Purification and Recovery of Curcuminoids from Curcuma longa

Apr 27, 2011 - A novel reactive sorption method has been devised for purification of curcuminoids from a complex multicomponent turmeric extract using...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/IECR

Purification and Recovery of Curcuminoids from Curcuma longa Extract by Reactive Sorption Using Polymeric Adsorbent Carrying Tertiary Amine Functional Group Anil R. Patil and Vilas G. Gaikar* Department of Chemical Engineering, Institute of Chemical Technology, Matunga, Mumbai 400 019, India

bS Supporting Information ABSTRACT: A novel reactive sorption method has been devised for purification of curcuminoids from a complex multicomponent turmeric extract using a weakly basic polymer as a selective adsorbent. The experimental results are further supported by molecular simulation to quantify the specific interactions of curcumins with tertiary amine group of the adsorbent in the presence of organic solvents of different polarities. Both experimental data and theoretical calculations showed that methanol was the best solvent for adsorption, while acetone was useful for desorbing curcumins from the adsorbed state. Dynamic adsorption and particularly desorption operations showed a significant effect of intraparticle diffusivity limitations on the rate of adsorption and effective utilization of the adsorbent. However, increased temperature and exposure to ultrasound could increase desorption of curcumins from the column. Nearly 50% and 85% of curcuminoids are desorbed from the bed with acetone alone and with simultaneous exposure to ultrasound, respectively. After a single adsorption and desorption cycle, the purity of recovered curcuminoids increases to 98%. Individual components of the extract and the purified curcuminoid product were characterized by liquid chromatographymass spectrometry.

’ INTRODUCTION Curcuminoids form a group of compounds of similar structures, i.e., curcumin, bisdemethoxycurcumin, and demethoxycurcumin, which are present as active phytochemicals in Curcuma longa (turmeric) species (Figure 1). Curcuminoids are extensively used as food coloring agents, natural antioxidants, and spice condiments. In recent years, curcuminoids have been increasingly investigated for their potent anticancer and antibacterial properties,1,2 and as modest inhibitors of HIV-1 and HIV-2 proteases.3 The curcuminoids are mostly obtained by extraction from turmeric with organic solvents such as ethanol, ethyl acetate, and acetone.4,5 A nonselective extraction of the active components along with other oleoresinous materials is the main problem associated with these solvents that results in a complex product mixture where the active component usually is a minor constituent. Extraction using the organic solvents is also commonly associated with an extensive loss of the solvents because of higher volatility even at low temperatures. The consumption of the solvent adds significantly to the cost of the final product apart from the environmental, health, and fire hazards. New techniques such as supercritical CO2 based extraction, microwave-assisted extraction, and aqueous hydrotropic solution based extraction have been reported in recent years for the extraction of curcuminoids from turmeric.68 The turmeric extract is typically a mixture of oleoresins comprised of essential oils and curcuminoids. The turmeric extract also contains volatile oil (27% w/w) consisting of bisabolane and sesquiterpenes, such as R-turmerone, curlone, β-turmerone, zingiberene, curcumenone, and curcumenol in different amounts.914 In many cases these components are present in larger amounts r 2011 American Chemical Society

than curcuminoids themselves in the extract. The presence of the essential oils also assists in extraction of the curcuminoids into the organic solutions because of molecular interactions between them. Various separation techniques have been reported for the recovery, purification, and isolation of curcuminoids from the turmeric extract.1521 Among these, chromatography, in different forms, is the most favored process for purification of curcuminoids. High-speed countercurrent chromatography (HSCCC)15 has been reported for separation of the curcuminoids from the extract using a multiple solvent system composed of n-hexane/ chloroform/methanol/water. Patel and Krishna16 have reported a pH zone refining method for the purification of curcuminoids taking advantage of the weakly acidic nature of the curcuminoids. Flash chromatography17,18 using a silica gel column also has been reported for the purification of curcuminoids, where the polarity of the mobile phase is gradually increased to desorb polar compounds of the mixture. Temperature-induced crystallization is another common method, and repeated crystallization of curcumin from ethanol is practiced industrially.19 Madsen and Hidalgo20 have shown an increased yield of the curcuminoids by additional stages of crystallization as the first crystallization step gave only 60% of the curcuminoids as the product. The multistep crystallization process still involves usage of multiple organic solvents and subzero temperatures which finally adds to the cost of the product.

Received: May 8, 2010 Accepted: April 27, 2011 Revised: March 25, 2011 Published: April 27, 2011 7452

dx.doi.org/10.1021/ie100998p | Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

Figure 1. Structures of curcuminoids.

We present here a unique approach based on understanding interactions of curcuminoids with the solvents used for the extraction by molecular modeling. Further, we have explored the reactive sorption of the curcuminoids on a basic polymer because of the phenolic nature of the curcuminoids. The process involves the use of a functionalized polymer with tertiary amine groups on a hydrophobic matrix as a selective adsorbent for curcumins from the turmeric extract in nonaqueous conditions. There has been considerable interest in purification of natural products by sorption on macroporous polymeric adsorbents due to easy recovery of the products and reusability of the polymeric adsorbents.2126 The functional group on the polymer matrix can be tailored for the desired separation and selectivity, giving an operational flexibility. The present work shows highly selective adsorption of curcumins from the crude turmeric extract on the polymer. The impurities in the crude turmeric extract are also characterized by mass spectroscopy.

’ MATERIALS AND METHODS All AR-grade solvents and chemicals were procured from S.D. Fine Chemicals, Mumbai, India. Dried roots of turmeric were purchased from a local market. A functionalized polymer, Indion 860, was obtained from Ion Exchange (I) Ltd., Mumbai, India. It is a weak base polymer with tertiary amino groups having a polystyrene divinylbenzyl copolymer as a backbone. The exchange capacity of the polymer is 4.2 mol/kg on a dry weight basis. The polymer, however, also showed the presence of a small amount of quaternary groups. It is a macroporous polymer having particle size ranging from 0.42 to 1.2 mm and specific surface area of 34 m2/g. Curcumin, isolated using a reported procedure19 and purified by a two step crystallization process, was used as the standard. Experimental Methods. Conditioning of the Adsorbent. The polymer beads were treated with an aqueous solution of sodium hydroxide (5% w/v) followed by a thorough washing with a large volume of deionized water to make the polymer completely free of alkali. The polymer was then further washed with methanol to

ARTICLE

remove the residual moisture and then dried in an oven at 333 ( 1 K, and then cooled to room temperature of 303 ( 1 K for storage. Kinetic Studies. The rate of adsorption of pure curcumin was studied separately in a three-necked fully baffled cylindrical glass reactor of capacity 150 cm3. A known quantity of the polymer beads was placed in the reactor, and a solution of curcumin in methanol was added to it. The suspension was agitated with a four flat blade turbine impeller at 304 K. Samples were withdrawn from the suspension at regular time intervals for the determination of residual concentration of curcumin. Batch Adsorption of Pure Curcumin. The equilibrium adsorption studies were carried out in tightly stoppered conical flasks to avoid the loss of solvent during the equilibrium process. A known amount of the polymer adsorbent was taken in the flask, and a solution of curcumin in an organic solvent was added it. The flasks were then kept in a constant-temperature bath with orbital shaking. After 4 h, when equilibrium was attained, the solutions were analyzed by high performance liquid chromatography (HPLC) to determine the residual concentration of curcumin. The equilibration time of 4 h was optimized by independent studies. Extraction from Turmeric. The extraction of curcuminoids was studied in a fully baffled cylindrical glass vessel (250 cm3) equipped with a four blade turbine impeller (2 cm diameter). Turmeric powder (15 g) of an average particle size of 400 μm was suspended in methanol (150 mL) by maintaining the solid/ solvent ratio of 10% (w/v), and the mixture was agitated for 2 h at a speed of 1000 rpm at 300 K. The suspension was then vacuum filtered with an ice bath to reduce the loss of methanol. The residue was washed with pure methanol for complete recovery of the extracted components from the cake. The filtrate was a mixture of curcuminoids along with oleoresins, terpenes, and other impurities. In a typical run, after evaporation of solvent, 1.108 g of a crude extract was recovered containing curcuminoids (0.717 g), turmerone (0.052 g), R-pinene (0.078 g), and β-bisabolane (0.165 g). Two additional unidentified components were also observed in the extract. The purity of curcumin and that of other components were established against pure curcumin as standard. Analytical Methods. The crude turmeric extract was analyzed using a high performance liquid chromatograph (HPLC) (Jasco 1090 series), equipped with a Hypersil C-18 column (length 250 mm, diameter 5 mm). A mixture of acetonitrilewater in 55:45 ratio was used as the mobile phase with a flow rate of 1.0 cm3/min. A PDA detector (JASCO MD 2010 series) was attached to the HPLC unit for detection. The product mixture was also subjected to LCMS analysis to characterize curcumin and its analogues, on a Finnigan LCQ Advantage Max mass spectrometer (LCQAD 30000, Thermo Electron Corp.), using the same column and mobile phase as the HPLC analysis. Pure and dry nitrogen was used as a sheath gas or nebulizer gas with a flow rate of 40 cm3/min, and an auxiliary nitrogen gas flow rate was maintained at 18 cm3/min. The capillary temperature was maintained at 548 K with the voltage at 420 V and the ion spray voltage at 5 kV. Column Studies. A glass column (diameter 1.4 cm) was packed with the pretreated adsorbent up to a height of 15 cm. The extract solution was pumped at 1.0 cm3/min flow rate in the upward direction. Samples were collected from the effluent stream of the column at frequent time intervals of 510 min at the beginning and 2530 min in the later stages. The operation 7453

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

Figure 2. Structure of weakly basic adsorbent.

was continued until the exit concentration of curcumins increased to the inlet concentration. When the column was exhausted completely, desorption of the curcumins was carried out using acetone, with and without ultrasound, followed by hot methanol wash for complete desorption.

’ RESULTS AND DISCUSSION Molecular Simulation for Interaction with the Adsorbent. The turmeric extract contains curcumin, its two analogues, i.e., demethoxycurcumin and bisdemethoxycurcumin, and a number of other impurities, mostly sesquiterpene compounds as turmeric oil. Figure 1 shows that curcuminoids are phenolic compounds with the acidic OH groups on two separate aromatic rings. In curcumin, both OH groups are flanked by two methoxy groups, while in demethoxycurcumin one of the methoxy groups is absent. In bisdemethoxycurcumin, both methoxy groups are absent. The presence of the methoxy groups has a direct bearing on the charges on the OH groups as well as the availability of the OH groups for interaction with other species in their vicinity, including the solvent molecules. For solvents, with H-bond acceptor characteristics, the ability to interact with the OH groups is influenced by the steric effect of the methoxy group(s). To separate curcuminoids from the crude extract from other components, such as essential oils and sesquiterpenes, it was decided to use a functional polymer as other components will not have any interaction other than van der Waals interactions. It was necessary to develop a suitable adsorbent that shows maximum interaction with the curcuminoids even in the presence of the other species. Curcuminoids being slightly acidic are expected to interact with the amino group of the functionalized polymer through Lewis acidbase interactions. To strengthen these interactions, we decided to maintain nonaqueous conditions for the adsorption. The strength of the interactions must be strong enough to give requisite selectivity in separation but also weak enough to recover the curcuminoids from the adsorbed state. In aqueous solutions, the ionic interactions are screened by the medium itself and alkaline conditions would cause degradation of curcumins. The lone pair of electrons on nitrogen of the amino group of Indion 860 polymer (Figure 2) can interact with the acidic hydrogen of the curcumin, forming an H-bond. We have used Material Studio (version 3.2, Acclerys) for characterizing the interactions between the interactive centers on the molecules using molecular modeling. Each curcuminoid and

ARTICLE

functional unit of the adsorbent were created separately in a vacuum and geometry optimized for their intrinsic energy using a COMPASS (Condensed-phase Optimized Molecular Potential for Atomistic Simulation Studies) force field.27 The calculation of the charges on different atoms of the structure was done using QEq.2730 COMPASS is a high-quality ab intio force field that consolidates parameters of organic and inorganic materials. In both these studies, van der Waals and electrostatic interactions were included in the calculation of the optimized energy in the FORCITE module of Material Studio. The basis of the QEq method is equilibration of atomic electrostatic potentials with respect to local charge distribution. The charge groups were applied with a cutoff distance of 3.2 Å for interactions. An in-built steepest descent algorithm with a root-mean-square gradient of 1  104 kcal/mol was used for convergence of the solution. The average energy of the system was normalized to kilocalories per mole. The simulation method of the geometry optimization was similar for all systems. The interaction between each curcuminoid with the polymer unit was studied again by optimizing the structure of a curcuminoid and two monomer units of the polymer, on both sides of the curcuminoid structure, in a vacuum using again a COMPASS force field. The interaction studies were further conducted in different solvent media, by simulating the solution with a solvent box of a fixed dielectric constant containing discrete solvent molecules. A cubic simulation box was created with periodic boundary conditions in all directions. Typically, the periodic box of size 35 Å  35 Å  35 Å contained 50 solvent molecules. One molecule of curcuminoids and two repeating units of the amine on a polymer’s aromatic structure were added to it. The entire system was then geometry optimized for the lowest free energy configuration. The optimum configuration was selected as the stable complex of curcuminoids with the adsorbent. The interaction energy (IE), in a vacuum, was calculated from energies of individual species: IE ¼ energy complex  ðenergycurcuminoids þ energyadsorbent Þ ð1aÞ The interaction energy in the solvated state was calculated as follows: IE ¼ energy solvated complex  ðenergy curcumin þ energyadsorbent þ energy solvent box Þ

ð1bÞ

The optimized energies of curcumin, demethoxycurcumin, and bisdemethoxycurcumin, in a vacuum, were 24, 36, and 47 kcal/mol, respectively, indicating a more stable structure for bisdemethoxycurcumin compared to the other two analogues. The charges on interactive centers, i.e., the phenolic OH and nitrogen of amino group on the adsorbent, were estimated from the optimized geometries. The nitrogen atom of the polymer carries a charge of 0.374, while the charge on the phenolic hydrogen of curcumin is 0.289. A hydrogen bond is thus expected by electrostatic interaction between these two centers. The charges on the phenolic hydrogens of the other two curcumin analogues are 0.287 and 0.285. The presence of methoxy groups on the adjacent carbons of OH groups makes curcumin a relatively stronger phenol compared to the other two components. A slightly higher charge on the acidic hydrogen of OH in the case of curcumin makes its interaction with the polymer stronger, while for the other species it would be weaker 7454

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

Figure 3. Interactions of curcumin and its analogues with polymeric adsorbent. (a) Curcumin with ligand. (b) Demethoxycurcumin with ligand. (c) Bisdemethoxycurcumin with ligand.

in a relative sense. The distance between the nitrogen atom of the Indion 860 and the phenolic hydrogens of the curcuminoids was also monitored after optimization of the structures of two polymer units and a single curcuminoid molecule. For both polymeric units, on interaction with curcumin, the distances are within the hydrogen bond length, 2.2 and 2.3 Å in a vacuum. In the case of demethoxycurcumin, the bond distances are 2.5 and 2.45 Å, whereas for bisdemethoxycurcumin, the bond distance increased to 2.52 and 2.55 Å. A shorter H-bond distance for curcumin suggests a greater affinity of curcumin toward the polymer as against its two analogues in a vacuum. This also indicates a possibility of even separating curcumin and its analogues from each other by using the same polymer. The molecular simulation of interactions between curcuminoids and the polymer indicated a stronger interaction and an attractive possibility of using the functional polymer for selective sorption of curcumins from the crude complex product obtained in the extraction process. Since the other components of the crude turmeric extract, such as terpenic compounds, are neutral in nature, their specific interaction with the polymer’s functional group was expected to be negligible.

ARTICLE

The bond distances, between the interacting centers, were also checked in solvated conditions to identify the solvent effect on the sorption. In methanol, the bond distance between nitrogen of the amino group and phenolic hydrogen of the curcumin decreased to 1.75 and 1.69 Å on either side of the molecule. In the cases of demethoxycurcumin (1.8, 1.87 Å) and bisdemethoxycurcumin (1.91, 1.93 Å), also, the bond distances decreased in methanol solutions, as if methanol drives the curcuminoids toward the polymer phase. Even in the solvated conditions the interaction of curcumin is apparently stronger than that of the other two species. In the cases of demethoxycurcumin and bisdemethoxycurcumin, the absence of one and two methoxy groups, respectively, next to the phenolic OH group helps in formation of intermolecular hydrogen bonds also with the OH group of methanol. However, in the case of curcumin, both methoxy groups sterically hinder the hydrogen bond formation with methanol. No specific interaction was observed between the polymer and the solvent molecules, however. In the case of acetone as a solvent, the bond distance between hydrogen of curcumin and nitrogen of polymeric adsorbent was 2.56 Å on one side and the other side it was 2.6 Å. Similarly, for demethoxycurcumin and bisdemethoxycurcumin the distances were 2.59 and 2.49 Å on one side and on the other side they were 2.46 and 2.51 Å, respectively. There was no significant change in the H-bond length in acetone solutions compared to the interaction between the two species in a vacuum. This gives the interesting possibility of identifying the solvent effect on adsorption of curcumins in each of these solvents. Since MeOH gave a better interaction between the adsorption centers, adsorption should be very favorable from methanolic solutions while acetone seems to reduce the interaction of curcuminoids with the polymer and thus should be useful to desorb the curcumins from the adsorbed state. MeOH is an H-bond donor as well an H-bond acceptor, but acetone with its carbonyl group is a stronger H-bond acceptor because of the basic carbonyl oxygen and should be able to interact with acidic phenolic hydrogen much better, reducing its interaction with nitrogen of the amine group on the polymer matrix. The interaction energy was the least for bisdemethoxycurcumin in a vacuum (21.3 kcal/mol), intermediate for demethoxycurcumin (23.1 kcal/mol), and maximum for curcumin (25.3 kcal/mol), indicating again a better interaction of the latter with the functional group. Figure 3 shows, in solvated state, the interaction of both OH groups of curcumin on either side of the molecule with the nitrogen atoms on two amino groups by hydrogen bonding. The solvent molecules are not shown for clarity. In the cases of demethoxycurcumin and bisdemethoxycurcumin, the absence of one and two methoxy groups, respectively, helps in the formation of intermolecular hydrogen bonding of the OH groups with methanol solvating them in a better way, whereas in the case of curcumin the presence of two methoxy groups on both sides sterically hinders the hydrogen bond formation with methanol and thereby reduces its tendency to remain in solution. After optimization in solvated conditions, the interaction energy for adsorption on the polymer was highest for curcumin (51 kcal/ mol), intermediate for demethoxycurcumin (41 kcal/mol), and the least for bisdemethoxycurcumin (35 kcal/mol), thereby suggesting a greater affinity of curcumin toward the adsorbent from the methanol solutions. In the case of acetone, the interaction energies for adsorption are lower than that of methanol and also the values in a vacuum. The interaction energy values for 7455

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

Figure 4. Electrostatic potential surfaces. (a) Curcumin with polymeric adsorbent. (b) Demethoxycurcumin with polymeric adsorbent. (c) Bisdemethoxycurcumin with polymeric adsorbent.

curcumin, demethoxycurcumin, and bisdemethoxycurcumin in acetone are 12, 7, and 3 kcal/mol, respectively. The comparison of these interaction energies in two solvents indicates that adsorption of curcumin will be selective from both solvents, but methanol appears to be a good solvent for adsorption while acetone should be a better solvent for desorption. The optimized structures were further studied for their quantitative structureactivity relationship (QSAR) properties, viz., heat of formation and electrostatic potential surface. Electrostatic potential surface (ESP) is a measure of electronegativity among various atoms in a molecule with contributions from both nuclei and electrons. ESP has been particularly useful for rationalizing the interactions between molecules and molecular recognition processes. Figure 4 shows electrophilic regions of curcumins that get attracted to the nucleophilic region of the nitrogen of the polymer. The electronic density on the molecules at different regions indicates the orientations these molecules would have toward interacting centers on the polymer matrix. A more useful quantity for comparison with experiments is the heat of formation, i.e., the enthalpy change on formation of a structure. In the case of the curcuminadsorbent pair, the heat of formation was less than the sum of their individual heats of formation. The decrease for the curcuminpolymer pair (117 kcal/mol) was greater than that for the demethoxycurcuminadsorbent pair (96 kcal/mol) and significantly lower for the bisdemethoxycurcuminadsorbent pair (64 kcal/mol). A lower decrease in the hearts of formation of bisdemethoxycurcumin and demethoxycurcumin indicates that those complexes have less stable natures. The entire exercise of molecular modeling points to use of methanol as a solvent for extraction of curcuminoids and their selective adsorption on the polymeric

ARTICLE

Figure 5. Langmuir adsorption isotherms for curcumin in different solvents with reproducibility at 298 K, on Indion 860 polymeric adsorbent. Initial curcumin concentration: 0.10.5 mol/dm3.

adsorbent with amino groups, along with a strong solvent effect. Characterization of the Components in the Extract. The HPLC analysis under the given analytical conditions showed the presence of β-bisabolane, R-pinene, and R-turmerone, which was further confirmed by LCMS. The MS spectrum revealed [M  H] peaks at 367, 337, and 307 for curcumin, demethoxycurcumin, and bisdemethoxycurcumin, respectively. R-Turmerone, β-bisabolane, and R-pinene were also identified as [M  H] peaks at 216, 203, and 135, respectively. The retention times of β-bisabolane, R-pinene, and R-turmerone are 2.3, 4.0, and 9.8 min, respectively, and those of demethoxycurcumin, bisdemethoxycurcumin, and curcumin are 6.5, 7.0, and 7.8 min, respectively. Batch Equilibrium Adsorption of Curcumin. For batch equilibrium studies the feed curcumin concentration was varied from 0.1 to 0.5 mol/dm3 to determine the sorption from solutions of curcumin in acetone, ethanol, and methanol, on Indion 860 adsorbent at 303 K. Curcumin is strongly adsorbed on the adsorbent from solutions in all three solvents. Figure 5 shows that the maximum uptake was from methanol solutions (1.4 mol/kg adsorbent) followed by ethanol solutions (1.3 mol/ kg adsorbent) and then acetone solutions (1.2 mol/kg adsorbent). Interestingly, these experimental results validate the predictions from the molecular simulations. Although the uptake was lower in acetone, it was still strong enough to pick up curcumin selectively and completely from the acetone solutions. Figure 5 also shows that the curcumin strongly adsorbs on the polymer even from polar solvents and follows a characteristic Langmuir isotherm. Table 1 shows the Langmuir equilibrium constant (K) and maximum loading capacity (qmax) for curcumin obtained by fitting the equilibrium data into the standard Langmuir isotherm equation. K is a measure of relative interactions of a 7456

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

ARTICLE

Table 1. Comparison between Langmuir and Modified Langmuir Constants with Different Solvents and at Different Temperatures solvent methanol

temp (K)

qmax (mol/kg)

K (dm3/mol)

ΔH (kcal/mol)

ΔS (kcal/mol)

ΔG (kcal/mol)

0.0086

3.45

0.0084 0.0086

3.40 3.33

298

1.4

343

310 315

0.79 0.77

252 95

6.02 6.02 6.02

321

0.72

186

6.02

0.0084

3.33

325

0.59

140

6.02

0.0087

3.19

ethanol

298

1.3

243





3.2

acetone

298

1.2

117





2.8

Figure 6. Langmuir adsorption isotherms for curcumin in methanol at different temperatures. Adsorbent: Indion 860 polymer. Initial curcumin concentration: 0.10.5 mol/dm3.

solute with an adsorbent and also depends on the solvent. It was higher in methanol, indicating better adsorption of curcumin from methanol solutions. The adsorption of an adsorbate on a given adsorbent has to compete with its solvation in the solution depending upon relative interactions of the adsorbate with the solvent. If a solute is preferably solvated by a solvent, its adsorption tendency becomes weaker while, in poorly solvated conditions, the tendency to get adsorbed by moving out of the solution is stronger. The preferential solvation also should affect the solubility of curcumin in different solvents. The solubility of curcumin in three solvents was measured separately at a room temperature of 303 K. The values are 203, 38.2, and 29.6 mmol/L, respectively, in acetone, ethanol, and methanol. Curcumin has the highest solubility in acetone and the least one in methanol. The higher solubility of curcumin in acetone indicates its favorable solvation by acetone and thereby reduces its tendency

to adsorb on the polymer from the acetone solutions. The solubility of curcumin is lowest in the methanolic solutions, indicating a less solvated curcumin which should, therefore, show a greater affinity for the adsorbent from methanolic solutions. Acetone can be used as an effective desorbent for the recovery of curcuminoids from the adsorbed phase as even predicted by molecular simulation. The interaction responsible for the adsorption of phenolic curcumins on amino groups of the polymer is of acidbase type. Hydrogen bond formation thus takes place between the lone pair of electrons on the nitrogen of the amine group of Indion 860 and the acidic hydrogen of curcumin. The carbonyl oxygen of acetone also is an H-bond acceptor and has stronger Lewis basicity because of the stronger electronegativity of oxygen. To a certain extent acetone interacts better with the acidic curcumin and, therefore, interferes in the adsorption of curcumin on the adsorbent. Temperature has a strong effect on the adsorption of curcumin on the polymeric adsorbent. Figure 6 shows that with increase in temperature the adsorption capacity of the polymer decreased, indicating the exothermic nature of the process. The Langmuir adsorption constants and loading capacity, therefore, show a decreasing trend with increase in temperature (Table 1). The thermodynamic parameters for the adsorption such as the free energy change (ΔG), enthalpy change (ΔH), and the entropy change (ΔS) were also calculated using the van’t Hoff equation and are reported in Table 1.31 The negative values of ΔG indicate spontaneous adsorption of curcumin on the adsorbent. The negative ΔH values imply the exothermic nature of the adsorption process, and higher temperature should reduce the extent of adsorption. Intraparticle Diffusitivity of Curcumin. The adsorption of curcumin from a solution on an adsorbent takes place basically in three steps. The first step consists of diffusion of curcumin through a liquid film surrounding the polymer particle to the solid surface. The second step consists of diffusion of curcumin into the internal surface area of a polymer particle, and the last step involves the adsorption of curcumin on the functional sites. The adsorption, if acid base type, should take place very quickly on the amine sites and should not control the sorption process. For the well-mixed conditions, the external mass transfer also must be faster than the intraparticle mass transfer. It is, therefore, the transport of the molecules in the polymer phase, through the intertwined polymer chains by hindered molecular diffusion, that decides the rate of the uptake of the curcumin. The kinetic studies were, therefore, essential to understanding the effective diffusion rate of the molecules in the polymer matrix. The mass balance in the bulk liquid solution phase describes a relation between the solution concentration and mass transfer 7457

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

ARTICLE

Figure 7. Dimensionless concentration of curcumins in effluent versus time. Temperature: 300 K.

Figure 8. Dimensionless concentrations of components in column effluent versus number of bed volume. Glass column: diameter 1.4 cm and length of 15 cm. Flow rate of 1.0 cm3/min.

and boundary conditions into the solid adsorbent phase and can be written as3243 dC0 ¼ ksl ap ½C0  Ci ðRp , tÞ V dt

ð2Þ

where V is the initial volume and ap is the total particle surface area. The solidliquid external mass transfer coefficient (ksl) was estimated, from an empirical correlation given by Levins and Glastonbury,44 to be 2.74  106 m/s. The mass balance within the particle at a distance r from the center of the particle reduces to   1 d 2 dCi dCi dq r ð3Þ þ Fp ð1  εi Þ D ¼ εi e 2 r dr dt dt dt where Ci = Ci(r,t), q is the adsorbed concentration per kilogram of the adsorbent, εi is internal porosity, and F is density of the adsorbent in g/cm3. A local equilibrium is assumed to be instantaneous at the active site; i.e., no resistance is assumed to exist in the actual adsorption step. In that case, the adsorbed concentration can be expressed in terms of the local solution concentration.    dq dq dCi ¼ ð4Þ dt dCi dt with initial conditions at t ¼ 0;

Ci ðr, 0Þ ¼ 0

and

C0 ð0Þ ¼ C0

De

dCi ¼ ksl ½C0  Ci ðRp ; tÞ dr dCi ð0; tÞ ¼0 dr

at r ¼ Rp

ð5Þ

at r ¼ 0

The plot of normalized curcumin concentration in methanol solutions versus time is shown in Figure 7. The curve denotes the fitted values from the model, while the points represent the experimental data. The polymeric adsorbent gave an effective diffusivity (De) of curcumin as 2  1010 m2/s in the solutions. The estimated internal mass transfer coefficient from Glueckauff’s correlation43 was on the order of 4  107 m/s for a particle of size 1 mm and an assumed value of tortuosity factor of 2.5. This value clearly indicates that the intraparticle diffusion, being at least an order of magnitude lower than external mass transfer, is dominating the adsorption process of curcumin in the polymer phase.44 Very poor intraparticle diffusivity puts a limit on the operating flow rate of the solution over the adsorbent. It also can limit the regeneration of the adsorbent as diffusion of molecules from inside the particles can be an extremely slow process. A curcumin molecule also has two points of attachment to the polymer structure. Dislodging both points of attachment simultaneously will be a difficult process. Thus, desorption of curcumin can be very slow. Indeed, experimentally the desorption of curcumin from the adsorbed state, as discussed later in this work, was found to be a limiting factor. Column Studies. The column adsorption studies were conducted at different initial concentrations of curcuminoids with a 7458

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research

ARTICLE

Figure 9. Impurity profiles in petroleum ether wash. Flow rate of 1.0 cm3/min. Total washing time: 120 min. Temperature: 298 K.

column filled with the adsorbent beads. Figure 8 shows the dimensionless effluent concentrations of six components of the turmeric extract versus the number of bed volumes of the solution that has been passed through the column. The uptake by the polymer from the methanol solutions was maximum for curcumin (70%), intermediate for demethoxycurcumin (62%), and minimum for bisdemethoxycurcumin (52%) until 10 bed volumes of the solution. Apart from curcuminoids, 34, 33, and 49% of R-turmerone, R-pinene, and β-bisabolane were also adsorbed, respectively, from the feed solution. R-Pinene and β-bisabolane are organic compounds of the terpene class and are found as oils in most natural products. As R-pinene and β-bisabolane are hydrocarbons, their interaction with the functional group of the polymeric adsorbents is very poor. At 10 bed volumes of the feed solution when the curcumin concentration in the exit effluent reached 6% of the feed concentration, the polymer had adsorbed a total of 0.516 mmol of curcumins, which corresponds to a loading capacity of 0.103 mol/kg adsorbent. Nearly 23 bed volumes of the solution was required for complete saturation of the column. After the adsorption run was over and the column was drained of the solution, a petroleum ether (4060 °C) wash was given to remove the impurities from the column. Since curcuminoids are insoluble in petroleum ether, their desorption is almost negligible, while 98% of the adsorbed impurities, i.e., R-pinene, R-turmerone, and β-bisabolane, were washed out of the column with petroleum ether (Figure 9). The differential uptakes indicate that curcumin has a greater affinity toward the polymer from a complex extract solution. The uptake of the curcuminoids by the polymer is dependent on the charges on the hydrogen atom of the hydroxyl group and their

Figure 10. Desorption of curcuminoids by acetone in the presence and absence of ultrasound. Filled symbols: ultrasound-assisted desorption; open symbols: desorption without ultrasound. Flow rate of 1.0 cm3/min. Temperature: 298 K.

ability to come close enough to the amino group to form hydrogen bonds and also on the ability of the solvent to form competitive hydrogen bonds. In the cases of demethoxycurcumin and bisdemethoxycurcumin the absence of one and two methoxy groups, respectively, results in strong or intermolecular hydrogen bonding with the methanol molecules compared to curcumin and thereby reducing the tendency of curcumin to remain in the solvent phase. The impurities being relatively less polar showed moderate adsorption profiles. Curcuminoids were recovered from the column by desorption using acetone after draining the column of the petroleum ether. Nearly 21 bed volume equivalents of acetone was required for 60% desorption of the curcuminoids (Figure 10) with 98% purity as a product. Due to intraparticle diffusion only 60% of curcuminoids was desorbed in the given time frame with acetone as a solvent. To remove the remaining curcuminoids, methanol at 323 K was passed through the column which increased the recovery to 95% of curcuminoids. However, using organic solvents at higher temperatures can lead to significant losses on a larger scale operation. The efficacy of acetone to desorb major amounts of curcuminoids also validates the results of molecular simulation. When compared with results of Koichi and Chihiro15 using high-speed countercurrent chromatography (HSCCC) where the percent recovery of curcumin, demethoxycurcumin, 7459

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research and bisdemethoxycurcumin was 4.4, 2.4, and 3.4%, respectively, the percent purity of total curcuminoids obtained with the functional polymer is more than 98% and the percent recovery of curcumin, demethoxycurcumin, and bisdemethoxycurcumin is 4, 2.5, and 2%, respectively, on the basis of initial turmeric feed. The disadvantage of HSCCC method is that the time required for the process is longer with concomitant dilution of the feed, but in the new method the resolution was completed within 3 h simply on the basis of selective adsorption without chromatographic operations. The new adsorptive process also can be scaled up easily compared to chromatographic processes. Patel and Krishna16 purified the curcuminoids by two methods, i.e., by countercurrent chromatography (CCC) and the pH zone refining method. However, the pH zone refining method is very confusing due to a large number of parameters of the mobile phase such as number of solvents, pH, and fraction ranges. Therefore, this method cannot be used for large scale purification of curcuminoids. Darrick17 have also purified the curcuminoids by using phenol protecting groups such as methanesulfonate, benzylsulfonate, etc., and recrystallization. The purity obtained by these methods is up to 99%. However, the recovery of curcumin was found to be very low. Another disadvantage of this method is that the use of different protecting groups increases the number of steps of processing in the final recovery of curcumin and ultimately the cost of the process. In one of the reported methods, temperature-induced crystallization19 was used wherein the purity obtained was up to 95%, using ethanol as a solvent. In this case, the recovery of the curcumin is very low, and solvent consumption is in huge amounts, so this method has also some limitations. Adsorption on functional polymers, therefore, seems to be an efficient and effective adsorption technology in which separation of curcuminoids can be carried out in an efficient manner. The recyclability of the adsorbent bed was checked by repeated adsorption and desorption cycles on the same polymer bed under the same conditions. When desorption was carried out using acetone but without ultrasound, the polymer bed could be used four times with 43% loss of adsorption capacity. The regeneration efficiency of the polymeric adsorbent decreased with the increasing number of runs even with desorption by methanol at higher temperatures. By the fourth run, the recovery of curcumin declined to 86%. It was suspected that, due to very slow intraparticle diffusion, complete desorption of curcuminoids was not taking place, reducing the capacity of the column with every cycle. In order to improve recyclability, removal of curcumin from the adsorbed state before the next cycle is essential. To enhance the percent adsorption of curcuminoids, the desorption studies were conducted by putting the adsorption column in an ultrasound bath. Nearly 85% of the adsorbed curcuminoids was desorbed by using simultaneous ultrasound while passing acetone over the bed and thereby increasing the percentage recovery of curcumin. After ultrasound-assisted desorption and methanol wash at 323 K, the polymer was efficiently used for four cycles with only 3% loss of curcuminoids. In this case with up to four continuous successive runs a 15% decrease in the adsorption capacity of the polymer was observed. Acoustic cavitations and thermal function have a direct effect on desorption of curcuminoids. Cavitation under ultrasound seems to be an important factor for efficient contact between adsorbent and the solvent. Under ultrasound irradiation microbubbles are formed in the solution phase. When negative pressure is high enough, microbubbles grow during the negative pressure and compress during the positive pressure. The expansion and

ARTICLE

compression causes constant pulsating or violent collapsing of the microbubbles. When collapse occurs near the solid adsorbent surface, it facilitates the release of the components and due to this the percent desorption of the curcuminoids increased in the presence of ultrasonication as shown in Figure 10. Ultrasound seems to be, therefore, an effective way of improving desorption of adsorbed solute and regeneration of the bed. The crude turmeric extract showed the presence of curcuminoids (65%), turmerone (4.68%), R-pinene (7.04%), and β-bisabolane (15.3%) as confirmed by LCMS. Apart from these components, two components present in the crude extract were unidentified. The purity of the recovered curcuminoids, after the column treatment, is 98% indicating effective removal of the impurities from the crude extract.

’ CONCLUSION Selective adsorption of curcuminoids was achieved from complex turmeric extract by using an amine functionalized polymer. Most importantly, molecular simulation could give guidelines for selection of a solvent for adsorption apart from predicting the type of the interactions present between the interacting molecules. By judicious selection of solvents, it is possible to manipulate adsorption and desorption of curcuminoids on the polymeric adsorbents. Nearly 50% adsorption of curcuminoids takes place with acetone alone, while 85% desorption takes place with simultaneous ultrasonication during acetone regeneration of the adsorbent. Almost complete desorption of curcuminoids was possible using methanol at 323 K. The purity of curcuminoids increased from 65 to 98% by a single cycle of adsorption and desorption. ’ ASSOCIATED CONTENT

bS

Supporting Information. HPLC chromatograms of crude extracts of turmeric and pure curcuminoids; LCMS spectra of crude extract of turmeric; tables detailing reusability of polymeric adsorbent without ultrasound and with ultrasound desorption. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Fax: 91-22-33612013. E-mail: [email protected].

’ ACKNOWLEDGMENT A.R.P. is grateful for the award of JRF from the University Grant Commission (UGC), India. The authors also acknowledge the financial support from Department of Science and Technology, GOI. ’ NOTATION ap = surface area of polymer particle, m2 C0 = concentration of the solute in external liquid phase, mol/m3 Ci = concentration of the solute in internal liquid phase, mol/m3 De = effective diffusivity, m2/s K = adsorption equilibrium constant, dm3/mol ksl = external solid liquid mass transfer coefficient, cm/s q = adsorbed solute concentration in the polymer, mol/kg qmax = maximum adsorption capacity, mol/kg r = radius, m Rp = particle radius, m 7460

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461

Industrial & Engineering Chemistry Research t = time, s T = temperature, K V = volume, m3 εi = internal voidage of particle

’ REFERENCES (1) Ramadasn, K.; Bhanumathy, P.; Nirmala, K.; George, M. C. Potential anticancer activity of turmeric (Curcuma longa). Cancer Lett. (Shannon, Irel.) 1985, 29, 197–202. (2) Majeed, M.; Badmaev, V.; Rajendran, R. Bioprotectant Composition Method of Use and Extraction Process of Curcuminoids. World Patent 003674, 1997. (3) Sui, Z.; Salto, R.; Li, J.; Craik, C.; Ortiz de Montellano, P. R. Inhibition of the HIV-1 and HIV-2 proteases by curcumin and curcumin boron complexes. Bioorg. Med. Chem. 1993, 1 (6), 415–422. (4) Verghese, J.; Joy, M. T. Isolation of coloring matter from dried Turmeric (Curcuma longa L) with Ethyl Acetate. Flavour Fragrance J. 1989, 4, 31–32. (5) Sastry, B. S. Curcumin content of turmeric. Res. Ind. 1970, 15, 258–260. (6) Marsin, S. M.; Ahmed, U. K.; Smith, R. M. Application of supercritical fluid extraction and chromatography to the analysis of turmeric. J. Chromatogr. Sci. 1993, 31, 20–25. (7) Dandekar, D. V.; Gaikar, V. G. Hydrotropic extraction of curcuminoids from turmeric. Sep. Sci. Technol. 2003, 38, 1185–1245. (8) Dandekar, D. V.; Gaikar, V. G. Microwave assisted extraction of curcuminoids from Curcuma longa. Sep. Sci. Technol. 2002, 37, 2669–2690. (9) Ohshiro, M.; Kuroyanag, A. Structures of sesquiterpenes from Curcuma longa. Phytochemistry 1990, 29, 2201–2205. (10) Golding, B. T.; Pombo, E. C. Turmerones isolation from turmeric and their structure determination. J. Chem. Soc., Chem. Commun. 1982, 363–364. (11) Coorey, R. V.; Hakansson, P. Application of mass spectrometry to characterize the components present in curcumin sample. Sri Lanka J. Phys. 2003, 4, 11–20. (12) Kiran, S.; Kapoor, V. P. Optimization of extraction and dyeing conditions for traditional turmeric dye. Indian J. Tradit. Knowl. 2007, 6, 270–278. (13) Taylor, S. J.; McDowell, I. J. Determination of curcuminoids pigments in turmeric (curcuma domestica Val) by reversed phase highperformance chromatography. Chromatographia 1992, 34, 73–77. (14) He, X. G.; Lin, L. Z.; Lain, L. Z. Analysis of flavonoids from red clover by liquid chromatography-electro spray mass spectrometer. J. Chromatogr., A 1996, 755, 127–132. (15) Koichi, I.; Chihiro, N. Purification of curcumin, demethoxy curcumin, bis-demethoxycurcumin by high-speed countercurrent chromatography. J. Agric. Food Chem. 2008, 56, 9328–9336. (16) Patel, K.; Krishna, G.; Sokoloski, E.; Ito, Y. Preparative separation of curcuminoids from crude curcumin and turmeric powder by pHzone-refining countercurrent chromatography. J. Liq. Chromatogr. Relat. Technol. 2000, 23 (14), 2209–2218. (17) Kim, D. Method to prepare pure curcumin. World Patent 143635, 2007. (18) Chearwae, W.; Wu, C. P.; Chu, H. Y. Curcuminoids Purified from Turmeric Powder Modulate the Function of Human Multidrug Resistance Protein 1 (ABCC1). Cancer Chemother. Pharmocol. 2006, 57, 376–388. (19) Gaikar, V. G.; Srivastava, S.; Leena, D.; Shettar, R. Process of Recovery of Pure Curcumins from Turmeric Rhizomes. Indian Patent 205792, 2007. (20) Madsen, B.; Hidalgo, G. Purification Process for Improving Total Yield of Curcuminoid Colouring Agent. Eur. Patent EP1313808, 2007. (21) Cao, Q. H.; Qu, W. J.; Deng, Y. X. Study on purification of total flavonoids from seed residue of Hippophae rhamnoides with macroporous resin. Chin. J. Chem. Eng. 2004, 29, 225–228.

ARTICLE

(22) Silva, E. M.; Pompeu, D. R.; Larondelle, Y.; Rogez, H. Optimization of the adsorption of polyphenols from Inga edulis leaves on macro porous resins using an experimental design methodology. Sep. Purif. Technol. 2007, 53, 274–280. (23) Jinhuan, Y.; Liang, S.; Guanjun, T.; Xingguo, W.; Aiyong, Q. Process research of macro porous resin chromatography for separation of N-(p-coumaroyl) serotonin and N-feruloylserotonin from Chinese safflower seed extract Qingzhe Jin. Sep. Purif. Technol. 2008, 62, 371–376. (24) Wan, J. B.; Zhang, Q. W.; Ye, W. C.; Wang, Y. T. Quantification and separation of protopanaxatriol and protopanaxadiol type saponins from Panax notoginseng with macro porous resins. Sep. Purif. Technol. 2008, 60, 198–205. (25) Liu, X. M.; Xiao, G. S.; Chen, W. D. Quantification, Purification of Mulberry Anthocyanins with Macro porous resins. J. Biomed. Biotechnol. 2004, 5, 326–331. (26) Ping, L.; Yanhui, W.; Runyu, M.; Xiaolin, Z. Separation of tea polyphenols from green tea leaves by a combined CATUFM-adsorption resin process. J. Food. Technol. 2005, 67, 253–260. (27) Sun, H. COMPASS: Ab initio forcefield optimized for condensed-phase applications—Overview with details on alkane and benzene Compounds. J. Phys .Chem. B 1998, 102, 7338–7364. (28) Sun, H.; Ren, P.; Fried, J. R. The COMPASS force field: Parameterization and validation for polyphosphazenes. Comput. Theor. Polym. Sci. 1998, 8, 229–246. (29) Rappe, A. K.; Goddard, W. A. Charge equilibration for molecular dynamics simulations. J. Phys. Chem. 1991, 95, 33–44. (30) Kabra, V. S.; Gaikar, V. G. Molecular simulation of sodium butyl benzene sulphonate at air-water interface and in aqueous solution. J. Mol. Liq. 2008, 142, 143–149. (31) Khan, A. R.; Alhaddad, A. Equilibrium studies of some aromatic pollutant from dilute aqueous solution on activated carbon at different temperature. J. Colloid Interface Sci. 1997, 154, 154–165. (32) Anasthas, H. M.; Gaikar, V. G. Removal of acetic acid impurities from ethyl acetate by adsorption on ion exchange resins. Sep. Sci. Technol. 2001, 36, 2623–2646. (33) Fredenslund, A.; Jones, R. L.; Prausnitz, J. M. Group-contribution estimation of activity coefficients of nonideal liquid mixtures. AIChE J. 1975, 21, 1086–1099. (34) Walas, M. S. Phase Equilibria in Chemical Engineering; Butterworth Publishers: Woburn, MA, USA, 1985. (35) Anasthas, H. M.; Gaikar, V. G. Adsorptive separation of alkyl phenols using ion exchange resins. React. Funct. Polym. 1999, 39, 227–237. (36) Akre, K.; Gaikar, V. G. Adsorptive separation of o-/p-aminoacetophenones using polymeric adsorbent and ion exchange resins. Sep. Sci. Technol. 2003, 41, 1593–1627. (37) Anasthas, H. M.; Gaikar, V. G. Adsorption of acetic acid on ionexchange resins in non-aqueous conditions. React. Funct. Polym . 2001, 47, 23–35. (38) Gaikar, V. G.; Debashish, M. Adsorptive recovery of napthenic acids using ion-exchange resins. React. Funct. Polym. 1996, 31, 155–164. (39) Garcia-Delgado, R. A.; Cotouelo, L. M. Equilibrium study of single-solution adsorption of anionic surfactants with polymeric XAD resins. Sep. Sci. Technol. 1992, 27, 975–987. (40) Kaili, L.; Jiayong, P.; Yiwei, C.; Rongming, C. Study the adsorption of phenol from aqueous solution on hydroxyl apatite nanopowders. J. Hazard. Mater. 2009, 161, 231–240. (41) Pon, S. Bard, A. J. Electro Analytical Chemistry; Marcel Dekker, Inc.: New York, NY, 1984; pp 571576. (42) Wilke, C. R.; Chang, P. Correlation of diffusion coefficients in dilute solutions. AIChE J. 1955, 1, 264–270. (43) Glueckauff, E. Part 10: Formulation for diffusion into spheres and their application to chromatography. Trans. Faraday Soc. 1982, 21, 446–457. (44) Levins, D. M.; Glastonbury, J. R. Particle-liquid hydrodynamics and mass transfer in a stirred vessel. Part 2—Mass transfer. Trans. Inst. Chem. Eng. 1972, 52, 132–139. 7461

dx.doi.org/10.1021/ie100998p |Ind. Eng. Chem. Res. 2011, 50, 7452–7461