Application of Ultrasound Measurements as PAT Tools for Industrial

Article pubs.acs.org/OPRD

Application of Ultrasound Measurements as PAT Tools for Industrial Crystallization Process Development of Pharmaceutical Compounds Lydia Helmdach,*,† Martin P. Feth,*,‡ and Joachim Ulrich*,† †

Zentrum für Ingenieurwissenschaften, Verfahrenstechnik/TVT, Martin-Luther-Universität Halle-Wittenberg, 06099 Halle, Germany Chemical & Biotechnological Development (C&BD) Frankfurt Chemistry, Sanofi-Aventis Deutschland GmbH, 65926 Frankfurt, Germany

‡

S Supporting Information *

ABSTRACT: Within this work, 12 different pharmaceutical compounds were analyzed by the single-frequency ultrasound measurement technique for its applicability to determine concentrations, as an important process parameter during crystallization processes, or to determine the metastable zone widths, as an important precondition for the development of crystallization processes. The results were compared to the applicability of inorganic and nonpharmaceutical compounds that have been discussed in the literature. It was found that according to the change of ultrasound velocity and adiabatic compressibility, a grouping of compounds can be derived. From this grouping it can be concluded that some organic compounds and especially inorganic compounds show an excellent applicability for concentration determination, while the application for pharmaceutical compounds is most often limited. Furthermore, a cost- and time-efficient possibility is shown for the integration of this technique in a pilot-plant-scale setup. A direct transferability of calibration models developed at the laboratory scale was found as long as the influence of undissolved air/gas was low in the pilot-plant setup.

1. INTRODUCTION The manufacturing of active pharmaceutical ingredients (APIs) often involves final or intermediate products in the solid state produced by precipitation or preferably by crystallization. Crystallization processes are established for purification and simultaneously induce mixture separation.1,2 Over 80% of drug products are produced by at least one crystallization step.3 The product quality of the solid product has to meet stringent specifications, such as particle size distribution, particle shape, crystallographic phase, and purity. These properties have the potential to impact bioavailability of the API.4 Often, however, batch-to-batch variations in crystallization processes are observed, which can lead to bad physicochemical product properties and a decrease in product quality (e.g., wrong size distribution). Such variations can have a strong impact on the processability of the API during subsequent physical treatments, such as filtration, drying, or milling of the material,5,6 and also on the efficiency and profitability of the process. Therefore, up-to-date manufacturing processes follow the quality-by-design approach, which involves the application of process analytical technology (PAT) strategies to reduce identified manufacturing risks that have a negative impact on the product quality. PAT tools such as NIR , MIR, and Raman spectroscopy, turbidity, and ultrasound offer the possibility to monitor and control important process parameters (e.g., concentrations) in real time,7,8 which can lead to increased and more constant product quality by real-time process decision making and process adjustment.9 In order to facilitate the challenging scale-up from laboratory to pilot-plant or industrial scale, the application of such techniques is not only preferable in the lab but also at higher scales. In industrial practice, however, only recently have some in situ sensors started to be applied at the pilot-plant or industrial level. In this study, the © XXXX American Chemical Society

application of single-frequency ultrasound to determine concentrations, as an important process parameter, and the metastable zone widths (MZWs), as an important precondition for process development, is evaluated and critically discussed. The technique is used at laboratory scale, and furthermore, a possibility for the integration of the ultrasound measurement technique at the pilotplant scale is presented. It is tested whether calibration models for concentration determination developed at the laboratory scale can be transferred directly to higher scales. This would be preferable for industrial practice since time for calibration model development and validation might be minimized.

2. MATERIALS, METHODS, AND PROCEDURES 2.1. Materials. Most of the materials used are pharmaceutical model compounds with different physical and chemical properties. The molecular weight ranges from 75 g/mol (small molecules) up to 46 °C; cooling: temperatures >40 °C). From the results, however, it can be concluded that the velocities measured in the pilot plant are significantly higher than those in the lab experiment. A possible reason for the unexpected behavior is given at the end of this section. The model compound acetylsalicylic acid, which belongs to group 2 (satisfactory to unsatisfactory applicability), was selected for further pilot plant investigations. The ultrasound velocity measurement (results not shown) shows the same limitations as E

dx.doi.org/10.1021/op4001803 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development

Article

the necessary measurement conditions is not given. As already shown by Helmdach et al.,26 air/gas bubbles can also influence solid-phase measurements in suspensions derived from optical reflectance measurements, focused-beam reflectance measurements, and single-frequency ultrasound. A further discussion on the effect of air/gas bubbles is given in section 3.2.2. 3.2. Concentration Measurements in the Liquid Phase. 3.2.1. Applicability for Pharmaceutical Compounds. According to the literature, the ultrasound technique can be used to measure concentrations, mostly of inorganic salts but also of organic compounds.27−29,24 Pharmaceutical compounds, however, have in comparison with previously tested materials often partly different properties, such as high molecular weights, low concentration ranges, and/or extremely small particle sizes. Therefore, the application of ultrasound might be limited for these kinds of materials. To evaluate the applicability of the ultrasound technique for pharmaceuticals, a total of 12 different test compounds were analyzed within different concentration ranges. Consequently, the data were compared with the measurements shown in the literature. An overview of the results and an evaluation of the result quality is to be seen in Table 3. As can be observed for most of the materials, the concentrations could be analyzed, except for the compounds artemesinic acid and SAR114137. The results are sorted by the sensitivity of the concentration measurement. The sensitivity is defined as the maximum change of ultrasound velocity for the maximum concentration difference analyzed under isothermal conditions, in order to exclude the influence of temperature on the measured velocity (T[°C]/Δc[g/100 g]/ΔVEL [m/s]). Furthermore, information on the change in velocity with increasing temperature [VEL trend (↑T)] and at the nucleation point (VEL trend at Tnucl) as well as of the root-mean-square error of calibration and/or prediction (RMSEC/RMSEP) and the molecular weights (MW) are given in Table 3. By sorting the results according to the concentration measurement sensitivity, a grouping is observed. The ultrasound velocity shows with increasing temperature and at the nucleation point the same trend for each group. Group 1, which is highlighted in light grey, has generally a high sensitivity for analysis of the concentration. The velocity trends with increasing temperature and at nucleation change in opposite directions (↑, ↓). Most of the materials in this group are inorganic or organic but nonpharmaceutical materials. Only two of the eight materials are compounds used for this study (pharmaceutical compounds). Consequently, it can be shown that the sensitivity of the ultrasound device for detection of concentration changes is higher for inorganic compounds with low molecular weights than for organic compounds with high molecular weights. Group 2 includes materials for which the velocity trend changes in the same direction (↑,↑ or ↓,↓). Within this grouping two subgroups can be distinguished: (a) both trends decreasing (↓,↓) and (b) both trends increasing (↑,↑). Furthermore, a group 3 is present for which the velocities change again in opposite directions (↓,↑) but vice versa compared with group 1. The materials in groups 2 and 3 are exclusively organic materials. Groups 2 and 3 show limited or no applicability for measurement of the concentration by the ultrasound measurement technique. From the combination of the results of section 3.1 (determination of MZW by ultrasound) and Table 3, important conclusions can be drawn concerning the application of the measurement technique to analyze the MZW and the concentration (see Table 4).

Figure 2. Determinations of MZW at the pilot-plant and lab scales for the compound paracetamol (group 3, excellent applicability) by means of ultrasound (concentration: 10 wt %). The arrows are directly assigned to the measurement in either the lab or pilot plant and indicate changes in ultrasound velocity, which can be used to determine nucleation and/ or solubility points. The disagreement between results for the lab and the pilot plant, especially for temperatures higher than the nucleation point (everything dissolved), can be explained by the presence of air in the pilot plant.

Figure 3. Determination of MZW at the pilot-plant and lab scales for the compound acetylsalicylic acid (group 2, satisfactory to unsatisfactory applicability) by means of ultrasound attenuation for a concentration of 22.5 g/100 g.

results present clear readings. In the pilot plant, however, the investigated compound shows similar limitations as in the case of paracetamol. Neither the ultrasound velocity nor the attenuation allowed the solubility or nucleation points to be determined. A graph of the ultrasound velocity trend measured in the pilot plant together with the lab results are given in section 6.8 in the Supporting Information. The highly fluctuating measurement signal and the measurement of significantly higher velocities in the pilot plant in comparison with the laboratory can be caused for temperatures lower than the nucleation temperature by particles of different sizes and/or different suspension densities. However, for conditions where no particles are present, this behavior is unexpected. A possible reason for this behavior might be the presence of undissolved air and/or gas (air and gas bubbles) during the pilot plant experiments, since the ultrasound sensor was integrated in a repump. If the stated reason is right, the measurement technique is fully functional, but the providence of F



Article

Table 3. Overview and Sensitivity of the Ultrasound Concentration Measurements for Pharmaceutical and Nonpharmaceutical (Inorganic, Organic) Compoundsa

a

Definitions: T[°C]/Δc [g/100 g]/ΔVEL [m/s], maximum change in the ultrasound velocity for the maximum concentration difference under isothermal conditions; VEL trend (↑T), ultrasound velocity trend with increasing temperature; VEL trend at Tnucl, ultrasound velocity trend at the nucleation point.

Since the trend of ultrasound velocity with increasing temperature and at the nucleation point is the same for each

group, all of the materials in one grouping have a specific shape of the temperature versus velocity curve. Furthermore, the graphs G



Article

Table 4. Evaluation of the Applicability of the Ultrasound Measurement Technique for the Measurement of MZW and Concentration for the Three Different Groups Derived from Table 3 (Pharmaceutical Compounds Most Often Belong to Group 2 or 3) group

determination of MZW

concentration measurement

1 2 3

excellent to good applicability (++) unsatisfactory applicability (−) excellent applicability (+++)

excellent applicability (+++) good to satisfactory applicability (+) unsatisfactory applicability (−)

Figure 4. Changes in (left) velocity and (right) density and adiabatic compressibility with temperature for glycine, a model compound of group 1.

Figure 5. Changes in (left) velocity and (right) density and adiabatic compressibility with temperature for citric acid, a model compound of group 3.

Although the density is always increasing with decreasing concentration, the adiabatic compressibility can change in both directions (increase or decrease). This is demonstrated for the model compound citric acid, which belongs to group 3 (Figure 5). It seems that the compressibility of the medium is the determining factor. Therefore, compounds of group 3 show the opposite behavior from compounds in group 1: the velocity is increasing with decreasing temperature and at the nucleation point. For compounds in groups 2a and 2b it is assumed that the adiabatic compressibility shows two different trends with decreasing temperature and at the nucleation point. This is presented in section 6.10 in the Supporting Information for the compounds irbesartan (group 2a) and SAR474832 (group 2b). For irbesartan, for instance, the velocity increases with decreasing temperature, and at the nucleation point a sudden decrease in velocity is detected. From the results measured for the compounds glycine (group 1) and citric acid (group 3), it can

including the calibration data or, in other words, the measured velocities for different concentrations only of the clear solutions (without particles) look the same. The temperature versus velocity graphs and the graphs including the calibration data of the analyzed pharmaceutical compounds can be seen in section 6.9 in the Supporting Information. The similarities in the shapes of the temperature versus velocity graphs can be related most likely to the changes in adiabatic compressibility and density. According to the Newton−Laplace equation (e.g., shown by Povey) the ultrasound velocity (v) is the square root of the reciprocal product of adiabatic compressibility and density.31 The equation can be used to calculate the compressibility from the measured ultrasound velocity and the measured density. Glycine, a model compound of group 1, shows with decreasing temperature an increase in density as well as adiabatic compressibility, as can be seen from Figure 4. This consequently leads to a decrease in velocity with decreasing temperature and at the nucleation point. H



Article

Figure 6. Calibration of concentration at the pilot-plant and lab scales for the compound acetylsalicylic acid (group 2, good to satisfactory applicability) by means of ultrasound.

Table 5. Ultrasound Calibration Model Summary (Pure Quadratic Equation) For Lab and Pilot-Plant Models of Acetylsalicylic Acid Dissolved in Ethanola

a

model

RMSEC

A

B

C

D

E

F

lab PP

0.11 0.12

1627.6750 1653.10524

−9.53813989 −10.0115147

−3.04597605 −3.09644158

0.00888890 0.00932317

0.01351588 0.01422938

0.00139963 0.00142365

Definitions: PP, pilot plant; RMSEC, root-mean-square error of calibration; A−F, constants for the calculation of concentration (referring to eq 1).

pressure can bring dissolved air/gas out of solution, creating bubbles. The velocity of sound in pure air is 322.16 m/s (20 °C). In the ethanol−water mixture used to dissolve the paracetamol, sound has a velocity of 1570 m/s (60 °C). The presence of air/gas bubbles therefore leads to a significant reduction in velocity in dependence on the undissolved air/gas content. In conclusion, a calibration transfer for the test compound paracetamol is not possible under the present conditions in the pilot plant setup as long as the air/gas bubbles cannot be reduced. Deaeration of the solvents [e.g., by sonication (high-power ultrasound), vacuum, or nitrogen displacement] might solve the problems.31 The circumstance that the experiments are performed at “big” scales, however, makes the application of these techniques more difficult. It should be noted that paracetamol has anyway a low sensitivity for concentration measurements with ultrasound, since the compound belongs to group 3. This leads to relatively high calibration and prediction errors, which are 1.2 and 2.1 wt % in case of the lab experiment and a concentration range from 9 to 43 wt %. In comparison with other concentration measurement techniques, such as MIR spectroscopy, the prediction error is significantly lower (7 times) for the same compound (e.g., 0.29 wt % for a pilot plant peak integration model), whereby a significantly lower concentration range (1−10 wt %) is used. Example: Acetylsalicylic Acid (Group 2). Acetylsalicylic acid dissolved in ethanol was used as an example of a compound in group 2 (good to satisfactory applicability). From the results shown in Figure 6 it can be observed that the velocities measured in the lab and in the pilot plant for the same concentration and temperature ranges show very good agreement. The low influence of air/gas for acetylsalicylic acid dissolved in ethanol in comparison with the system consisting of paracetamol dissolved in a 20 wt % ethanol−water mixture might be explained since gas is much more soluble in ethanol than in water.32 Therefore, the extent of nonsoluble gas, which forms air/gas bubbles, is smaller for acetylsalicylic acid. Furthermore, the

be assumed that the compressibility decreases with decreasing temperature. At the nucleation point, however, a small increase in adiabatic compressibility might take place. According to the presented results, the compounds in group 1 show an excellent applicability for the analysis of concentrations by means of ultrasound. Both variables (the density and the compressibility) change in the same direction. Therefore, the extent of the change in the velocity is maximal. For compounds in group 3, the applicability is unsatisfactory since the density and the compressibility change in opposite directions. Therefore, the velocity change is negligible or very low. Group 2 takes an intermediate position. 3.2.2. Applicability in the Pilot Plant. In the following, one compound of each group was selected in order to test the applicability of the ultrasound measurement technique for monitoring of the concentration at the pilot-plant scale, where the sensor is integrated in a horizontal direction in a flow-through cell. For this purpose, the ultrasound velocities of the different compound concentrations (solid material completely dissolved) were investigated in dependence on the temperature. Example: Paracetamol (Group 3). As a compound of group 3 (unsatisfactory applicability), paracetamol in an ethanol−water mixture (20:80 w/w) was chosen (results not shown). It was observed that there are significant differences in the analyzed velocities for the same concentration range. The velocities analyzed in the pilot plant are lower than the results of the lab experiment. The sensitivity to detect different concentrations is higher in the case of the laboratory experiments. From the pilot plant results it was seen that especially the measured velocities of the solvent are highly fluctuating. An explanation for the systematically lower velocities in the pilot plant is the presence of undissolved air/gas, which results from pumping of the solution in the pilot plant. According to Povey,31 even very small amounts of undissolved air/gas can have a dramatic influence on the velocity of sound. Although dissolved air/gas has a very small effect on the velocity, small changes in I



Article

viscosity of the medium might have an influence on the degassing behavior. From the measured ultrasound velocities in dependence on the concentration and temperature, calibration models can be developed. The concentration can be correlated using the following quadratic equation (T = 18−70 °C, v = 1000−1225 m/ s): c = A + BT + Cv + DTv + ET 2 + Fv 2

(1)

In this expression, c represents the concentration in g/100 g, T is the temperature in °C, and v is the ultrasound velocity in m/s. The rstool (interactive multidimensional response surface modeling) routine in MATLAB R2007a with the string “pure quadratic” was used to fit the interaction between velocity, temperature, and concentration. In Table 5 a calibration model summary for the lab and pilotplant models of acetylsalicylic acid is given. As can be seen, the calibration errors (RMSEC) are almost identical for the calibrations in the lab and pilot plant. The constants A−F for the calculation of concentration are in good accordance as well. A comparison of the ultrasound calibration model to a calibration method developed for ATR-MIR spectroscopy is given by Helmdach.33 It is shown that the RMSEP for the ultrasound calibration model is in agreement with the results of ATR-MIR spectroscopy (peak integration and partial least-squares model). Pilot Plant Transferability of the Ultrasound Lab Calibration Model. The ultrasound calibration model for acetylsalicylic acid dissolved in ethanol that was developed at the lab scale can also be used for predictions at the pilot-plant scale. This is shown by the recovery curve in section 6.11 in the Supporting Information. The results for the pilot plant data predicted by the pilot plant calibration model show ideal recovery behavior. The prediction error with 0.12 g/100 g is very low. Furthermore, the pilot plant data were predicted by a calibration model developed at the lab scale. The result of the recovery plot shows small deviations that increase with increasing concentration. The prediction error with 1.16 g/100 g is significantly higher than the prediction error for the samples predicted by the pilot plant calibration model. This might be explained since the concentrations in the pilot plant were not produced independently but by stepwise addition of solid material. Furthermore, an influence of air/gas bubbles cannot be excluded completely. As was already presented for the compound paracetamol, air/gas bubbles cause a reduction in the ultrasound velocity. This would lead in consequence to the prediction of lower concentrations. Nevertheless, it is shown that a direct calibration transfer from the lab to a pilot plant is possible for the system acetylsalicylic acid in ethanol with only small deviations being observed. Example: L-Glutamic Acid (Group 1). As a model compound of group 1 (excellent applicability), L-glutamic acid was chosen. The results of ultrasound measurements in the lab and pilot plant for different concentrations and temperatures can be seen in Figure 7. From the concentrations 0, 3, and 6 g/100 g it can be observed that there are differences between the lab and pilot plant experiments. On the one hand, the fluctuations of the ultrasound signal measured in the pilot plant are higher in comparison with the lab results, and on the other hand, the velocities analyzed in the pilot plant are significantly higher for temperatures lower than 82 °C compared with the lab results. Only at high temperatures (above 82 °C) can good agreement of the lab and pilot plant results be observed. By optical inspection of the product stream through the sight glasses it could be seen that from a temperature of 82 °C (and for lower temperatures) the content of air/gas

Figure 7. Calibration of concentration at the pilot-plant and lab scales for the compound L-glutamic acid (group 1, excellent applicability) by means of ultrasound.

bubbles started to increase suddenly. This might be due to the decrease in gas solubility with increasing temperature and would also explain the better signal quality for the cooling step in comparison with the heating step. Consequently, the calibration model can only be used within an extremely narrow temperature range under the present conditions in the pilot plant. As already stated for the compound paracetamol, the reduction or removal of the air/gas can lead to an improvement of the results in the pilot plant. Influence of the Process Conditions on the Air/Gas Bubble Content. From Figure 8 it can be observed that an optimization

Figure 8. Influence of the process conditions on the ultrasound velocity in the pilot plant for an L-glutamic acid concentration of 6 g/100 g.

of the process conditions can minimally reduce the effect of the increase in ultrasound velocity and the fluctuations of the measurement signal that might be caused by air/gas bubbles. It can be seen that cooling leads to a smoother signal with lower velocities that are in better agreement with the velocities of the lab experiment in comparison with the velocities measured during the heating cycle in the pilot plant. Furthermore, the signal can be improved by reduction of the pump stroke. This improvement leads to an excellent agreement of the analyzed velocities for temperatures higher than 82 °C. For temperatures lower than 82 °C, a sudden increase in the ultrasound velocity difference J



Article

(5) Feth, M. P.; Nagel, N.; Baumgartner, B.; Bröckelmann, M.; Rigala, D.; Otto, B.; Spitzenberg, M.; Schulz, M.; Becker, B.; Fischera, F.; Petzoldt, C. Eur. J. Pharm. Sci. 2011, 42, 116−129. (6) Kadam, S. S.; Vissers, J. A. W.; Forgione, M.; Geertman, R. M.; Daudey, P. J.; Stankiewicz, A. I.; Kramer, H. J. M. Org. Process Res. Dev. 2012, 16, 769−780. (7) Bakeev, K. Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries; Blackwell: Oxford, U.K, 2005. (8) Helmdach, L.; Feth, M. P.; Minnich, C.; Ulrich, J. Chem. Eng. Proc.: Process Intensification 2013, 70, 184−197. (9) Tawakkul, M.; Wu, H.; White, M.; Khan, M. Int. J. Pharm. 2009, 372, 39−48. (10) DrugBank: Open Data Drug & Drug Target Database. http:// www.drugbank.ca (accessed January 2013). (11) Bond, A. D.; Boese, R.; Desiraju, G. R. Angew. Chem., Int. Ed. 2007, 46, 615−617. (12) ChemSpider: The Free Chemical Database, ID 9097710. http:// www.chemspider.com/Chemical-Structure.9097710.html (accessed January 2013). (13) Feth, M. P.; Heyse, W.; Baumgartner, B.; Nagel, N.; Tappertzhofen, C.; Olpp, T.; Jurascheck, J.; Ulrich, J.; Helmdach, L.; Petzoldt, C. Eur. J. Pharm. Biopharm. 2013, 83 (3), 436−448. (14) Harris, R. K.; Ghi, P. Y.; Puschmann, H.; Apperley, D. C.; Griesser, U. J.; Hammond, R. B.; Ma, C.; Roberts, K. J.; Pearce, G. J.; Yates, J. R.; Pickard, C. J. Org. Process Res. Dev. 2005, 9, 902−910. (15) Helmdach, L.; Feth, M. P.; Ulrich, J. Cryst. Res. Technol. 2012, 47, 967−984. (16) Perlovich, G. L.; Hansen, L. K.; Bauer-Brandl, A. J. Therm. Anal. Calorim. 2001, 66, 699−715. (17) Dudognon, E.; Danéde, F.; Descamps, M.; Correia, N. T. Pharm. Res. 2008, 25, 2853−2858. (18) Delaney, S. P.; Pan, D.; Galella, M.; Yin, S. X.; Korter, T. M. Cryst. Growth Des. 2012, 12, 5017−5024. (19) Kitamura, M. J. Cryst. Growth 1989, 96, 541−546. (20) Kachrimanis, K.; Braun, D. E.; Griesser, U. J. J. Pharm. Biomed. Anal. 2007, 43, 407−412. (21) SensoTech GmbH. LiquiSonic OCM: Inline Analytical Technology for Online Crystallization Monitoring. http://www. sensotech.com/cms/fileadmin/SensoTech/Dokumente/LSM100/ LSM169_01.pdf (downloaded January 2013). (22) The MathWorks. Product Documentation, MATLAB R2012a. http://www.mathworks.de/de/help/stats/rstool.html (accessed January 2013). (23) Buchfink, R. Effects of Impurities on an Industrial Crystallization Process of Ammonium Sulfate. Dissertation, Martin-Luther-Universität Halle-Wittenberg, Shaker-Verlag, Halle, Germany, 2011. (24) Omar, W. Zur Bestimmung von Kristallisationskinetiken auch unter der Einwirkung von Additiven mittels Ultraschallmesstechnik. Dissertation, Universität Bremen, Bremen, Germany, 1999. (25) GEA Group Aktiengesellschaft. Inline Prozessanschlusse. http:// www.tuchenhagen.com/de/komponenten/in-line-prozessanschluesse. html (accessed January 2013). (26) Helmdach, L.; Pertig, D.; Rüdiger, S.; Lee, K.-S.; Stelzer, T.; Ulrich, J. Chem. Eng. Technol. 2011, 35, 1017−1023. (27) Glade, H.; Ilyaskarov, A. M.; Ulrich, J. Chem. Eng. Technol. 2004, 27, 736−740. (28) Heinrich, J. Determination of Crystallization Kinetics Using inSitu Measurement Techniques and Model-Based Experimental Design and Analysis. Dissertation, Martin-Luther-Universität Halle-Wittenberg, Halle, Germany, 2008; http://sundoc.bibliothek.uni-halle.de/dissonline/08/09H002/t1.pdf. (29) Strege, C. On (Pseudo-) Polymorphic Phase Transitions. Dissertation, Martin-Luther-Universität Halle-Wittenberg, Halle, Germany, 2004; http://sundoc.bibliothek.uni-halle.de/diss-online/04/ 04H318/prom.pdf. (30) Schwartz, A. M., Myerson, A. S. Solutions and Solution Properties. In Handbook of Industrial Crystallization; Myerson, A. S., Ed.; Butterworth-Heinemann: Boston, 2002; pp 1−32.

between the lab and pilot plant can be detected again. A decrease in the stirring speed with a reduced pump stroke does not lead to a further improvement of the measurement signal.

■

CONCLUSION Apart from the knowledge of the MZW, which is necessary for process development, it is important to monitor the concentration of the API solute in the solution during the crystallization process. The actual value of the concentration can provide the driving force for crystallization during the process. According to the results presented for the ultrasound sensor, the technique can be used to predict concentrations reliable for many substances. The application of this measurement technique for a number of pharmaceutical compounds (especially with high molecular weights), however, seems to be limited. On the contrary, for many inorganic compounds the technique promises most often an excellent applicability for the concentration and MZW determination due to the high sensitivity of the ultrasound velocity for these kind of compounds with low molecular weight. Whether a compound is analyzable by ultrasound can be determined very simply and rapidly in one experiment. For this purpose, the material is associated to a group (according to the grouping presented in Table 3) by the evaluation of the velocity change with increasing temperature and at the nucleation points. Concerning the transferability of calibration models (solute concentration) developed at the lab scale to the pilot-plant scale, it was found that a direct transfer is possible if the influence of air/ gas bubbles in the pilot plant is low. A reduction in undissolved air/gas can be achieved, for example, by degassing with highpower ultrasound. For materials that belong to groups 1 and 3, the technique can be used for accurate MZW determination.

■

ASSOCIATED CONTENT

S Supporting Information *

Ultrasound device and sample details and additional experimental results. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

[email protected] martinphilipp.feth@sanofi.com [email protected] Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors gratefully acknowledge Dr.-Ing. Ingo Benecke and Dipl.-Ing. Stefan Nitschke from SensoTech GmbH for providing the OCM device and fruitful discussions. The authors would like to thank the Sanofi C&BD Frankfurt Chemistry pilot plant team G839 for their continuous support of the experiments.

■

REFERENCES

(1) Mullin, J. W. Crystallization, 4th ed.; Butterworth-Heinemann: Oxford, U.K., 2001. (2) Lewiner, F.; Klein, J. P.; Puel, F.; Fevotte, G. Chem. Eng. Sci. 2001, 56, 2069−2084. (3) Reutzel-Edens, S. M. Curr. Opin. Drug Discovery Dev. 2006, 9, 806− 815. (4) Variankaval, N.; Cote, A. S.; Doherty, M. F. AIChE. J. 2008, 54, 1682−1688. K



Article

(31) Povey, M. J. W. Ultrasonic Techniques for Fluid Characterization; Acadamic Press: San Diego, CA, 1997. (32) Dalmolin, I.; Skovroinski, E.; Biasi, A.; Corazza, M. L.; Dariva, C.; Oliveira, J. V. Fluid Phase Equilib. 2006, 245, 193−200. (33) Helmdach, L. Application of Process Analytical Technology (PAT) Tools to Develop and Monitor Scalable Crystallization Processes of Pharmaceuticals. Dissertation, Martin-Luther-Universität HalleWittenberg, Halle, Germany, 2013.

L


Application of Ultrasound Measurements as PAT Tools for Industrial

Recommend Documents