Anal. Chem. 2003, 75, 6799-6806
Solid-Phase Extraction of Amphiphiles Based on Mixed Hemimicelle/Admicelle Formation: Application to the Concentration of Benzalkonium Surfactants in Sewage and River Water Francisco Merino, Soledad Rubio, and Dolores Pe´rez-Bendito*
Department of Analytical Chemistry, Facultad de Ciencias, Edificio Anexo Marie Curie Campus de Rabanales, University of Co´ rdoba, 14071 Co´ rdoba, Spain
The capability of surfactant-coated mineral oxides to aid the solid-phase extraction (SPE) of amphiphiles based on the formation of mixed hemimicelles/admicelles was investigated. The approach is illustrated by studying the adsolubilization of benzalkonium homologue (C12, C14, C16) surfactants (BAS) on sodium dodecyl sulfate (SDS)coated alumina. These oppositely charged surfactants form mixed aggregates on alumina causing retention of BAS by strong hydrophobic and ionic interactions. The recovery of BAS was found quantitative and independent of the alkyl chain length under a wide range of experimental conditions (3-200 mg of SDS/g of alumina; pH 2-11; sample flow rate 3-20 mL/min, and sample loading volume 0.025-1 L). Anionic and nonionic surfactants and electrolytes did not interfere to the levels found in raw sewage. Combination of BAS adsolubilization-based SPE with liquid chromatography/electrospray ionization in positive ion mode/ion trap mass spectrometry permitted the quantification of BAS with detection limits of 4 ng/L and their identification by isolation and subsequent fragmentation in the ion trap. The approach developed was applied to the determination of BAS in raw and treated sewage and river samples. The concentrations of benzalkonium surfactants found ranged between 0.1 and 49 µg/L. Solid-phase extraction (SPE) at present is the most popular sample preparation method.1 This technique applies to a wide range of compound classes and has undergone considerable development in the last years, with many improvements in format, automation, and introduction of new phases. Its acceptation in routine analysis as an alternative to liquid-liquid extraction is increasing in different areas, mainly environmental monitoring, where several official methods for analysis of organic compounds in drinking water and wastewater use SPE as the sample preparation method.2 * Corresponding author. Fax: 34-957-218644. E-mail:
[email protected]. (1) Hennion, M. C. J. Chromatogr., A 1999, 856, 3-54. (2) Hodgeson, J. W. Method 549, Determination of Diquat and Paraquat in Drinking Water by Liquid-solid Extraction and HPLC with UV detection; EPA: Cincinnati, OH, 1990; p 101. 10.1021/ac030224a CCC: $25.00 Published on Web 11/15/2003
© 2003 American Chemical Society
Supramolecular assemblies have been largely used in analytical extraction and concentration schemes.3-5 Techniques such as micellar-enhanced ultrafiltration,6,7 cloud point extraction,8-10 coacervation,11,12 surfactant-assisted transport of solutes across liquid membranes,13 and reverse micellar extractions14 have been extensively developed in the last years and their basic features and practical applications established. The use of supramolecular assemblies in SPE has been however hardly explored. Ionic surfactants adsorb on metal oxides such as alumina, silica, titanium dioxide, and ferric oxyhydroxides forming aggregates termed hemimicelles and admicelles,15 which present a high potential to be used as sorbent materials in SPE.5 These aggregates consist of monolayers of surfactants adsorbing head down on an oppositely charged surface (hemimicelles) and surfactant bilayers (admicelles).16-20 The process in which the analytes are partitioned between the bulk solution and these aggregates is referred to as adsolubilization, a phenomenon akin to micellar solubilization in bulk aqueous solutions. (3) Hinze, W. L., Armstrong, D. W., Eds. Ordered Media in Chemical Separation; ACS Symposium Series 342: American Chemical Society: Washington, DC, 1987. (4) Pramauro, E.; Pelizzetti, E. Surfactants in Analytical Chemisty, Applications of Organized Amphiphilic Media; Elsevier Science, Amsterdam, The Netherlands, 1996. (5) Rubio, S.; Pe´rez-Bendito, D. Trends Anal. Chem. 2003, 22, 470-485. (6) Dunn, R. O., Jr.; Scamehorn, J. F.; Christian, S. D. Sep. Sci. Technol. 1985, 20, 257-284. (7) Fillipi, B. R.; Scamehorn, J. F.; Christian, S. D.; Taylor, R. W. J. Membr. Sci. 1998, 145, 27-44. (8) Watanabe, H.; Tanaka, H. Talanta 1978, 25, 585-589. (9) Saitoh, T.; Hinze, W. L. Anal. Chem. 1991, 63, 2520-2525. (10) Martinez, R. C.; Gonzalo, E. R.; Cordero, B. M.; Pavo´n, J. L. P.; Pinto, C. G.; Laespada, E. F. J. Chromatogr., A 2000, 902, 251-265. (11) Casero, I.; Sicilia, D.; Rubio, S.; Pe´rez-Bendito, D. Anal. Chem. 1999, 71, 4519-4526. (12) Merino, F.; Rubio, S.; Pe´rez-Bendito, D. J. Chromatogr., A 2002, 962, 1-8. (13) Armstrong, D. W. Sep. Purif. Methods 1985, 14, 213-225. (14) Hinze, W. L.; Pramauro, E. Crit. Rev. Anal. Chem. 1993, 24, 133-177. (15) O’Haver, J. H.; Harwell, J. H.; Lobban, L. L.; O’Rear E. A. In Solubilization in Surfactant Aggregates; Christian, S. D., Scamehorn, J. F., Eds.; Dekker: New York, 1995; Chapter 8. (16) Bohmer, M. R.; Koopal, L. K. Langmuir 1992, 8, 2649-2659. (17) Bohmer M. R.; Koopal, L. K. Langmuir 1992, 8, 2660-2665. (18) Goloub, T. P.; Koopal, L. K.; Bijsterbosch, B. H.; Sidorova, M. P. Langmuir 1996, 12, 3188-3194. (19) Valsaraj, K. T. Sep. Sci. Technol. 1989, 24, 1191-1205. (20) Valsaraj, K. T. Sep. Sci. Technol. 1992, 27, 1633-1642.
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003 6799
The adsolubilization phenomenon has been applied in very different areas15 such as pharmacy, soil remediation, wastewater treatment, thin-film formation, etc. In analytical processes, the applications developed up to date have focused on admicellarenhanced chromatography4 where the main problem found has been the slow elution of the surfactant from the support, thus giving different retention performances with time, and on the preconcentration of heavy metals from aqueous samples, based on the formation of complexes with chelating agents previously adsolubilized,21-25 or on the adsolubilization of hydrophobic chelates previously formed in the bulk aqueous solution.26,27 Recently, application of admicelles to the concentration of chlorophenols in water prior to liquid chromatography has been reported.28 In the absence of ionic interactions, adsolubilization is predominantly controlled by the hydrophobicity of adsolubilizates. Thus, adsorption of pentachlorophenol on sodium dodecyl sulfate (SDS)-γ-alumina admicelles is ∼100% whereas adsorption of 2-chlorophenol is reduced to ∼16%.28 Since both ionic and hydrophobic interactions are possible in hemimicelles/admicelles, these should be especially suitable for the concentration of amphiphiles by the formation of mixed analytes-sorbent aggregates. This unexplored aspect is of interest because of the number of natural and synthetic amphiphiles present in different industrial fields, medicine, pharmacy, and environment. This article deals with the concentration of benzalkonium surfactants (BAS) in sewage and river water on SDS-γ alumina admicelles. Benzalkonium surfactants are alkyl (C12, C14, C16) dimethylbenzylammonium compounds widely used in cleaning and disinfection products. Their fate in the environment is of concern since there is a lack of data on their degradation29 and they are known to be toxic even at low concentrations.30 Few analytical methods have been reported in the literature for the determination of benzalkonium surfactants in aqueous environmental samples.31-33 Because of their strong affinity at the surface of particles, a significant proportion of surfactant disposal will be through adsorption in sewage sludge34 and sediments.35 So, to determine benzalkonium surfactants at the levels found in filtered aqueous samples (a few µg/L),31 the use of a preconcentration step prior to chromatographic analysis is mandatory. (21) Hiraide, M.; Sorouraddin, M. H.; Kawaguchi, H. Anal. Sci. 1994, 10, 125127. (22) Manzoori, J. L.; Sorouraddin, M. H.; Shemirani, F. Talanta 1995, 42, 11511155. (23) Hiraide, M.; Shibata, W. Anal. Sci. 1998, 14, 1085-1088. (24) Manzoori, J. L.; Sorouraddin, M. H.; Shabani, A. M. H. J. Anal. At. Spectrom. 1998, 13, 305-308. (25) Manzoori, J. L.; Sorouraddin, M. H.; Shabani, A. M. H. Microchem. J. 1999, 63, 295-301. (26) Hiraide, M.; Iwasawa, J.; Kawaguchi, H. Talanta 1997, 44, 231-237. (27) Hiraide, M.; Hori, J. Anal. Sci. 1999, 15, 1055-1058. (28) Saitoh, T.; Nakayama, Y.; Hiraide, M.; J. Chromatogr., A 2002, 972, 205209. (29) Mattew, J. S.; Malcolm, N. J. Biochim. Biophys. Acta 2000, 1508, 235-251. (30) Leal, J. S.; Gonzalez, J. J.; Kaiser, K. L. E.; Palabrica, V. S.; Comelles, F.; Garcı´a, M. T. Acta Hydrochim. Hydrobiol. 1994, 22, 13-18. (31) Ferrer, I.; Furlong, E. T. Environ. Sci. Technol. 2001, 35, 2583-2588. (32) Ford, M. J.; Tetler, L. W.; White, J.; Rimmer, D. J. Chromatogr., A 2002, 952, 165-172. (33) Norbeg, J.; Thordarson, E.; Mathiasson, L.; Jo¨nsson, J. A. J. Chromatogr., A 2000, 869, 523-529. (34) Merino, F.; Rubio, S.; Pe´rez-Bendito, D. J. Chromatogr., A 2003, 998, 143154. (35) Ferrer, I.; Furlong, E. T. Anal. Chem. 2002, 74, 1275-1280.
6800
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
Polymeric cartridges have been reported to achieve benzalkonium extraction recoveries of ∼75% from wastewaters provided that the percentage of acetonitrile in these samples was 25%.31 Also, their extraction into chlorobutane, after formation of an ion pair with heptanoic acid, by using a microporous membrane liquid-liquid extraction prior to normal-phase liquid chromatography has been described.33 The objective of the study presented here was to determine the suitability of hemimicelles/admicelles for the concentration of amphiphiles by developing a simple and reliable method for the identification and quantitation of benzalkonium surfactants in sewage and river water based on their adsorption on SDS-γ alumina admicelles and subsequent liquid chromatography/ electrospray ionization/ion trap mass spectrometry (LC/(ESI-IT)MS) analysis. Predominant factors influencing the formation of mixed analyte/sorbent aggregates were investigated. The feasibility of the method was proven by analysis of benzalkonium surfactants in river water samples and sewage samples from various wastewater treatment plants. To our knowledge this work is the first reported adsolubilization-based SPE based on the formation of mixed aggregates between analytes and extractant demonstrating that this approach is an interesting alternative for the concentration of amphiphiles in complex matrixes. EXPERIMENTAL SECTION Chemicals and Materials. All reagents were of analytical reagent-grade and were used as supplied. SDS, benzyldimethyldodecylammonium bromide (BDDA), benzyldimethyltetradecylammonium chloride (BDTA), and benzyldimethylhexadecylammonium chloride (BDHA) were obtained from Aldrich (Milwaukee, WI), and N-dodecylpyridinium chloride was obtained from Merck (Darmstadt, Germany). Stock solutions of the cationic surfactants were prepared in methanol. Nitric acid and HPLCgrade methanol were obtained from Panreac (Sevilla, Spain), and formic acid was obtained from Merck. Alumina (γ-form, for column chromatography) was supplied by Sigma (St. Louis, MO). The physical properties of this mineral oxide were as follows: surface area, 155 m2/g; point of zero charge pcz, 8.5; particle diameter range, 50-200 µm; mean value, 100 µm; mean pore size, 58 Å; density, 3.97 g/cm3. Bond Elut Jr. cartridge columns filled with 500 mg of alumina were obtained from Varian (Victoria, Australia). Sampling. Raw sewage and final effluent were collected from two wastewater treatment plants (WWTPs; Pozoblanco and Baena) in the south of Spain in March 2003. Pozoblanco WWTP receives mainly domestic effluents and Baena WWTP receives ∼30% industrial effluents (mainly from laundries and olive oil industries) mixed with ∼70% domestic wastewaters. River samples were taken from the Guadalquivir in Cordoba city in March 2003. Samples were collected in dark glass containers. Immediately they were filtered through 0.45-µm filters (Whatman GF/F, Osmonics, France) in order to remove suspended solids, and then they were adjusted to pH 2 by the addition of concentrated nitric acid. Finally, they were stored at 4 °C. Admicelles Extraction. The Bond Elut Jr. cartridge columns were conditioned with 10 mL of Milli-Q water. Afterward, admicelles were formed on the alumina by passing a 50-mL 0.01 M nitric acid solution containing 12 mg of SDS. Then, columns were washed with 10 mL of Milli-Q water. Samples (1 L of river water,
500 mL of treated sewage, 250 mL of raw sewage) were preconcentrated on the admicelles and benzalkonium surfactants were eluted with 1 mL of methanol. Solution and sample loading was performed by using a vacuum pump (Eyela A35, Rikakikai Co., Tokyo) at a flow rate of 20 mL/min. Aliquots of the eluate were injected into the LC/(ESI-IT)MS system. Liquid Chromatography/Mass Spectrometry. The analytes were separated, identified and quantified by using a LC/(ESI-IT)MS system (1100 Series LC/MSD, Agilent Technologies, Waldbronn, Germany), equipped with an automatic injector. The injection volume was set at 20 µL. The stationary-phase column was a 15-cm Nova Pack C8 column, with 3.9-mm i.d. and 5-µm particle diameter from Waters (Milford, MA). Methanol and 50 mM ammonium formate buffer (pH 3.5) were used as eluent solvents at a flow rate of 0.8 mL/min. The elution gradient program was as follows: 0-30 min, linear gradient from 70 to 90% of methanol; 30-45 min, isocratic conditions with methanol/ ammonium formate 90:10. The diver valve was programmed to send the mobile phase containing SDS and the most polar matrix compounds to waste. So, only 7 min after the initiation of the elution gradient program, the eluted component were sent to the ESI source. The surfactant analysis was carried out in the “ESI(+)” mode. The set of parameters used was as follows: capillary voltage, -5.0 kV; capillary exit voltage, 40 V; skimmer, 40 V; trap drive, 40; source temperature, 350 °C; drying gas, 10 L/min; nebulizer gas, 80 psi; maximal accumulation time, 150 ms. They were optimized by directly analyzing a mixture of benzalkonium surfactants (10 µg/L of each standard compound) in methanol/ammonium formate (80/20, v/v) using a KD Scientific, model 100, syringe pump (New Hope, MN) at 600 µL/h. Quantification was carried out under different conditions in order to select the best option in terms of sensitivity. Calibrations were run under (a) full-scan conditions (m/z scan range, 200-400), (b) reduced mass range conditions, generally termed SIM (the Agilent system used permits reduction of this range to 10 mass units, so the m/z of the ion parent (5 was the mass scan range selected for each homologue), and (c) specific MS/MS fragmentation (e.g., m/z 91). Homologues were quantified in all cases from the corresponding peak areas of extracted ion chromatograms. Smooth chromatograms were obtained by using the Gauss function (width, 5 points; cycles, 1). Correlation between peak areas and homologues concentrations (0.2-20 ng, absolute amount injected) were determined by linear regression and were in the range 0.9950.9991. The cationic surfactant N-dodecylpyridinium chloride (4 ng, absolute amount injected) was added to standards and samples just prior to instrumental analysis in order to be used as internal standard for quantitation. Structural identification was performed by MS/MS experiments after the parent ion was isolated and fragmented by using the ion trap mass spectrometer.34 The isolation width was set to 4 m/z units, and the resonance excitation was set to 1.1 V. Excitation of the ions was accomplished through collisions with helium. Adsorption Studies. The adsorption isotherm of SDS on γ-alumina was obtained by adding variable amounts of SDS (0300 mg) to an aqueous suspension of 1 g of γ-alumina. The pH of the solution was adjusted to 2.0 with nitric acid. The final aqueous volume of the solution was 50 mL. After vigorous stirring of the
solution for 5 min, the mixture was centrifuged at 5000 rpm for 5 min and the concentrations of SDS in the supernatants were determined by LC/(ESI-IT)MS. The stationary-phase column used was that specified above. The mobile phase was methanol/water 80:20. Under isocratic conditions, SDS eluted at 4.3 min. Quantitation was carried out in the ESI negative mode. The operational conditions of the ESI interface were optimized, and the following values were found: capillary voltage, 5.0 kV; capillary exit voltage, -100 V; skimmer, -40 V; trap drive, 38; source temperature, 350 °C; drying gas, 10 L/min; nebulizer gas, 80 psi; maximal accumulation time, 150 ms. Quantification was carried out under full-scan conditions (m/z scan range, 200-300) by using the extracted molecular ion chromatogram, and the corresponding peak area was measured. Correlation between peak areas and SDS concentration was obtained in the range 25-500 ng (absolute amount injected) with a correlation coefficient of ∼0.998. Adsorption of benzalkonium homologues on SDS admicelles as a function of pH and SDS concentration was carried out by adding 100 µL of an aqueous solution containing 100 mg/L of each homologue to the mixture SDS and γ-alumina (1 g) in a 50mL aqueous solution at a fixed pH value. The amount of anionic surfactant was varied in the range 0-300 mg, and the pH was studied between 2 and 11. The solution was vigorously stirred for 5 min, and then it was centrifuged at 5000 rpm for 5 min and the concentrations of benzalkonium homologues were determined in the supernatants by LC/(ESI-IT)MS, as specified above. RESULTS AND DISCUSSION Mixed Hemimicelle/Admicelle-Based SPE. Mixtures of surfactants have been known to undergo cooperative self-association to form mixed hemimicelles/admicelles.36 Hydrophobic and electrostatic interactions provide the basis for the ideal and nonideal mixing contributions, respectively, to the formation of the mixed aggregates. The nonideal interactions become progressively stronger in going from mixtures of the same surfactant type to those of opposite charge. Therefore, the formation of mixed aggregates between the anionic surfactant adsorbed on alumina and the benzalkonium surfactant by strong hydrophobic and electrostatic interactions should greatly favor the preconcentration of these cationic analytes. Below, the main factors affecting the SPE of benzalkonium surfactants on SDS-coated alumina are discussed. (a) Adsorption of SDS on Alumina. Figure 1 shows the experimental isotherm obtained for the adsorption of SDS on alumina at pH 2. This isotherm does not exhibit the four typical regions observed for the adsorption of surfactants on mineral oxides, which can be easily differentiated in the adsorption isotherm of SDS on alumina at higher pH values19,20 (Figure 1, inset). The behavior of SDS at pH 2, where there is no clear transition from the Henry’s law region (region I) to the hemimicelar region (region II), can be attributed to the large magnitude of the interaction between the alumina surface and the ionic surfactant headgroup, which masks the hydrophobic interaction that leads to hemimicelle formation. An important change of slope in the adsorption isotherm of SDS on alumina at pH 2 was observed for adsorbed SDS amounts (36) Roberts, B. L.; Scamehorn, J. F.; Harwell, J. H. Phenomena in Mixed Surfactant System; ACS Symposium Series 311; American Chemical Society: Washington, DC, 1988; p 200.
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
6801
Figure 1. Experimental adsorption isotherm for SDS on alumina at pH 2. Inset: typical adsorption isotherm of SDS on alumina at pH 8.19
above 150 mg (Figure 1), which could be related with alumina surface saturation. The cmc of SDS at pH 2 and 25 °C, calculated from surface tension measurements, was found to be 1417 mg/ L. Below this concentration only SDS monomers existed in solution. The concentration of SDS in aqueous solution was insignificant (lower than 5 × 10-3 mg/L) for adsorbed SDS concentrations below 100 mg of SDS/g of alumina (Figure 1). The maximum amount of SDS adsorbed on alumina was a strong function of pH. Thus, at pH 2 it was ∼180 mg of SDS/g of alumina (Figure 1). This amount decreased up to about 140 and 40 mg of SDS/g of alumina at pH 3.9 and 8.8, respectively. As a result, both the total amount of surfactant adsorbed and the ratio of surfactant adsorbed to surfactant in solution increases with decreasing pH, which should favor the adsolubilization of benzalkonium homologues on SDS admicelles. (b) Influence of the Amount of SDS on the Adsolubilization of BAS. In the absence of SDS, benzalkonium surfactants hardly adsorbed on γ-alumina below the pcz (8.5). Percentages of adsorption lower that ∼1% were obtained for all homologues. Above the pcz, the surface of aluminum oxide becomes more negatively charged with increasing pH, and as a result, a higher adsorption density of benzalkonium homologues was observed. Thus, the following adsorption percentages were obtained at pH 9, 14.7% BDDA, 33.4% BDTA, and 83.7% BDHTA, and at pH 11, ()65.1% BDDA, 81.9% BDTA, and 96.9% BDHA. As was expected, adsorption was also found to be dependent on the length of the alkyl chain; hydrophobicity favored the retention of homologues on the alumina surface. Figure 2 shows the dependence of the adsolubilization of benzalkonium homologues on SDS-coated alumina at pH 2 as a function of the amount of anionic surfactant added. Below 1 mg of SDS, which is more clearly seen in the inset to Figure 2, adsorption of cationic surfactants increased with adsorbed SDS concentration on alumina. The adsolubilization started from equimolecular amounts of SDS and benzalkonium surfactants 6802 Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
Figure 2. Effect of the amount of SDS on the sorption of BAS homologues on 1 g of alumina at pH 2: (b) BDDA, ([) BDTA, and (4) BDHA.
(e.g., 0.03 mg of SDS/g of alumina or 1 × 10-7 mol of SDS), and it was quantitative from 3 mg of SDS (SDS/benzalkonium homologue molar ratio above 100). Hydrophobicity was a factor predominant for adsolubilization by using such low amounts of SDS. Adsorption of benzalkonium homologues was found to be independent of the adsorbed amount of SDS in a wide interval (3-200 mg). Greater SDS amounts gradually decreased adsolubilization due to the formation of SDS micelles in the bulk aqueous solution. The cmc of SDS did not change in the presence of benzalkonium surfactants because of the low concentrations tested (∼1 × 10-7 M). Hydrophobicity favored the formation of SDS: benzalkonium mixed micelles, and therefore, the partition of BDHA to the aqueous phase was greater than for the lower alkyl chain homologues (Figure 2). Amounts of SDS between 3 and 100 mg/g alumina are recommended for adsolubilization of benzalkonium surfactants on the basis of both maximum adsorption of cationics and minimal SDS partition to the aqueous phase, which can be important for breakthrough of analytes. We selected 25 mg of SDS for further studies. (c) Influence of pH. The pH did not affect the adsolubilization of benzalkonium surfactants by SDS-γ-alumina between 2 and 11, being quantitative in this interval. The amount of SDS adsorbed, and accordingly the volume fraction of hemimicelles/admicelles produced, decreased slightly up to about pH 8 and then significantly decreased at higher pH values because of the reduction in positive charge of the γ-alumina surface. Two phenomena were thought to be responsible for the adsorption of benzalkonium surfactants on alumina around and above the pcz of the mineral oxide: adsolubilization and direct interaction of the cationics with the increased anionic charge of the alumina surface. Acid pH values are recommended for experimental work because precipitation of environmental water samples is generally observed at basic pHs and breakthrough of analytes should be delayed with decreasing pH because of the strong adsorption of SDS on alumina.
Table 1. Analytical Characteristics of the Method target compd BDDA
BDTA
BDHA
a
operation mode
m/z
slope ( sc
full scana SIMb MS/MS full scan,a internal standard full scan,a internal standard, addition standard on sewage full scana SIMb MS/MS full scan,a internal standard full scan,a internal standard, addition standard on sewage full scana SIMb MS/MS full scan,a internal standard full scan,a internal standard, addition standard on sewage
304 304 212 304/248 304/248
(12.0 ( 0.2) × 106 (11.9 ( 0.2) × 106 (377 ( 6) × 104 0.261 ( 0.005 0.252 ( 0.008
(-13 ( 4) × 105 (3 ( 2) × 105 (2 ( 0.9) × 104 0.019 ( 0.003 0.052 ( 0.007
0.998 0.9993 0.9991 0.9993 0.992
332 332 240 332/248 332/248
(15.1 ( 0.3) × 106 (14.8 ( 0.3) × 106 (12.4 ( 0.4) × 105 0.349 ( 0.009 0.353 ( 0.007
(-15 ( 5) × 105 (7 ( 4) × 105 (-2.1 ( 0.6) × 105 0.021 ( 0.006 0.07 ( 0.03
0.998 0.998 0.996 0.998 0.97
360 360 268 360/248 360/248
(16.1 ( 0.3) × 106 (16.6 ( 0.2) × 106 (33.2 ( 0.9) × 105 0.35 ( 0.02 0.36 ( 0.07
(-11 ( 5) × 105 (-9 ( 2) × 105 (-6 ( 2) × 105 0.05 ( 0.01 0.09 ( 0.08
0.998 0.9991 0.997 0.997 0.98
intercept ( sc
rd
m/z range, 200-400. b m/z range, ion molecular (5. c Standatd deviation. d Correlation coefficient (n ) 9).
(d) Desorption of BAS. Desorption of benzalkonium surfactants from the SDS-coated alumina column was studied using 2 mL of different eluents that should desorb the analytes on the basis of different interaction modes. Sodium hydroxide solutions between 0.01 and 2 M caused increased elution of SDS and BAS in proportion as the concentration of NaOH increased. Complete elution of SDS was observed from 0.1 M NaOH; however, recoveries of BAS below 20% were always obtained due to the strong ionic interactions between the negatively charged alumina surface and the cationic surfactants, which caused their retention in the absence of SDS. Elution of BAS by ammonium formate (0.1 M, pH 3.5) based on an ion exchange mechanism was also unsuccessful, indicating the importance of hydrophobic interactions between BAS and SDS. The most suitable eluents of BAS were organic solvents, which caused disruption of admicelles. Quantitative recovery of BAS was obtained by using methanol, which was selected for their desorption. A fraction of SDS was jointly eluted under these conditions; however, complete desorption of SDS required more methanol (above 20 mL) or 0.1 M NaOH. (e) Sample Loading and Flow Rate. Experiments to determine the sample loading volume were conducted by passing increasing volumes (0.025-1 L) of an aqueous solution at pH 2, containing 200 µg/L BAS, through an admicelle column filled with 500 mg of alumina and 12 mg of SDS. The loading of BAS ranged between 5 and 200 µg. Quantitative recovery of analytes was obtained in the interval investigated. These results indicate the high retention factor of BAS by formation of mixed hemimicelles/ admicelles with SDS and the strong adsorption of SDS to the alumina matrix at pH 2. In this way, preconcentration factors of 500 can be easily obtained by SPE of 1 L of sample and elution with 2 mL of methanol. Higher sample volumes were not investigated. The above experiments were conducted at three sample flow rates (3, 10, and 20 mL/min) in order to determine the effect of this parameter on recovery of BAS. The recovery was quantitative in the interval of flow rates investigated, thus permitting the rapid treatment of samples.
Analytical Performance. (a) Calibration Data. Calibration curves were run for each target compound, in the range 0.2-20 ng, using three operation modes: full-scan, reduced mass range (SIM), and specific MS/MS fragmentation. By using SIM and MS/ MS modes in liquid chromatography/ion trap mass spectrometry, gains in sensitivity are sometimes achieved compared to full-scan operation. The use of SIM shorts one part of the scan program (e.g., the time spent on production of the spectrum), which permits us to acquire a greater number of microscans and therefore to increase the ion current observed. However, only slight gains in sensitivity (about 2-3 times) are generally achieved using this operation mode. On the other hand, for many chemical systems, the use of tandem mass spectrometry enhances the sensitivity in terms of S/N ratio. Although the signal intensity in MS/MS is reduced in the collision-activated dissociation or fragmentation process following the first mass-selective stage, the reduction in noise level can be greater. Table 1 shows the figures of merit for the calibration curves obtained for BAS homologues. Quantifications were carried out in all cases from the corresponding peak areas of extracted ion chromatograms. The correlation coefficients obtained indicated good fits for all operation modes examined. No benefits in sensitivity were achieved by reducing the m/z scan range from 200 (full-scan mode) to 10 (SIM mode). On the other hand, BAS quantification by MS/MS using the m/z of the most abundant fragment ion (see Figure 3) also caused losses of sensitivity by a factor of about 1.5-5 compared to the measurement of parents ions (Table 1). On the basis of these results, we decided to work on full scan using N-dodecylpiridinium chloride as internal standard (Table 1). The instrumental detection limits were calculated from blank determinations by using a signal-to-noise ratio of 3 (the ratio between the peak areas for each BAS homologue and internal standard and peak area of noise). They were estimated to be ∼0.04 ng. From this value and considering the maximal volume of sample recommended for treatment (1 L), the volume of elution (2 mL), the volume of extract injected (20 µL), and the recovery Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
6803
Figure 3. Fragment ions and their relative abundance obtained from MS/MS analysis of BAS homologues using the ion trap at a resonance excitation of 1.1 V.
obtained from spiked samples (∼100%), the detection limit of BAS in environmental water samples was estimated to be ∼4 ng/L. Since detection limits calculated in this way are often optimistic, the practical detection limits37 were also estimated from six independent complete determinations of analyte concentrations in a typical matrix low-level material. In cases where this material could not be obtained, an estimate of the background signal was made at a representative part of the readout, adjacent to the analyte signal, in the analyte-containing sample. The practical detection limits ranged between 3 and 6 ng/L for raw sewage, 4 and 5 ng/L for sewage effluent, and 3-5 ng/L for river water. This high agreement between both types of detection limits could be explained by the low matrix influence in the developed method derived from the high selectivity of both the admicelle extraction and MS detection. By working with an automated SPE system coupled on-line with the LC, which permits the determination of the whole sample, detection limits of 40 pg/L could be achieved. (b) Selectivity Studies. The possible interference of matrix components that could affect the hemimicelle critical concentration (HMC) of SDS and accordingly the adsorption of BAS on the SDScoated alumina column was investigated. Special attention was paid to the effect of surfactants such as alkylbenzenesulfonates (LAS) and nonionic surfactants (nonylphenol ethoxylate, NPE, was selected as a model), which greatly exceed BAS concentration in sewage and can both compete with BAS for adsorption and decrease the HMC of SDS by formation of mixed micelles. Likewise, the effect of electrolytes (e.g., NaCl) on the adsorption of BAS was investigated since they are well known to cause decrease in the HMC of ionic surfactants. No interference was found from LAS and NPE at their concentration level in sewage effluents29 (mg/L). On the other hand, sodium chloride concentrations below 0.2 M did not affect the adsorption of BAS. Higher electrolyte concentrations, however, decreased the recoveries of the cationic surfactants. Thus, recoveries of 73% BDDA, 75% BDTA, and 71% BDHA and 54% BDDA, 55% BDTA, and 55% BDHA were found in the presence of 0.5 and 1 M NaCl, respectively. This decrease was directly (37) Thompson, M.; Ellison, S. L. R.; Wood, R.; Pure Appl. Chem. 2002, 74, 835-855.
6804
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
related to the decrease of the SDS critical micellar concentration. A cmc value of 85 mg/L was found under the following conditions: 1 M NaCl, pH 2, and 25 °C. Therefore, application of the proposed method to the control of BAS in the marine environment will require the establishment of the SDS adsorption isotherm under the salinity conditions of seawater in order to optimize the anionic surfactant concentration required for quantitative recovery of BAS. The influence of matrix components that could elute with the target compounds causing ion suppression or space-charge effects on the ion trap was assessed by comparison of the calibration curves obtained from standards and those obtained from sewage or river water samples fortified with known amounts of BAS. An internal standard was used for quantitation. The analytical characteristics of both types of calibration curves were similar (see typical data from sewage in Table 1), and therefore, matrix components did not interfere in the recovery of BAS or mass quantitation. The correlation coefficients indicating better fits for the calibration curves obtained in solution compared to those obtained from the wastewater matrix. According to these results, we recommended external calibration for determination of BAS in sewage/river waters. However, it is advisable to check the influence of matrix components if the sample analyzed is expected to have a composition very different from those investigated here. (c) Structural Information. Cationic surfactant identification was carried out though MS/MS spectrometry by using the ion trap instrument. For this purpose, the molecular ion corresponding to each BAS homologue was isolated and fragmented. Figure 3 shows the typical fragments obtained and their relative abundance under the proposed conditions of fragmentation. This relative abundance was similar to that found by using a triple quadrupole MS system or an increased cone voltage in a single quadrupole.32 Since the ion trap mass spectrometer can switch in the same chromatographic peak between full-scan MS and a collisioninduced dissociation (CID) product scan in the presence of helium collision gas, with no loss in signal or CID efficiency, we used this strategy to quantify and identify BAS in sewage/river samples. Analysis of Environmental Water Samples. To check the suitability of the method proposed for the analysis of BAS homologues, it was applied to the determination of these surfactants in river and wastewater samples. Table 2 lists the concentrations found for the target compounds, expressed as the mean value (n ) 3) and the corresponding standard deviation. Total method recoveries were assessed by analyzing spiked samples. As an example, the results obtained from river samples are shown in Table 2. Panels a and b of Figure 4 show respectively the MS extracted ion chromatograms from a standard solution and a wastewater sample. The MS/MS extracted ion chromatogram obtained from a river sample is depicted in Figure 4c. The distribution of BAS homologues found in the environmental samples was in agreement with the composition of commercial formulations. The BAS mole ratios C12/C14/C16 found for different Spanish products were 59:37:4 (softener, trade name Lenor); 57:39:4 (softener, trade name Flor), and 63:33:4 (detergent, trade name Ariel). In environmental samples, the BAS homologue concentrations also held the sequence C12 > C14 > C16 (see Table 2).
Table 2. Mean Concentration (µg/L) ( Standard Deviation of Benzalkonium Homologues in Wastewater and River Samples (Analyzed by SDS Admicelle-Based SPE-LC/ESI+-IT/MS sample location
DBDA m/z 304
TBDA m/z 332
HBDA m/z 360
raw sewage, Baena sewage effluent, Baena raw sewage, Pozoblanco sewage effluent, Pozoblanco river water, Co´rdoba spiked river water, 1µg/L spiked river water, 10 µg/L
49 ( 2 1.1 ( 0.1 4.3 ( 0.1 0.10 ( 0.02 detected 1.1 ( 0.1 9.7 ( 0.5
36 ( 1 detected 0.14 ( 0.03
