Dynamic Liquid−Liquid−Liquid Microextraction with Automated

Anal. Chem. 2005, 77, 1689-1695

Dynamic Liquid-Liquid-Liquid Microextraction with Automated Movement of the Acceptor Phase Xianmin Jiang, Sze Yin Oh, and Hian Kee Lee*

Department of Chemistry, National University of Singapore, 3 Science Drive 3, Republic of Singapore 117543

A new dynamic liquid-liquid-liquid microextraction procedure, with the automated movement of acceptor phase (LLLME/AMAP) to facilitate mass transfer, was developed in this study. Four compounds, 3-nitrophenol, 4-nitrophenol, 3,4-dinitrophenol, and 2,4-dichlorophenol, were used as model compounds to be preconcentrated from water samples. The extraction involved filling a 2-cm length of hollow fiber with 4 µL of acceptor solution using a conventional microsyringe, followed by impregnation of the pores of the fiber wall with 1-octanol. The fiber was then immersed in 4 mL of aqueous sample solution. The analytes in the sample solution were extracted into the organic solvent and then back-extracted into the acceptor solution. During extraction, the acceptor phase was repeatedly moved in and out of the hollow fiber channel and the syringe controlled by a syringe pump. Separation and quantitative analyses were then performed by using high-performance liquid chromatography. The results indicated that up to 400-fold enrichment of the analytes could be obtained under the optimized conditions. The enrichment factors were two times those of static liquidliquid-liquid microextraction. Good repeatabilities (RSD values below 9.30%) were obtained. The calibration linear range was from 10 to 1000 ng/mL with the square of the correlation coefficient (r2) >0.9916. Detection limits were in the range of 0.45-0.98 ng/mL. In addition, as compared with the previously reported dynamic three-phase microextraction in which there was no relative movement between the acceptor and the organic phase (which is not conducive to effective mass transfer), this new method shows much higher extraction efficiency. All these results suggest that this new dynamic LLLME/AMAP technique could be a better alternative to the previous LLLME for the extraction of analytes from aqueous samples. Conventional liquid-liquid extraction (LLE) has been widely used for sample pretreatment of organic compounds from aqueous samples.1-4 LLE is more suitable for the bulk extraction of large quantities of analytes and not for trace analysis. However, since * To whom correspondence should be addressed. E-mail: [email protected]. (1) Kamali F.; Herb B. J. Chromatogr., A 1990, 530, 222-225. (2) Durand, G.; Barcelo´, D. Anal. Chim. Acta 1991, 243, 259-271. (3) Harrison, I.; Leader, R. U.; Higgo, J. J. W.; Tjell, J. C. J. Chromatogr., A 1994, 688, 181-188. (4) Prosen, H.; Zupancic-Kralj, L. Trends Anal. Chem. 1999, 18, 272-282. 10.1021/ac040153v CCC: $30.25 Published on Web 02/15/2005

© 2005 American Chemical Society

relatively large amounts of solvents are used, the toxicity and environmental consequences of solvent disposal must be considered. In addition, when dirty aqueous samples are involved, the matrix effect has to be taken into account. In recent years, research activities have been oriented toward the development of efficient and miniaturized sample pretreatment methods. Solvent microextraction has been shown to be an alternative method to conventional LLE. Its main applications lie in the trace analysis of pollutants in the environment5-8 and drugs in biological fluids.9-11 In 1996, Cantwell and Jeannot12 developed a solvent microextraction method in which analytes in a 1-mL aqueous sample were extracted into an 8-µL organic solvent drop. This greatly reduced the amount of solvent used during the extraction. Due to the large volume ratio of the aqueous phase to the organic phase, a large enrichment factor was obtained. The disadvantage of this method was that extraction and injection were performed separately in two different instruments. Subsequently, He and Lee13 reported a method in which a conventional microsyringe was used for both extraction and injection. This technique was termed liquid-phase microextraction (LPME). Furthermore, two modes of LPME, static and dynamic, were developed. The difference was that the organic phase was constantly being withdrawn and discharged during dynamic LPME. Results indicated that dynamic LPME provided higher enrichment in a much shorter extraction time than the static mode. Due to the instability of the organic solvent drop, the use of a porous hollow fiber made of polypropylene was introduced to protect the solvent drop.8,9 In addition, hollow fiber membranes also offered a much larger surface area per volume, hence providing greater extraction efficiency. In 1999, Pedersen-Bjergaard and Rasmussen9 developed the concept of liquid-liquid-liquid microextraction (LLLME), which was used to extract ionizable and chargeable compounds from aqueous samples. In this method, the hollow fiber membrane was used to separate the three liquid phases and provide a high degree (5) Wang, Y.; Kwok, Y. C.; He, Y.; Lee, H. K. Anal. Chem. 1998, 70, 46104614. (6) Zhu, L. Y.; Zhu, L.; Lee, H. K. J. Chromatogr., A 2001, 924, 407-414. (7) Zhao, L.; Zhu, L.; Lee, H. K. J. Chromatogr., A 2002, 963, 239-248. (8) Shen, G.; Lee, H. K. Anal. Chem. 2002, 74, 648-654. (9) Pedersen-Bjergaard, S.; Rasmussen, K. E. Anal. Chem. 1999, 71, 26502656. (10) Ugland, H. G.; Krogh, M.; Rasmussen, K. E. J. Chromatogr., B 2000, 749, 85-92. (11) Pedersen-Bjergaard, S.; Rasmussen. K. E. Electrophoresis 2000, 21, 579585. (12) Cantwell, F. F.; Jeannot, M. A. Anal. Chem. 1996, 68, 2236-2240. (13) He, Y.; Lee, H. K. Anal. Chem. 1997, 69, 4634-4640.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005 1689

of sample cleanup since large molecules and particles were not extracted into the acceptor phase. The three phases involved were aqueous sample solution (donor solution), a very small volume of organic solvent immobilized in the pores of the hollow fiber, and a small volume of aqueous acceptor solution inside the hollow fiber (acceptor solution). More recently, Hou and Lee15 introduced the use of a syringe pump to enable repeated movement of the syringe plunger. In that technique, termed as dynamic three-phase microextraction, 5 µL of the aqueous acceptor solution was withdrawn in the syringe, followed by 2 µL of organic solvent that was confined in the hollow fiber channel. The pores of the fiber were preimpregnated with the organic solvent. The acceptor phase remained resident inside the syringe. During the extraction, the plunger was retracted, and 2 µL of aqueous donor solution was withdrawn into the fiber channel. After a 5-s dwell time, the plunger was depressed to refill the fiber with 2 µL of organic solvent and simultaneously expel the aqueous sample. The same cycles were repeated for a prescribed time. However, there are some limitations in this method. These limitations are that there is a very small contact area between acceptor phase and organic phase, there is no relative movement between acceptor phase and organic phase, which is not beneficial to mass transfer, and most of the acceptor phase is confined in the barrel of the microsyringe. To obviate the shortcomings mentioned above, an attempt should be made to increase greatly the interfacial area of contact between the organic and acceptor phases. In addition, relative movement between organic phase and acceptor phase to promote more efficient mass transfer should be effected. In the present work, we studied a new approach to LLLMEs that is, dynamic liquid-liquid-liquid microextraction with the automated movement of acceptor phase (LLLME/AMAP). The aim was to develop a better alternative to static LLLME and further improve high extraction efficiency. This new dynamic LLLME/ AMAP technique involved the automated movement of the acceptor phase in the hollow fiber using a syringe pump. The new technique was optimized with respect to the type of organic solvent used for immobilization in hollow fiber wall pores, the compositions of the donor and acceptor solutions, stirring speed, extraction time, and movement pattern of the syringe plunger. Nitrophenols were chosen as model compounds. They are widely present as water pollutants in the environment as a result of industrial discharges. EXPERIMENTAL SECTION Materials and Reagents. The Q3/2 Accurel polypropylene hollow fiber was purchased from Membrana (Wuppertal, Germany). The inner diameter of the fiber was 600 µm, the thickness of the wall was 200 µm, and the pore size was 0.2 µm. HPLCgrade acetonitrile and methanol were bought from J.T. Baker (Philipsburg, NJ). Sodium acetate, 1-octanol, n-octane, isooctane, n-hexane, toluene, and butyl acetate were bought from Merck (Darmstadt, Germany). 3-Nitrophenol (3-NP), 4-nitrophenol (4NP), 3,4-dinitrophenol (3,4-DNP), and 2,4-dichlorophenol (2,4DCP) were bought from Sigma-Aldrich (Steinheim, Germany). Each analyte was dissolved in HPLC-grade methanol to obtain a (14) Cussler. E. L. Diffusion: Mass Transfer in Fluid Systems, 2nd ed.; Cambridge University Press: Cambridge, U.K., 1997; Chapter 16. (15) Hou, L.; Lee, H. K. Anal. Chem. 2003, 75, 2784-2789.

1690 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

Figure 1. Schematic diagram of the dynamic LLLME/AMAP setup (not drawn to scale).

stock solution with a concentration of 1 mg/mL. These solutions were stored at 4 °C. Working solutions were prepared by dilution of stock standards with water purified by a Milli-Q ((Millipore, Bedford, MA) ultrapure water purification system. Instrument and Apparatus. The HPLC system consisted of a Hewlett-Packard (Palo Alto, CA) Series 1050 pump, an injector equipped with a 10-µL sample loop, a series 1050 UV detector, and LC1050 ChemStation software. A Spherisorb (PhaseSep, Norwalk, CT) ODS-2 200 mm × 4.6 mm i.d. column was used for separations. The mobile phase was acetonitrile-water (40:60) with 75 mM sodium acetate-glacial acetic acid (pH 3.50). A Harvard Apparatus (Holliston, MA) PHD 2000 programmable syringe pump was used to withdraw and discharge the acceptor solution into the hollow fiber at a prescribed rate. Extraction Procedure. The experimental setup is illustrated in Figure 1. The hollow fiber was cut into 2-cm segments. Each segment was flame-sealed at one end and had an internal volume of ∼4.4 µL. Prior to use, the segments were sonicated in HPLCgrade acetone for 10 min to remove any contaminants in the fiber. They were removed from the acetone and dried in the air. Each piece of porous hollow fiber was used for a single extraction only. The sample solution (donor phase) was adjusted to be acidic by HCl. The acceptor phase was adjusted to be basic by NaOH to ionize the analytes. A 4-mL aliquot of the sample solution was placed in a vial, which was placed on a magnetic stirrer plate (Heidolph, Kelheim, Germany). A 12 × 4 mm magnetic stirring bar was placed in the donor solution to ensure efficient stirring during the extraction. A conventional 10-µL HPLC syringe (SGE Scientific, Sydney, Australia) was used. A 4-µL aliquot of acceptor solution was introduced into a hollow fiber segment with a sealed end. The fiber was immersed in the organic solvent for 5 s to impregnate the pores of the fiber with the solvent. Then the fiber, which was still attached to the syringe, was placed into the donor solution.

The magnetic stirrer and the syringe pump were switched on to start the extraction. A 3-µL aliquot of acceptor solution was withdrawn into the syringe at a prescribed speed, followed by a pause for few seconds (dwell time) (as prescribed), then ejected from the syringe to the hollow fiber channel at the same speed, and followed by another pause for a few seconds (as prescribed), and the entire cycle repeated. After a prescribed extraction time, the assembly was removed from the donor solution, and the acceptor solution (1.5 µL) in the fiber was withdrawn back into the syringe. The hollow fiber was discarded at this juncture, and 2.5 µL of pure water was withdrawn into the syringe to combine with the acceptor solution so as to dilute the NaOH to minimize or eliminate the damage to the HPLC column. The entire solution in the syringe was then injected into the HPLC system for analysis. RESULTS AND DISCUSSION Basic Mechanism. The extraction process in LLLME involves three phases, namely, the aqueous donor phase, the organic phase, and the acceptor phase. This can be represented by the following equation. K1

K2

ia1 798 io 798 ia2

(1)

where i is the analyte, a1 is the aqueous donor phase, o is the organic phase immobilized in the pores of the hollow fiber, and a2 is the aqueous acceptor phase. In the first step of eq 1, the analyte i in a1 distributes between a1 and o. The distribution coefficient is given by

K1 ) Co,eq/Ca1,eq

(2)

In the second step, i distributes between o and a2. The distribution coefficient is expressed by

K2 ) Ca2,eq/Co,eq

(3)

where Ca1,eq, Co,eq, and Ca2,eq are the concentrations of i in a1, o, and a2, respectively, at equilibrium. The enrichment factor16 (EF) at a given time during the extraction can be determined as the ratio of the concentration of i in the aqueous acceptor phase at that time, to its initial concentration (Ca1,initial):

EF ) Ca2/Ca1,initial

(4)

Agitation of the aqueous donor solution (a1) is usually performed to enhance the extraction efficiency. According to the film theory,14,17 efficient stirring increases the mass-transfer coefficient in the aqueous phase (βa1) by decreasing the thickness (16) Ma, M.; Cantwell, F. F. Anal. Chem. 1999, 71, 388-393. (17) Cussler. E. L. Diffusion: Mass Transfer in Fluid Systems, 2nd Ed.; Cambridge University Press: Cambridge, U.K., 1997; Chapter 13.

(δa1) of the Nernst diffusion film. The relationship between βa1 and δa1 is given by18

βa1 ) Da1/δa1

(5)

This results in a higher rate of mass transfer into the organic phase by agitation. In addition, agitation can ensure that the aqueous sample in contact with the organic phase is constantly renewed. In this study, we invoke the film theory to explain the improved extraction efficiency in dynamic LLLME/AMAP. The film theory assumes that a stagnant film exists near every interface, and mass transfer across this film occurs via diffusion only.14 Therefore, we can assume that a very thin aqueous film (AF) of the acceptor solution is present between organic phase and acceptor phase. Based on the above theoretical consideration, repeated movement (withdrawal and discharge) of the acceptor phase is implemented to facilitate mass transfer into a2. The extraction process between o and a2 could be assumed in the following steps: (1) When the aqueous plug (AP) is withdrawn, a very thin AF is left on the inner surface of the fiber. (2) The analyte is transferred rapidly from the organic phase into the AF. (As AF has a very small volume, CAF and hence (CAF - Ca2) is very large. This results in a large value of N.) (3) The AP is completely discharged into the fiber. (4) The analyte is transferred rapidly from the AF into the AP. (AF works as an “analyte transporter” by which the analyte is transferred from the organic phase into the AP.) In static LLLME, the film between AP and organic phase is relatively stagnant. However, in LLLME/AMAP, we can suppose that the thickness of the film between two phases becomes thinner when the plunger is withdrawn, and there is a pause in movement for some time. According to film theory, the flux across the film can be calculated in terms of the diffusion coefficient (D) and diffusion thickness (L):14

N ) (D/L)(CAF - Ca2)

(6)

From the above equation, it can be seen that amount of the mass transfer per unit area is inversely proportional to the diffusion thickness. When the plunger moves down, the analytes in the film diffuse into the whole acceptor phase immediately. Thus, concentration of the analytes in the film decreased very rapidly. When the plunger is withdrawn again, the analytes in the organic phase diffuse into the thin film yet again and the cycle is repeated. An expanded view of the dynamic movement is shown in Figure 2. With the successive extractions using a very small volume of acceptor solution, the extraction efficiency can be enhanced significantly. In addition, for the method developed by Hou and Lee,15 the contact area between the acceptor phase and organic solvent was estimated to be 0.18 mm2 (syringe barrel cross-sectional area). In our method, as shown in Figure 2, the interfacial area of contact between the organic phase and the aqueous acceptor phase was maximized and can be calculated as S )πr2 + 2πrL, where r is the internal radius of the hollow fiber and L is the effective length (18) Cantwell, F. F.; Jeannot, M. A. Anal. Chem. 1997, 69, 235-239.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1691

Figure 2. Expanded view of dynamic LLLME with automated movement of the acceptor phase in the hollow fiber. (a) At the start of the sampling cycle, the AP resided in the hollow fiber. (b) The AP was withdrawn into the syringe. Analytes were extracted from the organic phase into AF. (c) The AP was discharged into the fiber. Analytes were transferred into from AF into AP.

of the fiber. Using the equation, the interfacial area is estimated to be 28.54 mm2. The great increase of contact area is beneficial for mass transfer between the organic solvent and acceptor phase. Selection of Organic Solvent. The type of solvent immobilized within the pores of the hollow fiber is of importance because it determines the efficiency of analyte preconcentration. The following factors should be considered: First, the solvent should have a low solubility in water to avoid dissolution into the aqueous phase, low volatility to prevent solvent loss due to evaporation, and effective immobilization in the pores of the fiber to ensure a functional intermediate phase. Second, the solubility of an analyte in the organic solvent should also be higher than that in the donor phase and lower than that in the acceptor phase.5 This promotes analyte migration from the donor phase through the pores of the hollow fiber and finally to the acceptor phase. Finally, the polarity of the organic solvent should match that of the polypropylene fiber, so that it can be easily immobilized within the pores of the fiber. Six organic solvents, namely, 1-octanol, toluene, hexane, n-octane, isooctane, and butyl acetate, were evaluated. Each 1692

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

extraction was performed for 20 min, using standard solutions containing 200 ng/mL aliquots of the four phenols dissolved in 0.1 M HCl. Toluene could only extract 3,4-dinitrophenol. Hexane, n-octane, and isooctane could only extract 2,4-dichlorophenol. Butyl acetate could only extract 4-nitrophenol and 2,4-dichlorophenol. Only 1-octanol could extract all four compounds, and the HPLC signals obtained were relatively higher than those of any other organic solvent studied. This could be attributed to its relatively higher polarity and greater affinity to the phenols. In addition, it also has low volatility, very poor solubility in water, and effectively immobilized in the membrane pores. Thus, 1-octanol was selected as the most suitable solvent. Compositions of the Donor and Acceptor Phases. In LLLME, the adjustment of the pH value of the donor and acceptor phases is another key to achieve high distribution ratio and enrichment factor. As the target compounds are acidic, the pH of the donor solution was adjusted in the proper acidic range to neutralize the compounds and reduce their solubility in the sample solution. To ensure efficient analyte transfer into the acceptor phase, the pH of the acceptor solution was adjusted to be in the

Figure 3. Effect of extraction time on extraction efficiency of dynamic LLLME/AMAP.

Figure 4. Effect of plunger speed on extraction efficiency of dynamic LLME/AMAP.

basic range. This leads to the ionization of the analytes and prevent them from re-entering the organic phase. With 1-octanol being immobilized in the pores of the fiber, a series of extractions were performed for 20 min each with stirring at 700 rpm, with HCl in the donor phase and NaOH in the acceptor phase. The pH of the donor phase was studied in the range of 0-3. When the pH was reduced from 3 to 1, the peak areas of all the analytes increased. This change could be explained by their pKa values. The pKa values of 4-NP, 3-NP, 3,4-DNP, and 2,4-DCP are 7.2, 8.4, 3.5, and 7.8, respectively. For almost complete deionization of an acidic analyte, the pH of the donor solution should be lower than its pKa by at least 2 units. As 3,4-DNP has the lowest pKa of 3.5, the pH of the donor solution should be 1.5 or below. Hence, the pH value at 1 gave the highest HPLC signals for all the analytes and was chosen as the optimum pH. It was also observed that, at pH 0, the peak areas were dramatically decreased. Thus, acidity that is too high is undesirable, as protonation of the analytes can occur and mass transfer into the organic phase is hindered. Sodium chloride from 10 to 40% (w/ v) was added to the donor solution to investigate the possibility of the salting-out effect from the donor phase, which may enhance extraction efficiency. No enhancement in enrichment factors was observed. For the acceptor phase, the pH was studied in the range of 11-14. The peak areas of all the analytes increased with the pH from 11 to 13. However, if the pH value of the acceptor phase is too high (pH >14), the HPLC column may be damaged. Thus, pH 13 was chosen as optimum as it gave the highest HPLC signals for all the analytes. Extraction Time. The effect of extraction time on the amount of analytes extracted was investigated. Each extraction was performed on a standard solution containing 200 ng/mL nitrophenols dissolved in 0.1 M HCl. The acceptor solution was 0.1 M NaOH, and the organic phase was 1-octanol. Since the extraction is not an exhaustive process, the main objective is to achieve sufficiently high extraction efficiency within a relatively short time. Furthermore, if the extraction time is too long, solvent loss and formation of air bubbles may occur, which would comprise the extraction. As shown in Figure 3, the amount of analytes extracted was found to increase with extraction time. The results indicated that the HPLC signals obtained for the analytes were sufficiently high after 20 min. In addition, it is not normally considered practicable to wait until equilibrium was established (>60 min).

Thus, 20 min was chosen as the extraction time for subsequent experiments. Agitation. Agitation was employed to facilitate the masstransfer process and extraction efficiency. In this study, the agitation speed was optimized for the extraction. The experimental results supported this explanation. The responses were found to increase with the stirring speed from 200 to 700 rpm. However, at 1000 rpm, the agitation became too vigorous and caused the formation of air bubbles that tended to adhere to the wall of the fiber, causing solvent evaporation and resulting in poor extraction efficiency and reproducibility. On the basis of these observations, 700 rpm was selected for subsequent experiments. Plunger Speed and Dwell Time. During the extraction, the syringe plunger was automatically manipulated up and down in the microsyringe barrel. Each sampling cycle consisted of four steps: withdrawal of the acceptor phase, pause (dwell time), discharge of the acceptor solution, and another pause. The effects of the plunger movement speed (sampling volume/withdrawal time) and the dwell time on extraction efficiency were studied. The plunger speed was studied in the range of 0.1-0.5 µL/s. From the results shown in Figure 4, it is observed that the amount of analytes extracted was enhanced with the increase of plunger speed. This is due to the fact that when a higher plunger speed was used, more samplings could be performed in a given time. Based on the observations, 0.5 µL/s, the maximum that could be set for the pump, was chosen as the optimum speed as the highest HPLC signals were obtained. To investigate the effect of dwell time on extraction efficiency, the plunger speed was kept at 0.5 µL/s and the dwell time was varied from 1 to 5 s. As shown in Figure 5, the results indicated that influence of dwell time on the amount of analytes extracted was found to be insignificant. This was due to the very thin aqueous acceptor film left on the inner wall of the fiber after each withdrawal movement of the syringe plunger. Hence, equilibrium was achieved very quickly between the organic phase and aqueous acceptor film. For practical reasons, 5 s was chosen as the dwell time for the rest of the study. Comparison with Static Liquid-Liquid-Liquid Microextraction. The enrichment factors, obtained by the new dynamic LLLME/AMAP technique using the optimum conditions, were compared to those of static LLLME. Static LLLME was performed under the same conditions: 1-octanol as the organic solvent, 0.1 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1693

Figure 5. Effect of dwell time on extraction efficiency of dynamic LLLME/AMAP. Table 1. Enrichment Factors of the Dynamic LLLME/ AMAP and Static LLLME enrichment factor analyte

static LLLME

dynamic LLLME/AMAP

4-NP 3-NP 3,4-DNP 2,4-DCP

93 97 112 227

194 190 201 347

improvement in extraction efficiency 2× 2× 2× 1.5×

M HCl donor solution, 0.1 M NaOH acceptor solution, a stirring speed of 700 rpm, and an extraction time of 20 min. Similarly, 1.5 µL of the acceptor solution with another 2.5 µL of pure water was injected into the HPLC system after each extraction. As shown in Table 1, the enrichment factors of static LLLME were from 93 to 227 while for the new dynamic mode the enrichment factors were from 194 to 347. In conclusion, the enrichment factor of the new dynamic technique was capable of achieving up to 2 times as high as that of static LLLME. Comparison between LLLME/AMAP and Dynamic ThreePhase Microextraction. Dynamic LLLME as developed by Hou and Lee15 was carried out in order to compare it with the present LLLME/AMAP. As shown in Figure 6, the new method LLLME/ AMAP had much higher enrichment factors than the latter. There are several ways to explain the results. First, as compared to dynamic three-phase microextraction, the LLLME/AMAP increased the interfacial area of contact between the acceptor phase and organic phase. In the previous dynamic three-phase microextraction, the interfacial area of contact between the organic solvent and the aqueous acceptor solution was very small. The area (S) can be calculated as S ) πr2, where r is the internal radius of the syringe barrel. Using this equation, the interfacial area was estimated to be 0.18 mm2 (syringe barrel cross-sectional area) while for LLLME/AMAP the area was much larger (28.54 mm2), in which the contact area can be calculated as S ) πr2 + 2πrL, where L is the effective length of the hollow fiber. This suggests that the mass transfer of the analytes from the organic phase into the acceptor phase could be compromised. Second, strictly speaking, for dynamic three-phase microextraction, there was no relative movement between the acceptor phase and the organic phase. Third, for dynamic three-phase microextraction, the acceptor phase was confined in the barrel of the microsyringe while for LLME/AMAP there is greater volume in the syringe barrel 1694 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

Figure 6. Comparison of extraction efficiency between LLLME/ AMAP and dynamic three-phase microextraction. Table 2. Performance of Dynamic LLLME/AMAP analyte

enrichment factor

linearity range (ng/mL)

RSD (%) (n ) 5)

LOD (ng/mL)

4-NP 3-NP 3,4-DNP 2,4-DCP

194 190 201 347

10-1000 10-1000 10-1000 10-1000

3.23 9.25 3.46 8.01

0.88 0.45 0.65 0.98

and channel of the hollow fiber for movement. The extraction efficiency of LLLME/AMAP is therefore much higher than that of dynamic three-phase microextraction. Method Evaluation. To evaluate the practical applicability of the new technique, its calibration linearity range, repeatability, and limits of detection were investigated. The performance of this technique is shown in Table 2. Enrichment factors of up to 347fold were achieved for the aqueous samples. The RSD was less than 9.30% for five replicate experimental results. All of the analytes exhibited good linearity over a range of 10-1000 ng/mL with the square of the correlation coefficient (r2) >0.9916. Good detection limits in the range of 0.45-0.98 ng/mL were also obtained, based on a signal-to-noise ratio of 3 (S/N ) 3). This new dynamic technique gave good repeatability, due to the automated movement of the syringe plunger by the syringe pump that provided the accuracy of control of the plunger movement and dwell time. Furthermore, the formation of air bubbles in the acceptor solution was reduced and the recovery of the acceptor solution after each extraction was also enhanced. CONCLUSION In the present study, a new method termed the dynamic liquid-liquid-liquid microextraction technique, with automated movement of the acceptor phase, was developed. As compared with static LLLME, this dynamic method could achieve much higher enrichment factors (up to 400-fold) under optimum condi-

tions. The procedure only needed several microliters of organic solvent and acceptor phase (4 µL) and 20-min extraction time. The newly developed procedure is a one-step microextraction technique that is fast, efficient, simple to operate, and requires a minimum amount of solvent. Moreover, with the automation, good precision and high enrichment were achieved. However, there is one limitation relating to the operation of the pump; the maximum speed of the syringe plunger was limited by the type of pump used. Therefore, the number of samplings over a given time could not be further increased to attain potentially higher enrichment of the analytes. Further studies will

involve the use of a pump that will have a higher plunger speed rating. ACKNOWLEDGMENT The authors gratefully acknowledge the financial support of this work by the National University of Singapore. X.J. is grateful to the university for the award of a research scholarship. Received for review August 25, 2004. Accepted December 8, 2004. AC040153V

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1695