Sequential extraction and separation using ionic liquids for stilbene glycoside and anthraquinones in Polygonum multiflorum

Journal of Molecular Liquids 241 (2017) 27–36

Contents lists available at ScienceDirect

Journal of Molecular Liquids journal homepage: www.elsevier.com/locate/molliq

Sequential extraction and separation using ionic liquids for stilbene glycoside and anthraquinones in Polygonum multiflorum Xueting Feng, Hang Song, Bing Dong, Yang Yang, Shun Yao ⁎ School of Chemical Engineering, Sichuan University, 610065 Chengdu, China

a r t i c l e

i n f o

Article history: Received 24 March 2017 Received in revised form 27 May 2017 Accepted 2 June 2017 Available online 03 June 2017 Keywords: Ionic liquid Polygonum multiflorum Sequential extraction Stilbene glycoside Anthraquinones

a b s t r a c t In recent years, ionic liquids (ILs) have received more attentions in green extraction and separation process as a kind of novel environment-friendly solvent, and their application in the field of natural products needs to be extended. In this paper, a new method of the sequential extraction for stilbene glycoside and anthraquinones from Polygonum multiflorum was developed for the first time and different ionic liquids was used and compared. It was found two benzothiazolium ILs including [BBth][Br] and [HBth][p-TSA] had good selectivity and high efficiency. In addition, the important extraction conditions were also investigated and optimized. Compared with the traditional acid/alcohol-water extraction process, the extraction assisted by ionic liquid not only can obtain higher yield in shorter time, but also can realize the selective extraction for two different kinds of constituents with less consumption. Furthermore, the recovery of target compounds and recycling of ionic liquid were also studied and it was found that solvent extraction combined with cation-exchange resin had the ideal recovery performance. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Stilbene glycoside (C20H22O9, abbreviated as SG) is a kind of polyhydroxy phenolic compound, which is one of particular constituents of Polygonum multiflorum in Polygonaceae family. It has been found with various bioactivities of eliminating free radicals, preventing cancer, lowering cholesterol, inhibiting atherosclerosis, protecting liver, vasodilating blood vessels and so on [1–4]. Because it is not stable, SG is easy to decompose in its aqueous solution under high temperature, and it is also unstable in acidic solution [5,6]. Anthraquinones is another kind of main components as natural medicine in Polygonum multiflorum besides SG, which is mainly composed of emodin, physcion, chrysophanol and rhein, etc. Especially, the content of emodin and physcion can reach 2.51% (dry weight) in total extract of Polygonum multiflorum [7]. They have antibacterial, anti-inflammatory, choleretic, diuretic, and immune regulation and cardiovascular protection functions [8–11]. Current methods of extracting anthraquinones mainly include water extraction, organic solvent extraction, acid extraction, and enzyme extraction. Generally, concentrated sulfuric acid or the mixture of concentrated sulfuric acid and chloroform is most popularly used [12], and both of the two solvents should be replaced in modern cleaner production for their well-known disadvantages. Moreover, their ⁎ Corresponding author. E-mail address: [email protected] (S. Yao).

http://dx.doi.org/10.1016/j.molliq.2017.06.012 0167-7322/© 2017 Elsevier B.V. All rights reserved.

extraction efficiency is low and duration is long. Furthermore, the polarity and stability of SG and anthraquinones are very different; the subsequent separation of them will be unavoidable if the nonselective system is employed (just like alcohol-water) after one-step extraction. In a word, the new and cleaner method for the extraction of the two different natural products is urgently needed at present. In recent years, ionic liquids (ILs) have received more attentions in green technology and process as a kind of novel solvent, and there have been enough successful examples. On the basis of their unique advantages of ideal thermal stability, selective extraction ability, comprehensive dissolving capacity, negligible vapor pressure, excellent structural designability and so on, ILs have been successfully applied in academia and industry fields. Especially in the extraction of bioactive natural products, ILs are being regarded as an attractive and effective replacement or alternative of conventional volatile organic solvents. Traditional alcohol-water system will extract most of constituents in herbs without any selectivity, and the great amount of coexisting impurities can result in the difficulty of obtaining target compounds. Moreover, if they belong to different structural types and need further separation, conventional extraction techniques cannot meet this requirement. Currently, the application of ILs in the field of natural bioactive products is attracting more and more attention for their extraction, separation together with hydrolysis [13,14]. For instance, 2.065 mol/L [BMim][HSO4] or 2.565 mol/L acidified [BMim]Br solution was selected for the extraction of myricetin and quercetin from Myrica rubra leaves

28

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

[15]; the extraction performance of three kinds of 1-alkyl-3methylimidazolium ILs was investigated and compared for isoflavones from Iris tectorum Maxim [16]; Wang and his coworkers used 2.0 mol/L [BMim]Br (1-butyl-3-methylimidazolium bromide) solution to extract anthraquinones in Rhubarb for their UPLC analysis [17]. More recently, imidazolium ILs were found to have good performance in the extraction for three antioxidant acids (gallic acid, vanillic acid and syringic acid) from their aqueous solution [18] and [BMim][BF4]based homogenate system was applied to extract orientin and vitexin from Trollius chinensis flowers [19]. Through the comprehensive searching in ISI Web of Science©2016 THOMSON REUTERS, [BMim][Br] and [BMim][BF4] are found as two most-frequently used ILs in the reported applications of extraction, and number of the literatures of flavonoids N phenols/acids N alkaloids N other types of natural products. However, most of these studies only focused on imidazolium-type ILs and their extraction ability for one kind of compounds with similar structures; and the separation of related products and the recovery of ILs were not be investigated and explored simultaneously. A comprehensive comparison is still necessary to search for the valuable relationships and regularity closely related with extraction. High selectivity, efficiency, recyclability and designability of ILs have not been exhibited fully in published research, so more and more new methods and findings about the structure-activity relationship of ILs are expected in their further applications for natural constituents. Based on the above background, different ionic liquids would be investigated and compared for the extraction of stilbene glycoside and anthraquinones from Polygonum multiflorum in this paper. The selectivity and recyclability of ILs would be fully utilized in order to develop the sequential extraction method. Then related effects of ILs structure and the other important extraction conditions would be explored on the extraction efficiency for target compounds. Moreover, the extraction process would be optimized by response surface method and compared with traditional ways. Finally, the recovery of target products and recycling of ionic liquid would be also studied. The following study is expected to provide new meaningful reference for the extraction of similar compounds. 2. Experimental 2.1. Materials All chemicals involved in this study were at least of analytical reagent grade. Methanol used for HPLC was of chromatographlic grade and purchased from Chemical reagents factory, Chengdu, China. Experimental water was redistilled and deionized. D001 resin was obtained from Bohong Technology Co, Ltd. (Tianjin, China), D113 and 732 resins were supplied by Guangfu chemical institute (Tianjin, China) and Kelong Chemical reagents factory, respectively. All of these cationexchange resins were pretreated before use, and their Na-form (carrying Na+ cation) could be converted to H-form (carrying H+ cation) through treatment with 5% HCl aqueous solution. All of standard compounds used for HPLC were purchased from RuiQi Biological Technology Company (Shanghai, China). Rhizome of Polygonum multiflorum originated from Danba county in Tibetan Autonomous Prefecture of Garzê in Sichuan province and was purchased from local drug market, and then the herbal raw materials were milled and dried. The sample powders passed through a stainless steel sieve and the particle size was controlled in 60 mesh. All samples were stored in closed desiccators until use. 2.2. Apparatus Morphology of herbal powders was observed with JSM-7001F scanning electron microscopy (JEOL Co., Ltd., Tokyo, Japan). SHA-CT thermostatic oscillator (Jinli instrumental Co., Ltd., Tianjin, China) and KQ2200DA ultrasonic extractor (Kunshanshumei Co., Ltd., Jiangsu, China)

were used in the extraction step. HPLC analysis was performed with an LC-20AT pump, a SPD-M20A PDA detector (Shimadzu, Kyoto, Japan), a Symmetry C18 column (Waters, Massachusetts, USA), and an HCT-360 LC columncooler/heater (HengaoTech & Dev, Tianjin, China). A Class-VP workstation (Shimadzu, Kyoto, Japan) was used for data acquisition. The pH meter was provided by ShiNuo physical optical instrument Co., Ltd. (Shanghai, China). 2.3. Synthesis of various ionic liquids Three great series of imidazolium, benzothiazolium and tropinium ILs (as shown in Table 1) were synthesized according to the reported procedures [20–22] except for [BnMim][Cl]. It was synthesized according to the following procedure: under the protection of highly pure N2, 4.15 g 1-methyl imidazole (0.05 mol) was mixed with 20 mL redistilled toluene, and 12.6 g benzyl chloride (0.l mol) was added dropwisely during the reaction at 40 °C. After refluxing for 12 h, the organic layer was washed with absolute ether repeatedly, and then the solvent was removed by evaporation. Finally, the pale yellow viscous liquid was obtained as the product of [BnMim][Cl]. All the ILs were dried for 4 h under vacuum at 90 °C and stored in closed desiccators before use. The purity of ILs was firstly checked by proton nuclear magnetic resonance, and then the HPLC analysis for various ILs was was carried out on Waters C18 chromatographic column (3.9 × 150 mm, 5 μm i.d.) at column temperature of 25 °C; the mobile phase was composed of methanol-water (23:77, V/V) for benzothiazolium ILs, or acetonitrilewater (20:80, V/V) for imidazolium ILs, or acetonitrile-water (12:88, V/V) for tropine-based ILs; the flow rate was 1.0 mL/min and injection volume was 10 μL. A 2000ES evaporative light scattering detector (Alltech, San Diego, USA) was used to analyze these ILs. As the result, the purities of all the investigated ILs were in the range of 96.5–99.2%. 2.4. Extraction procedure 0.50 g powders (60 mesh) of Polygonum multiflorum were weighed accurately and placed in 50 mL Erlenmeyer flask; the selected ionic liquid was dissolved in a certain amount of deionized water and then the concentration of 0.1–1.2 mol/L of its solution was obtained. Then: (1) Under the certain ratio of solid to liquid (g/mL, 1:5, 1:10, 1:20, 1:30, 1:40), the herbal powders was mixed with the IL solution and extracted for various duration (1, 3, 5, 10, 30, 60, 120 min) in the thermostatic oscillator under room temperature. After extraction, the solution was filtered and the filtrate was diluted with chromatographic methanol to 50 mL, and then the sample solution was filtered with a microporous membrane of 0.45 μm before HPLC analysis. (2) Under the certain ratio of solid to liquid (g/mL, 1:5, 1:10, 1:20, 1:30, 1:40), the herbal powders was mixed with the IL solution and extracted for various duration (5, 10, 20, 30, 60, 90, 120 min) in the ultrasonic extractor with the power of 40, 60, 80, 100 and 120 W. After extraction, the solution was filtered and the filtrate was diluted with methanol of chromatographic grade to 50 mL, and then the sample solution was filtered with a microporous membrane of 0.45 μm before HPLC analysis. 2.5. Quantitative analysis of stilbene glycoside and anthraquinones (1) The HPLC analysis for stilbene glycoside was developed according to Chinese Pharmacopoeia (2015 edition), which was carried out on C18 chromatographic column (3.9 × 150 mm, 5 μm i.d.) at column temperature of 25 °C; the mobile phase was composed of acetonitrile and water (25:75, V/V) and the flow rate was 1.0 mL/min. Detection wavelength was set at 320 nm and injection volume was 10 μL. Retention time of standard compound of stilbene glycoside was nearly 6 min.

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36 Table 1 The basic information of investigated ILs.

No.

Abbreviation of IL

Cations

Anions

Acid–base property

1

[C3tr][Br]

Br–

Neutral

2

[Bntr][Cl]

Cl–

Neutral

3

[BnMim][Br]

Br–

Neutral

4

[BnMim][Cl]

Cl–

Neutral

5

[BMim][Br]

Br–

Neutral

6

[BMim][PF6]

PF6–

Neutral

7

[HMim][HSO4]

HSO4–

Acid

29

was determined as y = 3.2 × 10−7x + 0.01314 (R2 = 0.9999), where y was its concentration (mg/mL) and x was the value of peak area, respectively. (2) The HPLC analysis for anthraquinones was developed according to previous study [23], which was also carried out on C18 chromatographic column (3.9 × 150 mm, 5 μm i.d.) at column temperature of 25 °C; the mobile phase was composed of methanol and water (80:20, V/V) and the flow rate was 1.0 mL/min. Detection wavelength was set at 254 nm and injection volume was 10 μL. It was found the major anthraquinones were emodin and physcion in the extracts, which could be well separated with coexisting ionic liquid and the latter had no significant effect on their retention time (see Fig. 1b). The standard curves of emodin and physcion were determined as y1 = 2.4 × 10−7x1 + 0.0111 (R2 = 0.9996) and y2 = 2.4 × 10−7x2 + 0.0286 (R2 = 0.9997) successively, where y was their concentration (mg/mL) and x was the value of peak area, respectively. The concentrations of target constituents (C) in extraction solution were determined by above HPLC methods and used to calculate the extraction efficiency (%), which could be obtained through the following equation: Extraction efficiency (%) = (C × V/M) × 100, where V was the volume of the extraction solution and M was the weight of herbal powders, respectively. Moreover, for the purpose of convenient comparison, relative extraction efficiency was also calculated when the maximum extraction efficiency (%) in comparison was taken to be 100%. 3. Results and discussion 3.1. Screening of ILs for the extraction of stilbene glycoside and anthraquinones

8

[PSMim][HSO4]

HSO4–

Acid

9

[BBth][Br]

Br–

Neutral

10

[BBth][BF4]

BF4–

Neutral

11

[BBth][PF6]

PF6–

Neutral

12

[BnBth][Cl]

Cl–

Neutral

13

[HBth][BF4]

BF4–

Neutral

14

[HBth][CH3SO3]

CH3SO3–

Acid

15

[HBth][p–TSA]

C7H8O3S–

Acid

Moreover, it could be well separated with coexisting ionic liquid and the latter had no significant effect on its retention time (see Fig. 1a). The standard curve of stilbene glycoside

The physical and chemical properties of ionic liquids can be greatly affected by the combination of cations and anions, so their extraction performance for different organic compounds is very different. Generally, the structural features of target constituents and the special characteristics of ILs as quaternary ammonium salts should be fully considered before the extraction, and then the potential ionic liquids are selected according to the principle of ‘like dissolves like’ to achieve efficient and highly selective extraction. Stilbene glycoside and anthraquinones have more free hydroxyl and phenyl groups in their structure, which are the possible interaction sites with ILs. Three series of cations with alkyl or aryl substituent group were − investigated, and those popular anions (including X−, BF− 4 , PF6 , − − − CH3SO3 , p-TSA and HSO4 ) were chosen to be coupled with them. As the result, the ILs with benzothiazolium cation were found to have the best extraction ability for the two kinds of different constituents at the same time. Fig. 2a and b show performance comparison of ILs solutions on the extraction efficiency of stilbene glycoside, and the results indicate the effect of anions is more significant than that of cations. Anions not only can affect the property of ILs more apparently, but also can provide stronger interaction with target compounds for their smaller volume. Moreover, in the comparison of three series of ILs with same anions, benzothiazolium ILs had the highest extraction efficiency, followed by tropinium ILs, and imidazolium ILs showed the weakest extraction ability. As the result, [BBth][Br] was found to be the best and the second was [C3Tro][Br]. Furthermore, in the comparison of various ILs with same cations, [BBth][BF4] and [BBth][Br] were found to be the most ideal ILs. As known to all, Cl−, Br− and BF− 4 are hydrophilic anions. As for [BBth][PF6], it has stronger hydrophobicity and is unsuitable for the extraction of glycosides with high polarity. In particular, [HBth][CH3SO3] and [HBth][p-TSA] belong to the acidic ILs, which had lower extraction efficiency than the neutral ionic liquids. Because stilbene glycoside is

30

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

Fig. 1. HPLC chromatogram of stilbene glycoside (a) and anthraquinones (b) in the extracts and standard solutions.

unstable, the glycosidic bond is easy to be hydrolyzed under the acidic condition. Considering the instability of [BBth][BF4] in its aqueous solution, [BBth][Br] was chosen in the following research. In the similar way, various kinds of ILs solutions on the extraction efficiency of emodin and physcion were also compared. The results indicated [HBth][p-TSA] had the best performance and the second was [HBth][CH3CO3]. Compared with immidazolium ILs, benzothiazolium ILs contain larger conjugated aromatic system as same as emodin and physcion, and stronger π-π interaction occurs between them, which is beneficial for the improvement of the extraction efficiency of emodin and physcion. So the performance of benzothiazolium ILs was better than that of those immidazolium ILs. On the other hand, acidic ionic liquids had higher extraction efficiency, and Brønsted acid strength sequence of tested eight ILs was [HBth][p-TSA] N [HBth][CH3CO3] N [HBth][BF4] N [PSMIM][HSO4] N [HMIM][HSO4] N [BMIM][PF6] ≈ [BMIM][Br] ≈ [BBth][Br] (see Table 2), which was determined according to the reference [21] and basically accorded with the order of their extraction ability. As known to all, anthraquinones have phenolic

hydroxyl groups, which will make them in their molecular status under acidic condition and undergo dissociation in basic condition. Above findings can prove both the π-π stacking effect and the undissociated status of target moleculars are very important for the extraction result for emodin and physcion. Then [HBth][p-TSA] was selected in the following investigation. 3.2. Effects of various conditions on the extraction of stilbene glycoside Concentration of ionic liquid solution has great influence on the destruction of cell wall and mass transfer process of extraction solution, which directly determines the extraction efficiency of ionic liquid on the target substance. In the following investigation, 10 mL of [BBth][Br] aqueous solutions with 0.1 mol/L ~ 1.2 mol/L were compared in the extraction of 0.5 g herbal powders for 1 h. As shown in Fig. 3a, the ionic liquid concentration was linearly proportional to the extraction amount of stilbene glycoside in the range of 0–0.8 mol/L; in the range of 0.8– 1.2 mol/L, the extraction efficiency of SG was still increased slowly and

Fig. 2. Performance comparison of cations (a) and anions (b) on the extraction efficiency of stilbene glycoside.

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36 Table 2 Brønsted acid strength of ionic liquids. Object

Amax (AU)

[I] (%)

[HI] (%)

H0

p-nitroaniline [BMIM][Br] [BBth][Br] [BMIM][PF6] [HMIM][HSO4] [PSMIM][HSO4] [HBth][BF4] [HBth][CH3CO3] [HBth][PTSA]

1.43 1.42 1.42 1.41 1.38 1.31 1.31 1.26 1.18

100 99.30 99.30 98.60 96.50 91.61 91.61 88.11 82.52

0 0.70 0.70 1.40 3.50 8.39 8.39 11.89 17.48

– 3.14 3.14 2.84 2.43 2.03 2.03 1.86 1.66

31

step of sequential extraction, which were found very unfavorable for high selectivity of the extraction of stilbene glucoside in above process. They can enhance mass transfer significantly and shorten extraction time; meanwhile the large amount of other constituents can be extracted together with stilbene glucoside. The advantage of selective extraction with [BBth][Br] would be completely lost, and the result would be as same as that in the extraction with conventional solvents. Moreover, 30 min was not longer than the extraction time (66, 80 or 90 min) in previous reports [24–26]. The first step is the key process in sequential extraction and separation for involved constituents from Polygonum multiflorum. 3.3. Effects of various conditions on the extraction of anthraquinones

then unchanged. For saving operation cost, 0.8 mol/L was the optimal ionic liquid concentration for the extraction. Besides the types and concentrations of ionic liquids, the solid-liquid ratio of the extraction system is also an important factor affecting the extraction efficiency. If the amount of the ionic liquid is too small, it will lead to incomplete extraction of target product; on the contrary, too much ionic liquid will result in the increase of solution viscosity, and then increase the difficulty of the mass transfer process and recovery of IL. Here 0.8 mol/L [BBth][Br] aqueous solution was used in the extraction of 0.5 g herbal powders for 1 h, and the suitable solid-liquid ratio was screened from 1:5 to 1:40 (g/mL) in the following investigation. It can be found from Fig. 3b that, in the range of 1:5–1:20, the extraction efficiency of stilbene glycoside will be higher with the increase of solid-liquid ratio. In the range of 1:20–1:30, the extraction performance will not be further improved with the increase of solid-liquid ratio. The result indicates that the solid-liquid ratio of 1:20 is enough for complete extraction of stilbene glycoside in the herbal sample. After exploring the effects of ionic liquid concentration and solidliquid ratio, the influence of extraction time on extraction efficiency was also studied. In general, too short extraction time will cause the incomplete extraction for target product; after reaching full extraction, excessive time will be unnecessary and unfavorable due to the instability of stilbene glycoside. In order to avoid its decomposition in the solution, the ideal extraction time should be explored. Then 0.5 g herbal powders were extracted with 10 mL of 0.8 mol/L [BBth][Br] aqueous solutions from 1 to 180 min. It was found the complete extraction could be achieved within 30 min without any auxiliary extraction means (see Fig. 3c). The fastest growing stage of extraction efficiency is the first 10 min; after 30 min, it will be lowered because the SG concentration difference inside and outside the cell becomes small. Finally, it should be noted that there were no extra conditions including heating and microwave or supersonic wave assistance in this

Fig. 4a–d show the effects of IL concentration, solid-liquid ratio, extraction time and ultrasonic power on the extraction result of anthraquinones, respectively. It can be found the major changing trend under related conditions is similar with that in the extraction of stilbene glycoside. In the concentration range of 0–0.5 mol/L, extraction efficiency of emodin and physcion became higher and higher with the increasing IL concentration; when the concentration was changed from 0.5 to 1.5 mol/L, the extraction amount of emodin and physcion began to decrease gradually. The main reason was the increase of the viscosity of the extraction liquid; for emodin and physcion the IL concentration of 0.5 mol/L was the most appropriate. As for the effect of solid-liquid ratio, 1:30 g/mL seemed to be the best choice for the two compounds. Through the comparison for different extraction duration, it was found the dissolution rate of emodin and physcion was slower than that of stilbene plycoside in the former study. It took 60–90 min to achieve the full extraction when 0.5 mol/L [HBth][p-TSA] with the solid-liquid ratio of 1:30 was used and ultrasonic power was 100 W. Predictably, the extraction speed can be higher in two-step sequential extraction, because the cells will have been swelled fully in the first step of extraction for SG and improved permeability of cell membrane will be beneficial for the rapid extraction in the second step of extraction for anthraquinones. Finally, ultrasonic power is also very important in extraction for cavitation, mechanical and thermal effects. High power is advantageous to enhance mass transfer by increasing the movement speed of the media molecules and their penetrability of cell wall. For emodin and physcion the ultrasonic power of 80 W was the most ideal, and excessive power will result in the more dissolution of possible impurities. 3.4. Optimization by response surface method and comparison with traditional methods Above procedures for extraction of target constituents were optimized by using a central composite design approach and all the

Fig. 3. Effects of IL concentration (a), solid-liquid ratio (b) and extraction time (c) on the extraction efficiency of stilbene glycoside.

32

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36 Table 3 The credibility analysis of the regression equations.

33

Table 4 Comparative study of extraction efficiency using different extraction methods.

Items

Values

Items

Values

Std. Dev. Mean C.V. % PRESS

5.32 × 10−3 0.95 0.56 2.18 × 10−3

R-Squared Adj R-Squared Pred R-Squared Adeq Precision

0.9859 0.9678 0.8449 27.089

related parameters were included for selection as various independent variables (each variable representing five levels). The experimental design was completed by Design Expert 7.1.3, which was also used to analyze the results through response surface. As the result, the regression second-degree complete model for the yield of stilbene glycoside (Z, mg) in terms of coded factors is presented as the following equation:

Z ¼ 0:96 þ 0:027A−0:026B þ 9:65  10−3 C þ 0:010AB þ 7:875  10−3 AC−0:010 BCEBC−7:173  10−3A2 −1:273  10−3 B2 −6; 972  10−3 C2

where A, B and C represents IL concentration, solid-liquid ratio and extraction time, respectively. The value of R2 (0.98567) of this model indicated a high degree of correlation between the observed and predicted values. The analysis of variance (ANOVA) of the regression model demonstrated that the model was statistically significant (p b 0.0001, Fvalue = 57.77). Furthermore, results of the error analysis indicated that the lack of fit was not significant (p N 0.05). B had the greatest influence on the extraction of stilbene glycoside as highly significant variables (p b 0.0001), followed by A and C. Experiments were performed to show that, when the IL concentration (A) was set at 1.15 mol/L, solid-liquid ratio (B) was 1:30 and extraction time (C) was 20 min, the measured value of SG yield from herbal powders was 25.82 mg/g, which was consistent with the predicted value as the maximum yield under optimal conditions. The credibility analysis of the regression equations in Table 3 also can prove the model can be used to predict the extraction result. In similar way, the optimal conditions for extracting anthraquinone can be obtained with response surface method. The optimal ionic liquid concentration was 0.64 mol/L, the extraction time was 52.57 min, the best solid-liquid ratio was 1:30 and the ideal ultrasonic power was 89.51 W (actual power was set at 90 W). Moreover, the influential order of the four variables was ultrasonic power N extraction time N solid-liquid ratio N IL concentration, which can prove the ultrasonic power always plays a very crucial role in ultrasonic assisted extraction again. Under above optimal conditions, the extraction efficiency of stilbene glycoside and anthraquinones resulting from the developed method was compared with corresponding performance of other traditional methods from previous research. In Table 4, all the listed methods used the same amount of herbal powders (0.5 g) and solid-liquid ratio (1:10), and their difference existed in the extraction media, time and auxiliary conditions. According to the comparison, the new extraction conditions can be verified and proved to be more efficient. The ionic liquid extraction method can greatly increase the yield of target constituents in Polygonum multiflorum, which indicates that the extraction method with ionic liquid has great potential. Of course, compared with the traditional organic solvent and water, the unit price of ionic liquids is higher; but the dosage of ionic liquid is low, and it can be reused after effective recovery. It cannot prevent it from becoming one of the very promising extraction methods in cleaner production.

Objects

Extraction conditionsa

Yield (mg)

Extraction efficiency (%)

Stilbene glycoside

Deionized water, 40 °C, 1.5 h 60% EtOH, r.t., 1.5 h 60% EtOH, 40 °C, 1.5 h [BBth][Br], r.t., 20 min MeOH, u.w., 1 h 80% MeOH, u.w., 1 h 80% MeOH, refluxing, 1 h [HBth][p-TSA], u.w., 1 h

8.6 12.1 12.2 12.9 64.5 76.1 89.5 149.0

0.017 0.024 0.024 0.026 0.129 0.152 0.179 0.298

Anthraquinones (emodin + physcion)

a r.t. is the abbreviation of room temperature; u.w. is the abbreviation of ultrasonic wave.

3.5. Sequential extraction method and recovery of products and ILs On the basis of above studies, a sequential extraction method was developed to extract and separate the two different kinds of target constituents including stilbene glycoside and anthraquinones. The herbal powders (60 mesh) were firstly extracted by 1.15 mol/L [BBth][Br] with the solid-liquid ratio of 1:30 in thermostatic oscillator for 20 min, and then the system was filtered to obtain the filtrate and herbal residue respectively. The filtrate was partitioned with immiscible organic solvent to recover stilbene glycoside, and the raffinate containing IL can be reused in the next extraction. The herbal residue continued to be extract with 0.64 mol/L [HBth][p-TSA] with the solid-liquid ratio of 1:30 in ultrasonic oscillator (90 W) for 52.57 min. Finally, the system was filtered and the filtrate was adsorbed by ion-exchange resin to recover IL for its recyclable use. The collected effluent was concentrated to obtain the product of anthraquinones. The scheme of general process flowchart was shown in Fig. 5, which has been scaled up to amplification experiment on the DC-NSG multifunctional (heating & ultrasonic assistance) extraction and recovery unit (actual filled amount of herbal materials: 500 g, solvent volume: 15 L, Dacheng Experimental Equipment Co. Ltd., Shanghai, China). The final extraction efficiency of SG and anthraquinones was 0.0254% and 0.283% respectively, which are close to the values in Table 4. Furthermore, extraction of herbs is usually based on mass transfer of internal phase, which occurs on the solidliquid two-phase interface from intracellular to extracellular. The photos of scanning electron microscope (SEM, see Fig. 6) showed the herbal particles were steadily swelled and destructed in the whole mass-transfer process of sequential extraction. In above process, solvent extraction was used to separate stilbene glucoside (SG) from ionic liquid solution. A variety of insoluble or slightly soluble organic reagents in water were tested for the extraction and phase ratio (Vextract:Vorganic solvent) = 1:3. IL recovery (%) can be calculated according to its concentrations before and after extraction, and the standard curve for its HPLC quantitative analysis was y = 5 × 10− 7x + 0.03586 (R2 = 0.9999, linear range: 0.1618– 6.969 μg) under the same chromatographic conditions as SG. The results are shown in Table 5, which indicate both n-butanol and pentanol can achieve good recovery of ILs. They can extract the large majority of stilbene glycoside (96.9%) from the extract, and other organic solvents cannot realize so effective separation and some of them have higher toxicity. Moreover, most ionic liquid can be remained in the solution for recyclable use. Furthermore, the effect of temperature on the extraction result was also investigated (see Table 6), and five different alcohols were compared. At 15 and 25 °C, there was no significant difference in the recovery of the stilbene glycoside when n-butanol was used, which could be maintained at nearly 97%. However, when the temperature reached 35 °C, the extraction rate of stilbene glycoside obviously decreased in all of tested solvents. On the contrary, the

Fig. 4. Performance comparison of various concentrations of ionic liquid (a), solid-liquid ratio (b), time (c) and ultrasonic power (d) for the extraction of emodin (left) and physcion (right).

34

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

Fig. 5. The scheme of sequential extraction method and recovery of products and ILs.

Fig. 6. SEM photos of Polygonum multiflorum particles before extraction (a, a1), after extraction with [BBth][Br] (b, b1) and then after extraction with [HBth][p-TSA] (c, c1).

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

35

Table 5 Results of extraction with different organic solvents. No.

Objects

Residual SG in the extract (mg)

Recovery of SG (%)*

Ionic liquid in the raffinate (mg)

Recovery of IL (%)*

1 2 3 4 5 6 7 8 9 10 11

Chloroform Ethyl acetate n-Hexane Ether Dichloromethane Toluene n-butanol Xylene Cyclohexane n-Pentanol Extract

12.49 4.02 11.85 11.77 11.78 11.72 0.39 12.13 11.92 0.59 12.68

1.29 68.2 6.34 6.99 6.92 7.43 96.9 4.19 5.81 95.4 0

2.19 1.90 2.12 2.12 2.12 2.05 2.14 2.09 2.10 2.15 2.19

99.8 86.6 96.7 96.9 96.9 93.3 97.8 95.3 95.9 98.1 0

* Recovery of SG (%) = (1-the residual amount of SG in extract ÷ the initial amount of SG in extract) × 100%. * Recovery of IL (%) = (the amount of residual ILs ÷ the initial amount of ILs) × 100%.

increase of temperature was beneficial to the recovery of IL, and the recovery in cyclohexanol even reached 99.9% at the temperature of 35 °C. Above two opposite changes can be understood easily because the water solubility of stilbene glycoside and ionic liquid will increase simultaneously under high temperature. Moreover, acidic cation-exchange resins were also considered to separate [BBth][Br] from the extract containing stilbene glucoside. However, the latter also could be partly adsorbed (even hydrolyzed) by these acidic polar resins together with the coexisting IL because its polarity is a little high. So the separation method based on the resins was not chosen finally. Comprehensively, the extraction with n-butanol under 25 °C was the most ideal selection. In the final step of Fig. 5, it adopts another way to separate target constituents and IL. Because there are many coexisting compounds together with IL at the end of the whole extraction process and only IL exists in ionic status in the mixture, the strategy is to recover IL for the mixture for its recyclable use. Here five different cation-exchange resins (732H, 732Na, D001H, D001Na and D113H) were compared, and the static adsorption experiments were carried out in thermostatic oscillator under room temperature. It should be noted that 732 resin is styrene-type strongly acidic gel resin functionalized with –SO3H, D001 resin belongs to styrene-type strongly acidic macroporous resin functionalized with –SO3H, and D113 resin is acrylic-type weakly acidic macroporous resin functionalized with –COOH, respectively. The solid-liquid ratio between resins and the extract was 1:10 (W/V) and adsorption time was 20 min. IL recovery (%) can be calculated according to its concentrations before and after adsorption, and the standard curve for its HPLC quantitative

Table 6 Results of ionic liquid recovery of different alcohols under different temperatures. Temperature (°C)

Alcohols

Recovery of SG (%)

Recovery of IL (%)

15

n-Butanol n-Pentanol Isobutanol Isopentanol n-Hexanol n-Butanol n-Pentanol Isobutanol Isopentanol n-Hexanol n-Butanol n-Pentanol Isobutanol Isopentanol n-Hexanol

96.3 95.7 94.1 95.0 88.3 96.9 95.4 94.9 95.0 89.1 92.3 78.2 65.3 34.7 45.8

92.3 94.4 90.4 94.6 97.8 97.5 98.3 94.3 96.2 99.5 96.2 98.5 95.3 98.2 99.9

25

35

Fig. 7. Adsorption process of [HBth][p-TSA] on 732H resin.

analysis was y = 4.5 × 10− 8x + 0.0051 (R2 = 0.9994, linear range: 0.1575–5.042 μg) under the same chromatographic conditions as anthraquinones. It was found the order of IL recovery (%) was 732H (96.85) N D001H (95.66) N 732Na (91.33) N D001Na (88.43) N D113H (78.55). Among them strongly acidic cation-exchange resins show better performance than weakly acidic cation-exchange (732H, D001H, D113H), and the resins in H type are more ideal than those resins in Na type. The adsorption process on 732H resin was investigated from 1 to 32 min, and the result in Fig. 7 proved the recovery of [HBth][p-TSA] was very rapid and complete adsorption could be achieved after nearly 20 min. After the adsorption, the IL was desorbed by HCl-H2O mixture solution (17.5:72.5, V/V/V) for recyclable use. Finally, the effect of temperature on the IL recovery was not obvious, so the adsorption can be performed under room temperature. 4. Conclusions In this paper, three different series of ionic liquids were studied and compared for the sequential extraction of stilbene glycoside and anthraquinones from Polygonum multiflorum. The results indicated that the structures of ILs had significant influence on the extraction efficiency for target compounds. [BBth][Br] and [HBth][p-TSA] was finally selected as the optimal ILs. In addition, the important extraction conditions were also investigated and optimized. Compared with the traditional alcoholwater extraction process, the extraction by ionic liquid not only can obtain higher yield in shorter time, but also can realize the selective extraction for two different kinds of constituents with less consumption. In addition, the recovery of target compounds and recycling of ionic liquid were also studied and it was found that solvent extraction combined with cation-exchange resin had the best recovery performance. Acknowledgements Preparation of this paper was supported by the National Natural Science Foundation of China (No. 81373284, 81673316). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

Y. Chen, M. Wang, R.T. Rosen, C.T. Ho, J. Agric, Food Chem. 47 (1999) 2226–2228. Y. Kimura, H. Okuda, J. Pharm. Pharmacol. 52 (2000) 1287–1295. L. Lv, X. Gu, J. Tang, C.T. Ho, Food Chem. 104 (2007) 1678–1681. X. Gao, Y.J. Hu, L.C. Fu, Chin. J. Chin. Mater. Med. 32 (2007) 323–326. G.F. Wang, X.L. Ren, H.Z. Ouyang, A.D. Qi, M. Zhao, Chin. J. Pharm. Anal. 31 (2011) 458–462. W.J. Wang, D.L. Liu, W. Wang, Global Trad. Chin. Med. 9 (2016) 414–418. Y.L. Xu, Q. Dong, Y.C. Xiao, L. Pi, F.Z. Hu, Chin. J. Anal. Suppl. 28 (2009) 327–329. H.Q. Zhang, C.H. Zhou, Y.Q. Wu, World J. Gastroenterol. 11 (2005) 3147–3150. Z.X. Wu, B.P. Yu, L. Xu, H. Xia, Basic Clin. Pharmacol. Toxicol. 105 (2009) 167–172. Y.S. Lee, O.H. Kang, J.G. Choi, Y.C. Oh, J.H. Keum, S.B. Kim, G.S. Jeong, Y.C. Kim, D.W. Shin, D.Y. Kwon, Pharm. Biol. 48 (2010) 1285–1290. A. Alisi, A. Pastore, S. Ceccarelli, N. Panera, D. Gnani, G. Bruscalupi, M. Massimi, G. Tozzi, F. Piemonte, V. Nobili, Int. J. Mol. Sci. 13 (2012) 2276–2289.

36

X. Feng et al. / Journal of Molecular Liquids 241 (2017) 27–36

[12] Z.H. Lou, G.Y. Lv, J.J. Yu, J. Zhejiang Chin. Med. Univ. 4 (2014) 495–500. [13] B. Tang, W. Bi, M. Tian, K.H. Row, J. Chromatogr. B 904 (2012) 1–21. [14] P. Wang, C.Y. Ma, S.W. Chen, S. Zhu, Z.X. Lou, H.X. Wang, J. Clean. Prod. 79 (2014) 265–270. [15] F.Y. Du, X.H. Xiao, G.K. Li, Biomed. Chromatogr. 25 (2011) 472–478. [16] Y. Sun, W. Li, J. Wang, J. Chromatogr. B 879 (2011) 975–980. [17] C.X. Lu, H.X. Wang, W.P. Lv, C.Y. Ma, P. Xu, J. Zhu, J. Xie, B. Liu, Q.L. Zhou, Chromatographia 74 (2011) 139–144. [18] Y.C. Fan, X.J. Li, L.L. Yan, J. Li, S.F. Hua, L.F. Song, R.P. Wang, S.N. Sha, Sep. Purif. Technol. 172 (2017) 480–488. [19] F.L. Chen, Q. Zhang, K.L. Mo, S.M. Fei, H.Y. Gu, L. Yang, Sep. Purif. Technol. 175 (2017) 147–157.

[20] Y.Y. Wang, W. Li, L.Y. Dai, Chem. Pap. 62 (2008) 313–317. [21] X.S. Zhou, J.B. Liu, W.F. Luo, Y.W. Zhang, H. Song, J. Serb. Chem. Soc. 76 (2011) 1607–1615. [22] J. Lu, H. Song, Y. Yang, G.F. Qian, L.R. Nie, S. Yao, J. Mol. Liq. 209 (2015) 648–656. [23] L.C. Lin, S.M. Nalawade, V. Mulabagal, M.S. Yeh, H.S. Tsay, Biol. Pharm. Bull. 26 (2003) 1467–1471. [24] L. Lv, Food Ferm. Ind. 30 (2004) 132–135. [25] T.L. Xue, Q. Wang, M. Ye, Chin. J. Ethnomed. Ethnopharm. 22 (2013) 39–41. [26] S.L. Zhao, R.H. Zhao, F.J. Xie, H.M. Wei, Chin. Tradit. Herb. Drug 37 (2006) 1496–1499.