Magnetic graphene-based cyanopropyltriethoxysilane as an adsorbent for simultaneous determination of polar and non-polar organophosphorus pesticides in cow’s milk

Abstract

A new adsorbent based on magnetic nanoparticles (Fe₃O₄), graphene (G) and cyanopropyltriethoxysilane (CNPrTEOS) was fabricated and applied in the magnetic solid phase extraction (MSPE) of organophosphorus pesticides. The synthesized adsorbent was characterized using Fourier transform infrared spectroscopy, scanning electron microscopy, energy-dispersive X-ray spectroscopy and transmission electron microscopy. The Fe₃O₄@G–CNPrTEOS MSPE was used for the simultaneous preconcentration of four polar and non-polar organophosphorus pesticides namely phosphamidon, dimethoate, diazinon and chlorpyrifos in fresh cow’s milk. Under optimum conditions, Fe₃O₄@G–CNPrTEOS MSPE shows good linearity in the concentration range of 0.1–200 ng mL⁻¹, low limits of detection (0.01–0.6 ng mL⁻¹) and high enrichment factors (2400). The precision of the MSPE method was investigated using intraday (% RSD 1.7–9.4%, n = 3) and interday (% RSD 10.7–17.6%, n = 15) studies. The synthesized adsorbent can be reused at least 10 times without significant loss of extraction efficiency. The Fe₃O₄@G–CNPrTEOS MSPE method offers good recovery (82–94%) for spiked milk samples as compared to commercial CNPr-SPE (42–59%) and C18-SPE (58–68%). The Fe₃O₄@G–CNPrTEOS MSPE method involves only dilution of the cow’s milk analysed without any treatment prior to analysis, and fewer interferences from protein fractions were observed.

---

1. Introduction

Organophosphorus pesticides (OPPs) are widely used in agriculture to control insects. They can be transferred to cow’s milk as a result of veterinary treatment against ectoparasite insects, inhalation or ingestion via grass/corn silage feed or the water supply.^1–3 OPP residues have been reported in several raw and commercial pasteurized milk samples in Italy, Mexico and Spain.^1,3–5 The presence of organochlorine pesticides has also been reported in cow’s milk.⁶ The presence of OPP residues in milk is a significant risk for human health since milk is widely drank by adults, children and infants.³ The maximum residual level (MRL) for OPPs in milk is between 0.01 mg kg⁻¹ to 0.05 mg kg⁻¹ as set by the European Union (EU) Commission 396/2005 (pesticides database, product milk).⁷ Due to the low MRL, a sensitive and accurate analytical method is imperative. Sample preparation is the critical stage in an analytical process (i.e., extraction, isolation and preconcentration) prior to instrument analysis.⁸ Due to the existence of complex matrices in biological samples, selective and cost effective sample preparation techniques are essential.^9–11

Graphene provides great potential benefits in analytical sample preparation. Graphene is likely to be a suitable sorbent for the removal/adsorption of organic compounds due to its high stability, large surface area, high adsorption capacity, rich sp² bonding and large delocalized π–π electrons.^12–14 An exceptional advantage of graphene is that both sides of the graphene planes are available for adsorption compared to other carbon-based nanomaterials. The hydrophobic properties of graphene also lead to high absorption affinity of non-polar compounds. Graphene has been shown to adsorb pesticides from water, juice, fruit, vegetables and oil.^15–23 Graphene was successfully used as a solid phase extraction (SPE) sorbent for phthalate and polycyclic aromatic hydrocarbons extraction from soy bean milk and dairy milk with high recovery and less interference.^24,25 In order to increase the graphene selectivity and sensitivity, it was modified using magnetic nanoparticles (MNPs), amine, composite nanoadsorbent, silica, titanium, alumina, polymers, membrane, carbon and carbon nanotubes (CNTs).^26–34 Even though graphene-based SPE methods provide great adsorption potential, the technique is tedious, time-consuming and there is a loss of graphene weight.

In 1999, SPE based on MNPs was developed for sample preparation which is called magnetic solid phase extraction (MSPE).³⁵ MSPE is fast, requires small amounts of solvent and is easy to separate the adsorbent from the sample by using an external magnet.^31–35 MSPE has been used successfully to preconcentrate analytes from biological samples.^33,34 For example, Fe₃O₄-functionalized C18 was used as an adsorbent for the extraction of aromatic amines from urine,³⁶ and Fe₃O₄/CNTs have been used to determine estrogen in milk.²⁸ Fe₃O₄–molecularly imprinted polymer (MIP) has also been used for the adsorption of melamine from milk,³² and Fe₃O₄@SiO₂–MIPs for estrogen preconcentration from human plasma.³⁷ Puerarin was extracted from rat plasma using Fe₃O₄@SiO₂–C18.³⁸

In this study, the combination of MNPs for ease of separation, and graphene’s large surface area for enhanced adsorption, were exploited to produce Fe₃O₄–graphene (Fe₃O₄–G) and then further modified with polar cyanopropyltriethoxysilane (CNPrTEOS) to enhance the capability for simultaneous preconcentration of polar and non-polar OPPs. First, Fe₃O₄@G was synthesized using a one-pot method, followed by reduction using lithium aluminum hydride and sodium borohydride. Then, CNPrTEOS was added onto the Fe₃O₄@G using a simple sol–gel process to produce Fe₃O₄@G–CNPrTEOS. The newly synthesized adsorbent was applied for the first time as an MSPE adsorbent for the simultaneous preconcentration of polar and non-polar OPPs from fresh cow’s milk samples.

The logarithm of the octanol–water partition coefficient, log K_O/W, identifies the polarity indices of organic compounds; log K_O/W value for polar compounds is <2.0, mid-polar is 2.0–3.0 and non-polar is >3.0.³⁹ The log K_O/W values of the OPPs studied are shown in Table 1.³⁹ Both polar and non-polar OPPs were successfully extracted simultaneously using the newly synthesized Fe₃O₄@G–CNPrTEOS due to the presence of both hydrophilic (–C≡N) and hydrophobic (graphene) properties. The proposed Fe₃O₄@G–CNPrTEOS MSPE method shows fast extraction, high recovery and high enrichment factors for the selected OPPs. Previous studies have reported that the fat and protein matrix in milk samples should be treated with acid in order to remove its effects.^40,41 However, in our study, we found that the as prepared Fe₃O₄@G–CNPrTEOS can be used directly to analyse fresh cow’s milk using simple dilution with deionized water without further sample preparation.

Table 1 The chemical structure of selected OPPs with their log K_O/W values and MRLs

OPPs name	Structure	log KO/W^39	MRLs in milk (mg kg^−1)
Phosphamidon		0.3	0.05
Dimethoate		0.7	0.05
Diazinon		3.8	0.02
Chlorpyrifos		4.7	0.01
Hexaconazole (internal standard)		3.6	0.02

---

2. Experimental

2.1 Materials

AR-grade acetone, HPLC grade methanol, ethanol (97%), sulfuric acid (98%) and nitric acid (65%) were obtained from QReC (Selangor, Malaysia). Acetonitrile, hydrogen peroxide, potassium permanganate and ammonia (32%) were from Merck (Schuchardt, Germany). Graphite powder and CNPrTEOS were obtained from Sigma Aldrich (St. Louis, MO, USA). Chlorpyrifos, hexaconazole, lithium aluminum hydride (LiAlH₄) and sodium borohydride (NaBH₄) were from Riedel-de Haen (Sleeze, Germany). Pestanal grade diazinon, dimethoate and phosphamidon, iron(III) chloride hexahydrate (FeCl₃·6H₂O) and ammonium iron(II) sulfate hexahydrate ((NH₄)₂Fe(SO₄)₂·6H₂O) were from Fluka (Hanover, Germany). Appropriate amounts of the selected OPPs (Table 1) were weighed and dissolved in 10 mL HPLC-grade methanol to produce a stock solution of 1000 mg L⁻¹. This was kept in a refrigerator until use. Commercial CNPr-SPE and C18-SPE cartridges (100 mg) were obtained from Supelco (PA, USA).

Fresh cow’s milk samples were obtained directly from the transfer truck tank from Johor Bahru, Malaysia, and diluted with deionized water prior to use.

2.2 Gas chromatography conditions

The four selected OPPs were determined using gas chromatography with micro-electron capture detector (GC-μECD). The column used was HP-5MS (30 m × 0.32 mm i.d. and 0.25 μm film thickness). Before operating the GC-μECD for OPPs analysis, oven temperature, temperature rate, inlet temperature, inlet gas flow rate, detector temperature and make up gas (nitrogen) volume were optimized. The optimum conditions obtained were as follows: nitrogen gas as back inlet carrier gas at a flow rate of 1 mL min⁻¹ and also back detector make-up gas (5 mL min⁻¹). The temperature was set at 280 °C and 300 °C for back inlet and back detector, respectively. The initial oven temperature was set at 70 °C (0 min) and it was then increased at a rate of 50 °C min⁻¹ to 180 °C (held for 1 min) and ramped at 30 °C min⁻¹ to 280 °C (held for 1 min).

2.3 Instruments

Deionized water was obtained from a Milli-Q system from Millipore (Bedford, MA, USA). The four OPPs were analyzed using an Agilent 7600A GC-μECD from Agilent Technologies Inc. (Santa Clara, CA, USA). A 1600 series Perkin-Elmer Spectrophotometer (MA, USA) was used for FTIR spectra recording in transmission mode in the range of 400–4000 cm⁻¹. The scanning electron microscope equipped with energy-dispersive X-ray spectroscopy (SEM/EDX) model JSM6390 from JEOL (Tokyo, Japan) was used for morphology and composition studies of the synthesized adsorbent. The transmission electron microscope (TEM) images were obtained using a JEM-2100 electron microscope from JEOL Ltd. (Tokyo, Japan) operated at 200 kV.

2.4 Synthesis of magnetic graphene-based cyanopropyltriethoxysilane

Natural graphite (1 g) was dispersed in 50 mL concentrated HNO₃/H₂SO₄ (1 : 3 v/v) mixture followed by slow addition of 3 g of KMnO₄ with stirring for 24 h to give a brown product. Then the mixture was poured into 200 g of ice followed by addition of 5 mL of H₂O₂ and a yellow product (graphene oxide, GO) was formed. To obtain graphene from GO, LiAlH₄ (0.5 g) and NaBH₄ (0.5 g) were applied as reducing agents followed by heating at 90 °C and stirring for 5 h. A dark color product (graphene) was obtained and was diluted with deionized water (400 mL), centrifuged and oven dried at 100 °C for 24 h. Previously hydrazine was used as a reducing agent, although it exhibited high efficiency it is highly toxic and hazardous to the environment. The graphene obtained was dispersed in 30 mL deionized water and sonicated for 30 min, then it was added to FeCl₃·6H₂O (0.05 g) and (NH₄)₂Fe(SO₄)₂·6H₂O (0.1 g). The mixture was degassed by bubbling nitrogen for 20 min and then 5 mL of ammonia solution (25%) was added slowly into it. The solution was stirred for 2 h. The dark product (Fe₃O₄@G) was washed with 500 mL deionized water and collected with the assistance of an external magnet.

The sol–gel process was then used to functionalize the Fe₃O₄@G with cyano groups to produce Fe₃O₄@G–CNPrTEOS. About 0.5 g of the Fe₃O₄@G was dispersed in 50 mL of water : methanol mixture (1 : 5 v/v) and sonicated for 30 min. Then, 1.5 mL of ammonia solution (32%) was added followed by the addition of 0.5 mL of CNPrTEOS with continuous stirring for 30 min. The mixture was left at room temperature until it formed a gel. The product was washed sequentially with methanol (20 mL), acetone (50 mL) and water (100 mL) before final drying at 100 °C for 24 h. The schematic of Fe₃O₄@G–CNPrTEOS synthesis is shown in Fig. 1.

Fig. 1 Schematic procedure for the synthesis of Fe₃O₄@G–CNPrTEOS.

2.5 GC-μECD instrument calibration

The instrument calibration was performed using the internal standard (hexaconazole 100 ng mL⁻¹) method using five point calibration for each OPP standard. The concentration range for standard chlorpyrifos and diazinon were from 10 to 500 ng mL⁻¹ and for standard phosphamidon and dimethoate from 200 to 10 000 ng mL⁻¹ (see ESI 1†). Injections were performed in triplicate (n = 3) for each calibration point. Calibration curves were obtained by plotting (A_S/A_IS) versus (C_S/C_IS), where A_S is the peak area of standard, A_IS is the peak area of internal standard (IS), C_S is the standard OPPs concentration, and C_IS is the concentration of the IS. The instrument limit of detection (LOD) and limit of quantification (LOQ) were obtained from (3 × SD)/m and (10 × SD)/m, respectively,⁴² where SD is the standard deviation of the lowest standard concentration of OPPs.⁴³

2.6 Pesticides standard and cow’s milk sample preparation

Stock solutions of OPPs were prepared in methanol (1000 mg L⁻¹) and diluted with deionized water for analysis of cow’s milk samples. For the optimization of effective parameters, standard OPP solutions (0.1 ng mL⁻¹ of each standard chlorpyrifos and diazinon; 10 ng mL⁻¹ of each phosphamidon and dimethoate) were spiked into a cow milk sample. The spiked level is 5–100× lower than the MRL set by the EU. However, the milk sample (10 mL) was diluted to 100 mL with deionized water to decrease the matrix effects from fat and protein (1 : 10 v/v, milk : water). The spiked OPPs were then extracted from the diluted cow’s milk samples by using the MSPE method based on the newly synthesized Fe₃O₄@G–CNPrTEOS sorbent.

2.7 Optimization of effective parameters on extraction performance

2.7.1 MSPE optimization procedure. The effective parameters for MSPE (types of desorption solvent, volume of desorption solvent, desorption time, extraction time, amount of adsorbent and sample volume) were optimized using one variable at a time procedure. Initial experiments were carried out using 10 mg of adsorbent and 30 min of extraction time. Extracted analytes were desorbed using 1 mL of different solvents (i.e., methanol, acetone, acetonitrile, ethyl acetate and n-hexane) with the aid of agitation for 5 min. In the next step, all parameters were fixed and the volume of desorption solvent was changed from 0.5 mL to 5 mL. Finally the optimized conditions obtained were as follows: 1 mL acetonitrile as desorption solvent, 2 min extraction time (vortex assisted time), 1 min agitation time during desorption process, 120 mL of cow’s milk (diluted milk, 1 : 10 v/v) and 100 mg of adsorbent. Efficiency of the parameters was evaluated using relative peak area ratio (peak area of OPPs/peak area of IS).

2.7.2 SPE optimization procedure. Commercial CNPr-SPE and C18-SPE cartridges were selected for comparison with MSPE. CNPr-SPE is more suitable for extraction of polar compounds and C18-SPE is suitable for non-polar compounds and has been used by many researchers for OPPs analysis, respectively.^44,45 The SPE cartridges were placed in a 12-port SPE vacuum manifold from Supelco (PA, USA). Three effective parameters – solvent type, volume of solvent and sample volume – were optimized one at a time. The cow’s milk sample was passed through the cartridge at a flow rate of 1 mL min⁻¹. Initially, 10 mL of the prepared cow’s milk sample was passed through the CNPr-SPE column and then washed with deionized water (5 mL) before it was dried by passing air over it for 30 min. The same procedure was repeated with the C18-SPE column. Finally, the optimized conditions for CNPr-SPE and C18-SPE cartridge were obtained with 4 mL and 5 mL of diluted cow’s milk (1 : 10 v/v), respectively, and acetonitrile (2 mL for CNPr-SPE and 3 mL for C18-SPE) as the eluting solvent. Efficiency of the parameters was also evaluated using relative peak area ratio (peak area of OPPs/peak area of IS).

2.8 Magnetic solid phase extraction procedure

The synthesized adsorbent (100 mg) was dispersed in 120 mL of prepared cow’s milk. The milk solution was vortexed for 2 min and the adsorbent was collected using an external magnet. The sample solution was decanted and the adsorbent was washed with 5 mL deionized water and finally transferred to a 15 mL centrifuge tube. Thereafter, the extracted OPPs were desorbed from the adsorbent using 1 mL acetonitrile with the aid of 1 min agitation time. The solvent was dried under a gentle stream of nitrogen flow and reconstituted with a solution of 50 μL of 100 ng mL⁻¹ hexaconazole (internal standard prepared in methanol). 1 μL of the reconstituted solution was injected into GC-μECD for analysis. Fig. 2 shows the schematic of MSPE for cow’s milk preparation.

Fig. 2 Schematic process of MSPE milk analysis.

Conventional SPE was performed for comparison as follows; SPE cartridges were conditioned by passing 5 mL methanol, followed by 10 mL deionized water. Then, 4 mL and 5 mL of diluted cow’s milk (each spiked with 10 ng mL⁻¹ of standard OPPs) were each passed through the CNPr-SPE and C18-SPE cartridge, respectively. After washing with water and 30 min air drying, the adsorbed OPPs were eluted from the CNPr-SPE and C18-SPE column using 2 and 3 mL acetonitrile, respectively. The eluate was then dried using a gentle stream of nitrogen gas (∼30 min) and the dried eluate was reconstituted with 50 μL of 100 ng mL⁻¹ hexaconazole (internal standard) prepared in methanol before 1 μL was injected into the GC-μECD for analysis.

2.9 Method validation

The linearity was studied by replicating three analyses (n = 3) of the concentration range (0.1–10 ng mL⁻¹ of each chlorpyrifos and diazinon; 2–200 ng mL⁻¹ of each phosphamidon and dimethoate). The LOD, LOQ, and % recovery were calculated using eqn (1)–(3), respectively.⁴² The precision (% RSD) was performed using intra-day (repeatability) and inter-day (reproducibility) method. Three extractions (n = 3) were performed in one day and continued for five consecutive days (n = 15). The reusability of the as synthesized Fe₃O₄@G–CNPrTEOS was assessed for 10 consecutive adsorption–desorption cycles (extraction) using 100 mg of the adsorbent (10 ng mL⁻¹ of each OPPs). The used adsorbent was washed sequentially using 2 mL methanol, 3 mL acetonitrile and finally 5 mL deionized water after each extraction.where SD is the standard deviation of the lowest concentration⁴³ in the cow’s milk samples, C_f is the found OPP concentration after extraction which was obtained from the method calibration curve and C_i is the initial OPP concentration in milk samples.

2.10 Milk sample analysis

The field applicability of the developed MSPE, C18-SPE and CNPr-SPE methods were studied for cow’s milk analysis. Unspiked cow’s milk samples (blank) were used for the three methods. The sample used for MSPE was spiked with 1 ng mL⁻¹ of diazinon and chlorpyrifos and 20 ng mL⁻¹ of phosphamidon and dimethoate. The spiked level was 5 ng mL⁻¹ of diazinon and chlorpyrifos, and 100 ng mL⁻¹ of phosphamidon and dimethoate for both the C18 SPE and CNPr-SPE. Optimized MSPE conditions were applied for these three extraction methods. Precision of the extraction process for all three methods was tested by repeating 3 times (% RSD 1.7–9.4%, n = 3) and 15 times (% RSD 10.7–17.6%, n = 15) and % R was calculated from the average data.

3. Results and discussion

3.1 Characterization

FTIR spectra of GO, graphene and Fe₃O₄@G–CNPrTEOS are shown in Fig. 3. For GO (Fig. 3A), a broad band at 3400 cm⁻¹ indicates O–H stretching vibration related to water. Due to its variety of functional groups, GO shows different bands; a peak at 1716 cm⁻¹ indicates C=O stretching for carboxylic groups, and carboxylic stretching vibration relates to C–OH at 1120 cm⁻¹. Stretching vibrations for alkoxy (C–O) were observed at 1036 cm⁻¹ and epoxy signals were observed at 1200 cm⁻¹ and 820 cm⁻¹. A peak at 1310 cm⁻¹ indicates aromatic C=C. Graphene shows fewer peaks in comparison to GO (Fig. 3B); the lump of the functional groups was removed from the GO in the reaction, followed by reduction using LiAlH₄ and NaBH₄. The as synthesized Fe₃O₄@G shows a peak at 1310 cm⁻¹ related to aromatic C=C on graphene sheets and a peak at 584 cm⁻¹ reveals Fe–O in Fe₃O₄. The FTIR spectra for synthesized Fe₃O₄@G–CNPrTEOS (Fig. 3C) shows two extra peaks at 2242 cm⁻¹ and 1070 cm⁻¹, identified as –C≡N and Si–O, respectively. The peak at 2800 cm⁻¹ is from the methyl groups of CNPrTEOS.

Fig. 3 FTIR spectra of (A) GO, (B) Fe₃O₄@graphene, and (C) Fe₃O₄@G–CNPrTEOS.

The SEM morphology shows that CNPrTEOS nanoparticles were successfully modified on the graphene sheets (Fig. 4A). The TEM micrograph (Fig. 4B) shows the small Fe₃O₄ particles size and thin graphene sheets produced. Fig. 4C shows the EDX analysis of the synthesized adsorbent and the expected elements including C, O, N, Si and Fe were observed in the spectra.

Fig. 4 (A) SEM image, (B) TEM image and (C) EDX spectra of the as synthesized magnetic Fe₃O₄@G–CNPrTEOS adsorbent.

3.2 GC-μECD instrument calibration

Linearity for standard diazinon and chlorpyrifos was obtained in the concentration range 10–500 ng mL⁻¹ and for standard phosphamidon and dimethoate 200–10 000 ng mL⁻¹ (see ESI 1†). The instrument LOD obtained was 1.7, 2.8, 26.7 and 52.8 ng mL⁻¹ for chlorpyrifos, diazinon, phosphamidon and dimethoate, respectively. The instrument LOQ obtained was 5.6, 9.3, 88.9 and 175.9 ng mL⁻¹ for chlorpyrifos, diazinon, phosphamidon and dimethoate, respectively. The dynamic range was carried out by limit of linearity obtained in the range of 5.6 to 10 000 ng mL⁻¹ (for all the OPPs).

3.3 Magnetic solid phase extraction optimization

3.3.1 Sample volume. Enrichment factor (EF) is an important parameter in extraction processes and it is improved with an increase in sample volume (Fig. 5). In order to obtain a significant EF, high volumes of sample (10–150 mL) with the same OPP concentration were investigated using MSPE based on Fe₃O₄@G–CNPrTEOS (Fig. 5A). It is clear that when the sample volume was increased up to 120 mL, the efficiency (peak area of OPPs std/peak area of IS) also increased. 120 mL diluted milk was used for further studies as a high EF was obtained. Commercial SPE cartridges were tested with sample loading volume in the range from 1 mL to 30 mL. Both C18-SPE (Fig. 5B) and CNPr-SPE (Fig. 5C) showed low breakthrough volume because their efficiency (peak area of OPPs std/peak area of IS) rapidly decreased after 5 mL sample loading and the trend continued to 30 mL. This could be due to column blocking from the milk fat and proteins. Thus, 5 mL sample volume was selected for C18-SPE and CNPr-SPE. Thus, the proposed MSPE based on synthesized Fe₃O₄@G–CNPrTEOS was demonstrated to have a significant EF in comparison with commercial SPE cartridges.

Fig. 5 Effect of sample volume on extraction performance using (A) Fe₃O₄@G–CNPrTEOS MSPE, (B) C18-SPE and (C) CNPr-SPE.

3.3.2 Types of elution solvent and eluent volume. The efficiency of MSPE is directly dependent on the desorption solvent. Thus, five different organic solvents with different polarity (acetone, acetonitrile, methanol, ethyl acetate, and hexane) were investigated for OPPs elution. Fig. 6A shows the effect of solvents on OPPs extraction by using the Fe₃O₄@G–CNPrTEOS MSPE. Acetonitrile showed high extraction efficiency (peak area of OPPs std/peak area of IS) for elution of the four selected OPPs from the adsorbent as compared to other solvents. Also, elution of the trapped OPPs from the C18-SPE and CNPr-SPE was carried out by using similar solvents (see ESI 2†). Acetonitrile was also found to be an efficient elution solvent due to the high peak area ratio obtained. Since the single solvent system was found to give good results for the four OPPs and this agrees with our previous results,¹⁰ mixed solvent systems were not attempted. In addition, different volumes of acetonitrile ranging from 0.5 to 5 mL were studied for desorption of the four OPPs from Fe₃O₄@G–CNPrTEOS, C18-SPE and CNPr-SPE. Finally 1 mL, 3 mL and 2 mL acetonitrile were selected as the optimum volume for Fe₃O₄@G–CNPrTEOS, C18-SPE and CNPr-SPE, respectively.

Fig. 6 Effect of (A) types of elution solvents, (B) extraction time and (C) mass of adsorbent on extraction performance (D) reusability of the as synthesized Fe₃O₄@G–CNPrTEOS as MSPE adsorbent for ten extractions.

3.3.3 Extraction time and desorption process. Extraction time is another important factor that affects the extraction performance. The effect of extraction time on OPPs preconcentration was carried out for a time ranging from 1 to 30 min. Fig. 6B shows that the extraction efficiency (peak area of OPPs std/peak area of IS) reached a maximum level at 2 min and thereafter the response did not increase significantly. However, the synthesized Fe₃O₄@G–CNPrTEOS MSPE adsorbent showed the fastest extraction (2 min) of the four selected OPPs from the milk samples. The fast extraction might be due to the porous nature of the sol–gel material.¹⁰ Desorption was performed with the aid of vortex agitation. The agitation time during desorption was tested at 4 different time settings from 1 to 10 min. The peak area ratio response was found to significantly increase at 1 min, but as the time was increased, the % recovery becomes constant. Thus, 2 min was selected as the optimum extraction time and 1 min as the optimum agitation time for the desorption process.

3.3.4 Mass of adsorbent. The other key parameter in the MSPE process is adsorbent dosage. The mass of Fe₃O₄@G–NPrTEOS (MSPE adsorbent) was studied using five different masses in the range of 10–200 mg (Fig. 6C). The peak area ratio of the extracted OPPs continuously increased from 10 to 100 mg then stayed stable up to 200 mg. Therefore, the optimum mass of adsorbent selected was 100 mg for further analysis.

3.3.5 Reusability study. Reusability of the synthesized Fe₃O₄@G–CNPrTEOS adsorbent was studied with 10 consecutive adsorption–desorption cycles. Fig. 6D shows the extraction performance of the first extraction, fifth and tenth extractions. The extraction efficiency did not change significantly after ten extractions. Good recovery of the OPPs (>85%) was obtained up to 10 extractions.

3.4 Method validation

Under the optimal experimental conditions, linearity, LOD, LOQ, precision (intra-day and inter-day), % recovery, and EF of the three proposed methods were studied. Optimum conditions for MSPE are as follows; 120 mL sample volume, 2 min extraction time, 1 mL acetonitrile as desorption solvent, 1 min agitation time during the desorption process and 100 mg adsorbent dosage. The results achieved were compared with those using C18-SPE and CNPr-SPE cartridges. The extraction methods were assessed using matrix-matched calibration (milk samples were spiked with different concentrations of OPPs). To obtain the method linearity, the standard solutions in milk samples (0.1–10 000 ng mL⁻¹) were analyzed by MSPE, C18-SPE and CNPr-SPE followed by GC-ECD (Table 2). The proposed three methods provided good coefficient of determination (R² ≥ 0.996) for all OPPs. The LOD values were calculated on the basis of (3 × SD)/m (n = 3) and the values obtained in the range from 0.01–0.58 ng mL⁻¹ using the Fe₃O₄@G–CNPrTEOS MSPE. The LOQ value was calculated on the basis of (10 × SD)/m (n = 3). The MSPE method exhibited lower LOD (20–65×) and higher EF (25–30×) as compared to the C18-SPE and CNPr-SPE cartridges. This indicates that both cartridges are probably easily influenced by the cow’s milk matrix and interfering species occupy the adsorption sites. Otherwise, the as synthesized adsorbent was less affected possibly due to the higher electrostatic interaction (δ⁺ and δ⁻) and large π–π interaction (phenyl ring) between the analytes and adsorbent. The LOD obtained for the Fe₃O₄@G–CNPrTEOS MSPE was well below (58–1000× lower) the MRLs (10–50 ng kg⁻¹) set by the EU.

Table 2 Statistical results of OPPs extraction from the milk sample, including linearity, LOD, limit of quantitation (LOD), response factor (RF) and EF

Analyte	MSPE Fe3O4@G–CNPrTEOS				C18-SPE				CNPr-SPE
	Linearity (ng mL^−1)	LOD (ng mL^−1)	LOQ (ng mL^−1)	EF^a	Linearity (ng mL^−1)	LOD (ng mL^−1)	LOQ (ng mL^−1)	EF^a	Linearity (ng mL^−1)	LOD (ng mL^−1)	LOQ (ng mL^−1)	EF^a
a Enrichment factor calculated using equation; , Vaq is sample volume (diluted milk) and Vorg is reconstituted organic solvent volume (50 μL).
Phosphamidon	2–200	0.56	1.8	2400	100–1000	22.3	74.2	80	100–1000	11.1	37.1	100
Dimethoate	2–200	0.58	1.9	2400	100–1000	18.8	62.3	80	100–1000	19.6	65.5	100
Diazinon	0.1–10	0.02	0.08	2400	5–50	1.2	3.9	80	5–50	1.3	4.1	100
Chlorpyrifos	0.1–10	0.01	0.05	2400	5–50	0.52	2.4	80	5–50	0.53	1.7	100

---

The repeatability of the proposed MSPE Fe₃O₄@G–CNPrTEOS was investigated using intra-day (n = 3) and inter-day precision (n = 15) studies. The repeatability obtained was as follows; intra-day % RSD 1.7–9.4%, n = 3 and inter-day % RSD 10.7–17.6%, n = 15. Although inter-day precision shows slightly high % RSD (17.6%) yet single factor ANOVA analysis shows reliable results (p > 0.05) that demonstrate that there is no significant difference between inter-day precisions.

3.5 Milk sample analysis

The MSPE method based on the newly synthesized Fe₃O₄@G–CNPrTEOS was applied for the determination of OPPs in fresh milk (120 mL). The OPP concentrations were calculated for MSPE, C18-SPE and CNPr-SPE methods using calibration curves, and the results are listed in Table 3. In our study we did not detect any of the studied OPPs in the milk analyzed, thus we spiked the samples to assess the suitability (% R) of the developed method. However, OPP residues have been found in Italian raw milk (0.005–0.018 mg kg⁻¹),^1,3 Mexican commercial pasteurized milk (0.005–0.02 mg kg⁻¹)⁴ and Spanish raw milk (0.005–0.2 mg kg⁻¹).⁵ The Fe₃O₄@G–CNPrTEOS MSPE method gave a much better recovery of the OPPs (82–94%) compared to C18-SPE (43–59%) and CNPr-SPE (58–68%). Fig. 7 shows the chromatogram of the (A) unspiked and (B, C and D) spiked milk samples.

Fig. 7 Chromatogram of (A) unspiked milk, and spiked milk: (B) 1 ng mL⁻¹ MSPE, (C) 100 ng mL⁻¹ CNPr-SPE and (D) 100 ng mL⁻¹ C18-SPE milk samples. Peaks: (1) phosphamidon, (2) dimethoate, (3) diazinon, (4) chlorpyrifos, and IS indicates hexaconazole.

Table 3 Recovery (% R) of spiked OPPs in milk sample and precision (% RSD, n = 3)^a

Analytes	Extraction methods
	Unspiked (for three methods)	Fe3O4@G–CNPrTEOS MSPE		C18-SPE		CNPr-SPE
		Spiked (ng mL^−1)	% R (% RSD, n = 3)^b	Spiked (ng mL^−1)	% R (% RSD, n = 3)	Spiked (ng mL^−1)	% R (% RSD, n = 3)
a nd = not detected (OPPs studied was not detected in the fresh milk samples using all three methods). Blank was used for the three methods.b (n = 3) extraction recovery were analyzed for three times.
Phosphamidon	nd	20	84 (4.1)	100	59 (6.3)	100	60 (9.3)
Dimethoate	nd	20	82 (5.6)	100	55 (4.4)	100	57 (6.2)
Diazinon	nd	1	94 (9.1)	5	49 (3.6)	5	68 (7.1)
Chlorpyrifos	nd	1	89 (7.2)	5	43 (3.8)	5	64 (9.9)

---

3.6 Proposed adsorption mechanism

Fig. 8 shows that the newly synthesized Fe₃O₄@G–CNPrTEOS adsorbent has five possible interaction points to adsorb OPPs. Hydrophobic interactions occur between the graphene sheets and the non-polar chlorpyrifos (log K_O/W = 4.6) and diazinon (log K_O/W = 3.5). Hydrophilic interactions occur between the hydrophilic tail of (CNPrTEOS) and polar phosphamidon (log K_O/W = 0.3) and dimethoate (log K_O/W = 0.7). Strong π–π interactions are found between the large delocalized electrons on graphene and the phenyl ring from diazinon and chlorpyrifos. These made Fe₃O₄@G–CNPrTEOS an effective adsorbent towards the non-polar OPPs. Also, hydrophilic interactions (dipole–dipole interactions (δ⁺ and δ⁻) and H-bonding) occur between OH bonded to Si from the CNPrTEOS chain and polar OPPs, causing good adsorption affinity towards the polar phosphamidon and dimethoate. Therefore, the MSPE Fe₃O₄@G–CNPrTEOS adsorbent is capable of simultaneous extraction of polar and non-polar OPPs from a complex fresh milk matrix with minimum matrix effects.

Fig. 8 Proposed adsorption mechanism between the Fe₃O₄@G–CNPrTEOS and OPPs.

3.7 Comparison with other techniques

Table 4^45–54 shows the LOD and recovery of the developed Fe₃O₄@G–CNPrTEOS MSPE method compared to other methods used for pesticide determination in milk. Pesticides isolated from milk samples using silica–C18 as a conventional SPE sorbent with LC-MS gave an LOD 20× lower⁴⁹ than the Fe₃O₄@G–CNPrTEOS. However, SPE is a tedious and time consuming method for milk sample preparation due to the need for protein and fat removal using acid.

The MSPE method and liquid–liquid extraction (LLE) followed by dispersive primary secondary amine clean-up (DPSA), and SPE with GC-MS gave an LOD (3.3–296 pg mL⁻¹) that is 1.5–30× lower than the current Fe₃O₄@G–CNPrTEOS MSPE method.⁴⁷ However, the LLE procedure is lengthier than the proposed Fe₃O₄@G–CNPrTEOS and the proposed MSPE method is greener.

Polydimethylsiloxane-SPME (PDMS-SPME) using gas chromatography-nitrogen phosphorus detector (GC-NPD) also gave a slightly lower LOD (2× lower) (0.05 ng mL⁻¹).⁵³ All the other methods listed in Table 4 produced LODs which are higher than the current LOD using Fe₃O₄@G–CNPrTEOS MSPE. For example triazines analysis in milk sample using diphasic dialysis extraction provided an LOD that is 100× higher⁵⁰ compared to the current MSPE method.

Dispersive liquid–liquid microextraction based on the solidification of floating organic droplets (DLLME-SFO),⁴⁵ solid matrix dispersion extraction (SMDE) method⁵¹ and Fe₃O₄@G–CNPrTEOS MSPE method provided good and comparable recovery for OPPs determination from fresh milk. The current Fe₃O₄@G–CNPrTEOS MSPE and DLLME-SFO method⁴⁵ gave the same LOD but the SMDE method⁵¹ gave a much higher LOD (30–2600×).

The proposed Fe₃O₄@G–CNPrTEOS MSPE method also shows a 10–250× lower LOD for OPPs from milk samples compared to polydimethylsiloxane-divinylbenzene (PDMS-DVB) SPME using GC-MS (2.1 ng mL⁻¹),⁴⁶ LLE using GC-NPD (1–25 ng g⁻¹)⁴⁸ and C18-SPE using GC-NPD (1.0 ng mL⁻¹).⁵⁴

MSPE using Fe₃O₄@SiO₂@C18 with GC-MS⁵² gave comparable LOD to the current method. This is due to π–π interactions between graphene and the OPPs and also electrostatic interactions. DLLME-SFO with GC-FPD⁴⁵ gave the same LOD (0.1 ng mL⁻¹) to the current Fe₃O₄@G–CNPrTEOS MSPE method. In contrast, C18-SPE provides only hydrophobic interactions and PDMS-DVB SPME provides less π–π interaction as compared to graphene-based sorbents. The proposed Fe₃O₄@G–CNPrTEOS MSPE method can be directly used in fresh milk samples without the need for tedious steps (removal of the fat and protein) and only a small amount of adsorbent (100 mg) is required to achieve the low LOD (ensuring compliance with MRL set by the EU) and good recovery of the OPPs.

Table 4 LOD and method recovery (% R) between different selected extraction methods used in OPPs analysis of milk

Adsorbent	Analyte	Sample	Method	LOD (ng mL^−1)	% R	Detector	Ref.
a Dispersive liquid–liquid microextraction based on solidification of the floating organic droplet.b Gas chromatography-flame photometric detection.c Headspace solid phase microextraction.d Dispersive primary secondary amine.e Gel permeation chromatography.f Dispersive solid phase extraction; microextraction.g Gas chromatography-nitrogen phosphorus detection.h Solid matrix dispersion extraction.i OCPs: organochlorine pesticides; PCBs: polychlorinated biphenyls.
Fe3O4@G–CNPrTEOS	OPPs	Milk	MSPE	0.1	82–104	GC-μECD	This study
—	OPPs	Milk	DLLME-SFO^a	0.1	80–106	GC-FPD^b	45
PDMS-DVB	OPPs	Milk	HS-SPME^c	2.1	—	GC-MS	46
—	OPPs	Milk	LLE-DPSA^d cleanup-SPE	0.0033–0.296	34–102	GC-MS	47
—	OPPs	Milk	LLE-GPC^e clean up	1–25 ng g^−1	50.3–139.4	GC-FPD^b	48
Silica–C18	Pesticides	Milk	DSPE^f	0.005	77.5	LC-MS	49
—	Triazines	Milk	Dialysis	10.0	89–116	GC-NPD^g	50
Flux-calcined macroporous diatomaceous material	OPPs	Milk	SMDE^h	3–260 ng g^−1	76–109	GC-FPD^b	51
Fe3O4@SiO2@C18	Endocrine disruptors (OCPs and PCBs)^i	Milk	MSPE	0.1–0.3	79–116	GC-MS	52
PDMS	OPPs	Milk	SPME	0.053	—	GC-NPD^g	53
C18	OPPs	Milk	SPE	1	44–117	GC-NPD^g	54

---

4. Conclusion

The current study aims to report the efficient extraction of OPPs in fresh cow’s milk based on MSPE using newly synthesized Fe₃O₄@G–CNPrTEOS as an adsorbent with minimal sample preparation. The proposed MSPE was used for simultaneous direct determination and preconcentration of polar and non-polar OPPs from fresh cow’s milk. The results obtained indicate that the Fe₃O₄@G–CNPrTEOS MSPE method exhibits 50× lower LOD and higher enrichment (25–30×) than both commercial CNPr-SPE and C18-SPE cartridges. Recovery of the spiked OPPs (0.1 ng mL⁻¹ of each standard chlorpyrifos and diazinon; 10 ng mL⁻¹ of each phosphamidon and dimethoate) from fresh cow’s milk is excellent for the Fe₃O₄@G–CNPrTEOS MSPE (>82%) compared to the recovery using CNPr-SPE (>42%) and C18-SPE (>58%). The Fe₃O₄@G–CNPrTEOS MSPE exhibits great potential as an alternative sorbent for simultaneous polar and non-polar OPPs extraction from complex matrices such as milk. The main advantages of the proposed MSPE method are that only a simple dilution using deionized water is needed for the analysis of OPPs in milk, emulsions do not occur and so the method is simple.

Acknowledgements

The authors would like to thank the Ministry of Education Malaysia for financial support through the Research Grants Number Q.J130000.262.10J43 and R.J130000.7826.3F452. Universiti Teknologi Malaysia is also acknowledged for the International Doctoral Fellowship (IDF) award received by H. R. Nodeh.

References

1. G. Pagliuca, A. Serraino, T. Gazzotti, E. Zironi and A. Borsari, J. Dairy Res., 2006, 73, 340–344.
2. T. Kiljanek, A. Niewiadowska and S. Semeniuk, Bull. Vet. Inst. Pulawy, 2013, 57, 185–189.
3. T. Gazzotti, P. Sticca, E. Zironi, B. Lugoboni, A. Serraino and G. Pagliuca, Bull. Environ. Contam. Toxicol., 2009, 82, 251–254.
4. I. O. J. H. Salas, M. M. González, M. Noa, N. A. Pérez, G. Díaz, R. Gutiérrez and H. Zazueta, J. Agric. Food Chem., 2003, 51, 4468–4471.
5. M. J. Melgar, M. Santaeufemia and M. A. Garcia, J. Environ. Sci. Health, Part B, 2010, 45, 595–600.
6. E. Kampire, B. T. Kiremire, S. A. Nyanzi and M. Kishimba, Chemosphere, 2011, 84, 923–927.
7. Regulation (EC) No 396/2005 of the European parliament and of the council of 23 February 2005 on maximum residue levels of pesticides in or on food and feed of plant and animal origin, Off. J. Eur. Union, 2005, L70, 1–16.
8. A. A. Elbashir, M. M. A. Omar, W. A. Wan Ibrahim, O. J. Schmitz and H. Y. Aboul-Enein, Crit. Rev. Anal. Chem., 2014, 44, 107–141.
9. H. Bagheri and S. Banihashemi, Anal. Chim. Acta, 2015, 886, 56–65.
10. W. A. Wan Ibrahim, K. V. Veloo and M. M. Sanagi, J. Chromatogr. A, 2012, 1229, 55–62.
11. M. M. A. Omar, W. A. Wan Ibrahim and A. A. Elbashir, Food Chem., 2014, 158, 302–309.
12. J. Wu, Y. Qian, C. Zhang, T. Zheng, L. Chen and Y. Lu, Food Analytical Methods, 2013, 6, 1448–1457.
13. Z. Li, R. Ma, S. Bai, C. Wang and Z. Wang, Talanta, 2014, 119, 498–504.
14. Q. Liu, M. Cheng, Y. Long, M. Yu, T. Wang and G. Jiang, J. Chromatogr. A, 2014, 1325, 1–7.
15. X. Ma, J. Wang, Q. Wu, C. Wang and Z. Wang, Food Chem., 2014, 157, 119–124.
16. X. Ma, J. Wang, M. Sun, W. Wang, Q. Wu and C. Wang, Anal. Methods, 2013, 5, 2809–2815.
17. D. P. Suhas, T. M. Aminabhavi, H. M. Jeongd and A. V. Raghu, RSC Adv., 2015, 5, 100984–100995.
18. Z. Li, M. Hou, S. Bai, C. Wang and Z. Wang, Anal. Sci., 2013, 29, 325–331.
19. H. R. Nodeh, W. A. Wan Ibrahim, M. A. Kamboh and M. M. Sanagi, RSC Adv., 2015, 5, 76424–76434.
20. A. Mehdinia, H. Khani and S. Mozaffari, Microchim. Acta, 2013, 181, 89–95.
21. W. Guan, C. Li, X. Liu, S. Zhou and Y. Ma, Food Addit. Contam., Part A: Chem., Anal., Control, Exposure Risk Assess., 2014, 31, 250–261.
22. H. Bagheri, A. Roostaie and M. Allahdadlalouni, Microchim. Acta, 2015, 182, 617–624.
23. W. A. Wan Ibrahim, H. R. Nodeh and M. M. Sanagi, Crit. Rev. Anal. Chem., 2015 DOI:10.1080/10408347.2015.1034354.
24. W. Wang, R. Ma, Q. Wu, C. Wang and Z. Wang, Talanta, 2013, 109, 133–140.
25. K. J. Huang, Y. J. Liu, J. Li, T. Gan and Y. M. Liu, Anal. Methods, 2014, 6, 194–201.
26. J. Ding, Q. Gao, X. S. Li, W. Huang, Z. G. Shi and Y. Q. Feng, J. Sep. Sci., 2011, 34, 2498–2504.
27. I. Ali, Z. A. Al-Othman and A. Al-Warthan, Desalin. Water Treat., 2016, 57, 10409–10421.
28. M. Sun, X. Ma, J. Wang, W. Wang, Q. Wu and C. Wang, J. Sep. Sci., 2013, 36, 1478–1485.
29. X. Zhang, Y. Liu, C. Sun, H. Ji, W. Zhao, S. Suna and C. Zhao, RSC Adv., 2015, 5, 100651–100662.
30. D. He, X. Zhang, B. Gao, L. Wang, Q. Zhao and H. Chen, Food Control, 2014, 36, 36–41.
31. A. Kaouk, T. P. Ruoko, Y. Gönüllü, K. Kaunisto, A. Mettenbörger, E. Gurevich, H. Lemmetyinen, A. Ostendorfc and S. Mathur, RSC Adv., 2015, 5, 101401–101407.
32. W. A. Wan Ibrahim, H. R. Nodeh, H. Y. Aboul-Enein and M. M. Sanagi, Crit. Rev. Anal. Chem., 2015, 45, 270–287.
33. F. Meng, W. Wei, J. Chen, X. Chen, X. Xu, M. Jiang, Y. Wang, J. Lua and Z. Zhou, RSC Adv., 2015, 5, 101121–101126.
34. W. Zhang, X. Shi, Y. Zhang, W. Gu, B. Li and Y. Xian, J. Mater. Chem. A, 2013, 1, 1745–1753.
35. M. Šafako and I. Šafak, J. Magn. Magn. Mater., 1999, 194, 108–112.
36. C. Jiang, Y. Sun, X. Yu, Y. Gao, L. Zhang and Y. Wang, J. Chromatogr. B, 2013, 947–948, 49–56.
37. S. Wang, R. Wang, X. Wu, Y. Wang, C. Xue and J. Wu, J. Chromatogr. B, 2012, 905, 105–112.
38. Q. Wang, L. Huang, P. Yu, J. Wang and S. Shen, J. Chromatogr. B, 2013, 912, 33–37.
39. M. J. Wells, Principles of Extraction and the Extraction of Semivolatile Organics from Liquids, Sample Preparation Techniques in Analytical Chemistry, 2003, vol. 162, p. 37.
40. A. V. H. Herrera, J. Hernández-Borges and M. A. Rodríguez-Delgado, J. Chromatogr. A, 2009, 1216, 7281–7287.
41. L. He, Y. Su, Y. Zheng, X. Huang, L. Wu and Y. Liu, J. Chromatogr. A, 2009, 1216, 6196–6203.
42. M. Ahmadvand, H. Sereshti and H. Parastar, J. Chromatogr. A, 2015, 1413, 117–126.
43. S. C. Moldoveanu and M. Kiser, J. Chromatogr. A, 2007, 1141, 90–97.
44. F. J. Schenck and J. Casanova, J. Environ. Sci. Health, Part B, 1999, 34, 349–362.
45. X. X. Miao, D. B. Liu, Y. R. Wang, U. Y. Yang, X. Y. Yang and H. U. Gong, J. Chromatogr. Sci., 2015, 53, 1813–1820.
46. F. M. Rodrigues, P. R. Mesquita, L. S. Oliveira, F. S. Oliveira, A. M. Filho, P. A. P. Pereira and J. B. Andradea, Microchem. J., 2011, 98, 56–61.
47. X. Chen, P. Panuwet, R. Hunter, A. M. Riederer, G. C. Bernoudy, D. B. Barr and P. B. Ryan, J. Chromatogr. B, 2014, 970, 121–130.
48. L. Yang, H. Li, F. Zeng, Y. Liu, R. Li, H. Chen, Y. Zhao, H. Miao and Y. Wu, J. Agric. Food Chem., 2012, 60, 1906–1913.
49. T. Dagnac, M. Garcia-Chao, P. Pulleiro, C. Garcia-Jares and M. Lompart, J. Chromatogr. A, 2009, 1216, 3702–3709.
50. M. A García, M. Santaeufemia and M. J. Melgar, Food Chem. Toxicol., 2012, 50, 503–510.
51. A. Di Muccio, P. Pelosi, I. Camoni, D. A. Barbini, R. Dommarco and T. Generali, J. Chromatogr. A, 1996, 754, 497–506.
52. M. S. Synaridou, V. A. Sakkas, C. D. Stalikas and T. A. Albanis, J. Chromatogr. A, 2014, 1348, 71–79.
53. L. Cardeal Zde and C. M. Dias Paes, J. Environ. Sci. Health, Part B, 2006, 41, 369–375.
54. G. Pagliuca, T. E. Zironi and P. Sticca, J. Chromatogr. A, 2005, 1071, 67–70.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra26742k

---