Application of preferential crystallization for racemic compound integrating thermodynamics, kinetics and optimization 3

Chapter Crystallization thermodynamics: solubility and metastable zone width of mandelic acid and ketoprofen 61 4.1 Introduction The prerequisite for direct crystallization from solution is the knowledge of the solid-liquid equilibrium of the two enantiomers in a solvent. Therefore, the solubility data is very important for the crystallization in solution. The knowledge and use of solubility properties of enantiomer mixtures is not only a matter of theoretical interest, but also allows one to carry out resolutions in a more rational manner than usually obtainable (Jacques et al., 1981). In Chapter 3, the binary phase diagram for mandelic acid and ketoprofen were constructed. But in the solution crystallization, the solubility ternary phase diagram is essential for design and optimization of a reasonable separation process. For each crystallization process including the chiral crystallization, the most important requirement is supersaturation, which is the driving force for nucleation to occur. The quantity often used to define the state of supersaturation is the metastable zone width. The metastable zone width is very essential parameter to control the crystal size and can be considered as a characteristic property of crystallization (Kim and Mersmann, 2001, Mullin, 2001). It is well known that supersaturated solution can exhibit a metastable zone, which constitutes the allowable supersaturation level during every crystallization process, shown in Fig. 4.1. 62 Fig 4.1 Typical super solubility diagram. Line AC: Cooling a saturated solution; Line AC': Concentrating a saturated solution by evaporation of solvent; Line AC'': A combination of cooling and evaporation. The diagram can be described in term of three distinct zones: 1. The stable zone of an undersaturated solution where no nucleation or crystal growth is possible. Existing crystal will simply dissolve, 2. The supersaturated metastable zone where growth may occur but spontaneous nucleation does not occur, and 3. The labile supersaturated zone of spontaneous and rapid nucleation. On the cooling crystallization, cooling the solution with concentration C from the initial point A will lead to the point B on the solubility curve. At this point the solution becomes saturated. On further cooling the system becomes supersaturated but spontaneous nucleation to form crystals does not occur immediately in this region until a point C on the metastable boundary is reached. 63 The metastable zone width is essential parameter for the growth of large size crystal from solution, since it is the direct measure of stability of the solution in its supersaturated region. Larger the zone width, the stability is higher (Bukley, 1951; Zaitseva, 1995). In addition, crystal growth rate is a function of supersaturation, that is, the higher the supersaturation, the higher the growth rate. Nucleation or formation of fine crystals is also driven by the supersaturation. Therefore, controlling the supersaturation level over time can aid control of the levels of growth and nucleation achieved, hence presenting the opportunity to control the size of the crystals produced. In view of thermodynamics aspects, for preferential crystallization process, it is crucial to keep the freedom of supersaturation of the racemate in its metastable zone to avoid its spontaneous nucleation. On the other hand, it is also important to control the supersaturation of the target enantiomer as the spontaneous nucleation of the target enantiomer may easily initiate the spontaneous nucleation of its racemate. Therefore, the solubility diagrams and supersaturation data are the thermodynamic basis for preferential crystallization for the racemic compound (Hongo et al., 1976, 1981; Jacques et al., 1981; Black et al., 1989; Collins et al., 1997; Collet, 1999). Though metastable zone width is very important factor for the whole crystallization process, it is thermodynamically not founded and kinetically not well defined. It depends on a number of parameters such as temperature level, rate of generating the supersaturation, solution history, fluid dynamics, the presence of foreign particles and impurities etc (Nyvlt et al.,1985; Ulrich and Strege, 2002).Therefore, the metastable zone width should be measured under a specific set of operation condition. The polythermal technique is the most widely used technique for determining the metastable zone width, involving cooling a saturated solution at a 64 fixed rate until nucleation occurs (Nyvlt, 1968; Sohnel and Nyvlt, 1975; Mullin and Jancic, 1979; Kim and Ryu, 1997; Mullin, 2001). In this chapter, the solubility data under different temperature, ternary phase diagram and metastable zone width under different condition were measured for the mandelic acid and ketoprofen. The newly developed Lasentec Focused Beam Reflectance Measurement (FBRM) was used to determine the nucleation and dissolution properties. 4.2 Method and Experiments 4.2.1 Apparatus The crystallization experiments were carried out in an automatic laboratory reactor system equipped with an l-L glass jacketed crystallizer, shown in Fig. 4.2 (Wang and Ching, 2006). 65 Fig 4.2 Experimental set-up: 1– Computer; 2- LabMax; 3- Julabo cooling/heating refrigerator; 4- Heavy duty stirrer; 5- Condenser; 6- Temperature sensor; 7- liter double-wall crystallizer; 8- FBRM; 9- Balance; 10- Solvent bottle; 11- Dosing pump. 4.2.2 Solubility and metastable zone width (MSZW) To determine the solubility data in solvent, the first step is to select a suitable solvent. The choice of solvent is crucial to the whole crystallization process. Many factors must be considered and some factors must inevitably be compromised in the selection process. Some of the main points include the following: (a) the solute to be crystallized should be readily soluble in the solvent. (b) it should also be easily 66 deposited from the solution in the desired crystalline form after cooling, evaporation, salting-out with an additive, etc. (c) a mixture of two or more solvents will occasionally be found to possess the best properties for a particular crystallization purpose (Mullin, 2001). In this study, the solubility is characteristic factor to get the pure enantiomer product. It should not too low and too high. According to the suitable solubility properties, the water was selected as solvent for mandelic acid and the mixture solvent of ethanol and water (volume 0.9:1.0) was selected for ketoprofen. For the solubility measurement, the polythermal method was used for determining the solubilities of mandelic acid with different mole percent of (S)-MA (100%, 80%, 70% and 50%) in water and the solubilities of ketoprofen with different mole percent of (S)-Kp (50%, 75%, 86%, 91%, 92%, 93%, 96%, 100%) in the mixture solvent of ethanol and water. The known amounts of samples were put into the double-wall crystallizer. The known volumes of solvents then were added. These volumes were not enough to dissolve the solute at the highest temperature of interest. The solutions were stirred, and the temperatures were maintained by LabMax. Subsequently, solvents were added in intervals of at least hours. On approaching the equilibrium, it was necessary to decrease the additions to 0.1ml, until all solute was dissolved. The solubility can be calculated accordingly. The error of this solubility mainly came from adding the solvent. It may bring the solvent’s volume error in the solubility calculation, so it was important to decrease the times of adding solvent. Therefore, it is necessary to measure the rough value of solubility before we began to measure the exact value using this adding solvent method. Error propagation was accomplished and the error of the experimental solubility data was less than 4%. For the mandelic acid, the experiments were performed in a temperature range between to 35 oC. The metastable zone widths of these solutions were determined 67 under different cooling rates 2, 5, and 10oC/h at a constant stirrer speed of 300 rpm. For the ketoprofen, the experiments temperature range was between 15 to 30 oC. The metastable zone widths of ketoprofen were determined under cooling rates 2oC/h at a constant stirrer speed rate of 300 rpm. The maximum allowable subcooling (Tdissolution – Tnucleation) was determined as MSZW. Nucleation and dissolution were detected by a Lasentec S400A Focused Beam Reflectance Method probe. With the consideration of possible metastable conglomerate formed and the fast transition rate into the stable form racemic compound under the studied experimental conditions, the measurement of saturation temperature was given enough time to reach the equilibrium with respect to the stable form racemic compound. The experimental data were all repeated at less three times to check the reproducibility. During the repeated measurements, the deviation of solubility and MSZW was less than 4%. 4.3 Results and discussion 4.3.1 Solubility and metastable zone width for the mandelic acid 4.3.1.1 Solubility and ternary phase diagram for mandelic acid The solubilities of 50%, 70%, 80% and 100% mole percent of (S)-MA were determined in water by polythermal method. During this process, the experimental time was long enough (more than two days) to avoid the appearance of unstable conglomerate mandelic acid (Profir et al., 2002). 68 According to the solubility theory, the solubility is corresponding to the melting point in some way. One species normally needs more energy to melt, when it has high melting temperature. On the other hand, this species is also difficult to be resolved and needs more solvent to resolve itself. Usually the higher melting point leads to lower solubility. Therefore, the ternary phase diagram may have a shape similar to that of the binary melting point phase diagram. All solubility results are presented in Table 4.1 and shown in Fig. 4.3, respectively. When the solubility data were transferred to the mass fraction of solvent, (S)-MA and (R)-MA according to the densities, the ternary phase diagram of MA in water was derived, shown in Fig. 4.4. Table 4.1 Solubility data (g/ml solvent) for different mole percent of (S)-MA Mole percent of (S)-MA o Temperature( C) 100% 80% 70% 50% 35 0.176 0.448 0.874 0.579 30 0.130 0.256 0.477 0.318 25 0.104 0.169 0.270 0.202 20 0.087 0.126 0.180 0.143 15 0.076 0.101 0.135 0.116 10 0.065 0.081 0.106 0.091 0.058 0.071 0.091 0.073 69 Solubility (g/ml water) S-MA 0.8 80% S-MA 70% S-MA 0.6 RS-MA 0.4 0.2 0 10 20 30 40 o Temperature ( C) Fig 4.3 Solubility of MA with different mole percent in water. Fig 4.4 The ternary phase diagram of MA in Water. 70 they did not, more time had to be allowed solution to get the equilibrium (Mullin, 2001). The experiments were carried out with a Shimadzu SCL-10AVP chromatographic system (Kyoto, Japan). The column used was chiralpak AD-H purchased from Daicel Chemical Industries, LTD. The experiment was conducted at room temperature about 24°C. The mobile phase contained 90%hexane and 10% IPA. The flow rate was 0.8ml/min. The solubility results obtained by HPLC and polythermal methods are agreement with each other. All solubility data are shown in Table 4.2. To draw the ternary phase diagram, the solubility data were transferred to the mass fraction of solvent, (S)-Kp and (R)-Kp. Fig. 4.9 shows the ternary phase diagram of ketoprofen. 77 Table 4.2 Solubility data for different mole percent of (S)-Kp Temperature (oC) 30 25 20 15 Mole percent of (S)-Kp Solubility (mg/ml solvent) 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 80.40 84.74 87.27 88.30 90.30 86.67 63.49 49.55 68.23 69.44 71.64 73.10 74.60 70.27 38.83 31.05 57.84 59.52 60.00 60.10 61.04 59.77 25.97 19.91 45.25 46.30 46.60 46.98 47.00 46.02 17.17 15.21 Mass Mass fraction of fraction of Solvent (S)-Kp (×100) (×100) 91.94 8.06 91.54 8.12 91.31 8.08 91.22 8.08 91.04 8.15 91.37 7.43 93.53 4.85 94.88 2.56 93.08 6.92 92.96 6.76 92.76 6.74 92.62 6.79 92.48 6.84 92.89 6.11 95.94 3.04 96.73 1.64 94.07 5.93 93.91 5.85 93.86 5.71 93.85 5.66 93.76 5.68 93.88 5.26 97.24 2.06 97.88 1.06 95.30 4.70 95.20 4.61 95.17 4.49 95.13 4.48 95.13 4.43 95.22 4.11 98.16 1.38 98.37 0.815 Mass fraction of (R)-Kp (×100) 0.34 0.61 0.70 0.81 1.20 1.62 2.56 0.28 0.50 0.59 0.68 1.00 1.02 1.63 0.24 0.43 0.49 0.56 0.86 0.70 1.06 0.19 0.34 0.39 0.44 0.67 0.46 0.815 78 Solvent 100 99 98 97 96 95 94 93 92 91 90 R-Kp 10 S-Kp Fig 4.9 Ternary phase diagram of (S)-Kp and 0.9:1.0 (vol) mixture solvent of ethanol and water, ■, 15oC; ♦, 20oC; ▲, 25oC; ●, 30oC. From the solubility data of Kp, it was found that all solubility of Kp increased greatly with the temperature, which may indicate that cooling crystallization method can be used for Kp. However the solubility of (RS)-Kp is much lower than that of (S)Kp, which means ketoprofen is kind of unfavorable racemic compound for the crystallization resolution. The ternary phase diagram of ketoprofen also can prove it because its shape was a typical ternary phase diagram of unfavorable racemic compound. From the ternary phase diagram, the eutectic point was about 91% mole percent of (S)-Kp, which was similar with that of binary phase diagram (91.6%). It suggests that the other separation method, such as chromatography, should first 79 enrich (S)-enantiomer composition to or above the eutectic point when the pure enantiomer would be obtained by the preferential crystallization. 4.3.2.2 Metastable zone width for ketoprofen For the sample from 1.0 to 0.86 mole fraction of (S)-Kp, the metastable zone widths were all narrow and about or below oC, while for the two samples of 75% and 50% mole fraction of (S)-Kp, the metastable zone widths were large, from 10oC to 16oC, shown in Fig. 4.10. Solubility(75%) metastable zone width(75%) o 35 Solubility(50%) 30 Metastable zone width(50%) ) C 25 ( e r 20 u t a r 15 e p m e 10 T 10 20 30 40 50 Solubility(mg/ml) 60 70 80 Fig 4.10 Solubility and Metastable zone width of 50% and 75% mole percent of S-Kp samples. According to the results of the metastable zone width of ketoprofen, the crystallization process may encounter with one problem that is narrow metastable zone width. Solution with narrow metastable zone width may bring problems to crystallization process. It is unstable solution in which it is hard to control the 80 crystallization process. The crystallization process very often ends up with secondary nucleation, which affects the growth of main crystal significantly (Rajesh, 2002). Therefore, effort was taken to study on the effects of the metastable zone of (RS)- and (S)-Kp, and 0.94 mole fraction of (S)-Kp in order to enhance the MSZW. Some factors were studied, such as temperature, cooling rate, stirring rate and volume ratio of mixed solvent. Through the design of experiment, the effect which may most greatly or most lightly enhance the metastable zone will be decided. 4.3.2.3 Fractional Experiment Design All results of any possible experiments that could be run in a complex process, varying all the controlling factors at all their levels, would add up to a large group of numbers. For example, in this study, the metastable zone is a process controlled by four main factors, namely temperature, cooling rate, stirring rate and volume ratio of mixed solvent. Each of the four factors can be used at three levels. For temperature, the three levels are 20 oC, 25 oC, and 30 oC; for cooling rate, these are 6.0 , 9.6, and 12.0 oC/h; 300, 500, and 700 rpm are for stirring rate and 1:0.6, 1:0.7, and 1:0.8 are for volume ratio of mixed solvent (water : ethanol). All the possible experiments would add up to a total of 81 runs. It is hard for anyone to take much time to every run of all 81 experiments. Fortunately, fractional experiment design could be a tremendously valuable tool for exploring new process or gain detailed comprehension of existing ones, and then optimizing those processes (Del Vecchio, 1977). With the exploration of the analysis of certain patterns of numbers known as matrices, scientists demonstrated that it was possible to extract a large fraction of the information in a matrix from a smaller fraction of the numbers in that matrix. Based 81 on this mathematical concept, the collection of experimental data, each one of which is the result of a single possible experiment, can be considered to be a mathematical matrix. This means that running a fraction of all those possible experiments will still allow an investigator to learn almost as much as if he or she had run them all. The 9run design (L9) was used in this study. That means that doing as few as of those 81 experiments can reveal a great deal about the process and furnish almost as much information as doing all 81might. In this study, for using conveniently, three levels of each main factor for L9 fractional experiment design are coded as -1, 0, and 1, which are shown in Table 4.3. The design of L9 with coded levels of factor is shown in Table 4.4. For example, in run 1, the level code of each of four factors is -1. Therefore, cross-checking with Table 4.3, the l level of the volume ratio of mixed solvent (water: ethanol) should be 1:0.6. The temperature, stirring rate and cooling rate could be 20 oC, 300rpm, and 6oC/h respectively. According to this design, the simplest way to analyze the data is to graph the average result of each level versus the other levels for each factor in the matrix. The details are described later. 82 Table 4.3 Levels for each of four factors Volume ratio of solvent Temperature Stirring rate Cooling rate ( water : ethanol) (oC) (rpm) (oC/h) 1:0.6 20 300 6.0 1:0.7 25 500 9.6 -1 1:0.8 30 700 12.0 Level Table 4.4 The design of L9 with coded levels of factor Run Volume ratio of solvent Temperature Stirring rate Cooling rate ( water : ethanol) -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 83 The solubility of (RS)- and (S)-Kp, and 0.94 mole fraction of (S)-Kp was measured and shown in Table 4.5. Some of these results were verified by HPLC method. The solubility values obtained by HPLC and adding solvent methods are in agreement with each other. Table 4.5 The solubility data of (RS)-Kp, 0.94 mole fraction of (S)-Kp, and (S)-Kp (mg/ml) Volume ratio of solvent ( water : ethanol) Temperature 20oC 25oC 30oC Sample 1:0.6 1:0.7 1:0.8 (RS)-Kp 4.70 7.62 12.02 0.94 (S)-Kp 13.95 19.50 24.22 (S)-Kp 13.38 19.23 23.95 (RS)-Kp 7.11 11.32 18.23 0.94 (S)-Kp 18.03 24.10 40.64 (S)-Kp 17.84 23.74 40.00 (RS)-Kp 9.07 14.45 31.95 0.94 (S)-Kp 21.86 28.95 50.28 (S)-Kp 21.38 28.44 51.23 4.3.2.4 Analysis of the effects influencing metastable zone width of ketoprofen According to the L9 experiment design, the metastable zone widths obtained for nine runs are shown in Table 4.6. These data obtained are regard as a mathematical matrix. According to the theory of fractional experiment design, the simplest way to 84 analyze the data is to compare the average result of each level with the other levels for each factor in the matrix. If no obvious contrast could be seen between the three level averages of one factor, then this factor is unlikely to exert any significant effect on the MSZW process. The difference between the highest and lowest of three averages of one factor may be used as an indicator on how this factor can affect the process. A distinct difference reveals that the corresponding factor has a large effect on MSZW process. Table 4.6 Worksheet for a L9 design for metastable zone widths Volume ratio of Metastable zone width Stirring solvent Temperature (water : (oC) Run Cooling (∆oC) rate rate (rpm) (oC/h) ethanol) (RS)- 0.94 (S)-Kp Kp (S)-Kp 1:0.6 20 300 6.0 12.0 4.5 4.5 1:0.6 25 500 9.6 13.5 4.8 4.8 1:0.6 30 700 12.0 13.0 5.0 5.0 1:0.7 20 500 12.0 13.5 2.0 2.0 1:0.7 25 700 6.0 12.5 1.8 1.8 1:0.7 30 300 9.6 10.0 1.5 1.5 1:0.8 20 700 9.6 11.5 1.8 1.8 1:0.8 25 300 12.0 11.0 1.5 1.5 1:0.8 30 500 6.0 9.0 1.5 1.5 85 Therefore, the analysis of the MSZW data in Table 4.6 would involve taking the averages of each subgroup of three according to the levels of each factor. For instance, for volume ratio of solvent (water: ethanol), the first three runs used the ratio 1:0.6 (-1 level). Therefore, these three runs belong to one subgroup for this factor. Run 4, 5, and run 7, 8, then belong to another two subgroups for the level and 1. The total and average values of each subgroup can be calculated. For the level -1 of sample (RS)-Kp, the total MSZW value and average MSZW value are respectively 38.5 oC and 12.83 oC for the volume ratio factor. If these calculations are done for all the four control factors, the resulting numbers come out as shown in Table 4.7. Table 4.7 Calculation average responses for three-level experiment for (RS)-Kp Volume ratio of Temperature Stirring rate Cooling rate solvent (water : ethanol) -1 total 38.5 37.0 33.0 33.5 total 36.0 37.0 36.0 35.0 total 31.5 32.0 37.0 37.5 -1 average 12.83 12.33 11.00 11.17 average 12.00 12.33 12.00 11.67 average 10.50 10.67 12.33 12.50 difference 2.33 1.66 1.33 1.33 86 For the samples 0.94 mole fraction of (S)-Kp and (S)-Kp, the calculation results were the same as shown in Table 4.8, because the MSZW results for these two samples were the same. Except these calculation results, most of the time, it is more convenient to examine the data in the sort of combined graph for all four factors illustrated in Fig. 4.11 for (RS)-Kp and Fig. 4.12 for 0.94 mole fraction of (S)-Kp and (S)-Kp. Table 4.8 Calculation average responses for three-level experiment for 0.94 mole fraction of (S)-Kp and (S)-Kp Volume ratio of Temperature Stirring rate Cooling rate solvent (water : ethanol) -1 total 14.3 8.3 7.5 7.8 total 5.3 8.1 8.3 8.1 total 4.8 8.0 8.6 8.5 -1 average 4.77 2.77 2.50 2.60 average 1.77 2.70 2.77 2.70 average 1.60 2.67 2.87 2.73 difference 3.17 0.10 0.37 0.23 87 14 M SZW (c) 13 12 11 Volume ratio of solvent Temperature 10 Stirring rate Cooling rate -1 Fig 4.11 Effect chart of L9 design for (RS)-Kp. 4.5 M SZW (c) 3.5 2.5 Volume ratio of solvent Temperature Stirring rate 1.5 Cooling rate -1 Fig 4.12 Effect chart of L9 design for 0.94 mole fraction of (S)-Kp and (S)-Kp. 88 It was found from Tables 4.7 and 4.8 that the average MSZW differences for ethanol volume ratio of solvent were very large and reached 2.33 for (RS)-Kp and 3.17 for 0.94 mole fraction of (S)-Kp and (S)-Kp. These results indicate that the factor of ethanol volume ratio may have a real effect on the MSZW process. From Figs. 4.11 and 4.12, it can be seen that the response lines for ethanol volume ratio of solvent declined sharply. With the increase of the volume ratio of ethanol, the MSZW decreased. The MSZW for three different levels of volume ratio showed very differently, with the first level (1:0.6) standing out clearly as promoting higher MSZW than the other two. All these results can be attributed to the influence of solution concentration on MSZW process. Adding ethanol can increase the concentration of Kp in the solution, so Kp may have lower concentration when the ethanol ratio is lower. Furthermore, the concentration of a solution can influence the MSZW greatly. The MSZW is the maximum undercooling of a solution expressed as the difference between saturation temperature and the temperature of the first solute precipitation. The first solute precipitation may depend on the probability of the two or more solute particles approaching each other at a small distance. Within such a distance, these solute particles can connect with each other and form clusters or nuclei. When the concentration of the solution is low, the solute particles may be relatively far away from each other and they need high supersaturation or undercooling to form the cluster. Therefore, the MSZW could be larger in a low solution concentration, that is to say, the MSZW could be larger in the low ethanol ratio solution. Compared with the ethanol volume ratio, temperature had smaller average differences of MSZW, which were 1.66 for (RS)-Kp and 0.10 for 0.94 mole fraction of (S)-Kp or (S)-Kp. This result suggests that the temperature may have less influence on the MSZW process than ethanol ratio has. In addition, the response lines for 89 temperature decreased very lightly. At lower temperature the MSZW was large and at higher temperature the MSZW was low. All these results may account for the change of the concentration of Kp with temperature. The Kp solubility increases with the increase of temperature. When the temperature is low, the Kp solubility or concentration is low. It is this low concentration that may result in high undercooling and large MSZW. For the stirring rate, the response lines increased very slightly and the average differences were also very small. These results indicate that the stirring rate may have very little influence on the MSZW process. Why the stirring rate can affect MSZW probably related to the mechanical action of the magnetic stirrer on the solution. When the solution is stirred by the magnetic stirrer, the solute particles can easily approach each other and form clusters or nuclei. Therefore, the first solute precipitation needs smaller undercooling and the MSZW may be lower. However, these mechanical actions exert an effect on the solution in some range of concentration. In the case of this experiment, the concentration may be out of this range, so the stirring rate has little effect on MSZW. The Cooling rate is also an important factor which can influence the MSZW. From Figs.4.11 and 4.12, it can be seen that the response lines for cooling rate both increased very lightly, which suggests that cooling rate may affect the MSZW process also very slightly. The larger the cooling rate was, the larger the MSZW was. A possible reason for this result is that the growth of a nucleus to a visible size requires a finite time. When cooling rate is rapid enough in the MSZW measurement process, the finite time for the growth of a nucleus to a visible size can be ignored, which may result in a large MSZW (Hussain, 2001). 90 Therefore, if we want to enhance the MSZW based on this study, the optimal experiment parameter selection in three different levels should be at the water/ethanol ratio of 1:0.6, at 20oC, at a stirring rate of 700rpm, and at a cooling rate of 12oC/h. In order to show that the MSZW can be further enhanced in these optimized conditions, the metastable zone width was measured at these conditions. The results showed that the MSZWs (∆0C) of (RS)-Kp and (S)-Kp were 15.0 and 5.5oC, respectively. Both of these two results obtained at the optimized conditions are higher than the maximum MSZW for (RS)-Kp and (S)-Kp (13.5 and 5.0oC) obtained in the fractional experiments. However, the MSZW was just 5.5oC even at the stirring rate 700 rpm and 12 o C/h cooling rate. It is still not high enough for the crystallization process. These conditions are not practical for the preferential crystallization process. It means that it is hard to control the supsaturation level of pure ketoprofen enantiomer to inhibit its spontaneous nucleation during the crystallization process. Therefore, it is hard to control the crystal size distribution and the shape of final products. 4.4 Conclusion The solubilities, ternary phase diagram and metastable zone width at different mole percent of S enantiomer were obtained for mandelic acid in water and ketoprofen in mixed solvent of ethanol and water with volume ration 0.9:1.0. The lasentec FBRM were use to detect nucleation and dissolution. The MSZWs of the racemic compound were different with racemate and pure enantiomer. For the mandelic acid, the solubility ratio of (RS) to (S)-MA decreases with the temperature in the range 35-5oC which suggests that the preferential crystallization of 91 mandelic acid in the chosen solvent is more favorable as temperature decreases. In addition, the solubility of (RS)-MA is much higher than that of (S)-MA, which shows the similar characteristics of solubility as more favorable racemic compound system. The ternary phase diagram of mandelic acid showed a typical shape of more favorable racemic compound, which further prove that the mandelic acid is a kind of more favorable racemic compound. All the MSZW results of mandelic Acid were higher than oC which are favorable for preferential crystallization process. The solubility curve and ternary phase diagram of ketoprofen show the typical properties of an unfavorable racemic compound system. The MSZW of high mole percent ketoprofen were all narrow. It is hard to control the supsaturation level of ketoprofen pure enantiomer to inhibit its spontaneous nucleation during the crystallization process. Therefore, it is hard to control the crystal size distribution and the shape of final products. Through the study of the main four effects of MSZW by fractional experiment design, it is obvious that the water/ethanol volume ratio in the solvent should have the most significant effect on the MSZW process of Kp. MSZW would increase with decreasing ethanol ratio. The temperature, stirring rate and cooling rate may affect the process slightly. MSZW would also be inversely proportional to the temperature, but would be directly proportional to the cooling rate. However, the MSZW of high mole percent ketoprofen are still not satisfied even in the optimal condition derived by fractional experiment design. 92 [...]... -1 total 38 .5 37 .0 33 .0 33 .5 0 total 36 .0 37 .0 36 .0 35 .0 1 total 31 .5 32 .0 37 .0 37 .5 -1 average 12. 83 12 .33 11.00 11.17 0 average 12.00 12 .33 12.00 11.67 1 average 10.50 10.67 12 .33 12.50 difference 2 .33 1.66 1 .33 1 .33 86 For the samples 0.94 mole fraction of (S)-Kp and (S)-Kp, the calculation results were the same as shown in Table 4.8, because the MSZW results for these two samples were the same Except... fraction of fraction of Solvent (S)-Kp (×100) (×100) 91.94 8.06 91.54 8.12 91 .31 8.08 91.22 8.08 91.04 8.15 91 .37 7. 43 93. 53 4.85 94.88 2.56 93. 08 6.92 92.96 6.76 92.76 6.74 92.62 6.79 92.48 6.84 92.89 6.11 95.94 3. 04 96. 73 1.64 94.07 5. 93 93. 91 5.85 93. 86 5.71 93. 85 5.66 93. 76 5.68 93. 88 5.26 97.24 2.06 97.88 1.06 95 .30 4.70 95.20 4.61 95.17 4.49 95. 13 4.48 95. 13 4. 43 95.22 4.11 98.16 1 .38 98 .37 0.815... (S)-Kp 13. 95 19.50 24.22 13. 38 19. 23 23. 95 (RS)-Kp 7.11 11 .32 18. 23 0.94 (S)-Kp 18. 03 24.10 40.64 (S)-Kp 17.84 23. 74 40.00 (RS)-Kp 9.07 14.45 31 .95 0.94 (S)-Kp 21.86 28.95 50.28 (S)-Kp 30 oC 4.70 (S)-Kp 25oC 1:0.8 (RS)-Kp 20oC 1:0.7 21 .38 28.44 51. 23 4 .3. 2.4 Analysis of the effects influencing metastable zone width of ketoprofen According to the L9 experiment design, the metastable zone widths obtained for. .. Effect chart of L9 design for 0.94 mole fraction of (S)-Kp and (S)-Kp 88 It was found from Tables 4.7 and 4.8 that the average MSZW differences for ethanol volume ratio of solvent were very large and reached 2 .33 for (RS)-Kp and 3. 17 for 0.94 mole fraction of (S)-Kp and (S)-Kp These results indicate that the factor of ethanol volume ratio may have a real effect on the MSZW process From Figs 4.11 and 4.12,... that of (S)-MA, which shows the similar characteristics of solubility as more favorable racemic compound system The ternary phase diagram of mandelic acid showed a typical shape of more favorable racemic compound, which further prove that the mandelic acid is a kind of more favorable racemic compound All the MSZW results of mandelic Acid were higher than 5 oC which are favorable for preferential crystallization. .. range from 5 -35 oC, which indicates that the chosen solvent are suitable for the crystallization operation and cooling preferential crystallization method can be used for the mandelic acid The solubility ratio of (RS) to (S)-MA decreases from circa 3. 3 at 35 oC to circa 1 .3 at 5 oC When the solubility ratio of (RS) to (S)-MA decreases, the destabilization process of the solution is decreased and the possibility... stirring rate and volume ratio of mixed solvent Each of the four factors can be used at three levels For temperature, the three levels are 20 oC, 25 oC, and 30 oC; for cooling rate, these are 6.0 , 9.6, and 12.0 oC/h; 30 0, 500, and 700 rpm are for stirring rate and 1:0.6, 1:0.7, and 1:0.8 are for volume ratio of mixed solvent (water : ethanol) All the possible experiments would add up to a total of 81 runs... mole percent of (S)-Kp Temperature (oC) 30 25 20 15 Mole percent of (S)-Kp Solubility (mg/ml solvent) 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 100% 96% 93% 92% 91% 86% 75% 50% 80.40 84.74 87.27 88 .30 90 .30 86.67 63. 49 49.55 68. 23 69.44 71.64 73. 10 74.60 70.27 38 . 83 31.05 57.84 59.52 60.00 60.10 61.04 59.77 25.97 19.91 45.25 46 .30 46.60 46.98... ethanol) -1 total 14 .3 8 .3 7.5 7.8 0 total 5 .3 8.1 8 .3 8.1 1 total 4.8 8.0 8.6 8.5 -1 average 4.77 2.77 2.50 2.60 0 average 1.77 2.70 2.77 2.70 1 average 1.60 2.67 2.87 2. 73 difference 3. 17 0.10 0 .37 0. 23 87 14 M SZW (c) 13 12 11 Volume ratio of solvent Temperature 10 Stirring rate Cooling rate 9 -1 0 1 Fig 4.11 Effect chart of L9 design for (RS)-Kp 5 4.5 4 M SZW (c) 3. 5 3 2.5 Volume ratio of solvent 2 Temperature... value and average MSZW value are respectively 38 .5 oC and 12. 83 oC for the volume ratio factor If these calculations are done for all the four control factors, the resulting numbers come out as shown in Table 4.7 Table 4.7 Calculation average responses for three-level experiment for (RS)-Kp Volume ratio of Temperature Stirring rate Cooling rate solvent (water : ethanol) -1 total 38 .5 37 .0 33 .0 33 .5 0 . 8.12 0 .34 93% 87.27 91 .31 8.08 0.61 92% 88 .30 91.22 8.08 0.70 91% 90 .30 91.04 8.15 0.81 86% 86.67 91 .37 7. 43 1.20 75% 63. 49 93. 53 4.85 1.62 30 50% 49.55 94.88 2.56 2.56 100% 68. 23 93. 08. of mandelic Acid are higher than 5 o C which are favorable for preferential crystallization process. 76 4 .3. 2 Solubility and metastable zone width for ketoprofen 4 .3. 2.1 Solubility and. 4%. 4 .3 Results and discussion 4 .3. 1 Solubility and metastable zone width for the mandelic acid 4 .3. 1.1 Solubility and ternary phase diagram for mandelic acid The solubilities of 50%,