Journal of Chromatography A, 889 (2000) 221229 www.elsevier.

com / locate / chroma

Determination of linear alkylbenzenesulfonates in wastewater treatment plants and coastal waters by automated solid-phase extraction followed by capillary electrophoresisUV detection and conrmation by capillary electrophoresismass spectrometry
J. Riu a , P. Eichhorn b , J.A. Guerrero c , Th.P. Knepper b , D. Barcelo a , *
a

Department of Environmental Chemistry, IIQAB-CSIC, c /Jordi Girona 18 26, 08034 Barcelona, Spain b ESWE, Sohnleinstrasse 158, 65201 Wiesbaden, Germany c Department of Chemistry, Universidad Nacional de Colombia, AA 14490, Bogota, Colombia

Abstract Linear alkylbenzenesulfonates (LASs) were determined in wastewaters and coastal waters by solid-phase extraction, using two different sample preparation protocols depending on the sample treated, followed by capillary electrophoresis and ultraviolet detection (CEUV). The linear range of the proposed method varied from 3 to 53 and from 25 to 495 mg / l, depending on the compound, with a limit of detection of 1 mg / l when 250 ml of coastal water was preconcentrated. [M2H] 2 ions were used for CEMS conrmation after quantication by CEUV. CEMS diagnostic ions were the same ones used in LCelectrospray (ESI) MS and corresponded to m /z 297, 311, 325 and 339 for C 10 , C 11 , C 12 and C 13 LASs, respectively. LASs were determined in wastewater samples of the inuent and efuent of three wastewater treatment plants (WWTPs), two of them using biological treatment with secondary settlement and receiving mainly domestic wastewaters whereas one of the plants was operated with physicochemical treatment and received mainly industrial wastewaters. LASs were also analyzed in two samples from coastal waters of the bay of Cadiz (Spain) receiving untreated domestic efuents. All samples were also analyzed by LCESI-MS and the results are compared with the CEUV method developed in this work. The concentration levels of total LASs varied from 988 to 1309 mg / l in the inuents of WWTPs, whereas in the efuents the concentrations varied from 136 to 197 mg / l. The levels of LASs in coastal wastewaters of the bay of Cadiz varied from 739 to 911 mg / l, indicating that the wastewaters discharged into the bay did not undergo any treatment at all. 2000 Elsevier Science B.V. All rights reserved. Keywords: Water analysis; Environmental analysis; Surfactants; Alkylbenzenesulfonates, linear

1. Introduction Linear alkylbenzenesulfonates (LASs) are the most commonly used anionic surfactants. From the 2.8?10 9 kg of surfactants produced in Europe in
*Corresponding author. Tel.: 134-93-4006-118; fax: 134-932045-904. E-mail address: [email protected] (D. Barcelo).

1998, almost 0.42?10 9 kg correspond to LASs [1]. Commercially available LASs are mixtures of secondary isomers, with alkyl chain lengths of 1013 carbon atoms. The formula of the commercial product is shown in Fig. 1. It corresponds to a mixture of homologues, most with chain lengths varying between 10 and 13 carbon atoms. Each of these homologues consists of a varying number of positional isomers.

0021-9673 / 00 / $ see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0021-9673( 00 )00400-3

222

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229

Fig. 1. General chemical structure of linear alkylbenzenesulfonates.

After use, LASs are discharged into domestic or industrial wastewaters. Routine determination of LASs in surface waters uses solid-phase extraction (SPE) followed by time-consuming and tedious derivatization prior to gas chromatographymass spectrometry (GCMS) [2,3]. Continuous ow fast atom bombardment MS was also developed and permitted the direct determination of LASs in wastewater and river samples [4]. Capillary electrophoresis (CE) is a good technique for the analysis of ionic compounds. Because of its high separation efciency and low solvents costs, CE is appropriate for the analysis of many dyes and related compounds employed in the dye industry [5]. This efciency of the CE technique can be additionally enhanced by an automated SPE including simultaneous sample enrichment and clean-up with elimination of interferences from wastewaters. An automated SPE method involving the ASPEC XL system, followed by CEUV and CEMS, was developed by our group for the determination of sulfonated azo dyes [6]. CE has been increasingly used for the analysis of LASs in environmental matrices in recent years [7 10]. For the separation of LAS homologues an organic modier, like acetonitrile or g-cyclodextrin is added to a phosphate or borate buffer [7,8]. Without adding organic solvents all LAS homologues and isomers produce only one peak in the electropherogram. This method is useful for the analysis of total LASs [9]. For the determination of isomeric compounds, CE has been suggested as one of the best methods available. Normally additives like sodium dodecyl sulfate (SDS) [7] or a-cyclodextrins [8] are used with a phosphate or borate buffer and acetonitrile as organic modier, although some authors found a better resolution of some

isomeric LASs using capillary zone electrophoresis (CZE) with no SDS present in the buffer [10]. CEUV is the preferred method for analysis of LASs. In environmental analysis the use of MS detection has been demonstrated to be an indispensable tool because it can overcome the large number of interfering substances and thus can avoid overestimation as compared to less sensitive detectors like UV. In CEMS additives like SDS cannot be used because of sensitivity problems. Moreover, the separation and determination of isomers in environmental water samples with CE is difcult although it is not essential to know LAS isomeric distribution in wastewater samples for risk assessment. Taking into account these facts the objectives of this work were: (1) to develop a methodology for the trace determination of the four commonly used LAS homologues by SPE, using two different protocols depending on the sample treated (coastal waters receiving untreated domestic efuents from the bay of Cadiz or industrial wastewaters from inuent and efuent of water treatment plants from Catalonia, Spain), followed by CEUV. (2) To compare the former methodology with the reference liquid chromatographyelectrospray mass spectrometry (LC ESI-MS) methodology. (3) To use CEMS for conrmation purposes of CEUV environmental traces.

2. Experimental

2.1. Study area of the bay of Cadiz and sample pretreatment

The study area and sample pretreatment are the same as those described in a previous paper [11]. Coastal marine water samples receiving untreated wastewater discharges from the bay of Cadiz, in the southwest of Spain, were analyzed with the method developed in this work. Water samples (250 ml) were acidied to pH 3, then puried, and concentrated by SPE using a C 18 hydrophobic-type mini-column and eluted with methanol onto a SAX strong anionic exchanger. The second elution was carried out with 3 ml of 2 M HCl in methanol. The eluate was evaporated until dryness and was re-dissolved with 1

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229 Table 1 Recoveries and standard deviations (n56) for LASs using twostage SPE (C 18 and SAX)a Compound C 10 C 11 C 12 C 13 LAS LAS LAS LAS Average recovery (%) 94 96 98 94 SD (%) 2 1 2 1

223

a A 200-ml volume of water spiked with a nal concentration of 50 mg / l (LAS mixture) was preconcentrated.

duced, which took about 30 min by using a Baker Spe 12G combined with a vacuum system (15 p.s.i. negative pressure; 1 p.s.i.56894.76 Pa). The target compounds were eluted with two aliquots of 5 ml of methanol with a waiting time of 5 min between the two elution steps. The eluate was evaporated under gentle stream of nitrogen at 258C. The samples were reconstituted in 1 ml water. Table 2 shows the recoveries of the different LASs obtained with this SPE method.

2.3. CEUV conditions

ml of water. Table 1 shows the recoveries obtained with this method. CE was carried out with a Beckman P/ACE 5000 CE system (Beckman Instruments, Palo Alto, CA, USA). The UV detector operated at a wavelength of 214 nm. The separation was performed in a 47 cm (40 cm effective length to detector)375 mm I.D. fused-silica capillary (Beckman Instruments). The electrophoresis buffer solution was 50 mM ammonium acetate in water at pH 5.6 and with an organic modier of 30% isopropanol. The capillary was regenerated subsequently with 0.1 M NaOH (15 min), water (5 min), and working buffer solution (10 min) before each analysis. The temperature of the capillary column was set at 258C. After a pressure injection (0.5 p.s.i.) over 5 s, a voltage of 20 kV was applied across the capillary. Data analysis was performed by System Gold software. The pH of the electrolyte and the spiked water samples was adjusted by adding ammonia and acetic acid, respectively and it was measured with a Model 691 pH meter (Metrohm, Herisau, Switzerland) connected with pH glass electrode containing 3 M KCl and silver chloride. In order to optimize the separation several parameters were tested. Different
Table 2 Mean recoveries and standard deviations (SDs), for LASs using Isolute (n54)a Compound C 10 C 11 C 12 C 13 LAS LAS LAS LAS Mean recovery (%) 78 93 82 77 SD (%) 3 4 3 3

2.2. Treatment plants and sample preparation

Sample collection took place at three wastewater treatment plants (WWTPs). Two of the plants carry out biological treatment (WWT1D and WWT2D) which receive domestic efuents. The third one (WWT3I), consisted of primary settlement involving only physicochemical treatment, which received industrial efuents from various types of industries the major discharges (around 60%). Water samples from the inuent and efuent were collected corresponding to 24-h composite samples. The 24-h composite samples were obtained by the treatment plant operators. All water samples were collected in glass bottles. Samples were transported to the laboratory and were stored at 48C prior to the analysis that took place within 12 days. Before analysis, the pH values of the samples were adjusted to the neutral range and samples were ltered when necessary. Disposable 6-ml cartridges packed with 200 mg Isolute ENV1 (International Sorbent Technology, UK) were attached to the ASPEC XL system (Gilson, Villiers-le-Bel, France) which is tted with an external Model 306 LC pump and connected with a Model 817 switching valve for the selection of samples. The SPE cartridges were conditioned with 7 ml methanol followed by 3 ml water (pH 3) with a ow-rate of 1 ml / min. After the spiked groundwater samples (200 ml) were dispensed with a ow-rate of 5 ml / min through the cartridge columns, a clean-up step with 1 ml water (1 ml / min, pH 7) was performed. Subsequently a drying step was intro-

a A 200-ml volume of water spiked with a nal concentration of 50 mg / l (LAS mixture) was preconcentrated.

224

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229

percentages of organic modier were used (20, 30 and 40%) with different buffer concentrations (ammonium acetate, 10, 25, 50 and 80 mM). Finally, several pH values from 5 to 9 were tested. Acetonitrile instead of isopropanol was also used as organic modier (20, 30 and 40%) with ammonium acetate in water as buffer system. The most signicant advantage of acetonitrile is the reduction of the analysis time (analytes could be analyzed in less than 10 min) but resolution was not as good as with isopropanol and, in some real sample analysis, there were some interferences coeluting with the target compounds. Due to the fact that the number of samples in this study was not high, the separation with isopropanol was chosen.

2.4. CEMS conditions

For CEMS operation the voltage applied was 20 kV and the capillary length was 80 to 100 cm in order to extend it to the probe tip through the stainless steel sheath capillary. The CE system was connected to a VG Platform mass spectrometer from Micromass (Manchester, UK) equipped with a CE probe and an ESI interface. The design of this probe consists of a triaxial ow arrangement whereby the CE eluent is mixed with a suitable make-up solvent at the probe tip, and then nebulized using nitrogen gas. The CE capillary extends fully to the probe tip through the stainless steel sheath capillary, which carries the make-up solvent. Around the sheath capillary is the nebulizer capillary through which the nitrogen gas ows to the probe tip. The make-up solvent performs two functions: to supplement the CE ow to a level adequate for electrospray operation (the CE electroosmotic ow is in the range of 10 nl / min and is too low for ESI operation without make up solvent) and to make electrical contact between the CE buffer and the probe tip. The nitrogen gas that ows through the probe tip maximizes the efciency of the nebulization. The design of the source is not different from the system used in the normal megaow ESI operation which has been previously described by our group [12]. The make-up solvent, consisting on isopropanol water (80:20) with 0.1% of ammonia, was delivered at a ow-rate of 10 ml / min by a gradient system

used in isocratic condition from a Waters 616 pump controlled by a Waters 600S Controller from Waters Millipore (Milford, MA, USA). The MS instrument was tuned by lling the capillary with the studied compounds and monitoring the signal corresponding to the mass of the tuning ion while the voltage of the CE was applied to introduce the sample into the MS system. The operating parameters were adjusted in order to achieve maximum sensitivity (with the consequent loss of fragmentation and structural information). In this study a nebulizer gas of 2530 l / h was used and the drying gas was set at a low value (in the order of 50 l / h or less). The source temperature was set at 758C. The cone voltage was set at 20 V in order not to produce fragmentation and achieve the best sensitivity. The instrument control and data processing utilities included the use of the MassLynx application software installed on a Digital DEC personal computer 466. The operational parameters of the LCESI-MS method are explained in a former paper [13].

3. Results and discussion

3.1. Calibration graphs

The linearity of the system with UV detection was studied by external calibration carried out by quantication of spiked groundwater with different concentrations of the LASs in the range from 3 to 500 mg / l depending on the compound. This was followed by SPE preconcentration. Calibrations were performed by a ve-point calibration curve over the concentration range. The linear regression equations with slopes and correlation coefcients are summarized in Table 3. Repeatability varied between 2 and 5% for an average of six injections in the same day and reproducibility between 6 and 12% for one injection in six different days. The relationship between the concentration of each compound and its peak area was found to be linear, as indicated by correlation coefcients between 0.9807 and 0.9952. The minimum detectable amount (limit of detection, LOD) was about 1 mg / l at a signal-to-noise ratio of 3 for the four LASs.

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229 Table 3 Calibration data obtained with CEUV at 214 nm Compound C 10 C 11 C 12 C 13 LAS LAS LAS LAS Calibration equation y50.000282x20.000737 y50.000255x20.00517 y50.000245x20.00592 y50.000232x20.00531 r2 0.981 0.995 0.992 0.984

225

Linear range (mg / l) 353 25495 23469 15306

Table 4 Concentrations of LASs found in several coastal water samples from the bay of Cadiz and inuent (Inf.) and efuent (Eff.) wastewaters of several wastewater treatment plants by CEUV Concentration (mg / l) C 10 LAS WWT1D Inf. WWT1D Eff. WWT2D Inf. WWT2D Eff. WWT3I Eff. Cadiz 1 Cadiz 2 271 12 275 31 9 162 188 C 11 LAS 721 57 839 134 92 622 663 C 12 LAS 359 99 406 102 69 321 291 C 13 LAS 139 53 110 41 58 75 146

Table 5 Concentrations of LASs found in several coastal water samples from the bay of Cadiz and inuent (Inf.) and efuent (Eff.) wastewaters of several wastewater treatment plants by LCESI-MS Concentration (mg / l) C 10 LAS WWT1D Inf. WWT1D Eff. WWT2D Inf. WWT2D Eff. WWT3I Eff. Cadiz 1 Cadiz 2 194 41 121 47 35 118 118 C 11 LAS 536 66 399 75 87 347 310 C 12 LAS 385 25 319 73 38 306 232 C 13 LAS 194 4 149 31 6 140 79

3.2. Environmental samples

The concentrations obtained for the samples from the inuent and efuents of the wastewater treatment plants and the bay of Cadiz are summarized in Table 4. These samples were also determined by LCESIMS in order to compare the CEUV method with the previously reported one and the results are shown in Table 5. From the data reported in Tables 4, 5 and 6 we can conclude that although the samples analyzed were the same with both systems, the concentrations do not fully agree. Differences of 27% average were found. There are two explanations for such disagree-

Table 6 Comparison of concentrations of total LASs (by combining the data of Tables 4 and 5) found in several coastal water samples from the bay of Cadiz and inuent (Inf.) and efuent (Eff.) wastewaters of several wastewater treatment plants by LCESIMS and CEUV Concentration (mg / l) CEUV WWT1D Inf. WWT1D Eff. WWT2D Inf. WWT2D Eff. WWT3I Eff. Cadiz 1 Cadiz 2 1490 221 1630 308 228 1180 1288 LCMS 1309 136 988 197 167 911 739 9 34 35 31 22 18 38 Discrepancy (%)

226

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229

ment. First, the cases of overestimation that CEUV shows can be attributed to coeluting substances, whereas LCMS is much more selective. However, for individual LASs, some samples show lower concentration values in CEUV than in LCESIMS, so in that case it cannot be attributed to coelution. In this case the explanation can be found in the existence of positional isomers of each one of the four individual LASs which are a complex mixture of 20 possible isomers. If not all the isomers of each individual LAS homologue coelute in one peak, this can affect on the quantication and can be an explanation of the observed disagreement. In MS detection, all the isomers of each homologue are quantied together because they have the same m /z value. However, when comparing total LAS concentrations with both methods then this problem does not exist. This hypothesis can be conrmed by data in Table 6, where the comparison between total

LASs determined with two methods gives overestimation of the CEUV method in all cases. More agreement was also found in some samples as compared to other ones. This can be attributed to the instability of the LCESI-MS system, that can lead to reproducibilities of 1520%. In some cases the signal has much more uctuations than with CEUV, and consequently, will affect quantication. Fig. 2 shows an electropherogram of the analysis of the inuent and efuent of WWT2D by CEUV.

3.3. Conrmation with CEMS

In the previous section, the disagreement between the CE and the LC data has shown the need for the MS detection with the CE methodology. For this reason the conrmation with MS for the samples analyzed by CEUV has been developed. The parameters of the mass spectrometer were optimized in order to get the maximum sensitivity. By changing the cone voltage it is possible to induce fragmentation in the source region, and consequently, to obtain structural information (but at the expense of sensitivity). Due to the low amounts injected in CE (nl) the sensitivity is a serious drawback of this technique. This is the reason why we needed to use low cone voltages (20 V) in order to avoid fragmentation and achieve the highest detection limits. All the compounds studied were detected as anions and the negative mode of ionization in time-scheduled selected ion monitoring (SIM) conditions was used for all of them. The ions employed to detect the compounds by SIM conditions were the deprotonated molecular ions, which give the best signal-to-noise ratio for each compound and corresponded to m /z 297 for C 10 LAS, 311 for C 11 LAS, 325 for C 12 LAS and 339 for C 13 LAS. Fig. 3 shows an electropherogram of a water extract from the inuent and the efuent of WWT2D (Ripoll, Girona, Spain). In Fig. 4, the selected ion electropherograms of the inuent of the same sample of Fig. 2 are shown. No positional isomers are apparent in the CEMS chromatographic trace, but this could be attributed to the lower sensitivity of CEMS versus CEUV. So, we cannot discount the existence of coeluting positional isomer peaks in the CEUV analysis being a problem for the quantitative analysis of LASs.

Fig. 2. CEUV (214 nm) electropherogram of a water extract from an efuent (a) and inuent (b) of WWT2D (Ripoll, Girona, Spain).

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229 Fig. 3. CEESI-MS electropherogram of water extract from the inuent and the efuent of WWT2D (Ripoll, Girona, Spain). Ions monitored corresponded to m /z 297 for C 10 LAS, 311 for C 11 LAS, 325 for C 12 LAS and 339 for C 13 LAS. x-Axis: migration time in min. 227

228

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229

Fig. 4. Selected ion electropherograms of an extracted sample from the efuent of WWT2D (Ripoll, Girona, Spain), where no isomer peaks can be observed. Ions monitored corresponded to m /z 297 for C 10 LAS, 311 for C 11 LAS, 325 for C 12 LAS and 339 for C 13 LAS. x-Axis: migration time in min.

4. Conclusions Overall, we have shown that the present method reported here allowed the determination of LASs in wastewaters and coastal waters by using CEUV. By the use of CEMS it was feasible to conrm the presence of the different analytes. It was not a problem to determine LASs by CEUV since the LODs of the CE system are in the 1 ppb range and

most of the samples had much higher values. CE UV and CEMS, having much poorer detection limits than LCESI-MS, are useful techniques in wastewater monitoring. They offer good selectivity and they permit the quantication of the different LASs. The use of CEUV and CEMS for other types of environmental matrices, such as groundwater or surface water samples, is more complicated because sensitivity is poorer as compared to LC

J. Riu et al. / J. Chromatogr. A 889 (2000) 221 229

229

ESI-MS. For the analysis of wastewater, the samples are much more concentrated with organic pollutants, so sensitivity is not a problem. In addition, CE offers good selectivity and low operating costs, which offer potential for environmental trace analysis. The need for MS detection in environmental applications for CE can be derived from the disagreement observed between CEUV and LCESI-MS data. So further work to improve the stability of CEMS is needed. Future work is planned in order to quantify LASs by CEMS, so it will improve the quantitative data obtained by CEUV in environmental samples. In the present work only qualitative data for conrmation could be obtained.

References
[1] Anaerobic Biodegradation of Surfactants, AISE-CEFIC, February 1999, Brussels, Belgium, pp. 175. [2] J.A. Field, Th.M. Field, Th. Ploguer, W. Giger, Environ. Sci. Technol. 28 (1994) 497. [3] J.A. Field, D.J. Miller, Th.M. Field, S.B. Hawthorne, W. Giger, Anal. Chem. 64 (1992) 3161. [4] A.J. Borgerding, R. Hites, Anal. Chem. 64 (1992) 1449. [5] H. Siren, R. Sulkava, J. Chromatogr. A 717 (1995) 149. [6] J. Riu, I. Schonsee, D. Barcelo, J. Mass Spectrom. 33 (1998) 653. [7] K. Heining, C. Vogt, G. Werner, Analyst 123 (1998) 349. [8] K. Heining, C. Vogt, G. Werner, J. Chromatogr. A 745 (1996) 281. [9] C. Vogt, K. Heining, B. Langer, J. Mattusch, G. Werner, Fresenius J. Anal. Chem. 352 (1995) 508. ` ` [10] P.L. Desbene, C. Rony, B. Desmazieres, J.C. Jacquier, J. Chromatogr. 608 (1992) 375. [11] E. Gonzalez-Mazo, M. Honing, D. Barcelo, A. Gomez-Parra, Environ. Sci. Technol. 59 (1997) 77. [12] C. Molina, M. Honing, D. Barcelo, Anal. Chem. 66 (1994) 4444. [13] J. Riu, E. Gonzalez-Mazo, A. Gomez-Parra, D. Barcelo, Chromatographia 50 (1999) 275.

Acknowledgements This work has been supported by the Environmental and Climate Program of the European Commission PRISTINE (Contract No. ENV4-CT97-494) from Wastewater Cluster and CICYT (AMB99-0167CE). We thank International Sorbent Technology for providing SPE cartridges.