English

• More than 38.8 million people had been infected with human immunodeficiency virus (HIV)-1 in 2015 [1]. The anti-AIDS drug market will experience a period of strong growth due to the rising infected rate of HIV patients and the launch of new drugs. As an integrase inhibitor, dolutegravir (DTG-7) was developed for treating HIV-1 infection [2]. Literature data on pharmaceutical production indicates that over 90% of the active pharmaceutical ingredients (APIs) exist in forms of organic crystals [3, 4]. Therefore, crystallization process serves as one of key unit operations that importantly determines the production quality. Moreover, the outcome of crystallization directly impacts the downstream operations of filtration, drying, milling and compaction [4, 5]. In consideration of needle-like dolutegravir sodium crystals obtained only by industrial batch crystallization, it is promising to improve crystal quality and optimize morphology of dolutegravir sodium.

  Batch crystallization is a widely adopted operation in pharmaceutical crystallization [6]. Differences in process control between multiple batches can cause inconsistent crystal size distributions (CSD), structures and habits, thus affecting downstream operation and corresponding product quality [7]. Recently, continuous flow is widely used in electrochemistry and crystallization. Peng and coworkers utilized continuous-flow electrochemistry for promoting Aza-Wacker cyclization with high selectivity [8]. According to a report by the US Food and Drug Administration (FDA), continuous crystallization enables quality control improvement of pharmaceutical production [9, 10]. Moreover, continuous crystallization technology has been recognized in a rapidly increasing trend due to its intrinsic advantages, such as enhanced stability in production quality, improved consistency in crystal morphology, reduced expenditures, as well as excellent potential for transportation and scaling-up due to the modular design [11]. In recent years, it is becoming increasingly popular for continuous crystallization in both industrial and academic research laboratories [12]. However, existing continuous crystallization techniques suffer from poor bioavailability [13-15], low production efficiency [16, 17], unsatisfying yield, fouling and blockage issues, as well as small production scale [18-20]. These inherent limitations drive innovation in the development of continuous crystallizer [21-23].

  From a pharmaceutical application perspective, a desired reaction crystallization system can address following aspects: 1) understanding kinetics of reaction and crystallization to regulate crystal morphology and size distribution; 2) equipping in-situ visualization tools to control quality [24, 25] such as microscope imaging system, X-ray diffraction [26], ultraviolet-visible (UV–vis) spectroscopy [27] and Raman detector [28, 29]; 3) simplifying operation process to improve production efficiency and precise synthesis to meet high-end customized demand. On the other hand, microreactor continuous crystallization is a highly competitive method to obtain high value-added API, but it has not yet been realized to extensively produce dolutegravir sodium.

  Microreactors are becoming an emerging technique for flow chemistry processes, especially in pharmaceutical production where it has proven their uniqueness and high applicability [30]. Microreactors enable precise control of flow [31, 32], mass transfer, heat transfer [30, 33] and reaction within confined internal space with typical characteristic diameters of 10−500 µm. Compared with batch crystallization, continuous microreactor crystallization demonstrates notable advantages, which include shorter reaction time, improved purity, smaller sample volume [34], precise process control [34, 35], as well as better selectivity [8] owing to high-throughput heat and mass transfer efficiency [36, 37]. Li et al. presented a novel microreactor technology that continuous preparation of nanoporous particles in microfluidic chip was monitored in situ by ultraviolet–visible absorption spectroscopy [27]. In addition, droplet microreactors sequence can provide homogeneous mixing [38-40] for continuous crystallization and screening conditions [14, 41], easily benefiting from confined internal space in droplet. However, separating API from droplets and carrier fluid is a hurdle for the industrial application of continuous crystallization due to redundant capital expenditures. It is worth noting that liquid-gas heterogeneous microreactors can eliminate separation step of carrier fluid and chemical slurry to improve the production efficiency. Introducing gas can generate Taylor flow, which isolates the chemicals reagents in liquid phase within microchannels. The chaotic advection appears in liquid phase by internal circulations with Taylor flow pattern [42-44], which reinforces mixing performance [45] of liquid phase, aggrandizes surface ratios of area to volume [44] and improves mass transfer efficiency in microchannels [46]. Furthermore, liquid-gas heterogeneous microreactors can avoid tedious manual operation and inconsistent product attributes of crystals. In recent studies, it has been shown that the introduction of gas promotes the mixing of polymer solutions [44]. The tubular crystallizer via creating liquid-gas flow gets smaller insulin crystals in comparison with other methods [47], but there is a potential risk that the crystals may incessantly adhere to the inner wall of microchannels. Continuous crystallization using liquid-gas heterogeneous microreactors can eliminate the issue of crystal blocking microchannels by strictly regulating supersaturation, gas flow rate and residence time. We proposed that crystal nucleation stage of API is isolated in the microchannels and crystal nuclei are swept away by liquid-gas flow in time to avoid problem of crystal blockage. The continuous manufacture using liquid-gas heterogeneous microreactor will be able to put into production and bring prosperity for pharmaceutical production.

  This work focuses on developing a liquid-gas heterogeneous microreactor and associated approach which simultaneously unravels reaction kinetics and mixing performance, as well as continuously synthesizes dolutegravir sodium crystals. The reaction kinetics and mixing intensify mechanism were directly quantified through transforming fluorescent signals mapping to kinetics and mixing parameters by involvement of fluorescent dye tracer. We proposed a liquid-gas dispersing and mixing strategy to reduce risk of crystal blockage in microchannels. Furthermore, we investigated the influence of initial supersaturation, gas flow rate, and ultrasound treatment on crystal size distribution, determining the optimal continuous crystallization conditions of dolutegravir sodium. In addition, we demonstrated initial supersaturation can modify the crystal morphology, and revealed the crystal morphology mechanisms.

  The continuous microreactor reactive crystallization was conducted to yield microscale particles of dolutegravir sodium. As Fig. 1a depicts, the formulation project was divided into serval modules for different purposes. Gas pressure regulator was connected to gas buffer and gas flow regulator, which guided nitrogen to microreactor chip using polyfluoroalkoxy (PFA) tubing. Microscope with camera was connected to computer so as to record the experimental data. Microreactor glass chip was placed on customized Hastelloy manifold that is heated by temperature regulator. The outlet of microreactor chip was associated with slurry collector. The design of Hastelloy manifold is illustrated on Supporting information.

  Figure 1

  

  Figure 1.  Illustration of experimental setup for liquid-gas heterogeneous microreactor. (a) The schematic diagram of microreactor apparatus for dolutegravir sodium continuous crystallization. The magnified image Ⅰ represents the Y-junction flow pattern of nitrogen and sodium hydroxide. Ⅱ is the mixture of liquid-gas flow and dolutegravir within T-junction structure. Orange area and blue area of syringe pump represent dolutegravir (DTG-7) dissolved in iso-propyl alcohol and sodium hydroxide solution, respectively. (b) The microreactor glass chip with microchannels sealed in manifold is displayed perpendicular to the microscope view.

  DownLoad: Full-Size Img PowerPoint

  As Fig. 1b depicts, the two reagents were injected by syringe pumps (Harvard Pump 11 Elite, United States) to microchannel A (depth 230 µm, width 600 µm) and microchannel B (depth 190 µm, width 490 µm), respectively. Nitrogen was supplied to another microchannel C (depth 190 µm, width 490 µm) at preset pressure via a constant pressure regulator (FluidicLab PC1, China) and gas flow rate was precisely controlled by gas flow regulator (Alicat KM7100, United States). Based on liquid-gas mixing method, the sodium hydroxide solution sheared off gas to shape gas-fluid Taylor flow through Y-junction micromixer and the adjacent nitrogen slug filled with NaOH solution. Then liquid-gas heterogeneous flow mixed with dolutegravir (DTG-7) solution through the second T-junction micromixer inside which violently mixing occurred and supersaturated solution of dolutegravir sodium (DTG) appeared. In this mode, supersaturated solution was taken away quickly by Taylor flow through the second micromixer under excellent micromixing conditions, hardly causing crystals to clog in downstream microchannel (depth 230 µm, width 800 µm). In the case of controlling the supersaturated solution concentration and gas flow rate, only nucleation occurred in microreactors chip. The liquid-gas heterogeneous crystallization procedure and mechanisms characterization were recorded by a CMOS camera (POC panda 4.2 M−USB−PCO, United States) paired with a microscope (Nikon Eclipse E400, Japan). The crystal size distributions were obtained from a laser particle size analyzer (Microtrac SYNC S3500SI, United States). A scanning electron microscope (Nova Nano SEM 450, United States) was employed to identify crystal morphology. The preparation pathway of dolutegravir sodium (Fig. S1 in Supporting information) and preparation method are presented in Supporting information.

  In order to reduce measurement time and improve measurement accuracy, we characterized the reaction kinetics in the first Y-junction microchannel of glass chip. BCFL acid fluorescein as a pH-sensitive fluorescent indicator was added to NaOH solution in the microchannel, where the light intensity increases as the pH value increases. The two streams of NaOH solution containing BCFL Acid and DTG-7 solution were severally injected into microchannels at 0.02 mL/min and 0.02 mL/min, maintained at 66.1 ℃, as Fig. 2a depicts. In this image, one pixel indicates 1.605 µm, and different analysis pixel positions of downstream mean different reaction time in this picture. Therefore, selecting an area in axial direction at downstream mixing section for gray value analysis established relation between NaOH concentration field and reaction time, where the width is 490 µm and the depth is 190 µm. The complete time of reaction can be calculated by measuring length from Y-junction to end of reaction (Fig. 2b). In order to obtain the correlation between light intensity and NaOH concentration, the calibration equation of light intensity and pH value was characterized in the pH range of 6.0-12.45, as shown in Fig. 3a. The variation of fluorescence light intensity corresponded to neutralization reaction process. The concentration profile at selected area of axial direction for reaction time is shown in Fig. 3b at reaction temperature of 66.1 ℃. Fitting trend of experimental data displays that curve prefers the exponential model than linear model.

  Figure 2

  

  Figure 2.  pH-sensitive dye indicates in situ the reaction kinetics of the dolutegravir with sodium hydroxide under steady-state conditions. (a) The blue arrows represent the direction of inlet streams entering Y-junction microchannel. Light region and dark region mean NaOH solution and dolutegravir solution, respectively. Fluorescence light intensity of red dashed line along axial direction of microchannel was selected for indicating changes of time and sodium hydroxide concentration. (b) Red dashed rectangle area represents the end of reaction process downstream.

  DownLoad: Full-Size Img PowerPoint

  Figure 3

  

  Figure 3.  The processing of experimental data plots for reaction kinetics parameters. (a) Fluorescence light intensity as a calibration curve of pH value. (b) Concentration profiles (C/C₀) at selecting certain analysis distance for reaction time of sodium hydroxide downstream.

  DownLoad: Full-Size Img PowerPoint

  Besides, the research of Duan et al. [48] demonstrates that neutralization reaction of acetic acid corresponds to first order reaction kinetics model. Specifically, the determination coefficients R² and reaction rate constant k of zeroth order reaction and first order reaction kinetics models are calculated by fitting the experimental data at 66.1, 50 and 40 ℃. Summaries results are listed in Table S1 (Supporting information), in which a series of R₁² simulated by first order reaction kinetics are higher than those of the zero-order kinetics R₀¹ at 66.1, 50 and 40 ℃. Dolutegravir sodium reaction kinetics is more consistent with the first order kinetics:

  where C_OH⁻ is the existing concentration of sodium hydroxide which can be calculated by pH value. k represents reaction rate constant of first order kinetics and t is reaction time.

  Further we studied the activation energy E_a of dolutegravir sodium reaction system to evaluate the barrier of the reaction occurring. The linear fit of experiment data is illustrated in Fig. S2 (Supporting information). The activation energy value and exponential factor of dolutegravir sodium reaction are 12.630 kJ/mol and 813.300, respectively. Generally, owing to dominant position of diffusion-control reaction, there is a lower activation energy of less than 20 kJ/mol. Whereas reaction-rate controlled reaction kinetics has higher activation energy [48, 49]. As a result, the activation energy acquired by pH-sensitive fluorescent dyes further sustains that dolutegravir sodium reaction belongs to a diffusion-control reaction. Since the reaction rate is larger than the diffusion rate, the key factor to improve the reaction procedure needs to enhance the mixing effect of two fluid streams.

  When concentration of OH¯ is decreased by 97% of original concentration, it is considered as the complete time (t_0.97) which takes for the reaction to be basically completed can be presented by the following equation:

  After completion of reaction, t = t_0.97 and the relationship exists here:

  here k_evg represents the average reaction rate constant at 66.1 ℃. We obtained the complete time (t_0.97) of reaction is 0.381 s. Simultaneously, Fig. 2b displays that the calculated time end of reaction process is 0.655 s. The complete time (t_0.97) is close to the calculated time that image is captured, which indicates the imaging of adding fluorescent dyes is suitable for in situ characterizing reaction kinetics.

  For fluid flow in microchannels, a great deal of work has been devoted to improving and characterizing mixing performance. The key factors that affect mixing performance include microchannel structure and introduction of the gas. We further quantified mixing effect, which is enhanced by gas flow rate in heterogeneous continuous microreactor. BCFL acid, a pH-sensitive fluorescent indicator, was applied to determine OH¯ concentration field in microchannel. When the fluorescence molecules were excited by light source, the mixing performance in microchannels can be characterized in situ under the optical imaging system. The fluorescence intensity distribution map of the mixed two fluids can be obtained with advantages of rapid measurement and simple operation.

  In order to avoid the interference of reaction facilitation, DTG-7 was not added to fluid 1. Two entrances of Y-junction were nitrogen (0.02 mL/min) and sodium hydroxide solution with fluorescence indicator (0.02 mL/min), respectively. Then liquid-gas Taylor flow formed in Y-junction. In the T-junction, one inlet was the mixture of iso-propyl alcohol and ultrapure water without DTG-7, the other was light-dark Taylor flow. We characterized uniform mixing on the side of the channel away from upstream that was sodium hydroxide solution with fluorescent tracer. Further, under the conditions of the same experience temperature and the same flow rate, we used the feature points where fluorescence intensities started to fluctuate steadily between 900 and 1050 as the representation of uniform mixing. In view of microchannel edge there is shadow presented in image, and the measured curves offset along the channel edge for 40 µm. From the absence of nitrogen to introducing nitrogen and increasing flow rate, mixing performance is observed in Fig. 4a. Meanwhile, the relationship between fluorescence intensity and mixing distance was obtained by ImageJ, as indicated in Fig. 4b. Velocity of flow and mixing distances in the channel were calculated to quantify the mixing time at different flow rates, as shown in Fig. 4c. It can be seen that the mixing time reduces significantly from 1.170 s to 0.111 s when the nitrogen flow rate increases from 0 mL/min to 0.3 mL/min. The two streams are greatly enhanced in mixing by liquid-gas heterogeneous microreactor. In addition, with the continuous increase of flow rate, the mixing efficiency has been improving. However, mixing time reduces weakly to 0.026 s as flow rate keeps increasing to 1.25 mL/min. Therefore, the initial introduction of gas phase makes mixing more violent in the microchannel. However, the diminishing marginal utility of reducing mixing time becomes very poor as the gas flow rate continues to increase.

  Figure 4

  

  Figure 4.  The effect of varying gas flow rates on mixing performance of the two fluids. (a) Internal fluorescence light intensity distribution captured by camera corresponds to changes of local concentrations. Yellow dashed curves designate measuring parts. (b) Changing tendency of emitted fluorescence intensity is analyzed by ImageJ at selected mixing distances. (c) Fitting curve for the relation between gas flow rate and mixing time in microchannel.

  DownLoad: Full-Size Img PowerPoint

  We gained solubility curve of dolutegravir sodium dissolved in ultrapure water and iso-propyl alcohol to explore the potential correlations with microreactor continuous crystallization process. The solubility curve and measurement operation are presented in Fig. 5a and Supporting information, respectively. Experimental operation zone exists on condition that volume fraction of iso-propyl alcohol ranges from 70% to 80%. In this area, the solubility curve was fitted well with solubility data.

  Figure 5

  

  Figure 5.  Investigating the effect of initial supersaturation on crystal size. (a) The change of DTG solubility along with iso-propyl alcohol at 66.1 ℃ in the mixture of ultrapure water and iso-propyl alcohol. (b) Crystal D90 as tendency curves of initial supersaturation at conditions of ultrasound and without ultrasound.

  DownLoad: Full-Size Img PowerPoint

  Liquid-gas heterogeneous continuous microreactor reactive crystallization ran in the microchannel. Flow rate of nitrogen and DTG-7 solution were maintained at 0.95 mL/min and 0.075 mL/min, respectively. Initial supersaturation in the reaction system was further controlled by regulating the flow rate of sodium hydroxide solution. As the crystals nucleated constantly, the slurry was carried out by liquid-gas flow into slurry collector. The gathered sample slurry was centrifuged as soon as the continuous crystallization lasted for 60 min. Each continuous crystallization of DTG can run for 5 h. The operation conditions of liquid-gas heterogeneous continuous microreactor reactive crystallization are given in Table S2 (Supporting information).

  The D90 of samples without ultrasound treatment exhibit the following patterns in Fig. 5b, when the initial supersaturation reaches the range of 1.25-2.00. The effect of initial supersaturation on crystal size distribution (CSD) is found that the crystal size distribution of dolutegravir sodium remains almost unchanged at D90, then it goes up sharply to peak and down rapidly with initial supersaturation increase. Due to the agglomeration, the D90 reversely increases. The D90 reaches a maximum of 109.6 µm where the initial supersaturation is 1.58. The trend indicates that the nucleation rate and crystal growth rate are both low under low initial supersaturation, where there is no significant gain to change initial supersaturation from 1.26 to 1.35 for D90. However, the D90 increases rapidly corresponding to nucleation rate and crystal growth rate, and finally reaches the peak value with the increase of initial supersaturation from 1.35 to 1.58. After, the positive impact of nucleation rate on decreasing crystal size is more significant than crystal growth rate. At high supersaturation, a large number of nuclei are generated, which have more superficial area to provide attachment sites for crystal growing. However, the DTG concentration around the crystal nucleus supplies deficiently for the crystal growth surface, causing precipitous reduction in crystal size and continuous decline to the smallest D90. As the initial supersaturation further increases, agglomeration occurs on the smaller crystals in the supersaturation range of 1.70-2.00, resulting in the larger crystal sizes and a reverse increase in D90. The effect of the initial supersaturation of 2.00 on agglomeration is presented in Fig. S3 (Supporting information).

  Ultrasound is an important factor to influence crystal size and has the potential to alleviate particle agglomeration. Therefore, liquid-gas heterogeneous continuous crystal slurry samples with supersaturation in the range of 1.58-2.47 were processed by centrifugation and were added ultrasonic treatment for 20 min. The D90 data obtained by the laser particle size analyzer is plotted as a red curve in Fig. 5b. Since the D90 does not increase in reverse at high supersaturation, it can be seen that ultrasound treatment can alleviate crystal agglomeration to reduce crystal size. On the one hand, the trend of D90 has a peak with the increase of supersaturation, which is partially consistent with the pattern of control group samples (without ultrasound treatment). On the other hand, a decline in peak of D90 appears under higher supersaturation, opposed to control group. A series of data for crystal D90 with no ultrasound treatment or ultrasound treatment are listed in Table S3 (Supporting information).

  The introduction of the gas flow enhances the mixing performance within the microchannel, attributed to liquid-gas inhomogeneous flow providing a strong mixing condition. At a constant liquid flow rate, we explored the effect of nitrogen on the crystal distribution size by controlling the nitrogen flow rates of 0.3, 0.6, 0.95 and 1.25 mL/min, respectively. The liquid-gas flow captured from camera shows that when the nitrogen flow rate reaches up to 0.95 mL/min, the Taylor flow begins to deform due to high gas flow rate (Fig. S4 in Supporting information). Here infers the unstable flow pattern situates between Taylor flow and annular flow. As the gas flow rate further increases, it tends to evolve to Taylor-annular flow so that mixing performance improves faintly. Fig. 6a conveys the correlation between the CSD and gas flow rate without ultrasound treatment. We used (D90 − D10)/D50 as an indicator to measure the peak width of the CSD. When initial supersaturation is 1.85, the peak width is calculated to be 1.84, 1.61, 1.26 and 1.41 at gas flow rate of 0.3, 0.6, 0.95 and 1.25 mL/min, respectively. Under the condition of the same initial supersaturation, larger gas flow rate results in narrower CSD when the gas flow rate is controlled to 0.3, 0.6 or 0.95 mL /min. In contrast, as the gas flow rate continues to increase, the CSD widens. Moreover, the greater gas flow rate obtains the smaller mean crystal size in the experimental system, in which initial supersaturation is 1.35, 1.58 and 1.85 with ultrasound treatment, as shown in Fig. 6b. Meanwhile, except at gas flow rate of 1.25 mL/min, the trends of crystal mean size having the peaks at 0.3, 0.6, 0.95 mL/min are consistent with the previous experimental results (Fig. 5b). Based on the research, introducing a higher gas flow rate tends to arouse a smaller crystal size and a narrower crystal size distribution, but too high gas flow rate will make the crystal size distribution wider. Therefore, a narrow crystal size distribution requires that the gas flow rate should be controlled to a suitable value.

  Figure 6

  

  Figure 6.  The effect of gas flow rate on continuous crystallization. (a) Effect of different gas flow rates on crystal size distribution. (b) Effect of gas flow rate on crystal mean size at initial supersaturation of 1.35 mL/min, 1.58 mL/min, 1.85 mL/min.

  DownLoad: Full-Size Img PowerPoint

  It is vitally important to control the crystal morphology and size of DTG that can affect the bioavailability of active pharmaceutical ingredients. The factors impacting crystal morphology mainly include crystallization process, polarity of solution, additive and concentration [50]. Optimizing the crystal morphology can not only improve the efficacy of active pharmaceutical ingredients, but also avert the problems caused by needle-like crystals for downstream pharmaceutical process, including challenges in crystal filtration, tableting and breakage [5, 50, 51]. Consequently, we investigated the effect of initial supersaturation on the crystal habit of DTG with 0.95 mL/min of gas flow rate and ultrasound treatment. The rod-shaped crystals were observed by optical microscopy. The average aspect ratios obtained in liquid-gas heterogeneous continuous crystallization are presented in Fig. 7 with the increase of initial supersaturation. When supersaturation reaches 1.58, the average aspect ratio of DTG is 2.81. However, when supersaturation is 2.46, the average aspect ratio is 4.82. The aspect ratio of the crystal increases with an increase in supersaturation. This information detail implied that supersaturation has larger impact on long crystal faces than on wide faces. Based on the findings by Klapwijk et al. [51], a possible mechanism explaining on results is that growth rate perpendicular to long axis direction of crystal faces can develop faster due to demanding smaller growth area than other wide axis directions of crystal face with increase of initial supersaturation. The verification of crystal form and discussion of crystal surface are illustrated in Fig. S4 (Supporting information). In terms of obtaining small aspect ratio of crystal to ensure subsequent compaction properties and improve bioavailability of medicines, it is necessary to select the optimal supersaturation conditions to control crystal habits into rod-like shape instead of needle-like shape.

  Figure 7

  

  Figure 7.  The crystal aspect ratio distributions and microscope images for four initial supersaturations of DTG. The crystals in images are randomly selected for measurement. S, N severally represent initial supersaturation and the counted number of crystals. AAR means average aspect ratio of crystals.

  DownLoad: Full-Size Img PowerPoint

  According to the proportion of ingredients and manipulation provided by the manufacturer, the yield of dolutegravir sodium is 96.15% in batch crystallization. A series of yields for dolutegravir sodium prepared by liquid-gas heterogeneous microreactor are shown in Fig. 8. When supersaturation reaches to 1.58, yield of dolutegravir sodium is 98.1%, which exceeds the yield of dolutegravir sodium obtained by existing batch crystallization.

  Figure 8

  

  Figure 8.  The yields at different supersaturation for dolutegravir sodium of continuous preparation in liquid-gas heterogeneous microreactor.

  DownLoad: Full-Size Img PowerPoint

  Compared with batch crystallization, continuous preparation of dolutegravir sodium using liquid-gas heterogenous microreactor has superior qualities in crystal size, morphology and crystal size distribution. The optimized dolutegravir crystals using liquid-gas heterogenous microreactor and commercial product by batch crystallization were characterized as shown in Figs. 9a and b, respectively. When supersaturation is 1.58, the crystal D90 of dolutegravir sodium is 20.14 µm, which is smaller than the crystal D90 (32.75 µm) of commercial product. In addition, dolutegravir sodium tends towards narrower crystal size distribution and smaller aspect ratio of crystals, which are attributed to continuous preparation and supersaturation control in microreactor. The crystal morphology is successfully controlled to rod-like shape instead of needle-like shape by continuous crystallization using liquid-gas heterogenous microreactor, as illustrated in Fig. S5 (Supporting information). Thus, crystal qualities of dolutegravir sodium are significantly improved by liquid-gas heterogenous microreactor under conditions of guaranteeing high yield.

  Figure 9

  

  Figure 9.  The crystal properties of dolutegravir sodium. (a) Continuous preparation of dolutegravir sodium in microreactor. (b) Commercial products of dolutegravir sodium in batch crystallization.

  DownLoad: Full-Size Img PowerPoint

  In this study, our novel liquid-gas heterogeneous microreactor achieves the continuous crystallization of dolutegravir sodium for more than 5 h, as well as integrates a rapid evaluation of reaction kinetics characterization and mixing intensify mechanism. The reaction process conforms to the first order kinetics model with a low activation energy value. The introduction of gas can greatly improve the mixing efficiency of reactants while reducing the deposition of crystals in the microchannel. With the increase of supersaturation, D90 of dolutegravir sodium remains almost unchanged at first, then spikes to a peak and drops sharply, and finally increases reversely, which is associated with agglomeration of crystals. Ultrasonic treatment can relieve crystal agglomeration and increase the crystal dispersion, so that effect of high supersaturation on D90 is suppressed. Large gas flow rate tends to produce small crystals and narrow crystal size distributions, whereas the continuous augment of gas flow rate causes wider crystal size distributions. This work informs that excessively increasing gas flow rate should be controlled in industry, not only to ensure mixing intensity and small residence time, but also to obtain small crystal size and a narrow crystal size distribution. In terms of crystal morphology, large supersaturation can increase the aspect ratio of crystal and make the crystal surface rough. Modifying morphology under moderate supersaturation without paying the price for yield, can improve pharmaceutical preparation of dolutegravir sodium and prevent crystal clogging in the microchannel.

  In brief, besides ensuring high yield of dolutegravir sodium, continuous crystallization of liquid-gas heterogeneous microreactor can optimize crystal size distribution, reduce crystal size and improve crystal quality. The optimal supersaturation of continuous crystallization is 1.58 and the optimal gas flow rate is 0.95 mL/min with ultrasound treatment by using the liquid-gas heterogeneous microreactor.

  The fundamental insights of the crystallization behavior and underlying mechanisms can provide significant guidance for the existing continuous industrial crystallization. We expect to see the findings of continuous crystallization using liquid-gas heterogenous microreactor can be extended to industry and other research fields.

  Declaration of competing interest

  We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.

  Acknowledgments

  This work is supported by the Natural Science Foundation of the Science and Technology Commission of Shanghai Municipality (No. 19ZR1472200), and National Natural Science Foundation of China (No. 22178072).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.06.059.

  
  
• 1. [1]

     S.W. Keeshin, J. Feinberg, Expert Opin. Drug Saf. 14 (2015) 141–147. doi: 10.1517/14740338.2015.973845

  2. [2]

     W. Albrecht, J. Geier, D. Perez, WO 2016102078 A1, 2016.

  3. [3]

     T. Wang, H. Lu, J. Wang, et al., J. Ind. Eng. Chem. 54 (2017) 14–29. doi: 10.1016/j.jiec.2017.06.009

  4. [4]

     A.J. Alvarez, A.S. Myerson, Cryst. Growth Des. 10 (2010) 2219–2228. doi: 10.1021/cg901496s

  5. [5]

     D. Han, T. Karmakar, Z. Bjelobrk, J. Gong, M. Parrinello, Chem. Eng. Sci. 204 (2019) 320–328. doi: 10.1016/j.ces.2018.10.022

  6. [6]

     H.H. Tung, E.L. Paul, M. Midler, J.A. Mccauley, Critical issues in crystallization practice, in: H.H. Tung, E.L. Paul, M. Midler, J.A. Mccauley (Eds. ), Crystallization of Organic Compounds: An Industrial Perspective, WILEY, 2009, pp. 101–116.

  7. [7]

     S. Lawton, G. Steele, P. Shering, et al., Org. Process Res. Dev. 13 (2009) 1357–1363. doi: 10.1021/op900237x

  8. [8]

     X. Peng, P. Feng, Chinese J. Org. Chem. 41 (2021) 2918–2919. doi: 10.6023/cjoc202100051

  9. [9]

     K. Plumb, Chem. Eng. Res. Des. 83 (2005) 730–738. doi: 10.1205/cherd.04359

  10. [10]

      S.L. Lee, T.F. O'Connor, X. Yang, et al., J. Pharm. Innov. 10 (2015) 191–199. doi: 10.1007/s12247-015-9215-8

  11. [11]

      B. Wood, K.P. Girard, C.S. Polster, D.M. Croker, Org. Process Res. Dev. 23 (2019) 122–144. doi: 10.1021/acs.oprd.8b00319

  12. [12]

      K. A. Powell, A.N. Saleemi, C.D. Rielly, Z.K. Nagy, Chem. Eng. Process. Process Intensif. 97(2015) 195–212. doi: 10.1016/j.cep.2015.01.002

  13. [13]

      K.R. Vandana, Y. Prasanna Raju, V. Harini Chowdary, M. Sushma, N. Vijay Kumar, Saudi Pharm. J. 22 (2014) 283–289. doi: 10.1016/j.jsps.2013.05.004

  14. [14]

      R.M. Tona, T.A.O. McDonald, N. Akhavein, J.D. Larkin, D. Lai, Lab Chip 19 (2019) 2127–2137. doi: 10.1039/c9lc00204a

  15. [15]

      N. Blagden, M. de Matas, P.T. Gavan, P. York, Adv. Drug Deliv. Rev. 59 (2007) 617–630. doi: 10.1016/j.addr.2007.05.011

  16. [16]

      C. Rougeot, J.E. Hein, Org. Process Res. Dev. 19 (2015) 1809–1819. doi: 10.1021/acs.oprd.5b00141

  17. [17]

      M. Fujiwara, Z.K. Nagy, J.W. Chew, R.D. Braatz, J. Process Control. 15 (2005) 493–504. doi: 10.1016/j.jprocont.2004.08.003

  18. [18]

      M.D. Johnson, S.A. May, J.R. Calvin, et al., Org. Process Res. Dev. 16 (2012) 1017–1038. doi: 10.1021/op200362h

  19. [19]

      X. Yang, D. Acevedo, A. Mohammad, et al., Org. Process Res. Dev. 21 (2017) 1021–1033. doi: 10.1021/acs.oprd.7b00130

  20. [20]

      K.A. Powell, A.N. Saleemi, C.D. Rielly, Z.K. Nagy, Org. Process Res. Dev. 20 (2016) 626–636. doi: 10.1021/acs.oprd.5b00373

  21. [21]

      R. Lakerveld, J.J.H. Van Krochten, H.J.M. Kramer, Cryst. Growth Des. 14 (2014) 3264–3275. doi: 10.1021/cg500090g

  22. [22]

      X. Jiang, D. Lu, W. Xiao, et al., AIChE J. 62 (2015) 829–841. doi: 10.1038/hr.2015.73

  23. [23]

      W.J. Liu, C.Y. Ma, X.Z. Wang, Procedia Eng. 102 (2015) 499–507. doi: 10.1016/j.proeng.2015.01.199

  24. [24]

      S. Mascia, P.L. Heider, H. Zhang, et al., Angew. Chem. Int. Ed. 52 (2013) 12359–12363. doi: 10.1002/anie.201305429

  25. [25]

      A.S. Myerson, M. Krumme, M. Nasr, et al., Science 104 (2015) 832–839. doi: 10.1002/jps.24311

  26. [26]

      Y. Ke, Q. Zhang, T. Wang, et al., Nano Energy 73 (2020) 104785. doi: 10.1016/j.nanoen.2020.104785

  27. [27]

      W. Li, Y. Li, W. Zhang, et al., Chin. Chem. Lett. 32 (2021) 1131–1134. doi: 10.1016/j.cclet.2020.09.039

  28. [28]

      L.Y. Pfund, A.J. Matzger, ACS Comb. Sci. 16 (2014) 309–313. doi: 10.1021/co500043q

  29. [29]

      R. Storey, R. Docherty, P. Higginson, et al., Crystallogr. Rev. 10 (2004) 45–56. doi: 10.1080/08893110410001664846

  30. [30]

      K.S. Elvira, X.C. I Solvas, R.C.R. Wootton, A.J. Demello, Nat. Chem. 5 (2013) 905–915. doi: 10.1038/nchem.1753

  31. [31]

      N. Hao, Z. Xu, Y. Nie, et al., Chem. Eng. J. 378 (2019) 122222. doi: 10.1016/j.cej.2019.122222

  32. [32]

      S. Marre, K.F. Jensen, Chem. Soc. Rev. 39 (2010) 1183–1202. doi: 10.1039/b821324k

  33. [33]

      N. Hao, Y. Nie, J.X.J. Zhang, Biomater. Sci. 7 (2019) 2218–2240. doi: 10.1039/c9bm00238c

  34. [34]

      A. Shimoyama, Y. Fujimoto, K. Fukase, Synlett (2011) 2359–2362.

  35. [35]

      T. Asai, A. Takata, Y. Ushiogi, et al., Chem. Lett. 40 (2011) 393–395. doi: 10.1246/cl.2011.393

  36. [36]

      P.D.I. Fletcher, S.J. Haswell, E. Pombo-Villar, et al., Tetrahedron 58 (2002) 4735–4757. doi: 10.1016/S0040-4020(02)00432-5

  37. [37]

      V. Hessel, H. Löwe, Chem. Eng. Technol. 28 (2005) 267–284. doi: 10.1002/ceat.200407167

  38. [38]

      K. Wang, L. Li, P. Xie, G. Luo, React. Chem. Eng. 2 (2017) 611–627. doi: 10.1039/C7RE00082K

  39. [39]

      I. Shestopalov, J.D. Tice, R.F. Ismagilov, Lab Chip 4 (2004) 316–321. doi: 10.1039/b403378g

  40. [40]

      E.M. Chan, A.P. Alivisatos, R.A. Mathies, J. Am. Chem. Soc. 127 (2005) 13854–13861. doi: 10.1021/ja051381p

  41. [41]

      P. Laval, N. Lisai, J.B. Salmon, M. Joanicot, Lab Chip 7 (2007) 829–834. doi: 10.1039/b700799j

  42. [42]

      Z. Qiao, Z. Wang, C. Zhang, et al., AIChE J. 59 (2012) 215–228.

  43. [43]

      N. Shao, A. Gavriilidis, P. Angeli, Chem. Eng. Sci. 64 (2009) 2749–2761. doi: 10.1016/j.ces.2009.01.067

  44. [44]

      L. Zha, M. Shang, M. Qiu, H. Zhang, Y. Su, Chem. Eng. Sci. 195 (2019) 62–73. doi: 10.1016/j.ces.2018.11.043

  45. [45]

      J.S. Zhang, K. Wang, Y.C. Lu, G.S. Luo, Chem. Eng. Process. Process Intensif. 49 (2010) 740–747. doi: 10.1016/j.cep.2009.10.009

  46. [46]

      A. Gunther, S. Khan, M. Thalmann, F. Trachsel, K.F. Jensen, Lab Chip 4 (2004) 278–286. doi: 10.1039/B403982C

  47. [47]

      R. Chen, J. Weng, S.F. Chow, R. Lakerveld, Cryst. Growth Des. 21 (2020) 501–511.

  48. [48]

      Y. Duan, R. Rausa, P. Fiaschi, K.D. Papadopoulos, Tribiology Int. 98 (2016) 94–99. doi: 10.1016/j.triboint.2016.01.053

  49. [49]

      M.M. Scherer, J.C. Westall, M. Ziomek-moroz, P. Tratnyek, Environ. Sci. Technol. 31 (1997) 2385–2391. doi: 10.1021/es960999j

  50. [50]

      L.E. Hatcher, W. Li, P. Payne, et al., Cryst. Growth Des. 20 (2020) 5854–5862. doi: 10.1021/acs.cgd.0c00470

  51. [51]

      A.R. Klapwijk, E. Simone, Z.K. Nagy, C.C. Wilson, Cryst. Growth Des. 16 (2016) 4349–4359. doi: 10.1021/acs.cgd.6b00465

• 

  Figure 1  Illustration of experimental setup for liquid-gas heterogeneous microreactor. (a) The schematic diagram of microreactor apparatus for dolutegravir sodium continuous crystallization. The magnified image Ⅰ represents the Y-junction flow pattern of nitrogen and sodium hydroxide. Ⅱ is the mixture of liquid-gas flow and dolutegravir within T-junction structure. Orange area and blue area of syringe pump represent dolutegravir (DTG-7) dissolved in iso-propyl alcohol and sodium hydroxide solution, respectively. (b) The microreactor glass chip with microchannels sealed in manifold is displayed perpendicular to the microscope view.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  pH-sensitive dye indicates in situ the reaction kinetics of the dolutegravir with sodium hydroxide under steady-state conditions. (a) The blue arrows represent the direction of inlet streams entering Y-junction microchannel. Light region and dark region mean NaOH solution and dolutegravir solution, respectively. Fluorescence light intensity of red dashed line along axial direction of microchannel was selected for indicating changes of time and sodium hydroxide concentration. (b) Red dashed rectangle area represents the end of reaction process downstream.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  The processing of experimental data plots for reaction kinetics parameters. (a) Fluorescence light intensity as a calibration curve of pH value. (b) Concentration profiles (C/C₀) at selecting certain analysis distance for reaction time of sodium hydroxide downstream.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  The effect of varying gas flow rates on mixing performance of the two fluids. (a) Internal fluorescence light intensity distribution captured by camera corresponds to changes of local concentrations. Yellow dashed curves designate measuring parts. (b) Changing tendency of emitted fluorescence intensity is analyzed by ImageJ at selected mixing distances. (c) Fitting curve for the relation between gas flow rate and mixing time in microchannel.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  Investigating the effect of initial supersaturation on crystal size. (a) The change of DTG solubility along with iso-propyl alcohol at 66.1 ℃ in the mixture of ultrapure water and iso-propyl alcohol. (b) Crystal D90 as tendency curves of initial supersaturation at conditions of ultrasound and without ultrasound.

  下载: 全尺寸图片 幻灯片
  

  Figure 6  The effect of gas flow rate on continuous crystallization. (a) Effect of different gas flow rates on crystal size distribution. (b) Effect of gas flow rate on crystal mean size at initial supersaturation of 1.35 mL/min, 1.58 mL/min, 1.85 mL/min.

  下载: 全尺寸图片 幻灯片
  

  Figure 7  The crystal aspect ratio distributions and microscope images for four initial supersaturations of DTG. The crystals in images are randomly selected for measurement. S, N severally represent initial supersaturation and the counted number of crystals. AAR means average aspect ratio of crystals.

  下载: 全尺寸图片 幻灯片
  

  Figure 8  The yields at different supersaturation for dolutegravir sodium of continuous preparation in liquid-gas heterogeneous microreactor.

  下载: 全尺寸图片 幻灯片
  

  Figure 9  The crystal properties of dolutegravir sodium. (a) Continuous preparation of dolutegravir sodium in microreactor. (b) Commercial products of dolutegravir sodium in batch crystallization.

  下载: 全尺寸图片 幻灯片
•