Figure 9, respectively. Figure 8 confirms that CLD shifts towards larger size from steady-state 1 to steady-state 2 due to reduced primary nucleation rate, and CLD does not change from steady-state 2 to steady-state 3 though disturbance exists. In addition, the Raman spectra of samples presented in Figure 9 indicate that monoclinic paracetamol (Form I, stable form, octahedrons) rather than orthorhombic paracetamol (Form II, metastable form, needle-like) crystals are obtained in all three steady-states.46

48

This

observation agrees with literature in which wet mill was found to facilitate the formation of stable form crystals.49

16


Page 17 of 32

Figure 8. Microscope images of crystals obtained in (a) steady-states 1, (b) steady-state 2 and (c) steady-state 3 in process 1 with WMADNC implemented upstream.

12000

9000

Counts [a.u.]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Steady-state 1

6000

Steady-state 2

3000

Steady-state 3

Form I

0 400

800

1200 1600 2000 2400 2800 3200 -1

Raman shift [cm ] Figure 9. Raman spectra of crystals obtained in steady-state 1, steady-state 2 and steady-state 3 in process 1 with WMADNC implemented upstream.

3.2. Downstream application of the WMADNC approach to achieve closed-loop controlled secondary nucleation (process 2) In process 2 (Figure 1(b)), the wet mill works as a downstream size reduction unit under high

17



shear secondary nucleation and breakage mechanisms. It can efficiently reduce crystal size. The WMADNC approach is used downstream to achieve closed-loop controlled secondary nucleation kinetics. The application of the WMADNC approach in process 2 under startup, steady-state and disturbance are investigated next. 3.2.1. Startup and steady-state dynamics Process variables during operation of process 2 with the WMADNC approach enabled are presented in Figure 10(a). The upper and lower bounds of the WMADNC set-point are selected as 13,000 and 10,000, respectively. Within the first RT, the particle total chord counts increases very fast due to the high shear enhanced secondary nucleation kinetics, which leads to an overshoot of the total counts. As a result, the WMADNC approach heats up the wet mill jacket to reduce supersaturation and secondary nucleation rate. Finally the total counts automatically converges to the selected set-point after one oscillation. Figure 10(b) presents a comparison of total counts in process 2 without control (i.e. constant mill temperature) and with WMADNC approach. Similarly as in process 1, WMADNC approach does not improve startup in process 2 either. This again confirms that change in milling temperature is unfavorable to stabilize the startup dynamics since the nucleation kinetics in the wet mill is too fast and too sensitive to the milling temperature. (a) 55

45

20000 16000 12000

o

Temperature [ C]

50

100

24000

40

8000

35

4000

30

0

25

-4000

80

60

40

-8000

20 0

1

2

3

4 5 6 7 No. of Residence Time (RT) [-]

18


8

9

10

-12000

20



Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 32

Page 19 of 32

(b) 30000

Total Counts Without control Total Counts With WMADNC

25000

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


20000 15000 10000 5000 0

0

1

2

3


8

9

10

11

Figure 10. (a) Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and SWMCL during startup and steady-state operation in process 2 with WMADNC approach. (b) Comparison of total particle chord counts dynamics in process 2 without control and with WMADNC approach.

3.2.2. Set-point change and control of CLD As shown in Figure 11, the upper and lower bounds of the WMADNC set-point are reduced to 8,000 and 6,000, respectively, when the process is at steady-state 1. This set-point change results in the WMADNC adapting the heating of the wet mill jacket to decrease supersaturation and secondary nucleation rate. The total counts is able to reach the new set-point after a few RT. At the same time, it can be seen from Figure 11 that the SWMCL increases from steady-state 1 to steadystate 2. In addition, the square weighted CLDs of both steady-states are compared in Figure 12. It is clear that the number of fine crystals decreases significantly from steady-state 1 to steady-state 2. This also provides an excellent approach to tailor CLD and control the number of fine crystals in the system by adjusting the WMADNC set-point.

19



56


Steady-state 4

52

o

Temperature [ C]

Steady-state 5

44

90 8000

40 36

4000

32 28

0

24 20 8

9

10

11


16

17

18

80 70 60 50


48

100 12000

Counts [#/s]

40

Figure 11. Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and SWMCL in process 2 with WMADNC approach under set-point change.


6 5

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 32

4 3 2 1 0

10

100 Chord Length [m]

Figure 12. Square weighted CLDs of steady-state 4 and steady-state 5 in process 2 with WMADNC approach.

3.2.3. Disturbance rejection The closed-loop disturbance rejection performance of the WMADNC approach in process 2 is also tested. As shown in Figure 13, the wet mill angular speed is suddenly changed from 10,000 rpm to 13,000 rpm when the system is at steady-state. The two opposite effects of changing wet mill angular speed described in process 1 also exist in this process. It is clear that in this process the temperature effect of wet milling is dominating, and total counts and SWMCL begin to 20


Page 21 of 32

decrease and increase respectively due to increased milling temperature and reduced secondary nucleation rate. The decrease in total counts automatically activates the WMADNC approach, which cools down the wet mill jacket immediately and finally brings total counts and SWMCL back to their initial steady-state values. The square weighted CLDs shown in Figure 14 confirm that the step disturbance is completely rejected by WMADNC approach and same CLD is achieved

o

72 68 Mill@ Mill@ 64 10k rpm 13k rpm 60 56 Steady-state 5 52 48 44 40 36 32 28 24 20 22 23


Steady-state 6

100 10000 90 8000 6000 4000 2000

24 25 No. of Residence Time (RT) [-]

26

0 27

80 70 60


Temperature [ C]

consistently under disturbance.

Counts [#/s]

50

Figure 13. Milling outlet temperature and jacket temperature set-point, FBRM total particle chord counts, WMADNC total counts set-point, and SWMCL in process 2 with WMADNC approach under disturbance.

6


5

Counts [#/s]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4 3 2 1 0

Chord Length [m]

Figure 14. Square weighted CLDs of steady-state 5 and steady-state 6 in process 2 with WMADNC implemented downstream.

21



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 32

Figure 15 and Figure 16 present the microscope images and Raman spectra of crystals obtained in steady-state 4, steady-state 5 and steady-state 6 in process 2, respectively. The results confirm that less fine particles are present in steady-state 5 than steady-state 4 due to reduced secondary nucleation rate, and the disturbance introduced does not change the CLD from steady-state 5 to steady-state 6. In addition, the stable form of paracetamol (Form I) is obtained in all three steadystates, indicated by the identical Raman spectra of all samples, shown in Figure 16.

22


Page 23 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 15. Microscope images of crystals obtained in (a) steady-states 4, (b) steady-state 5 and (c) steady-state 6 in process 2 with WMADNC implemented downstream.

23



15000 12000

Counts [a.u.]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 32

Steady-state 4

9000 Steady-state 5

6000 Steady-state 6

3000

Form I

0 400

800

1200 1600 2000 2400 2800 3200 -1

Raman shift [cm ] Figure 16. Raman spectra of crystals obtained in steady-state 4, steady-state 5 and steady-state 6 in process 2 with WMADNC implemented downstream.

The FBRM based closed-loop WMADNC approach described in this work should be applicable to not only continuous MSMPRC, but also other types of continuous crystallizers such as continuous plug flow crystallizer. It can be envisioned that both upstream and downstream applications of WMADNC can be integrated with continuous plug flow crystallization processes to achieve closed-loop controlled primary or secondary nucleation kinetics, therefore provide superior control on CLD and reject disturbances. In addition, the proposed WMADNC approach does not require any modelling effort or knowledge about solubility or crystallization kinetics of the compound. The only requirement is a few experiments to tune the control parameters such as heating and cooling rates. Therefore the WMADNC approach provides a very fast approach for highly automated recipe design for continuous crystallization processes.

24


Page 25 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4. Conclusions A closed-loop WMADNC approach is proposed and implemented in laboratory scale integrated continuous MSMPRC-wet mill system. The proposed approach is demonstrated to provide closedloop controlled primary or secondary nucleation kinetics by using it upstream or downstream in continuous cooling crystallization. Closed-loop WMADNC approach does not show any improvement in the startup of either applications, which indicates that closed-loop control is redundant for wet milling during startup. However, during steady-state operation, the proposed WMADNC is proved to be an excellent approach to tailor CLD by adjusting the total chord counts set-point in both applications therefore can directly control the CQA. In addition, WMADNC can provide very fast and effective disturbance rejections. Same CLD can be achieved consistently under disturbances in both processes using WMADNC upstream or downstream.

Acknowledgements Eli Lilly and Company is acknowledged for providing financial support and MSMPRC transfer units. In addition, we gratefully acknowledge Christopher Burcham, Christopher Polster and Daniel Jarmer of Eli Lilly and Company for input and discussions.

References

25



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 32

(1) Acevedo, D.; Nagy, Z. K. Systematic Classification of Unseeded Batch Crystallization Systems for Achievable Shape and Size Analysis. J. Cryst. Growth 2014, 394, 97. (2) King, J. C.; Li, H.; Grover, M. A.; Kawajiri, Y.; Rousseau, R. W. Optimization of Two-Stage Cooling Profile in Unseeded Batch Cooling Crystallization. IFAC-PapersOnLine 2015, 48, 297. (3) Yang, Y.; Nagy, Z. K. Model-Based Systematic Design and Analysis Approach for Unseeded Combined Cooling and Antisolvent Crystallization (CCAC) Systems. Cryst. Growth Des. 2014, 14, 687. (4) Woo, X. Y.; Tan, R. B. H.; Braatz, R. D. Precise Tailoring of the Crystal Size Distribution by Controlled Growth and Continuous Seeding from Impinging Jet Crystallizers. CrystEngComm 2011, 13, 2006. (5) Li, H.; Kawajiri, Y.; Grover, M. A.; Rousseau, R. W. Application of an Empirical FBRM Model to Estimate Crystal Size Distributions in Batch Crystallization. Cryst. Growth Des. 2014, 14, 607. (6) Li, H.; Grover, M. A.; Kawajiri, Y.; Rousseau, R. W. Development of an Empirical Method Relating Crystal Size Distributions and FBRM Measurements. Chem. Eng. Sci. 2013, 89, 142. (7 ) Simone, E.; Zhang, W.; Nagy, Z. K. Application of Process Analytical Technology-Based Feedback Control Strategies to Improve Purity and Size Distribution in Biopharmaceutical Crystallization. Cryst. Growth Des. 2015, 15, 2908. (8) Saleemi, A. N.; Rielly, C. D.; Nagy, Z. K. Comparative Investigation of Supersaturation and Automated Direct Nucleation Control of Crystal Size Distributions Using ATR-UV/vis Spectroscopy and FBRM. Cryst. Growth Des. 2012, 12, 1792. (9 ) Burcham, C. L.; Collins, P. C.; Jarmer, D. J.; Seibert, K. D. Formulation and Analytical Development for Low-Dose Oral Drug Products; Zheng, J., Eds.; John Wiley & Sons Inc.: Hoboken, NJ, 2008; Chapter 8, pp 205.

26


Page 27 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(10) Engstrom, J.; Wang, C.; Lai, C.; Sweeney, J. Introduction of a New Scaling Approach for Particle Size Reduction in Toothed Rotor-Stator Wet Mills. Int. J. Pharm. 2013, 456, 261. (11) Harter, A.; Schenck, L.; Lee, I.; Cote, A. High-Shear Rotor-Stator Wet Milling for Drug Substances: Expanding Capability with Improved Scalability. Org. Process Res. Dev. 2013, 17, 1335. (12) Kamahara, T.; Takasuga, M.; Tung, H. H.; Hanaki, K.; Fukunaka, T.; Izzo, B.; Nakada, J.; Yabuki, Y.; Kato, Y. Generation of Fine Pharmaceutical Particles via Controlled Secondary Nucleation under High Shear Environment During Crystallization

Process Development and

Scale-up. Org. Process Res. Dev. 2007, 11, 699. (13 ) Gradl, J.; Schwarzer, H. C.; Schwertfirm, F.; Manhart, M.; Peukert, W. Precipitation of Nanoparticles in a T-Mixer: Coupling the Particle Population Dynamics with Hydrodynamics through Direct Numerical Simulation. Chem. Eng. Process. 2006, 45, 908. (14) Choi, Y.; Chung, S.; Oh, M.; Kim, H. Investigation of Crystallization in a Jet Y-Mixer by a Hybrid Computational Fluid Dynamics and Process Simulation Approach. Cryst. Growth Des. 2005, 5, 959. (15) Woo, X. Y.; Tan, R. B. H.; Braatz, R. D. Modeling and Computational Fluid Dynamics Population Balance Equation Micromixing Simulation of Impinging Jet Crystallizers. Cryst. Growth Des. 2009, 9, 156. (16) Jiang, Mo.; Wong, M. H.; Zhu, Z.; Zhang, J.; Zhou, L.; Wang, K.; Versypt, A. N. F.; Si, T.; Hasenberg, L. M.; Li, Y.; Braatz, R. D. Towards Achieving a Flattop Crystal Size Distribution by Continuous Seeding and Controlled Growth. Chem. Eng. Sci. 2012, 77, 2. (17) Pharmaceutical Quality for the 21st Century: A Risk-Based Approach; U.S. Food and Drug Administration/Department of Health and Human Services: Silver Springs, MD, 2007.

27



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 32

(18 ) Yang, Y.; Nagy, Z. K. Combined Cooling and Antisolvent Crystallization in Continuous Mixed Suspension, Mixed Product Removal Cascade Crystallizers: Steady-State and Startup Optimization. Ind. Eng. Chem. Res. 2015, 54, 5673. (19 ) Yang, Y.; Nagy, Z. K. Advanced Control Approaches for Combined Cooling/Antisolvent Crystallization in Continuous Mixed Suspension Mixed Product Removal Cascade Crystallizers. Chem. Eng. Sci. 2015, 127, 362. (20) Koswara, A.; Nagy, Z. K. Anti-Fouling Control of Plug-Flow Crystallization via Heating and Cooling Cycle. IFAC-PapersOnLine 2015, 48, 193. (21) Peña, R.; Nagy, Z. K. Process Intensification through Continuous Spherical Crystallization Using a Two-Stage Mixed Suspension Mixed Product Removal (MSMPR) System. Cryst. Growth Des. 2015, 15, 4225. (22 ) Yang, Y.; Nagy, Z. K. Application of Nonlinear Model Predictive Control in Continuous Crystallization Systems. American Control Conference (ACC) 2015, 4282. (23) Li, J.; Trout, B. L.; Myerson, A. S. Multistage Continuous Mixed-Suspension, Mixed Product Removal (MSMPR) Crystallization with Solids Recycle. Org. Process Res. Dev. 2016, 20, 510. (24) McGlone, T.; Briggs, N. E. B.; Clark, C. A.; Brown, C. J.; Sefcik, J.; Florence, A. J. Oscillatory Flow Reactors (OFRs) for Continuous Manufacturing and Crystallization. Org. Process Res. Dev. 2015, 19, 1186. (25) Jiang, M.; Zhu, Z.; Jimenez, E.; Papageorgiou, C. D.; Waetzig, J.; Hardy, A.; Langston, M.; Braatz, R. D. Continuous-Flow Tubular Crystallization in Slugs Spontaneously Induced by Hydrodynamics. Cryst. Growth. Des. 2014, 14, 851. (26) Jiang, M.; Papageorgiou, C. D.; Waetzig, J.; Hardy, A.; Langston, M.; Braatz, R. D. Indirect Ultrasonication in Continuous Slug-Flow Crystallization. Cryst. Growth Des. 2015, 15, 2486.

28


Page 29 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(27) Randolph, A. D. The Mixed Suspension, Mixed Product Removal Crystallizer as a Concept in Crystallizer Design. AICHE J. 1965, 11, 424. (28) Alvarez, A. J.; Singh, A.; Myerson, A. S. Crystallization of Cyclosporine in a Multistage Continuous MSMPR Crystallizer. Cryst. Growth Des. 2011, 11, 4392. (29) Ferguson, S.; Ortner, F.; Quon, J.; Peeva, L.; Livingston, A.; Trout, B. L.; Myerson, A. S. Use of Continuous MSMPR Crystallization with Integrated Nanofiltration Membrane Recycle for Enhanced Yield and Purity in API Crystallization. Cryst. Growth Des. 2014, 14, 617. (30 ) Hou, G.; Power, G.; Barrett, M.; Glennon, B.; Morris, G.; Zhao, Y. Development and Characterization of a Single Stage Mixed-Suspension, Mixed-Product-Removal Crystallization Process with a Novel Transfer Unit. Cryst. Growth Des. 2014, 14, 1782. (31) Polster, C. S.; Cole, K. P.; Burcham, C. L.; Campbell, B. M.; Frederick, A. L.; Hansen, M. M.; Harding, M.; Heller, M. R.; Miller, M. T.; Phillips, J. L.; Pollock, P. M.; Zaborenko, N. Pilot-Scale Continuous Production of LY2886721: Amide Formation and Reactive Crystallization. Org. Process Res. Dev. 2014, 18, 1295. ( 32 ) Power, G.; Hou, G.; Kamaraju, V. K.; Morris, G.; Zhao, Y.; Glennon, B. Design and Optimization of a Multistage Continuous Cooling Mixed Suspension, Mixed Product Removal Crystallizer. Chem. Eng. Sci. 2015, 133, 125. (33) Quon, J. L.; Zhang, H.; Alvarez, A.; Evans, J.; Myerson, A. S.; Trout, B. L. Continuous Crystallization of Aliskiren Hemifumarate. Cryst. Growth Des. 2012, 12, 3036. (34) Su, Q.; Nagy, Z. K.; Rielly, C. D. Pharmaceutical Crystallization Processes from Batch to Continuous Operation Using MSMPR Stages: Modelling, Design, and Control. Chem. Eng. Process. Process Intensif. 2015, 89, 41. (35 ) Wong, S. Y.; Tatusko, A. P.; Trout, B. L.; Myerson, A. S. Development of Continuous

29



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 32

Processes Using a Single-Stage Mixed-Suspension, Mixed-Product Removal Crystallizer with Recycle. Cryst. Growth Des. 2012, 12, 5701. (36) Zhang, H.; Quon, J.; Alvarez, A. J.; Evans, J.; Myerson, A. S.; Trout, B. Development of Continuous Anti-Solvent/Cooling Crystallization Process Using Cascaded Mixed Suspension, Mixed Product Removal Crystallizers. Org. Process Res. Dev. 2012, 16, 915. (37) Vetter, T.; Burcham, C. L.; Doherty, M. F. Regions of Attainable Particle Size in Continuous and Batch Crystallization Processes. Chem. Eng. Sci. 2014, 106, 167. (38) Mydlarz, J.; Briedis, D. Modeling of Crystal Size Distribution in a Mixed Suspension Mixed Product Removal Crystallizer When Growth is Size Dependent. Sep. Technol. 1993, 3, 212. (39 ) Mashingaidze, T. A.; Garside, J.; Tavare, N. S. A Continuous Mixed Suspension Mixed Product Removal Dissolver. Chem. Eng. Comm. 1987, 62, 221. (40) Nishio, S.; Kavanagh, J. P.; Garside, J. A Small-scale Continuous Mixed Suspension Mixed Product Removal Crystallizer. Chem. Eng. Sci. 1991, 46, 709. (41) Chemaly, Z.; Muhr, H.; Fick, M. Crystallization Kinetics of Calcium Lactate in a MixedSuspension-Mixed-Product Removal Crystallizer. Ind. Eng. Chem. Res. 1999, 38, 2803. (42) Yang, Y.; Song, L.; Gao, T.; Nagy, Z. K. Integrated Upstream and Downstream Application of Wet Milling with Continuous Mixed Suspension Mixed Product Removal Crystallization. Cryst. Growth Des. 2015, 15, 5879. (43) Yang, Y.; Song, L.; Nagy, Z. K. Automated Direct Nucleation Control in Continuous Mixed Suspension Mixed Product Removal Cooling Crystallization. Cryst. Growth Des. 2015, 15, 5839. (44) Mitchell, N. A.; Frawley, P. J. Nucleation Kinetics of Paracetamol Ethanol Solutions from Metastable Zone Widths. J. Cryst. Growth 2010, 312, 2740. (45) Frawley, P. J.; Mitchell, N. A.; Ó’Ciardhá, C. T.; Hutton, K. W. The Effects of Supersaturation,

30


Page 31 of 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Temperature, Agitation and Seed Surface Area on the Secondary Nucleation of Paracetamol in Ethanol Solutions. Chem. Eng. Sci. 2012, 75, 183. (46) Nanubolu, J. B.; Burley, J. C. Investigating the Recrystallization Behavior of Amorphous Paracetamol by Variable Temperature Raman Studies and Surface Raman Mapping. Mol. Pharmaceutics 2012, 9, 1544. (47) Thorley, F. C.; Baldwin, K. J.; Lee, D. C.; Batchelder, D. N. Dependence of the Raman Spectra of Drug Substances upon Laser Excitation Wavelength. J. Raman Spectrosc. 2006, 37, 335. (48) Barthe, S. C.; Grover, M. A.; Rousseau, R. W. Observation of Polymorphic Change through Analysis of FBRM Data: Transformation of Paracetamol from Form II to Form I. Cryst. Growth Des. 2008, 8, 3316. (49) Lo, E.; Mattas, E.; Wei, C.; Kacsur, D.; Chen, C. Simultaneous API Particle Size Reduction and Polymorph Transformation Using High Shear. Org. Process Res. Dev. 2012, 16, 102.

31



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For Table of Contents Only

32


Page 32 of 32

Application of Wet Milling-Based Automated Direct Nucleation Control

Recommend Documents