Adsorption of soy isoflavones by activated carbon: Kinetics, thermodynamics and influence of soy oligosaccharides

Chemical Engineering Journal 215–216 (2013) 113–121

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Adsorption of soy isoflavones by activated carbon: Kinetics, thermodynamics and influence of soy oligosaccharides Yun Shi, Xiangzhen Kong, Caimeng Zhang, Yeming Chen, Yufei Hua ⇑ State Key Laboratory of Food Science and Technology, Jiangnan University, Wuxi, 1800 Lihu Avenue, Jiangsu Province, PR China School of Food Science and Technology, Jiangnan University, Wuxi, 1800 Lihu Avenue, Jiangsu Province, PR China

h i g h l i g h t s " Systematic studied the adsorption of soy isoflavones by activated carbon. " The adsorption characters vary with the chemical forms of soy isoflavones. " The adsorption of soy isoflavones by activated carbon is endothermic. " Soy oligosaccharides have negative influence to the diffusion of isoflavones.

a r t i c l e

i n f o

Article history: Received 12 September 2012 Received in revised form 10 October 2012 Accepted 11 October 2012 Available online 10 November 2012 Keywords: Adsorption Activated carbon Isoflavones Soy molasses

a b s t r a c t Adsorption behavior of daidzin, genistin, 600 -O-malonyldaidzin and 600 -O-malonylgenistin, the four major soy isoflavones presented in soy molasses centrifugation supernatant on activated carbon was studied in this paper so as to provide theoretical basis for the purification of soy oligosaccharides from soy molasses. Kinetic experiments showed that the adsorption processes obeyed pseudo-second-order kinetics and equilibrium was nearly achieved in 90 min. Weber–Morris model fitting showed that adsorption process consisted of 3 stages: boundary layer diffusion and two intra-particle diffusions. Experimental adsorption data for every isoflavone components could be described separately by the Langmuir isotherm model and the calculated maximum adsorptions were in the order of genistin > daidzin > 600 -O-malonylgenistin > 600 O-malonyldaidzin, indicating that the adsorption driving forces were due to dispersion interactions between the aromatic ring of isoflavone and the aromatic structure of the activated carbon. Adsorption behaviors of isoflavones on activated carbon in sugar free solutions were compared. It was found that, by removing sugar from the system, diffusion rate constants and the sum of the maximum adsorption capacity increased. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction Soy molasses is a by-product generated in the production of soy protein concentrate, in which soy oligosaccharides, isoflavones, saponins, and other phytochemicals are enriched [1,2]. The byproduct has been a popular fermentation medium for bio-ethanol [3] as well as lactic acid [4] production and a good resource of soybean phytochemicals. Several researchers reported isolating isoflavones and other phytochemicals from the insoluble precipitates of soy molasses suspension [5–7] while there are few reports concerned with the centrifugation supernatant of soy molasses. The supernatant contains most of the water soluble com⇑ Corresponding author at: State Key Laboratory of Food Science and Technology, School of Food Science and Technology, Jiangnan University, Wuxi 214122, 1800 Lihu Avenue, Jiangsu Province, PR China. Tel./fax: +86 510 85917812. E-mail address: [email protected] (Y. Hua). 1385-8947/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2012.10.100

ponents existed in soy molasses, predominantly sucrose, raffinose and stachyose. Raffinose and stachyose have been proved to be functional oligosaccharides, as they can stimulate the growth of bifidobacteria and other kinds of lactic acid bacteria, promote the competitive exclusion of potential pathogens [8] and reduce the levels of some colonic enzymes (b-glucuronidase, nitroreductase, azoreductase and glycoholic acid hydrolase) which are implicated in the conversion of procarcinogens to carcinogens [9]. Currently, commercial soy oligosaccharides are isolated from soybean whey which is generated from soy protein isolate production [10]. Extraction of soy oligosaccharides from soy molasses supernatant could be more beneficial economically because of its higher oligosaccharide concentration, thus less energy is needed in water evaporation. As Galanakis [11] pointed out, recovery of high-added value compounds usually follows 4–5 stages of macroscopic pretreatment, macro- and micro-molecules separation, extraction, isolation and purification, as well as finally product

114

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

Nomenclature P Po pH(PZC) Qe Co Ce V A Dq S2 Qem N F-test SBET Vp t

pressure of N2 (MPa) saturated vapor pressure of N2 (MPa) point of zero charge adsorbed isoflavones on activated carbon at equilibrium (mg/g) initial concentration of isoflavone in solution (mg/L) equilibrium concentration of isoflavone in solution (mg/ L) volume of solution (L) dry mass of activated carbon (g) standard deviation model variance adsorbed isoflavones on activate carbon calculated from model number of experimental data Fisher statistical test BET surface area of activated carbon (m2/g) total pore volume of activated carbon (cm3/g) contact time (h)

formation. In the purification process of oligosaccharides extracted from soy molasses, isoflavones are one of the major components which should be removed from the supernatant so as to prevent the development of bitter taste and dark color to the product. In nature, soy isoflavones exist predominantly as malonylglucosides forms, but acetylglucosides, glucosides and aglycons forms generated from the malonylglucosides during processing of soybeans or sample preparation and analysis [12]. Thus, four different chemical forms of soy isoflavones could be found in processed soy products, i.e. the aglycons (daidzein, glycitein and genistein), the glucosides (daidzin, glycitin and genistin), the acetylglucosides (600 -O-acetyldaidzin, 600 -O-acetylglycitin and 600 -O-acetylgenistin), as well as the malonylglucosides (600 -O-malonyldaidzin, 600 -Omalonylglycitin, and 600 -O-malonylgenistin) [13,14]. Adsorptions using activated carbon, resins and other porous substances are commonly practiced in product decoloration and purification. Several studies reported the adsorption behavior of dyes [15,16], phenolic compounds [17,18], organic acid [19,20] and metal ions [21,22] on activated carbon. On the other hand, Gugger et al. [23] reported the adsorption of soy isoflavone in ultrafiltration permeate fraction of soy molasses or soybean whey using resins (divinylbenzene, ethylvinylbenzene copolymer resin). At present, there are no systematic studies about the adsorption of soy isoflavones on activated carbon. This paper reports the adsorption behavior of four predominant isoflavones, daidzin, genistin, 600 -O-malonyldaidzin and 600 -O-malonylgenistin in soy molasses centrifugal supernatant on the activated carbon. The kinetic and thermodynamic characteristic adsorption constants were determined using accepted theoretic models. Effects of the presence of soy oligosaccharides on the adsorption behavior were also studied.

2. Materials and methods 2.1. Chemicals and regents Activated carbon (AC) powder (purchased from Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) was dried at 105 °C overnight for pre-activation. Oligosaccharides standards (sucrose, raffinose and stachyose) were purchased from Sigma Chemical Co. Ltd. Commercial soy oligosaccharides syrup was purchased

Qt

j kid I kd Ci KF n Qm KL R2 pKa DG DH DS Ko R T

concentration of adsorbed isoflavone at contact time t (mg/g) pseudo-second order kinetic parameter (g/mg/h) intra-particle diffusion rate constant (mg/g/h1/2) boundary layer constant (mg/g) diffusion rate constant (mg/g/h1/2) initial concentration of isoflavone in solution at the start of the linear portion (mg/L) Freundlich isotherm coefficient Freundlich adsorption coefficient Langmuir maximum adsorption capacity (mg/g) Langmuir adsorption coefficient (L/mg) correlation coefficient dissociation constant Gibs free energy (kJ/mol) enthalpy change (kJ/mol) entropy change (kJ/mol/K) adsorption equilibrium constant universal gas constant (J/mol/K) adsorption temperature (K)

from Shansong Biological Product Co. Ltd. (Linyi, China). Soy isoflavone standards (daidzin, glycitin, genistin, 600 -O-malonyldaidzin and 600 -O-malonylgenistin) were purchased from Wako Pure Chemical Industries, Ltd. 2.2. Analysis and measurement methods The high-performance liquid chromatography (HPLC) apparatus (Hitachi, Tokyo, Japan) consisted of a HITACHI L-2490 RI detector, a HITACHI L-2400 UV detector, a HITACHI L-2300 column oven, and a HITACHI L-2130 pump. Reversed-phase HPLC analysis of isoflavones was carried out on a 250  4.6 mm, 5 lm YMC-Pack Pro C18 column (YMC Co., Ltd., Kyoto, Japan). A linear HPLC gradient was composed of (A) 0.05% trifluoroacetic acid in water and (B) 0.05% trifluoroacetic acid in ACN. Following injection of 20 lL of sample, solvent B was increased from 15% to 20% over 15 min with the flow rate 0.8 mL/ min. In the second 15 min, solvent B was increased from 20% to 30%. In the third 15 min, solvent B was increased from 20% to 35%. In the following 20 min, solvent B was decreased from 35% to 15%, and then held at 15% for 5 min. The flow rate was kept at 1.0 mL/min between 15 min and 70 min. Eluted isoflavones were detected by their absorbance at 254 nm. The experiment was carried out at 34 °C. Analysis of carbohydrates was conducted on a 250  4.6 mm, 5 lm Alltima Amino column (Grace, Waukegan, USA). The mobile phase was solvent of 68% ACN in water, and the flow rate was 1.0 mL/min. Eluted carbohydrates were detected by RI detector and the analysis was performed at 40 °C. The textural characterization of the activated carbon was carried out by N2 adsorption at 77 K using an ASAP 2020 Micromeritics instrument. The surface area was calculated using BET method and the total pore volume was calculated from the amount of N2 adsorbed at P/Po = 0.95 [24]. The pH(PZC) (point of zero charge) of the activated carbon was measured using the method suggested by Franz et al. [25]. 2.3. Preparation of soy molasses centrifugation supernatant Soy molasses (kindly provided by Wonderful Industrial Group Co., Ltd., Shandong, China) was diluted with distilled water and

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

adjusted to acidic pH with HCl. The suspension was left at room temperature for a few minutes before centrifuged at 3200g for 20 min. Finally hazel and clear liquid supernatant was obtained as experimental material, which was called sample solution in the following experiments.

115

where Co and Ce are the initial and equilibrium concentrations (mg/ L), respectively, V is the volume of solution (L), and A is the dry mass of activated carbon (g). Each experimental point was an average of two independent adsorption experiments. 2.6. Statistical analysis

2.4. Preparation of oligosaccharide-free soy isoflavone mixture Sample solution was concentrated to over 50 BX at 70 °C under reduced pressure, and then absolute ethanol was added into the concentrated solution with the volume ratio 1:5. After centrifuging and filtering the mixture to remove most of the sugar components, the isoflavone enriched ethanol solution was evaporated to nearly dryness under reduced pressure. The residue was re-extracted with methanol followed by vacuum evaporation. Dissolve the residue with distilled water to obtain soy isoflavone solution which was substantially free from oligosaccharides.

Error analysis was carried out using the normalized standard deviation (Dq, %) [15]. If the data from the model are similar to the experimental data, the values of Dq will be lower.

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u  N  u 1 X Q e;i  Q em;i 2 Dqð%Þ ¼ 100  t N  1 i¼1 Q em;i

ð2Þ

The model variance (S2) was calculated using following equation [26]:

XN 2

i¼1

ðQ e;i  Q em;i Þ2

2.5. Adsorption studies

S ¼

In this study, all activated carbon adsorption experiments were carried out at pH 4.0, by contacting the desired amount of activated carbon with 25.00 mL of isoflavone solutions in hermetically closed 150 mL Erlenmeyer flasks and shaked at 150 rpm in a thermostatic rotary shaker. After the selected contact time, the suspension was filtered through filter paper and cellulose nitrate membranes (0.22 lm) sequentially to obtain samples for HPLC analysis. The adsorbed isoflavones on activated carbon at equilibrium (denoted as Qe, mg/g) was defined as

where Qem and Qe are the adsorbed isoflavones on activated carbon calculated from the model and the experimental data (mg/g) respectively, while N is the number of experimental data. Evaluation of statistical significant difference between models was carried out using the F-test. The calculated F value was defined as follow:

Q e ¼ ðC o  C e Þ 

V A

ð1Þ

F cal ¼

N1

S2H S2L

ð3Þ

ð4Þ

where S2H is the higher model variance and S2L is the lower model variance.

Fig. 1. HPLC chromatogram of sample solution recorded by RI (a) and at 254 nm (b). For (a), peaks 1–5 are fructose, glucose, sucrose, raffinose and stachyose, respectively. For (b), peaks 1–5 are daidzin, glycitin, genistin, 600 -O-Malonyldaidzin and 600 -O-Malonylgenistin, respectively.

116

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

If the calculated F value Fcal is higher than the tabulated F value Ftab with the chosen level of probability (95%), there is significant difference between the two models.

3. Results and discussion 3.1. Proximate analysis of sample solution The HPLC chromatograms confirmed that the predominant oligosaccharides in sample solution were sucrose, raffinose and stachyose (Fig. 1a), with the concentrations of 31.150 ± 1.076, 5.938 ± 0.106, and 25.378 ± 0.891 mg/mL, respectively. The content of oligosaccharides may vary as processing technology and varieties of soybeans. As reported by Siqueira et al. [3], the percent of sucrose, raffinose and stachyose in soybean molasses were 28.4, 9.68 and 18.6 in dry basis, respectively. While for soybeans grown in China, the content of raffinose and stachyose were about 0.50 and 3.30 g/100 g soybean, respectively [27]. At the same time, five isoflavones, daidzin, glycitin, genistin, 600 -O-malonyldaidzin and 600 -O-malonylgenistin with typical concentrations of 73.211 ± 1.307, 11.260 ± 0.252, 92.863 ± 0.776, 107.410 ± 2.291 and 138.430 ± 2.408 mg/L, respectively, were found (Fig. 1b). The results agreed quantitatively with the typical isoflavone distribution in soy molasses reported by Chajuss [28] who pointed out that the weight percent of daidzin, glycitin, genistin, 600 -O-malonyldaidzin and 600 O-malonylgenistin in soy molasses were 0.23, 0.06, 0.36, 0.30 and 0.45, respectively. Glycitin was excluded in the following adsorption experiments because its concentration was too low to be determined with the same accuracy as other four isoflavones.

Fig. 2. Time dependence of isoflavones adsorbed on activated carbon (Qt) from sample solution at 303.16 K.

3.2. Characterization of the activated carbon The activated carbon used in this study was a classic commercial activated carbon with a large surface area (SBET) of 1390 m2/ g, and the SBET of other commercial activated carbons were reported to be about 1000–2000 m2/g [18,19,25]. Besides, the total pore volume (Vp) of activated carbon was 0.60 cm3/g, which was similar with other commercial activated carbons mentioned above. The pH(PZC) (point of zero charge) of the activated carbon was measured to be 7.30, which is a reasonable value for unmodified activated carbons [24,29].

3.4. Adsorption mechanism

3.3. Adsorption kinetics and contact time Fig. 2 shows the time course of the adsorption of isoflavones on activated carbon. As is shown in Fig. 3, the data were found to fit to pseudo-second order model with high regression coefficient (R2 > 0.999).

t 1 1 ¼ þ t Q t j  Q 2e Q e

Fig. 3. Plots of the pseudo-second order linearized kinetic model for the adsorption of isoflavone by activated carbon from sample solution at 303.16 K.

In order to know whether the diffusion was the controlling process for isoflavone adsorption on AC, the adsorption data of Fig. 2 was evaluated using Weber–Morris model:

Q t ¼ kid  t 1=2 þ I

ð6Þ 1/2

ð5Þ

In the above pseudo-second order model, Qt and Qe are the concentrations of adsorbed isoflavone (mg/g) at contact time t (h) and equilibrium, respectively, and j is the pseudo-second order kinetic parameter (g/mg/h) [30]. The concentrations of adsorbed isoflavone at equilibrium (Qe) were calculated to be 11.197, 12.561, 14.898 and 20.967 mg/g for daidzin, genistin, 600 -O-malonyldaidzin, and 600 -O-malonylgenistin, respectively. To determine a workable contact time for the experiment, Qt/Qe at different time was calculated (data not shown). It was found that for different isoflavones, values of about 98% could be reached after 90 min of adsorption. In the following, Qe and Ce were actually Qt and Ct at 90 min respectively.

where kid is the intra-particle diffusion rate constant (mg/g/h ) and I (mg/g) is a constant that gives idea about the thickness of the boundary layer, i.e., the larger the value of I, the greater the boundary layer effect is [31]. The plots (Fig. 4) shows that the adsorption processes consists of two linear sections with different slopes, indicating that two intra-particle diffusion steps occurred in the adsorption process. The first straight portion (about from 0.316 h1/2 to 0.707 h1/2) could be attributed to the diffusion in the macro-pore of AC, and the second portion (after 0.707 h1/2) described the micro-pore diffusion [32], characteristic of the final equilibrium step. However, the adsorption process was not controlled solely by the intra-particle diffusion since the extrapolation of the first linear portion does not pass through the origin, suggesting the involvement of external mass transfer, or film diffusion in the adsorption process. For a well

117

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

Langmuir isotherm models (Fig. 5b). In other words, equilibrium isothermal models were applied to each solute neglecting the possible interference from other solute(s). The Freundlich isotherm assumes that an infinite adsorption can occur, and describes the equilibrium conditions on heterogeneous sites with different adsorption energies [35]. The Freundlich isotherm is represented by the following equation:

Q e ¼ K F  C 1=n e

ð7Þ

KF is related to the degree of adsorption (depending on temperature, surface area of the adsorbent, and the relative attraction of the solutes in a mixture for the solid surface), and n provides the rough estimation of the intensity of the adsorption (depending on the temperature). The Langmuir isotherm with the mathematical form as follow is used to describe a complete monolayer adsorption, with the assumption of homogeneous distribution of active sites on the activated carbon surface. Fig. 4. Weber–Morris intra-particle diffusion plots for the adsorption of isoflavones by activated carbon from sample solution at 303.16 K.

agitated system, mass transfer from the bulk solution to the liquid film surrounding the adsorbent could be neglected. However, the present study showed that the diffusion through the liquid film surrounding the particles has an important effect on the initial sorption kinetics of isoflavones. Namasivayam and Yamuna [33] also suggested that film or intra-particle diffusion would be the rate limiting step while mass transfer from bulk solution to the out surface of liquid film could be neglected in a vigorously agitated adsorption system as Mohan and Singh [34] pointed out. Two intra-particle diffusion rate constants were obtained from slopes of two linear portions while the diffusion rate constant in the boundary layer was approximated as the slope of the line starting from origin and ending at Qt (0.224 h1/2). The obtained constants (kd) divide by the concentration corresponding to those at the start of each linear portion (Ci) so as to standardize the data (Table 1). As is expected, all isoflavones had highest diffusion constants in the boundary layer but lowest in micropores. Among isoflavones, genistin had remarkably higher diffusion constants than others in all stages but daidzin gave lowest value in the final (micropore diffusion) stage. Besides, the lowest values of micropore diffusion rate constants revealed that its the rate limiting process. 3.5. Adsorption isotherms Equilibrium adsorption data (Qt and Ct values at 90 min) for daidzin, genistin, 600 -O-malonyldaidzin, and 600 -O-malonylgenistin were analyzed separately by classical Freundlich (Fig. 5a) and

Ce 1 Ce ¼ þ Q e Q m  KL Q m

ð8Þ

where Qm is the maximum adsorption capacity (mg/g) corresponding to complete monolayer coverage, and KL (L/mg) is the adsorption coefficient which is the ratio of the rate constants of adsorption and desorption and related to the energy of adsorption [36]. Table 2 shows that for all isoflavones analyzed here, the R2 values of Langmuir model were similar to those of Freundlich model. For further confirmation, error analysis were carried out using standard deviation (Dq) to compare the differences between experimental data and predicted values by models. As seen in Table 2, the values of Dq of Langmuir model were lower than that of Freundlich model which suggested that Langmuir model was the better one for describing the experimental data. However, by comparing the values of Fcal and Ftab, it is obvious that there’s no statistical significant difference between the Langmuir and Freundlich model for describing the adsorption of isoflavones on AC except for 600 -O-malonyldaidzin. There’s report showed that Langmuir model was more applicable to the adsorption of multi-solute systems than Freundlich model [37]. Thus, the number of adsorption sites on activated carbon is limited and that the isoflavone mixtures form a monomolecular layer on the adsorbent at saturation. The result also suggested that isoflavones may not be adsorbed onto mutually common adsorption sites but that each isoflavone may have selective adsorption sites. From Table 3 it can be observed that the maximum adsorption Qm values was in the order of genistin > daidzin > 600 -O-malonylgenistin > 600 -O-malonyldaidzin. Interestingly enough, for two pairs of isoflavones derived from daidzein and genistein aglycon respec-

Table 1 The diffusion constants for the adsorption of isoflavones by activated carbon from sample solution at 303.16 K. Diffusion section

Isoflavone

kd (mg/g/h1/2)

Ci (mg/L)

kbd/Ci

Dq (%)

Boundary layer diffusion 0–0.05 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

37.222 43.063 45.654 65.274

61.962 64.821 84.166 119.490

0.6007 0.6643 0.5424 0.5652

– – – –

Macropore diffusion 0.05–0.50 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

3.804 3.361 6.224 6.858

16.757 11.952 29.186 33.974

0.2270 0.2812 0.2133 0.2019

0.131 0.111 0.262 0.388

Micropore diffusion 0.50–1.50 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

0.871 1.212 2.840 3.892

9.260 5.443 17.155 20.707

0.0940 0.2227 0.1655 0.1880

0.154 0.189 0.226 0.340

118

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

structure. Comparing to daidzein, genistein aglycon has an extra hydroxyl group at position 5 of ring A, which forms a very strong intramolecular hydrogen bond with the carbonyl group at position 4 of ring C. The hydrogen bond decreases the partial negative charge on the carbonyl oxygen atom and modified the distribution of electron on ring A and ring C [39]. Since the adsorption of adsorbates that contain aromatic ring (e.g. phenols and isoflavones) on activated carbon is governed by ‘‘p–p dispersion interaction’’, the electron dispersion over ring A and ring C of isoflavones may determine the category of sites which they can be adsorbed to [17,18,40]. The fact that malonyl-glucoside isoflavones occupied a fixed fraction of sites in their respective category suggested that malonyl-group, in spite of linking to different molecules, modified the adsorption behavior in the same way. Table 3 also shows that 600 -O-malonylgenistin had substantially higher KL than other isoflavones, indicating that 600 -O-malonylgenistin interacted with activated carbon much stronger. A parallel can be drawn between KL and the retention time on C18 column. Lee and Row [41] found that the retention time factor of soy isoflavones on RP-HPLC was determined by physicochemical properties such as water solubility, hydrophobicity, hydrophilic-lipophlic balance, hydrophilic surface area, total energy, and connectivity index. 3.6. Adsorption thermodynamics The free-energy (DG), the enthalpy change (DH), and the entropy change (DS) were calculated from the temperature-dependent adsorption isotherms to predict the adsorption process.

DG ¼ RT ln K o

ð9Þ

where Ko is the adsorption equilibrium constant, R is the universal gas constant (8.314 J/mol/K) and T is the temperature (K) [42]. Since the adsorption of isoflavones by activated carbon fitted to the Langmuir model which represented micropore filling, free energy change (DG) of the adsorption can be calculated by following equation [43]: Fig. 5. Linearized plots of Freundlich (a) and Langmuir (b) equations for the adsorption of isoflavones by activated carbon from sample solution at 303.16 K.

DG ¼ RT ln K o ¼ RT ln

  Q m  KL Vp

ð10Þ

where Qm (mg/g) and KL (L/mg) are the Langmuir parameters, and Vp (cm3/g) is the total pore volume of AC. Enthalpy change (DH) and entropy change (DS) can be obtained by van’t Hoff plot:

tively, two ratios, Qm (600 -O-malonyldaidzin)/Qm (daidzin) and Qm (600 -O-malonylgenistin)/Qm (genistin) give very similar value, about 0.47. It is thus reasonable to assume that there are two categories of adsorption sites on AC, corresponding to genistein and daidzein aglycon containing isoflavone, respectively. As Rickert et al. [38] reported, the pKa of daidzin, genistin and malonyl-glucoside isoflavones are about 9.55, 9.20 and 5.70, respectively. At the adsorption pH of 4.0, daidzin and genistin were essentially unionized while malonyl-glucoside isoflavones were slightly negatively charged. In both cases, electrostatic interaction between solutes and adsorbents did not play an important role. The different adsorption behavior between genistein and daidzein aglycon containing isoflavone could be resulted from their differences in

ln K o ¼ ln

  Q m  KL DS DH ¼  R RT Vp

ð11Þ

Table 3 summarized the Langmuir isothermal parameters and the adsorption thermodynamic analysis results for isoflavones. The negative values of DG and positive values of DH for all the four isoflavones indicated that the adsorption process was spontaneous and endothermic in nature. The pure adsorption process was expected to be exothermic because bond formation between adsorbate and adsorbent is accompanied by the release of heat.

Table 2 Isotherm parameters obtained by fitting equilibrium data with Langmuir and Freundlich models. Isoflavone

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin a b

Langmuir

Freundlich

KLa (L/mg)

Qma (mg/g)

R2a

Dqb (%)

S2b

KF a

na

R2a

Dqb (%)

S2b

Fcal

Ftab

0.0681 0.0888 0.0936 0.3211

55.866 91.743 25.707 43.860

0.9979 0.9999 0.9996 0.9986

0.480 0.684 0.623 0.742

0.66 3.76 0.34 1.92

4.078 4.671 7.299 15.314

16.549 32.865 13.622 32.806

0.9553 0.9951 0.9657 0.9797

0.676 0.760 6.953 0.761

1.23 5.01 36.17 2.08

1.88 1.33 106.86 1.08

2.40 2.40 2.40 2.40

Parameters obtained at 303.16 K. Statistical analysis for the data of four experimental temperatures (303.16, 313.16, 323.16 and 333.16 K).

119

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121 Table 3 Langmuir and Thermodynamic parameters for the adsorption of isoflavones by activated carbon from sample solution. Isoflavones

T (K)

Daidzin

303.16 313.16 323.16 333.16

Genistin

Qm (mg/g)

KL (L/mg)

DG (kJ/mol)

DH (kJ/mol)

DS (kJ/mol/K)

55.866 56.818 59.172 63.694

0.0681 0.0718 0.0719 0.0721

21.340 22.241 23.823 24.769

3.705

0.085

303.16 313.16 323.16 333.16

91.743 92.593 97.087 102.041

0.0888 0.0890 0.0892 0.0901

23.193 24.013 25.731 26.693

2.563

0.088

600 -O-molonyldaidzin

303.16 313.16 323.16 333.16

25.707 26.596 29.155 33.333

0.0936 0.1156 0.1425 0.1450

20.220 21.530 23.758 24.913

15.051

0.120

600 -O-molonylgenistin

303.16 313.16 323.16 333.16

43.860 47.619 55.249 57.803

0.3211 0.3684 0.7016 0.8737

24.528 25.919 29.758 31.412

29.437

0.184

Table 4 The diffusion constants for the adsorption of isoflavones by activated carbon from sugar-free isoflavone solution at 303.16 K. Diffusion section

Isoflavone

kd (mg/g/h1/2)

Ci (mg/L)

kbd/Ci

Dq (%)

Boundary layer diffusion 0–0.05 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

110.770 90.974 47.895 30.971

72.780 50.532 45.237 31.961

1.5220 1.8003 1.0588 0.9690

– – – –

Macropore diffusion 0.05–0.20 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

14.669 7.432 8.063 6.655

23.244 9.847 23.818 18.110

0.6311 0.7547 0.3385 0.3674

0.280 0.300 0.775 0.771

Micropore diffusion 0.25–1.00 h

Daidzin Genistin 600 -O-malonyldaidzin 600 -O-malonylgenistin

6.164 2.885 7.214 3.568

15.733 6.112 19.558 14.627

0.3918 0.4720 0.3688 0.2439

0.623 0.425 0.980 0.716

However, a diffusion controlled adsorption process is endothermic since diffusion rate is favored at higher temperature. It is further backed up by the above kinetics studies in which an intra-particle diffusion adsorption mechanism was proposed [44]. In fact, while some systems did show exothermic adsorption of activated carbon [45,46], more adsorption studies reported endothermic behaviors, such as Sureshkumar and Namasivayam [47], Ahmad et al. [48], and Bulut et al. [49]. Similarly, the values of DS for pure adsorption are negative due to the decrease in randomness. The positive DS suggested a net gaining of degree of freedom which was originated from the release of bonded water or oligosaccharides on AC accompanied with the adsorption of isoflavone molecules. Similar results (DS > 0) have also been found for the adsorption of direct red 12 B by biogas residual slurry [33] and the removal of Congo red by activated carbon [50]. 3.7. Influence of oligosaccharides As revealed in Section 3.1., there were about 60 mg/mL soy oligosaccharides and 0.4 mg/mL soy isoflavones in sample solution. It is commonly considered that activated carbons prefer to adsorb those components with lower polarity, such as isoflavones, but not oligosaccharides which carry many hydroxide groups. However, the preference might be offset by the high concentration of oligosaccharides. Hence it was necessary to study the influence of oligosaccharides by comparing the adsorption behavior of isoflavones by activated carbon in the sugar-free solution. For sugar-free isoflavone solution, adsorption kinetics could also be described by Weber-Morris model very well (Table 4).

Diffusion rate constants for the three stages show elevated values in sugar-free solution. By comparing the data to Table 1, it can be observed that diffusion rate constants for malonyl-group conjugated isoflavones increased to a less extent than non-conjugated isoflavones, indicating that diffusion behavior does not influenced solely by the bulk solution properties but also determined by some micro-environmental conditions. Experiments showed that, in sugar-free solution, the isoflavones-AC equilibrium adsorption data fit the Langmuir adsorption isotherms very well (Table 5). The sum of the maximum adsorption capacity for the four isoflavones increased from 217.176 mg/g at 60 g/L of oligosaccharides to 219.587 mg/g at 0 g/L of oligosaccharides, indicating that on removing sugar from the solution, a small portion of sites occupied by sugar molecules were released. Compared to the total Qm, changes of Qm for individual isoflavone would be more impressive. We can find that on shifting from 60 g/L of oligosaccharides solution to 0 g/L of oligosaccharides solution, the maximum adsorption capacity of daidzin and genistin increased about 2.0 mg/g and 3.5 mg/g respectively while that for 600 -O-malonyldaidzin and 600 -O-malonylgenistin decreased about 2.5 mg/g and 0.5 mg/g respectively. Considering the results in a reverse direction, it seems to suggest that, on conducting sorption in sugar solution, only a small portion of those adsorption sites designed for daidzin and genistin was occupied by soy oligosaccharide molecules and some of the sugar occupied sites were transformed to adsorption sites for malonyl-conjugated isoflavones. To the best of our knowledge, no study reported the similar results. A possible explanation was that the hydration of soy oligosaccharides reduced the amount of ‘‘free water’’ in which isoflavones were

120

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121

Table 5 The maximum adsorption capacities for the adsorption of isoflavones by activated carbon from isoflavone solution with different concentrations of oligosaccharides. Isoflavones

Daidzin Genistin 600 -O-molonyldaidzin 600 -O-molonylgenistin

0 g/L oligosaccharides

60 g/L oligosaccharides

Qm (mg/g)

KL (L/mg)

R

Qm (mg/g)

KL (L/mg)

R2

57.803 95.238 23.256 43.290

0.0686 0.0889 0.1924 0.3554

0.9994 0.9999 0.9971 0.9995

55.866 91.743 25.707 43.860

0.0681 0.0888 0.0936 0.3211

0.9979 0.9999 0.9996 0.9986

dissolved and increased [H+] correspondingly. The lower ‘‘effective pH’’ could suppress the dissociation of carboxyl group of malonylconjugated isoflavone but increase the positive charge on the surface of AC, thus may result in increased electrostatic attraction between adsorbates and adsorbent. Further studies are needed to get deeper understanding of the mechanism. 4. Conclusion The kinetic experiments show that the adsorption processes of four major soy isoflavones in soy molasses supernatant, daidzin, genistin, 600 -O-malonyldaidzin and 600 -O-malonylgenistin on activated carbon obeyed pseudo-second-order kinetics and equilibrium adsorptions were nearly achieved in 90 min. Weber-Morris model fitting showed that adsorption process consisted of 3 stages: boundary layer diffusion and two intra-particle diffusions. The concentration calibrated diffusion rate constants for boundary layer diffusion were higher than those for intra-particle diffusions, while those for daidzin and genistin were higher than those for malonyl-conjugated counterpart respectively. Experimental adsorption data for every isoflavone components in the molasses supernatant were successfully described by the Langmuir equilibrium isotherm model, suggesting that each isoflavone may have selective adsorption sites. The calculated maximum adsorptions were in the order of genistin > daidzin > 600 -O-malonylgenistin > 600 -O-malonyldaidzin. Thus, the adsorption driving forces were due to dispersion interactions between the aromatic ring of isoflavone and the aromatic structure of the activated carbon. To study the influence of soy oligosaccharides, adsorption behaviors of isoflavones on activated carbon in sugar-free solutions were compared. It was found that, by removing sugar from the system, diffusion rate constants and the sum of the maximum adsorption capacity increased. Acknowledgements We gratefully acknowledge the financial support received from Ministry of Science and Technology, RP China (2012BAD34B04-1), Natural Science Foundation of China (21146002) and the Open Project Program of State Key Laboratory of Food Science and Technology, Jiangnan University (SKLF-ZZB-201202). References [1] M. Hosny, J.P.N. Rosazza, Novel isoflavone, cinnamic acid, and triterpenoid glycosides in soybean molasses, J. Nat. Prod. 62 (1999) 853–858. [2] A.M. Nash, A.C. Eldridge, W.J. Wolf, Fractionation and characterization of alcohol extractables associated with soybean proteins: nonprotein components, J. Agri. Food Chem. 15 (1967) 102–108. [3] P.F. Siqueira, S.G. Karp, J.C. Carvalho, Production of bio-ethanol from soybean molasses by Saccharomyces cerevisiae at laboratory, pilot and industrial scales, Bioresour. Technol. 99 (2008) 8156–8163. [4] M. José-Luis, B.M. Chassy, J.D. McCord, Lactobacillus salivarius for conversion of soy molasses into lactic acid, J. Food Sci. 58 (1993) 863–866. [5] D.H. Waggle, B.A. Bryan, Recovery of isoflavones from soy molasses, US Patent 20030129263A1, 2003. [6] T.A. Dobbins, A.H. Konwinski, Soy isoflavone concentrate process and product, US Patent 20010003781A1, 2001. [7] O. Ramot, T.M. eredith, Process for obtaining solid soy isoflavone-containing products, US Patent 20030104084A1, 2003.

2

[8] Y. Lan, B.A. Williams, M.W.A. Verstegen, Soy oligosaccharides in vitro fermentation characteristics and its effect on caecal microorganisms of young broiler chickens, Anim. Feed Sci. Technol. 133 (2007) 286–297. [9] K.D. Arunachalam, Role of Bifidobacteria in nutrition, medicine and technology, Nutr. Res. 19 (1999) 1559–1597. [10] S. Kim, W. Kim, I.K. Hwang, Optimization of the extraction and purification of oligosaccharides from defatted soybean meal, Int. J. Food Sci. Technol. 38 (2003) 337–342. [11] C.M. Galanakis, Recovery of high added-value components from food wastes: conventional, emerging technologies and commercialized applications, Trends Food Sci. Technol. 26 (2012) 68–87. [12] A.P. Griffith, M.W. Collison, Improved methods for the extraction and analysis of isoflavones from soy-containing foods and nutritional supplements by reversed-phase high-performance liquid chromatography and liquid chromatography-mass spectrometry, J. Chromatogr. A 913 (2001) 397–413. [13] Huei-ju Wang, P.A. Murphy, Isoflavone content in commercial soybean foods, J. Agric. Food Chem. 42 (1994) 1666–1673. [14] S. Barnes, M. Kirk, L. Coward, Isoflavones and their conjugates in soy foods: extraction conditioins and analysis by HPLC-Mass Spectrometry, J. Agric. Food Chem. 42 (1994) 2466–2474. [15] N. Nasuha, B.H. Hameed, Adsorption of methylene blue from aqueous solution onto NaOH-modified rejected tea, Chem. Eng. J. 166 (2011) 783–786. [16] E.N. El Qada, S.J. Allen, G.M. Walker, Adsorption of basic dyes from aqueous solution onto activated carbons, Chem. Eng. J. 135 (2008) 174–184. [17] M.L. Soto, A. Moure, Charcoal adsorption of phenolic compounds present in distilled grape pomace, J. Food Eng. 84 (2008) 156–163. [18] D. Richard, M. de Bellefon, L.D. Núñez, Adsorption of complex phenolic compounds on active charcoal: adsorption capacity and isotherms, Chem. Eng. J. 148 (2009) 1–7. [19] Y. El-Sayed, T.J. Bandosz, Adsorption of valeric acid from aqueous solution onto activated carbons: role of surface basic sites, J. Colloid Interface Sci. 273 (2004) 64–72. [20] S.Y. Ou, Y.L. Luo, F. Xue, Seperation and purification of Ferulic acid in alkalinehydrolysate from sugarcane bagasse by activated charcoal adsorption/ anion macroporous resin exchange chromatography, J. Food Eng. 78 (2007) 1298– 1304. [21] A.H. Sulaymon, B.A. Abid, J.A. Al-Najar, Removal of lead copper chromium and cobalt ions onto granular activated carbon in batch and fixed-bed adsrobers, Chem. Eng. J. 155 (2009) 647–653. [22] J. Goel, K. Kadirvelu, C. Rajagopal, V.K. Garg, Removal of lead (II) by adsorption using treated granular activated carbon: batch and column studies, J. Hazard. Mater. 125 (2005) 211–220. [23] E. Gugger, R. Grabiel, Production of isoflavone enriched fractions from soy protein extracts, US Patent 006171638B1, 2001. [24] Y.S. Al-Degs, G.M. Walker, Effect of solution pH, ionic strength, and temperature on adsorption behavior of reactive dyes on activated carbon, Dyes Pigm. 77 (2008) 16–23. [25] M. Franz, H.A. Arafat, N.G. Pinto, Effect of chemical surface heterogeneity on the adsorption mechanism of dissolved aromatics on activated carbon, Carbon 38 (2000) 1807–1819. [26] A. Esposito, F. Pagnanelli, F. Vegliò, PH-related equilibria models for biosorption in single metal systems, Chem. Eng. Sci. 57 (2002) 307–313. [27] Z. Jiao, X.X. Si, Z.M. Zhang, G.K. Li, Z.W. Cai, Compositional study of different soybean (Glycine max L.) varieties by 1H NMR spectroscopy, chromatographic and spectrometric techniques, Food Chem. 135 (2012) 285–291. [28] D. Chajuss, Soy molasses: processing and utilization as a functional food, in: Soybeans as Functional Foods and Ingredients, AOCS Press, Illinois, 2006 (Chapter 10). [29] F. Haghseresht, S. Nouri, G.Q.M. Lu, Effect of carbon surface chemistry and solution pH on the adsorption of binary aromatic solutes, Carbon 41 (2003) 881–892. [30] B.H. Hameed, I.A.W. Tan, A.L. Ahmad, Adsorption isotherm, kinetic modeling and mechanism of 2,4,6-trichlorophenol on coconut husk-based activated carbon, Chem. Eng. J. 144 (2008) 235–244. [31] I.D. Mall, V.C. Srivastava, N.K. Agarwal, Removal of congo red from aqueous solution by bagasse fly ash and activated carbon: kinetic study and equilibrium isotherm anlyses, Chemosphere 61 (2005) 492–501. [32] Y. Önal, C. Akmil-Basßar, C. Sarıcı-Özdemir, Investigation kinetics mechanisms of adsorption malachite green onto activated carbon, J. Hazard. Mater. 146 (2007) 194–203. [33] C. Namasivayam, R.T. Yamuna, Adsorption of Direct Red 12B by biogas residual slurry: equilibrium and rate process, Environ. Pollut. 89 (1995) 1–7.

Y. Shi et al. / Chemical Engineering Journal 215–216 (2013) 113–121 [34] D. Mohan, K.P. Singh, Single- and multi-component adsorption of cadmium and zinc using activated carbon derived from bagasse—an agricultural waste, Water Res. 36 (2002) 2304–2318. [35] E.N. El Qada, S.J. Allen, G.M. Walker, Adsorption of Methylene Blue onto activated carbon produced from steam activated bituminous coal: a study of equilibrium adsorption isotherm, Chem. Eng. J. 124 (2006) 103–110. [36] K. Kadirvelu, K. Thamaraiselvi, C. Namasivayam, Adsorption of nickel (II) from aqueous solution onto activated carbon prepared from coirpith, Sep. Purif. Technol. 24 (2001) 497–505. [37] Y. Al-Degs, M.A.M. Khraisheh, S.J. Allen, Competitive adsorption of reactive dyes from solution: equilibrium isotherm studies in single and multisolute systems, Chem. Eng. J. 128 (2007) 163–167. [38] D.A. Rickert, M.A. Meyer, J. Hu, P.A. Murphy, Effect of extraction pH and temperature on isoflavone and saponin partitioning and profile during soy protein isolate production, J. Food Sci. 69 (2004) C623–C631. [39] W.L. Whaley, J.D. Rummel, K. Niksa, Interactions of genistein and related isoflavones with lipid micelles, Langmuir 22 (2006) 7175–7184. [40] S.K. Lee, K.H. Row, Estimation for retention factor of isoflavones in physicochemical properties, Bull. Korean Chem. Soc. 24 (2003) 1265–1268. [41] A. Da˛browski, P. Podkos´cielny, Z. Hubicki, M. Barczak, Adsorption of phenolic compounds by activated carbon—a critical review, Chemosphere 58 (2005) 1049–1070. [42] X.J. Peng, Y.H. Li, Z.K. Luan, Adsorption of 1,2-dichlorobenzene from water to carbon nanotubes, Chem. Phys. Lett. 376 (2003) 154–158.

121

[43] I. Abe, K. Hayashi, M. Kitagawa, Adsorptive mechanism on activated carbon in the liquid phase. I. Free energy change for adsorption of organic compounds from aqueous solution on activated carbon, Bull. Chem. Soc. Jpn. 52 (1979) 1899–1904. [44] V.C. Srivastava, M.M. Swamy, I.D. Mall, Adsorptive removal of phenol by bagasse fly ash and activated carbon: equilibrium, kinetics and thermodynamics, Colloids Surf. A 272 (2006) 89–104. [45] M.J. Iqbal, M.N. Ashiq, Adsorption of dyes from aqueous solutions on activated charcoal, J. Hazard. Mater. 139 (2007) 57–66. [46] K.U. Ahamad, M. Jawed, Kinetics, equilibrium and breakthrough studies for Fe (II) removal by wooden charcoal: a low-cost adsorbent, Desalination 251 (2010) 137–145. [47] M.V. Sureshkumar, C. Namasivayam, Adsorption behavior of Direct Red 12B and Rhodamine B from water onto surfactant-modified coconut coir pith, Colloids Surf. A 317 (2008) 277–283. [48] A.A. Ahmad, B.H. Hameed, N. Aziz, Adsorption of direct dye on palm ash: kinetic and equilibrium modeling, J. Hazard. Mater. 141 (2007) 70–76. [49] Y. Bulut, N. Gözübenli, H. Aydın, Equilibrium and kinetics studies for adsorption of direct blue 71 from aqueous solution by wheat shells, J. Hazard. Mater. 144 (2007) 300–306. [50] C. Namasivayam, D. Kavitha, Removal of congo red from water by adsorption onto activated carbon prepared from coir pith, an agricultural solid waste, Dyes Pigm. 54 (2002) 47–58.