Microsoft Word - Contents.doc HUNGARIAN JOURNAL OF INDUSTRIAL CHEMISTRY VESZPRÉM Vol. 34. pp. 21-26 (2006) SEPARATION OF ORGANIC COMPOUNDS BY GRADIENT SIMULATED MOVING BED CHROMATOGRAPHY M. NAGY1, , T. SZÁNYA2, Z. MOLNÁR2, G. TURZA2, G. GÁL2, L. HANÁK2, J. ARGYELÁN2, A. ARANYI1, K. TEMESVÁRI1 and Z. HORVÁTH1 1Gedeon Richter Ltd., H-1475 Budapest 10. PO Box 27. HUNGARY; E-Mail: NagyMM@richter.hu 2University of Pannon, Department of Chemical Engineering POB 158, Veszprém, H-8201, HUNGARY A promising solution for the separation of organic compounds in pharmaceutical industry is the gradient simulated moving bed chromatography (SMB). The above mentioned process can be used for the production of high purity materials (isomers, optical isomers, biomolecules). We assumed isotherm, isochor equilibrium adsorption (competitive multicomponent Langmuir-adsorption equilibrium) in the mathematical model and neglected the effects of axial dispersion. In our present work we extended the mathematical model with solvent adsorption-desorption processes (acetone-dichloromethane eluent, steroid compounds, silicagel). The mathematical model was solved by finite differences numerical mathematical method using PC. The gradient SMB separations were carried out with a laboratory scale four column equipment in open loop system at 1:1:2:0 column configuration. The SMB equipment (L=25 cm, I.D.=1 cm) was planned and constructed for separation of a steroid mixture using YMC S-50 silica gel as adsorbent (specific surface=798.63 m2 g-1). The effect of switching time change (5.5 min, 9 min, 11.5 min, 22.5 min) on component separation was examined. During our measurements the amount of acetone in dichloromethane was 55 % v/v in the fresh eluent and pure dichloromethane in the feed. The process variables of the gradient SMB (product purity, yield, productivity, specific solvent consumption) are very favourable. Our conclusion is that the measured and the calculated data agree well and we have produced more than 99.9 %m/m purity and 90 % yield , productivity 1300-3000 g steroid kg-1 adsorbent day at 0.1-0.3 m3 fresh eluent kg-1 steroid eluent consumption. Keywords: Preparative liquid chromatography, Simulated moving bed chromatography, Solvent gradient SMB Introduction The SMB process was developed in the early 1960s by Brougthon and Gerhold [1] and has been used for many years in the petrochemical and sugar industries for large scale separation [2]. Recently, the separation of enantiomers due to the demand for high purity products has become necessary [3]. The SMB (Fig. 2.) is a continuous unit operated in cyclic mode which reproduces the performance of the equivalent true moving bed (TMB) unit (Fig. 1.). The SMB is divided into four sections, each constituted of a counter current adsorption column which plays a specific role in the separation. Let us consider a feed mixture consisting of more retained species (A) and a less retained one (B) dissolved in the eluent. The separation is obtained in the two central sections, where B is carried by the mobile phase to the raffinate outlet. The fresh eluent is fed to the bottom of the section I, so as to desorb A and regenerate the adsorbent solid, before it is recycled to section IV. On the other hand B is retained in section IV and the pure eluent is recycled to section I. In practice the movement of the solid is not feasible and the TMB is replaced by the SMB configuration where the adsorption beds are fixed and the counter-current solid-liquid movement is simulated by periodical switching the inlet and outlet ports of the unit [4]. I. Feed (A +B) Extract (A) Raffinate (B) IV. III. II. A ds or be nt r ec ir cu la ti on L iq ui d re ci rc ul at io n S (E-F+R ) D=S+REC REC Fresh eluent Fig. 1.: True moving bed (TMB) adsorber – I, II, III, IV – zones; S – desorbent (solvent, eluent); Rec – recirculated eluent; E – extract stream with the better adsorbed component A; F – feed stream with the components A and B; R – raffinate (liquid outlet) stream with the less adsorbed component B 22 I II III IV D=S+REC E (A) F (A+B) R (B) Liquid phase Adsorbent phase Fig. 2.: Simulated moving bed (SMB) adsorber – I, II, III, IV – zones, respectively HPLC columns Use of gradient elution is recommended, if conditions must be changed during separation. The applied gradient can be the pressure, temperature or solution composition. According to the literature of solution composition changing gradient SMB, where the solution composition is chosen, four scientific „schools” exist; the MPI Magdeburg Inst. Dyn. Komplex Techn. System in Germany [5,6,7], the Kluyer Laboratory for Biotechnology at Delft University of Technology in the Netherlands [8,9,10]. ETH Zürich, Inst. Verfahrenstechnik in Switzerland [4, 11] and there are significant scientific results in France [12], too. We calculated the initial parameters with the method of Morbidelli et al. [13]. After the authors in a two component system can be determined a the mII-mIII area from the geometric data of the SMB equipment, the volumetric velocities, the parameters of Langmuir-type isotherms and the concentrations of the mixture to be separated. The actual operating conditions determine a point on the mII-mIII diagram. The variable operating conditions are: fresh eluent, recirculated eluent, feed, extract and raffinate flow rates, switching time. Mathematical models and joined computer programmes published in our earlier papers [14,15] have been developed and applied for the calculation of SMB. Experiments All necessary additional information for this project, like the adsorption equilibrium data, the Number of Theoretical Plates (NTP) and Height of Equivalent Theoretical Plate (HETP), the frontal adsorption - elution measurement was published earlier [16]. Parameters of SMB-LC Measurements We applied an 1-1-2-0 column configuration in the SMB system. This way the steroid B has “bigger space” along the length of the columns. In this case we did not use the raffinate pump, we adjusted the flow rate in segment III so that the steroid B appeared in the LROUT flow in the open-looped SMB system. The function of segment I is the regeneration of the adsorbent and the production of the steroid A in the extract. As fresh eluent in the segment I we used 1.2:1 (v/v) acetone - dichloromethane (gradient). The function of segment II is the separation of steroid A and B and the extraction of the steroid A in this segment. The function of segment III is to separate steroid A and B and to produce steroid B in the LROUT flow. We fed the mixture of steroid A and B into the third segment, the steroids were soluted in pure dichloromethane because this improved the solution of the steroids and helped us to use gradient SMB. Each of the columns in the four-column SMB equipment were previously equilibrated with pure dichloromethane at 293 K, we wanted to separate the dichloromethane mixture of 42 g dm-3 steroid B and 18 g dm-3 steroid A. The fresh eluent 1.2:1 v/v acetone-dichloromethane volumetric ratio was 4.25 cm3min-1. The feed flow rate of the steroid sample was 0.8 cm3 min-1. Steroid A was extracted with 2.2 cm3 min-1 parameter, so the LROUT flow was 2.85 cm3min-1 and the switching time was 22.5 min. Process parameters were determined by Morbidelli method (Fig. 3. and Table 1). Fig. 3.: The measurement points placed in the Morbidelli triangle at different switching times When we decreased the switching time to 11.25 min, we doubled the original flow rates. When we used 5.5 min for switching time, we doubled once again the flow rates. In case of 9 min as switching time, we used computer simulation to optimize the work point to get as close to the points of Morbidelli triangle as it was possible (F= 1.8 cm3min-1, D= 14.4 cm3min-1, E=7.8 cm3min-1, LROUT= 8.4 cm3min-1). 23 Table 1: Input data of the software SMB-KROM-N Number of components k = 3 Column inner diameter ID.=1 cm Column length L=25 cm Number of columns N=4 Free volume coefficient EPS= 0.8018 cm3 liquid free volume cm-3 column Bulk density ROH= 0.4045 g silica gel cm-3 column Langmuir constants (same as on page ) Feed concentration liquid cmacetonemg -3Facetone 0=c liquid cmcomponentBmg -3FB 42=c liquid cmcomponentAmg -3FA 18=c Flow rates * Optimized by simulation liquidcomponentAorBmgS A S B 30 −== cmcc Simulation SMB 8/41 SMB 8/42 SMB 8/46* SMB 8/44 Switching time(min) 22.5 11.25 9 5.5 F (cm3 min-1) 0.8 1.6 1.8 3.2 D (cm3 min-1) 4.25 8.5 14.4 17.0 E (cm3 min-1) 2.2 4.4 7.8 8.8 R (cm3 min-1) 0 0 0 0 Fresh eluent liquid cmacetonemg. -3Facetone 6435=c Number of Theoretical Plates NTP=200 / 25 cm column Calculation time 585 min Simulation Data of 1:1:2:0 Column Configuration SMB Simulation was done by SMB-KROM-N software. The input data are summarized in Table III. The software takes into account the solvent adsorption- desorption phenomena, too. Fig. 4 shows the profile of A and B component along the length of columns and Fig. 5 shows the acetone gradient along the length of columns, at quasi- stationary state in case of RG SMB8/46 simulation. 0 5 10 15 20 25 30 35 40 45 0 200 400 600 800 NTP c A , B c om p on en t ( g dm -3 ) B component A component Fig. 4.: Profile of A and B components at quasi- stationary state 24 Fig. 5.: Profile of acetone at quasi-stationary state Fig. 6 and 7 shows the acetone concentration change in the raffinate and extract outlet versus simulation time in case of RG SMB8/46 simulation. 200 250 300 350 400 450 711 720 729 Simulation time (min) C on ce nt ra tio n a ce to ne (g d m -3 ) Acetone Raffinate Fig. 6.: Acetone concentration versus two period of simulation time in the raffinate outlet 428 429 430 431 432 433 434 435 436 437 711 720 729 Simulation time (min) C on ce nt ra tio n a ce to ne (g d m -3 ) Acetone Extract Fig. 7.: Acetone concentration versus two period of simulation time in the extract outlet Results Measurement Results of 1:1:2:0 Column Configuration SMB Compared to Simulation On the Fig. 8, 9, 10 and 11 those measurements are marked with grey colour where the desired results can be seen, with white colours the measurements which are out of our requirements. 0 20 40 60 80 100 Prep HPL GSMB8/ (22.50 GSMB8/ (11.25 GSMB8/ (5.50 GSMB8/ (9.00 pu rit y "B "( % / Raffinate > 99.9 % m/m B η B > 90 % Raffinate < 99.9 % m/m B η B < 90 % Fig. 8.: Comparing the purity of steroid B in different measurements 0 10 20 30 40 50 60 70 80 90 100 Prep. HPLC GSMB8/41 (22.50 min) GSMB8/42 (11.25 min) GSMB8/44 (5.50 min) GSMB8/46 (9.00 min) yi el d "B "( % ) Raffinate > 99.9 % m/m B ηB > 90 % Raffinate < 99.9 % m/m B or ηB < 90 % Fig. 9.: Comparing the yield of steroid B in different measurements 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Prep. HPLC GSMB8/41 (22.50 min) GSMB8/42 (11.25 min) GSMB8/44 (5.50 min) GSMB8/46 (9.00 min) P B (m g B /g s ili ca g el m in ) Raffinate > 99.9 % m/m B ηB > 90 % Raffinate < 99.9 % m/m B or ηB < 90 % Fig. 10.: Comparing the productivity of steroid B in different measurements 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Prep. HPLC GSMB8/41 (22.50 min) GSMB8/42 (11.25 min) GSMB8/44 (5.50 min) GSMB8/46 (9.00 min) SF B (d m 3 e lu en t g -1 B ) Raffinate > 99.9 % m/m B ηB > 90 % Raffinate < 99.9 % m/m B or ηB < 90 % Fig. 11.: Comparing the eluent consumption of steroid B in different measurements 25 The result of the experiments shows that the decrease of the switching time highly improves the productivity, especially at the optimized 9 min switching time experiment. In this case the eluent consumption is also properly low. To describe numerically the above GSMB 8/41 and GSMB 8/46 the productivity was increased from 0.95 mg steroid B g-1 adsorbent min-1 to 2.64 mg steroid B g-1 adsorbent min-1. The eluent consumption was changed from 0.12 cm3 eluent mg-1 steroid B to 0.18 cm3 eluent mg-1 steroid B. We achieved the best favourable result in the RG- 1040-GSMB 8/46 measurement at 9 min switching time. In this case the calculated values of measurement were compared to data given in Table 2. We note that the LROUT 1, 2 are given by halving the LROUT liquid stream for 0-4,5 min and 4,5-9 min fractions. Table 2: Results of the RG 1040 GSMB 8/46 measurement compared to result of the simulation Measurement Simulation Purity of raffinate (LROUT) LROUT1 96.96 %m/m B LROUT2 > 99.9 %m/m B LROUT 1 +LROUT 2 > 99.9 %m/m B Purity of extract (E) 88.89 %m/m A 93.5 %m/m A Yield LROUT1 6.46 %B LROUT2 92.6 %B Extract 68.1 %A LROUT 1 +LROUT 2 > 99.99 % 98.3 % Productivity LROUT 2 2.14 mingelsilicag Bmg Extract 0.54 mingelsilicag Amg LROUT 1 +LROUT 2 2.16 mingelsilicag Bmg Extract 0.69 mingelsilicag Amg Eluent consumption LROUT 2 0.134 Bmg eluentfreshcm3 Extract 0.403 Amg eluentfreshcm3 LROUT 1 +LROUT 2 0.197 Bmg eluentfreshcm3 Extract 0.61 Amg eluentfreshcm3 Discussion Our conclusion on the basis of laboratory gradient, 1:1:2:0 column configuration, open loop SMB measurements is that varying the switching time of SMB process results big changes in productivity. In our case we get the highest productivity at 9 min switching time. We significantly increased the operation results according to the preparative HPLC measurement (Table 3). Table 3: Results of the RG 1040 GSMB 8/46 measurement compared to preparative HPLC measurement Preparative HPLC RG1040 SMB8/46 measurement (9 min) Purity of raffinate (LROUT 2) >99.9 % m/m B >99.9 % m/m B Yield (LROUT 2) ~ 95 % B >92.6 % B Productivity (LROUT 2) 0.303 mingelsilicag Bmg 2.14 mingelsilicag Bmg Eluent consumption (LROUT 2) 0.354 Bmg eluentfreshcm3 0.134 Bmg eluentfreshcm3 26 We achieved 700% productivity increase and 50 % eluent consumption decrease beside more than 99.9 % m/m B purity and 90 % B yield. It could be seen, that the developed SMB process was outstandingly economic, the value of the productivity was successfully increased to 3.082 kg B/ kg silica gel within a day, at 0.134 m3 fresh eluent/kg B steroid eluent consumption. SYMBOLS cAcetone concentration of acetone, g dm -3 cB concentration of steroid B, g dm -3 cA concentration of steroid A, g dm -3 k number of components k’ capacity factor ID column inner diameter, cm L column length, cm F flow rate of feed, cm3 min-1 D flow rate of fresh eluent, cm3 min-1 E flow rate of extract, cm3 min-1 R flow rate of raffinate, cm3 min-1 LROUT outlet liquid at 1:1:2:0 column configuration at R=0 T switching time, min ε free volume coefficient, cm3 liquid free volume cm-3 column ROH bulk density, g silica gel cm-3column NTP Number of Theoretical Plates N Number of columns REFERENCES 1. BROUGHTON D. B., GERHOLD C. G. (1961) US Pat. 2 985589 2. RUTHVEN D. M., CHING C. B., (1989) Chem. Eng. Sci. 44 (5), 1011-1038 3. JUPKE A., EPPING A., SCHMIDT-TRAUB H., (2002) J. Chromatogr. A, 944 93-117 4. GIOVANNI O., MAZZOTTI M., MORBIDELLI M., DENET F., HAUCK W., NICOUD R., (2001) J. Chromatogr. A, 919 1-12 5. ANTOS D., SEIDEL-MORGENSTERN A., (2002) J. Chromatogr., 944 77-91 6. ANTOS D., SEIDEL-MORGENSTERN A., (2001) Chem. Eng. Sci., 56 6667-6682 7. ZIOMEK G., KASPEREIT M., JEZOWSKI J., SEIDEL- MORGENSTERN A., ANTOS D., (2005) J. Chromatogr A, 1070 111-124 8. JENSEN T., REIJNS T., BILLIET H., WIELEN L., (2000) J. Chromatogr A, 873 149-162 9. HOUWING J., BILLIET H., WIELEN L., (2002) J. Chromatogr A, 944 189-201 10. HOUWING J., HATEREN S., BILLIET H., WIELEN L., (2002) J. Chromatogr A, 952 85-98 11. ABEL S., MAZZOTTI M., MORBIDELLI M. (2002), J. Chromatogr A, 944 23-29 12. HAUCK W., LUDEMANN-HOMBOURGER O., NICOUD R. M., DI GIOVANNI O., MAZZOTTI M., MORBIDELLI M., PREP 2001 14th International Symposium Exhibit & Workshop on PREPARATIVE/PROCESS CHROMATOGRAPHY, Washington, DC. USA 13. MIGLIORINI C., MAZZOTTI M., MORBIDELLI M., (1998) J Chromatogr. A, 827:2 161-173 14. SZÁNYA T., ARGYELÁN J., KOVÁTS S., HANÁK L., (2001) J Chromatogr. A, 908 265-272 15. NAGY M., MOLNÁR Z., HANÁK L., ARGYELÁN J., SZÁNYA T., RAVASZ B., ARANYI A., TEMESVÁRI K., (2004) Chromatographia, 60 S 181-187. 16. NAGY M., MOLNÁR Z., HANÁK L., ARGYELÁN J., SZÁNYA T., RAVASZ B., TURZA G., ARANYI A., TEMESVÁRI K., (2004) Hungarian Journal of Industrial Chemistry, 32 13-21