Improved electromembrane microextraction efficiency of chloramphenicol in dairy products: the cooperation of reduced graphene oxide and a cationic surfactant

Abstract

A novel method of electromembrane extraction combined with HPLC analysis was applied for preconcentration–determination of chloramphenicol (CAP) residues in dairy products. It is shown that the presence of the cationic surfactant cetyltrimethylammonium bromide (CTAB) in the sample solution, and reduced graphene oxide (RGO) in the solvent membrane, results in the enhancement of the anionic analyte extraction into the acceptor phase. In fact, the addition of CTAB leads the accumulation of the analyte at the sample solution/supported liquid membrane (SLM) interface, and thus amelioration of the analyte transfer rate to the acceptor phase. The effect of RGO on this process comes from its efficient sorbent characteristic which augments the overall analyte partition coefficient into the SLM. The main parameters that affect the extraction efficiency were examined and optimized. Under the optimized conditions, the proposed method provided acceptable linearity (0.04–250 ng g⁻¹), satisfactory repeatability and reproducibility (CV% < 5.5), low LODs (0.012–0.021 ng g⁻¹) with high preconcentration factors (195–255) in the different samples. Finally, the proposed method was successfully applied to the evaluation of CAP residues in dairy products.

---

1. Introduction

Chloramphenicol (CAP) is a broad-spectrum antibiotic of the Amphenicol drug family. It has been previously used in veterinary medicine for the treatment of several infections.^1,2 It is efficient against a wide range of microorganisms and due to its ready availability and low cost has been broadly used.¹ Nevertheless, because of its toxicity in humans e.g. aplastic anemia, bone marrow depression and hypersensitivity, the use of CAP in food-producing animals and animal feed products has been strictly banned in the European Union and United States.³ The European Union has assigned a maximum residue limit (MRL) of 0.3 ng g⁻¹ for CAP residues in different food matrices, including meat and milk.⁴

As the illegal use of chloramphenicol in foods is still a real problem worldwide, the development of novel techniques for the determination of this antibiotic residues in foodstuffs is an essential task. A variety of analytical methods have been developed in order to monitoring of CAP levels in milk and food matrices using gas-chromatography (GC)^4,5 and liquid chromatography (LC).^6–9 Application of these methods is usually needs to employ primarily a liquid–liquid extraction (LLE) or a clean-up step using solid-phase extraction. These conventional extraction techniques have some disadvantages that motivated the analytical chemists to develop new miniaturized sample preparation methods.¹⁰

Hollow fiber liquid phase microextraction (HF-LPME) and electromembrane extraction (EME) are two highlighted achievements in the field of miniaturized sample preparation techniques and associated with rapidly progressing.^11,12 In both of these techniques, an appropriate water–immiscible organic solvent, as supported liquid membrane (SLM), is immobilized in the wall pores of a propylene hollow fiber. The SLM is sandwiched between donor and acceptor solutions and acts as a selective barrier for the transfer of specific analytes from one into the other solution. SLMs usually offer effective elimination of sample matrix components, e.g. large macromolecules and particles in real samples.¹³

The EME setup is similar to HF-LPME, except in the former an electrical field is employed by immersing two platinum electrodes in the sample and acceptor solutions. The mechanisms of analyte transport in HF-LPME and EME are based on passive diffusion and electrokinetic migration, respectively. In comparison with passive diffusion, an electrokinetic migration is significantly efficient transport mechanism and extracts the charged analytes in a very short time relative to the long-time extraction needed for HF-LPME (typically 20–60 min).^14,15 To this advantage one can added the elimination of further sample preparation steps. This is considered as the most attractive advantages of EME technique. It is supposed that the electrical field as main driving force aids the isolation of charged analytes from the sample matrix.

To improve the extraction efficiency of EME some developments comprising the modification of SLM composition or donor phase have been reported.^15,16 Reduced graphene oxide (RGO) as a relatively novel class of carbon nanomaterials can immobilized in the SLM. RGO has found its way to material science owing to the special characteristics such as layered structure, high specific surface area, excellent adsorptive properties and electrical conductivity.^17–19 RGO is composed of a hexagonal carbon network bearing oxygen-containing functional groups.¹⁸ These functional groups improve the dispersibility of RGO in organic solvent and also enhance its adsorption efficiency.

Surfactants are a large group of surface active substances with a great number of applications.²⁰ Surfactants bear a hydrophobic part and a hydrophilic part. Depend on the nature of the hydrophilic part the surfactants are classified as anionic, non-ionic, cationic or amphoteric. Owing to their special structure, there has been a growing interest in their application in sample preparation methods to improve extraction efficiency.^21,22

Following to our recent study concerning on the application of EME for preconcentration–determination pramipexole in urine samples,²³ the present report describes the effect of the cooperation of reduced graphene oxide (RGO) and CTAB (as a cationic surfactant) on the EME efficiency of CAP as an acidic compound. The combination of CTAB in donor phase and RGO in membrane promotes the migration of anionic analytes into the acceptor phase. The application of the proposed EME method was examined for the determination of CAP in some dairy products; raw cow's and goat's milk samples, skim milk powder and whole milk powder, using HPLC-UV analysis.

2. Experimental

2.1. Chemicals, materials and solutions

All the chemicals used were of analytical reagent grade. Chloramphenicol (acidic compound with pK_a = 5.5) was purchased from Sigma-Aldrich (Saint Louis, Missouri, USA). RGO was purchased from Sigma-Aldrich (Steinheim, Germany). 2-Nitrophenyloctyl ether (NPOE), 1-hexanol, 1-heptanol and 1-octanol were obtained from Fluka (Buchs, Switzerland). Triton X-100, Triton X-114 and cetyltrimethylammonium bromide (CTAB), hydrochloric acid and sodium hydroxide were purchased from Merck (Darmstadt, Germany). HPLC grade water used in all experiments was purified on a Milli-Q water purification system (Madrid, Spain).

A 1000 mg L⁻¹ chloramphenicol in methanol was prepared as the stock solution and stored at 4 °C in the dark. All working standard solutions of different concentrations were freshly prepared by diluting stock solutions with HPLC grade water. Dairy products including raw cow's milk, raw goat's milk, skim milk powder and whole milk powder were purchased from the market in Zanjan-Iran. 5.0 g of milk powder sample was weighed into a 100 mL plastic centrifuge tube and then 50 mL of HPLC water was added. The tube was located in water bath at 40 °C and mixed about 10 min to achieve a homogeneous sample.

Calibration curves were built by spiking blank milk samples at eight concentration levels (with certain amount of standard solution) to achieve the desired concentrations. Then, the spiked samples were diluted with HPLC grade water (1 : 5) and the pH was adjusted to 7.5 using NaOH solution. These samples were then submitted to the EME cell.

2.2. Chromatographic determination procedure

Separation and detection of chloramphenicol were performed by a Shimadzu (Japan) HPLC system comprising a LC-10AD HPLC pump, a six-port Rheodyne valve with a 20 μL sample loop and a YL 9120 UV-Vis detector. The chromatographic separation was run on a C18 analytical column (Hector, 5 μm, 10 mm × 4.6 mm) with a C18 guard column (Hector, 5 μm, 250 mm × 4.6 mm) under isocratic elution conditions. Chromatographic data were recorded and analyzed using YL clarity software. The mobile phase consisted of 10 mmol L⁻¹ ammonium acetate buffer (pH = 4.6) and acetonitrile (70 : 30, v/v). The flow rate was 1.0 mL min⁻¹ and temperature was ambient. The detector wavelength was set at the wavelength 260 nm. All pH measurements were accomplished using a pH meter 827 Metrohm (Herisau, Switzerland).

2.3. Equipment for EME

The equipment used for the electromembrane extraction was similar that described previously.²⁴ The used DC power supply was an EPS 600 model (Paya Pajoohesh Pars, Tehran, Iran) with a variable voltage in the range of 0–600 V, providing a current output in the range of 0–0.5 A. The platinum wires (Pars Platin, Tehran, Iran) with a diameter of 0.25 mm were employed as electrodes and connected to the power supply. A glass vial with an internal diameter of 1 cm and height of 10 cm was used as the donor solution compartment (Fig. 1). The porous hollow fiber for immobilization of SLM and housing the acceptor solution was PPQ3/2 polypropylene fiber (Membrana, Wuppertal, Germany), with 0.6 mm an internal diameter, 200 μm wall thickness and 0.2 μm pore size. The fiber was cut to 8.0 cm pieces, cleaned with acetone in an ultrasonic bath to eliminate any impurities and dried prior to use. A magnetic stirrer Heidolph model 301 (Kelheim, Germany) was used to stir the donor solution at a stirring rate of 300–1250 rpm during the extraction process.

Fig. 1 Schematic presentation of EME set up.

2.4. Electromembrane procedure

A graphical illustration of EME procedure has been revealed in Fig. 1. In order to prepare the RGO reinforced hollow fiber (RGO-HF), the homogeneous dispersed mixture of RGO in 1-octanol (3 mg mL⁻¹) was slowly injected into a 8 cm piece of HF, using a HPLC syringe. Then, the HF was entirely immersed in the RGO dispersed mixture and placed in ultrasonic bath for 15 min until the mixture was dispersed and spread within pores of the fiber. The excess amount of RGO adsorbed on surface the HF was then carefully washed with 1-octanol, until no RGO was seen in the washing solvent. Fig. 2 shows the scanning electron microscopy (SEM) images which represent the presence of the RGO in the wall pores of the fiber. A 20 μL volume of NaOH solution (pH = 10) as the acceptor phase was inserted into the lumen of the RGO-HF using a microsyringe and its end was thermally sealed. The anode (positive electrode) was placed in the lumen of the RGO-HF. The fiber containing the SLM (dispersed RGO in 1-octanol), the acceptor phase and anode electrode was dipped in 6.5 mL of donor solution (pH = 7.5) containing 10 ng mL⁻¹ of CAP and 0.15 mM of CTAB. Then, the cathode (negative electrode) was inserted directly into the donor solution. The electrodes were connected to a power supply and the sample compartment was placed on a stirrer. The power supply was turned on and the extraction accomplished by applying a voltage of 30 V for 15 min at 1000 rpm. Under the applied voltage, the charged analyte (negative charge) migrated from the donor solution through the SLM, and then transferred to the acceptor phase. After the extraction, the sealed end of the HF-RGO was cut with scissors, and the acceptor phase was collected using a 25 μL microsyringe and injected into the HPLC instrument.

Fig. 2 SEM images of (A) the HF surface (B) the HF only with 1-octanol and (C) the HF containing the dispersed RGO solution in 1-octanol.

2.5. Calculation of enrichment factor, extraction recovery, and relative recovery

The enrichment factor (EF) and the extraction recovery (ER%) were calculated according to following equations:C_f,a is the final concentration of CAP in the acceptor phase after extraction, and C_i,s is the initial analyte concentration in the donor phase before extraction. V_f,a and V_i,s define the corresponding solutions volume. C_f,a was calculated from a calibration graph achieved from direct injection of standard solutions.

The relative recovery (RR%) was calculated according to eqn (3):

C_found, C_real, and C_added are the determined concentration of CAP after adding of a known amount of standard solution into the real sample, the initial concentration of CAP in the real sample, and the spiked known concentration of standard to the real sample, respectively.

3. Results and discussion

3.1. Experimental design

The flux of the charged substance through the SLM (J_i) has been defined by means of a mathematical model based on the Nernst–Planck equation (eqn (4)):²⁵where D_i represents the diffusion coefficient of the charged substance, h is the thickness of the SLM, ν is a function of electrical potential, C_i is the concentration of charged substance at the SLM/sample interface, C_i,0 is the concentration of charged substance at the acceptor/SLM interface and χ (ion balance) is the total ionic concentration in the donor phase divided by the total ionic concentration in the acceptor phase. The model describes how different factors affect the flux of analytes across the SLM. In order to achieve the maximum extraction efficiency for determination of CAP in dairy products, the effective parameters the EME procedure were investigated.

Some preliminary experiments were performed to investigate possible interactions between the effective parameters. The obtained results indicated that there was no significant binary interaction between the parameters. Thus, one-variable-at-a-time (OVAT) methodology was used for optimization of extraction process. In this strategy, the response is investigated for each factor in turn while all the other factors are held at a constant level. A standard solution containing 10 ng mL⁻¹ of the analyte dissolved in HPLC water was used for all optimization experiments.

In this study the reported pattern of new articles were used to design of experiments.^15,16,23 In this order, considering to importance of affected parameters, SLM composition, amount of RGO in SLM, type of surfactant, amount of surfactant, pH of donor phase and acceptor, voltage, extraction time, agitation speed and ionic strength of donor solution, respectively were investigated and optimized. Then, selected conditions were applied in dairy products. Experimental conditions in details are given in related sections.

3.1.1. Composition of SLM. Selecting the composition of SLM is one of the most crucial steps for a successful EME process. An optimal organic solvent must have a suitable dipole moment or electrical conductivity in order to create an electrical field between the donor and acceptor solutions and also a relatively low volatility and solubility in water to prevent solvent loss during the EME process.^26,27 In addition, RGO must be well dispersed in the organic solvent. Based upon literature reports, nitro-aromatic solvents and aliphatic alcohols are efficient organic solvents for extraction of basic drugs and acidic drugs, respectively.²⁸ Based on the above considerations, four organic solvents 2-nitrophenyloctyl ether (NPOE), 1-hexanol, 1-heptanol and 1-octanol were tested. The results shown in Fig. 3A reveal that 1-octanol provided the highest EF for extraction of CAP. The obtained results were in agreement with the previous studies because 1-octanol has been the best candidate for acidic drugs in EME up to now.²⁸ More suitability of 1-octanol in comparison with other aliphatic alcohols may be justified by its higher electrical resistance, which decreases the electrical current and improves the stability or repeatability of the EME system.²⁹ Thus, 1-octanol was selected as the principal SLM constituent in the subsequent experiments.

Fig. 3 The effect of the extraction parameters including (A) the organic solvent composition, (B) the amount of RGO, (C) type of surfactant (D), the concentration of surfactant (CTAB) (E), the donor phase pH ((a) the acceptor phase pH (b)) and (F) applied voltage on the EF of EME system. EME conditions; extraction time: 15 min (for A–F), stirring rate: 1000 rpm (for A–F), concentration of surfactant: 0.1 mM (for C) and 0.15 mM (for D–F), the amount of RGO: 3 mg mL⁻¹ (for C–F), applied voltage: 20 V (for A–E), pH of acceptor phase: 10 (for A–D, E(a) and F), pH of donor phase: 7.5 (for A–D, E(b) and F) and analyte concentration: 10 ng mL⁻¹ (for A–F). Error bars were achieved based on three independent experiments.

3.1.2. Effect of the amount of RGO. The adsorption capacity of the analyte in SLM can be affected by the amount of dispersed RGO in the membrane. The presence of oxygen bearing binding groups on the surface of RGO sheets and the large delocalized π-electron system enhance its adsorption efficiency to absorb and interact with analyte.¹⁷ The analyte would be adsorbed on the surface of RGO sheets from the donor solution and subsequently desorbed into the acceptor solution, under applied electrical field. The results in Fig. 3B show that the additions of RGO into 1-octanol significantly improve the EF of the EME procedure. By increasing the amount of dispersed RGO in the range 0–3 mg mL⁻¹, the EF of the analyte increased. Beyond 3 mg mL⁻¹ of the dispersed RGO the value of EF diminished, owing to the aggregation of RGO sheets in the wall pores of fiber which would block the pores.²⁶ Based on this investigation an amount of 3 mg mL⁻¹ of the RGO in 1-octanol was chosen for performing the subsequent experiments.

3.1.3. Effect of type and concentration of surfactant. It was suggested that the presence of surfactant can improve the migration of the analyte through the liquid membrane, under applied electrical field. According to the nature of the surfactant, its hydrophobic head of orients to the organic solvent and the hydrophilic head remains in the aqueous phase. In other words, surfactant molecules accumulate in the organic solvent/sample solution interface.³⁰ Selection of appropriate surfactant is fundamental for obtaining a satisfactory preconcentration in EME process. The results of the investigation of the effect of nonionic and ionic surfactants (CTAB, Triton X-100 and Triton X-114) on the extraction efficiency are shown in Fig. 3C. These results revealed that CTAB is more suitable than the other tested surfactants. The hydrophilic head of CTAB bears a positive charge and therefore this fact causes the SLM surface to gain a positive charge. This induces the migration of anionic analyte molecules towards the SLM and under applied voltage, analyte can be simply released into the acceptor solution.

Surfactant concentration in sample solution is an essential factor for effective extraction of analyte into acceptor solution. Therefore, the effect of CTAB concentration on the EF of the presented method was investigated in the range of 0.0–0.3 mmol L⁻¹ (Fig. 3D). The efficiency of the extraction process was improved with CTAB concentration up to 0.15 mmol L⁻¹ due to the increase of free surfactant monomers. Beyond this concentration, the EF value was declined sharply. The optimized concentration was 0.15 mmol L⁻¹, lower than the critical micelle concentration (CMC) of CTAB (near 1 mmol L⁻¹). It is owing to great interaction of analyte molecules with formed micelles and also to an increase in the viscosity of the sample solution.^16,30 Therefore a concentration of 0.15 mmol L⁻¹ was selected as the optimized CTAB concentration throughout the study.

3.1.4. Optimization of the sample and acceptor solutions pH. Strong ionization of the acidic compounds in the sample solution is vital to obtain an efficient electrokinetic migration under an electrical field. The sample solution should be adequately basic. This condition allows the acidic compounds bear negative charge and to be transferred toward the positive electrode (anode). The completely ionized form of CAP could be created when the pH value exceeded its pK_a value (pK_a: 5.5). Thus, in order to study the effect of pH of the sample solution, pH of the acceptor phase was kept constant at 10 and the pH of the sample solution was changed within the range of 5–9 (Fig. 3E(a)). The results indicated that, maximum amount of CAP was extracted when pH of the sample solution was adjusted at 7.5. Beyond this pH, the EF value dropped. This can be attributed to an increase in ion balance of the system and also may be related to the competition between charged analytes and hydroxide ions to transfer across SLM.^26,31 Hence, pH of 7.5 was chosen for the sample solution in further experiments.

The acceptor solution pH is also a key factor during EME procedure. The acceptor solution should be intensely alkaline to guarantee the ionization of the analytes and simplify desorption of them from the RGO and SLM. For this purpose, pH of the sample solution was kept constant at 7.5 and the acceptor pH was changed in the range of 8–12 with NaOH solution. The results revealed that an increase in the pH of the acceptor phase to 10 improves extraction efficiency of CAP (Fig. 3E(b)). In the case of the acceptor solution, decreasing the pH increases the protonation probability of the drug and accelerates their back diffusion extraction probability into sample solution.³² On the other hand, increasing the pH of the acceptor solution promotes the release rate of acidic drug into the acceptor solution at the SLM/acceptor solution interface and resulting in the increase of the extraction efficiency. Nevertheless, there are some restrictions for the application of high concentrations of NaOH in the acceptor solution such as the increase in the risk of bubble formation.²⁸ Bubble formation leads to intensive fluctuation in the volume of acceptor phase and raises uncertainties of the result achieved by EME technique.²⁸ The results in this part of study conduct us to use a pH of 10 for the acceptor phase.

3.1.5. Effect of applied potential. The major driving force for the migration of charged analytes across the SLM is created by applied voltage. Based on previous reports, an acidic drug needs a low electrical potential, while a basic drug requires a relatively high electrical potential to be extracted.³³ This may be owing to the fact that 1-octanol has a high dielectric constant while NPOE has a lower dielectric constant.³⁴ Therefore, in order to optimize the applied potential, a series of experiments with various extraction voltages in the range 0–60 V were performed (Fig. 3E). In a totally quiescent assembly (0 V) the extraction efficiency of CAP was very low. This low extraction was observed because in the presence of the CTAB, CAP can be possibly extracted into acceptor solution by passive diffusion mechanism. As it is shown in Fig. 3F relatively high extraction efficiency was achieved when a 30 V potential was applied. Nevertheless, at higher voltages a decrease in the extraction efficiency of the process was observed. This was due to the production of air bubbles at the surface of the electrodes (electrolysis phenomena).²⁴ Therefore, a 30 V voltage was chosen as the applied voltage across the SLM in the subsequent experiments.

3.1.6. Time dependency of the process. The extraction time is an important factor for the EME methods, and should be examined. For investigation of the time dependency of the proposed method and optimization of this factor, the effect of extraction time is studied in the range 2–25 min. The EF of CAP was increased by increasing the extraction time from 2 to 15 min. A further increase in the extraction time caused a reduction in the EF of the method. This phenomenon may be owing to back diffusion of analyte into sample solution, a built-up of a boundary layer of ions at the interfaces of the SLM and local loosening of the SLM as a result of its evaporation.^24,28 SLM evaporation leads to increase the electrical current level and increasing the electrolysis reactions. Hence, 15 min was selected as the optimal extraction time.

3.1.7. Effect of stirring rate. In this work, the volume of the sample solution was adjusted at 6.5 mL that it is a relative large one for EME process. Thus, suitable agitation should be applied to increase the convection effect between the two electrodes. The stirring rate was examined within 300 to 1250 rpm. The extraction efficiency of CAP was enhanced by stirring speed up to 1000 rpm, however it was diminished at higher stirring rates. This reduction of the extraction efficiency at the higher stirring rate can be interpreted by considering the bubble formation in the acceptor phase leads to the loss of sorbent/SLM in the fiber pores. Based on these results, the stirring rate of 1000 rpm was chosen for further studies.

3.1.8. Influence of salt addition. In conventional LLE, salt addition decreases the solubility of analytes in sample solution and therefore it causes an increase in their partition into the organic solution.³⁵ Nevertheless, in EME technique, salt addition may have a different effect. According to eqn (4), the presence of high contents of ionic species in the sample solution leads to a decrease in the extraction efficiency of charged analytes due to the competition with analyte ions in migration toward the anode electrode, and thus an increase in ion balance and instability problems in EME system. To study the effect of this factor, the addition of sodium chloride salt into the sample solution in the range of 0–10% (w/v) was evaluated. It was found a significant drop in the EF of the analyte by increasing the salt concentration. Such observation has been reported in other investigations.^24,28 Consequently, no salt was added to the sample solution in the subsequent experiments.

3.2. The mechanism of the process

To verify the mechanism of the improvement of extraction efficiency of CAP by cooperative effect of RGO and CTAB, four different modes of EME technique namely, EME based on the 1-octanol (EME), EME based on the immobilized RGO into membrane (RGO/EME), surfactant assisted EME (surfactant/EME) and surfactant assisted EME based on immobilized RGO into membrane (surfactant/RGO/EME), were designed and examined. The EFs of CAP in the different EME modes were shown in Fig. 4. The experimental results showed that the EF of analyte in the system surfactant/RGO/EME is the highest among the tested methods. Thus, it was assumed that CAP was extracted and enriched by both RGO and CTAB. The extraction principle is a cooperation effect based on surfactant/LPME and SPME under electrical field rather than only on surfactant/LPME or SPME. Also, it is suggested that, in surfactant/RGO/EME mode, each immobilized RGO molecule in the membrane forms a SPME unit, each 1-octanol molecule is a LPME unit, each RGO molecule adhering 1-octanol comprises a solid/liquid microextraction unit, and many solid/liquid microextraction units form an array. Hence, the multipath or diversity EME process can enhance the extraction capability of CAP.

Fig. 4 The EFs of CAP in different EME modes under optimized extraction conditions.

3.3. Evaluation of the method performance

Quantitative parameters of the EME method including EF, relative recovery (RR%), extraction recovery (ER%), limit of detection (LOD), limit of quantification (LOQ), linearity (LR) and repeatability were examined for extraction of CAP in analyte-free water and dairy products under the optimum conditions described above (Section 3.1.). The calculated figures of merit are given in Table 1.

Table 1 Figures of merit of proposed EME method

Sample	LOD^a	LOQ^a	L. R.^a	EF^b	CV% (n = 3)^b		ER^b%
					Intra-day	Inter-day
a In ng g^−1.b Using a concentration of 10 ng g^−1 for the analyte.
Water	0.012	0.04	0.04–250	255	3.9	4.7	78.4
Raw milk	0.018	0.06	0.06–250	210	4.3	5.5	64.4
Milk powder	0.021	0.07	0.07–250	195	4.6	5.3	59.8

---

Calibration curves were constructed by plotting peak area against the concentration of the analyte. The developed method revealed good linearity with correlation coefficient values greater than 0.999 in the range 0.04–250 in water, 0.06–250 in raw milk and 0.07–250 ng g⁻¹ in milk powder samples. The LODs (S/N: 3) were obtained in the range 0.012–0.021 ng g⁻¹ while the LOQs (S/N: 10) were obtained in the range of 0.04–0.07 ng g⁻¹ in different samples. Intra- and inter-day precision (CV%) were found to be lower than 5.5%. This confirms the satisfactory precision of the proposed technique for the preconcentration–determination of trace amounts of CAP in water samples and dairy products. The extraction recoveries for dairy products were obtained in the range 59–64%, which corresponds to enrichment factors of 195–210. These enrichment factors were significantly lower than those found for the water samples. The lower extraction recoveries for dairy products were essentially created due to interactions of proteins and fat with CAP molecules that are called the suppressing effect. Nevertheless, this effect was reduced with adjusting the applied voltage. Application of suitable voltage allowed the proposed EME was also capable for extraction–determination of CAP in dairy products with acceptable sample clean up into an aqueous extract. Fig. 5 depicts the HPLC-UV chromatograms of CAP in raw milk (A) and milk powder (B) samples before (a) and after spiking (b) 10 ng g⁻¹ of CAP, under optimal extraction conditions. The experimental results show the lack of CAP in the examined dairy products. To investigate the matrix effect in dairy products, four concentration levels (0.1, 50, 100 and 200 ng g⁻¹) of the analyte were separately spiked into milk samples and their relative recoveries (RR%) were found as 88–97% (Table 2). The high relative recoveries showed a negligible matrix effect on EME efficiency in different milk samples. The comparison of the proposed EME method with some previously reported techniques for the determination of CAP indicated that the EME method has a lower detection limit, good precision and an extensive linear dynamic range (Table 3). Also, the results reveal that the method is reliable, reproducible and meets the requirement for the European Union regulation of 0.3 ng g⁻¹ for dairy products.

Fig. 5 HPLC-UV chromatograms of CAP in the raw milk (A) and in the milk powder (B) before (a) and after spiking (b) 10 ng g⁻¹ of CAP, under optimized EME conditions.

Table 2 Determination of CAP in different milk samples

Sample	Cadded^a	Cfound^a	CV% (n = 3)	RR%
a All concentrations are based on ng g^−1.
Raw cow's milk	0.1	0.093	3.7	93
	50	45	4.1	90
	100	97	5.0	97
	200	188	4.8	94
Raw goat's milk	0.1	0.089	3.8	89
	50	46.5	4.7	93
	100	94	5.3	94
	200	182	5.4	91
Skim milk powder	0.1	0.092	3.8	92
	50	44	5.1	88
	100	89	4.9	89
	200	184	3.6	92
Whole milk powder	0.1	0.094	4.9	94
	50	45.5	4.8	91
	100	89	5.4	89
	200	180	3.7	90

---

Table 3 Comparison of the proposed method with other techniques for quantitative determination of CAP in milk samples

Technique	Matrix	LOD^a	L. R.^a	CV%	Reference
a In ng g^−1.
QuEChERS/LC-MS-MS	Milk	0.045	0.1–15	<8.7	36
LLE & SPE/HPLC-MS-MS	Milk	0.3	1.5–25	<13.4	37
LLE/LC-MS-MS	Milk powder	0.09	Up to 0.9	<15	38
EME-HPLC-UV	Milk	0.018	0.06–250	<5.5	This work
EME-HPLC-UV	Milk powder	0.021	0.07–250	<5.5	This work

---

4. Conclusions

The present work describes a new approach of electromembrane extraction combined with HPLC-UV detection was developed for the determination of chloramphenicol as a model of acidic compounds from dairy products. In comparison with the conventional hollow fiber EME technique, the immobilization of RGO in the membrane and the presence of the CTAB in the donor solution provide an effective method for the improvement of the electromembrane extraction performance. The process bears some advantages include short time, cost effectiveness, simplicity, and ease of operation and minimal consumption of expensive and hazardous solvents. These features are of key interest for analytical laboratories doing routine trace analysis of CAP in dairy products by EME technique.

Acknowledgements

The authors thank for the research council of the University of Zanjan for financial support of this study.

References

1. E. C. P. do Rego, E. d. F. Guimaraes, A. L. M. dos Santos, E. d. S. M. Mothe, J. M. Rodrigues and A. D. Pereira Netto, Anal. Methods, 2015, 7, 4699–4707.
2. L. Rodziewicz and I. Zawadzka, Talanta, 2008, 75, 846–850.
3. J. Ferguson, A. Baxter, P. Young, G. Kennedy, C. Elliott, S. Weigel, R. Gatermann, H. Ashwin, S. Stead and M. Sharman, Anal. Chim. Acta, 2005, 529, 109–113.
4. M. Rejtharová and L. Rejthar, J. Chromatogr. A, 2009, 1216, 8246–8253.
5. S. Ding, J. Shen, S. Zhang, H. Jiang and Z. Sun, J. AOAC Int., 2005, 88, 57–60.
6. F. Barreto, C. Ribeiro, R. B. Hoff and T. D. Costa, Food Addit. Contam., 2012, 29, 550–558.
7. D. R. Rezende, N. F. Filho and G. L. Rocha, Food Addit. Contam., Part A, 2012, 29, 559–570.
8. C. Douny, J. Widart, E. Pauw, G. Maghuin-Rogister and M. L. Scippo, Food Anal. Methods, 2013, 6, 1458–1465.
9. S. Liu, X. Z. Wu, Z. H. Gao and F. Jiao, Anal. Methods, 2013, 5, 1150–1154.
10. M. T. Jafari, M. Saraji and H. Sherafatmand, Anal. Bioanal. Chem., 2011, 399, 3555–3564.
11. S. Pedersen-Bjergaard and K. E. Rasmussen, Anal. Chem., 1999, 71, 2650–2656.
12. A. Gjelstad, K. E. Rasmussen and S. Pedersen-Bjergaard, J. Chromatogr. A, 2006, 1124, 29–34.
13. S. Shariati, Y. Yamini and A. Esrafili, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2009, 877, 393–400.
14. J. Y. Lee, H. K. Lee, K. E. Rasmussen and S. Pedersen-Bjergaard, Anal. Chim. Acta, 2008, 624, 253–268.
15. K. S. Hasheminasab and A. R. Fakhari, Anal. Chim. Acta, 2013, 767, 75–80.
16. H. Bagheri, A. R. Fakhari and A. Sahragard, RSC Adv., 2016, 6, 4843.
17. H. Tabani, A. R. Fakhari, A. Shahsavani, M. Behbahani, M. Salarian, A. Bagheri and S. Nojavan, J. Chromatogr. A, 2013, 1300, 227–235.
18. S. Pei and H. M. Cheng, Carbon, 2012, 50, 3210–3228.
19. A. R. Fakhari, A. Sahragard, H. Ahmar and H. Tabani, J. Electroanal. Chem., 2015, 747, 12–19.
20. A. Sarafraz Yazdi, Trends Anal. Chem., 2011, 30, 6.
21. E. K. Paleologos, D. L. Giokas and M. I. Karayannis, TrAC, Trends Anal. Chem., 2005, 24, 426–436.
22. M. Moradi, Y. Yamini, A. Esrafili and S. Seidi, Talanta, 2010, 82, 1864–1869.
23. A. Fashi, F. Khanban, M. R. Yaftian and A. Zamani, Talanta, 2017, 162, 210–217.
24. A. Fashi, M. R. Yaftian and A. Zamani, Food Chem., 2015, 188, 92–98.
25. A. Gjelstad, K. E. Rasmussen and S. Pedersen-Bjergaard, J. Chromatogr. A, 2007, 1174, 104–111.
26. X. Y. Song, J. Chen and Y. P. Shi, New J. Chem., 2015, 39, 9191–9199.
27. M. Bazregar, M. Rajabi, Y. Yamini, A. Asghari and Y. Abdossalami asl, J. Chromatogr. A, 2015, 1410, 35–43.
28. Y. Yamini, S. Seidi and M. Rezazadeh, Anal. Chim. Acta, 2014, 814, 1–22.
29. S. Seidi, M. Rezazadeh, Y. Yamini, N. Zamani and S. Esmaili, Analyst, 2014, 139, 5531.
30. P. Zahedi, S. S. Davarani, H. R. Moazami and S. Nojavan, J. Pharm. Biomed. Anal., 2016, 117, 485–491.
31. K. Sadat Hasheminasab, A. R. Fakhari, A. Shahsavani and H. Ahmar, J. Chromatogr. A, 2013, 1285, 1–6.
32. J. N. Sun, Y. P. Shi and J. Chen, RSC Adv., 2015, 5, 37682.
33. H. Tabani, A. R. Fakhari and A. Shahsavani, Electrophoresis, 2013, 34, 269–276.
34. M. H. Koruni, H. Tabani, H. Gharari and A. R. Fakhari, J. Chromatogr. A, 2014, 1361, 95–99.
35. S. Seidi, Y. Yamini, A. Heydari, M. Moradi, A. Esrafili and M. Rezazadeh, Anal. Chim. Acta, 2011, 701, 181–188.
36. H. Y. Liu, S. L. Lin and M. R. Fuh, Talanta, 2016, 150, 233–239.
37. H. Tian, Chemosphere, 2011, 83, 349–355.
38. L. Rodziewicz and I. Zawadzka, Talanta, 2008, 75, 846–850.

---

Footnote

† Present address: School of Chemistry, The University of Melbourne, Victoria 3010, Australia.

---