Supramolecular nanosolvent-based hollow fiber liquid phase microextraction as a novel method for simultaneous preconcentration of acidic, basic and amphiprotic pollutants

Abstract

The coextraction of acidic, basic and amphiprotic pollutants from various matrixes is a significant and disputable concept in sample preparation strategies. In this study, for the first time, coextraction of acidic, basic and amphiprotic pollutants was performed using supramolecular nanosolvent-based hollow fiber liquid phase microextraction (SS-HF-LPME) as an efficient method followed by high performance liquid chromatography-photodiode array detection. The supramolecular solvent (SUPRAS) is formed through coacervation of decanoic acid aqueous vesicles in the presence of tetrabutylammonium hydroxide. The results revealed that 40% SUPRAS in 1-decanol has the best extraction efficiency for three selected model analytes (4-nitrophenol, 3-nitroaniline and 1-amino-2-naphthol). The extraction process was accomplished in two-phase mode and the unique interactions between the solvent and polar analytes (hydrophobic, electrostatic, hydrogen bonding and π–cation interactions) resulted in elevated coextraction efficiency. Central composite design methodology combined with the desirability function approach was applied to develop predictive models for simulation and optimization of the SS-HF-LPME procedure. The optimized conditions were: pH of the sample, 9.0; percentage of SUPRAS in 1-decanol, 40%; extraction time, 30 min; salt concentration, 20% w/v; stirring rate, 1250 rpm. Under the optimum conditions, detection limits and linear dynamic ranges were achieved in the range of 0.1–0.2 μg L⁻¹ and 0.5–400 μg L⁻¹, respectively. The percentage extraction recoveries and relative standard deviations (n = 5) were in the range of 56.1–71.1 and 4.1–6.9, respectively. Finally, the applicability of this method was successfully confirmed by analyzing rain, snow, river, dam and wastewater samples.

---

1. Introduction

It has been reported that aniline, phenol and their derivatives are acute environmental pollutants, and they are classified as hazardous waste and priority toxic pollutants by the Environmental Protection Agency of America;^1,2 moreover, they have been suspected to be carcinogenic agents.^2–4 They are consumed in diverse manufacturing processes such as pesticides and herbicides, pharmaceuticals, plastics, dyestuff, pigments, wood preservatives, rubber chemicals, and explosives.^5–7 Anilines and phenols can easily permeate through soil and contaminate ground water due to their high solubility in water.² Herein, coextraction of these pollutants is addressed.

Nitrophenols are a class of the most important pollutants present in the environment. Nitrophenols are formed in the atmosphere through the photochemical reaction of benzene with nitrogen monoxide in highly polluted air.^2,8 4-Nitrophenol (4-NP), for instance, is one of the 129 organic pollutants listed by the United States Environmental Protection Agency as carcinogens and hazardous to human beings as well as the environment.⁹ Furthermore, 4-NP damages mitochondria and inhibits energy metabolism in humans and animals.^5,9 Hence, exploring a simple, rapid, sensitive, environmentally friendly and cost effective method for 4-NP determination is crucial.

Azo dyes are synthetic organic colorants generally produced by coupling a diazonium compound with an aromatic amine or a phenol, and they are utilized in various areas such as nutrition, cosmetics, paper, pharmaceutical, printing ink, textile and tanning industries.¹⁰ Several azo dyes used as colorants for food, drugs and cosmetics can be reduced by cell suspensions of predominant intestinal anaerobes;¹¹ therefore, it can be assumed that the ingestion of certain azo dyes is indeed a risk for human health. In this respect, 1-amino-2-naphthol (1-A2N), produced by the reduction of acid orange 7, has been reported to induce bladder tumors.¹² The high toxicity of 1-A2N (EC₅₀ 0.1 ± 0.03 mg L⁻¹) is probably due to its high solubility in lipids.¹³

Nitroaniline isomers such as 3-nitroaniline (3-NA), as nitro-substituted derivatives of aromatic amines, have become more and more significant in environmental science due to their high toxicity and their suspected carcinogenic properties.^14,15 These pollutants are mainly used as intermediates in the synthesis of dyestuff, pharmaceuticals, pesticides, and herbicides,^6,15 and then they are released into the environment directly as industrial wastes or indirectly as breakdown products of pesticides and herbicides.^15–17

Several analytical methods, such as high-performance liquid chromatography (HPLC) with ultraviolet,^2,8 mass spectrometry¹⁸ or electrochemical detection,¹⁹ gas chromatography with flame ionization²⁰ or mass spectrometry detection,²¹ and capillary zone electrophoresis, have been utilized for the determination of phenol, aniline and their derivatives.^22,23 All the named methods have been successfully applied for the routine analysis of each category, but none of them affords simultaneous quantification of the mentioned acidic, basic and amphiprotic pollutants in a single step.

Sample preparation procedures play a dominant role in chemical analyses. Extensive sample cleanup procedures are usually required to remove matrix components which may interfere with the analysis.²⁴ Liquid–liquid extraction and solid phase extraction are commonly applied as sample pretreatment techniques in analytical chemistry.^15,25,26 However, these methods are time-consuming, generally labor-intensive, and require large quantities of expensive, toxic and environmentally unfriendly organic solvents.²⁷ Solvent microextraction techniques, which are commonly faster and simpler than conventional methods, effectively overcome these problems by reducing the amount of organic solvent consumption.²⁸ Moreover, extraction, preconcentration, and sample introduction into the analytical instrument are performed in a single step.^29,30 In 1999, a novel and efficient liquid phase microextraction technique based on applying hollow fiber membrane (HF-LPME) was developed.³¹ Using this microporous hollow fiber membrane provides the merits of the protection of the acceptor phase as well as efficient sample microfiltration through the pores of the hollow fiber.^32,33 HF-LPME can be done in either two- or three-phase configuration. In the two-phase sampling configuration (HF-LPME), the analytes of interest are extracted from an aqueous sample into a water-immiscible extraction solvent which is immobilized in the pores and lumen of the hollow fiber. In contrast, in the three-phase sampling configuration (HF-LLLME), limited to ionizable analytes, the analytes are extracted from an aqueous sample through a water-immiscible extractant which is immobilized in the pores of the hollow fiber and ultimately back-extracted into an acceptor aqueous phase inside the lumen of the hollow fiber.^28,34

Various extractants, including common solvents (i.e. long-chain aliphatic alcohols, long-chain hydrocarbons, ethers),^28,29,31,33 ionic liquids,^35,36 and supramolecular nanosolvents (SUPRASs),^32,37 have been applied in HF-LPME. SUPRASs are of interest due to their unique properties. SUPRASs, also referred to as coacervates,³⁸ which are used in surfactant liquid–liquid phase separation,³⁹ are nanostructured liquids constructed from three-dimensional aggregates of amphiphilic compounds. The supramolecular solvent produced from coacervation of decanoic acid aqueous vesicles in the presence of the tetrabutylammonium (Bu₄N⁺) cation has been utilized as an extraction solvent in numerous literature.^{32,37,40–44} Two characteristics give the alkyl carboxylic acid-based coacervates a high potential for analytical extraction processes. First, the polar region of molecular aggregates comprises protonated and deprotonated carboxylic groups and ammonium groups; hence, various types of interactions (e.g., electrostatic, π–cation, hydrogen bonds, formation of mixed aggregates, etc.) can be established with the analytes of interest, in addition to the hydrophobic interactions in the hydrocarbon region.³⁷ Second, vesicles have a number of available solubilization moieties; therefore, high concentrations of polar and non-polar analytes can be solubilized in each aggregate.^32,37,41

In this context, the aim is to develop an HF-LPME method based on applying supramolecular nanosolvent for coextraction and determination of some priority acidic, basic and amphiprotic pollutants in various samples. To the best of our knowledge, there is no previous report on the coextraction of acidic, basic and amphiprotic pollutants using a supramolecular solvent-based hollow fiber liquid phase microextraction method. The unique properties of this type of solvent made the coextraction of the analytes of interest feasible. Although direct extraction with the supramolecular solvent may be easier and faster than the SS-HF-LPME method, the selectivity and repeatability of SS-HF-LPME can be greatly improved due to the protection of the acceptor phase as well as efficient sample microfiltration through the pores of the hollow fiber. Central composite design (CCD) in combination with the desirability function (DF) approach has been utilized to develop a predictive model for simulation and optimization of the SS-HF-LPME method. Finally, the optimized procedure was applied to determine the analytes in various real samples, with satisfactory results.

2. Experimental

2.1. Chemicals and reagents

4-NP, 3-NA, 1-A2N, acid red 88, alizarin yellow GG and methylene blue were purchased from Sigma-Aldrich (Milwaukee, WI, USA). Diphenhydramine and sodium diclofenac were kindly donated by Darou Pakhsh (Tehran, Iran) and used without further purification. Decanoic acid (DA), tetrabutylammonium hydroxide (Bu₄N⁺), ammonium hydroxide (28% w/v), NaCl, 1-octanol, 1-nonanol, 1-decanol, 1-undecanol, n-hexadecane, which were all of analytical grade, were supplied by Merck (Darmstadt, Germany). HPLC-grade acetonitrile (ACN) and methanol (MeOH) were purchased from Caledon (Georgetown, Ontario, Canada). Ultrapure water was prepared using a Milli-Q system from Millipore (Bedford, MA, USA). Rain and snow water samples were collected during April 2013 and February 2014, respectively. A river water sample was collected from the Karaj River (Karaj, Iran). A wastewater sample was obtained from a pharmaceutical factory (Tehran, Iran) and a dam water sample was collected at the Ilam Dam (Ilam, Iran).

2.2. Equipment

2.2.1. Chromatographic conditions and equipment. Analysis of the standard and test samples was performed using a Shimadzu SCL-10AVP HPLC instrument from the Shimadzu Company (Tokyo, Japan) combined with an LC-10AVP pump, SPD-M10AVP diode array detector (DAD), a Rheodyne 7725i (PerkinElmer, USA) injector, along with a 20 μL sample loop. The LC-solution program for LC was used to perform data processing. A Capital HPLC column (Scotland, UK) ODS-H C₁₈ (250 mm × 4.6 mm, i.d. 5 μm) was employed for all separations. The mobile phase was a mixture of deionized water and acetonitrile (50 : 50, v/v) for 12 min and 100% acetonitrile for 3 min at a flow rate of 1 mL min⁻¹, and the detector wavelength was set at 230, 240 and 315 nm for 3-NA, 1-A2N and 4-NP, respectively. The pH of solutions was measured by using a Metrohm digital pH meter 827 equipped with a glass calomel electrode. In the extraction procedure, an 8.5 mL sample vial, and an MR 3001 heating-magnetic stirrer from the Heidolph Company (Kelheim, Germany) were used. An EBA 20 Hettich centrifuge (Oxford, UK) and a 50 μL Hamilton HPLC syringe (Reno, NV, USA) were employed too.

2.2.2. Dynamic light scattering measurements. Dynamic light-scattering (DLS) measurements were carried out with a Malvern Zetasizer Nano ZS using Dispersion Technology.

2.3. Preparation of standard solutions and real samples

Stock solutions of pollutants, dyes and drugs (1000 mg L⁻¹) were prepared in HPLC-grade methanol, stored in a fridge at 4 °C and brought to ambient temperature just prior to use. Mixed working solutions of the analytes at different concentrations were prepared by dilution with ultrapure water or deionized water containing various NaCl concentrations. The water samples were filtered through a Millipore 0.22 μm cellulose acetate filter before the extraction process. Spiked/non-spiked rain water samples (8 mL) were used without any dilution.

2.4. Preparation of the supramolecular solvent

The SUPRAS was prepared by mixing 5.15 g of DA and 15.6 mL of tetrabutylammonium hydroxide in 200 mL distilled water at pH 7 ± 0.1. The mixture was stirred at 1200 rpm for 10 min to dissolve the DA.^32,37 Phase separation was performed by centrifugation of the mixture at 4000 rpm for 5 min and the obtained SUPRAS was used for further experiments.

2.5. SS-HF-LPME procedure

Accurel Q3/2 polypropylene hollow fiber membrane (200 μm wall thickness, 600 μm I.D. and 0.2 μm pore size) was obtained from the Membrana Company (Wuppertal, Germany) and used for all experiments. Hollow fibers were ultrasonically cleaned with acetone for 5 min. Each dried fiber was cut manually into 10.0 cm segments, which accommodate approximately 27 μL of the receiving phase. Afterward, 8.0 mL of the sample solution (pH, 9.0, adjusted with a dilute NaOH solution; NaCl concentration, 20% w/v) containing 0.1 mg L⁻¹ of the target analytes was transferred into an 8.5 mL vial (48 mm height × 7.5 mm diameter) with a 4 mm × 7 mm magnetic stirrer bar. The sample vial was placed on the magnetic stirrer and a 50 μL Hamilton microsyringe (Bonaduz, Switzerland) was used to introduce the receiving phase (40% SUPRAS in 1-decanol) into the hollow fiber. A 35 μL aliquot of the receiving phase was then withdrawn into the microsyringe and its needle was inserted into the lumen of the hollow fiber. Thereafter, the fiber was inserted in the organic phase (40% SUPRAS in 1-decanol) for 90 s and the excess of the organic phase was carefully removed by washing the outside of the hollow fiber with ultrapure water. Subsequently, the receiving phase was injected into the lumen of the hollow fiber and the end of the hollow fiber was sealed with a piece of aluminum. The U-shaped hollow fiber was immersed into the sample solution. The extraction was performed at room temperature and the sample was stirred at 1250 rpm during extraction for 30 min. After extraction, the fiber was removed from the sample vial, the end of the hollow fiber was opened, and the receiving phase was retracted into the microsyringe. Finally, 20 μL of receiving phase was injected into the HPLC-PDA system for subsequent analysis.

2.6. Response surface methodology and desirability function

Traditional optimization methods with successive variations in variables such as a one-factor-at-a-time (OVAT) approach are still used, although it is well accepted that they are relatively time-consuming and expensive for a large number of variables and frequently fail to predict the optimum condition.^45,46 The major drawback of the OVAT approach is the lack of inclusion of the interactive effects among variables.⁴⁷ Therefore, in order to optimize the preconcentration of the model analytes by the proposed method, a central composite design (CCD) in combination with desirability function (DF) was employed. It is worthy of note that, for an experimental design involving four variables expressed by CCD, linear, quadratic and cross terms can be involved. The precise optimum point can be obtained with the aid of response surface methodologies, exhibiting relationships between variables and responses graphically.⁴⁸

Finding optimum conditions for a single response is usually relatively simple, but in practice the problems are often more complex and the studied phenomena are described by a number of responses. Certain responses can oppose one another: changes in a factor which promote one response may have a suppressing effect on the others, etc.⁴⁹ To solve this problem, in 1980, Derringer and Suich applied an overall response to optimize multiple responses by developing the DF.^50,51 Therefore, in the case of multiple response optimizations, the Derringer function or DF can be employed, since it is the most critical and most widely applied multi-criteria methodology in analytical procedures.⁵¹ At first, in the DF approach, each predicted response is transformed to a dimensionless desirability value (d) and then all transformed responses are combined into one particular response. The scale of the individual DF ranges between 0 and 1, while for the most desirable response d is equal to 1 and for a completely undesired response d is 0.⁵² Different transformations on data may be implemented depending on whether the response is optimum when it is maximized, minimized, or at a predefined value.⁵³

In this work, the experimental design matrix and data analysis were carried out by the Design-Expert statistical software program (7.0.0 trial version).

3. Results and discussion

3.1. Size determination of SUPRAS

The size and morphology of the nano-sized aggregates was explored by the DLS technique. The DLS size distribution of aggregates is depicted in Fig. 1S (ESI†). The peak centered at approximately 1–2 nm corresponds to aqueous micelles. The peaks appearing at 28–59 and 342–531 nm are related to vesicles. Moreover, the results revealed that vesicles are the dominant type of aggregate in the SUPRASs.

3.2. Optimization of SS-HF-LPME parameters

Before defining any specific limits for performing CCD, some pilot experiments should be carried out to evaluate the approximate domains for each factor. The factors influencing the extraction capability of the proposed method, such as pH of sample, membrane solvent, percentage of SUPRAS, extraction time, salt content of sample solution and stirring rate, were investigated and optimized. Out of these six factors, membrane solvent and stirring rate were selected using the one-variable-at-a-time method. Stirring rate was fixed at 1250 rpm, since observations showed that by increasing stirring rate up to 1250 rpm, the extraction of the target analytes was increased as well. The volume and shape of the vial was such that no air bubbles were formed at such a high speed and extraction kinetics would be promoted. The optimization of the four other factors was performed using central composite design in combination with the desirability function approach (CCD-DF).

3.2.1. Selection of membrane solvent. Compatibility with the lipophilic polypropylene hollow fiber, low water solubility to prevent dissolution into the aqueous phase, affinity for target compounds, reasonably higher solubility of analytes in the organic phase than in the aqueous phase, and low volatility, which will restrict solvent evaporation during extraction, are several important criteria for the selection of organic solvent as a liquid membrane in order to achieve the highest enrichment factor.^28,54 Based on the required characteristics, it was observed (Fig. 1) that 1-decanolcontaining SUPRAS was the most appropriate not only for less risk of solvent loss during longer extraction time but also due to the unique interactions between the solvent and polar analytes (hydrophobic, electrostatic, hydrogen bonding and π–cation interactions) that result in elevated coextraction efficiency. Besides, in the case of 1-decanol the results were more reproducible than for the other solvents. It is worth noting that all solvents tested contained 50% SUPRAS. The viscosity of 1-decanol (viscosity = 12.05 cP, polarity index = 0.37) is higher than those of 1-octanol (viscosity = 7.77 cP, polarity index = 0.54) and 1-nonanol (11.7 cP, polarity index = 0.41) and is lower than that of 1-undecanol (viscosity = 17.2 cP, polarity index = 0.27). It can be claimed that the higher viscosity of 1-decanol leads to its stability during the extraction process. Moreover, the polarity of 1-decanol is higher than that of undecanol. However, most of the target analytes have low partition coefficients, and so there was no possibility of good extraction capability with non-polar solvents such as n-hexadecane (viscosity = 3.45, polarity index = 0.21).

Fig. 1 Effect of organic solvent on the extraction efficiency. Conditions: sample volume, 8.0 mL; stirring rate, 1250 rpm; extraction time, 45 min; concentration of analytes, 0.5 mg L⁻¹; pH of sample, 10; 50% v/v SUPRA, without salt addition.

3.2.2. Central composite design and desirability function. In the next step, the affecting factors were selected based on preliminary experiments and optimized by a CCD experiment. In other words, CCD was utilized to optimize the effect of four factors (pH of sample, extraction time, percentage of SUPRAS and salt content of sample solution). According to the experimental equation obeying CCD – N = 2^f + 2f + C₀, where f is the number of variables and C₀ is the number of center points – f and C₀ were set at 4 and 6, respectively, which meant that 30 trials should be performed.⁵⁵

The following equation is implemented in order to find the best joint response acquisition (DF), also named the geometric mean (Geo mean):

where r_i is the importance of each variable relative to the others. A matter of the utmost importance is maximization of DF in the optimization procedure, i.e. when DF (ranging from 0 to 1) is a non-zero value, all the variables which are simultaneously optimized can be supposed to have a desirable value.⁴⁸ Obtaining an appropriate set of conditions that will meet all the determined criteria is the main goal of an optimization procedure, and achieving a DF value of 1 is not required. The results of the CCD were investigated according to the criteria assigned based on desirable levels of factors and responses (Table 1) in order to find the best extraction conditions. To obtain the desired extraction efficiency as an objective function, Geo mean, as an indicator of extraction efficiency, was maximized. It is worth noting that initial data preprocessing, i.e., normalizing the related responses of each analyte, is necessary before data analysis. Subsequently, the obtained DF would be an input value for CCD.⁵⁶

Table 1 Experimental variables and levels of the central composite design (CCD)

	Level			Star points (α = 2)
	Lower	Central	Upper	−α	+α
A: pH	6.0	7.5	9.0	4.5	10.5
B: extraction time (min)	20	30	40	10	50
C: SUPRAS (%, v/v)	20	40	60	0	80
D: salt content (%, w/v)	10	15	20	5	25

---

The experimental data were in good accordance with the quadratic polynomial equation (Table 2). Analysis of variance (ANOVA) was used to evaluate the significant terms in the model for each response and the related significances were judged by the F-statistic calculated from the data (Table 1S, ESI†). The model F-value of 6.49 (p-value = 0.0004) implies that the model is significant, and there is only a 0.04% chance that a model F-value of 6.49 could occur due to noise. The p-value for lack of fit (LOF) in the ANOVA table was higher than 0.05, which confirms the LOF is not significant relative to the pure error.⁵⁶

Table 2 Sequential model sum of squares

Source	Sum of squares	df	Mean square	F-Value	p-Value prob. > F
Mean vs. total	16.47	1	16.47
Linear vs. mean	0.17	4	0.042	4.58	0.0065
2FI vs. linear	1.608 × 10^−4	6	2.679 × 10^−5	2.200 × 10^−3	1.0000
Quadratic vs. 2FI	0.17	4	0.044	11.51	0.0002	Suggested
Cubic vs. quadratic	0.012	8	1.500 × 10^−3	0.23	0.9706	Aliased
Residual	0.045	7	6.411 × 10^−3
Total	16.87	30	0.56

---

Two-dimensional (2-D) color maps are depicted in Fig. 2, representing high desirability with warm “red” and low desirability with cold “green” colors. The optimum point can be selected from the constructed design space by visual examination, which is in accordance with the highest desirability value condition. In consequence, the highest d value of 0.916 was obtained at pH = 9.0, extraction time = 30 min, SUPRAS percentage = 40% v/v in decanol and salt content = 20% w/v as the optimum conditions.

Fig. 2 2-D model depicting the overall desirability function and the response surfaces obtained for the global desirability function.

The sample pH determines the form of analytes in aqueous solution, which plays an important role in the coextraction of target analytes. At pH 9.0, 4-NP exists in anionic form and the other analytes are in their neutral forms. Therefore, 4-NP can interact with SUPRAS and 1-decanol through hydrophobic, electrostatic (between negative charge of 4-NP and positive sites of TBA), hydrogen bonding and π–cation interactions.^37,43,57 3-NA and 1-A2N can interact through hydrophobic interaction, hydrogen bonding and π–cation interaction. These mixed-mode mechanisms and multiple binding sites would provide a good solubilization of the model analytes in SUPRAS, thus assisting efficient extraction of the analytes. The extraction efficiencies of target analytes were improved dramatically by increasing SUPRAS content from 0 to 40%, and then decreased, which may be due to an increase in solvent viscosity which decreases the mass transfer rate.⁵⁷ Furthermore, the results showed that the coextraction of the analytes is possible in acidic medium. The extraction of positively charged (protonated) species can be a result of ion pair formation between decanoate and protonated 3-NA and 1-A2N species.³⁷ However, the best extraction efficiency was obtained in basic medium. The extraction efficiency of target analytes was augmented dramatically by increasing extraction time from 10 to 30 min, and then a decrease occurred which may have been due to solvent loss and the formation of air bubbles, which would suppress the extraction efficiency. The extraction efficiency of the analytes increased by addition of NaCl to the aqueous solution up to 20% w/v. According to the salting-out effect, the solubility of analytes in the aqueous phase will be decreased and their partitioning into the organic phase will be increased. In higher NaCl concentrations, the viscosity of the aqueous solution may hinder the mass transfer process and lead to lower extraction efficiency of the analytes.⁶¹

Through the statistical processes, the response surfaces obtained for the global desirability function based on the design and modeled CCD are depicted in Fig. 2, in which some of the surfaces obtained for the different factor combinations are presented. As can be appreciated, the global desirability function value was about 0.916 for all these possible experimental conditions. According to the overall results of the optimization study, pH = 9.0, extraction time = 30 min, SUPRAS percentage = 40% v/v in decanol and salt content = 20% w/v were selected as the optimum values.

3.3. Applicability of SS-HF-LPME method for coextraction of other compounds

Under the optimized conditions, the performance of the proposed method was explored for simultaneous extraction of some other basic and acidic compounds. For this purpose, diphenhydramine (DPH, pK_a = 9.0) as a basic drug and sodium diclofenac (DIC, pK_a = 4.2) as an acidic drug were extracted under the optimized conditions (obtained for 4-NP, 3-NA and 1-A2N) and acceptable results were achieved. Under these conditions, preconcentration factors of 75 and 110 for DPH and DIC were obtained, respectively. Moreover, the applicability of this method for extraction of acidic dyes (acid red 88 and alizarin yellow GG) and a basic dye (methylene blue) was explored. As depicted in Fig. 3, a color change was observed before and after the extraction process, indicating that the dyes were successfully extracted into the acceptor phase. For more clarity, it is worth noting that 25 mg L⁻¹ of each dye was subjected to the extraction protocol. All the results obtained confirmed the applicability of SS-HF-LPME for coextraction of various compounds, due to the mixed-mode mechanisms and multiple binding sites of SUPRAS.

Fig. 3 Photographs of dye preconcentration under the optimal conditions: (a and b) acid red 88, (c and d) alizarin yellow GG and (e and f) methylene blue; (a), (c) and (e) are for before the extraction initiation; (b), (d) and (f) are for after an extraction time of 30 min.

3.4. Analytical figures of merit of SS-HF-LPME

The analytical performance of the proposed method is tabulated in Table 3. Quality characteristics of the current method were evaluated under the final optimized conditions. Under the optimized conditions, limit of detection (LOD), regression equation, correlation of determination (r²), dynamic linear range (DLR), preconcentration factor (PF), and extraction recovery (R%) of each analyte were evaluated. LOD values were calculated at the signal to noise ratio of 3. The repeatability (within-day RSDs, n = 5 samples, at 30 μg L⁻¹ level of the analytes) and reproducibility (between-day RSDs, n = 3 days, at 30 μg L⁻¹ of the analytes) of the method for the determination of the target analytes were equal to or less than 6.9% and 12.9%, respectively. Enrichment factor (EF) values were calculated as the ratio of the slopes of the calibration curves before and after preconcentration. The extraction recoveries were calculated by the following equation:²⁸where EF is the enrichment factor and V_f and V_i are the organic phase and aqueous sample volume, respectively.

Table 3 Analytical figures of merit of SS-HF-LPME method

Analyte	LOD (μg L^−1)	LOQ^a (μg L^−1)	DLR (μg L^−1)	Regression equation	r^2	EF^b	ER^c (%)	RSD^d (%) (within-day)	RSD^d (%) (between-day)
a Limit of quantitation.b The enrichment factor for each analyte was calculated as the ratio of the slopes of the calibration curves with and without preconcentration.c Extraction recovery.d The relative standard deviation (n = 5 samples for within-day and n = 3 days for between-day) was obtained at 30 μg L^−1 level of the analytes.e Concentration in μg L^−1.
4-NP	0.20	0.5	0.5–400	y = 20 473C^e − 4736.5	0.9998	166	56.1	6.9	12.9
3-NA	0.15	0.5	0.5–400	y = 23 590C + 1857.5	0.9986	178	60.2	5.4	9.1
1-A2N	0.10	0.5	0.5–400	y = 33 802C + 6254.2	0.9995	211	71.1	4.1	8.5

---

3.5. Analysis of real samples

To evaluate the accuracy and also applicability of the procedure for complicated samples, the coextraction of the aforementioned model compounds from real water samples (snow water, rain water, river water, dam water and pharmaceutical wastewater) was performed. Fig. 4 and 2S† present the chromatograms of the rain, snow, river, dam and wastewater samples before and after spiking. Nitrophenols such as 4-NP are formed in the atmosphere from the photochemical reaction of benzene with nitrogen monoxide in highly polluted air. Hence the presence of 4-NP in snow and rain water samples in highly polluted areas is expected, in contrast to the river water, which may be polluted or not polluted due to, firstly, probably originating from an unpolluted area, and secondly, probably containing 4-NP even lower than the LOD of the method. Table 4 shows that the results of the three replicate analyses of each real sample obtained by the proposed method are in good agreement with the spiked levels.

Fig. 4 The chromatograms of (A) the snow water sample (a) before spiking, (b) spiked at 10 μg L⁻¹ of each analyte; (B) the rain water sample (a) before spiking, (b) spiked at 10 μg L⁻¹ of each analyte; and (C) the river water sample (a) before spiking, (b) spiked at 10 μg L⁻¹ of each analyte after SS-HF-LPME under optimized conditions.

Table 4 Determination of the target analytes in various matrixes^a

Sample	Analyte	Cadded	Cfound	RR^b (%)	RSD (%) (n = 3)
a All concentrations are based on μg L^−1.b Relative recovery.
Snow water	4-NP	—	11.2	—	7.0
		10.0	21.9	107	8.8
	3-NA	—	n.d.	—	—
		10.0	9.7	97	6.4
	1-A2N	—	n.d.	—	—
		10.0	9.0	90	7.0
Rain water	4-NP	—	7.8	—	6.0
		10.0	16.9	91	6.8
	3-NA	—	n.d.	—	—
		10.0	10.9	109	7.5
	1-A2N	—	n.d.	—	—
		10.0	10.2	102	5.3
River water	4-NP	—	n.d.	—	—
		10.0	9.5	95	4.6
	3-NA	—	n.d.	—	—
		10.0	8.9	89	6.4
	1-A2N	—	n.d.	—	—
		10.0	10.4	104	5.0
Dam water	4-NP	—	n.d.	—	—
		10.0	10.6	106	7.1
	3-NA	—	n.d.	—	—
		10.0	9.3	93	5.8
	1-A2N	—	n.d.	—	—
		10.0	9.7	97	4.1
Wastewater	4-NP	—	n.d.	—	—
		10.0	8.6	86	8.0
	3-NA	—	n.d.	—	—
		10.0	9.1	91	6.3
	1-A2N	—	n.d.	—	—
		10.0	9.5	95	7.5

---

3.6. Comparison of SS-HF-LMPE with other alternative methods

Table 5 compares the figures of merit of the SS-HF-LPME method and the alternative methods for the extraction of the target analytes in various matrixes. The results of the comparison demonstrate that the current method has a wide linear dynamic range and low detection limit and also entails the advantage of the coextraction of acidic, basic and amphiprotic compounds in contrast to most of the other methods. In addition, this method requires only a very small amount of an environmentally friendly organic solvent. Utilizing fresh acceptor phase and discarding the hollow fiber after each extraction eliminates the possibility of sample carryover and ensures repeatability and reproducibility.

Table 5 Comparison of SS-HF-LPME with alternative methods used for the extraction and determination of the target analytes^f

Analytes	Method	Sample	DLR	LOD	LOQ	RSD (%)	Ref.
a Ion pair-based surfactant-assisted microextraction.b Capillary liquid chromatography.c Dispersive micro-solid phase extraction.d Magnetic solid phase extraction.e Directly suspended droplet liquid–liquid–liquid microextraction.f All concentrations are based on μg L^−1.
4-NP	IP-LPME^a-HPLC-DAD	Tap, mineral and rain water	0.2–75	0.1	—	≤6.3	2
4-NP	HF-LPME-CLC^b	Sea water	1–200	0.5	—	≤6.2	8
3-NA, 4-NP 1-A2N	D-μ-SPE^c-HPLC-DAD	Rain, snow and river water	0.5–600	0.1–0.25	0.5–1	≤8.5	56
4-NP	MSPE^d-HPLC-UV	Tap, river and rain water	0.75–100	0.3	0.75	4.9	58
3-NA	HF-LPME-HPLC-UV	Tap, river and ground water	1–1000	0.1	1	4.1	59
3-NA	DSD-LLLME^e-HPLC-UV	Tap, river and ground water	5–1500	1.0	5	4.9	60
3-NA, 4-NP 1-A2N	SS-HF-LPME-HPLC-DAD	Rain, snow, river, dam and wastewater	0.5–500	0.1–0.2	0.5	≤6.9	Current method

---

4. Conclusion

In the current method, for the first time, a novel strategy for coextraction of acidic, basic and amphiprotic pollutants (with different polarities) using supramolecular nanosolvent-based hollow fiber liquid phase microextraction was proposed. The polar region of the nanosolvent is composed of protonated and deprotonated carboxylic groups and ammonium groups; therefore, various types of interactions (e.g., electrostatic, π–cation, hydrogen bonds, formation of mixed aggregates, etc.) can be established with analytes of interest, in addition to hydrophobic interactions in the hydrocarbon region. Moreover, vesicles have a number of available solubilization moieties; therefore, high concentrations of polar and non-polar analytes with different natures (acidic, basic or amphiprotic) can be solubilized in each aggregate. The method described is simple, fast and cheap. Regarding the few microliters of organic solvent consumed and its environmentally friendly nature, this strategy can be considered a green technique. Utilizing fresh acceptor phase and discarding the hollow fiber after each extraction has led to high reproducibility and repeatability of the method as well as avoiding carryover problems.

References

1. T. Zhang, Q. Lang, D. Yang, L. Li, L. Zeng, C. Zheng, T. Li, M. Wei and A. Liu, Electrochim. Acta, 2013, 106, 127–134.
2. M. Moradi, Y. Yamini, J. Kakehmam, A. Esrafili and M. Ghambarian, J. Chromatogr. A, 2011, 1218, 3945–3951.
3. J. Yang and H.-C. Liu, Analyst, 2000, 125, 1605–1610.
4. H. Wu and L. M. Du, Spectrochim. Acta, Part A, 2007, 67, 976–979.
5. Q. Yang, X. Chen and X. Jiang, Chromatographia, 2013, 76, 1641–1647.
6. H. Kataoka, J. Chromatogr. A, 1996, 733, 19–34.
7. A. Young, G. Lai, B. Hung, A. Yuen and Y. He, Chromatographia, 2011, 74, 83–88.
8. L. Zhu, L. Zhu and H. K. Lee, J. Chromatogr. A, 2001, 924, 407–414.
9. T. J. Edison and M. G. Sethuraman, Spectrochim. Acta, Part A, 2013, 104, 262–264.
10. M. N. Sánchez, P. M. Santos, C. P. Sappó, J. L. Pérez Pavón and B. M. Cordero, Talanta, 2014, 119, 375–384.
11. K. T. Chung, G. E. Fulk and M. Egan, Appl. Environ. Microbiol., 1978, 35, 558–562.
12. D. Méndez-Paz, F. Omil and J. M. Lema, Enzyme Microb. Technol., 2005, 36, 264–272.
13. A. Gottlieb, C. Shaw, A. Smith, A. Wheatley and S. Forsythe, J. Biotechnol., 2003, 101, 49–56.
14. A. D. Dimou, V. A. Sakkas and T. A. Albanis, Int. J. Environ. Anal. Chem., 2004, 84, 173–182.
15. C. Tong, Y. Guo and W. Liu, Chemosphere, 2010, 81, 430–435.
16. J. Patsias and E. Papadopoulou-Mourkidou, J. Chromatogr. A, 2000, 904, 171–188.
17. Y. Zhu, M. Wang, H. Du, F. Wang, S. Mou and P. R. Haddad, J. Chromatogr. A, 2002, 956, 215–220.
18. M.-Q. Cai, X.-Q. Wei, C.-H. Du, X.-M. Ma and M.-C. Jin, J. Chromatogr. A, 2014, 1349, 24–29.
19. O. Jáuregui and M. T. Galceran, Anal. Chim. Acta, 1997, 340, 191–199.
20. A. Sarafraz-Yazdi, M. Sayyar Ardaki and A. Amiri, J. Sep. Sci., 2013, 36, 1629–1635.
21. S. Mishra, V. Singh, A. Jain and K. K. Verma, Analyst, 2001, 126, 1663–1668.
22. C. M. Knapp and J. J. Breen, J. Chromatogr. A, 1998, 799, 289–299.
23. X. Guo, J. Lv, W. Zhang, Q. Wang, P. He and Y. Fang, Talanta, 2006, 69, 121–125.
24. A. A. Asgharinezhad, N. Mollazadeh, H. Ebrahimzadeh, F. Mirbabaei and N. Shekari, J. Chromatogr. A, 2014, 1338, 1–8.
25. Q. Zhou, G. Jiang, J. Liu and Y. Cai, Anal. Chim. Acta, 2004, 509, 55–62.
26. N. Li and H. K. Lee, J. Chromatogr. A, 2001, 921, 255–263.
27. H. Ebrahimzadeh, N. Mollazadeh, A. A. Asgharinezhad, N. Shekari and F. Mirbabaei, J. Sep. Sci., 2013, 36, 3783–3790.
28. H. Ebrahimzadeh, N. Shekari, Z. Saharkhiz and A. A. Asgharinezhad, Talanta, 2012, 94, 77–83.
29. H. Ebrahimzadeh, A. A. Asgharinezhad, H. Abedi and F. Kamarei, Talanta, 2011, 85, 1043–1049.
30. L. Xu, C. Basheer and H. K. Lee, Developments in single-drop microextraction, J. Chromatogr. A, 2007, 1152, 184–192.
31. S. Pedersen-Bjergaard and K. E. Rasmussen, Anal. Chem., 1999, 71, 2650–2656.
32. M. Moradi, Y. Yamini, F. Rezaei, E. Tahmasebi and A. Esrafili, Analyst, 2012, 137, 3549–3557.
33. T. G. Halvorsen, S. Pedersen-Bjergaard and K. E. Rasmussen, J. Chromatogr. A, 2001, 909, 87–93.
34. E. Psillakis and N. Kalogerakis, TrAC, Trends Anal. Chem., 2003, 22, 565–574.
35. R. Fortunato, C. A. M. Afonso, M. A. M. Reis and J. G. Crespo, J. Membr. Sci., 2004, 242, 197–209.
36. J. F. Peng, J. F. Liu, X. L. Hu and G. B. Jiang, J. Chromatogr. A, 2007, 1139, 165–170.
37. F. Rezaei, Y. Yamini, M. Moradi and B. Daraei, Anal. Chim. Acta, 2013, 804, 135–142.
38. E. Yilmaz and M. Soylak, RSC Adv., 2014, 4, 47396–47401.
39. M. Khan and M. Soylak, RSC Adv., 2015, 5, 62433–62438.
40. F. J. López-Jiménez, A. Ballesteros-Gómez and S. Rubio, Food Chem., 2014, 143, 341–347.
41. A. Moral, M. D. Sicilia and S. Rubio, J. Chromatogr. A, 2009, 1216, 3740–3745.
42. F. Rezaei, Y. Yamini and M. Moradi, J. Chromatogr. A, 2014, 1327, 155–159.
43. M. Moradi and Y. Yamini, J. Chromatogr. A, 2012, 1229, 30–37.
44. A. Moral, M. D. Sicilia and S. Rubio, Anal. Chim. Acta, 2009, 650, 207–213.
45. S. J. Kalil, F. Maugeri and M. I. Rodrigues, Process Biochem., 2000, 35, 539–550.
46. R. Gheshlaghi, J. M. Scharer, M. Moo-Young and P. L. Douglas, Anal. Biochem., 2008, 383, 93–102.
47. J. Zolgharnein, A. Shahmoradi and J. B. Ghasemi, J. Chemom., 2013, 27, 12–20.
48. D. D. Stephan, J. Werner and R. P. Yeater, Essential Regression and Experimental Design for Chemists and Engineers, MS Excel Add in Software Package, 1998–2001.
49. A. M. Carro and R. A. Lorenzo, Analyst, 2001, 126, 1005–1010.
50. L. V. Candioti, M. M. De Zan, M. S. Cámara and H. C. Goicoechea, Talanta, 2014, 124, 123–138.
51. G. Derringer and R. Suich, J. Qual. Tech., 1980, 12, 214–219.
52. H. Heidari, H. Razmi and A. Jouyban, J. Sep. Sci., 2014, 37, 1467–1474.
53. B. Dejaegher and Y. Vander Heyden, Acta Chromatogr., 2009, 21, 161–201.
54. E. Psillakis and N. Kalogerakis, J. Chromatogr. A, 2001, 938, 113–120.
55. H. Ebrahimzadeh, F. Mirbabaei, A. A. Asgharinezhad, N. Shekari and N. Mollazadeh, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2014, 947–948, 75–82.
56. A. A. Asgharinezhad and H. Ebrahimzadeh, J. Chromatogr. A, 2015, 1412, 1–11.
57. M. Moradi, Y. Yamini, F. Rezaei, E. Tahmasebi and A. Esrafili, Analyst, 2012, 137, 3549–3557.
58. E. Tahmasebi, Y. Yamini, S. Seidi and M. Rezazadeh, J. Chromatogr. A, 2013, 1314, 15–23.
59. A. Sarafraz-Yazdi, F. Mofazzeli and Z. Es'haghi, Talanta, 2009, 79, 472–478.
60. A. Sarafraz Yazdi, F. Mofazzeli and Z. Es'haghi, J. Chromatogr. A, 2009, 1216, 5086–5091.
61. A. A. Asgharinezhad and H. Ebrahimzadeh, Anal. Bioanal. Chem., 2016, 408, 473–486.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra23488c

---