Ionic liquid-based dispersive liquid-liquid microextraction combined with functionalized magnetic nanoparticle solid-phase extraction for determination of industrial dyes in water

N-butyl pyridinium bis((trifluoromethyl)sulfonyl)imide ([Hpy]NTf2) functionalized core/shell magnetic nanoparticles (MNPs, Fe3O4@SiO2@[Hpy]NTf2)) were prepared and applied as an adsorbent for magnetic solid phase extraction (MSPE) of three commonly used industrial dyes including malachite green, crystal violet and methylene blue. Extraction solution was mixed with 100 mg extraction material of Fe3O4@SiO2@[Hpy]NTf2, and 1 mL of acetonitrile was used to elute target analytes for further extraction and purification. [Hpy]NTf2 was used as extraction solution, and 500 μL methanol was selected as dispersive solvent in ionic liquid (IL) dispersive liquid–liquid microextraction (DLLME) method. After sonication for 5 min and centrifugation at 447 g for 10 min, 20 μL of sedimented phase was injected into HPLC-UV system. The limit of detection (LOD) and limit of quantification (LOQ) of current method were 0.03 and 0.16 μg·L−1, respectively, which indicated the sensitivity was comparable or even superior to other reported methods. The relative recoveries of the target analytes ranged from 86.1% to 100.3% with relative standard deviations between 0.3% and 4.5%. The developed method has been successfully applied to determine the level of three industrial dyes in different water samples.

physical properties, such as high polarity and thermal stability, low volatility and wonderful miscibility. Recently there are several novel sorbents using ionic liquids coated magnetic nanoparticles have been applied for trace enrichment of drug residues in water samples 14,15 . DLLME is a microextraction technique applying appropriate extraction solvent and dispersant for the microextraction of analytes in aqueous sample with merits of rapidity, low cost, high recovery and simplicity of operation. However, there are some limitations when using DLLME such as highly toxic extractive and dispersive solvents. To overcome these difficulties, the usage of ILs has been applied as green solvent to replace the conventional organic solvents to extract pollutants which has shown great interests. There are several reports using ionic liquid based DLLME for trace enrichment of organic chemicals in foods, biological, and water samples [16][17][18] . In our previous investigations, we developed the method which combined SPE with DLLME (SPE-DLLME) as an efficient hyphenated technique so as to successfully improve the selectivity and sensitivity in pretreatment process 13,19,20 .
[Hpy]NTf 2 has been applied as extraction solution for further purification in IL-DLLME method. To the best of our knowledge, this is the first report about preparation of ionic liquid based magnetic nanoparticles (MNPs), and their applications of SPE sorbents on simultaneous adsorption of MG, CV and MB in water.
The drinking water was obtained from bottled water in Shenyang and samples of running water were taken from the tap in the laboratory. River water (RW) samples were collected from the South Canal of Shenyang. Influent (IWW) and effluent (EWW) wastewaters were collected from the urban wastewater treatment plant of Shenyang. All samples were collected in December 2015 and filtered by 0.45 μm nylon membrane and stored in amber glass bottles at −20 °C before analysis.
Instrumentations and analytical conditions. An LC-20AT liquid chromatography equipped with a UV detector (Shimadzu Co., Ltd., Kyoto, Japan) was used for the chromatographic analysis and the detection wavelength was 615 nm. Separations were performed on a Kromasil C 18 column (250 mm × 4.6 mm, 5 μm) and the column temperature was maintained at 30 °C. A good chromatographic separation was implemented with a mobile phase consisting of acetonitrile and 0.1% formic acid in water solution (60:40, v/v) at a flow rate of 1 mL·min −1 . An X'Pert PRO X-ray diffractometer (PANalytical Co., Netherlands), an HCT-1 TA instruments (Beijing Hengjiu Scientific Instrument Factory, Beijing, China), an SSX-550 SEM instruments (Shimadzu Co., Ltd., Kyoto, Japan) and a Waters 2414 Infrared spectrometer (Waters Corp., Milford, MA, USA) were conducted for characterization of the materials. A KQ5200DE ultrasonic cleaner bath (Kunshan ultrasonic instrument co., Ltd., Kunshan, China) and JJ-1 electric mixer (Jiangsu Ronghua instrument manufacturing co., Ltd., Jiangsu, China) were used for synthesis of the materials. A pH meter (PHS-3CF, Shanghai, China) was employed for pH adjustment. A vortex mixer (XW-80A, Jiangsu, China) was used for thoroughly mixing the solution. added to a round-bottom flask which contained 50 mL ethanol and 15 mL pure water. The mixture was sonicated for 15 min to ensure homogeneous mixture. After adjusting the pH to 9 using ammonia, 2 mL of TEOs was added dropwise to the flask. The mixture was mechanically stirred under nitrogen protection for 24 h to perform the silica coating. The produced Fe 3 O 4 @SiO 2 was separated and washed with deionized water, to which was added 1 mL HCl solution (1 mol·L −1 ) and mechanically stirred for 12 h at room temperature. The MNPs were collected by a magnet, washed with water, and vacuum-dried at 60 °C for 12 h. Extraction procedure. One hundred milligram of Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 was added to 100 mL of water sample (pH was adjusted to 4.0 by using HCl) in a conical flask. The mixture was shaken for 20 min and an external magnet was applied to isolate the material from the water. 1 mL of acetonitrile was used to elute target analytes from the material by shaking for 20 min. The elute solvent was transferred to a 10-mL glass conical tube after being separated by a magnet. The elution was evaporated to dry on a rotary vacuum evaporator at 35 °C, and 5 mL of pure water (adjusted pH to 4.0 using HCl) was added to the test tube to reconstitute the residue. A mixture of 70 μL of [Hpy]NTf 2 and 500 μL of methanol was rapidly injected into the aqueous solution to form a cloudy solution, which was sonicated for 5 min to increase the extraction efficiency. The mixture was centrifuged at 447 g for 10 min and the dispersed fine particles of the extraction phase were deposited at the bottom of the test tube. The supernatant was removed using a microsyringe. The remaining sedimented phase was diluted by mobile phase solution (acetonitrile/water, v/v, 60/40), 20 μL of which was injected into HPLC system for analysis.
As shown in Fig In this study, TGA was conducted in a nitrogen atmosphere, and the temperature was increased from 25 to 1000 °C at a rate of 5 °C·min −1 . As can be seen in Fig. 1B, an additional weight loss of 8% after heating to 150 °C, corresponded to water and organic solvent content. Besides, an obvious weight loss of 25% was observed from 400 to 600 °C due to the decomposition of the ILs, which was consistent with the decomposition temperature of Bis(trifluoromethane sulfonimide) ILs 21 . This observation indicated that [Hpy]NTf 2 was located on the surface of Fe 3 O 4 @SiO 2 . There was no significant weight loss when the temperature increasing up to 500 °C, which indicated the yield of final products was about 70%. Figure 2A showed   Scanning electron microscope (SEM) produces microscopic images of sample surface material. As shown in Fig. 2B, these are SEM pictures of Fe 3 O 4 , Fe 3 O 4 @SiO 2 , and Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 . SEM spectrum of Fe 3 O 4 shows small particle size, spherical shape, no accumulation, dispersed particles, and 100 nm-pore size, which are consistent with the parameters in label of commercial product. The particle size of Fe 3 O 4 @SiO 2 is about 200 nm, which is larger than that of Fe 3 O 4 . Most of the particles are spherical, but the size is large which may due to poor grinding and dispersion forming during synthesis process. The particle size of Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 is between 200~800 nm, but aggregated which affects the particle size. It is due to high viscosity of ionic liquid, which can be further demonstrated that the synthesized material is Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 .
Optimization of chemical synthesis. Chemical synthesis of Fe 3 O 4 @SiO 2 . When large amount of TEOs was added to the reaction system, most of the synthesized SiO 2 would disperse into the solution. This was because of limited binding sites on the surface of Fe 3 O 4 , whereas extra TEOs would lead to more SiO 2 into the free state. The amount of TEOs was also optimized and results indicated that 2 mL of TEOs was the most suitable volume which was applied in the further investigation. Besides, pH showed great influence on the productivity. When the pH was below 9, the final productivity was extremely low, however when pH was adjusted to 9-10, productivity was improved from 40% to 70%. This indicated the reaction should be proceeded under basic conditions, because hydroxyl group (-OH) group could provide binding sites for SiO 2 . Insufficient basicity could interfere the number of binding sites, whereas the basicity showed no influence on the binding sites when it was saturated. When the temperature was below 25 °C, ILs could not be efficiently attached on the surface of Fe 3 O 4 @SiO 2 . However, productivity will not increase above 30 °C, with the possible reasons of molecular thermal movement; therefore, 25 °C was selected as the reaction temperature.
Optimization of MSPE. The effects of several parameters on extraction efficiency by MSPE were studied, including pH, sorbent amount, extraction time, type of desorptive solvent and volume. The influence of all these parameters was evaluated in terms of peak areas.
Effect of pH. The sample pH is a crucial factor in the extraction process which could affect the existing state of the targets and extraction efficiency. Considering the pKa values of 6.9, 0.8 and 4.5 for MG, CV and MB, respectively, the water samples should be acidic, so as to make the targets to be presented in molecular form in water samples. In this work, the effect of pH was investigated over the pH range of 2.0-7.0. As shown in Fig. 3A, the recoveries of three targets increased and reached a maximum value at a pH of 4.0, and subsequently decreased, which indicated that three targets were successfully extracted to the IL sediment phase. Therefore, pH 4.0 was selected for the subsequent assays.
Effect of amount. The effect of amount of Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 MNPs was investigated with the amount varying from 70 to 120 mg. Figure 3B demonstrated that the peak areas of almost all the 3 dyes increased continuously with increasing the sorbent amount from 80 to 100 mg. Further increasing the sorbent amount showed no obvious increasing in the peak area of the targets. These results indicated that 100 mg of sorbent was sufficient to extract dyes in water, and therefore, 100 mg of sorbent was used for the following experiments.

Effect of extraction time.
To evaluate the effect of extraction time on the recoveries of targets, different extraction time ranging from 5 to 30 min were tested (Fig. 3C). The recoveries gradually increased up to 20 min, and slightly increased when extraction time was increased to 30 min. Thus, 20 min was selected as the optimal extraction time.
Effect of the type of elution and its volume. An appropriate type of elution solvent is important for perfect extraction efficiency by Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 MNPs. Four types of solvents were selected as desorptive solvent, including methanol, ethanol, acetonitrile and acetone. As shown in Fig. 3D, acetonitrile yields higher recovery than that of methanol, ethanol, and acetone. The effect of sample volume on extraction efficiency of targeted dyes was investigated with the volume changing from 700 to 1100 μL. As shown in Fig. 3E, the recoveries of all these three dyes declined with increasing the sample volume from 1000 to 1100 μL. This is because when increasing the sample volume, especially for 700-1000 μL, the magnetic sorbents would be dispersed more widely and sorbents lost as long as volume increased. Then, 1000 μL was selected as sample volume.
Based on the above discussions, the optimal conditions for enrichment of targets were determined as follows: Optimization of DLLME. Effect of extraction solvent volume. In order to evaluate the influence of extraction solvent volume on extraction efficiencies of targets, different volumes (40, 50, 60, 70, 80, 90 μL) of [Hpy]NTf 2 were tested using the same MSPE-DLLME procedure. As shown in Fig. 4(A), the recoveries of targets increased with increasing the volume of [Hpy]NTf 2 from 40 to 70 μL, but above 70 μL, the recoveries remained a constant level. Considering the enrichment factor decreased when the IL solvent volume was increased. Thus, 70 μL of [Hpy]NTf 2 was selected in the following studies, since the higher recoveries were obtained and the EFs were acceptable.
Effect of dispersive solvent and its volume. The dispersive solvent should efficiently disperse the extraction solvent into the aqueous sample to increase the contact area. The main point to be considered in selecting dispersive solvent is its miscibility in both organic and aqueous phases. Methanol, ethanol, acetone and acetonitrile were widely used as dispersive solvents. Based on Fig. 4(B) we concluded that methanol gave the highest recoveries among the solvents. Thus, methanol was chosen as the disperser solvent.
In the meanwhile, the volume of disperser solvent affects the dispersion degree of the extraction solvent in aqueous phase, consequently affecting the extraction efficiency. Therefore, the effect of dispersive solvent volume on the extraction recovery was also investigated in the range of 0.4-1.2 mL. The obtained results (Fig. 4C) showed that the extraction recovery increased with the increasing volume of methanol up to 0.5 mL and slightly decreased as the volume increasing. Thus, 0.5 mL of acetonitrile was selected as the optimum volume. Effect of ultrasound time. The effect of ultrasonic time on the extraction efficiency was examined within the range from 2 to 6 min. As shown in Fig. 4(D), ultrasonic time of 5 min was the optimal condition and longer ultrasonic time could not obtain higher extraction efficiency. When the mixture of ionic liquid extraction solvent and dispersive solvent were injected to the aqueous sample, the fine droplets of extraction solvent provided an infinitely large contact surface area between aqueous phase and extraction solvent to speed up the mass-transfer process. It would take a short period of time for the analytes to transfer from aqueous phase to extraction phase and to reach an equilibrium state. However, the volatilization of organic solvent and consequent thermal effect of analytes would lead to the loss of analytes as the reaction time went longer. Hence, 5 min was chosen as the optimum ultrasonic time.
Effect of ionic strength. The effect of ionic strength of the sample solution was evaluated by adding NaCl (0-10%, w/v) under the previous optimum conditions. Generally, addition of moderate salt could decrease the solubility of targets in the aqueous sample, thus increasing the extraction efficiency. However, in the current investigation, we concluded that ionic strength did not have significant effect on the extraction of target dyes. Therefore, the MSPE-IL-DLLME procedure was performed without salt addition to the sample solutions.
Analytical performance of the method. Under the optimized conditions, the method was applied to determination of dyes in water samples. The analytical results can be found in Table 1. The calibration curves were established by plotting peak area against concentration. The regression coefficients were determined to be 0.998. The limit of detection (LOD) and limit of quantification (LOQ) were 0.03-0.05 μg·L −1 and 0.11-0.16 μg·L −1 , respectively. The precisions of the method ranged from 1.2% to 2.6% (n = 6) and accuracies varied between 0.3% and 4.5%, indicating that the method showed high sensitivity and reproducibility.
Analysis of real water samples. To further evaluate the method applicability, South Canal river water, tap water, drinking water and waste water samples were analyzed using established method. Accuracy of a method was  Table 1. The performance characteristics of MSPE-DLLME.
evaluated as the percentage of deviation from the known added amount of analyte in the sample. It is generally investigated by recovery experiments. Analyte recovery was calculated according to the equation as follows: where A is the concentration of analyte standards prepared in mobile phase, B is the concentration of blank (unspiked) sample, C is the concentration of analyte detected from the sample spiked before extraction.
In recovery experiments, the recovery value will not exceed 100% in theory, however the system error, operation error and other errors are inevitable in practice. So, the measured value (C-B) may be greater than the actual amount added (A), resulting in recovery values higher than 100%, which is a common phenomenon in method validation [22][23][24] . Generally, the recovery value of range of 85-115% can be acceptable, which means that the analytical error of the method is ±15%, and that is in full compliance with the requirements in analysis.
In our study, unspiked water samples and water samples spiked with 3 dyes at three concentration levels (0.4, 2 and 10 μg·L −1 ) were analyzed by using proposed method (n = 3). The results are listed in   Table 4. Comparsion of MSPE-DLLME-HPLC/UV with other methods for the determination of industrial dyes.
and MB were detected in river water and influent wastewater samples (Table 3). These results indicated that the established method could be successfully applied to analysis of target dyes at trace level in real samples.

Comparison of MSPE-IL-DLLME method to previous methods. The extraction effects of MG, CV
and MB were significantly enhanced by the proposed MSPE-IL-DLLME coupled with HPLC-UV method. Furthermore, the current protocol can be evaluated by comparing with previously reported work from the viewpoints of extraction solvent, LODs, enrichment factor (EF), and recoveries etc. The extraction solvent using in DLLME is [Hpy]NTf 2 , which is more environmental and provides new insight in extraction techniques. According to the Table 4, the established method provides relatively more sensitive, higher EF and better recoveries than previous reports 3, [25][26][27][28] . These merits make our method as a reliable tool in monitoring MG, CV and MB pollution in water samples.

Conclusions
A novel sorbent based [Hpy]NTf 2 functionalized MNPs was synthesized for the pretreatment through MSPE combined DLLME for the enrichment of 3 dyes in water. The obtained Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 adsorbents were characterized via FTIR, TGA, XRD and SEM. Under the optimized conditions, the method provided suitable analytical characteristics including extraction efficiency up to 86%, enrichment factor over 480, and LOD between 0.03-0.05 μg·L −1 . Based on the results, we can concluded that Fe 3 O 4 @SiO 2 @[Hpy]NTf 2 MNPs could be introduced as a novel sorbent for rapid extraction of dyes from aqueous solution.