Research Note pubs.acs.org/IECR
Modeling the Quasi-Equilibrium of Multistage Phytoextractions Björn Dreisewerd, Juliane Merz,* and Gerhard Schembecker Laboratory of Plant and Process Design, Department of Biochemical and Chemical Engineering, TU Dortmund University, Emil-Figge-Straße 70, 44227 Dortmund, Germany ABSTRACT: A model describing multistage leaching phytoextractions based on a nonideal quasi-equilibrium line of an inverted McCabe−Thiele diagram is proposed in this study. The model is able to simulate the temperature dependence of the amount of extractable target component. It was applied to the extraction of artemisinin from Artemisia annua with ethanol at different temperatures. All model parameters were determined via single-stage extractions only. The model was able to represent the experimental data of multistage leaching extractions, including the temperature-dependent extractable amount of artemisinin. Therefore, the model can be used to reduce the number of experiments and accelerate the process design of phytoextraction processes.
1. INTRODUCTION Phytochemicals (i.e., components originated from plants) have a high importance in the pharmaceutical, food, and cosmetics industries.1,2 The target components are often produced from plant material by solid−liquid extraction, called phytoextraction, in which the solid plant material containing the target component of interest is contacted with a liquid solvent. In this context, the batch maceration mode, in which the solid plant material is suspended in the liquid phase, is the industrially most applied operating mode.3,4 One major drawback of this operating mode is the limited degree of extraction, posed by the quasiequilibrium of the extraction, which corresponds to the state of a constant extract concentration.5 Therefore, single-stage extractions are mostly insufficient to achieve acceptable yields and multistage crossflow extractions are commonly applied. To display the quasi-equilibrium, the right-angled triangular diagram, the Ponchon−Savarit diagram, the equilateral triangular diagram, or the McCabe−Thiele diagram can be used.6−12 In all cases, the complex multicomponent system of the phytoextraction is reduced to a pseudo-three-component system, consisting of the extracted target component T, the carrier component C (the plant matrix), and the liquid solvent component LS. Depending on the diagram, the mass fractions of component i in phase j (wi,j, eq 1), or the reduced mass fractions with a carrier-free basis (w*i,j , eq 2), which is based on the liquid phase only, is utilized. mi , j wi , j = mC, j + mLS, j + m T, j (1) wi*, j =
mass fractions are usually equal, since it can be assumed that the plant matrix is insoluble. Multistage extractions are mainly designed graphically and calculations are found only if an ideal quasi-equilibrium is present, i.e., if the target component is completely transferred into the liquid phase.8,11 This is due to the fact that a thermodynamic equilibrium is hardly ever reached and, accordingly, no physical relationship between the concentration in the extract and the overall raffinate concentration exists.6,13 With regard to process design, a mathematical model describing nonideal multistage extractions with model parameters determined from single-stage extraction experiments can reduce the experimental effort and shorten the process synthesis time. Hence, a model based on mass balances and an empirical equation describing the quasi-equilibrium line of an inverted McCabe−Thiele diagram is proposed to predict the behavior of multistage extraction processes. Moreover, the model takes into account that the target component might be partly unextractable, e.g., caused by a temperature-dependent physisorption or chemisorption or a particle-size-dependent trapping of the target component in the inner structure of the plant matrix.5 The extraction of the antimalarial artemisinin from Artemisia annua is used as model system in this study, since preliminary experiments with ethanol showed that the extractable amount of artemisinin is temperature-dependent.
2. MATERIALS AND METHODS 2.1. Model Equations. The components present during extraction are divided into three classes: inert components (I) (e.g., the insoluble plant matrix, i.e. the carrier component C), target components (T), and liquid solvent components (LS). In this study, we propose a nonideal quasi-equilibrium line, which lies below the diagonal of the inverted McCabe−Thiele diagram and is represented by the following empirical equation:
mi , j mLS, j + m T, j
(2)
Especially for the raffinate phase R, which consists of the partly depleted solid plant material and a certain amount of adhering extract solution (the remaining humidity), wT,R and w*T,R might differ strongly. Both refer to the total amount of target component present in the raffinate phase (mT,R), consisting of the target component in the remaining humidity, the unextracted target component, and the target component adsorbed on the surface of the plant material. In case of the extract phase E, both © 2016 American Chemical Society
Received: Revised: Accepted: Published: 1808
November 26, 2015 January 26, 2016 January 26, 2016 January 27, 2016 DOI: 10.1021/acs.iecr.5b04506 Ind. Eng. Chem. Res. 2016, 55, 1808−1812
Research Note
Industrial & Engineering Chemistry Research * = wT,E
* a1wT,R − a3 * (a1 − 1) 1 + a 2wT,R
3. RESULTS AND DISCUSSION The results of the extractions of artemisinin from A. annua with ethanol at 30, 50, and 60 °C are given in Figure 1. Aside from the
(3)
The empirical parameters a1, a2, and a3 are determined experimentally via single-stage extractions with different liquidto-solid ratios, which corresponds to the determination of different tie lines. The root of the equation refers to the unextractable part of the target component. Left-sided to the root, where extract concentrations are negative, the concen* ) in the tration is set to zero. Equation 3 can have a pole (wT,R,pole range of 0 ≤ w*T,R ≤ 1, according to eq 4, which defines the maximum range of validity of the function. * wT,R,pole =−
1 a 2(a1 − 1)
(4)
Furthermore, the ratio of the amount of remaining humidity per amount of solid material, which is referred to as the remaining humidity coefficient (f H), must be determined in the same experiments to close the mass balances. The dependence of f H on the liquid-to-solid ratio (χ) was found to be empirically describable by a power-law function with the parameters b1 and b2 (see eq 5).
fH = b1χ b2
(5)
The remaining model equations are mass balances that describe a multistage leaching extraction. 2.2. Raw Materials and Chemicals. Dried and ground leaves of A. annua were purchased from CPQBA-UNICAMP, Brazil. Artemisinin standard (98 wt %) was obtained from Xi’an BoSheng Biological Technology, China. Super gradient grade acetonitrile for HPLC-analysis and undenatured ethanol (96 vol %) were purchased from VWR, Germany. Ultrapure water was produced using a Milli-Q Purification System with 0.22 μm Millipak express filters from Merck Millipore, Germany. 2.3. Sample Analysis. The extracts are analyzed with an Agilent 1290 Infinity UHPLC system (Agilent Technologies, USA). A detailed description of the analytical method applied, the UHPLC system, and the sample preparation has been given in the earlier publication by Dreisewerd et al.14 2.4. Extraction Experiments. All experiments were conducted in closed 250 mL laboratory bottles that had been sealed with PTFE winding tape, and the total amount of solid material and solvent was 0.100 kg. The liquid-to-solid ratio was varied from 3 kg kg−1 to 50 kg kg−1, with each experiment being conducted in duplicate. The exact masses of the solid plant material and the solvent are controlled by weighing. The bottles are placed in a temperature-controlled shaking water bath (Type GFL 1083, GFL, Germany) with extraction temperatures of 30, 50, and 60 °C and a shaking frequency of ∼200 min−1. The extraction is run for 90 min until the quasi-equilibrium is reached.14 Afterward, the extracts and raffinates are separated by decantation and weighed, and samples of the extracts are analyzed via ultrahigh-performance liquid chromatography (UHPLC). Multistage leaching extractions are conducted with 9.09 × 10−3 kg of A. annua at the same temperatures as the single-stage extraction experiments. The experimental proceedings are in accordance with the procedure described by Dreisewerd et al.14
Figure 1. Influence of the liquid-to-solid ratio on (gray ●) the extract concentration at the quasi-equilibrium, (gray ▲) the experimental yield, and (△) the ideal yield of the extraction of artemisinin from A. annua with ethanol at different temperatures: (a) 30 °C, (b) 50 °C, and (c) 60 °C.
concentrations at the quasi-equilibrium, the ideal yield (Yid) and the experimental yield (Yex) of artemisinin are shown. The ideal yield corresponds to a full recovery of the extract solution from the raffinate (eq 6), i.e., a complete separation between the liquid phase and the solid phase. Such a separation is technically unfeasible, since a part of the extract solution always sticks to the solid plant material. Therefore, the experimental yield corresponds to the amount of separated extract in the experiment. 1809
DOI: 10.1021/acs.iecr.5b04506 Ind. Eng. Chem. Res. 2016, 55, 1808−1812
Research Note
Industrial & Engineering Chemistry Research
Both the increasing amount of remaining humidity with increasing liquid-to-solid ratio and with increasing temperature can be explained by a decrease in viscosity and density of the liquid phase and, hence, easier penetration of the solid plant material by the solvent. Using the maximum initial amount of artemisinin in the plant material and the extracted amount, the reduced mass fractions of the corresponding raffinate phases were calculated via mass balancing. Afterward, the model equation of the quasiequilibrium line (cf. eq 3) was fitted to the experimental data via the MS Excel solver’s evolutionary algorithm. Figure 3 shows that the empirical equation fitted the experimental data well for all temperatures. Obviously, the slopes of the quasi-equilibrium lines differ from an ideal extraction (marked by a dashed line in Figure 3), in which the total amount of target component is
Thus, the ideal amount of extract (mE,id) in eq 6 must be substituted by the real amount of extract (mE). m T,E,id Yid (%) = × 100 m T,S,0 =
wT,E(mL + mT,E,id ) wT,S,0mS
× 100 (6)
where, in the numerator of the second equation line, (mL + mT,E,id) = mE,id. Then, through substitution, one obtains the expression wT,EmL
Yid (%) =
1 − wT,E
wT,S,0mS
× 100
The initial mass fraction of artemisinin determined in the plant material used is represented as wT,S,0, while mT,S,0 is the corresponding amount (mass). The former parameter was determined via multistage leaching extractions, as described by Dreisewerd et al.14 for the production of a completely leached plant material and resulted in 0.68 wt % of artemisinin. The masses of added liquid solvent and solid material are represented by mL and mS, respectively, while mT,E,id is the amount of artemisinin in the extract for an extraction with a complete separation of liquid and solid (in the case of experimental yield, mT,E,id must be substituted by mT,E). The corresponding mass fraction of artemisinin in the extract is represented as wT,E. Comparing the diagrams in Figures 1a−c, for all temperatures investigated, the extract concentration at the quasi-equilibrium decreased as the liquid-to-solid ratios increased, because of a stronger dilution of artemisinin. The highest extract concentrations were determined at 50 °C. The ideal yields were slightly influenced by the liquid-to-solid ratio only, indicating that the capacity of the solvent was not reached yet. However, the experimental yields strongly increased as the liquid-to-solid ratio increased, since the amount of remaining humidity decreased in relation to the total amount of solvent used in the experiment (e.g., from ∼75.6% at χ = 3 kg kg−1 to ∼6.4% at χ = 50 kg kg−1, 50 °C). Nevertheless, relative to the amount of solid material, the amount of remaining humidity increased as the liquid-to-solid ratio increased (cf. Figure 2). Furthermore, the remaining humidity coefficients increased as the temperature increased.
Figure 3. Quasi-equilibrium lines for the extraction of artemisinin from A. annua with ethanol at different temperatures: (a) 30 °C, (b) 50 °C, and (c) 60 °C. (Legend: (gray ●) experimental data, () fitted quasiequilibrium lines, (···) the negative part of the fitted quasi-equilibrium lines, and (- - -) diagonal (corresponding to an ideal quasi-equilibrium line.)
Figure 2. Remaining humidity coefficients for the extraction of artemisinin from A. annua with ethanol at (gray ●,) 30 °C, (○;- - -) 50 °C, and (●;· · ·) 60 °C. (Data points represent experimental data, and lines represent fitted curves.) 1810
DOI: 10.1021/acs.iecr.5b04506 Ind. Eng. Chem. Res. 2016, 55, 1808−1812
Research Note
Industrial & Engineering Chemistry Research transferred to the liquid phase and, therefore, would result in values of a1 = 1, a2 = 0, and a3 = 0 kg kg−1. Since extraction with ethanol at 50 °C led to ideal yields close to 100% for high liquidto-solid ratios (Figure 1b), the corresponding quasi-equilibrium line intercepted the abscissa at a zero concentration and, thus, showed that all artemisinin was extracted from the plant material. The other extractions had an a3 value higher than zero (cf. Table 1), meaning that artemisinin was partly unextractable at 30 and Table 1. Model Parameters Determined for the Extraction of Artemisinin from A. annua with Ethanol at Different Temperatures temperature [°C]
a1
a2
a3 [× 10−4 kg kg−1]
b1
b2
30 50 60
0.857 0.774 0.773
9.591 1.052 25.923
2.463 0.000 0.943
1.765 2.068 2.115
0.133 0.119 0.122
60 °C. Accordingly, the empirical equation chosen seems to be adequate for the representation of the nonideal quasi-equilibrium line. An explanation for the temperature-dependent extractability of artemisinin could not be finally determined in this study. The reason might be temperature-dependent chemisorption. Thermal degradation of artemisinin might be another explanation for the reduced extractable amount of artemisinin at 60 °C, compared to 50 °C. Physisorption can be excluded, since artemisinin does not adsorb onto the solid plant material when extracted with ethanol.14 Trapping of artemisinin in the internal structure of the plant material is also most unlikely, as artemisinin is produced and stored in the glandular trichomes,15 which are located on the surface of the plant’s leaves. The proposed model aims at predicting multistage extractions. Hence, the fitted model parameters of the single-stage extractions given in Table 1 were used to calculate multistage leaching extractions, and the results were compared to experimental data, which is shown in Figure 4. The model was able to predict the stage-dependent concentrations and accumulated masses. Particularly, the extractions at 30 and 60 °C resulted in reduced extractable amounts of artemisinin, which was also predicted by the model. Therefore, the concept of a nonideal quasi-equilibrium line with an unextractable part seems to be meaningful to describe phytoextraction processes and particularly for the simulation of multistage leaching extractions.
Figure 4. Comparison between predicted multistage leaching experiments and experimental data for extraction of artemisinin from 9.09 × 10−3 kg of A. annua with ethanol at a liquid-to-solid ratio of 10 kg kg−1 and different temperatures: (a) 30 °C, (b) 50 °C, and (c) 60 °C. (Legend: (gray ●) experimental concentrations, (○) predicted concentrations, (gray ▲) experimental accumulated masses of artemisinin extracted, and (△) predicted accumulated masses of artemisinin extracted.)
4. CONCLUSIONS The quasi-equilibrium model proposed was shown to be applicable for the prediction of multistage leaching extractions with a temperature-dependent extractable amount of a target component. Accordingly, the concept of a nonideal quasiequilibrium line seems to be meaningful for phytoextractions. Moreover, all model parameters were determined from singlestage extractions only and, thus, the number of experiments needed for process design can be reduced, since different combinations of the liquid-to-solid ratios and the number of stages can be predicted by the model. Nevertheless, the transferability to and applicability for other model systems should be investigated in the future. Another target of future works could be the application of the model to other operating modes, e.g., countercurrent extractions, which should be possible using small modifications of the mass balances only.
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: +49 (0)231/755-4325. Fax: +49 (0)231/755-2341. Email:
[email protected]. Notes
The authors declare no competing financial interest.
■
NOMENCLATURE
Latin Letters
a1 = first model parameter of the empirical quasi-equilibrium line equation
1811
DOI: 10.1021/acs.iecr.5b04506 Ind. Eng. Chem. Res. 2016, 55, 1808−1812
Research Note
Industrial & Engineering Chemistry Research
(10) Prabhudesai, R. K. Leaching. In Handbook of Separation Techniques for Chemical Engineers; Schweitzer, P. A., Ed.; McGraw− Hill: New York, 1997; pp 5.3−5.32. (11) Treybal, R. E. Mass-Transfer Operations; McGraw−Hill: New York, 1980. (12) Toledo, R. T. Fundamentals of Food Process Engineering; Springer: New York, 2007. (13) Takeuchi, T. M.; Pereira, C. G.; Braga, M. E. M.; Maróstica, M. R., Jr.; Leal, P. F.; Meireles, M. A. A. Low-Pressure Solvent Extraction (Solid−Liquid Extraction, Microwave Assisted, and Ultrasound Assisted) from Condimentary Plants. In Extracting Bioactive Compounds for Food Products: Theory and Applications; Meireles, M. A. A., Ed.; CRC Press: Boca Raton, FL, 2009; pp 137−218. (14) Dreisewerd, B.; Merz, J.; Schembecker, G. Determining the Solute−Solid Interactions in Phytoextraction. Chem. Eng. Sci. 2015, 134, 287. (15) Olsson, M. E.; Olofsson, L. M.; Lindahl, A.-L.; Lundgren, A.; Brodelius, M.; Brodelius, P. E. Localization of Enzymes of Artemisinin Biosynthesis to the Apical Cells of Glandular Secretory Trichomes of Artemisia annua L. Phytochemistry 2009, 70, 1123.
a2 = second model parameter of the empirical quasiequilibrium line equation a3 = third model parameter of the empirical quasi-equilibrium line equation (unextractable amount of target component), kg kg−1 b1 = first parameter of the remaining humidity coefficients b2 = second parameter of the remaining humidity coefficients f H = remaining humidity coefficient, kg kg−1 m = mass, kg w = mass fraction, kg kg−1 w* = reduced mass fraction (carrier-free basis), kg kg−1 Y = yield, % Greek Letters
χ = liquid-to-solid ratio, kg kg−1 Subscripts
0 = initial C = carrier component E = extract ex = experimental H = remaining humidity i = refers to component i I = inert component id = ideal j = refers to phase j L = liquid solvent LS = liquid solvent component pole = pole R = raffinate S = solid material T = target component
■
REFERENCES
(1) Bart, H.-J. Extraction of Natural Products from Plants−An Introduction. In Industrial Scale Natural Products Extraction; Bart, H.-J., Pilz, S., Eds.; Wiley−VCH: Weinheim, Germany, 2011; pp 1−25. (2) Bart, H.-J.; Hagels, H.; Kassing, M.; Jenelten, U.; Johannisbauer, W.; Jordan, V.; Pfeiffer, D.; Pfennig, A.; Tegtmeier, M.; Schäffler, M.; Strube, J. Positionspapier der ProcessNet Fachgruppe, “Phytoextrakte− Produkte und Prozesse”: Vorschlag für einen neuen, fachübergreifenden Forschungsschwerpunkt, 2012. Available via the Internet at the ProcessNet Web site: http://www.processnet.org/processnet_media/ PP_Phytoextrakte_2012_web-p-4026.pdf (accessed March 12, 2015). (3) Kassing, M.; Jenelten, U.; Schenk, J.; Strube, J. A New Approach for Process Development of Plant-Based Extraction Processes. Chem. Eng. Technol. 2010, 33, 377. (4) Bol, J. B.; Pfennig, A. Wissensbasierte Designmethode zur Auslegung von maßgeschneiderten Feststoffextraktoren auf der Basis von Laborversuchen; Final Report to the IGF-Project 16146 N; Forschungs-Gesellschaft Verfahrens-Technik e.V.: Frankfurt am Main, Germany, 2012. (5) Pfennig, A.; Delinski, D.; Johannisbauer, W.; Josten, H. Extraction Technology. In Industrial Scale Natural Products Extraction; Bart, H.-J., Pilz, S., Eds.; Wiley−VCH: Weinheim, Germany, 2011; pp 181−220. (6) Sattler, K.; Feindt, H. J. Thermal Separation Processes: Principles and Design; Wiley−VCH: Weinheim, Germany, 1995. (7) Brown, G. G.; Foust, A. S.; Katz, D. L.; Schneidewind, R.; White, R. R.; Wood, W. P.; Brown, G. M.; Brownell, L. E.; Martin, J. J.; Williams, G. B.; Banchero, J. T.; York, J. L. Unit Operations; John Wiley & Sons, Inc.: New York, 1967. (8) McCabe, W. L.; Smith, J. C.; Harriott, P. Unit Operations of Chemical Engineering; McGraw−Hill: Boston, MA, 2005. (9) Mersmann, A.; Kind, M.; Stichlmair, J. Thermal Separation Technology: Principles, Methods, Process Design; Springer: New York, 2011. 1812
DOI: 10.1021/acs.iecr.5b04506 Ind. Eng. Chem. Res. 2016, 55, 1808−1812