Fully Automatic Sample Treatment by Integration of Microextraction by

Mar 13, 2009 - Fully Automatic Sample Treatment by Integration of Microextraction by Packed Sorbents into ... The suggested integration allows the aut...
0 downloads 0 Views 427KB Size
Anal. Chem. 2009, 81, 3188–3193

Fully Automatic Sample Treatment by Integration of Microextraction by Packed Sorbents into Commercial Capillary Electrophoresis-Mass Spectrometry Equipment: Application to the Determination of Fluoroquinolones in Urine Gabriel Morales-Cid, Soledad Ca ´ rdenas, Bartolome ´ M. Simonet, and Miguel Valca ´ rcel* Department of Analytical Chemistry, Marie Curie Building (Annex), Campus de Rabanales, University of Cordoba, E-14071 Cordoba, Spain This paper describes a new and innovative way to integrate microextraction by packed sorbents (MEPS) into commercial CE equipment. The suggested integration allows the automatic sample cleanup and preconcentration requiring only a few microliters of sample and no additional hardware and software. The MEPS was integrated in the outlet region of a commercial CE equipment cartridge in order to provide easy manipulation and exchange. The robustness of the proposed integration was demonstrated by the design and use of a (MEPS)-nonaqueous capillary electrophoresis (NACE)-MS method used to determine fluoroquinolones “FQs” (namely, ofloxacin, marbofloxacin, enrofloxacin, danofloxacin, and difloxacin) in urine. The method allows the analysis of micrograms per liter of FQs to be carried out with only 48 µL of urine sample. The obtained LODs were in the range 6.3-10.6 µg/L. An analysis of spiked urine samples was used to validate the method. Absolute recoveries were in the range of 71-109% while the precision expressed as repetitivity of peak area was lower than 5.9%. The sample treatment step in analytical procedure ought to be simplified using automated and simple hardware configurations. The coupling between the sample treatment unit and the analytical instrument is critical for this purpose. It can be performed discontinuously with two possible configurations (i) off-line, in which the operator collects the sample aliquots and introduces them into the analyzer, and (ii) at-line, in which a robotic interface is programmed to carry out the transfer step. The treated sample can also be introduced into the analyzer onand in-line by means of an automated and continuous transfer line.1 In this scenario, both on- and in-line setup offer several advantages over the discontinuous couplings, since the sample treatment can be fully automated, and therefore both operator supervision and the implementation of secondary devices are unnecessary. Alternatively, this paper describes a discontinuous * To whom correspondence should be addressed. Phone/fax: +34 957 218616. E-mail: [email protected]. (1) Santos, B.; Simonet, B. M.; Rios, A.; Valcarcel, M. TrAC, Trends Anal. Chem. 2006, 25, 968–976.

3188

Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

coupling where the sample treatment unit is integrated in the commercial equipment, enabling the use of a fully automated system. The development of new strategies for on-capillary sample preconcentration and cleanup proved to be very interesting for the analysis of complex matrixes. Although on-capillary SPECE-MS coupling feasibility has previously been demonstrated,2-7 certain disadvantages still remain. Two of the most important drawbacks of on-capillary SPE-CE couplings are (i) high-speed loading is not possible and (ii) the sample matrix is introduced into the capillary, which can result in the adsorption of matrix compounds on the capillary walls. This in turn makes it necessary to introduce an additional cleanup step. These disadvantages can be easily balanced out by implementing online coupling. For a complete overview about SPE-CE hyphenation, readers are referred to ref 8. In this work, Tempels et al. perform an exhaustive review about the interface designs for in-line and online SPE-CE. For instance, the use of on-line coupling of SPE and CE-MS for peptide analysis has been demonstrated elsewhere.9 For this article, the authors coupled an SPE microcolumn with CE via a valve interface. Although the system is very robust, there are some drawbacks. The coupling via a switching valve is quite challenging, and a number of safety recommendations are essential in order to avoid unintended electrical discharges. In addition, they concluded that it is difficult to accommodate the elution volumes and the CE injection volumes since only 4-5% of the SPE elution volume is injected. (2) Waterval, J. C. M.; Hommels, G.; Bestebreurtje, P.; Versluis, C.; Heck, A. J. R.; Bult, A.; Lingeman, H.; Underberg, W. J. M. Electrophoresis 2001, 22, 2709–2716. (3) Viberg, P.; Nilsson, S.; Skog, K. Anal. Bioanal. Chem. 2004, 378, 1729– 1734. (4) Lara, F. J.; Garcia-Campana, A. M.; Ales-Barrero, F.; Bosque-Sendra, J. M. Electrophoresis 2008, 29, 2117–2125. (5) Macia, A.; Borrull, F.; Calull, M.; Benavente, F.; Hernandez, E.; Sanz-Nebot, V.; Barbosa, J.; Aguilar, C. J. Sep. Sci. 2008, 31, 872–880. (6) Benavente, F.; Vescina, M. C.; Hernandez, E.; Sanz-Nebot, V.; Barbosa, J.; Guzman, N. A. J. Chromatogr., A 2007, 1140, 205–212. (7) Puig, P.; Tempels, F. W. A.; Somsen, G. W.; de Jong, G. J.; Borrull, F.; Aguilar, C.; Calull, M. Electrophoresis 2008, 29, 1339–1346. (8) Tempels, F. W. A.; Underberg, W. J. M.; Somsen, G. W.; de Jong, G. J. Electrophoresis 2008, 29, 108–128. (9) Tempels, F. W.; Underberg, W. J.; Somsen, G. W.; de Jong, G. J. Electrophoresis 2007, 28, 1319–1326. 10.1021/ac900234j CCC: $40.75  2009 American Chemical Society Published on Web 03/13/2009

To date, in- and on-line SPE-CE coupled equipment has been scarcely applied in routine. Probably because these couplings still need considerable human participation, especially to make the packed sorbents. Microextraction by packed sorbents (MEPS), which is also called microextraction by packed syringe, is the integration of a miniaturized SPE bed with a microvolume syringe (100-250 µL). Commercially available MEPS uses an integrated barrel insert and needle (BIN) device to reduce SPE to a microscale suitable for small sample volumes and to dramatically reduce the eluent volume. Furthermore, MEPS is a portable SPE device which can be operated manually or incorporated into robotics samplers.10 Compared with liquid-liquid extraction and SPE, MEPS reduces the sample preparation time and the organic solvent consumption.11,12 Fluoroquinolones (FQs) are broad spectrum antibiotics, which are included in the quinolones (Qs) group. At present, there are four generations of Qs.13,14 The latest FQ generations are replacing the first-generation of Qs since they have lost their activity against bacteria, as some of these microorganisms have developed a resistance.15,16 Nowadays, FQs are the most extensively used drugs for the treatment of bacterial infections both in human and veterinary medicine. The action mechanism is slightly different depending on each group. Quinolones target bacteria type II DNA topoisomerases,17 while FQs act by binding to gyrase or topoisomerase IV in the presence of DNA.18 The determination of FQs and Qs by LC in biological samples19,20 and, furthermore, their determination using a number of separation techniques, including LC and CE in food samples and the environment, have recently been reviewed.21 In addition, FQs have been determined by CE-MS in chicken and fish,22 fish and livestock,23 and human fluids such as human urine.24,25 The aim of this work is to study the potential of a new, fully automated and integrated MEPS-CE-MS system designed to discriminate and quantify FQs in human urine. In addition, the coupling setup between MEPS and CE-MS is also discussed below. EXPERIMENTAL SECTION Reagents and Standards. FQs standards ofloxacin, marbofloxacin, enrofloxacin, danofloxacin, and difloxacin (analytes are abbreviated as OFLO, MARBO, ENRO, DANO, and DIFLO, respectively) were purchased from Sigma-Aldrich (Madrid, Spain). (10) Wynne, P.; Hibbert, R.; DiFeo, D.; Dawes, P. Column 2008, 12–17. (11) El-Beqqali, A.; Abdel-Rehim, M. J. Sep. Sci. 2007, 30, 2501–2505. (12) Morales-Cid, G.; Cardenas, S.; Simonet, B. M.; Valcarcel, M. Electrophoresis 2009, 30, 1–8. (13) Andriole, V. T. Clin. Infect. Dis. 2005, 41 (Suppl. 2), S113–119. (14) Oliphant, C. M.; Green, G. M. Am. Fam. Physician 2002, 65, 455–464. (15) Madurga, S.; Sanchez-Cespedes, J.; Belda, I.; Vila, J.; Giralt, E. ChemBioChem 2008, 9, 2081–2086. (16) Van Bambeke, F.; Michot, J. M.; Van Eldere, J.; Tulkens, P. M. Clin. Microbiol. Infect. 2005, 11, 256–280. (17) Marians, K. J.; Hiasa, H. J. Biol. Chem. 1997, 272, 9401–9409. (18) Drlica, K. Curr. Opin. Microbiol. 1999, 2, 504–508. (19) Carlucci, G. J. Chromatogr., A 1998, 812, 343–367. (20) Samanidou, V. F.; Christodoulou, E. A.; Papadoyannis, I. N. Curr. Pharm. Anal. 2005, 283–308. (21) Andreu, V.; Blasco, C.; Pico, Y. TrAC, Trends Anal. Chem. 2007, 26, 534– 556. (22) Juan-Garcia, A.; Font, G.; Pico, Y. Electrophoresis 2006, 27, 2240–2249. (23) Juan-Garcia, A.; Font, G.; Pico, Y. Electrophoresis 2007, 28, 4180–4191. (24) Wei, S.; Lin, J.; Li, H.; Lin, J. M. J. Chromatogr., A 2007, 1163, 333–336. (25) Liu, Y. M.; Jia, Y. X.; Tian, W. J. Sep. Sci. 2008, 31, 3765–3771.

Ammonium acetate, formic acid (Sigma-Aldrich; Madrid, Spain), methanol, and acetonitrile (Panreac Quı´mica, S.A.; Barcelona, Spain) were used to prepare the background electrolyte (BGE). Sodium hydroxide and hydrochloric acid (Panreac Quı´mica, S.A.; Barcelona, Spain) were used for capillary conditioning. Standard stock solutions of the target analytes were prepared by dissolving the required amount in methanol to obtain a 100 mg/L concentration of each compound and subsequently stored at 4 °C. Standard solutions of appropriate concentrations were prepared on a daily basis by diluting the stock solutions in methanol or Milli-Q water. Instrumentation. An Agilent HP3D capillary electrophoresis system (Waldbronn, Germany) coupled to an Agilent 1100 series LC/MSD mass spectrometer via an electrospray-API interface was used. The makeup flow of sheath liquid was delivered by an Agilent 1100 isocratic pump, which was operated at a 1:100 split ratio. The uncoated fused silica capillaries used (60 cm × 50 µm i.d., 375 µm o.d.) were supplied by Beckman (Fullerton, CA). Agilent ChemStation software was used to control the CE-MS system and to acquire and process data. pKa values were performed with Pallas 3.1 for Windows with pKalc 5.1 (CompuDrug, Budapest, Hungary). Instrumental Operating Conditions. The selected BGE consisted of 20 mM ammonium acetate in a 50:50 (v/v) acetonitrile/methanol solution adjusted to pH 4 with formic acid. It was prepared on a weekly basis. Each new capillary was sequentially conditioned at 2 bar using 1 M HCl (5 min), 0.1 M NaOH (10 min), Milli-Q water (5 min), and BGE (10 min). At the beginning of each day, the capillary was conditioned, applying 930 mbar as follows: 0.1 M NaOH (10 min), Milli-Q water (5 min), running buffer (10 min). Between runs, the capillary was filled and conditioned with the BGE for 6 min at 930 mbar. Analyses were performed at 25 kV using the standard polarity mode (outlet polarity is negative). The capillary cartridge was kept at a temperature of 20 °C. Mass spectrometric conditions were as follows: the sheath liquid composition was 50:50 (v/v) methanol-acetonitrile plus 2% formic acid. In order to ensure reproducibility and to avoid alterations caused by the evaporation of methanol and formic acid from the solution, the sheath liquid was prepared on a weekly basis. It was pumped into the sprayer at a flow rate of 4 µL min-1. The fragmentor voltage was set at 80 V. The selected nebulizer pressure was 3 psi, and the N2 (drying gas) was delivered at a flow rate of 10 L min-1 at 200 °C. The capillary voltage was kept constant at 4 kV. For data collection, the ions with the m/z values corresponding to [M + H]+ for each FQ were monitored. The peak area of the corresponding ions was measured for each single analyte. Values were 362, 363, 360, 358, and 400 for OFLO, MARBO, ENRO, DANO, and DIFLO, respectively. The spray chamber was cleaned with methanol and Milli-Q water on a weekly basis. Daily cleaning with Milli-Q water is also recommended. MEPS-CE-MS Setup. Sample treatment was performed with a microextraction by packed sorbents system from SGE Analytical Science (Melbourne, Australia); each barrel insert and needle (BIN) contained ∼4 mg of C18 packing with a particle size of 45 µm and pore size of 60 Å. In order to perform the sample treatment and the measurement of urine samples, a commercially available Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

3189

Figure 1. CE-MS capillary cartridge connected to MEPS.

C18 BIN was fitted to the outlet position of the CE-MS cartridge. A Teflon tube measuring 0.3 mm i.d. × 4.25 m length was connected to the BIN and inserted into the CE-MS cartridge. This loop worked as a reservoir (∼300 µL volume) where the solvents could be stored in order to allow for the conditioning of the C18, the preconcentration, and the elution step (Figure 1). One of the capillary tips was placed on the inlet position of the CE cartridge following the usual procedure for connecting CE equipment and a mass spectrometer, and the other tip was connected to the ionization chamber of the mass spectrometer. The CE instrument had two possible pressure sources: (i) an external one, which made it possible to apply pressure ranging between 2 and 15 bar, and (ii) an internal one, allowing pressure to range from -50 to 50 mbar. It also allowed the capillary to be flushed at a pressure of ∼930 mbar. In order to create a fully automated method, the hardware which controlled the fluid/pressure module of the CE equipment was slightly modified. Prior to that, it was necessary to select the CEC+MS mode of the instrument software. If you have an external pressure supply connected to the instrument, the CEC mode allows applying high pressure on both vial lifts during operation. However, internal pressure cannot be applied on the outlet lift since it is only available on the inlet lift. Therefore, only external pressure can be applied on the outlet lift, while both high and low pressures can be used on the inlet lift. Moreover, in the fluid hardware both external and internal pressures converge on the same line before connecting with the inlet lift. We aimed to divide these different pressures lines into three differentiated lines, including one internal and two external lines. Using these three lines, we achieved two objectives: sample preconcentration in MEPS using internal pressure and automation using two external pressure lines. An additional three-way valve which could be activated by air was connected to the pressure hardware of the equipment. In Figure 2A, the common setup of the pressure/ liquid handling hardware is shown. As can be seen, there are two tubes which connect the valve module where the external pressure is connected to the vial lifts (tubes 1 and 2). These tubes were removed from their connection to the valve block and subsequently connected to the two opposite inlets of the three-way valve. In addition, the internal pressure coming from the air pump (tube 3) was also connected to the valve on its last free inlet. The final configuration is shown in Figure 2B. The outlets of the external pressure valve module which are joint to the inlet and outlet lifts 3190

Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

(see Figure 2A) were connected to the free inlets opposite the additional valve (tubes 4 and 5). These additional inlets of the valve allowed the valve position to be switched by increasing pressure (g2 bar). It is important to note that because internal pressure was disconnected from the external pressure valve block, only external pressure could be applied through both outlets. In the final setup, the use of external pressure on the inlet lift switches the valve position and when applied on the outlet lifts it switches the valve back to its previous position. Furthermore, applying internal pressure on the inlet lift made it possible to apply internal pressure on the outlet lift or inlet lift, depending on the position of the valve. To sum up, we were able to apply both positive and negative internal pressures to aspirate and deliver fluids, respectively, in either the inlet lift or outlet lift. Internal pressure was used for two purposes: (i) to flush the CE capillary (at 930 mbar) and (ii) to deliver or aspirate the sample and eluents through MEPS at -50 and 50 mbar, respectively. Procedure for the Automated Sample Preconcentration and Cleanup. The procedure requires four basic steps. All were performed at either 50 or -50 mbar in order to aspirate or to deliver solution respectively. Step 1, conditioning and cleaning of the C18 packing; 100 µL of methanol was aspirated and later delivered to a waste vial. Then, 100 µL of Milli-Q water containing 2% formic acid was aspirated and delivered to the waste vial. Step 2, concentrating the sample aliquot; a mixture of 48 µL of a urine sample with 192 µL of 2.5% formic acid solution was passed through the MEPS and delivered to the waste vial through MEPS, as seen before. Analytes were trapped on the sorbent. Step 3, drying the sorbent; a small volume of air was aspirated in order to dry the C18 packing, 50 mbar for 2 min. Step 4, eluting analytes; 40 µL of 5% formic acid solution prepared in methanol was passed through MEPS and then delivered and collected in a mini glass vial in the autosampler. RESULTS AND DISCUSSION Nonaqueous Capillary Electrophoresis (NACE) Separation and MS Detection. First we studied the use of different BGEs prepared in both aqueous and nonaqueous media containing ammonium acetate as the electrolyte. Aqueous BGEs including water and water-methanol mixtures resulted in a lower sensitivity compared with nonaqueous BGEs. In terms of resolution, the differences between the studied BGEs were not so critical. Nonetheless, the resolution was slightly improved when NACEs were used. Furthermore, the use of organic solvents as BGEs makes the system more compatible with the MS detector. The electrophoretic conditions for the separation of quinolones using methanol-acetonitrile mixtures such as BGEs containing ammonium acetate have already been discussed in a previous article.26 The author suggested a 50: 50 (v/v) methanol-acetonitre buffer containing 20 mM ammonium acetate, which was adjusted to apparent pH (pH*) 5.4. Preliminary tests showed that sensitivity was clearly higher than with the other tested buffers, and resolution was not significantly affected. However, it is necessary to modify some of these parameters when, instead of a UV detector, a mass spectrometer is used because of the presence of a hydrodynamic flow produced by the siphoning effect of the ESI (26) Hernandez, M.; Borrull, F.; Calull, M. Electrophoresis 2002, 23, 506–511.

Figure 2. Schematic representation of normal and modified setup of liquid/pressure module of the CE equipment. Table 1. Analytical Figures of Merit of the Whole Method for the Determination of Five Fluoroquinolonesa analyte

slope

intercept

R2

LODb (µg/L)

LOQc (µg/L)

RSDd (%)

RSDe (%)

OFLO MARBO ENRO DANO DIFLO

293.8 380.7 546.7 556.9 288.7

5263.9 7208.4 7508.9 6845.8 7192.6

0.998 0.997 0.998 0.998 0.994

6.5 7.3 6.3 7.0 10.6

21.8 24.2 21.1 23.3 35.4

3.4 4.1 5.5 5.2 5.9

1.4 1.4 1.5 1.2 1.4

a Calibration graphs were obtained at seven different concentration levels, ranging between 12.5 and 500 µg/L. The repeatability of the peak area and migration time measured as RSD was calculated at 100 µg/L. b Calculated as 3Sy/x/slope. c Calculated as 10Sy/x/slope. d Relative standard deviation of peak area in six measurements. e Relative standard deviation of migration time in six measurements.

interface. In order to achieve a more volatile BGE to improve ionization in the spray chamber, we used formic acid instead of acetic acid for pH adjustment because of its higher vapor pressure. In addition, the pH* also had to be studied. When the pH* was at 5.4, a partial ionization occurred. pKas of the studied quinolones are in the range of 5.66-6.21. With the use of a stronger acidic medium, analytes were easily protonated, which facilitated their separation as cations. For this reason pH* ranging from 5.4 to 3.5 (such values were 5.4, 4.6, 4, 3.5) were studied. When the pH* was reduced to 4, the peak shape improved, since the presence of other species is avoided at this pH* value. On the other hand, extremely low pH levels can affect the integrity of the

capillary walls. This is why a pH* 4 was selected to continue the study. Lower pH* levels were discarded. To speed up the separation process, high voltages were studied. It is not recommended to apply more than 25 kV, because that usually leads to dramatic drops in current intensity due to the Joule effect. As a result, 25 kV was selected to continue the optimization process. With the use of this voltage and the proposed capillary length, each CE analysis (without MEPS sample treatment) was completed in less than 6 min. The injection time was studied, ranging from 3-15 s at 50 mbar. When we introduced a large sample plug, the peak width increased and so did the peak area. However, the resolution and the sensitivity expressed as LOD were not improved. The optimum value was 10 s followed by 3 s of BGE, also carried out at 50 mbar, to ensure the repeatability of the injection. The distance that the capillary protrudes out of the sprayer tip was also studied. The recommended position was at approximately 0.1 mm. We also tested 0.2 and 0.3 mm. In the end, the best position turned out to be the one recommended initially (0.1 mm). When we increased the distance, ionization became less efficient. Drying gas temperatures ranging from 150 to 250 °C were also studied. When this temperature was increased to over 200 °C, sensitivity levels decreased. Apart from that, no significant differences were observed when temperature was reduced to 150 °C. In the end, 200 °C was adopted with the aim of ensuring an appropriate ionization of the analytes and a favorable sample matrix volatilization. The sheath liquid composition has been known to affect both electrospray stability and ionization. We studied various mixtures Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

3191

Table 2. Recoveries Obtained from the Analysis Spiked Human Urine Samples sample

added (µg/L)

1 2 3 4 5

12.5 50 100 50 0

1 2 3 4 5

12.5 50 100 50 0

1 2 3 4 5

12.5 50 100 50 0

1 2 3 4 5

12.5 50 100 50 0

1 2 3 4 5

12.5 50 100 50 0

found (µg/L)

absolute recovery (%)

Difloxacin 11.5 54.7 98.4 49.2

91.8 109.3 98.4 98.5

Marbofloxacin 10.8 52.7 98.8 35.4

86.5 105.5 98.8 70.8

Danofloxacin 10.2 51.3 99.2 48.4

81.7 102.6 99.2 96.8

Enrofloxacin 10.9 52.3 99.0 38.7

87.5 104.6 99.0 77.5

Ofloxacin 10.8 49.3 99.9 40.7

86.2 98.6 99.9 81.4

of methanol and water and formic acid. The results show that a 50:50 (v/v) methanol-water mixture containing 2% formic acid provided the best results in terms of electric contact and sensitivity. On the basis of these results, 2% formic acid seemed enough to provide good electrical contact at low drying gas pressure, and the achieved sensitivity was higher than with only 1% formic acid. Sheath liquid flow rates ranging from 3 to 5 µL/min were studied. The highest tested flow rate resulted in lower sensitivity than the lowest one. A 4 µL/min value was adopted as optimal, since it provided the best sensitivity levels. Nebulizer pressure is a critical parameter in both the resolution of the mixture and the sensitivity of the method. Pressures from 1 to 10 psi were studied. Pressures above 3 psi resulted in a marked peak overlap, which was complete at 10 psi. Resolution was better at lower pressure levels, because the influence of the Venturi’s effect caused by the nebulizer gas at the capillary tip was limited. However, pressure levels of less than 3 psi sometimes resulted in an interruption of the electrical contact and the loss of sensitivity due to the increase of baseline noise. According to these results, 3 psi is the recommended value. We also examined the influence of the flow rate of the drying gas, which was studied for 1-13 L/min. This parameter turned out not to be as critical as the previous one though. The best sensitivity level was achieved at 10 L/min judging from the results. Values over 13 L/min sometimes resulted in a sudden current drop and a loss in sensitivity and resolution. At the end, the influence of the fragmentor voltage on the ionization process was studied applying 60-120 V. Finally, a value of 80 V was selected. 3192

Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

Sample Treatment Optimization. Preliminary batch MEPS tests were performed using a glass syringe to fit MEPS on water samples spiked with 100 µg/L of each standard. An aliquot of 240 µL of this standard solution was concentrated through MEPS. Before elution, the sorbent was dried with air and then eluted with methanol three times in a row (25 µL × 3). Its subsequent analysis revealed the presence of traces of the analytes in the second eluate but none in the third. We analyzed the effect of adding one very volatile acid or one base, ammonium hydroxide or formic acid, in order to minimize the carry-over. The presence of acids or bases altered the distribution in the sorbent and eluent due to a modification of the pH level of the matrix. This is why the addition of ammonium hydroxide between 1 and 10% vol was looked at more closely. However, adding 1% the elution was not complete, and the peak shape was asymmetric. When 5 to 10% was added, no changes could be observed. Formic acid was added to the eluent at concentrations between 1 and 10% vol. With the use of formic acid, the elution was complete. In the end, the selected value was 5%, because higher ones did not increase the elution levels of the analytes. Adding this amount of formic acid reduces the carry-over down to concentration levels which are not quantifiable. Afterward, a closer look was taken at the preconcentration step. To this end, formic acid was added to the sample and tested. The results confirmed that when we added a small concentration of formic acid in the sample (2% vol), the preconcentration improved. Although the off-line optimization of the sample treatment is basic for the coupling of MEPS in the CE-MS equipment, the integration required some specifications to be taken into account. In order to calculate the volume per time that the system could aspirate by applying internal pressure, the device had to be calibrated. We then validated the model doing a number of repetitions at different time intervals in order to calculate RSD values. The calculated flow rate obtained when applying a pressure of 50 mbar was 37.5 µL/min. These recommendations should be reoptimized when this kind of system is integrated into a piece of equipment. The obtained within-day RSD in terms of aspirated or delivered eluent volume was e4.0%. After that, the investigation of the preconcentration factor was carried out. In order to obtain reproducible results, we calculated the RSD in terms of peak area for analytes using standard solutions. The RSD reached unacceptable values when the elution of analytes was carried out using volumes which were 12 times lower than the volume which passed through the C18 packing. Therefore, and as a compromise between the RSD and the elution volume, an eluent volume 6 times lower than the one used for the preconcentration step was selected (40 µL). Using this ratio, we were able to achieve RSDs lower than 5.9% in terms of peak area. The procedure for the automatic sample treatment is described in the Experimental Section. Finally, each BIN can be used more than 40 times for this application. As a result, this MEPS behavior increases the robustness of the proposed method and helps reduce costs. Analytical Figures of Merit. Table 1 shows the analytical figures of merit for the proposed method. The calibration curves were calculated for seven concentration levels (each concentration was injected in triplicate). Such concentrations were 12.5, 25, 50,

100, 200, 350, and 500 µg/L. LODs and LOQs were calculated as 3Sy/x/slope and 10Sy/x/slope, respectively. LOD ranged from 6.3 to 10.6 µg/L. The repeatability of the method was also studied by means of repetitive analysis (n ) 4) of the same sample following the recommended procedure. The obtained RSD values for peak areas and migration times were lower than 5.9% and 1.5%, respectively. These results demonstrate the suitability of the suggested method, since these errors are not more critical than those obtained from conventional CE-MS methods including SPE as sample treatment. Linearity was in the range of 0.994-0.998. Analysis of Urine Samples. Human urine samples were diluted 1:5 with Milli-Q water, spiked with a known amount of each FQ, and acidified with 2% formic acid. The expected concentration after the sample treatment was twice as high as the concentration in the urine sample. Table 2 shows the absolute recoveries (calculated by using the calibration equations) for all the analytes. As can be seen, there were no interferences in unspiked human urine analyses. The absolute recoveries ranged from 70.8 to 109.3%. The separation window was free of interferences due to the matrix. Therefore, analytes can be easily discriminated and quantified using the m/z value of each single ion.

demonstrated. This new tool offers several advantages over the previous micro-SPE-CE-MS couplings, since its use is easy and inexpensive. These aspects make this strategy suitable to be applied to routine laboratories. The use of a totally integrated pressure switching valve in the CE instrument helps avoid problems caused by current discharges and makes the methodology more robust as well as feasible. The reutilization of MEPS for more than 40 analyses also reduces the cost of each analysis. Moreover, even an unskilled worker can easily replace MEPS. The operator simply has to remove the old MEPS from the CE cartridge and manually attach the new one by connecting the loop. Thus, access to the inner components of the CE apparatus is unnecessary. Finally, the automation of the procedure reduces the necessary manpower as well as the risk of human error.

CONCLUSION The ability to integrate MEPS in CE equipment to perform automatic sample cleanup and preconcentration tasks has been

Received for review February 2, 2009. Accepted February 27, 2009.

ACKNOWLEDGMENT The authors would like thank the Spanish Ministry of Innovation and Science for Project CTQ2007-60426 and the Junta of Andalusia for Project FQM02300. G. Morales-Cid also wishes to thank the Ministry for the award of a Research Training Fellowship (Grant BES-2005-10603).

AC900234J

Analytical Chemistry, Vol. 81, No. 8, April 15, 2009

3193