Platform construction and extraction mechanism study of magnetic mixed hemimicelles solid-phase extraction

Simple, accurate and high-throughput pretreatment method would facilitate large-scale studies of trace analysis in complex samples. Magnetic mixed hemimicelles solid-phase extraction has the power to become a key pretreatment method in biological, environmental and clinical research. However, lacking of experimental predictability and unsharpness of extraction mechanism limit the development of this promising method. Herein, this work tries to establish theoretical-based experimental designs for extraction of trace analytes from complex samples using magnetic mixed hemimicelles solid-phase extraction. We selected three categories and six sub-types of compounds for systematic comparative study of extraction mechanism, and comprehensively illustrated the roles of different force (hydrophobic interaction, π-π stacking interactions, hydrogen-bonding interaction, electrostatic interaction) for the first time. What’s more, the application guidelines for supporting materials, surfactants and sample matrix were also summarized. The extraction mechanism and platform established in the study render its future promising for foreseeable and efficient pretreatment under theoretical based experimental design for trace analytes from environmental, biological and clinical samples.

In recent years, a new solid-phase extraction (SPE) method based on mixed hemimicelles assemblies (hemimicelles/admicelles) have been attracting more and more attention [1][2][3][4][5] . The employed sorbents are produced by adsorbing ionic liquid (such as 1-hexadecyl-3-methylimidazolium tetrafluoroborate) or ionic surfactants (such as sodium dodecyl sulfate and cetyltrimethylammonium bromide) on the surface of metal oxides (such as alumina, silica, titanium dioxide, and ferric oxyhydroxides). Compared with the normal sorbents, one feature of these sorbents is that the outer surface of hemimicelles is hydrophobic and the admicelles is ionic, which offers regions of distinct polarity, allowing the adsolubilization of diverse analytes. The inoic liquid or surfactant hydrocarbon chains provide hydrophobic or π -π stacking interactions for hydrophobic analytes, while the polar groups adsorb ionic analytes via electrostatic interaction or hydrogen-bonding interaction.
What makes hemimicelle-based SPE special is its versatility (types of interactions can be changed according to the region of the isotherm), the high number of sites for solubilization, and the dynamic character of the surfactant or ionic liquid aggregates adsorbed onto the mineral oxide. Any scientist with a good basis on the chemistry involved can select the best supporting materials and surfactant for a specific application. Up to now, Various SPE methods based on mixed hemimicelles have been described in the literatures. Early in 2003, Merino 6 et al. firstly developed a simple and reliable SPE method for the identification and quantitation of benzalkonium surfactants based on their adsorption on sodium dodecylsulphate (SDS)-alumina mixed hemimicelles assemblies in sewage and river water complex matrixes, demonstrating that this approach is an interesting alternative for the concentration of amphiphiles in complex matrixes. In 2006, Cantero 7 et al. successfully used hemimicelles/admicelles of SDS-alumina and cetylpyridinium chloride (CPC)-silica as sorbent materials and filled it in the SPE cartridge columns for the concentration of alkylphenol polyethoxylate (APE) biodegradation products from wastewater and river water samples. Niu 3 et al. adopted the cationic surfactant cetyltrimethylammonium bromide (CTAB)-titanate nanotube system to form hemimicelles/admicelles for extracting nonpolar phthalate Scientific RepoRts | 6:38106 | DOI: 10.1038/srep38106 tetrafluoroborate (C 16 mimBF 4 ), 1-dodecyl-3-imidazole tetrafluoroborate (C 12 mimBF 4 ), 1-octyl-3-methylimidazolium tetrafluoroborate (C 8 mimBF 4 ), 1-butyl-3-methylimidazolium tetrafluoroborate (C 4 mimBF 4 )), and N-cetylpyridinium bromide were purchased from CAS Lanzhou Institute of Chemical Physics, and their chemical structures are as shown in Table 4. Ferric chloride hexahydrate (FeCl 3 .6H 2 O), ethylene glycol (EG), diethylene glycol (DEG), sodium acrylate (CH 2 =CHCOONa), sodium acetate (CH 3 COONa), disodium hydrogen phosphate (NaH 2 PO 4 ) and phosphoric acid sodium dihydrogen phosphate (Na 2 HPO 4 ) were purchased from Aladdin.
Instruments. The morphology of the as-synthesized nanoparticles was studied by using a FEI TECNAI G2F20 transmission electron microscope (TEM). The surface groups on the as-synthesized nanoparticles were measured with a 8400 s FT-IR spectrometer (Shimadzu Corporation, Japan) by using KBr as diluting agent. The magnetic properties were studied by using a LDJ 9600-1 vibrating sample magnetometer (VSM) operating at room temperature with applied fields up to 10 kOe. Zeta-potential measurements of as-synthesized nanoparticles were performed with ZetaPlus Zeta Potential Analyzer (Brookhaven, USA). Fe 3 O 4 NPs were prepared by solvothermal method 19 . Briefly, 2.4 g FeCl 3 ·6H 2 O, 3.4 g CH 2 CHCOONa and 3.4 g CH 3 COONa were added into a mixture of ethylene glycol (EG, 22.5 mL) and diethylene glycol (DEG, 22.5 mL) under ultrasonication for about 1 h. The resulting homogeneous black solution was then transferred and sealed into a teflon-lined stainless-steel autoclave. The autoclave was heated at 200 °C for 10 h, and then cooled to room temperature. After the reaction, the obtained Fe 3 O 4 NPs were washed with ethanol and water for several times, and then dried in vacuum at 60 °C for 10 h. Fig. 1, the MSPE procedure based on mixed hemimicelles using Fe 3 O 4 NPs allowed rapid preconcentration of analytes in complex aqueous systems by a simple magnetic extraction method. First, a certain amount of Fe 3 O 4 NPs, C 16 mimBr and 2 mL of phosphate buffer were added sequentially into a 4 mL centrifuge tube. The mixture was sonicated for 5 min to suspend the Fe 3 O 4 NPs to form mixed hemimicelles, then 2 mL aqueous sample was added in the mixed solution. In order to absorb analytes, the mixture was sonicated for 5 min, and then equilibrated for 90 min in an oscillator. In the external magnetic field, the magnetic nanomaterial with analytes was adsorbed onto the centrifuge tube wall. When the suspension became limpid (about 2 s later), the supernatant was decanted, and the isolated nanoparticles were eluted with 1 mL methanol containing 1% acetic acid (HAc) (0.5 mL every time and washed twice) to desorb the preconcentrated target analytes. The eluted solution was collected and then dried under a stream of nitrogen at 60 °C, and redissolved in 500 μ L methanol. The methanol solution was vibrated in an ultrasonic bath for 2 min in order to insure that the target analytes can be dissolved into methanol completely. Finally, after filtration through 0.45 μ L membrane, 10 μ L of the solution was injected into the HPLC system for analysis.

Magnetic solid phase extraction (MSPE) based on mixed hemimicelles. As shown in
For the contrast study of methods, we extracted flavonoid compounds by magnetic retrieval of chitosan solid-phase extraction technique at the same time. Firstly, 1 mL sample was added into a 5 mL vessel, the pH of which was adjusted to 7.0 by using 0.02 M PBS buffered solution. Secondly, 15 mg chitosan was added into the above solution and was oscillated slightly for 30 s. And then a certain amount of Fe 3 O 4 NPs was added, the mixture was sonicated for 1 min. In the external magnetic field, the chitosan Fe 3 O 4 composites adsorbed with analytes was isolated from solution. After 2 s the suspension became limpid, the supernatant was decanted, and the collected nanoparticles were eluted with 1 mL acetonitrile containing 1% HAc to desorb composite. Finally, the eluted solution was collected and then dried under a stream of nitrogen at 60 °C, and redissolved in 500 μ L methanol. The methanol solution was vibrated in an ultrasonic bath for 2 min in order to insure that the target analytes can be dissolved into methanol completely. After filtration through 0.45 μ L membrane, 10 μ L of the solution was injected into the HPLC system for analysis.    Table 5 (similarly hereinafter). Three chlorophenols (4-chlorophenol, 2,4-dichlorophenol and 2,4,6-trichlorophenol) were quantified by HPLC-DAD analysis and the mobile phase was methanol-0.1% formic acid (25: 75, V/V). Three fluoroquinolones (ofloxacin, ciprofloxacin, gatifloxacin) were quantified by high-performance liquid chromatography with diode array detector (HPLC-DAD) and the mobile phase was methanol -1% aqueous solution of triethylamine (adjusted with phosphoric acid to pH 4.3) (25: 75, V/V). Three cephalosporins (ceftazidime, cefotaxime, cephradine) were quantified by HPLC-DAD analysis and the mobile phase was methanol-0.1% formic acid (25:75, V/V). Vitamin A acetate was quantified by HPLC-DAD analysis and the mobile phase was methanol -water (96: 4, V/V). Myristic acid was quantified by HPLC-RID quantitative analysis. HPLC system (Shimadzu LC-20AT), comprising a quaternary pump, column oven, autosampler and quaternary detectors, in which columns produced by Agilent ZORBAX Eclipse XDB-C 18 column (150 mm × 4.6 mm, 5 μ m). The mobile phase was methanol -water (90:10, V/V).

Results and Discussion
Characterization. The TEM image and SEM image of the morphology of the Fe 3 O 4 NPs were shown in Fig. 2. It can be seen that most of the particles were uniform with a spherical shape from different angles as reported before. As shown in Fig. 2(A), the TEM image suggested that the size of the particle was about 170 nm in diameter. In the present experiment, it can be uniformly suspended in solution during the extraction process, and therefore suitable for magnetic mixed hemimicelles solid phase extraction. For the metal oxide, the isoelectric point measured by zeta-potential is an important property 20 . The study shows that the charge density and the type of metal oxide will change with the pH of the buffer solution. When the pH is below the isoelectric point, Fe 3 O 4 NPs surface is positively charged, otherwise negatively charged. As shown in Fig. 3, the isoelectric point (IEP) of Fe 3 O 4 NPs was found to be at pH 3.47, which approached to previously reported data for natural silica (3.0) 12 . When the pH is 6, the Fe 3 O 4 which is negatively charged produced the electrostatic adsorption with ionic liquid to form the magnetic mixed hemimicelles SPE system. Figure 4 showed VSM magnetization curves of Fe 3 O 4 NPs at room temperature. The maximum saturation magnetization was 64.60 emu g −1 for Fe 3 O 4 NPs which indicated their magnetism was sufficient for MSPE. Figure 4 also demonstrated that the prepared Fe 3 O 4 NPs could be separated from their dispersion quickly once an external magnetic field was applied.
For demonstrating C 16    Zeta-potential isotherm. The adsorption isotherms of C 16 mimBr on Fe 3 O 4 NPs are useful for not only understanding the principle of mixed hemimicelles but also selecting the optimal surfactant amount, which can be demonstrated by zeta-potential isotherm 8,21 . Figure 6 presents the zeta-potential change of Fe 3 O 4 NPs suspensions when C 16 mimBr is absorbed at pH 6.0. Generally, the isotherms can be divided into three regions including hemimicelles, mixed hemimicelles and admicelles. And the process involves four adsorption mechanisms such as hydrophobic interaction, electrostatic interaction, hydrogen-bonding interaction and π -π stacking interactions. As shown in Fig. 6(D), in the first region (C 16 mimBr added from 0 to 0.1 mg), due to electrostatic attraction, single-layer coverage (hemimicelles) is gradually formed, while the added ionic surfactant molecules are adsorbed onto the surface of negatively charged Fe 3 O 4 NPs under pH 6.0 (above the IEP). The zeta-potential of mineral oxides increases from negative to zero. In the second region, the formation of bilayers admicelles involves hydrophobic and electrostatic interactions. With the increasing amount of ionic surfactant molecules, the zeta-potential of mineral oxides reverses and increases after passing through a zero zeta-potential. Both the above-mentioned regions are suitable for SPE method. When the amount of C 16 mimBr is above 0.8 mg, the zeta-potential of mineral oxides remains constant, manifesting that the entire surface of metal oxides is saturated and surfactants begin to form micelles in solution. In this region, aqueous surfactant micelles could be formed in the aqueous solution and be in equilibrium with admicelles, which may result in the solubilization of analytes into micelles in bulk solution again and cause the isotherm region span unsuitable for SPE. However, it should be pointed out that the structure of the surfactant aggregates could not be simply defined as monolayers (hemimicelles) or bilayers (admicelles), which need further evidences for the formation of discrete surface aggregates.  As shown in Fig. 6(B,C,D,F and G), the trend of the formation of the mixed hemimicelles is basically similar, the longer the hydrophobic chain is, the less ionic liquid is needed. From the Fig. 6(A), it can be seen that the zeta potential of Fe 3 O 4 NPs cannot be affected by the amount of the C 4 mimBF 4 , which might be explained by the fact that the C 4 mimBF 4 has poor affinity with Fe 3 O 4 NPs and are more difficult to form micelles because of short carbon chain. So they are not appropriate for magnetic mixed hemimicelles solid phase extraction.

Study of extraction mechanism.
In recent years, regarding to the extraction mechanism of the magnetic mixed hemimicelles SPE, there are sporadic speculation including hydrophobic interaction, electrostatic interaction, hydrogen-bonding interaction and π -π stacking interactions. In order to fully elaborate extraction mechanism, we carried out a systematic investigation by using the following three categories and six sub-types of compounds.
Research on hydrophobic compounds containing aromatic ring (flavonoids and chlorophenol compounds). In this work, two sub-types of compounds consisting of five flavonoids and there chlorophenol were examined. As shown in Table 1, these two sub-types of compounds both have high hydrophobicity constant (log P), and also have a gap between five flavonoids and three chlorophenol compounds in the hydrophobicity constant respectively. So the hydrophobic interaction can be investigated in the magnetic mixed hemimicelles solid phase extraction. More importantly, though flavonoids and chlorophenol compounds all have relatively high hydrophobicity constant, there are some differences in their structures. Compared to the chlorophenol compounds, the flavonoids posses a longer aromatic chain, by which the π -π stacking interactions on the magnetic mixed hemimicelles solid phase extraction can be obtained. In the study, the C 16 mimBr coated on the surface of the magnetic Fe 3 O 4 NPs was chosen as solid phase adsorbents, and five flavonoids and three chlorophenol compounds were studied and compared.
As shown in Fig. 7(A), in the absence of C 16 mimBr, the flavonoids were hardly adsorbed onto the surface of Fe 3 O 4 NPs, because they only exist weaker hydrogen-bonding interaction and electrostatic repulsion at present. In contrast, with the increasing amount of C 16 mimBr, the adsorption ratio of five flavonoids increased remarkably. In the first region (0-0.1 mg of C 16 minBr, hemimicelles) the sorbent offer four types of interactions: hydrophobic  interactions from the hydrocarbon chain, π -π stacking interactions from aromatic ring, hydrogen-bonding interaction and electrostatic repulsion from the uncovered groups in the mineral oxide surface. In the second region (below CMC, mixed hemimicelles) the sorbent offer three types of interactions: hydrophobic interactions from the hydrocarbon chain, π -π stacking interactions from aromatic ring and electrostatic attraction between C 16 minBr and flavonoids. In the third region (above CMC, admicelles), the recovery of flavonoids decreased due to the formation of micelles in the bulk aqueous solution, which led to a redistribution of the flavonoids into the solution. It is worth noting that when the concentration of C 16 mimBr is above CMC, the recovery of the apigenin and chrysin is relatively lower than the other three flavonoids compounds. It can be explained that the more  hydrophobic apigenin and chrysin is, the more prone to interact with the redundant C 16 mimBr which is in the aqueous solution.
For three chlorophenol compounds, the trend of the recoveries is similar to the flavonoids. The recoveries increased with increasing C 16 mimBr, thus reaching a maximum level when 0.4 mg C 16 mimBr coats on the surface of the 8 mg Fe 3 O 4 NPs. However, the recovery of 4-chlorophenol always maintained unsatisfactory even when increasing the amount of the C 16 mimBr. Under optimum conditions, the maximum adsorption rate of 2,4,6-trichlorophenol, 2,4-dichloro phenol, 4-chlorophenol can reach to 90%, 70% and 20%, respectively. Through literature review and analysis 14 , we found that Log P of 2,4,6-trichlorophenol, 2,4-dichlorophenol, 4-chlorophenol were 3.48, 2.88, 2.27 respectively. Therefore, it was speculated that with the amount of chlorine atom decreasing, the log P of chlorophenol compounds decreased and the hydrophobic properties of the chlorophenol compounds decreased gradually, thus the effect between compounds and mixed hemimicelles weakened by degrees, leading to the result that the recovery of 4-chlorophenol was below than that of 2,4,6-trichlorophenol, 2,4-dichlorophenol.
When conducting the overall comparison between the two sub-types of the compounds, it can be seen that the recovery of flavonoids is generally higher than the chlorophenol compounds. Particularly, it is unusual that 4-chlorophenol hydrophobic constant (2.27) is higher than that of quercetin (2.16), while the extraction recovery (20%) was significantly lower than that of quercetin (90%), suggesting that the hydrophobic constant cannot fully determine the level of extraction efficiency. The hydrophobic constant (Log P) value refers to partition coefficient values in n-octanol/water two-phase system, which only considers partition coefficient in two phases, without considering the characteristics of the compound structure. By comparing their structures, the flavonoids have larger aromatic system than chlorophenol compounds, making it easier to produce π -π stacking interactions with aromatic structure of ionic liquids (such as imidazole ring). These results suggest that it is necessary to consider the impact of hydrophobic interactions and π -π stacking interactions when investigating the mechanism of the magnetic mixed hemimicelles solid phase extraction.
Research on hydrophilic compounds containing aromatic ring (quinolone and cephalosporin compounds). To examine the impact of electrostatic interactions further, two sub-types of hydrophilic compounds (including three quinolone compounds and three cephalosporins compounds) were selected. As shown in Table 2, the hydrophobic constants (log P) of the two sub-types of compounds are relatively low. The effect of the C 16 mimBr was studied by varying the amount of the C 16 mimBr from 0 to 6 mg, which was shown in the Fig. 8. In the absence of C 16 mimBr, pure Fe 3 O 4 NPs possess a certain adsorption towards the quinolone and cephalosporin compounds, because they exist weaker electrostatic attraction and hydrogen-bonding interaction at present. However, with the increasing amount of ionic liquids, the recoveries of the two kinds of compounds decreases, which may be attributed to the fact that the C 16 mimBr changed the charge properties of sorbents surface. Therefore the interactions between sorbents and analytes changed from electrostatic attraction to electrostatic repulsion, resulting in the decreases of recoveries. This result demonstrated that quinolone and cephalosporin compounds are not suitable for the magnetic mixed hemimicelles solid phase extraction. It may reveal a very useful rule that trying to analyze hydrophilic compounds by mixed hemimicelles SPE technique is not a good strategy in the future, especially when surfactant and analytes are like charged. Also it is worth noting that although the two sub-types of compounds possess a large aromatic system, they still cannot obtain high recoveries by mixed hemimicelles SPE systems. These results suggested that π -π stacking interactions only plays a supporting role, and hydrophobic interaction may play the dominant role during the extraction process.
Research on hydrophobic compounds excluding aromatic ring (retinyl acetate and myristic acid). To further verify the dominant role of hydrophobic interaction and the supporting role of π -π stacking interactions, retinol acetate and myristic acid were investigated. It can be predicted that these two compounds hold hydrophobic interaction with the mixed hemimicelles because of the great hydrophobicity constant. The difference is that the retinol acetate contains long conjugate structure while myristic acid doesn't. As shown in the Fig. 9, the retinol acetate and myristic acid were hardly adsorbed onto the surface of Fe 3 O 4 NPs in the absence of the C 16 mimBr, then the adsorption ratio increased remarkably and finally decreased as the increasing of the amount of the C 16 mimBr, which is similar to the trend of the flavonoids and chlorophenol compounds. As shown in the Table 3, the two sub-types of the compounds have good extraction efficiency due to the high hydrophobic constant. The maximum recovery of retinol acetate is slightly higher than myristic acid, which implying that the conjugation reaction plays a little role in the extraction when the recoveries could reach a high level. These results verified the dominant role of hydrophobic interaction and the supporting role of conjugation reaction.
Comparative research on two methods for flavonoids enrichment. In order to investigate the influence of hydrogen-bonding interaction, we studied two methods for flavonoids enrichment. Recently, some researchers have studied a two-step magnetic retrieval SPE technique in many matrices. Zhang et al. have applied the magnetic retrieval of chitosan solid phase extraction in green tea beverage samples 22 . Li et al. used the magnetic retrieval of ionic liquid in environmental water samples 23 . Therefore, we chose magnetic retrieval of chitosan solid-phase extraction technique and magnetic mixed hemimicelles solid-phase extraction technique for comparative research for flavonoids enrichment. It was found that the recoveries of flavonoids had obvious difference between the two methods. By using magnetic retrieval of chitosan solid-phase extraction, quercetin, luteolin and kaempferol had high extraction recoveries 24 . However, the recoveries of apigenin and chrysin were relatively low, which could only reach about 70% even under optimal extraction conditions. The main reason was that magnetic retrieval of chitosan solid-phase extraction process is mainly based on hydrogen-bonding interaction between chitosan and compounds. However, compared with other compounds, apigenin and chrysin have relatively fewer functional groups which can form hydrogen bonds. Therefore, the chitosan adsorption degrees of apigenin and chrysin were relatively low, resulting in relatively low extraction recoveries. However, by using magnetic mixed hemimicelles solid-phase extraction system, the phenomenon did not take place. The extraction recoveries of five flavonoids were very similar (Fig. 7), they could all achieve a relatively high level. Hence, the extraction mechanisms of magnetic retrieval of chitosan solid-phase extraction technique and magnetic mixed hemimicelles solid-phase extraction technique are different and hydrogen bonding interactions had no significant effect in the latter technique. This could be easily understood because magnetic mixed hemimicelles solid-phase extraction system did not contain a lot of hydrogen bonding sites. Based on this point, during the design process of magnetic mixed hemimicelles solid-phase extraction project, there is no need to consider hydrogen-bonding interaction, except the ionic liquid or ionic surfactants with a large amount of hydrogen bonding site.
Comparative research on the structure of ionic liquids. Effect of different carbon chain lengths. In this part, effect of different carbon chain lengths on the enrichment of five flavonoids was investigated based on the same parent nucleus structure and anionic species. Different chain lengths of ionic liquids (C 4 mimBF 4 , C 8 mimBF 4 , C 12 mimBF 4 and C 16 mimBF 4 ) on the surface of the Fe 3 O 4 NPs were compared. The same as above, in the mixed hemimicelles region (below CMC) the sorbent offer three types of interactions: hydrophobic interactions from the hydrocarbon chain, π -π stacking interactions from aromatic ring and electrostatic attraction between C 16 minBr and flavonoids. As shown in Fig. 10(A), with the increasing amount of C 4 mimBF 4 , the recovery of five flavonoids became lower and lower, indicating that the C 4 mimBF 4 is prone to interact with flavonoids in the aqueous solution, rather than being adsorbed onto the surface of the Fe 3 O 4 NPs. With the increase of carbon chain length, the maximum recoveries of five flavonoids gradually increased and the required amount of ionic liquid to reach the maximum recovery gradually decreased. Therefore, the ionic liquids with a longer carbon chain are more suitable for magnetic mixed hemimicelles solid phase extraction. Because the longer carbon chain of ionic liquids are, the stronger hydrophobic interaction is. These results further confirm that the hydrophobic interactions play a dominant role in the magnetic mixed hemimicelles solid phase extraction system.
Effect of different parent nucleus structure. In order to investigate the impact of the different parent nucleus structure of ionic liquids on the enrichment of flavonoids, two kinds of ionic liquids (C 16 mimBr, C 21 H 38 NBr) and surfactants cetyltrimethyl ammonium bromide (CTAB) were selected. As shown in Table 4, they have the same carbon chain length and the same anion, the only difference is their parent nucleus structure. Therefore, in the mixed hemimicelles region (below CMC) the π -π stacking interactions happened between ionic liquids and flavonoids, while CTAB didn't. As can be seen from Fig. 11(A,B), only 0.3 mg C 16 mimBr and C 21 H 38 NBr and 0.5 mg surfactant CTAB were needed to achieve high recoveries, respectively. Very interestingly, they need relatively more amount of surfactant CTAB. After calculating and considering their relative molecular weight, we found it is reasonable to believe that ionic liquids containing aromatic action are more applicable to magnetic mixed hemimicelles solid phase extraction. This is important because it confirms once again from another angle that the π -π stacking interactions plays a supporting role in the magnetic mixed hemimicelles solid phase extraction system.
Effect of different anions. In order to investigate the effect of different anions, we chose and compared two kinds of ionic liquids (C 16 mimBF 4 and C 16 mimBr) with different anions. As shown in the Fig. 12(A) and (B), the five flavonoids can achieve high extraction efficiency. It can be drawn that different anions of the ionic liquid have no significant effect to form mixed hemimicelles solid phase extraction system.  After comparing the different types of ionic liquids with different carbon chain length, parent nucleus structure and different anions, it can be concluded that the carbon chain length of the ionic liquid plays a decisive role in forming mixed hemimicelles solid phase extraction system. The longer the carbon chain is, the easier to form the mixed hemimicelles solid phase extraction system. In another word, for the same analytes, the extraction efficiency increased obviously with the increase of hydrophobic constant of ionic liquid. We also found that the ionic liquid containing aromatic action is more suitable for magnetic mixed hemimicelles solid phase extraction system, which provides a guiding role in the process of choosing the ionic liquids.
Optimization of extraction conditions. Effect of solution pH. Metallic oxide surface charge is an important factor in the formation of magnetic mixed hemimicelles solid-phase extraction adsorbents and extraction process. The pH value of the buffer plays a decisive role in the type and density of surface charge. While it will also make effects on the existing forms of analytes in the solution, thereby affecting the interaction between analytes and magnetic mixed hemimicelles solid-phase extracting agent.
As shown in Fig. 13(A), the effect of pH on enrichment of flavonoids and chlorophenol compounds was examined by varying the pH from 2.0 to 10.0. When the pH was around the IEP of Fe 3 O 4 NPs, the charge density of Fe 3 O 4 NPs surface was very low, and a small number of positive ions of C 16 mimBr were adsorbed onto the surface of Fe 3 O 4 NPs to form mixed hemimicelles. When the pH value was above the IEP, the surfaces of Fe 3 O 4 NPs became negatively charged, which led to strong electrostatic attraction between C 16 mimBr and the charged Fe 3 O 4 NPs surface. The results indicated that the maximum adsorption performance occurred at pH 6.0, but when the pH was over 6.0, the recoveries of quercetin, luteolin and kaempferol decreased, while the recoveries of apigenin and chrysin remained unchanged. This can be explained by the fact that apigenin and chrysin posses higher hydrophobic constant (log P) and stronger hydrophobic interaction. On the other hand, because  apigenin and chrysin have relatively high pKa, they still remain in molecular form when pH is high, therefore, the hydrophobic interaction with mixed micelles system is stronger, leading to higher extraction recoveries. As shown in Fig. 13(B), the recoveries of three chlorophenol compounds reached the maximum at pH 7.0. When the pH is above 7.0, the recoveries decreasd, which can be explained by the fact that the three compounds would be dissociated at higher pH conditions. Similarly, as shown in Fig. 13(C and D), the maximum adsorption of myristic acid and retinol acetate occurred at pH 6.0 and 7.0, respectively. From these results, It can be concluded that the recovery reduced when the analyte is dissociated. This is essential for researchers to quickly develop mixed hemimicelles solid phase extraction method. This result confirms once again from another perspective that hydrophobic interaction plays the dominant role in magnetic mixed hemimicelles solid phase extraction system.

Effect of kinds of magnetic supporting materials.
Compared to the micro-sized solid-phase extraction support materials, magnetic nanoparticles have a high surface area and strong magnetism, therefore they are more suitable to construct magnetic solid phase extraction platform 25,26 . The magnetic nanomaterials (such as iron oxide magnetic nanoparticles) and magnetic composite materials (such as magnetic carbon nanotubes, magnetic graphene oxide) have been systematically compared in our group. The results showed that the advantage of magnetic composites is the greater surface area, and therefore the amount of adsorbents is less 12 . But the disadvantage is the complex synthetic process and high manufacturing cost. In this experiment, in order to explore the mechanism of magnetic mixed hemimicelles, we do not hope the support materials themselves have adsorption site of hydrophobic interaction, π -π stacking interactions, electrostatic interactions and hydrogen bonding interactions with analytes. So we chose iron oxide magnetic nanoparticles as a support material which has simple preparation process and a single type of surface functional groups, establishing magnetic mixed hemimicelles solid-phase extraction platform for the study of extraction mechanism.
Effect of the amount of Fe 3 O 4 NPs. The designers should take into account that hemimicelles/mixed hemimicelles/admicelles are dynamic sorbents and the actual structure depends on the experimental conditions. If designers change the amount of nanoparticles while the amount of ionic liquid keeps constant, then the coverage of the surface and so the type of sorbent changed. Therefore, in order to optimize the amount of Fe 3 O 4 NPs, we increased both Fe 3 O 4 NPs and ionic liquid in the same proportion (20:1, mass ratio) in order to keep the type of sorbent. As shown in Fig. 14, through   sample solution and the total cost were less than 0.1 dollar. Therefore, compared with the conventional methods, less requirement on support material and surfactant is a significant advantage of magnetic mixed hemimicelles solid-phase extraction technique [31][32][33] .
Effect of ionic strength. The ionic strength influence the surface charge density. Therefore, we studied the effect of ionic strength by varying the sodium ions concentrations between 0.02 and 0.5 mol/L. With the increase of the ionic strength, recoveries of the analytes decreased. Therefore, 0.02 mol/L was selected for the further studies.
Effect of extraction time. In order to obtain satisfactory recoveries, the effect of extraction time on adsorption and magnetic separating time were investigated under the optimal conditions mentioned above. The effect of extraction time on the adsorption was examined by changing the time from 5 to 120 min. The results suggested that 10 min was sufficient for the highest extraction recoveries. However, when the extraction time was more than 10 min, recoveries of analytes decreased. This phenomenon may be attributed to the fact that with the increase of extraction time, some ionic liquid or ionic surfactants redissolved into the solution. Therefore, 10 min was selected for extraction time, which enabled these analytes to be completely adsorbed.
Effect of desorption conditions. To obtain satisfactories recovery of the above there categories of analytes, different kinds of organic solvents (methanol, methanol containing 1% HAc, methanol containing 2% HAc, acetonitrile, acetonitrile containing 1% HAc and acetonitrile containing 2% HAc) which were widely used to completely disrupt the hemimicelles and admicelles were investigated. We found that methanol containing 1% HAc had the best desorption ability. This may be attributed to the fact that the addition of acetic acid caused the analytes to exist in the molecular form which was prone to be soluble in organic solvent. Besides, when the pH value of the desorption solvent was below or around the IEP of Fe 3 O 4 NPs, charge density on Fe 3 O 4 NPs surface was low, which was favorable for the disruption of mixed hemimicelles.
Method validation. According to the above experiments, we found that the four sub-types of hydrophobic compounds including flavonoids, chlorophenol compounds, retinol acetate and myristic acid can achieve good extraction efficiency in the magnetic mixed hemimicelles solid phase extraction system. Therefore, to evaluate the accuracy and feasibility of the magnetic mixed hemimicelles SPE platform developed, the four sub-types of hydrophobic compounds were enriched and detected in complex samples.
Firstly, the magnetic mixed hemimicelles SPE was applied in the enrichment of flavonoids in the biological system. 6.0 mg Fe 3 O 4 NPs and 0.3 mg C 16 mimBr were added into urine samples, extraction time was 10 min, and methanol containing 1% HAc was used for the desorption. To evaluate the accuracy and precision of the method, urine samples spiked with the five flavonoids were analyzed. The samples were dealt with the proposed MSPE method before injection. Chromatogram was shown in Fig. 15(A). It can be seen that no interfering peaks at the elution times of analytes in urine samples were observed. The retention times of quercetin, luteolin, kaempferol, apigenin and chrysin were 7.96, 9.35, 15.17, 17.23, 26.93 min respectively. The results implied the proposed method based on C 16 mimBr-coated Fe 3 O 4 NPs was appropriate for the determination of quercetin, luteolin, kaempferol, apigenin and chrysin.
Secondly, the MSPE was applied in the enrichment of chlorophenol compounds in the environmental system. 8.0 mg Fe 3 O 4 NPs and 0.4 mg C 16 mimBr were added into environmental samples, extraction time was 10 min, and methanol containing 1% HAc was used for the desorption. To evaluate the accuracy and precision, tap water samples spiked with the 2,4-dichlorophenol and 2,4,6-trichlorophenol were analyzed. The samples were dealt with the proposed MSPE method before injection. Chromatogram was shown in Fig. 15(B). The retention times of 2,4-dichlorophenol, 2,4,6-trichlorophenol were 10.19 and 22.56 min respectively. The results implied the proposed method based on C 16 mimBr-coated Fe 3 O 4 NPs was appropriate for the determination of 2,4-dichlorophenol and 2,4, 6-trichlorophenol.
Finally, the MSPE was applied in the enrichment of retinol acetate and myristic compounds in biological systems. 6.0 mg Fe 3 O 4 NPs and 0.3 mg C 16 mimBr were added into urine samples, extraction time was 10 min, and methanol containing 1% HAc was used for the desorption. To evaluate the accuracy and precision, urine samples spiked with the retinol acetate and myristic were analyzed. The samples were dealt with the proposed MSPE method before injection. Chromatogram was shown in Fig. 15(C,D). It can be seen that no interfering peaks at the elution times of analytes in urine samples were observed. The retention times of retinol acetate and myristic were 3.93 and 7.45 min respectively. The results implied the proposed method based on C 16 mimBr-coated Fe 3 O 4 NPs was appropriate for the determination of retinol acetate and myristic in biological systems.
Linearity, LOD and LOQ. As is shown in Table 6, under the optimized conditions, the quantitative parameters of the proposed method, including linear range, correlation coefficients were evaluated. The correlation coefficient (R 2 ) ranged from 0.9912 to 0.9992. The LOD and LOQ were considered as the analyte minimum concentrations that can be confidently identified and quantified by the method. The LOD calculated on the basis of signal-to-noise ratio of 3 (S/N = 3) for luteolin can be achieved to 0.5 ng mL −1 .
Precision. The precision were evaluated by applying the proposed magnetic mixed hemimicelles SPE method to six replicate spiked samples at three different concentration levels (low, middle and high quantification concentrations). The precision results were shown in Table 7. These results demonstrated that the magnetic mixed hemimicelles SPE method had good precision. Recovery. The recovery of the method was assessed by using the standard addition method. Table 8 showed the recoveries that were calculated after spiking three different concentration levels (low, middle and high quantification concentrations), respectively. The recoveries were expressed as the mean value of three independent determinations. Recoveries were in the range of 80.1-103.7%, with the relative standard deviations of the recoveries varying between 2.0% and 8.7%. These results demonstrated the method was suitable for analyzing the four sub-types of hydrophobic compounds in complex samples.
The extraction mechanism of magnetic mixed hemimicelles SPE system. This work developed a series of simple, cheap and accurate extraction methods for the determination of trace amounts of compounds in complex samples. Most importantly, we hope to, through our systematic study, firstly illustrate the extraction mechanism of magnetic mixed hemimicelles SPE system. The results were summed up as follows: In order to study the influence of hydrophobic interaction, four sub-types of hydrophobic compounds including flavonoids, chlorophenols, retinol acetate and myristic acid were selected and examined. The results demonstrated that the hydrophobic interaction plays a dominate role in the magnetic mixed hemimicelles solid phase extraction. For different analytes, the extraction efficiency increased obviously with the increase of hydrophobic constant of analytes. For the same analytes, the extraction efficiency increased obviously with the increase of  hydrophobic constant of ionic liquid (C 16 mimBr, C 12 mimCl, C 12 mimBF 4 ) or ionic surfactants (such as sodium dodecyl sulfate or cetyltrimethylammonium bromide). In order to study the effect of π -π stacking interactions, two sub-types of hydrophobic compounds consisting of flavonoids and chlorophenols were compared, the extraction efficiency increased with the larger conjugation structure, which indicated that π -π stacking interactions plays a role in the extraction process. However, it is worth noting that although quinolone and cephalosporin compounds possess a large aromatic system, they still cannot obtain high recoveries by mixed hemimicelles SPE systems. These results suggested that π -π stacking interactions only plays a supporting role.
In order to study the effect of hydrogen-bonding interaction, magnetic mixed hemimicelles SPE method and magnetic retrieval of chitosan SPE method were compared to extract five flavonoids compounds. It was found that the recoveries of flavonoids had clear difference in magnetic retrieval of chitosan solid-phase extraction. The compounds contain smaller groups which can make hydrogen bonds, and the extraction recoveries are lower. The main reason was that magnetic retrieval of chitosan solid-phase extraction process is mainly based on hydrogen-bonding interaction between chitosan and compounds. However, in magnetic mixed hemimicelles solid-phase extraction system, the extraction recoveries of five flavonoids had little difference, which indicated that hydrogen bonding interactions in magnetic mixed hemimicelles solid-phase extraction system had no significant effect, except the ionic liquid or ionic surfactants with a large amount of hydrogen bonding site.
In order to study the effect of electrostatic interaction, two sub-types of hydrophilic cephalosporins and quinolones compounds were chosen and investigated. Interestingly, compared with the pure Fe 3 O 4 NPs, the recoveries of these hydrophilic compounds decreased observably when mixed hemimicelles were formed by adding ionic liquids, revealing a very useful rule that trying to analyze hydrophilic compounds by mixed hemimicelles SPE technique is not a good strategy in the future, especially when surfactant and analytes are like charged.
In a word, the extraction mechanism may be speculated that hydrophobic interaction plays a dominant role, while π -π stacking interactions plays a supporting role, hydrogen bonds usually do not play a role, electrostatic interaction sometimes even plays as negative antagonism (electrostatic repulsion) in the magnetic mixed hemimicelles solid phase extraction. Therefore, if researchers need to develop an analysis method based on magnetic mixed hemimicelles solid phase extraction technique, it is essential to grasp these simple guidelines: if log P < 0, hydrophilic compound is not suitable for the magnetic mixed hemimicelles solid phase extraction because of the absence of hydrophobic interaction; if 0 ≤ log P ≤ 3, weak hydrophobic compound may be suitable for the magnetic mixed hemimicelles solid phase extraction system under the π -π stacking interactions reaction; if Log P > 3,

Analytes
Conc. (ng/mL) Mean accuracy (%) RSD (%)  strong hydrophobic compound is suitable for the magnetic mixed hemimicelles solid phase extraction system on the basis of a strong hydrophobic interaction. It is worth noting that these values are empirical data which are speculated according to the experimental results. More accurate values are pending for efforts of more researchers and supports by more experimental data.
The application guidelines of magnetic mixed hemimicelles SPE platform. To construct mixed hemimicelles SPE platform, supporting materials, surfactants and sample matrix should be noticed except considering properties of analytes. Therefore, we summarized representative literature of mixed hemimicelles SPE in Table 9 and provided the following recommendations: For support materials, compared with conventional support materials, some nanomaterials (such as carbon nanotubes and graphene oxide) have obvious advantages in saving the dosage based on their high specific surface area 12 . Moreover, when selecting some multifunctional support materials, especially magnetic materials for example, the magnetic mixed hemimicelles SPE platform can be easily collected and separated by an external magnetic field without additional centrifugation or filtration, which makes separation easier and faster.
For surfactants, this work revealed a rule that the extraction efficiency of analytes increased obviously with the increase of hydrophobic constant of ionic liquid (mainly depends on the length of the carbon chain). This rule might be applicable to other ionic liquid (C 16 mimBr, C 12 mimCl, C 12 mimBF 4 ) or ionic surfactants (such as sodium dodecyl sulfate or cetyltrimethylammonium bromide). It's also found the ionic liquid containing π -π stacking interactions is more suitable for magnetic mixed hemimicelles solid phase extraction system.
For sample matrix, the mixed hemimicelles SPE platform might be applicable to analyze samples in aqueous matrix (hydrophilic), rather than oily matrix (hydrophobic). Because based on our group experience and literature 12,18 in this area (as summarized in Table 9), It's found mixed hemimicelles SPE method have been successfully applied in aqueous matrix (such as water, sewage, urine, serum, plasma, red wine). Interestingly, so far there has no application in oily matrix (such as oil, plant oil, gasoline). Actually this is understandable that, compared with in aqueous matrix, the hydrophobic interaction between analytes and mixed hemimicelles are competition and interference in oily matrix.

Conclusions
The formation of magnetic mixed hemimicelles is a valuable strategy for the extraction and concentration of analytes in complex samples. This work developed a series of simple, cheap and accurate extraction methods for the determination of three categories and six sub-types of compounds. Most importantly, to the best of our knowledge, this is the first attempt to systematically study and comprehensively illustrate the extraction mechanism of   mixed hemimicelles solid-phase extraction. The application guidelines about supporting materials, surfactants and sample matrix were also summarized. With the extraction mechanism and platform established in the study, confidence is gained and increased purposefulness is achieved, providing a stepping stone in the realization of the widespread application of magnetic mixed hemimicelles solid-phase extraction for preconcentration of trace amounts of analytes from complex samples.