Anal. Chem. 1999, 71, 2650-2656
Liquid-Liquid-Liquid Microextraction for Sample Preparation of Biological Fluids Prior to Capillary Electrophoresis Stig Pedersen-Bjergaard* and Knut Einar Rasmussen
School of Pharmacy, University of Oslo, P.O. Box 1068 Blindern, 0316 Oslo, Norway
Methamphetamine as a model compound was extracted from 2.5-mL aqueous samples adjusted to pH 13 (donor solution) through a thin phase of 1-octanol inside the pores of a polypropylene hollow fiber and finally into a 25-µL acidic acceptor solution inside the hollow fiber. Following this liquid-liquid-liquid microextraction (LLLME), the acceptor solutions were analyzed by capillary zone electrophoresis (CE). Extractions were performed in simple disposable devices each consisting of a conventional 4-mL sample vial, two needles for introduction and collection of the acceptor solution, and a 8-cm piece of a porous polypropylene hollow fiber. From 5 to 20 different samples were extracted in parallel for 45 min, providing a high sample capacity. Methamphetamine was preconcentrated by a factor of 75 from aqueous standard solutions, human urine, and human plasma utilizing 10-1 M HCl as the acceptor phase and 10-1 M NaOH in the donor solution. In addition to preconcentration, LLLME also served as a technique for sample cleanup since large molecules, acidic compounds, and neutral components were not extracted into the acceptor phase. Utilizing diphenhydramine hydrochloride as internal standard, repetitive extractions varied less than 5.2% RSD (n ) 6), while the calibration curve for methamphetamine was linear within the range 20 ng/µL to 10 µg/mL (r ) 0.9983). The detection limit of methamphetamine utilizing LLLME/CE was 5 ng/mL (S/N ) 3) in both human urine and plasma.
During the last 10 years, capillary electrophoresis (CE) has been developed into a highly attractive separation technique for both ionic and neutral compounds. Owing to high separation efficiencies, low separation times, a low consumption of reagents, and relatively simple method development, CE has been implemented for research and routine analysis of pharmaceuticals, peptides, proteins, agrochemicals, raw materials, water, and DNA as well as for many other applications.1 However, because most CE is performed with UV detection directly on the narrow fusedsilica capillaries used for the separation and because only nanoliter volumes of sample are injected in traditional CE, most concentra(1) Altria, K. D.; Bryant, S. M. LC-GC 1997, 15, 448.
2650 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
tion detection limits obtained in CE are relatively high. Therefore, CE is principally used for compounds present at relatively high concentration levels, while the advantages of the technique are difficult to utilize for trace analysis applications in connection with biological and environmental samples. Several attempts to improve concentration detection limits in CE have been published. A number of workers have developed advanced injection techniques including analyte stacking, field amplification, and transient isotachophoresis that facilitate the analysis of larger sample volumes.2-6 All of these are based on voltage applied across the CE capillary, resulting in stacking or focusing due to variation of ion mobilities in various field strengths or chemical microenvironments. As a result, relatively large sample volumes can be analyzed with minimal loss of analyte resolution or separation efficiency. However, since the advanced injection techniques are carried out within conventional CE capillaries, injection volumes are normally limited to 99%) were purchased from Norsk Medisinaldepot (Oslo, Norway). 1-octanol (>99.5%) and 2-octanone (>97%) were obtained from Fluka (Buchs, Switzerland), hexyl ether (purity not reported) was purchased from Sigma (St. Louis, MO), and isooctane (>99.5%) was obtained from Merck. Standard Solutions and Biological Samples. Standard solutions of methamphetamine was prepared in different concentrations of NaOH by dilution from a 0.2 mg/mL stock solution of methamphetamine in water. For the validation studies, diphenhydramine hydrochloride was added as an internal standard from a 0.2 mg/mL stock solution in water. The stock solutions of methamphetamine and diphenhydramine hydrochloride were stored at +5 °C and protected from light. Prior to LLLME, samples of human urine and plasma were prepared by the following procedure: 2.5 mL of sample was spiked with 3 µL of the 0.2 mg/mL stock solution of methamphetamine and subsequently 125 µL of 2 M NaOH was added. RESULTS AND DISCUSSION Basic Principle. Metamphetamine was selected as a model compound (analyte) in the present work and was extracted from aqueous standard solutions, human urine, and human plasma. The basic principle of LLLME is illustrated in Figure 2. Prior to extraction, the sample solution was made strongly alkaline in order to deionize the analyte and consequently to reduce the solubility within the sample solution (donor solution). A small piece of a porous hollow fiber of polypropylene was treated with an organic solvent (typically 1-octanol) immiscible with water and filled with a small volume of diluted hydrochloric acid (acceptor solution). The exposure to the organic solvent served to fill the pores of the hollow fiber with the organic solvent. Subsequently, the hollow fiber was placed within the sample solution, resulting in a threephase system that consisted of a large volume of an alkaline 2652 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
sample solution (aqueous donor solution), a very small volume of organic solvent immobilized in the pores of the hollow fiber, and a small volume of diluted hydrochloric acid inside the hollow fiber (aqueous acceptor solution). Because the organic solvent in the pores of the hollow fiber was immiscible with water, it served as an effective barrier between the donor phase and the acceptor phase. Owing to the low solubility of the analyte within the alkaline donor phase and because of the correspondingly high solubility in the acidic acceptor solution, the analyte was extracted from the donor solution through the organic solvent in the pores of the hollow fiber and into the acceptor solution. Because the analyte molecules were ionized within the acceptor solution, they were prevented from re-entering the organic solvent in the pores of the hollow fiber. Since the volume of acceptor solution inside the hollow fiber was very small compared to the volume of the donor solution (sample), the analyte was preconcentrated within the acceptor solution with time. In addition to analyte enrichment, LLLME served as a method for sample cleanup. Large molecules (like proteins) were unsoluble in the immobilized organic solvent and remained in the donor solution (sample). Furthermore, anionic and neutral sample constituents were not extracted into the acidic acceptor phase; acidic compounds were ionized and were unable to penetrate the organic solvent in the pores of the hollow fiber, polar neutral components were also unsoluble in the organic phase, and hydrophobic neutral compounds principally were distributed within the organic phase and the donor solution. Equilibrium Considerations and Enrichment Factor. As discussed above, the LLLME process involves a series of two reversible extractions. In the first step, the analyte is extracted from the sample solution (donor solution) and into the organic phase immobilized within the pores of the hollow fiber. In the second step, the analyte is back-extracted into the aqueous acceptor phase inside the hollow fiber. For an analyte i, the extraction process may be illustrated with the equation
ia1 T io T ia2
(1)
where the subscript a1 represents the aqueous donor phase (sample solution), o the organic phase within the pores of the hollow fiber, and a2 the aqueous acceptor phase. At equilibrium, the distribution ratios for the analyte i in the three-phase system are
K1 ) Co,eq/Ca1,eq
(2)
K2 ) Co,eq/Ca2,eq
(3)
and
where Co,eq is the equilibrium concentration of i in the organic phase, Ca1,eq is the equilibrium concentration of i in the donor phase, and Ca2,eq is the equilibrium concentration of i in the acceptor phase. At equilibrium, the mass-balance relationship
Table 1. Enrichment Factor (Ee) as a Function of the Donor/Acceptor Volume Ratio (Va1/Va2) and the Acceptor/Donor Partition Coefficient K phase ratio (Va1/Va2)
K ) 10
1000 100 40 20 10 1
9.9 9 8 7 5 0.91
enrichment factor (Ee) K ) 100 K ) 1000 91 50 29 17 9.1 0.99
500 91 39 19 9.9 1
K)∞ 1000 100 40 20 10 1
for i is given by20,21
Ca1,initial ) (K2Ca2,eq)/K1 + (K2Ca2,eqVo)/Va1 + (Ca2,eqVa2)/Va1 (4) where Ca1,initial is the initial concentration of i in the donor phase (sample), Va1 is the volume of donor solution (sample), Vo is the volume of organic solvent in the pores of the hollow fiber, and Va2 is the volume of acceptor solution inside the hollow fiber. The enrichment factor (Ee), defined as the ratio Ca2,eq/Ca1,initial, may be calculated by rearrangement of eq 4 to
Ee ) 1/[K2/K1 + (K2Vo)/Va1 + Va2/Va1]
(5)
In LLLME, the volume of organic solvent immobilized in the pores of the hollow fiber (Vo) is small, and eq 5 may be simplified to
Ee ) 1/[1/K + Va2/Va1]
(6)
K ) Ca2,eq/Ca1,eq
(7)
where
Enrichment factors based on eq 6 were calculated for different K values and volume ratios between the donor and acceptor phase (Table 1). On the basis of these calculations, K values of 100 or more are required for the analytes of interest in order to obtain high enrichment factors (>50). In addition, the donor/acceptor volume ratio should not be below 100 in order to obtain high enrichments and to ensure analyte preconcentrations attractive for practical work with biological samples. This information was of high importance during construction of the technical setup for LLLME. Technical Setup. The porous hollow fiber used was commercially available and made of polypropylene. A polypropylene fiber was selected owing to the excellent compatibility with a broad range of organic solvents. With this material, no degradation of the hollow fiber was observed following impregnation. The inner diameter of the hollow fiber was 600 µm. This was appropriate for the microliter volumes of acceptor solution utilized in the present work (discussed below). The wall of the hollow fiber was (20) Ma, M.; Cantwell, F. F. Anal. Chem. 1998, 70, 3912. (21) Pawliszyn, J. Solid-Phase Microextraction: Theory and Practice; Wiley-VCH: New York, 1997.
relatively thick (200 µm), which simplified the preparation of extraction units; the mechanical stability of the hollow fibers was excellent and the fibers were easily connected to the two syringe needles of the simple extraction unit prior to LLLME. A pore size of 0.2 µm was selected to ensure efficient penetration of small molecules such as methamphetamine while blocking for larger matrix components (like proteins in plasma samples). Following extraction, the extracts (acceptor solutions) were collected in small inserts placed in the vials for the CE instrument. For the instrumentation used in the present work (MDQ from Beckman), 200-µL inserts were selected among the commercially available alternatives providing the lowest sample volume. From this type of insert, 25 µL was found to be the minimum volume providing repeatable injections, and consequently, 25 µL was selected as the acceptor volume. The volume of acceptor solution was measured exactly and introduced into the hollow fiber by a microliter syringe. With a 25-µL acceptor volume and a desire to achieve enrichment factors in the range 50-100, LLLME was performed from sample volumes of 2.5 mL throughout the work (donor/acceptor volume ratio of 100; see Table 1). Commercially available 4-mL vials with screw caps were utilized as extraction units. Two medical needles were penetrated through the screw cap and were connected to the ends of the porous hollow fiber; one served to guide the syringe for injection of the acceptor solution while the other was utilized for collection of the acceptor solution after extraction. In addition to the extraction unit, a microliter syringe was required to introduce the acceptor solution as described above, and an inexpensive 2.5-mL syringe filled with air was utilized for collection of the acceptor solution into the CE vials. This simple setup combined with the low price of the porous hollow fiber resulted in very low costs for each extraction unit; each in-vial extraction unit was a disposable device utilized only for a single extraction. This effectively prevented carry-over effects during large series of extractions. Impregnation. The type of solvent immobilized within the pores of the hollow fiber was of high importance in order to achieve efficient analyte preconcentration by LLLME within reasonable time. The solubility of the analyte in the immobilized solvent should be higher than the solubility in the donor phase in order to promote analyte migration through the pores of the hollow fiber. In addition, the solvent should be immiscible with water to avoid dissolution during the extraction, nonvolatile to prevent evaporation during sample preparation, and effectively immobilized on surfaces of polypropylene to ensure a high quality of the intermediate phase. On the basis of these criteria, 1-octanol, 2-octanone, hexyl ether, and isooctane were evaluated as solvents for immobilization. Enrichment was accomplished by 45 min of LLLME from standard solutions of 100 ng/mL methamphetamine dissolved in 0.1 M NaOH. With 1-octanol, 2-octanone, and hexyl ether, methamphetamine was enriched by a factor of 62-70 (Table 2). These solvents were easily immobilized on the hollow fiber by a 5-s exposure, the volatility was low, and the solvents remained effectively immobilized during the extraction (no leakage to the sample). With isooctane in contrast, the highly nonpolar nature complicated immobilization within the porous polypropylene hollow fiber. On the basis of the immobilization experiments, 1-octanol was selected for the rest of the work. Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
2653
Table 2. Preconcentration of Methamphetamine Utilizing Different Solvents for Impregnation of the Porous Hollow Fiber solvent
preconcn factora
1-octanol 2-octanone hexyl ether isooctane
70 65 62 -b
a Results varied within 20% RSD (n ) 6). b No signal was observed for methampetamine.
Table 3. Preconcentration of Methamphetamine Utilizing Different Donor and Acceptor Solutions donor, NaOH (M)
acceptor, HCl (M)
preconcn factora
1 10-1 10-2 10-3 10-4 10-1 10-1 10-1 10-1 10-1
10-1 10-1 10-1 10-1 10-1 1 10-1 10-2 10-3 10-4
80 74 65 64 40 b 74 40 17 2
a Results varied within 20% RSD (n ) 6). b Problems related to high ionic strength of sample during capillary electrophoresis.
Donor and Acceptor Solutions. With the hollow fiber impregnated with 1-octanol, subsequent experiments were conducted to optimize the composition (pH) of both the donor and acceptor solutions. For all of the experiments, LLLME was accomplished for 45 min with NaOH in the donor solution (sample) and HCl in the acceptor solution. For the donor phase, the concentration NaOH was varied between 10-4 and 1 M. As illustrated in Table 3, the enrichment factor for methamphetamine increased with increasing pH. With 10-4 M NaOH, the pH of the solution was close to the pKa value of methamphetamine (pKa ) 10.1), but despite this the analyte was enriched by a factor of 40. While 1 M NaOH provided the highest enrichment factor for methamphetamine (approximately 80), 10-1 M NaOH was selected for the rest of the work for both stability and practical reasons. In addition to 10-1 M NaOH, NaCl was added to the donor solution in a subsequent experiment to evaluate the possibility of out-salting the analyte from the donor phase. However, although NaCl was added at concentration levels of 10, 20, and 30%, no further enrichment was obtained. While the composition of the donor solution was not very critical, the enrichment was more sensitive to compositional (pH) variations of the acceptor solution. As illustrated in Table 3, the concentration of HCl in the acceptor solution was varied between 10-4 and 1 M. With 1 M HCl, problems were observed during capillary electrophoresis of the extract; owing to the high ionic strength of the sample extract (acceptor solution) as compared with the 50 mM phosphate buffer (pH 2.75) utilized as separation medium, the methamphetamine peak was heavily distored and defocused due to antistacking. With 10-1 M HCl in contrast, the peak shape was significantly improved and methamphetamine was enriched by a factor of approximately 74. As the concentration of 2654 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
Figure 3. Preconcentration of methamphetamine versus extraction time.
HCl was decreased below 10-1 M, the enrichment factor for methamphetamine was dramatically reduced. Although the initial pH of the acceptor solution of 10-4 M HCl was far below the pKa value of methamphetamine (pKa ) 10.1), no enrichment was observed because the absolute buffer capacity within the small volume of acceptor solution was low. Thus, on the basis of the results in Table 3, 10-1 M HCl was selected as the acceptor solution for the rest of the work. Extraction Time. With the hollow fiber impregnated with 1-octanol, 10-1 M NaOH in the donor solution, and 10-1 M HCl as the acceptor solution, the extraction time was optimized for methamphetamine in Figure 3. The amount of methamphetamine extracted by LLLME increased with increasing exposure time from 5 to 45 min, while the enrichment factor stabilized at about 70 utilizing extraction times above 45 min. On the basis of these results, 45 min was selected as the extraction time. Although the extraction time was relatively long, a large number of different samples (5-20) were extracted simultaneously resulting in a very high sample capacity. Extraction Efficiency and Distribution Ratio. On the basis of the experiments discussed above, optimal LLLME of methamphetamine was obtained by utilizing a porous hollow fiber impregnated with 1-octanol, an acceptor solution of 10-1 M HCl, a donor solution containing 10-1 M NaOH, and an extraction time of 45 min. In this case, methamphetamine was typically enriched by a factor of 75 from aqueous samples (Figure 4). This corresponded to a 75% extraction efficiency. With Va2 ) 25 µL, Va1 ) 2.5 mL, and Ca1,initial ) 100 ng/mL, a value of 293 was obtained for the partition coefficient between the acceptor phase and the donor phase (K) for methamphetamine by utilizing eq 6. This value effectively explained the significant enrichments obtained in the present work. Validation. To evaluate the practical applicability of the proposed LLLME technique, repeatability, linearity, detection limit, and limit of quantification were investigated by utilizing standard solutions of methamphetamine in water. For this purpose, diphenhydramine was added as internal standard at the 50 ng/mL level. This compound was appropiate because it was enriched approximately by the same factor as methamphetamine (Ee ≈ 75) and it was easily separated from methamphetamine by the CE system. In a first experiment, six replicates of methamphetamine (100 ng/mL) and diphenhydramine (50 ng/mL) were treated by
Figure 4. Preconcentration of 100 ng/mL methamphetamine from a standard solution in water: CE buffer, 50 mM phosphate pH 2.75; separation voltage, 15 kV; capillary, 10 cm (effective length) × 75 µm i.d.; detection, UV at 200 nm.
LLLME/CE. The absolute peak areas for methamphetamine and the internal standard varied within 20% RSD, but the repeability was significantly improved when the results for methamphetamine were corrected with the internal standard. In the latter case, the results varied within 6.8% RSD when peak areas were utilized and within 5.2% RSD based on peak heights. The repeatability was acceptable and comparable with other microextraction techniques reported in the litterature.19 In a subsequent experiment, the linearity was evaluated within the range 20 ng/mL to 10 µg/mL where a linear regression coefficient of 0.9983 was obtained. The limit of detection (S/N ) 3) for methamphetamine was 5 ng/mL, and the limit of quantitation was 17 ng/nL (S/N ) 10). The broad linear range combined with the low detection limit suggests a high potential for monitoring methamphetamine in human urine and plasma by LLLME/CE. Preconcentration of Methamphetamine from Human Urine and Plasma. To finish the evaluation of LLLME, methamphetamine was preconcentrated from both human urine and plasma. Aliquots (2.5 mL) of both samples were spiked with 100 ng/mL methamphetamine, and subsequently, 125 µL of 2 M NaOH was added to each to provide an approximately 10-1 M concentration in the donor solution. The low volume of a strong NaOH solution served to effectively reduce dilution of the sample prior to LLLME. LLLME was performed for 45 min by utilizing a 10-1 M solution of HCl inside the porous hollow fiber. As illustrated in Figures 5 and 6, methamphetamine was effectively preconcentrated also from the biological samples. In both cases, an enrichment factor of approximately 75 was obtained, which provided a 5 ng/mL detection limit (S/N ) 3). The enrichment factors indicated that LLLME of methamphetamine was not influenced by the high ionic strength of urine or by the proteins present in human plasma. The concentration range available with LLLME/CE was highly relevant for the monitoring of methamphetamine in human urine and plasma. In addition to excellent enrichment, a high sample cleanup potential was observed for LLLME in connection with the biological samples. For drug-free plasma, only a few other peaks emerged in the electropherograms
Figure 5. Preconcentration of 100 ng/mL methamphetamine from human urine: CE buffer, 50 mM phosphate pH 2.75; separation voltage, 15 kV; capillary, 30 cm (effective length) × 75 µm i.d.; detection, UV at 200 nm.
Figure 6. Preconcentration of 100 ng/mL methamphetamine from human plasma: CE buffer, 50 mM phosphate pH 2.75; separation voltage, 15 kV; capillary, 10 cm (effective length) × 75 µm i.d.; detection, UV at 200 nm.
(Figure 6), which enabled very rapid CE runs in a 10-cm effective length capillary. For human urine, the drug-free samples were also very clean when exposed to LLLME (Figure 5), but a single peak present in the blanks comigrated with methamphetamine when analyzed in 10-cm CE capillaries. Thus, for urine samples, a 30-cm capillary was utilized to ensure sufficient separation of methamphetamine. Advantages of LLLME/CE. The present work has demonstrated a high potential of LLLME/CE based on a porous polypropylene fiber. For methamphetamine present in both human urine and plasma, a 65-80 times enrichment and effective sample cleanup was accomplished with 45 min of LLLME. Owing to the off-line nature of the concept and the low costs of each extraction Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
2655
unit, a large number of samples were prepared simultaneously. With LLLME in combination with the short CE runs, a large number of samples were analyzed within short time. The cost of each extraction unit was low and each unit was a disposable device utilized only for a single extraction. This was a major advantage because cross-contamination and carry-over effects were totally eliminated. Work is in progress to evaluate the concept for other
2656 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999
basic compounds and for acidic compounds and to fully automate LLLME as an attractive off-line sample preparation technique for CE. Received for review January 20, 1999. Accepted April 20, 1999. AC990055N
