Application of Ni:ZnS nanoparticles loaded on magnetic multi-walled carbon nanotubes as a sorbent for dispersive micro-solid phase extraction of phenobarbital and phenytoin prior to HPLC analysis: experimental design

Abstract

In this study, an efficient magnetic sorbent was prepared for dispersive micro-solid phase extraction (DMSPE) by incorporation of Ni:ZnS nanoparticles in magnetic multi-walled carbon nanotubes (Ni:ZnS-MMWCNTs). This sorbent was used for the extraction of phenobarbital (PHB) and phenytoin (PHN) prior to their determination by high performance liquid chromatography (HPLC) equipped with an ultra violet detector. In order to obtain the best extraction performance of PHB and PHN, the factors affecting the extraction efficiency of DMSPE including pH of the sample solution, amount of adsorbent, vortex time, desorption volume, ultrasonic time, ultrasonic temperature as well as ionic strength were investigated. The main factors were found using the Plackett–Burman (P–B) method and were optimized via central composite design (CCD). Under the optimum experimental conditions, the calibration graphs were found to be linear over the concentration ranges of 2–400 μg L⁻¹ and 3–500 μg L⁻¹ with the limit of detection (LOD) and limit of quantification (LOQ) of 0.5 and 1.8 μg L⁻¹ for PHB and 0.8 and 2.8 μg L⁻¹ for PHN, respectively. The good relative standard deviations (RSDs, at a level of 10 μg L⁻¹, n = 6) of 3.7 and 4.5% were achieved for PHB and PHN, respectively and the preconcentration factor for the proposed method was found to be 200. Ultimately, the developed method was successfully applied to the determination of PHB and PHN in plasma, urine and water samples.

---

1. Introduction

Antiepileptic drugs (AEDs) such as phenobarbital (PHB) and phenytoin (PHN) have been used in mono or polytherapy in the treatment of epilepsy and psychological conditions such as anxiety, depression, autism, and post-traumatic stress disorders.¹ Although the preferable approach to epilepsy treatment is monotherapy, the monotherapy with AEDs often fails and patients with refractory seizures must use multiple AEDs to control the incidence of unpredictable epileptic seizures.^2–4 Due to the drug–drug interactions in polytherapy, undesirable side effects such as itching, headaches, dizziness, fever, depression, double vision, nausea and hallucinations are often caused.¹ Nowadays, therapeutic drug monitoring (TDM) is a widely accepted procedure for the improvement of the effectiveness of antiepileptic therapy and the optimization of the patients' clinical outcomes as well as the reduction of undesirable therapeutic side effects.^5,6 TDM is a clinical method for the measurement of drug concentrations in the patients' plasma, serum or urine samples for the determination of drug dose and usage. Since there is a significant relation between the blood drug level and its effects, TDM is one of the most essential subjects in antiepileptic therapy.

Several analytical instrumental techniques such as immune assays,⁷ spectrophotometry,⁸ high performance liquid chromatography (HPLC),^9–11 micellar electrokinetic chromatography¹² and capillary electrophoresis¹³ have been reported for the determination of AEDs. However, their determination at low levels in complicated biological samples with high concentration of interfering matrix requires a sample preparation step for the separation and the enrichment prior to their instrumental analysis. Thus, the availability of a simple, sensitive, rapid, and highly accurate sample preparation technique is necessary for the measurement of AEDs. The most popular sample preparation techniques used for the separation and preconcentration of drugs are liquid–liquid extraction^14,15 and solid phase extraction.^16–18 However, classical methods have experienced some disadvantages such as time consuming and multistage operations, requirement of relatively large sample and organic solvent volumes, labor intensiveness, and difficulty in automation.^19–21 To cope with these limitations, recent studies have focused on the development of simple, highly efficient, rapid, inexpensive, and miniaturized microextraction techniques. Up to now, some microextraction techniques including liquid phase microextraction (LPME),²² solidified floating organic drop microextraction (SFODME),^23,24 hollow fiber solid phase microextraction (HF-SPME),²⁵ and stir bar sorptive extraction (SBSE)²⁶ have been applied for the extraction of AEDs from various samples, however, there is no report available concerning the use of dispersive micro-solid phase extraction (DMSPE). DMSPE^27,28 is based on the dispersion of a small amount of sorbent in the sample solution and have the advantages of simplicity, low consumption of organic solvent and sorbent, high efficiency of recovery, and the extraction time shorter than the traditional SPE.^29–31 The choice of the sorbent is the main factor affecting the extraction efficiency in SPE methods.^32–34 In recent years, the application of nanomaterials such as multi-walled carbon nanotubes (MWCNTs) and metal nanoparticles has drawn much attention in SPE due to their special characteristics including high surface area, high adsorption capacity, high chemical and thermal stability and short diffusion routes.^29,35,36 However, one of the limitations facing nanomaterials in their applications concerns the small particle size, which brings about difficulties in their separation from the liquid phase. The limitation of tedious and complicated centrifugation and filtration steps has been overcome through the modification of these sorbents with magnetic nanoparticles (MNPs)³⁷ which simplify the gathering of sorbent in solution through the application of an external magnetic field.

Ni:ZnS nanoparticles have been supported on activated carbon and are used for the extraction of bendiocarb and promecarb as well as losartan and valsartan.^29,38 In this work, successive to the previous one, the possibility of loading of Ni:ZnS nanoparticles on nano support materials was investigated and it was loaded on magnetic MWCNTs (Ni:ZnS-MMWCNTs). The new prepared sorbent was then used for DMSPE of PHB and PHN at trace levels from biological and water samples. The main factors affecting the DMSPE procedure were found by Plackett–Burman (P–B) screening design and then optimized by central composite design (CCD).

2. Experimental

2.1. Reagents and materials

Standards of PHB and PHN were purchased from Sigma Company (St. Louis, MO, USA). Stock standard solutions of the drugs (200 mg L⁻¹) were separately prepared by dissolving appropriate amounts of PHB and PHN in HPLC grade methanol and stored at 4 °C taken to the room temperature just before use. The working solutions were prepared freshly by the appropriate dilution of the standard solutions with HPLC grade water. Methanol, ethanol and acetone (HPLC grade), hydrochloric acid, sulfuric acid (98% w/w), nitric acid (65% w/w), aqueous ammonia (25% w/w), sodium chloride, sodium hydroxide, nickel acetate, zinc acetate, Na₂EDTA, thioacetamide, ferrous chloride tetrahydrate (FeCl₂·4H₂O), and ferric chloride hexahydrate (FeCl₃·6H₂O) were all purchased from the Merck company (Darmstadt, Germany). The MWCNTs with 30–50 nm diameters and ∼20 μm length were purchased from the Research Institute of Petroleum Industry (Tehran, Iran). Human samples were obtained from patients in Parastar Hospital (Behbahan, Iran) and the informed consent was obtained for any experimentation with human subjects. All experiments were performed in compliance with relevant laws and guidelines of research committee of chemistry department of Yazd University and the institutional committees have approved the experiments.

2.2. Instrumentation and software

The chromatographic runs for the separation and determination of PHB and PHN drugs were performed using a KNAUER Smart line HPLC system equipped with a model 2550 UV-VIS detector (the wavelength was set at 220 nm), micro vacuum degasser, LPG system and a Zorbax SB-C₁₈ (150 mm × 4.6 mm, 5 μm) (Agilent) column. The EZChrom Elite software applied to process the chromatographic data. The HPLC conditions for the determination of PHB and PHN were optimized and the mobile phase was isocratic binary of acetonitrile : water (40 : 60, v/v) with a flow rate of 1 mL min⁻¹. An ultrasonic water bath (TECNO-GAZ, 60 Hz, 130 W, Italy) equipped with a digital timer and a temperature controller was used in this work. Adjustment and measurement of the pH values for the sample solutions were carried out using a Metrohm 774 pH-meter equipped with a combined glass-calomel electrode.

2.3. Data processing

The experimental designs as well as regressional analysis were carried out by means of the STATISTICA software (StatSoft Inc., Tulsa, USA: version 7.0). An analysis of variance (ANOVA) was performed in the design to assess the significance of the model. The quality of the polynomial model equation was judged statistically via the lack of fit (LOF) while its statistical significance was determined by the F-test and P-value.

2.4. Preparation of Ni:ZnS-MMWCNTs

In the first step, untreated MWCNTs were purified while heated at 350 °C for 35 min to remove the amorphous carbon and the other impurities. After thermal treatment, acid functionalized MWCNTs (AF-MWCNTs) were prepared by blending the MWCNTs (2.5 g) with 200 mL of sulfuric acid and nitric acid mixture (3 : 1, v/v, 6 mol L⁻¹ acids) followed by refluxing at 60 °C for 4 h. Then, the dispersed solution was filtered and AF-MWCNTs were washed with several portions of deionized water to neutrality, and dried at 80 °C overnight.³⁹ Subsequently, 2 g of AF-MWCNTs was added to 100 mL of deionized water containing FeCl₂·4H₂O (2.1 g) and FeCl₃·6H₂O (5.8 g), and the mixture was sonicated for 10 min under N₂ atmosphere at 70 °C. After that, 20 mL of ammonia solution (25% w/w) was added and the suspension was stirred for 30 min. Finally, the magnetic MWCNTs (MMWCNTs) were separated from the mixture by the application of an external magnetic field, washed with distilled water and ethanol (99.9%) respectively, and dried at 100 °C for 4 h for the subsequent step.

Besides, 2.5 g of the prepared magnetic MWCNTs was added to the Ni:ZnS nanoparticles solution synthesized according to the previously given procedure³⁸ and was mixed at 80 °C for about 30 min. The mixture was left at 80 °C for 12 h to complete the reaction, and the Ni:ZnS-MMWCNTs were filtered, washed extensively with double distilled water and dried at 100 °C for 12 h. The successful coating of Ni:ZnS nanoparticles (NPs) on the surface of MMWCNTs was confirmed by comparing the Transmission Electron Microscopy (TEM) images of the MWCNTs and Ni:ZnS-MMWCNTs (Fig. 1a and b).

Fig. 1 TEM image of (a) MMWCNT and (b) Ni:ZnS-MMWCNT.

2.5. Sample preparation

Urine and plasma samples were obtained from healthy volunteers not taking any drugs within 4 months. The samples were stored in the refrigerator at −20 °C prior to analysis. The frozen samples were left to thaw at room temperature and centrifuged at 5000 rpm for 10 min in order to precipitate the white lipids. The supernatant fractions were filtered through a 0.45 μm filter. Then, the urine (5.0 mL) and plasma (2.0 mL) samples were diluted with deionized water at volume ratios of 1 : 2 and 1 : 5 respectively and were treated by DMSPE process as described in Section 2.6. Water samples (tap and well water) were filtered through a 0.45 μm Millipore filter and then stored in glass containers at 4 °C prior to analysis. Next, the pH of water samples was adjusted to ∼7.5, and treated according to the DMSPE procedure.

2.6. DMSPE procedure

The prepared Ni:ZnS-MMWCNTs sorbent (11.5 mg) was added to 10.0 mL of the sample or the standard solution containing not more than 4 μg of PHB or PHN with pH of 7.5. The mixture was shaken by vortex for 8.0 min and after the completion of the extraction, the magnetic sorbent was isolated from the solution by the application of an external magnetic field. The supernatant solution was decanted and the extracted analytes were desorbed from the sorbent upon the addition of 0.75 mL of methanol and sonical agitation for 7.0 min. Subsequently, the sorbent was held by an external magnet and the supernatant was transferred into a glass vial. The eluent was evaporated under a mild nitrogen stream and the residue was dissolved in 50.0 μL of methanol. Finally, 20.0 μL of resultant solution was injected into the HPLC-UV system for the separation and quantification of the analytes.

3. Results and discussion

In the preliminary experiments, the capability of the Ni:ZnS-MMWCNT for the extraction of PHB and PHN was investigated and was found that the synthesized sorbent has high affinity for the analytes. This is due to the interaction of sorbent with analytes through the electrostatic interaction with metallic site, π electrons of PHB and PHN with reactive site of S in ZnS or the NH groups of analytes with Zn site of Ni:ZnS nanoparticles.³⁸ Furthermore, the effect of Ni:ZnS nanoparticles on extraction capability of sorbent was confirmed by carrying the same procedure with magnetic MWCNTs. The results revealed that with magnetic MWCNT the recoveries of analytes were much lower than with Ni:ZnS-MMWCNT, (∼20–30% of Ni:ZnS-MMWCNT). Then, the experiments were performed to select the appropriate eluent. For this purpose, 1.0 mL of each of the three kinds of organic solvents including methanol, ethanol and acetone was examined and the extraction recovery (ER%) was found to be as follows: methanol (80.5 ± 4.5%), acetone (68.3 ± 3.4%), and ethanol (58.2 ± 5.5%). According to these results, the methanol was chosen as the eluent in the subsequent experiments. The experimental factors affecting the ER of the analytes by DMSPE were investigated and optimized using the Plackett–Burman (P–B) design and the response surface methodology (RSM).

3.1. Plackett–Burman design (P–B)

Since the extraction procedure depends on numerous factors, carrying out the optimization via one-factor at-a-time (OFAT) of the classical optimization approach seems tedious and time consuming. Thus, the P–B design was applied for the determination of the main factors that affect the ER% of the analytes through the DMSPE method to alleviate this problem.⁴⁰ The P–B design allows examining at most f = N − 1 factor in N experiments, where N is a multiple of four (N = 8, 12, 16, 20…). Each independent factor is tested at two levels of high and low which are denoted by (+) and (−), respectively. The experimental design with the factors name, symbol code, and actual level of the factors is provided in Table 1.

Table 1 Factors, symbols and levels in 2⁷⁻⁴ P–B design matrix^a

Factors	Levels
	Low (−1)	Center point (0)	High (+1)
a C: center point.
(X1) pH of sample solution	3	6	9
(X2) adsorbent (mg)	5	10	15
(X3) vortex time (min)	2	6	10
(X4) desorption volume (mL)	0.5	0.75	1
(X5) ultrasonic time (min)	2	5	8
(X6) ultrasonic temperature (°C)	20	30	40
(X7) ionic strength (NaCl% (w/v))	0	3	6

---

Run	X1	X2	X3	X4	X5	X6	X7	ER%
1	9	5	2	1	2	20	6	29.58
2	3	5	10	0.5	8	20	6	47.56
3	3	15	2	1	8	20	0	60.23
4	9	15	10	0.5	2	20	0	80.34
5	9	5	2	0.5	8	40	0	34.53
6	3	5	10	1	2	40	0	42.85
7	3	15	2	0.5	2	40	6	50.35
8	9	15	10	1	8	40	6	89.73
9 (C)	6	10	6	0.75	5	30	3	76.56
10 (C)	6	10	6	0.75	5	30	3	77.53
11 (C)	6	10	6	0.75	5	30	3	76.15
12 (C)	6	10	6	0.75	5	30	3	75.56

---

The experiments were run in a random manner to minimize the effect of the uncontrolled variables.⁴⁰ The effects of the studied factors in the screening experiment have been presented in the form of a Pareto chart in Fig. 2 where, the bar length in this figure is proportional to the significance of each factor on ER of PHB and PHN by DMSPE method. The result indicates that the amounts of sorbent (Ni:ZnS-MMWCNTs; mg), pH, vortex and ultrasonic time (min) were the most important parameter affecting the ER of PHB and PHN. Therefore, these factors were evaluated by the CCD for further assessment. Furthermore, the other factors including the desorption volume, ionic strength and ultrasonic temperature had no significant effect on the ER of the analytes and were eliminated from further consideration. In the subsequent experiments desorption volume, ionic strength and ultrasonic temperature were set at 0.75 mL, 0% and 25 °C, respectively.

Fig. 2 Standardized main effect Pareto chart for the P–B of screening experiment. Vertical line in the chart defines 95% confidence level.

3.2. Central composite design (CCD)

The chosen factors by P–B design, the pH (X₁), the amount of Ni:ZnS-AC nanoparticles (X₂), the vortex time (X₃) and the ultrasonic time (X₄) were then optimized using a CCD. The CCD consisted of a combination of a factorial design (2^f), a star design (2f) and the center of each factor.³⁸ The central points are used to determine the experimental error and the reproducibility of the data. The experimental design points with factors coded values used in matrix of experiments (−2, −1, 0, +1, +2) are given in Table 2. The total number of the design points (N) required has been predicted by the following equation:⁴¹

N = 2^f + 2f + N (1)

Table 2 Design matrix for the 2⁴ central composite designs^a

Factors	Levels
	−α (−2)	Low (−1)	Central (0)	High (+1)	+α (+2)
a ER%: extraction recovery percentage, (C): center point.
(X1) pH value	2	4	6	8	10
(X2) adsorbent (mg)	5	8	11	14	17
(X3) vortex time (min)	3	5	7	9	11
(X4) ultrasonic time (min)	2	4	6	8	10

---

Run	(X1)	(X2)	(X3)	(X4)	(ER%)
1	4	8	5	4	37.6
2	4	8	5	8	38.7
3	4	8	9	4	70.4
4	4	8	9	8	69.5
5	4	14	5	4	62.7
6	4	14	5	8	61.7
7	4	14	9	4	75.2
8	4	14	9	8	78.4
9	8	8	5	4	63.2
10	8	8	5	8	69.4
11	8	8	9	4	84.6
12	8	8	9	8	88.5
13	8	14	5	4	81.4
14	8	14	5	8	86.7
15	8	14	9	4	78.8
16	8	14	9	8	94.6
17	2	11	7	6	51.4
18	10	11	7	6	87.8
19	6	5	7	6	64.3
20	6	17	7	6	87.5
21	6	11	3	6	57.4
22	6	11	11	6	90.3
23	6	11	7	2	67.6
24	6	11	7	10	80.8
25 (C)	6	11	7	6	87.7
26 (C)	6	11	7	6	86.4
27 (C)	6	11	7	6	86.8
28 (C)	6	11	7	6	87.5

---

Subsequently, the obtained results were examined by the analysis of variance (ANOVA) to identify the main effect of factors and their interactions (Table 3). The plot of tridimensional graph led to the generation of surface response which was used for the prediction of the best operating conditions according to the P and F-value. The influence of the four selected factors on the ER of PHB and PHN was investigated through the consideration of RSM. The results of ANOVA and regression coefficients (Table 3) indicated that the contribution of the quadratic model was significant (P < 0.05 at 95% confidence level). In order to evaluate the adequacy of the model, the determination coefficient (R² = 0.989) and the lack of fit (LOF = 0.273) were also calculated. The data analysis of the results, according to CCD, gave the following equation for ER% of PHB and PHN:

ER% = 87.05 + 9.41X₁ + 6.00X₂ + 8.52X₃ + 2.50X₄ − 4.70X₁² − 3.13X₂² − 3.64X₃² − 3.55X₄² − 2.94X₁X₃ − 4.35X₂X₃ (2)

Table 3 Analysis of variance (ANOVA) for central composite design

Source of variation	Sum of square	Df^a	Mean square	F-Value^b	P-Value
a Df: degrees of freedom.b Test for comparing model variance with residual (error) variance.
X1	2124.402	1	2124.402	2514.085	0.012695
X2	864.000	1	864.000	1022.485	0.019903
X3	1740.807	1	1740.807	2060.126	0.014024
X4	150.000	1	150.000	177.515	0.047692
X1^2	386.255	1	386.255	457.106	0.029755
X2^2	170.909	1	170.909	202.260	0.044690
X3^2	231.478	1	231.478	273.938	0.038417
X4^2	220.488	1	220.488	260.932	0.039361
X1X2	42.250	1	42.250	50.000	0.089439
X1X3	138.063	1	138.063	163.388	0.049703
X1X4	51.840	1	51.840	61.349	0.080841
X2X3	302.760	1	302.760	358.296	0.033601
X2X4	10.563	1	10.563	12.500	0.175480
X3X4	6.760	1	6.760	8.000	0.216347
Lack of fit	65.373	10	6.537	7.736	0.273323
Pure error	0.845	1	0.845
Total SS	5957.726	25

---

The plots of the predicted ER% and raw residuals versus the observed and predicted ER% are shown in Fig. 3a and b, respectively. Close inspection of these figures reveals that the predicted ER% values generally fall on a straight line (Fig. 3a) whereas, the residual versus predicted response follows no obvious pattern (Fig. 3b). These plots are required for checking the normal distribution of errors.

Fig. 3 (a) Plot of predicted value vs. observed value for extraction of PHB and PHN. (b) Plot of residuals versus predicted response for extraction of PHB and PHN.

Then, for optimizing the main factors (pH, amount of adsorbent, vortex and ultrasonic time) and obtaining the maximum ER%, the response surface was drawn by CCD study. The results demonstrated in Fig. 4 depict the most relevant fitted response surface plots of ER% versus the affecting factors. The curvatures of these plots demonstrate the interaction between the optimized factors of the DMSPE for PHB and PHN. Due to the chemical structure of PHB and PHN, the pH value plays an important role in their interaction with the sorbent surfaces. The results of Fig. 4a–c clearly show that the extraction efficiency for PHB and PHN increases with an increase in the pH of solution and reaches a plateau at a pH of 7–8, respectively. This could be explained by the pK_a values of PHB (pK_a1 = 7.30, pK_a2 = 11.8) and PHN (8.33). Thus, when the pH is between 7 and 8 the neutral forms of both analytes dominate and can easily extract into the sorbent.

Fig. 4 Response surfaces for the 2⁴ central composite designs: (a) adsorbent (mg)–pH; (b) vortex time (min)–pH; (c) ultrasonic time (min)–pH; (d) vortex time (min)–adsorbent (mg); (e) ultrasonic time (min)–adsorbent (mg) and (f) ultrasonic time (min)–vortex time (min).

The surface plots of Fig. 4 also show that with low amount of sorbent, along with low vortex and ultrasonic time, the ER is low. This may be related to the fact that the surface areas of the sorbent are not sufficient for the sorption of analytes and the vortex and ultrasonic time are not adequate for complete extraction or desorption process. According to the Fig. 4a–f, the ER% of PHB and PHN are maximized in the regions where the pH value, amount of sorbent, vortex time and ultrasonic time are 7–9, 10–13 mg, 7–9 min and 6–8 min respectively. Thus, an optimum pH of 7.5, amount of sorbent of 11.5 mg, vortex time of 8 min and ultrasonic time of 7 min were selected for further experiments. Finally, triplicate experiments were conducted using the optimized factors for validation and the ER% was found to be 95.3 ± 2.3% which indicated that the CCD is suitable for optimization of extraction condition for PHB and PHN.

3.3. Analytical performance

Under the optimized conditions, the analytical features of the developed method consisting linearity (correlation coefficient (r²) and linear range (LR)), limit of quantification (LOQ), limit of detection (LOD), and precision (RSD%) were evaluated. LOD and LOQ were calculated as 3 and 10 times of the standard deviation of blank signal divided by the slope of the calibration curve, respectively. The calibration curves were constructed by simultaneous extraction and determination of PHB and PHB at seven concentration levels. Good linearity was observed over the range of 2–400 μg L⁻¹ and 3–500 μg L⁻¹ with the correlation coefficient of 0.9996 and 0.9991 for PHB and PHN, respectively. The LODs and LOQs were found to be 0.5 and 1.8 μg L⁻¹ for PHB and 0.8 and 2.8 μg L⁻¹ for PHN, respectively. The precision expressed as the relative standard deviations (RSDs) was determined by the analysis of the standard samples containing 10 μg L⁻¹ of PHB and PHN (n = 6) which were found to be 3.7 and 4.5% for PHB and PHN, respectively. The stability and reusability of the Ni:ZnS-MMWCNT were also evaluated. For this purpose, after elution the analytes from the magnetic sorbent, it was reused for the same process in a subsequent cycle. The experimental results showed that the extraction efficiency remained approximately constant after 8 cycles with RSD% of 3.1%, indicating that the Ni:ZnS-MMWCNT had excellent stability and reusability for repeated usages. The preconcentration factor defined as the ratio of the maximum volume of initial solution (10.0 mL) to the final extraction volume (50.0 μL) was found to be 200 for the proposed method.

3.4. Validation of the method

To investigate the applicability of the developed extraction method, water, plasma and urine samples were analyzed via the DMSPE method, the water samples were then spiked at three concentration levels of PHB and PHN (5, 15 and 25 μg L⁻¹) and the accuracy of the method was evaluated through the recovery experiments. For each concentration, three replicate experiments were performed and the results were averaged. The results summarized in Table 4 indicate that the ER% for the spiked samples is good confirming the applicability of the method to the sample type examined. Furthermore, the repeatability and reproducibility of the DMSPE method were investigated by intra- and inter-day analysis of the spiked urine and the plasma samples. The intra and inter-day RSDs were determined by five replicate analyses of the spiked samples at three different concentration levels (low, middle, and high) in one day as well as five consecutive days. The results (Table 5) clearly demonstrate that the DMSPE method has suitable precision and accuracy in the analysis of real samples.

Table 4 Extraction recoveries of PHB and PHN in spiked water samples extracted with the DMSPE method (n = 3)

Analyte	Spiked (μg L^−1)	Tap water		Well water
		Found (μg L^−1)	ER ± RSD (%)	Found (μg L^−1)	ER ± RSD (%)
PHB	—	—	—	—	—
	5.0	4.76	95.2 ± 3.1	4.67	93.4 ± 4.3
	15.0	14.36	95.7 ± 2.8	14.23	94.9 ± 3.8
	25.0	24.12	96.5 ± 2.5	23.85	95.4 ± 2.6
PHN	—	—	—	—	—
	5.0	4.71	94.2 ± 4.7	4.64	92.8 ± 5.6
	15.0	14.50	96.7 ± 3.4	14.16	94.4 ± 4.0
	25.0	24.20	96.8 ± 2.3	23.76	95.0 ± 3.3

---

Table 5 ER%, inter- and intra-day precision of blank urine and plasma spiked with PHB and PHN after DMSPE

Analyte	Spiked (μg L^−1)	Urine				Plasma
		Inter-day		Intra-day		Inter-day		Intra-day
		Mean found (μg L^−1)	ER% ± RSD%	Mean found (μg L^−1)	ER% ± RSD%	Mean found (μg L^−1)	ER% ± RSD%	Mean found (μg L^−1)	ER% ± RSD%
PHB	5.0	4.64	92.8 ± 4.5	4.57	91.4 ± 5.3	4.58	91.6 ± 4.3	4.52	90.4 ± 5.6
	15.0	14.45	96.3 ± 3.4	14.25	95.0 ± 4.8	14.38	95.9 ± 3.7	14.20	94.6 ± 4.4
	25.0	24.15	96.6 ± 2.9	23.70	94.8 ± 3.6	23.80	95.2 ± 3.5	23.35	93.4 ± 4.1
PHN	5.0	4.68	93.6 ± 4.8	4.60	92.0 ± 5.1	4.56	91.2 ± 4.6	4.50	90.0 ± 5.4
	15.0	14.35	95.7 ± 4.0	14.15	94.3 ± 4.5	14.25	95.0 ± 3.7	14.13	94.2 ± 4.2
	25.0	24.24	97.0 ± 3.5	24.00	96.0 ± 4.0	24.04	96.2 ± 3.8	23.65	94.6 ± 4.6

---

Furthermore, the chromatogram of standard (prepared in mobile phase), blank and spiked urine samples after the extraction by DMSPE method are shown in Fig. 5. It clearly indicates that the matrix of the extracted sample has no significant effect on the chromatogram behavior of PHB and PHN and the retention time of both analytes in standard and real samples is the same. This further confirmed the suitability of the method for TDM analysis. To demonstrate the superiority of the presented method, a comparison between the important analytical parameters of the developed DMSPE and those reported in the literature was made as summarized in Table 6.^9–11,35 The results revealed that the developed DMSPE-HPLC-UV exhibits suitable dynamic range, good repeatability (RSD%) and a detection limit which is almost by a factor of ten lower than other methods, and can improve the TDM analysis.

Fig. 5 Chromatograms of extracted PHB (1) and PHN (2) after DMSPE-HPLC-UV at optimum extraction condition from standard solution (a), blank urine (b) and spiked urine at 5 (c) and 15 (d) μg L⁻¹.

Table 6 Comparison of the DMSPE method with other methods for determination of the PHB and PHN

Method	Analyte	Matrix	LDR (μg L^−1)	LOD (μg L^−1)	RSD%	Ref
a Molecularly imprinted solid-phase extraction.
SPE-HPLC-UV	PHB	Plasma	250 to 10^5	25	12.15	9
	PHN		500 to 50 × 10^3	50
LLE-HPLC-DAD	PHB	Serum	—	110	5.1	10
MISPE^a-HPLC-UV	PHN	Plasma	2500 to 40 × 10^3	—	<15.0	11
SPE-HPLC-DAD	PHB	Plasma	4 × 10^3 to 40 × 10^3	60	<7.5	35
	PHN		4 × 10^3 to 40 × 10^3	80	<10.3
DMSPE-HPLC-UV	PHB	Urine	2.0–400	0.5	<5.0	This work
	PHN	Plasma	3.0–500	0.8

---

4. Conclusion

In this research, Ni:ZnS nanoparticles were loaded on magnetized multi-walled carbon nanotubes and Ni:ZnS-MMWCNTs were prepared and introduced as sorbent in DMSPE method for simultaneous separation and preconcentration of two antiepileptic drugs from biological samples for the first time. In the first stage of DMSPE optimization, the relative influence of the seven factors was considered by a Plackett–Burman experimental design which minimized the experimental effort in finding the most significant variables in complex systems. Then, the curvature of the response surface and the accurate position of the optimum conditions were evaluated by means of central composite design. At optimum conditions values of variables, the calibration graphs were found to be linear over the wide concentration ranges of 2–400 μg L⁻¹ and 3–500 μg L⁻¹ with the limit of detections of 0.5 and 0.8 μg L⁻¹ for PHB and PHN, respectively, which is almost by a factor of ten lower than other methods, and can improve the TDM analysis. Thus, combination of DMSPE with HPLC-UV in the developed method provides a safe, easy and time saving procedure for monitoring the PHB and PHN in water and biological samples and can be a candidate in clinical laboratory for TDM analysis.

Conflict of interest

The authors report no conflicts of interest in this work.

Acknowledgements

We are grateful to Yazd University and Behbahan Khatam Alanbia University of Technology for the financial support.

References

1. J. Kuhn and C. Knabbe, Talanta, 2013, 110, 71–80.
2. H. Levert, P. Odou and H. Robert, Biomed. Chromatogr., 2002, 16, 19–24.
3. C. L. Deckers, S. J. Czuczwar, Y. A. Hekster, A. Keyser, H. Kubova, H. Meinardi, P. N. Patsalos, W. O. Renier and C. M. Van Rijn, Epilepsia, 2000, 41, 1364–1374.
4. F. Bugamelli, C. Sabbioni, R. Mandrioli, E. Kenndler, F. Albani and M. A. Raggi, Anal. Chim. Acta, 2002, 472, 1–10.
5. E. Yukawa, Clin. Pharmacokinet., 1996, 31, 120–130.
6. S. I. Johannessen, CNS Drugs, 1997, 7, 349–365.
7. J. M. Juenke, J. P. McGraw, G. A. McMillin and K. L. Johnson-Davis, Clin. Chim. Acta, 2012, 413, 529–531.
8. N. Rahman and M. Hoda, Anal. Bioanal. Chem., 2002, 374, 484–489.
9. A. Serralheiro, G. Alves, A. Fortuna, M. Rocha and A. Falcão, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2013, 925, 1–9.
10. M. Vosough, S. Ghafghazi and M. Sabetkasaei, Talanta, 2014, 119, 17–23.
11. A. Bereczki, A. Tolokan, G. Horvai, V. Horvath, F. Lanza, A. John Hall and B. Sellergren, J. Chromatogr. A, 2001, 930, 31–38.
12. M. Imazawa and Y. Hatanaka, J. Pharm. Biomed. Anal., 1997, 15, 1503–1508.
13. A. Haque, X. H. Xu and J. T. Stewart, J. Pharm. Biomed. Anal., 1999, 21, 1063–1067.
14. K. B. Borges, E. F. Freire, I. Martins and M. E. Pereira Bastos de Siqueira, Talanta, 2009, 78, 233–241.
15. A. Esrafili, Y. Yamini and S. Shariati, Anal. Chim. Acta, 2007, 604, 127–133.
16. S. S. C. Tai, C. Y. Yeh and K. W. Phinney, Anal. Bioanal. Chem., 2011, 401, 1915–1922.
17. F. Rezaei, Y. Yamini, M. Moradi and B. Ebrahimpour, Talanta, 2013, 105, 173–178.
18. S. Farzaneh, E. Asadi, M. Abdouss, A. Barghi-Lish, S. Azodi-Deilami, H. A. Khonakdar and M. Gharghabi, RSC Adv., 2015, 5, 9154–9166.
19. A. Amiri Pebdani, A. M. Haji Shabani, S. Dadfarnia and S. Khodadoust, Spectrochim. Acta, Part A, 2015, 147, 26–30.
20. M. R. Hadjmohammadi and H. Ghambari, J. Pharm. Biomed. Anal., 2012, 61, 44–49.
21. H. R. Sobhi, Y. Yamini and R. Haji Hosseini Baghdad Abadi, J. Pharm. Biomed. Anal., 2007, 45, 769–774.
22. F. L. Roveri, B. A. P. B. Paranhos and M. Yonamine, Forensic Sci. Int., 2015, 265, 75–80.
23. M. Asadi, S. Dadfarnia, A. M. Haji Shabani and B. Abbasi, J. Sep. Sci., 2015, 38, 2510–2516.
24. A. Amiri Pebdani, A. M. Haji Shabani, S. Dadfarnia, M. S. Talebianpoor and S. Khodadoust, J. Sep. Sci., 2016, 39, 1935–1944.
25. Z. Es’haghi, Z. Rezaeifar, G. H. Rounaghi, Z. Alian Nezhadi and M. Ahmadi Golsefidi, Anal. Chim. Acta, 2011, 689, 122–128.
26. R. Helena, C. Queiroz, C. Bertucci, W. R. Malfara, S. A. C. Dreossi, A. R. Chaves, D. A. R. Valerio and M. E. C. Queiroz, J. Pharm. Biomed. Anal., 2008, 48, 428–434.
27. A. A. Asgharinezhad, N. Mollazadeh, H. Ebrahimzadeh, F. Mirbabaei and N. Shekari, J. Chromatogr. A, 2014, 1338, 1–8.
28. N. N. Naing, S. F. Y. Li and H. K. Lee, J. Chromatogr. A, 2015, 1426, 69–76.
29. S. Khodadoust, M. Ghaedi and M. R. Hadjmohammadi, Talanta, 2013, 116, 637–646.
30. A. Amiri Pebdani, S. Khodadoust, M. Akbartabar Toori, V. Zarezade and M. S. Talebianpoor, RSC Adv., 2016, 6, 36238–36247.
31. Y. G. Zhao, X. H. Chen, S. D. Pan, H. Zhu, H. Y. Shen and M. C. Jin, Talanta, 2013, 115, 787–797.
32. A. Amiri Pebdani, A. M. Haji Shabani and S. Dadfarnia, J. Iran. Chem. Soc., 2016, 13, 155–164.
33. J. Casado, G. Castro, I. Rodríguez, M. Ramil and R. Cela, Anal. Bioanal. Chem., 2015, 407, 907–917.
34. W. H. Chung, S. H. Tzing and W. H. Ding, J. Chromatogr. A, 2013, 1307, 34–40.
35. F. Bugamelli, C. Sabbioni, R. Mandrioli, E. Kenndler, F. Albani and M. A. Raggi, Anal. Chim. Acta, 2002, 72, 1–10.
36. Q. Zhou and Z. Fang, RSC Adv., 2014, 4, 7471–7475.
37. G. D. Tarigh and F. Shemirani, Talanta, 2015, 144, 899–907.
38. A. Amiri Pebdani, S. Dadfarnia, A. M. Haji Shabani, S. Khodadoust and S. Haghgoo, J. Chromatogr. A, 2016, 1437, 15–24.
39. M. Amoli-Diva, K. Pourghazi and S. Hajjaran, Mater. Sci. Eng., C, 2016, 60, 30–36.
40. B. Dejaegher and Y. Vander Heyden, J. Pharm. Biomed. Anal., 2011, 56, 141–158.
41. M. Ghaedi, S. Khodadoust, H. Sadeghi, M. A. Khodadoust, R. Armand and A. Fatehi, Spectrochim. Acta, Part A, 2015, 136, 1069–1075.

---