Journal of Chromatography A, 1361 (2014) 291–298

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Dispersive liquid–liquid microextraction for the determination of phenols by acetonitrile stacking coupled with sweeping-micellar electrokinetic chromatography with large-volume injection Hui He a , Shuhui Liu a,b,∗ , Zhaofu Meng c , Shibing Hu c a

College of Science, Northwest A&F University, Yangling, China State Key Laboratory of Crop Stress Biology in Arid Areas, Yangling, China c College of Natural Resources and Environment, Northwest A&F University, Yangling, China b

a r t i c l e

i n f o

Article history: Received 5 May 2014 Received in revised form 30 July 2014 Accepted 5 August 2014 Available online 13 August 2014 Keywords: Dispersive liquid–liquid microextraction Capillary electrophoresis Acetonitrile stacking Sweeping Phenols

a b s t r a c t The current routes to couple dispersive liquid–liquid microextraction (DLLME) with capillary electrophoresis (CE) are evaporation of water immiscible extractants and backextraction of analytes. The former is not applicable to extractants with high boiling points, the latter being effective only for acidic or basic analytes, both of which limit the further application of DLLME-CE. In this study, with 1-octanol as a model DLLME extractant and six phenols as model analytes, a novel method based on acetonitrile stacking and sweeping is proposed to accomplish large-volume injection of 1-octanol diluted with a solvent–saline mixture before micellar electrokinetic chromatography. Brij-35 and ␤-cyclodextrin were employed as pseudostationary phases for sweeping and also for improving the compatibility of sample zone and aqueous running buffer. A short solvent–saline plug was used to offset the adverse effect of the water immiscible extractant on focusing efficiency. The key parameters affecting separation and concentration were systematically optimized; the effect of Brij-35 and 1-octanol on focusing mechanism was discussed. Under the optimized conditions, with ∼30-fold concentration enrichment by DLLME, the diluted extractant (8×) was then injected into the capillary with a length of 21 cm (42% of the total length), which yielded the overall improvements in sensitivity of 170–460. Limits of detection and qualification ranged from 0.2 to 1.0 ng/mL and 1.0 to 3.4 g/mL, respectively. Acceptable repeatability lower than 3.0% for migration time and 9.0% for peak areas were obtained. The developed method was successfully applied for analysis of the phenol pollutants in real water samples. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Since its inception in 2006 [1], dispersive liquid–liquid microextraction (DLLME) has attracted much attention due to simplicity of operation, low solvent usage and cost, high enrichment efficiency and very short extraction time compared to other liquid phase microextraction techniques (LPME) [2]. It has been widely applied in determining analytes at trace levels in environmental, food and biological samples [3]. In recent years the commonly used DLLME extractants, such as halogenated hydrocarbons and aromatic hydrocarbons are gradually being replaced by low toxic ones, typically ionic liquids (ILs) [4] and low-density organic solvents [5].

∗ Corresponding author at: College of Science, Northwest A&F University, No. 3 Taicheng Road, Yangling, Shaanxi 712100, China. Tel.: +86 29 87092226; fax: +86 29 87092226. E-mail address: [email protected] (S. Liu). http://dx.doi.org/10.1016/j.chroma.2014.08.013 0021-9673/© 2014 Elsevier B.V. All rights reserved.

DLLME is mainly linked up with gas chromatography (GC) because water-immiscible extractants are volatile in GC. For highperformance liquid chromatography (HPLC), the extractants are either injected directly or evaporated to dryness before reconstitution and injection, which depends on their compatibility with the mobile phase. Capillary electrophoresis is a powerful complementary technique to GC and HPLC. Its combination with DLLME is regarded as a very attractive environmentally sustainable analytical tool and should have broad application prospects since both of them consume very small amount of organic solvents. However, due to the incompatibility of water immiscible extractants and aqueous background electrolytes (BGEs), the implementation of DLLME with CE has been somewhat slower to develop. A recent review summarized the use of microextraction combined with CE in bio-analysis [6]. For DLLME, the main mediate routines to couple with CE are evaporation of immiscible extractants [7–13] and backextraction of analytes [14–18]. However, some problems lying in these two methods limit their application.

292

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

Extraction solvents are evaporated under the air or nitrogen steam followed by reconstitution of analytes into an aqueous solution before injection. That requires much time and brings some loss to volatile and semi-volatile compounds, and is not applicable to the extractants with high boiling points including ionic liquid (ILs) and long-chain alkanols. On the other hand, DLLME followed by backextraction exists to transfer analytes from extraction solvents to aqueous solutions primarily according to the changes in hydrophobic characteristics after ionization. Nevertheless, it is impracticable for neutral analytes and less effective for hydrophobic and ionizable compounds, such as pentachlorophenol [19]. Recently, formation of a microemulsion of the DLLME extractant (chloroform) with an aqueous methanol (MeOH) allowed analytes to be introduced directly into the capillary through electrokinetic injection (EKI) [20]. Whereas, the inherent shortcomings of EKI could not be easily overcome, e.g., discriminatory sample loading amounts, unsatisfactory reproducibility, and for uncharged analytes, the effectiveness is inhibited despite the inclusion of them in ionic micelles [21]. Apparently, these drawbacks can be removed via hydrodynamic injection of immiscible extractants containing target analytes, but in fact it is not easy to combine direct injection of DLLME extractants with capillary zone electrophoresis (CZE) or micellar electrokinetic chromatography (MEKC) coupled with common on-line concentration strategies, such as stacking [22–24], SDS-sweeping [25,26] and stacking with reverse migrating micelles [27]. The problem of incompatibility between DLLME extractants and sample solutions/BGEs required by the on-line concentration techniques remains unsolved. Notably, an on-line enrichment technique—transient “pseudoisotachophoresis (p-ITP)”, better known as “acetonitrile (ACN) stacking”, has several distinctive characteristics compared with others [28,29]. For instance, (i) ca. 66% ACN and high amount of NaCl added into a sample matrix induce isotachophoresis focus for charged analytes, (ii) rapid completion of the enrichment process quickly removes the ionized analytes into the buffer solution, which detaches them from the sample zone avoiding serious influence on the subsequent separation and detection of the analytes, and (iii) it allows large-volume injection of the sample matrix to achieve a high enrichment factor (EF). Inspired by its unique merits, we came up with the following idea. A large proportion of ACN in the sample matrix produces on-line stacking, and could enable water immiscible DLLME extractants to be dissolved and directly introduced into a CE system. In addition, the loss of DLLME EF due to the dilution could be compensated by large-volume sample injection. In this context, it would be a practical and feasible approach to combine DLLME with CE by use of ACN stacking. In this study, with 1-octanol as a model DLLME extractant and six phenols as model analytes, a novel method based on ACN stacking and sweeping was proposed to accomplish large-volume injection of 1-octanol diluted with a solvent–saline mixture before micellar electrokinetic chromatography. A solvent–saline plug and a buffer containing Brij-35 and ␤-cyclodextrin (␤-CD) were used to counteract the deleterious effect of 1-octanol on stacking and improve the compatibility of the sample zone and aqueous buffer. Parameters affecting the separation and stacking efficiency were systematically optimized to achieve the best analytical performance. Once validated, the method was successfully applied for analysis of these phenols in real water samples. 2. Experimental 2.1. Chemicals Analytical standards of 2,4,6-trichlorophenol (2,4,6-TCP, 98%), 2,4-dichlorophenol (2,4-DCP, 98%), 2,5-dichlorophenol (2,5-DCP, 98%), bisphenol A (BPA, 99.8%), 4-chlorophenol (4-CP, 99%),

3-methylphenol (3-MP, 99%) were purchased from Aladdin Chemistry (Shanghai, China). Hydrochloric acid, sodium hydroxide, sodium tetraborate and ␤-CD were provided by Bodi Chemical Reagent (Tianjin, China). NaCl was obtained from Xilong Chemical Reagent (Shantou, China). Sodium dodecyl sulfate (SDS) was obtained from Sanland (Los Angeles CA, USA). Brij-35, Tween-20, cetyltrimethylammonium bromide (CTAB) and 1-octanol (99.5% pure) were purchased from Aladdin Chemistry (Shanghai, China). All reagents were of analytical grade unless indicated otherwise. HPLC-grade MeOH (≥99.9% pure), ACN (≥99.9% pure) and isopropanol (≥99.9% pure) were purchased from Kermel Chemical Reagent (Tianjin, China). Standard stock solutions of the phenols at concentration of 2 mg/mL were prepared in MeOH and stored in brown bottles at −18 ◦ C. A mixed standard stock solution of six phenols was prepared with MeOH at the concentration of 200 ␮g/mL. Working standard solutions were prepared daily by diluting the mixed stock standard solution with MeOH to the required concentrations. 2.2. Apparatus and conditions All electrophoresis experiments were performed on a Beckman P/ACETM MDQ Capillary Electrophoresis System (Beckman coulter, Fullerton, CA) equipped with a photodiode array detector (PAD). Data acquisition and instrument control were carried out using 32 Karat software (Version 8.0, Beckman Coulter). ANOVA was performed using SPSS software (Version 19.0, IBM). Uncoated fused-silica capillaries (50 cm × 50 ␮m I.D. or 50 cm × 75 ␮m I.D., with an effective length of 40 cm) were purchased from Ruifeng Optical Fiber Factory (Yongnian, Hebei Province, China). The detector wavelength was operated at 214 nm for optimization of separation, and the wavelength used for quantitative analysis was set at 214 nm for 2,4,6-TCP, 4-CP and 3-MP and 254 nm for 2,4DCP, 2,5-DCP and BPA. 214 nm, 206 nm and 254 nm were used to evaluate the purity of the phenols in real samples. Electrophoretic separations were carried out with an applied voltage of 20 kV (75 ␮m I.D.) or 28 kV (50 ␮m I.D.) under normal polarity. All operations were carried out at 25 ◦ C. The pH values of BGE solutions were adjusted by a pH meter (HANNA Instrument, Italy). Water was purified using a Millipore Direct-Q 3 system (Millipore Corporation, Bedford, MA, USA). New capillaries were rinsed at 20 psi successively with MeOH for 8 min, water for 3 min, 0.1 M HCl for 5 min, water for 3 min, 0.1 M NaOH for 30 min, water for 5 min, and finally with the running buffer for 20 min and equilibrated at 20 kV with running buffer for 10 min. Between runs, the capillary was rinsed at 20 psi sequentially with the mixture of MeOH and water (1:1, v/v) for 1 min, 0.1 M NaOH for 2 min, water for 1 min and the running buffer for 3 min. All solutions were prepared freshly everyday, submitted to ultrasonic treatment for 5 min and filtered through a cellulose acetate filter (0.22 ␮m). In ACN stacking and sweeping-MEKC, the capillary was first conditioned with the BGE consisting of 10 mM ␤-CD, 40 mM Brij-35 and 10% MeOH in 25 mM borate buffer (pH 11.2). Then the extractant (1octanol) obtained from DLLME step diluted (8×) in a 100 mM NaCl solution containing 8% (v/v) MeOH and 54% (v/v) ACN was directly injected into the capillary at 3 psi for 40 s (75 ␮m I.D.), immediately, followed by injection of the solvent–saline plug at 0.5 psi for 10 s. A voltage of 20 kV was then applied for both sample stacking and subsequent separation. With a sample injection at 0.5 psi for 3 s, the sample and the BGE used for conventional MEKC procedure was the same as in the ACN stacking and sweeping-MEKC. 2.3. Sample preparation Three environmental water samples were collected from the Huaqing pool (Xian, Shaanxi), the Weihei river (Xian, Shaanxi) and

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

the Gaoganqu river (Yangling, Shaanxi). Two sewage water samples were collected from local sewers around the Northwest A&F University, and a dirty sewage water sample from a sewage treatment plant (Ordos city, Inner Mongolia). The water samples were centrifuged at 10,000 rpm for 20 min, then the supernatant was filtrated through 0.45 ␮m disposable membranes and stored at 4 ◦ C in darkness until analysis. For migration test of BPA, boiling pure water was poured into five used and five new polycarbonate bottle; the water samples was kept in the bottles overnight at room temperature, and then transferred into glass bottles and kept at 4 ◦ C.

2.4. Surfactant-assisted DLLME procedure This DLLME step was based on Moradi’s study [30] and our previous research [31]. In brief, the ionic strength and pH of a sample were adjusted to an appropriate value (10% (w/v) NaCl; pH 2.0). An aliquot of 3.3 mL water sample was added into a disposable polyethylene pipette with a long and narrow neck, followed by the addition of 0.09 mM CTAB (as dispersive solvent) and 120 ␮L 1-octanol (as extraction solvent). The pipette was shaken for 5 min, and the resulting cloudy solution was centrifuged at 5000 rpm for 3 min. The dispersed fine droplets of the extraction phase (115 ␮L ± 2) were collected on the top of the pipette using a microsyringe, and quantitatively transferred to a 1.5 mL microtube. The extractant phase was diluted (8×) with a 100 mM NaCl solution containing 8% MeOH and 54% ACN before analysis by CE.

3. Result and discussion 3.1. Optimization of buffer composition For optimization of the key factors influencing separation and stacking of the analytes, an aliquot of spiked 1-octanol (2 ␮g/mL of each analyte) that was diluted (10×) with 5% (v/v) MeOH and 55% (v/v) ACN in 100 mM NaCl was used as a sample solution, and injection condition was kept constant at 7 psi 30 s (50 ␮m I.D. capillary) to produce sufficient stacking. The preliminary test showed that CZE mode with a borate buffer solution failed as a result of frequent current failure with the sample solutions containing 1-octanol. MEKC was therefore selected to increase the compatibility between sample zone and aqueous buffer. A set of 25 mM borate buffer solutions comprising of 10 mM ␤-CD, 25 mM Brij-35 and 10% MeOH was compared at three pH levels (9, 10.5 and 11.2). The results indicated that resolution was improved at pH 11.2. The investigation of the effect of borate buffer concentration (20–35 mM) showed that a high concentration enhanced separation and enrichment, and yet led to a longer analysis time and much Joule heating, with 25 mM borate as optimum. Three organic solvents including MeOH, ACN and isopropanol were evaluated in the range of 0–15%, and 10% of MeOH was selected in terms of good resolution. To investigate the effect of pseudostationary phase on separation and preconcentration, ionic-, nonionic- and mixed micelles including SDS, Tween-20, Brij-35 and the match among each other (SDS—Brij-35, SDS—Tween-20, Brij-35—Tween-20) were tested at a concentration of 15 mM. These results (Supplementary Fig. S1) indicated that addition of SDS in the borax buffer resulted in fluctuated baseline and broaden peaks, but good resolution and focusing effect were obtained with the nonionic surfactants, particularly Brij-35. Subsequently, the effect of the content of Brij-35 was examined; this showed when the content of Brij-35 was increased from 15 to 40 mM, the peak shapes were sharpened obviously in conjunction with a reduced migration time, and a higher concentration of

293

Brij-35 (>40 mM) gave rise to irregular current break. It was apparent that 40 mM Brij-35 exerted the best analytical performance. The enhancement of focusing effect in the presence of Brij-35 could be explained by the combined mechanism of ACN stacking and sweeping. When a positive voltage was applied, the ACN worked as “pseudo-termination ions” driving the anionic phenols quickly toward the anode direction to keep pace with the leading ion (Cl− ) speed [32] until enriching them on the boundary between sample zone and buffer zone. Upon continued electrophoresis, once the enriched analyte ions entered the buffer zone, they encountered reversely migrating Brij-35 micelles and were swept into a narrower zone. It is reasonable to assume that an increase in the Brij-35 concentration intensified the affinity of analytes and micelles, consequently enhancing sweeping efficiency. Previous studies have showed that Brij-35 could serve as pseudostationary phase for sweeping and separating phenol derivatives [33] and glycosides [34]. The effect of Brij-35 or ␤-CD used alone on sweeping was also investigated; this indicated that ␤-CD could produce the same sweeping effect as Brij-35 with the same concentration (10 mM). However, because of limited solubility of unsubstituted ␤-CD (about 16 mM), 10 mM ␤-CD used in this study was considered mainly for improving the resolution of 2,4-DCP and 2,5-DCP, while 40 mM Brij-35 for sweeping the phenols. Clearly both of them were favorable for improving the compatibility between sample zone containing 1-octanol and aqueous BGE. Apart from sweeping from Brij-35, viscosity-induced stacking might also contribute to narrowing peaks. In this part of the experiment, hydrodynamic injection velocity of the analytes was measured (7 psi). It was observed that when the buffer contained 0, 25 and 40 mM Brij-35, velocity of the injection was 34.8, 30.8 and 26.7 cm/min, respectively, which evidently revealed the greater viscosity of the BGE containing higher contents of Brij-35. Addition of polyethylene oxide (PEO) in the buffer was proved to be beneficial for field-amplified stacking [35] and ACN stacking because PEO acted as a viscosity modifier [36]. It was possible that Brij-35 had the same effect: after migrating through the low-viscous sample zone containing ACN and 1-octanol at a high speed, the target ions slowed down at the boundary between sample zone and BGE with higher viscosity, thus stacking more efficiently.

3.2. Optimization of concentration conditions 3.2.1. Sample composition In classical ACN stacking, samples were dissolved in 66% ACN solution containing 1% NaCl [28]. Here, 1-octanol was not soluble in this typical ACN-saline mixture until a small amount of MeOH was added. Since dilution factor of 1-octanol affected enrichment and separation efficiency, different dilution factors (5–12×) were firstly tested and ultimately eight-fold dilution was used in terms of better peak shape and separation. Given that ACN was usually employed in organic solvent highfield amplified stacking [37] and good compatibility of MeOH with DLLME extractants, different ratios of ACN to MeOH (54%:8%, 31%:31%, and 0%:62%) was examined to find an appropriate sample matrix for 1-octanol (Supplementary Fig. S2). The peak shapes of BPA and 3-MP were observed to be deteriorated extremely when the MeOH content was 31% or 62%; the ratio 54%:8% was therefore selected in this experiment. These results implied that though addition of MeOH played an important part in dissolving 1-octanol, the excessive amount could substantially decrease stacking efficiency and even produce bubbles due to its low boiling point. For dilution of different DLLME extractants, ratios of MeOH to ACN should be adjusted finely according to solubility of the extractants meanwhile the MeOH content should be maintained as little as possible.

294

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

A

B

Fig. 1. Effect of 1-octanol and solvent–saline plug on stacking: chromatogram (A) and current (B). Sample composition: 1-octanol diluted (8×) with 100 mM NaCl solution containing 8% (v/v) MeOH and 54% (v/v) ACN; sample injection: 13 cm (30 s at 7 psi); plug length: 0.5 cm (15 s at 0.5 psi); BGE: 10 mM ␤-CD, 40 mM Brij-35 and 10% MeOH in 25 mM borate (pH 11.2); capillaries: 50 ␮m I.D., 50 cm (total), 40 cm (effective length); separation voltage: 28 kV; detection wavelength: 214 nm; peaks: 1 (2,4,6-TCP), 2 (2,4-DCP), 3 (2,5-DCP), 4 (BPA), 5 (4-CP), 6 (3-MP), 0.5 ␮g/mL for each phenol.

The NaCl content in the sample matrix was another factor affecting p-ITP. Generally, its concentration was no more than 170 mM (approximately 1%) [32], so it was evaluated over the range of 50–150 mM in this study. Good resolution was observed with 100 mM NaCl (Supplementary Fig. S3); when the salt content was too low (50 mM) or too high (150 mM), the current drop or poor stacking occurred. Ultimately, 100 mM NaCl was adopted. 3.2.2. Effect of 1-octanol and solvent–saline plug on ACN stacking In this study, 1-octanol diluted with the solvent–saline mixture was directly introduced into the capillary to achieve large-volume sample injection. The effect of 1-octanol on stacking was in need of examination via comparison between the sample solution without

1-octanol, close to the classical ACN–saline matrix, and the diluted 1-octanol (see Section 2.2). As illustrated in Fig. 1A, in the latter case the migration time was prolonged and the peak shapes became poor and broaden; meanwhile, current difference between the BGE and the sample solution was increased, as shown in Fig. 1B. Because of large-volume injection of the sample containing 1octanol, its comparatively high viscosity reduced the EOF, which explained the longer migration time. Likewise, viscosity difference between the sample and the buffer was cut down eventually leading to poor stacking efficiency. It was important to note that in spite of the increased conduction difference between sample zone and buffer zone, stacking efficiency was not as good as that for the classical ACN-saline matrix. This result revealed that the

Table 1 Analytical performance of DLLME–ACN-stacking-MEKC. Compound

LOD (ng/mL) LOQ (ng/mL) Precision (RSD %, n = 5) Migration time 5 ng/mL 10 ng/mL 50 ng/mL Peak area 5 ng/mL 10 ng/mL 50 ng/mL Recovery ± RSD % (n = 3) 5 ng/mL 10 ng/mL 50 ng/mL EFs DLLME ACN stacking DLLME–ACN stacking a b

2,4,6-TCP

2,4-DCP

2,5-DCP

BPA

4-CP

3-MP

5–500 182.74 ± 12.74 (0.997) 196.39 ± 8.00 (0.998) 180.79 ± 7.09 (0.997)

82.37 ± 4.84 (0.999) 82.19 ± 3.95 (0.997) 82.39 ± 5.28 (0.998)

74.53 ± 4.95 (0.999) 70.93 ± 4.22 (0.998) 74.92 ± 3.40 (0.998)

107.86 ± 7.63 (0.999) 99.88 ± 6.42 (0.998) 98.23 ± 1.22 (0.998)

194.38 ± 15.82 (0.997) 184.25 ± 15.39 (0.999) 180.31 ± 4.73 (0.998)

314.98 ± 11.05 (0.999) 297.33 ± 25.65 (0.999) 296.52 ± 2.84 (0.999)

0.2 0.6

0.5 1.6

1.0 3.2

0.3 1.1

0.5 1.8

1.0 3.4

1.2 2.4 1.3

1.2 3.7 1.3

1.2 2.1 1.3

1.1 2.1 1.9

1.0 2.7 1.7

1.8 3.4 2.5

9.0 5.2 6.2

4.9 3.8 6.9

4.3 8.6 5.8

2.5 8.9 4.7

8.1 3.0 5.3

8.0 3.9 4.4

99 ± 8.5 104 ± 5.5 99 ± 6.3

104 ± 1.6 108 ± 4.3 92 ± 5.2

89 ± 1.9 97 ± 4.6 89 ± 1.4

106 ± 1.5 113 ± 3.0 107 ± 1.7

97 ± 5.7 94 ± 0.9 108 ± 0.6

100 ± 2.9 91 ± 5.6 102 ± 3.2

30 111 459

30 80 231

32 101 256

27 77 250

19 60 172

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

Linearity range (ng/mL) Slopea ± SDb (r) Pure water River water Sewage

25 95 281

Slope of linearity equation. Calculated in six replicates for each concentration level.

295

296

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

Fig. 2. Electropherograms of the phenols in different water samples analyzed by the proposed method. Sample composition: 1-octanol diluted (8×) in 100 mM NaCl solution containing 8% (v/v) MeOH and 54% (v/v) ACN; sample injection: 21 cm (40 s at 3 psi); plug length: 1 cm (10 s at 0.5 psi); capillaries: 75 ␮m I.D., 50 cm (total), 40 cm (effective length); separation voltage: 20 kV. BGE and peak identification were the same as in Fig. 1. Peak quantification: 214 nm for peaks 1, 5 and 6; 254 nm for peaks 2, 3 and 4.

viscosity difference played a more important role in ACN stacking than the conduction difference, which might just support this process. Galli et al. [38] and Cao et al. [31,34] proposed the same speculation when analyzing organic acids in natural rubber latex and glycosides in plant extracts by use of ACN stacking coupled with CE, respectively. In Kong’s study on exploring the focusing mechanism of ACN stacking, MeOH, ethanol, isopropanol and acetone were tested to replace ACN to perform stacking; no comparative results were obtained [29]. In view of the poor stacking due to the presence of 1-octanol, a 0.5-cm solvent–saline plug with the same composition as the dilution of 1-octanol was introduced after a sample to eliminate the adverse impact of 1-octanol. As illustrated in Fig. 1B, although the current profile obtained with the post-plug followed the same trace as that without the plug, use of the plug resulted in much better peak shapes, which could be comparable to that obtained without 1-octanol (Fig. 1A). Insert of the post-plug created a new boundary between the BGE and the plug zone. Owing to the quickness of ACN stacking [29], the analytes ions following the leading ion (Cl− ) migrated rapidly through the 1-octanol sample zone and the plug zone successively and finally stacked on this new boundary rather than the original one. Thus, introduction of the plug avoided the interference of 1-octanol, increased the compatibility of 1-octanol and BGE, and allowed better stacking. Furthermore, it prevented frequent current failure and thus improved robustness and repeatability. The plug length was also investigated and these results indicated that a longer plug did not further improve stacking. 3.2.3. Injection volume In this study, the maximum injection volume of the diluted 1octanol was examined with 50 and 75 ␮m I.D. capillaries. For 50 ␮m I.D. capillaries, the magnitude of the stacked peak heights gradually increased when the injection volume was changed from 180 nL to 400 nL (18–40% of the total volume), but decreased when the volume was increased to 470 nL (48% of the total volume); for

75 ␮m I.D. capillaries, all the target analytes exhibited the continued increase of peak heights when the injection volume was varied from 360 nL to 960 nL (16–42% of the total volume), and the peaks broadened at a volume of 1070 nL (48% of the total volume). Apparently, the larger the size of capillaries, the greater the maximum injection volume was. It was found out that good current stability and reproducibility could be achieved more easily when injection size was expanded. Thus, to offset the substantial loss of EF of DLLME due to dilution of the extractant, injection condition was set at 3 psi for 40 s (960 nL, 42% of the total volume) with a 75 ␮m I.D. capillary. 3.3. Validation of the method 3.3.1. Calibration curve and characteristics A comparison of the calibration curve slopes was performed to evaluate the matrix effect in pure water, a local river (clean) and sewage (close to black color and dirty). Under the optimum conditions, the curves were constructed at six different concentration levels corresponding to 5, 20, 50, 100, 200 and 500 ng/mL. Spiked samples at each level were prepared (Section 2.4) in triplicate and injected twice. As can be seen in Table 1, statistically equal slopes were obtained by using Duncan’s multiple range tests, which indicated that no significant differences were observed at a confidence level of 95% for each analyte in these three matrices. Therefore, the calibration curve obtained in pure water could be used for quantification. LODs between 0.2 and 1.0 ng/mL and LOQs between 0.6 and 3.4 ng/mL were calculated as the minimum spiked concentration yielding the S/N ratio equal to 3 and 10, respectively, which were comparable with those provided by the HPLC methods equipped with UV detector [30,31]. In this method, DLLME alone contributed 19–32 folds enrichment, and the on-line concentration played a major role in increasing EF by approximately 100 folds. In spite of the loss of EF of DLLME due to dilution of the extractant (8×), the entire procedure provided improvements in sensitivity of 172–459.

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

The relative standard deviations (RSDs) were examined at three levels of 5, 10 and 50 ng/mL, and lower than 3.0% for migration time and 9.0% for peak areas were obtained. Recoveries varied from 89% to 108%. 3.3.2. Comparison among different direct injection modes To date, three methods were reported on dealing with hydrodynamic introduction of water immiscible extractants into the CE. Conventional direct injection was demonstrated to be suitable for a low viscous ionic liquid (1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide) [39]; for a highly viscous ionic liquid (1-Butyl-3-methylimidazoium hexafluorophosphate), the rinsing buffer containing 66.7% MeOH facilitated compatibility between the ionic liquid and aqueous BGE [40]. Chung et al. [41] combined single drop microextraction (SDME) with nonaqueous CE, achieving full-capillary injection of the extractant (pentanol) and LODs of 0.06–0.19 ng/mL for weakly acidic and hydrophobic analytes. This presented method allowed large-volume injection (42% of the total volume) of the diluted (8×) DLLME extractant into an aqueous BGE by employing ACN stacking coupled with sweeping. By comparison, this procedure avoided the use of a nonaqueous BGE and refrained from the incompatibility of water-immiscible extractant and aqueous BGE. As a simple, convenient and sensitive method to combine DLLME and CE, it can serve as a universal approach in introducing other DLLME extractants into a CE system. 3.4. Analysis of the real sample To demonstrate the practicability of the proposed method, it was exploited for analysis of the phenols in 3 river water samples, 3 sewage water samples, and 10 bottle water samples prepared according to Section 2.3 for the migration test of BPA. Typical chromatograms are depicted in Fig. 2. No target peaks were observed in the river water and sewage samples. The amounts of BPA of 5 water samples stored in used polycarbonate bottles were 4.9 ± 0.03 (mean ± SD, n = 3), 12.7 ± 1.12, 356.8 ± 2.30, 15.5 ± 0.04 and 8.3 ± 2.22 ng/mL, respectively, while it was not detected in those stored in new bottles. 4. Conclusion In this study, a simple, convenient and sensitive strategy based on ACN stacking and sweeping was successfully developed to accomplish large-volume injection of the diluted (8×) DLLME extractant and on-line concentration before MEKC analysis. With addition of Brij-35 and solvent–saline plug, focusing effect was enhanced and the incompatibility of sample and aqueous buffer was diminished. This approach avoids the drawbacks of evaporation and backextraction steps, and allows the combination of CE and DLLME with high viscous and low violate extractants. Compared to other injection modes, this method does not only conquer the shortcomings of electrokinetic injection but also surmounts the limit of conventional direct injection of water–immiscible extractants and refrains from use of a nonaqueous buffer system. It can serve as a universal approach to couple LPME with CE, thus expanding the application of LPME–CE in green analytical chemistry. Acknowledgments The authors acknowledge with gratitude and appreciation financial support from Northwest A&F University (No. Z111021005) and the National Natural Science Foundation of China (NSFC no. 41271244).

297

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2014.08.013.

References [1] M. Rezaee, Y. Assadi, M.R. Milani Hosseini, E. Aghaee, F. Ahmadi, S. Berijani, Determination of organic compounds in water using dispersive liquid–liquid microextraction, J. Chromatogr. A 1116 (2006) 1–9. [2] M. Rezaee, Y. Yamini, M. Faraji, Evolution of dispersive liquid–liquid microextraction method, J. Chromatogr. A 1217 (2010) 2342–2357. [3] D. Han, K.H. Row, Trends in liquid-phase microextraction, and its application to environmental and biological samples, Microchim. Acta 176 (2012) 1–22. [4] M.J. Trujillo-Rodríguez, P. Rocío-Bautista, V. Pino, A.M. Afonso, Ionic liquids in dispersive liquid–liquid microextraction, Trends Anal. Chem. 51 (2013) 87–106. ˇ [5] L. Kocúrová, I.S. Balogh, J. Sandrejová, V. Andruch, Recent advances in dispersive liquid–liquid microextraction using organic solvents lighter than water. A review, Microchem. J. 102 (2012) 11–17. [6] I. Kohler, J. Schappler, S. Rudaz, Microextraction techniques combined with capillary electrophoresis in bioanalysis, Anal. Bioanal. Chem. 405 (2013) 125–141. [7] Y. Wen, J. Li, W. Zhang, L. Chen, Dispersive liquid–liquid microextraction coupled with capillary electrophoresis for simultaneous determination of sulfonamides with the aid of experimental design, Electrophoresis 32 (2011) 2131–2138. [8] S.S. Fortes, T. Barth, N.A. Furtado, M.T. Pupo, C.M. de Gaitani, A.R. de Oliveira, Evaluation of dispersive liquid–liquid microextraction in the stereoselective determination of cetirizine following the fungal biotransformation of hydroxyzine and analysis by capillary electrophoresis, Talanta 116 (2013) 743–752. [9] I. Kohler, J. Schappler, T. Sierro, S. Rudaz, Dispersive liquid–liquid microextraction combined with capillary electrophoresis and time-of-flight mass spectrometry for urine analysis, J. Pharm. Biomed. Anal. 73 (2013) 82–89. [10] S. Zhang, X. Yin, Q. Yang, C. Wang, Z. Wang, Determination of some sulfonylurea herbicides in soil by a novel liquid-phase microextraction combined with sweeping micellar electrokinetic chromatography, Anal. Bioanal. Chem. 401 (2011) 1071–1081. [11] S. Zhang, X. Yang, X. Yin, C. Wang, Z. Wang, Dispersive liquid–liquid microextraction combined with sweeping micellar electrokinetic chromatography for the determination of some neonicotinoid insecticides in cucumber samples, Food Chem. 133 (2012) 544–550. [12] M. Hernandez-Mesa, C. Cruces-Blanco, A.M. Garcia-Campana, Green methodology based on dispersive liquid–liquid microextraction and micellar electrokinetic chromatography for 5-nitroimidazole analysis in water samples, J. Sep. Sci. 36 (2013) 3050–3058. ˜ M. del Olmo-Iruela, Eval[13] M.D. Víctor-Ortega, F.J. Lara, A.M. García-Campana, uation of dispersive liquid–liquid microextraction for the determination of patulin in apple juices using micellar electrokinetic capillary chromatography, Food Control 31 (2013) 353–358. [14] Y. Zhang, B. Jiao, Dispersive liquid–liquid microextraction combined with online preconcentration MEKC for the determination of some phenoxyacetic acids in drinking water, J. Sep. Sci. 36 (2013) 3067–3074. [15] P. Soisungnoen, R. Burakham, S. Srijaranai, Determination of organophosphorus pesticides using dispersive liquid–liquid microextraction combined with reversed electrode polarity stacking mode—micellar electrokinetic chromatography, Talanta 98 (2012) 62–68. [16] A.V. Herrera-Herrera, J. Hernandez-Borges, T.M. Borges-Miquel, M.A. Rodriguez-Delgado, Dispersive liquid–liquid microextraction combined with nonaqueous capillary electrophoresis for the determination of fluoroquinolone antibiotics in waters, Electrophoresis 31 (2010) 3457–3465. [17] D. Airado-Rodriguez, C. Cruces-Blanco, A.M. Garcia-Campana, Dispersive liquid–liquid microextraction prior to field-amplified sample injection for the sensitive analysis of 3,4-methylenedioxymethamphetamine, phencyclidine and lysergic acid diethylamide by capillary electrophoresis in human urine, J. Chromatogr. A 1267 (2012) 189–197. [18] T.T. Liang, Z.H. Lv, T.F. Jiang, Y.H. Wang, High-density extraction solvent-based solvent de-emulsification dispersive liquid–liquid microextraction combined with MEKC for detection of chlorophenols in water samples, Electrophoresis 34 (2013) 345–352. [19] P. de Morais, T. Stoichev, M.C. Basto, M.T. Vasconcelos, Extraction and preconcentration techniques for chromatographic determination of chlorophenols in environmental and food samples, Talanta 89 (2012) 1–11. [20] U. Alshana, N.G. Goger, N. Ertas, Ultrasound-assisted emulsification microextraction for the determination of ephedrines in human urine by capillary electrophoresis with direct injection. Comparison with dispersive liquid–liquid microextraction, J. Sep. Sci. 35 (2012) 2114–2121. [21] S.Y. Hsieh, C.C. Wang, S.M. Wu, Microemulsion electrokinetic chromatography for analysis of phthalates in soft drinks, Food Chem. 141 (2013) 3486–3491. [22] F.E.P. Mikkers, F.M. Everaerts, T.P.E.M. Verheggen, Concentration distributions in free zone electrophoresis, J. Chromatogr. A 169 (1979) 1–10. [23] F.E.P. Mikkers, F.M. Everaerts, T.P.E.M. Verheggen, High-performance zone electrophoresis, J. Chromatogr. A 169 (1979) 11–20.

298

H. He et al. / J. Chromatogr. A 1361 (2014) 291–298

[24] R.L. Chien, D.S. Burgi, Sample stacking of an extremely large injection volume in high-performance capillary electrophoresis, Anal. Chem. 64 (1992) 1046–1050. [25] J.P. Quirino, S. Terabe, Sweeping of analyte zones in electrokinetic chromatography, Anal. Chem. 71 (1999) 1638–1644. [26] J.P. Quirino, S. Terabe, Exceeding 5000-fold concentration of dilute analytes in micellar electrokinetic chromatography, Science 282 (1998) 465–468. [27] J.P. Quirino, S. Terabe, On-line concentration of neutral analytes for micellar electrokinetic chromatography. 3. Stacking with reverse migrating micelles, Anal. Chem. 70 (1998) 149–157. [28] Z.K. Shihabi, Stacking in capillary zone electrophoresis, J. Chromatogr. A 902 (2000) 107–117. [29] Y. Kong, Studies on the mechanism of the acetonitrile–salt stacking method in capillary electrophoresis, J. Liq. Chromatogr. Relat. Technol. 29 (2006) 1561–1573. [30] M. Moradi, Y. Yamini, A. Esrafili, S. Seidi, Application of surfactant assisted dispersive liquid–liquid microextraction for sample preparation of chlorophenols in water samples, Talanta 82 (2010) 1864–1869. [31] J. Cao, S. Liu, W. Bai, J. Chen, Q. Xie, Determination of the migration of bisphenol A from polycarbonate by dispersive liquid–liquid microextraction combined with high performance liquid chromatography, Anal. Lett. 46 (2013) 1342–1354. [32] Z.K. Shihabi, Transient pseudo-isotachophoresis for sample concentration in capillary electrophoresis, Electrophoresis 23 (2002) 1612–1617. [33] M.R.N. Monton, J.P. Quirino, K. Otsuka, S. Terabe, Separation and online preconcentration by sweeping of charged analytes in electrokinetic

[34]

[35]

[36] [37] [38]

[39]

[40]

[41]

chromatography with nonionic micelles, J. Chromatogr. A 939 (2001) 99– 108. J. Cao, B. Li, Y.X. Chang, P. Li, Direct on-line analysis of neutral analytes by dual sweeping via complexation and organic solvent field enhancement in nonionic MEKC, Electrophoresis 30 (2009) 1372–1379. W.L. Tseng, H.T. Chang, On-line concentration and separation of proteins by capillary electrophoresis using polymer solutions, Anal. Chem. 72 (2000) 4805–4811. Z.K. Shihabi, Stacking of weakly cationic compounds by acetonitrile for capillary electrophoresis, J. Chromatogr. A 817 (1998) 25–30. Z.K. Shihabi, Organic solvent high-field amplified stacking for basic compounds in capillary electrophoresis, J. Chromatogr. A 1066 (2005) 205–210. V. Galli, C. Barbas, Optimization in sample stacking for the measurement of short chain organic acids in serum of natural rubber latex by capillary electrophoresis, Anal. Chim. Acta 482 (2003) 37–45. M.C. Breadmore, Ionic liquid-based liquid phase microextraction with direct injection for capillary electrophoresis, J. Chromatogr. A 1218 (2011) 1347–1352. Q. Wang, H. Qiu, J. Li, X. Liu, S. Jiang, On-line coupling of ionic liquid-based single-drop microextraction with capillary electrophoresis for sensitive detection of phenols, J. Chromatogr. A 1217 (2010) 5434–5439. K. Choi, Y.G. Jin, D.S. Chung, In-line coupling of two-phase single drop microextraction and large volume stacking using an electroosmotic flow pump in nonaqueous capillary electrophoresis, J. Chromatogr. A 1216 (2009) 6466–6470.