Solid-Phase Extraction of Aristolochic Acid I from Natural Plant Using Dual Ionic Liquid-Immobilized ZIF-67 as Sorbent

: (1) Background: ZIF-67 is one of the most intriguing metal–organic frameworks already applied in liquid adsorption. To increase its adsorption performance, dual ionic liquids were immobilized on ZIF-67 in this research; (2) Methods: The obtained sorbent was used to adsorb aristolochic acid I (AAI) in standard solutions. Then, the sorbent was applied in solid-phase extraction to remove AAI from Fibraurea Recisa Pierre extracted solution. (3) Results: By analyzing the adsorption models, the highest adsorption capacity of immobilized sorbent (50.9 mg/g) was obtained at 25 ◦ C within 120 min. In the SPE process, 0.02 mg of AAI was removed per gram of herbal plant, the adequate recoveries were in the range of 96.2–100.0%, and RSDs were 3.5–4.0%; (4) Conclusions: The provided experimental data revealed that ZIF-67@EIM-MIM was an excellent potential sorbent to adsorb and remove AAI from herbal plant extract, and the successful separation indicated that this sorbent could be an ideal material for the pretreatment of herbal plants containing AAI.


Introduction
Bioactive compounds from herbal medicines are widely used; however, not all bioactive compounds in natural plants are safe. For example, Aristolochia and related plants are frequently used to cure diseases of the respiratory system in Asian countries. However, aristolochic acid I (AAI), a quite harmful compound in genus of Aristolochia plants, can cause a variety of kidney diseases and even cancer in humans [1]. The U.S. Food and Drug Administration issued a safety alert that dietary supplements or other botanicalcontaining products containing AAI have been associated with nephropathy. Wu et al. used ultra-high-performance liquid chromatography-mass spectrometer (UHPLC-HR-MS) to detect 289 samples and the results demonstrated that most representatives from the Aristolochiaceae family contain toxic AAs [2]. Hence, the AAI as a hazardous compound should be removed before using the herbal plants.
In a novel separation process, adsorption techniques using unique materials are widely applied. One of emerging porous materials is metal-organic frameworks (MOFs). They are advanced porous materials with several advantages such as simple synthesis, high porosity, and potential applications [3][4][5]. ZIF-67, one of the most intriguing MOF materials, exhibits excellent thermal and chemical stability due to its microporous structure, large surface area, and tunable pore aperture [6,7]. It has been fabricated in many types of porous composites and applied in electrochemistry [8], organic compounds separation [9], and adsorption [10]. Particularly in liquid-phase adsorption, ZIF-67 has attracted much attention for water purification. Lin and Chang removed malachite green from water using ZIF-67 with 2430.0 mg/g of adsorption capacity at 20 • C [11]. Yang et al. used magnetic embedded ZIF-67 to remove nitrophenol and methyl orange from water, and 52.5 mg/g and 738.0 mg/g of adsorption capacities were obtained, respectively [12,13]. Another graphene/ZIF-67 composite was applied by Yang et al. to remove ionic dyes in water, and the maximum adsorption capacity is up to 1714.2 mg/g [14].
Previous studies demonstrated that ZIF-67 was a great potential sorbent and the adsorption results on its modifications were desirable [15]. Some researchers have evaluated several modified ZIF-67 sorbents. For example, ZIF-67 was sulfureted by Liang et al. and a hollow sorbent was prepared, and the adsorption capacity (471.7 mg/g) of ciprofloxacin antibiotics was obtained on the sorbent [16]. Yang et al. synthesized a layered doublehydroxide composited ZIF-67 and applied it in CO 2 adsorption with a 22.2 mg/g of adsorption capacity [17]. Zhao et al. synthesized a core-shell ZIF-67@ZIF-8 sorbent and obtained a 101.9 mg/g of adsorption capacity of Rhodamine B in aqueous solution [18]. Knebel et al. applied ZIF-8-on-ZIF-67 membrane in H 2 /CO 2 gas separation [19].
Ionic liquids (ILs) are recognized as green solvents, and have been used in efficient extraction or sorbent modification. IL has some unique physicochemical properties, such as negligible vapor pressure, wide chemical and electrochemical windows, non-flammability, tunable viscosity, and miscibility with several inorganic/organic solvents, which make them potential replacements for organic solvents in several areas [20,21]. One of the IL applications is the sorbent modifier. When the ILs are modified on the surface of sorbents, their hydrophilicity/hydrophobicity properties with different cations or anions can provide various chemical bonds (hydrogen, ionic, π-π, et al.), which create chemisorption during the adsorption process, and increased the interactions between sorbents and target compounds [22]. Two methods such as impregnation and chemical modification were used to prepare modified sorbents. In the preparation of an impregnated sorbent, IL and substrate were mixed and stirred in a solvent for several hours. The process was quite simple, but the bonds were mainly van der Waals force or hydrogen bonds, which were neither strong nor stable enough [23]. Hence, some researchers preferred the chemical modification of sorbents by covalent bonding of IL [24,25].
However, there is no report about the application of dual ionic liquid-immobilized ZIF-67 to adsorb and remove AAI. Hence, in this research, three dual IL-immobilized ZIF-67 sorbents were synthesized and the adsorption capacities of AAI on them were evaluated. Finally, the sorbent was applied in SPE to remove AAI from Fibraurea Recisa Pierre extracted solution.

Apparatus
Scanning electron microscopy (SEM) images were obtained using a MIRA3 scanning microscope (TESCAN, Brno, Czech Republic). Fourier transform infrared (FT-IR, Nicolet 6700, Thermo Fisher, Waltham, MA, USA) spectroscopy was performed using KBr pellets in the range of 400.0-4000.0 cm −1 with a scan rate of 20.0 scans/min. Thermogravimetric analysis (Labsys evo, Setaram, Caluire, France) was used with a heating rate of 10 • C/min under N 2 . The powder X-ray diffraction (XRD) patterns were performed on an Empyrean diffractometer (Malvern Panalytical, Eindhoven, Netherlands) at room temperature using Cu radiation (60.0 kV and 55.0 mA) over the range 5.0-60.0 • for characterizing the crystalline structure.

Preparation of ZIF-67, ZIF-67@Br and ZIF-67@Imidazole
ZIF-67 was prepared based on a previous reported method with modifications [3]. Cobalt-nitrate hexahydrate Co(NO 3 ) 2 ·6H 2 O (1.5 g) and 2-methylimidazole (6.6 g) were dissolved in 100.0 mL of methanol, respectively. The Co(NO 3 ) 2 ·6H 2 O solution was slowly added into the 2-methyimidazole solution. The resulting mixture was stirred for 8 h and aged for 16 h at the ambient temperature. After that, the precipitate ZIF-67 was collected by centrifugation and washed repeatedly by methanol for three times, and dried at 60 • C for 12 h. Finally, 7.7 g of ZIF-67 can be obtained. NBS (7.4 g) was dispersed in 200.0 mL of dichloromethane with the assistance of ultrasonication. Then, the solution was mixed with 6.7 g of ZIF-67 and 0.3 g of AIBN in a 250.0 mL flask and stirred at 45 • C for 12 h. The solid brominated product (ZIF-67@Br) was collected by centrifugation. After rinsing with dichloromethane for four times, the 13.6 g of ZIF-67@Br was obtained after drying at 40 • C for 24 h.
Then, the imidazole modified ZIF-67 (ZIF-67@imidazole) was obtained by the reaction of ZIF-67@Br (12.5 g) and imidazole (3.2 g) in ethanol at 70 • C for 8 h. The final product (11.4 g) was collected by centrifugation and washed repeatedly with ethanol, and dried at 70 • C for 12 h.

Adsorption Isothermal and Kinetics Studies
For the adsorption isothermal study, 0.005 g of ZIF-67, ZIF-67@EIM-IM, ZIF-67@EIM-MIM, ZIF-67@EIM-EIM was immersed in 4.0 mL of AAI methanol solution with concentration in the range of 1.2-132.1 mg/L at 5, 25, 35 and 45 • C. After 20 h, the solutions after adsorption were filtered with 0.2 µm Millipore membrane and the concentrations of AAI were determined. The adsorption amount was calculated according to Equation (1) and the adsorption efficiencies were obtained using different concentrations with Equation (2). Additionally, the adsorption mechanism was investigated using the adsorption capacity from the Langmuir model (Equation (3)) and Freundlich model (Equation (4)).
where Q e (mg/g) is the adsorption capacity of AAI adsorbed onto sorbent at equilibrium. C 0 and C e (mg/mL) are the initial concentrations and equilibrium concentrations after adsorption, respectively. V (mL) is the solution volume and m (g) is the weight of the sorbents. ER (%) is the adsorption efficiency. The Langmuir parameters K L (L/mol) is related to the adsorption rate and Q max (mg/g) is the adsorption capacity. The Freundlich constant n is related to the heterogeneity of the adsorption and KF is the adsorption capacity of the adsorbent.  (5)) and pseudo-second-order (Equation (6)) kinetics models were used to fit the experimental data to evaluate the kinetics of the adsorption process.
Q t (mg/g) is the amount of AAI adsorbed at different times. k 1 (1/min) and k 2 (g/mg·min) are the rate constants for pseudo-first-order and pseudo-second-order models, respectively.

Solid-Phase Extraction of AAI from Fibraurea Recisa Pierre
First, 0.5 g of Fibraurea Recisa Pierre powder was dipped into 4.0 mL of methanol for 8 h at room temperature. The extracted solution was obtained after it was filtered with 0.2 µm Millipore membrane. A SPE cartridge (Ø 0.9 cm) was fixed in a closed container at a certain temperature, 0.05 g of ZIF-67@EIM-IL sorbent was packed into it and pretreated with 4.0 mL of methanol. Then the extracted solution was poured into the cartridge and the solution from the cartridge bottom was collected. The flow was controlled as 0.2 mL/min and the collected solution was poured back into the cartridge. After 8 times repetitions, hexane, water, methanol, and methanol/0.1 M aqueous NaOH (90.0:10.0, v/v) with different elution abilities were used to separate AAI from interferences with a flow 1.0 mL/min.

Characterization
The results of SEM analyses of the synthesized composites are presented in Figure 2. In Figure 2A, ZIF-67 showed a regular polygon-shape with a mean diameter of about 180.0 nm. When the IL was immobilized, the polygon-shape of ZIF-67 was covered ( Figure 2B) and the average diameter was slightly increased to 195.0 nm according to the scale-plate in the images. After the aggregation of dual ILs on ZIF-67, the irregular surface of sorbent was observed and the average diameter increased to 245.0 nm ( Figure 2C). In Figure S1, the variation tendencies of surface and diameter were also observed on other ILs-immobilized sorbents.
with different elution abilities were used to separate AAI from interferences with a flow 1.0 mL/min.

Characterization
The results of SEM analyses of the synthesized composites are presented in Figure 2. In Figure 2 (A), ZIF-67 showed a regular polygon-shape with a mean diameter of about 180.0 nm. When the IL was immobilized, the polygon-shape of ZIF-67 was covered ( Figure  2 (B)) and the average diameter was slightly increased to 195.0 nm according to the scaleplate in the images. After the aggregation of dual ILs on ZIF-67, the irregular surface of sorbent was observed and the average diameter increased to 245.0 nm (Figure 2 (C)). In Figure S1, the variation tendencies of surface and diameter were also observed on other ILs-immobilized sorbents. The FT-IR of all materials in Figure S2 showed a wide peak at 1400.3 cm −1 belongs to the imidazole group in ZIF-67. The -Cl peak (belongs to 1,2-dichloroethane) at 668.0 cm −1 and -C-N peaks (1165.8 cm −1 and 1235.1 cm −1 ) proved that the IL was immobilized on ZIF- The FT-IR of all materials in Figure S2 showed a wide peak at 1400.3 cm −1 belongs to the imidazole group in ZIF-67. The -Cl peak (belongs to 1,2-dichloroethane) at 668.0 cm −1 and -C-N peaks (1165.8 cm −1 and 1235.1 cm −1 ) proved that the IL was immobilized on ZIF-67. Dual ILs-immobilized ZIF-67 had larger peak area of -C-N than ZIF-67@EIM-Cl because of the double imidazole groups. More peaks of carbon chain bending appeared at 1087.2 cm −1 proved that dual ILs were immobilized. Additionally, among the three dual ILs (EIM-IM, EIM-MIM and EIM-EIM), -C-H group on the outlay of dual IL EIM-EIM was larger than the other two, so the peak in the range of 1450.5-1502.2 cm −1 had the largest area.
TGA can determine the thermal degradation of ZIF-67 and immobilized ILs. Some results of sorbents are shown in Figure S3. From 400 • C to 800 • C, 1.1 mg, 1.4 mg and 2.1 mg weight loss were observed for ZIF-67, ZIF-67@EIM-Cl and ZIF-67@EIM-MIM, respectively. ZIF-67 decomposed obviously above 400 • C due to the decomposition of the ligand, an extra weight loss was observed for ZIF-67@EIM-Cl could be resulted from the decomposition of C-C bond [30]. Similar tendency of ZIF-67@EIM-MIM were observed and more weight loss could be corresponded to decomposition of more C-C bonds.
After imidazole and ILs-immobilized on ZIF-67, the XRD spectrogram showed several weak peaks were also detected at 2θ values which were matching to ZIF-67 ( Figure S4). After immobilization, the peaks were mostly weakened conforming the structure of the functional layers placed on the surface of ZIF-67.

Adsorption Isothermal and Kinetics Studies on the Three ZIF-67@EIM-ILs
The isothermal adsorption is used to explain the relationship between the adsorption capacity of adsorbent and the adsorbates concentration when the adsorption is equilibrium at a fixed temperature. The adsorption capacities of AAI onto the three dual IL-immobilized ZIF-67 sorbents at four temperatures (5,25,35 and 45 • C) are shown in Figure 3. According to the molecular structure of AAI, -O-and -NO 2 groups could create a weak hydrogen bond and -COOH was an electron-withdrawing group, so it could form strong hydrogen bonds with IL. The physicochemical characteristic of AAI (pKa value is 3.3) dominated its anionic form in methanol solution at natural pH, so the ionic interaction between IL and AAI resulted higher concentration of AAI on sorbents [31]. AAI also contained hydrophobic phenanthrene rings and a -OCH 3 group. The phenanthrene ring interacted with ILs by π-π bond and -OCH 3 group interacted with the carbon chain of IL due to the low polarity. The polarity of the out-layer IL decreased with carbon chain length increasing. The optimum equilibrium of the interaction and hydrophobic property of AAI was obtained on methylimidazole group of the out-layer IL. So ZIF-67@EIM-MIM had higher adsorption amount than the other two sorbents.
The isothermal adsorption is used to explain the relationship between the adsorption capacity of adsorbent and the adsorbates concentration when the adsorption is equilibrium at a fixed temperature. The adsorption capacities of AAI onto the three dual IL-immobilized ZIF-67 sorbents at four temperatures (5, 25, 35 and 45 °C) are shown in Figure  3. According to the molecular structure of AAI, -O-and -NO2 groups could create a weak hydrogen bond and -COOH was an electron-withdrawing group, so it could form strong hydrogen bonds with IL. The physicochemical characteristic of AAI (pKa value is 3.3) dominated its anionic form in methanol solution at natural pH, so the ionic interaction between IL and AAI resulted higher concentration of AAI on sorbents [31]. AAI also contained hydrophobic phenanthrene rings and a -OCH3 group. The phenanthrene ring interacted with ILs by π-π bond and -OCH3 group interacted with the carbon chain of IL due to the low polarity. The polarity of the out-layer IL decreased with carbon chain length increasing. The optimum equilibrium of the interaction and hydrophobic property of AAI was obtained on methylimidazole group of the out-layer IL. So ZIF-67@EIM-MIM had higher adsorption amount than the other two sorbents.  Figure S5 showed the relationship between C0 and adsorption efficiency ER (%). It was observed that with the fixed volume of solution, the saturation of the sorbents was achieved more easily at a higher initial concentration, so the ER (%) was lowered with the increasing initial concentration. In addition, the adsorption amounts on all sorbents increased from 5 to 25 °C, and then decreased when the temperature was over 25 °C. This is because at a low temperature, motion of the molecules was too slow to obtain a fully adsorption. Then the temperature over 25 °C corresponded to a faster motion of the molecules, and the strength of interaction was not enough to bind the AAI molecules on  Figure S5 showed the relationship between C 0 and adsorption efficiency ER (%). It was observed that with the fixed volume of solution, the saturation of the sorbents was achieved more easily at a higher initial concentration, so the ER (%) was lowered with the increasing initial concentration. In addition, the adsorption amounts on all sorbents increased from 5 to 25 • C, and then decreased when the temperature was over 25 • C. This is because at a low temperature, motion of the molecules was too slow to obtain a fully adsorption. Then the temperature over 25 • C corresponded to a faster motion of the molecules, and the strength of interaction was not enough to bind the AAI molecules on sorbents. Otherwise, high temperature inhibited the adsorption of AAI on sorbents through hydrophobic interactions [32]. Figures S6 and S7 showed the relationships between equilibrium concentration C e and adsorption amount Q e by fitting the data to the Langmuir model and the Freundlich model. The results showed that the initial concentration and temperatures played important roles in the adsorption process. In addition, according to the adsorption theory, the adsorption process of AAI on the sorbents was monolayer adsorption at the homogeneous surface.
The adsorption capacity of AAI on the three sorbents as a function of time at 25 • C is shown in Figure 4. Also, the fit of pseudo-first-order and pseudo-second-order equations to the adsorption data are shown in Figures 4 and S8. The data suggested that the adsorption possibly included multi-diffusion processes. The amount of AAI adsorbed increased with time increasing, and higher initial concentration needed more time to reach the equilibrium. sorbents. Otherwise, high temperature inhibited the adsorption of AAI on sorbents through hydrophobic interactions [32]. Figure S6 and S7 showed the relationships between equilibrium concentration Ce and adsorption amount Qe by fitting the data to the Langmuir model and the Freundlich model. The results showed that the initial concentration and temperatures played important roles in the adsorption process. In addition, according to the adsorption theory, the adsorption process of AAI on the sorbents was monolayer adsorption at the homogeneous surface. The adsorption capacity of AAI on the three sorbents as a function of time at 25 °C is shown in Figure 4. Also, the fit of pseudo-first-order and pseudo-second-order equations to the adsorption data are shown in Figure 4 and Figure S8. The data suggested that the adsorption possibly included multi-diffusion processes. The amount of AAI adsorbed increased with time increasing, and higher initial concentration needed more time to reach the equilibrium.   From the theoretical model analysis, the Langmuir model fitt data better and exhibited a higher correlation coefficient than the Fre 67@EIM-MIM showed the highest adsorption capacity of 50.88 mg/ adsorption conditions were 25 °C with 120 min. To prove the optimu ity of ZIF-67@EIM-MIM, the maximum adsorption amounts of ZIF 67@imidazole, ZIF-67@EIM-Cl and the other two immobilized sorb under the same conditions. The results showed in Figure 5 proved th tion of IL can promote the adsorption performance obviously, esp MIM with dual ILs.

Solid-Phase Extraction of AAI from Real Sample
After loading 3.0 mL of extracted solution onto the sorbent, the filter solution was collected and analyzed. Figure 6A,B,G showed the chromatograms of standard solution, extraction solution before and after loading. Compared (A) and (B), peak areas of interferences and AAI obviously decreased. It revealed that both interferences and AAI were adsorbed on the sorbent. Then the effects of the washing and elution solutions were assessed. Large amounts of non-polar and polar interference were washed out using hexane ( Figure 6C), water ( Figure 6D), and methanol ( Figure 6E) because of weak interactions between these interference and sorbent. Therefore, the volume of these three solvents needed to be increased until no more interference was detected. Accordingly, 6.0 mL, 4.0 mL and 6.0 mL of hexane, water and methanol were applied, respectively. Finally, methanol/0.1 M aqueous NaOH (90.0:10.0, v/v) with higher elution ability was applied, and 6.0 mL of it can remove all the adsorbed AAI from the sorbent ( Figure 6F). The result proved that AAI was successfully separated from interferences and 0.02 mg of AAI can be removed per gram of Fibraurea Recisa Pierre.
teractions between these interference and sorbent. Therefore, the volume of these three solvents needed to be increased until no more interference was detected. Accordingly, 6.0 mL, 4.0 mL and 6.0 mL of hexane, water and methanol were applied, respectively. Finally, methanol/0.1 M aqueous NaOH (90.0:10.0, v/v) with higher elution ability was applied, and 6.0 mL of it can remove all the adsorbed AAI from the sorbent (Figure 6 (F)). The result proved that AAI was successfully separated from interferences and 0.02 mg of AAI can be removed per gram of Fibraurea Recisa Pierre.

Validation
The standard solution was diluted in methanol to obtain a series of solution with nine concentration levels 1.2, 15.0, 30.0, 45.0, 60.0, 75.0, 85.0, 100.0, 132.0 mg/L. The calibration curves were constructed using the chromatographic peak areas with three injection at each level. Good linearity of the calibration curves was obtained, and the correlation equation was y = 0.5x − 0.7 (R 2 = 0.98) (y is the peak area and x is the concentration in solution). Based on signal-to-noise ratios of 3 (LOD) and 10 (LOQ), the LOD and LOQ were 0.02 mg/L and 0.09 mg/L, respectively.
Then, a certain concentration was spiked into Fibraurea Recisa Pierre extracted solution and analyzed to calculate the recoveries of SPE [33]. Moreover, the intra-day and the inter-day relative standard deviations (RSDs) were calculated by repeating the SPE process with exactly same conditions 5 times in the same day and injecting for 5 consecutive days. Good recoveries of 96.2-100.0% and RSDs of 3.5-4.0% were obtained ( Table 1). The satisfactory recoveries and RSD implied that this method was accurate and precise for analyzing AAI. To evaluate the reusability of ZIF-67@EIM-MIM, it was reused in SPE process, and after recycling seven times the recoveries of AAI were larger than 63.2%.
Additionally, ZIF-67@EIM-MIM was synthesized under the same conditions every two days and repeated for 5 times. The adsorption capacity (under the optimized adsorption condition: 25 °C with 120 min) and SPE process were repeated on each obtained

Validation
The standard solution was diluted in methanol to obtain a series of solution with nine concentration levels 1.2, 15.0, 30.0, 45.0, 60.0, 75.0, 85.0, 100.0, 132.0 mg/L. The calibration curves were constructed using the chromatographic peak areas with three injection at each level. Good linearity of the calibration curves was obtained, and the correlation equation was y = 0.5x − 0.7 (R 2 = 0.98) (y is the peak area and x is the concentration in solution). Based on signal-to-noise ratios of 3 (LOD) and 10 (LOQ), the LOD and LOQ were 0.02 mg/L and 0.09 mg/L, respectively.
Then, a certain concentration was spiked into Fibraurea Recisa Pierre extracted solution and analyzed to calculate the recoveries of SPE [33]. Moreover, the intra-day and the interday relative standard deviations (RSDs) were calculated by repeating the SPE process with exactly same conditions 5 times in the same day and injecting for 5 consecutive days. Good recoveries of 96.2-100.0% and RSDs of 3.5-4.0% were obtained ( Table 1). The satisfactory recoveries and RSD implied that this method was accurate and precise for analyzing AAI. To evaluate the reusability of ZIF-67@EIM-MIM, it was reused in SPE process, and after recycling seven times the recoveries of AAI were larger than 63.2%. Additionally, ZIF-67@EIM-MIM was synthesized under the same conditions every two days and repeated for 5 times. The adsorption capacity (under the optimized adsorption condition: 25 • C with 120 min) and SPE process were repeated on each obtained sorbent. The rate of reproducibility is between 92.4-98.8% and the results revealed that the sorbent was reliable for the extraction sample.
The performance of proposed sorbent was compared with previous studies in Table 2. Original ZIF-67 was applied by Lin et al. to adsorb malachite green with a capacity of 240.0 mg/g [11]. Du et al. used the original ZIF-67 to adsorb 21 organic dyes, according to different size and functional groups, the capacities were in the range of 4.3-800.0 mg/g [34]. Liu et al. used sulfonated ZIF-67 to adsorb 9 organic dyes, the maximum capacity increased to 1250.0 mg/g [35]. Wei et al. immobilized ZIF-67 using 1-hexyl-3-methyl imidazolium chloride ionic liquid modified magnetic Fe 3 O 4 (IL-Fe 3 O 4 ) nanoparticles and hydroxylated multiwall carbon nanotubes (MWCNTs-OH), and the sorbent ZIF-67@MWCNTs-OH@IL-Fe 3 O 4 was applied in α-chymotrypsin adsorption with capacity of 637.7 mg/g [36]. The results of two immobilized ZIF-67 sorbents approved that the capacity of immobilized ZIF-67 increased efficiently. To separate AAI, several composite sorbents were investigated. Molecular imprinted polymer (MIP) was a common sorbent and Xiao et al. applied the MIP to remove AAI from a plant extract; however, the capacity was only 1.3 mg/g [37]. Silica, as a stable substrate, was immobilized by various functional groups. Wang et al. developed a dummy molecularly imprinted silica (MIS) with a capacity of 8.1 mg/g [38]. Fang et al. immobilized silica with imidazolium chloride-ethylimidazolium chloride (Sil@IM-BIM), and the sorbent was obtained a capacity of 16.7 mg/g [24]. When magnetic carbon nanotubes with adenine was used to immobilize silica, the capacity of obtained sorbent SiO2-A@CNT/Fe3O4 increased to 24.5 mg/g [39]. UIO-66, as another MOF material, was composited with Nmethylolacrylamide (NMA) by Zhang et al. and it was used to separate AAI as a monolith sorbent with a capacity of 35.4 mg/g [33]. Shu et al. prepared a chitosan modified carbon microcoils (CMCs@CS), the sorbent exhibited a good binding ability (77.7 mg/g) and satisfactory selectivity [40]. Overall, the sorbent developed in the present study exhibited higher adsorption capacity.

Conclusions
In this study, three dual ILs-immobilized ZIF-67 sorbents were prepared and applied in SPE to remove AAI from Fibraurea Recisa Pierre which was used as a herbal medicine to treat cough. Using the analysis of adsorption models, the adsorption capacities of immobilized sorbents were efficiently increased, and the highest amount of 50.9 mg/g was obtained on ZIF-67@EIM-MIM at 25 • C within 120 min among the three sorbents. When ZIF-67@EIM-MIM was applied in SPE, 6.0 mL, 4.0 mL and 6.0 mL of hexane, water, and methanol can wash all interferences and 6.0 mL of methanol/0.1 M aqueous NaOH (90.0:10.0, v/v) can remove all adsorbed AAI. The provided experimental data revealed that ZIF-67@EIM-MIM was a potential sorbent to remove AAI from herbal plant, and it could be an ideal material for the pretreatment of herbal plant extracted solution containing AAI.