Surfactant assisted magnetic dispersive micro solid phase extraction-HPLC as a straightforward and green procedure for preconcentrating and determining Caffeine, Lidocaine, and Chlorpromazine in biological and water samples

ABSTRACT Magnetic dispersive micro solid-phase extraction as a proper and selective sample preparation was developed to extract and clean up three drugs, including Caffeine (CAF), Lidocaine (LID), and Chlorpromazine (CPZ), before determining by HPLC-DAD. Nickel-doped BiFeO3 was synthesised using the sol–gel method as a magnetic sorbent core and coated with tetraethyl orthosilicate (TEOS) to enhance the sorbent stability and the sorbent selectivity. In the extraction procedure, the sorbent was dispersed into acetonitrile as disperser solvent and sodium dodecyl sulphate (SDS) as a disperser agent before injecting into the sample solution to extract the analytes. Several influential factors in the method were evaluated using an experimental design strategy. Under the optimum conditions, the method displayed wide linearity in the concentration ranges of 20.8–365, 1.4–389, and 2.7–355 ng mL−1 to determine CAF, LID, and CPZ with a suitable R2 between 0.9963 and 0.9971, respectively. Intra-day and inter-day relative standard deviations were in the ranges of 3.5–3.8 and 3.9–4.3%. LODs and LOQs for the determination of CAF, LID, and CPZ were lower than 6.3 and 20.8 ng mL−1. The method is convenient for determining the analytes in real water and biological samples with RSD (n = 3) and recovery in the ranges of 3.0–4.3% and 87.5–96.9%, respectively.


Introduction
Caffeine (CAF), 1,3,7-trimethylpurine-2,6-dione, is a natural alkaloid that is primarily found in some plants such as cocoa, coffee, and tea.CAF and its derivatives are prescribed as a psychoactive drug to treat neonatal apnoea [1], androgenetic alopecia [2], postural hypotension [3], and apnoea of prematurity [4].A study of the effects of high caffeine consumption, especially coffee, indicated that high caffeine consumption significantly increases anxiety, depression, and stress in high school children [5].Overdose of CAF has resulted in acute renal failure, arrhythmia, and severe rhabdomyolysis [6,7].Lidocaine (LID), 2-(diethylamino)-N- (2,6-dimethylphenyl) acetamide is widely used as a local anaesthetic in surgery where the patient's consciousness is not lost [8].Recently, LID showed anti-tumour effects on breast cancer [9] and colon cancer in vitro [10].It also enhanced the effects of other anticancer drugs on bladder cancer in humans [11].Overdose of LID may cause human death through cardiac arrest, ventricular fibrillation, and brain death [12,13].Chlorpromazine (CPZ), 3-(2-chlorophenothiazin-10-yl)-N, N-dimethylpropan-1-amine, is the most significant compound in the great group of phenothiazine derivatives.It is extensively utilised as a therapeutic agent for controlling agitation, excitement, and other psychomotor disturbances in schizophrenic patients and decreases the manic phase of manic depressive modes [14].Adverse effects of CPZ include dry mouth, urine retention, constipation, blurred vision, and dizziness.Overdose of CPZ leads to kidney injury, liver damage in humans, and brain damage in mice [15].The chemical structures of CAF, LID, and CPZ are presented in Figure S1.
The sample preparation procedure is a critical stage to determine an analyte in biological and environmental samples due to the high matrix effects of these real samples, low concentration of analyte in the samples, and convert the sample to suitable form to inject into detection systems, such as high-performance liquid chromatography (HPLC), gas chromatography (GC), or capillary electrophoresis [16][17][18].Dispersive microsolid-phase extraction (DμSPE) is widely utilised as the sample preparation method with a straightforward and practical way, appropriate preconcentration factor, and clean up [19][20][21][22][23][24].In the DμSPE procedure, a sorbent with suitable properties to interact with analytes is poured into samples and then dispersed to adsorb analytes before separating the sorbent.The analytes adsorbed on the sorbent are usually desorbed to these determinations in a proper solvent [25].In DμSPE, sorbent plays an essential role in extracting an analyte through a selective and effective interaction with the analyte [26].Therefore, the preparation or selection of a sorbent is very effective in the success of the analyte extraction process, especially in real samples [27,28].
Also, the sorbent consumption and extraction time depend on the sorbent dispersion efficiency in the sample solution.In other words, as the sorbent dispersion efficiency in the sample solution increases, the sorbent contact area with the analyte increases, which leads to a reduction in the extraction time and sorbent consumption.Usage ultrasonic wave is a proper method to increase the sorbent dispersion efficiency in the sample solution [29].The sorbent degradation is the vital drawback of ultrasonic waves to disperse the sorbent due to the formation of microreactors and the realisation of high energy produced with these waves in the sample solution [21].Surfactants as disperser agents can reduce the sorbent degradation in the presence of ultrasonic waves by decreasing the dispersion time (ultrasonic time) [19].Although increasing the sorbent dispersion efficiency increases the extraction efficiency of the analyte at its surface, it is difficult to separate the sorbent from the sample solution after the analyte extraction.It increases the centrifuge time to separate the sorbent due to the formation of suspensions with fine sorbent particles [30].The use of magnetic sorbents can enhance the sorbent separation in the presence of an external magnetic field and eliminate the centrifuge stage.
The perovskite-type bismuth ferrite (BFO) is known as one of the several compounds that shows the coexistence of ferroelectric and antiferromagnetic orders above room temperature with Curie temperature (T C n 830°C) and Neel temperature (T N n 375°C) [31,32].The multiferroic BFO exhibits a rhombohedral distortion structure, and it belongs to the R3c group.Also, BFO as a non-toxicity material can be used as a sorbent core, leading to easy and rapid separation of the sorbent from the sample solution in the presence of a strong magnet [33].However, BFO doesn't have suitable functional groups to selectively interact with the analyte and thermodynamic instability in its structure.So, it is essential to modify the BFO nanoparticles by functionalising their surface to increase their selectivity and stability.
Our study shows that no DμSPE method for the simultaneous extraction of CAF, LID, and CPZ in urine and water samples has been proposed so far.Therefore, providing a suitable sample preparation method for measuring these drugs to study their adverse effects in humans or environment with good sensitivity and appropriate selectivity was considered.This research prepared Ni-doped BFO/TEOS as a novel and green sorbent to extract CAF, LID, and CPZ based on the DμSPE procedure before determining with HPLC.Although BFO has a magnetic property, Ni as a dependent in the BFO lattice structure leads to an increase in magnetic properties and better separation of the prepared sorbent from the sample solution in the presence of an external magnetic field.Besides, functionalization of Ni-doped BFO as a magnetic sorbent core with SiO 2 nanoparticles to prepare the green sorbent increased the extraction efficiency and selectivity of the sorbent through suitable interaction between the analytes with silanol groups of TEOS on the sorbent surface.A sol-gel strategy was utilised to prepare and functionalise Ni-doped BFO, increasing the sorbent stability and enhancing the uniformity of the sorbent structure at the molecular level by forming the chemical bond.In the DμSPE method, a surfactant as a disperser agent was applied to enhance the sorbent dispersion efficiency and reduce the ultrasonic time.The effective parameters on the DμSPE process were optimised by an experimental design with few experiments.The most advantages of experimental design to optimise effective parameters are the reduction of the number of tests, the consumption of reagents, the cost and time of optimisation, and the possibility of investigating the effect of interaction between parameters on the extraction efficiency [34,35].

Instruments and materials
The drugs (CAF, LID, and CPZ) in standard and real samples were determined using a Knauer HPLC system (Germany) equipped with a photodiode array detector (K-2600), an EZ-Chrom Elite software, and a Eurospher 100/5C18 analytical column (4.6 mm ×250 mm, 5 µm).The mobile phase contained two components: phosphate buffer (pH 7.30, 10 mM) and acetonitrile with a ratio of 60:40% v/v, a total run time of 10 min, and a flow rate was 1.0 mL min −1 .The wavelength of 254 nm was set in the DAD detector to determine CAF, LID, and CPZ, respectively.The pH of the sample solution was determined and adjusted by a pH metre (Metrohm 780, Switzerland).The crystal structure, morphology, and functional groups of the prepared sorbent were investigated using an X-ray diffractometer (Bruker D8 ADVANCE, Germany), a field-emission scanning electron microscopy (FE-SEM, TESCAN MIRA3, the Czech Republic), and Fourier-transform infrared spectroscopy (FT-IR, Bruker Optik GmbH TENSOR 27, Germany).The optimisation process was performed using Minitab software (V 14).

Synthesis of Ni-doped BFO nanoparticles
A simple sol-gel method using tartaric acid was performed to prepare Ni-doped BFO nanoparticles (BiFe 0.95 Ni 0.05 O 3 ) [36,37].In the process, a suitable amount of Ni(NO 3 ) 2 .6H 2 O, Fe(NO 3 ) 3 • 9H 2 O, and Bi(NO 3 ) 3 • 5H 2 O were poured in a solution containing 1.0 mL of concentrated nitric acid and distilled water (20 mL), followed by stirring at 750 rpm for 1 h at a temperature of 54 ± 3°C until all compounds were wholly dissolved in the solution.Tartaric acid was added to the obtained solution and stirred for 2 h to form a suspension with a milky colour.An ammonium solution (1.0 mol L −1 ) was dropwise added into the suspension under stirring to adjust the pH of the suspension to 2.0.The suspension was dried by rising temperature to 200°C to form a powder, then calcining at 400°C for 3 h in an electric furnace.The obtained powder was washed several times with a nitric acid solution (0.1 mol L −1 ) and deionised water to remove impurity phases, including Bi 25 FeO 39 and Bi 2 Fe 4 O 9 , and unreacted materials, respectively.The powder was finally calcined at 650°C for 4 h in the electric furnace to form Ni-doped BFO nanoparticles with a dark brown colour.

Synthesis of Ni-doped BFO nanoparticles/TEOS
Tetraethyl orthosilicate (TEOS, 2.0 mL) was poured in 4.0 mL of methanol and stirred for 20 min to form a uniform solution.Concentration hydrochloric acid was added dropwise to the solution, followed by stirring for 2 h to create a uniform sol [16].Ni-doped BFO nanoparticles (0.3 g) were added to the sol, and the vial cup was closed entirely to prevent methanol evaporation.The mixture was stirred at 1000 rpm for 24 h at 45 ± 3 O C. The mixture was centrifuged, and the obtained nanoparticles (Ni-doped BFO/TEOS) were separated from the methanol phase.Ni-doped BFO nanoparticles/TEOS was heated at 80 ± 5 O C for 20 h to dry the product and then well ground in a mortar.

Characterisation of Ni-doped BFO nanoparticles/TEOS
FESEM image of Ni-doped BFO nanoparticles/TEOS is shown in Figure S2, indicating that the sorbent surface consists of Ni-doped BFO nanoparticles coated with TEOS.The use of TEOS binds Ni-doped BFO nanoparticles in its lattice structure, which leads to increased sorbent stability.Also, the sorbent surface has pores of different sizes that can extract analyte through physisorption.However, the presence of silanol groups on the sorbent surface plays a major role in the chemisorption of the analytes and the selectivity of the sorbent.
EDX pattern of Ni-doped BFO nanoparticles/TEOS is presented in Figure S3.The pattern indicated that the sorbent is composed of bismuth, iron, oxygen, silica and nickel, and other elements are not present as impurities in its structure.The quantitive results in Table S1 also show that two elements (bismuth and iron) have the highest weight percentage in the sorbent structure, respectively.
FTIR image of Ni-doped BFO nanoparticles/TEOS is shown in Figure S4.A peak at 3422 Cm −1 is related to the stretching vibration of the OH functional groups.The vibration peak at 1096 Cm −1 corresponded to BiO in the BFO lattice.Two peaks at 946 and 806 Cm −1 are assigned to the stretching vibration of SiO groups and the rocking vibration of the Si-CH 3 group, confirming that TEOS was coated on the Ni-doped BFO nanoparticles.Two other peaks at 561 and 464 Cm −1 are related to the stretching vibration and bending vibration of FeO in the BFO lattice, respectively.
XRD pattern of BFO and Ni-doped BFO nanoparticles/TEOS is shown in Figure S5.BFO has several peaks at 22.4°, 31.7°,32.1°, 39.0°, 39.4°, 45.7°, 51.3°, 51.8°, 57.1° and 66.3° with the planes of 012, 104, 110, 006, 202, 024, 116, 122, 241/300, and 220 respectively.The intensities of the two peaks with the planes of 110 and 012 were increased and decreased with adding Ni as the dopant in the XRD pattern of the Ni-doped BFO nanoparticles/TEOS [38].BFO and Ni-doped BFO nanoparticles/TEOS displayed rhombohedral distortion perovskite structures with a reference code of 01-072-2035 for BFO lattice.No suitable peak was observed for TEOS in the XRD pattern of the Ni-doped BFO nanoparticles/TEOS, which may be due to the lack of a suitable crystalline structure of TEOS at the time of coating the sorbent core (the Ni-doped BFO nanoparticles).

Microextraction procedure
The sorbent (Ni-doped BFO nanoparticles/TEOS, 3 mg) was added to 1.0 mL of acetonitrile and 1.0 mL of SDS solution (0.032 mol L −1 ), followed by dispersing for 10 min under ultrasonication.Next, 15.0 mL of a sample solution containing CAF, LID, and CPZ was poured in a falcon vial, and its pH was adjusted to 8.1 using an aqueous sodium hydroxide solution (0.1 mol L −1 ).The sorbent suspension was injected into the sample solution, followed by sonicating for 3 min.The sorbent was then separated from the sample solution by a strong magnet after the extraction of the analytes.THF (250 μL) as a desorption solvent was added to the sorbent, then sonicated for 8 min.The THF phase was separated in the presence of the strong magnet, and 20 μL of it was injected into HPLC to determine CAF, LID, and CPZ.The extraction recovery (ER%) was calculated using the following equation: where C s and V s are the analyte concentration in the sample and the sample volume, and C o and V o are the analyte concentration in the desorption solvent and the desorption solvent volume, respectively [29].

Optimisation strategy
The microextraction method depends on various factors, which their optimisation can have a significant effect on the extraction efficiency of the analytes.Therefore, several factors such as type of disperser solvent, extraction time, pH, type and volume of desorption solvent, desorption time, amount of sorbent, amount of disperser agent were evaluated using one factor at a time and experimental design strategy.Two factors, type of disperser solvent and desorption solvent, were studied by one factor at a time method because their evaluation in the experimental design leads to a significant increase in the number of runs required for the optimisation procedure.Other factors were investigated in two steps: the screening step using a Plackett-Burman design and the optimisation step using a Box-Behnken design.

Sample preparation of water and urine samples
The application of the method was evaluated to determine CAF, LID, and CPZ by analysing several real samples, including well, tap, river water, and urine samples.The well water, tap water, and river water samples were obtained from a well outside Mashhad, Research Analytical Chemistry Laboratory (Azad University of Mashhad, Iran), and the Mayan River (Torqabeh-Shandiz, Iran), respectively.Drug-free urine samples were obtained from a laboratory (Mashhad, Iran) from healthy volunteers.The conditions and reason for sampling were explained to the volunteers, and sampling was performed with their permission.Sampling was performed at least 20 h after consuming any caffeinated beverage such as tea or coffee.All water and urine samples were centrifuged for 10 min at 7000 rpm and filtered through a Whatman filter with a pore size of 0.45 μm before analysis.Then, the water samples were spiked with a standard solution of CAF, LID, and CPZ to prepare the spiked samples at three concentrations of 6.0, 40 and 80 ng mL −1 and urine samples at two concentrations of 40 and 80 ng mL −1 before reanalysing with the proposed methods.

Type of disperser solvent
In a mode of the DμSPE method, the sorbent disperses into a suitable solvent under ultrasonic wave to prepare a proper sorbent suspension with fine particles before injecting into the sample solution.The technique reduces the extraction time and enhances the method selectivity by preventing the extraction of interfering species with a slow mass transfer in the sample solution.Four solvents, including methanol, acetonitrile, ethanol, and tetrahydrofuran, were selected as the disperser solvent (Figure 1).The results indicated that acetonitrile is the best disperser solvent in extracting CAF, LID, and CPZ.This may be due to the highest dielectric constant of acetonitrile compared to other selected solvents.However, using acetonitrile in the presence of SDS leads to an increase in the extraction efficiencies of the analytes because SDS as a disperser agent improved the dispersion efficiency of the sorbent into acetonitrile.

Type of desorption solvent
After the extraction procedure, the desorption of the analyte into a suitable solvent from the sorbent surface is a critical stage to its determination by HPLC.Thus, several solvents, including methanol, ethanol, acetonitrile, and tetrahydrofuran, were chosen as the desorption solvent to desorb the extracted analytes on the sorbent surface.The results are presented in Figure 2, showing the highest peak areas for the determination of CAF, LID, and CPZ were obtained using tetrahydrofuran as the desorption solvent.The results also  indicated that acetonitrile displayed the lowest extraction efficiencies of the analytes.In other words, the peak areas for all analytes decrease as the polarity of the selected solvents increases.CAF, LID, and CPZ have a positive charge at pH 7 due to their pK a for amine function groups are 10.4,9.3, and 9.4, respectively.Therefore, a proper interaction between the positive charge of the analytes with the unbonded electron pairs of the oxygen atom on methanol, ethanol, and THF can lead to desorb the analyte from the sorbent surface.This interaction occurs between the positive charge of the analytes and the unbonded electron pair of nitrogen atoms with lower strengths than the oxygen atoms of the other solvents studied.Besides, the tetrahydrofuran heterocyclic ring can interact with analytes through non-polar-non-polar interactions, increasing the analyte extraction efficiency compared to methanol and ethanol.

Screening stage
Other factors, including sample solution volume, extraction time, pH, volume of desorption solvent, desorption time, amount of sorbent, amount of disperser agent, were investigated using an experimental design.In the first stage, selective factors were screened to determine the significant factors on the extraction of CAF, LID, and CPZ to reduce the experimental runs to optimise the microextraction procedure.In the second stage, the significant factors were evaluated and optimised.Therefore, a Plackett-Burman design was generated for screening the factors.The runs in this design were performed in a random order to eliminate the effects of unknown and uncontrollable factors.The factors, their levels, and the Plackett-Burman design are presented in Table 1.The response of each run was determined based on three repetitions of each run under the same conditions and calculating the average of their results.Significant factors were determined using analysis of variance (ANOVA) at a 95% confidence limit (Table S2).
The p-value as a suitable parameter in ANOVA Table is applied to evaluate the impact of factors on the extraction of CAF, LID, and CPZ.At the 95% confidence limit, a factor with a p-value lower than 0.05 is considered a significant factor in the extraction procedure.Significant factors are marked with a positive sign in the ANOVA table (Table S2).Besides, the Pareto chart is a graphical representation for the simple determination of significant factors (Figure 3).According to Figure 3, three factors, including pH (B), desorption time (H), desorption solvent volume (J), have significant effects on the extraction of CAF, LID, and CPZ.The results also indicated that pH has the highest impact, and desorption solvent volume is the next factor with a high effect on the extraction of CAF, LID, and CPZ.Therefore, these factors were selected for the optimisation stage, and other factors were fixed in a constant amount.The selected amount of non-significant factors, including sample solution volume, disperser time, extraction time, sorbent amount, volume of disperser agent, and amount of salt, were constant at 15 mL, 10 min, 3 min, 3 mg, 1.0 mL, and 0 w/v% based on the obtained optimisation plot (Figure S6).

Optimisation stage
Box-Behnken design is a proper and widely used design for the optimisation of factors.This design requires fewer experiments than other designs, such as central composite design for the optimisation process, which reduces sample and sorbent consumption, optimisation time, and cost.Therefore, a Box-Behnken design was generated to optimise three significant factors selected from the screening stage.The selected factors and Box-Behnken design are shown in Table 2.The design consisted of 15 runs in which the runs were performed in a random order to eliminate the effects of unknown and uncontrollable factors.The response of each run was determined based on three repetitions of each run under the same conditions and calculating the average of their results.The results were investigated with ANOVA at a 95% confidence limit (Table S3).Based on the obtained p-value for evaluating the effect of each factor in the extraction of CAF, LID, and CPZ, all three selected factors had significant impacts on the extraction of analytes because their p-value was lower than 0.05 at a 95% confidence limit.In contrast, two-way interactions between factors were not significant parameters with meaningful effects on the extraction of CAF, LID, and CPZ.In other words, the extraction of analytes is mainly influenced by each factor independently, and the interaction between them does not have a significant effect.Significant factors are marked with a positive sign in the ANOVA table (Table S3).Also, the model presented by investigating the obtained results is significant because the p-value of the model (0.000) is lower than 0.05, and the p-value of lack of fit (0.674) is higher than 0.05 (lack of fit is a non-significant parameter).The lack of fit indicated the changes in the obtained responses around the proposed model.Therefore, the lack of fit is a non-significant parameter (p-Value ≥ 0.05) when the variations in responses with the fitted model are small and meaningless.A quadratic equation in uncoded units was obtained with the analysis of the results from Box-Behnken design to express the relationship between the factors of the interactions with the analyte extraction efficiency as follows: The fit of the equation with the results was evaluated using the determination coefficient (R 2 ) and the adjusted determination coefficient (R 2 adj ), and the predicted determination coefficient (R 2 pred ).The results indicated that the equation fit the results well due to the high value of R 2 (98.85%) and R 2 adj (96.79%).Besides, a suitable R 2 pred of 89.95% indicated that the equation can predict the future results well.The equation showed that pH (B) has a high and positive effect on the extraction of CAF, LID, and CPZ due to the large value of the coefficient.Obviously, the analyte adsorption on the sorbent to its extract depends on the interaction between them.The strength of these interactions is affected by the pH of the sample solution through a change in the polarity of the functional groups at the sorbent surface and the analyte.At pH 8.1, the amine groups of the analytes can interact with the silanol groups of the sorbent surface by forming hydrogen bonds.At higher pH, a reduction in extraction efficiencies of the analytes was displayed due to instability of sol-gel structure and degradation of sorbent.The desorption time (H) and desorption solvent volume (J) are other significant factors that increased the extraction efficiencies of the analytes with increasing this factor.The desorption process is an equilibrium and time-dependent process that requires a minimum time to reach equilibrium and achieve maximum extraction efficiency of the analytes.Therefore, the extraction efficiency of analytes increases with increasing desorption time.Also, a suitable volume of desorption solvent is required to distribute the analyte between the two phases of sorbent and desorption solvent.However, by further increasing the amount of desorption solvent, the extraction efficiency decreases due to the dilution of the analyte.Effects of simultaneous change of two factors on ER% are shown in surface plots (Figure 4), indicating that the extraction efficiencies of CAF, LID, and CPZ were increased with an increasing amount of factors and reduced afterwards.The slope of the curve is higher than the other curves for the effects of change of pH and desorption time on the extraction efficiency.The optimum values of three significant factors were determined using an optimisation plot (Figure 5).According to Figure 5, the optimum values of pH, desorption time, and desorption solvent volume were 8.1, 8 min, and 250 μL, respectively.The optimum value of all factors is presented in Table S4.

Figure of merits
Linearity, LOD, LOQ, RSD, and preconcentration factors to determine CAF, LID, and CPZ were investigated in distilled water and urine samples.The matrix-matched calibration was used to evaluate the calibration curve in the urine sample for all analytes (Table 3).The calibration curves were linear in the concentration ranges of 20.8-365, 1.4-389, and 2.7-355 ng mL −1 with R 2 between 0.9963-0.9971 to determine CAF, LID, and CPZ in the distilled water, respectively.LODs and LOQs to determine CAF, LID, and CPZ were calculated based on 3S b /m and 10S b /m.Where S b and m are the standard deviation for three determinations of blank samples and slope of the calibration curve, respectively.LODs and LOQs were lower than 6.3 and 20.8 ng mL −1 for the analyte determination in the distilled water, respectively.Intra-day and inter-day RSDs were determined based on five determinations of each analyte with a concentration of 40 ng mL −1 under the optimum conditions on one day and three consecutive days.Intra-day and inter-day RSDs were in the ranges of 3.5-3.8%and 3.9-4.3%,respectively.The preconcentration factor (PE) was obtained using the following equation: Where C o and C s are the concentration of the analyte before and after performing the extraction procedure.Besides, V s and V o are the volumes of desorption solvent and water sample.PF for the determination of CAF, LID, and CPZ in the distilled water was in the range of 569-633.

Real sample analysis
Several real samples, including well, tap, river water, and urine samples, were prepared based on section 2.7.Each real sample was analysed three times under the same condition.Recovery was calculated from the mean concentrations obtained for the three analyses.The results are shown in Table 4, indicating that the technique has suitable recoveries for determining CAF, LID, and CPZ in real water and biological samples with proper RSDs.The obtained recoveries were in the range of 87.5-96.9%,with RSDs between 3.0% and 4.3% for three determinations of CAF, LID, and CPZ in real water and biological samples under the optimum conditions.Typical chromatograms of CAF, LID, and CPZ in non-spiked and spiked samples were presented in Figure S7.

Comparison with other procedures
The ability of the method to determine CAF, LID, and CPZ in real water and biological samples was compared by other published methods.All selected methods include a microextraction procedure as a sample preparation method and have been published recently.Several parameters, such as LOD, LOQ, Linearity, and RSD, were compared and presented in Table 5.The results indicated that the suggested method has LODs and LOQs better than other published methods for determining CAF, LID, and CPZ.It is possible to measure analytes at low concentrations by reducing the LOD and LOQ method, which is an important parameter for measuring drugs in real samples where the concentration of analytes is very low.However, LOD and LOQ obtained for the LID determination are slightly better than the suggested method.In the procedure, a combination of extraction and microextraction methods, including molecular imprinting solid-phase extraction and biodispersive liquid-liquid microextraction, were used to determine LID, leading to a better LOD and LOQ.However, performing two extraction methods caused an increase in the procedure time and material and reagent consumption and reduced reproducibility of the method due to a rise in the procedure stages.The suggested method has good and wide linearity compared with other methods.However, providing a method with a wide linear range can help to measure the analyte in different samples without the need for other additional processes such as sample dilution.Besides, RSDs of the presented method are equal or slightly better than other methods.This reduction in the method reproducibility may be due to the limitation of sorbent separation after the extraction procedure and separation of the acceptor phase after the desorption procedure under the external magnetic field.However, the suggested method showed suitable LODs, LOQs, linearity, and RSD towards the analytes.Besides, the method requires an appropriate sample volume, short extraction time, high preconcentration factor, and low desorption solvent volume with low toxicity.The sorbent was shown to have a proper extraction efficiency towards the analytes, high chemical and mechanical stability, and simple separation from the samples.Therefore, it can apply as an excellent sample preparation method to determine CAF, LID, and CPZ in real water and biological samples.

Conclusion
A magnetic dispersive micro solid-phase extraction procedure was developed to extract and clean up three drugs, including CAF, LID, and CPZ, in real water and biological samples before determining by HPLC.In the procedure, the sorbent was dispersed into acetonitrile as disperser solvent before injecting it into the sample solution.SDS as a disperser agent was applied for increasing the dispersion efficiency of sorbent.Usage of acetonitrile and SDS leads to a reduction in the dispersion time of the sorbent and an increase in the dispersion efficiency of the sorbent.Ni-doped BFO/TEOS was synthesised as a novel and green sorbent and  characterised with various methods.Increasing Ni nanoparticles as a dependent in the BFO lattice structure enhances magnetic properties of Ni-doped BFO as the magnetic sorbent core.Also, coating Ni-doped BFO with TEOS using the sol-gel strategy increases the sorbent stability and sorbent selectivity towards the analytes.Many influential factors were investigated and optimised to obtain the best extraction efficiency towards the analytes using experimental design and one factor at a time procedure.The experimental design reduces the experiments and the cost and time for the optimisation procedure.This strategy made it possible to investigate the interaction between the factors, indicating that the interaction between

Figure 3 .
Figure 3. Pareto chart obtained from screening stage for the extraction of CAF, LID, and CPZ.

8 a
Limit of detection b Limit of quantitation c Relative standard deviation d Homogeneous liquid-liquid microextraction method based on solvents volume ratio alteration e Deep eutectic solvent-based liquid-phase microextraction method; 6 Thin film microextraction f Ultrasound-assisted emulsification-microextraction g High-performance liquid chromatography -photodiode array detector h Dispersive liquid-liquid microextraction based on solidification of floating organic drop i Single-interface hollow fber liquid-phase microextraction and electromembrane extraction j On-line electrochemically controlled solid phase microextraction h Ion-pair extraction procedure

Table 1 .
The selected factors and Plackett-Burman design.

Table 2 .
The significant factors and Box-Behnken design.
water, tap water, river water, and human urine samples indicated that the process has excellent ability to analyse real water and biological sample with proper relative standard deviations and recoveries.

Table 3 .
Figure of merit for the extraction of CAF, LID, and CPZ.

Table 4 .
Determination of CAF, LID, and CPZ in real water and urine samples.
a Not detect.

Table 5 .
Comparison of the method with other procedures.