Cloud Point Extraction with Surfactant Derivatization as an Enrichment

Anal. Chem. 2009, 81, 7113–7122

Cloud Point Extraction with Surfactant Derivatization as an Enrichment Step Prior to Gas Chromatographic or Gas Chromatography-Mass Spectrometric Analysis Yoshitaka Takagai*,†,‡ and Willie L. Hinze*,† Department of Chemistry, Wake Forest University, P.O. Box 7486, Winston-Salem, North Carolina 27109, and Cluster of Science and Technology, Faculty of Symbiotic Systems Science, Fukushima University, Kanayagawa 1, Fukushima 960-1296, Japan Cloud point extraction (CPE) using Triton X-114 was successfully applied as an extractive preconcentration step prior to gas chromatographic-mass spectrometric analysis. No liquid chromatographic or back-extraction steps were required to remove the target analyte(s) from the surfactant-rich extractant phase. Instead a postextraction derivatization step is employed in which the surfactant of the surfactant-rich phase is reacted with N,Obis(trimethylsilyl)trifluoroacetamide (BSTFA) prior to injection into the gas chromatograph. Such derivatization of the Triton X-114 surfactant following CPE was found to provide improved chromatographic performance yielding a reasonable elution time window that is free of derivatized surfactant signals, reproducible analyte retention times, and quantitative results. Mixtures of polycyclic aromatic hydrocarbons (PAHs), herbicides, and profens were utilized to demonstrate the feasibility and performance of this approach. The retention times of six PAHs (acenaphthene, acenaphthylene, anthracene, biphenyl, dibenzofuran, and fluorene) were found to be very reproducible with relative standard deviations (RSDs) in the range of 0.5-0.8%. Quantitative gas chromatographymass spectrometry (GC/MS) analysis of a herbicide test mixture (composed of alachlor, atrazine, butachlor, hexachlorocyclopentadiene, metolachlor, and simazine) following their CPE from spiked water samples yielded detection limits in the range of 6.6-97 ng/L (except for that of hexachlorocyclopentadiene which was 482 ng/L). The enrichment factors achieved for these herbicides ranged from 17 to 33. The recovery of the herbicides from spiked water samples ranged from 90 to 100% except for simazine and atrazine which were 50% and 74%, respectively. The BSFTA derivatization step can serve not only to derivatize the surfactant but also appropriate nonvolatile (or less volatile) analytes. An ibuprofen and flurbiprofen test mix was utilized to demonstrate this feature. The proposed protocol offers an attractive alternative means by which surfactant-mediated extractions can be * To whom correspondence should be addressed. E-mail: takagai@ sss.fukushima-u.ac.jp (Y.T.), [email protected] (W.L.H.). † Wake Forest University. ‡ Fukushima University. 10.1021/ac9009963 CCC: $40.75  2009 American Chemical Society Published on Web 07/21/2009

utilized as an enrichment step prior to gas chromatographic or gas chromatographic-mass spectrometric analysis of analytes which should serve to expand the scope of CPE in gas chromatographic (GC) analysis. Nonionic surfactant micelle solutions exist as a single homogeneous isotropic phase at temperatures below their cloud point. However, if the solution temperature is raised above the cloud point, such solutions become turbid and phase separate to yield a surfactant-lean phase (typically dubbed as the aqueous phase) and small volume surfactant-rich (coacervate) phase. Target analyte species often differentially partition between these two phases.1 Extractions based on such phenomenon are referred to as cloud point extractions (CPEs). A plethora of reports have concerned the utilization of such CPE approach as a means by which to extract and enrich inorganic, organic, and biological analytes prior to their analysis by spectroscopic, capillary electrophoretic or liquid chromatographic means.2-13 In contrast, only a relatively few articles have reported the use of CPE prior to gas chromatographic (GC) or gas chromatography-mass spectrometry (GC/MS) analysis. The primary reason for this stems from the fact that direct introduction of the surfactant-rich extractant phase into a GC system causes (1) Watanabe, H.; Tanaka, H. Talanta 1978, 25, 585–589. (2) Hinze, W. L.; Pramauro, E. A. Crit. Rev. Anal. Chem. 1993, 24, 133–177. (3) Cordero, B. M.; Pavon, J. L. P.; Pinto, C. G.; Laespada, M. E. F. Talanta 1993, 40, 1703–1710. (4) Tani, H.; Kamidate, T.; Watanabe, H. J. Chromatogr. A 1997, 780, 229– 241. (5) Quina, F. H.; Hinze, W. L. Ind. Eng. Chem. Res. 1999, 38, 4150–4168. (6) Carabias-Martinez, R.; Rodriguez-Gonzalo, E.; Moreno-Cordero, B.; PerezPavon, J. L.; Garcia-Pinto, C.; Laespada, E. F. J. Chromatogr. 2000, 902, 251–265. (7) Halko, R.; Hutta, M. Chemicke Listy 2000, 94, 990–993. (8) Stalikas, C. D. Trends Anal. Chem. 2002, 21, 343–355. (9) Paleologos, E. K.; Giokas, D. L.; Karayannis, M. I. Trends Anal. Chem 2005, 24, 426–436. (10) Bezerra, M. D. A.; Arruda, M. A. Z.; Ferreira, S. L. C. Appl. Spectrosc. Rev. 2005, 40, 269–299. (11) Silva, M. F.; Cerutti, E. S.; Martinez, L. D. Microchim. Acta 2006, 155, 349–364. (12) Shen, J. C.; Shao, X. G. Prog. Chem. 2006, 18, 482–487. (13) Kushchevskaya, N. F.; Gorbachevskii, A. N.; Doroshchuk, V. A.; Kulichenko, S. A. J. Water Chem. Technol. 2008, 30, 296–308.

Analytical Chemistry, Vol. 81, No. 16, August 15, 2009

7113

difficulties.6,14-23 Namely, the surfactant can (i) adsorb onto (coat) the stationary phase and alter its polarity which causes changing non-reproducible analyte retention times with subsequent injections and/or (ii) itself elute as a series of peaks over a period of time from the column such that it overlaps with and obscures the analyte peak(s) of interest. In addition, it was thought that the introduction of the surfactant-rich extractant phase could “clog or block” the GC column. Several approaches have been employed to circumvent such difficulties. The first involves the use of mini-column liquid chromatography to separate and recover the target analyte(s) from the surfactant-rich extractant phase. Cation exchangers or silica gel and Florisil stationary phases have been employed in conjunction with aqueous methanolic, methanol-hexane or hexane mobile phases for this purpose. The analyte-containing eluate obtained is typically dried by a stream of dry nitrogen gas and then dissolved in a suitable solvent (or its volume reduced by rotary evaporation) prior to introduction into the GC system. Via use of such protocol, CPE has been successfully employed as an extractive preconcentration step for the GC determination of polychlorinated biphenyls in water, phenothiazine in human serum, and disulfoton in water.15-17 The second approach involves conventional liquid-liquid extraction of the target analyte(s) from the surfactant-rich extractant phase using a water-immiscible organic solvent. Following the CPE step, the analyte(s) are back-extracted from the surfactant-rich phase into organic solvents in which the surfactant has a very low solubility. Microwave or ultrasound irradiation is sometimes also employed to assist and speed up this backextraction process. The resulting organic extract is then directly injected into the GC system; or in some instances, after volume reduction and/or supplemental cleanup via an additional centrifugation step.18-23 It should be cautioned that this approach does not always result in complete elimination of the surfactant, and surfactant peaks often still appear in the gas chromatograms.19,20,22 Polycyclic aromatic hydrocarbons, plasticizers, tobacco alkaloids, organophosphorous pesticides, 1,4-dicholorbenzene, and sunscreen product residues are among the analytes that have been successfully preconcentrated by CPE prior to GC or GC/MS analysis using this approach. For highly volatile analytes, the surfactant-rich extractant phase obtained after the CPE step can be placed in small sealed vials with subsequent static headspace sampling employed as the (14) Hinze, W. L.; Singh, H. N.; Fu, Z. S.; Williams, R. W.; Kippenberger, D. J.; Morris, M. D.; Sadek, F. S. In Chemical Analysis of Polycyclic Aromatic Compounds; Vo-Dinh, T., Ed.; John Wiley & Sons: New York, 1989; Chapter 5, pp 151-169. (15) Froschl, B.; Stangl, G.; Niessner, R. Fresenius’ J. Anal. Chem. 1997, 357, 743–746. (16) Ohashi, A.; Ogiwara, M.; Ikeda, R.; Okada, H.; Ohashi, K. Anal. Sci. 2004, 20, 1353–1357. (17) Faria, A. M.; Dardengo, R. P.; Lima, C. F.; Neves, A. A.; Queiroz, M. E. L. R. Int. J. Environ. Anal. Chem. 2007, 87, 249–258. (18) Zygoura, P. D.; Paleologos, E. K.; Riganakos, K. A.; Kontominas, M. G. J. Chromatogr., A 2005, 1093, 29–35. (19) Giokas, D. L.; Sakkas, V. A.; Albanis, T. A.; Lampropoulou, D. A. J. Chromatogr., A 2005, 1077, 19–27. (20) Sikalos, T. I.; Paleologos, E. K. Anal. Chem. 2005, 77, 2544–2549. (21) Paleologos, E. K.; Giannakopoulos, S. S.; Zygoura, P. D.; Kontominas, M. G. J. Agric. Food Chem. 2006, 54, 5236–5240. (22) Shen, J.; Shao, X. Anal. Chim. Acta 2006, 561, 83–87. (23) Jia, G. F.; Lv, C. G.; Zhu, W. T.; Qiu, J.; Wang, X. Q.; Zhou, Z. Q. J. Hazard. Mater. 2007, 159, 300–305.

7114


injection technique.24-27 In such static headspace GC analysis, only the volatized analyte species are introduced into the GC column. This approach has been employed for determination of the concentrations of species such as benzene, toluene, ethylbenzene, and chlorinated ethanes in the surfactant-rich phase, typically in the context of the design of CPE systems for the removal of volatile organics from wastewaters.24-27 In this note, an alternative approach for GC or GC/MS analysis of analytes present in the surfactant-rich extractant phase following CPE is proposed. A post-extraction derivatization step is employed in which the surfactant in the surfactant-rich extractant phase is derivatized with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) prior to introduction into the GC. Such derivatization step improved the chromatographic performance yielding a fairly wide elution time window absent of surfactant peaks, reproducible analyte retention times, and quantitative results. This approach enables the direct use of the surfactant-rich phase following CPE without any subsequent laborious column chromatographic or back-extraction analyte isolation procedures. It should prove to be an attractive alternative approach for the GC or GC/MS analysis of analytes following their preconcentration by CPE in many situations. EXPERIMENTAL SECTION Apparatus. Samples were analyzed by gas chromatographymass spectrometry using an Agilent 6850 series gas chromatograph (Agilent Technologies, Palo Alto, CA) linked to an Agilent 5973N quadrupole mass spectrometer operating in the positive ion electron impact ionization mode. A DB-5 ms fused-silica capillary GC column (30 m × 0.25 mm i.d. with a film thickness of 0.25 µm) coated with 5% phenyl, 95% methylpolysiloxane (Agilent J&W, Folsom, CA) was employed. The carrier gas was helium (carrier gas pressure 75 kPa). Samples were introduced into the GC system by manual injection of 1.0 µL volume in the split mode (split ratio 50:50). In the typical GC/MS run, both the injector and the detector temperatures were set at 320 °C with a He carrier gas flow rate of 1.00 mL/min. It is important to note that the injector inlet temperature needs to be in the range of 300 to 325 °C to prevent sorption of the Triton surfactants to the inlet liner. The column temperature programmed run was started at an initial oven temperature of 80 °C (2.0 min hold time) followed by a temperature ramp of 5 °C/min up to a final temperature of 295 °C (with 35.0 min hold), resulting in a total analysis run time of 80 min. At the conclusion of each run, the oven temperature was increased to 320 °C for 2.0 min at an increased He carrier gas flow rate of 2.61 mL/min as a column cleanup step. A mass range of m/z 50 to 800 was scanned in each run. The GC/MS data were processed using Enhanced ChemStation Version C Software (Agilent Technologies). A second GC/MS system was used for the quantitative determination of herbicides. Namely, an Agilent 7890A gas chromatograph with an Agilent 5975c quadrupole mass spectrom(24) Trakultamupatan, P.; Scamehorn, J. F.; Osuwan, S. Sep. Sci. Technol. 2002, 37, 1291–1305. (25) Sakulwongyai, S.; Trakultamupatam, P.; Scamehorn, J. F.; Osuwan, S.; Christian, S. D. Langmuir 2000, 16, 8226–8230. (26) Weschayanwiwat, P.; Kunanupap, O.; Scamehorn, J. F. Chemosphere 2008, 72, 1043–1048. (27) Kungsanant, S.; Kitiyanan, B.; Rirksomboon, T.; Osuwan, S.; Scamehorn, J. F. Sep. Purif. Technol. 2008, 63, 370–378.

eter and an Agilent 7863B series auto sample injector system was employed. The GC column, temperature program, and other parameters were the same as previously outlined except that the final column temperature attained was 300 °C (rather than 295 °C), the atom ionization unit temperature was 250 °C, and the sample was injected by an autosampler in the split mode (split ratio of 50:1). In addition, the mass selective detector was operated in the selective ion monitoring mode (rather than total ion monitoring mode) for the herbicide quantification. A Precision micro-semi micro Centricone centrifuge (Precision Scientific Co., Chicago) was employed for the CPE centrifugation step to facilitate phase separation and isolation of the surfactantrich phase. Reagents. Triton X-100 (polyethylene glycol mono[4-(1,1,3,3tetramethylbutyl)-phenyl] ether with an average ethylene oxide (EO) chain length of n ) 9.5; TX-100), Triton X-114 (polyethylene glycol mono[4-(1,1,3,3-tetramethylbutyl)-phenyl] ether with an average ethylene oxide (EO) chain length of n ) 7.5; TX-114) and Triton X-114 reduced (hydrogenated form of Triton X-114, TX-114R) were purchased from Sigma-Aldrich (St. Louis, MO) and used without further purification. The N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) and 1-(trimethylsilyl)-imidazole (TMSI) silylating reagents were obtained from Aldrich Chemical Co. (Milwaukee, WI) and utilized without further purification. HPLC grade acetonitrile was obtained from Fisher Scientific Co. (Raleigh, NC). All other organic solvents (analytical grade materials) were purchased from Sigma-Aldrich and used without further purification. A certified pH 2.00 (0.05 M KCl-HCl) buffer solution was obtained from Fisher Scientific Co. In-house distilled water was further purified using a Milli-Q Reagent Water System (Millipore Corp., Milford, MA). Analytical grade biphenyl, acenaphthene, acenaphthylene, anthracene, dibenzofuran, flurbiprofen, fluorene, and ibuprofen were all purchased from Sigma-Aldrich. Concentrated 1000.0 mg/L stock solutions of these aromatic compounds were prepared using acetone as solvent with the more dilute working solutions obtained by appropriate dilution of these stock solutions using water (in preparation of standards for the CPE experiments) or acetone (in preparation of standards for GC/MS analysis). An EPA herbicide solution (EPA 508.1 mixture containing the six herbicides alachlor, atrazine, butachlor, hexachlorocyclopentadiene, metolachlor, and simazine all present at a concentration level of 1000 mg/L in acetone) was purchased from Supelco (Bellefonte, PA). Working herbicide solutions were prepared by appropriate serial dilution of this EPA mixture using water as the solvent (for CPE experiments) or acetone (standards for GC/MS analysis). Isotopically labeled acenaphthene-d10 (analytical grade, Sigma-Aldrich) was employed as the internal standard for the quantitative GC/MS analysis of the herbicides. A standard stock solution of 1000 mg/mL acenaphthene-d10 was prepared in pyridine, and this was further diluted with pyridine as solvent to prepare working solutions. The silylation reagents and all of the prepared analyte standard and working solutions were stored in the dark and refrigerated at 5 °C. Procedures. Determination of the GC/MS Behavior of Standard Solutions of Triton X-Series Surfactants. The gas chromatogram of each of the Triton X surfactants was obtained by directly

injecting 1.0 µL of the standard surfactant solution (10.0 g/L in acetonitrile) into the GC. It should be cautioned that injection of the standard surfactant solutions without derivatization led to significant sorption of the surfactant in the GC injector inlet with subsequent GC runs producing many spurious unknown peaks. To clean the inlet, repeated injections (at least five) of 5.0 µL of the trimethylsilylation reagent mixture (equal volumes of BSTFA/ pyridine) was required. Trimethylsilylation Reaction of Standard Triton X-Series Surfactant Solutions. A 0.10 mL aliquot of the Triton X surfactant standard solution (10.0 g/L in acetonitrile) was transferred to a small vial followed by the addition of 0.10 mL neat pyridine and 0.10 mL neat trimethylsilylation reagent. The vial was capped using an inert screw cap, gently shaken, and placed in a 80 °C constant temperature water bath for 5 min. After cooling to room temperature, 1.0 µL of this reaction solution was directly injected into GC/MS system for analysis. Cloud Point Extraction from Aqueous Solutions. In a typical cloud point extraction, 4.70 mL of the aqueous analyte-containing solution at room temperature (ca. 18-20 °C) was placed in a 5.00 mL glass conical-bottom centrifuge tube. Next, 0.30 mL of a 100 g/L aqueous Triton X surfactant solution was added, the tube capped with a glass cap, and the mixture heated at 80 °C in a water bath for 10 min. The tube was then centrifuged for 3 min at 2500 rpm (908g) to facilitate the phase separation process. The surfactant-rich phase (ca. 50 µL) is more dense and appears at the bottom of the tube. The upper phase (surfactant-lean) was removed using a disposable Pasteur pipet. The moisture (water) present in the surfactant-rich phase was removed by reheating the tube again for 2.0 min in the 80 °C water bath followed by centrifugation for 1.0 min at 2500 rpm after which the aqueous phase was carefully removed with a microliter syringe. This heating/centrifugation/moisture removal processs was repeated an additional two times. A Hamilton microsyringe was then employed to remove 30 µL of the resulting surfactant-rich phase which was transferred to a special GC glass vial. For the cloud point extraction of ibuprofen and flurbiprofen, this same general procedure was employed except that the pH of the sample solution was adjusted to a value of 2.00 via use of a certified Fisher pH 2.00 buffer solution. Trimethylsilylation Reaction of the Surfactant in the SurfactantRich Phase Following the CPE Step. To 30 µL of the surfactantrich phase (resulting from the aforementioned CPE step) in the 2.0 mL GC vial was added 30 µL of neat pyridine and 30 µL of the neat trimethylsilylation reagent. The contents of the vial were gently mixed by shaking and the vial placed in an 80 °C constant temperature water bath for 5 min. A 1.0 µL portion of this solution was directly injected into the GC/MS system. It is important to note that the trimethylsilylation reagent, BSTFA and pyridine, readily dissolve in the Triton X-114 surfactant-rich phase provided that the moisture has been removed as described in the CPE step. If water is still present in the surfactant-rich phase, then the BSTFA does not completely dissolve, the resulting mixture appears as a two phase system, and the trimethylsilylation reaction does not proceed to any appreciable extent. To achieve complete reaction, a properly stored or fresh BSTFA reagent must be used. Analytical Chemistry, Vol. 81, No. 16, August 15, 2009

7115

Table 1. Diagnostic Ions and Time Windows used for Mass Spectrometric Quantification of Herbicides compound hexachlorocyclopentadiene acenaphthene-d10 simazine atrazine alachlor metolachlor butachlor

SIM ions (m/z) 236.8, 164.1 186.1, 200.1, 160.1, 162.1, 176.1,

271.7 201.0 215.0 269.1 238.1 311.1

time window (min) 5.00-15.00 15.00-20.00 20.00-25.00 20.00-25.00 25.00-29.00 25.00-29.00 29.00-final

Quantitative Determination of Herbicides by GC/MS following their Cloud Point Extraction. Spiked aqueous solutions containing nano- to microgram levels of the six herbicides were subjected to CPE using the experimental procedure as previously outlined. A 30 µL aliquot of the surfactant-rich phase thus obtained was placed in a special 2.0 mL GC vial to which was added 30 µL of 100 mg/L of acenaphthene-d10 (the internal standard) dissolved in pyridine, and 30 µL of fresh BSTFA. The vial was placed in an 80 °C constant temperature water bath for 5 min after which it was removed. After the vial reached room temperature, 1.0 µL of the solution was autoinjected (split ratio 50:1) into the Agilent 7890A GC/MS system. The mass selective detector was operated in the selective ion monitoring mode. The diagnostic ions and the corresponding time frames utilized are summarized in Table 1. All quantitative data was based on peak area measurements and every data point represents the average of duplicate determinations. Internal standard calibration with acenaphthene-d10 as the internal standard (present at a final concentraton of 33.3 mg/L) was employed for determination of the respective herbicide analytical figures of merit. The limit of detection (LOD) was calculated as the analyte concentration that would yield a detector signal equal to the mean plus three times the standard deviation of replicate blank measurements, that is, extraction of just distilled water containing no herbicide (n ) 7). The percent extraction recovery, %R, was calculated as

R(%) )

Csurfactant rich phaseVsurfactant rich phase × 100 C0V0

where C0 and V0 represent the initial herbicide concentration and volume prior to the CPE step and Csurf and Vsurf represent the final herbicide concentration in the surfactant-rich phase and the volume of that surfactant-rich phase. The enrichment (preconcentration) factor is equal to the ratio of the analyte concentration present in the surfactant-rich phase following the CPE step divided by its initial concentration in the solution prior to extraction. Under optimized CPE conditions, the TX-114 phase ratio was 0.01 [i.e., final volume of the surfactantrich phase (50.0 µL) divided by the initial volume of the aqueous phase (5.00 mL) ) 0.01]. The theoretical maximum enrichment factor is thus 100 for analytes that are quantitatively extracted. However, this surfactant-rich phase was subsequently diluted by the addition of the required BSTFA and pyridine derivatization reagents; hence 33.3 is the theoretical maximum preconcentration factor attainable under our experimental conditions. However, this 7116


value can be greatly increased by merely increasing the volume of the sample solution that is to be extracted. RESULTS AND DISCUSSION Gas Chromatography-Mass Spectrometric Behavior of Triton X-Series Surfactants and Their Trimethylsilylated Derivatives. Preliminary experiments were conducted in which the GC/MS chromatograms of standard solutions of TX-100, TX114, and TX-114 (reduced form) in acetonitrile were each recorded. The chromatogram of each surfactant (not shown) consisted of a series of peaks because each of the commercial preparations actually consist of a polydisperse mixture of oligomers which differ in their number of ethylene oxide repeat units, that is, different n values in formula C8H17C6H4(CH2CH2O)nH; for Triton X-100 (where on average, n ) 9.5) or Triton X-114 (where on average, n ) 7.5). Similar chromatographic profiles had been previously reported for such nonionic surfactants.22,28-31 The chromatographic peaks were quite broad and tailed. In addition, appreciable amounts of surfactant sorbed onto the injector inlet liner such that subsequent injections yielded chromatographs which exhibited many extra spurious peaks because of this memory effect. In an attempt to eliminate surfactant sorption and improve the appearance of the chromatographic peaks, the effect of derivatization of the Triton surfactants with trimethylsilylation (TMS) reagents was examined. The general order of reactivity for TMS reagents is N-trimethylsilylimidazole (TMSI) > N,O-bis(trimethylsilyl)trifluoroacetoamide (BSTFA) > trimethylsilyl azide (TMSA).32,33 Thus, both TMSI and BSTFA were investigated. The reaction of the Tritons with TMSI produced an insoluble salt byproduct, and the chromatograms exhibited extra peaks with an unstable baseline. However, the reaction with BSTFA proved effective as it completely reacted with the Triton X surfactants within 5 min on heating in the presence of pyridine. This reaction also produced a highly volatile early eluting byproduct that did not cause any chromatographic interference. The chromatograms obtained for the trimethysilylated derivatives of TX-100, TX-114, and TX-114R are shown in Figure 1. After derivatization with BSTFA, the TMS derivatives showed improved GC behavior with respect to peak shapes and peak tailing. The retention times of the TMS-Triton surfactants were slightly longer, compared to those of the underivatized forms, and separated as sharp peaks with no tailing. Fourteen peaks were observed in the chromatogram of TMS-Triton X-114R (Figure 1-A), seven for TMSTriton X-114 (Figure 1-B), and nine for TMS-Triton X-100 (Figure 1-C). The retention times for the seven TMS-Triton X-114 peaks are similar to each set of corresponding peaks seen in the chromatogram for TMS-Triton X-114R (see Supporting Information section for a discussion regarding the origin of these sets of peaks). In the analysis of TMS-Triton X-100, the GC conditions (28) Asmusse, C.; Stan, H. J. J. High Resol. Chromatogr. 1998, 21, 597–604. (29) Nadeau, H. G.; Oaks, D. M.; Nichols, W. A.; Carr, L. P. Anal. Chem. 1964, 36, 1914–1917. (30) Petrovic, M.; Barcelo, D. J. Mass Spectrom. 2001, 36, 1173–1185. (31) Cross, J. In Nonionic Surfactants - Chemical Analysis; Cross, J., Ed.; Marcel Dekker: New York, 1987; Chapter 6, pp 169-224. (32) Blau, K.; Halket, J. Handbook of Derivatives for Chromatography, 2nd ed.; John Wiley & Sons: New York, 1993. (33) Knapp, D. R. Handbook of Analytical Derivatization Reactions; John Wiley & Sons: New York, 1979.

Figure 1. GC/MS chromatograms of the trimethylsilylated derivatives of [A] Triton X-114R, [B] Triton X-114, and [C] Triton X-100 obtained upon reaction of standard solutions of these Triton surfactants in acetonitrile with BSTFA/pyridine at 80 °C for 5 min. Conditions are as noted in the experimental section except that in the case of TMS-Triton X-100, the programmed temperature ramp was at a rate of 15.0 °C/min from 80 to 320 °C during the chromatographic run (compared to 5.0 °C/min from 80 to 295 °C for TMS-Triton X-114 and TMS-Triton X-114R). The MS was operated in the total ion monitoring mode. The n refers to the number of oxyethylene repeat units.

differed (i.e., the temperature ramp rate was greater) from the general protocol noted in the experimental section because the elution time would otherwise have been very long (>90 min) as compared to the GC analysis of TMS-Triton X-114 or TMS-Triton X-114R using the general procedure. Regarding these TMS-Triton derivatives, there is an elution time window during which no TMSsurfactant peaks appear; in the case of TMS-Triton X-114 (or TX114R), this window encompasses the interval from 5 to 30 min under the GC conditions employed. Target analytes that elute within this time frame should be amendable to cloud point extraction and preconcentration with subsequent GC analysis. It is important to note that this elution window can be manipulated by changing the column temperature program (for example, this time window would be lengthened (increased) by decreasing the temperature ramp rate). All of the chromatographic peaks were identified by their mass spectra. The typical mass spectrum of such polyethoxylate surfactants shows the characteristic pattern of equally spaced signals with mass differences of 44 Da (corresponding mass of one ethylene oxide unit).34 In this study, mass spec peaks corresponding to homologous with n ) 2 to n ) 8 were observed for both TMS-Triton X-114 and TMS-Triton 114R while peaks corresponding to n ) 2 to n ) 10 were seen for TMS-Triton X-100. Shen and Shao had previously assigned mass spec peaks corresponding to n ) 2 to 5 for underivatized Triton X-114 in dichloromethane as solvent.22 (34) Laughlin, J. B.; Cassady, C. J.; Cox, J. A. Rapid Commun. Mass Spectrom. 1997, 11, 1505–1508.

The mass spectrum of the most intense peak (peak #5 corresponding to the species with n ) 5 in Figure 1-B, retention time ) 43.40 min) for TMS-Triton X-114 is shown in Figure 2. The molecular ion peak is observed at m/z ) 498. The largest peak at m/z ) 427 corresponds to the fragment remaining after the dissociation of the TMS from the molecule. The peaks whose m/z ) 117, 161, 205, 249, and 293 arise from the TMSpolyoxyethylene species that differ in their number of ethylene oxide moieties present. The m/z ) 73 and 57 peaks are due to the presence of the TMS and tert-butyl group fragments, respectively. The chromatographic and mass spectrometric results indicated that the reaction between BSTFA and the different Triton X series surfactants was essentially complete (>99.9% reacted). If the reaction between the Tritons and BSTFA is not complete, then no fragment peak at m/z ) 73 is observed in the mass spectrum. Optimization of the Derivatization Reaction between BSTFA and the Triton X-114 Surfactant in the SurfactantRich Phase and its Gas Chromatography-Mass Spectrometric Behavior. (i) Selection of the Proper Additive for the SurfactantRich Phase. As noted in the literature, the surfactant-rich phase is typically too viscous for direct injection into any chromatographic system. The viscosities of neat Triton X-114 and Triton X-100 are reported to be 260 and 240 cP, respectively, at 25 °C35 and that of the TX-114 coacervate phase 205 cP27 which makes it difficult to reproducibly and accurately draw and dispense this phase using (35) Technical Data Sheets for Triton X-114 Surfactant and Triton X-100 Surfactant, The Dow Chemical Company, Midland, MI; see: http:// www.dow.com/surfactants/products/octyl.htm.


7117

Figure 2. Mass spectrum of the component giving rise to the most intense peak eluting at about 43.40 min in Figure 1-B which corresponds to TMS-Triton X-114 (n ) 5).

a microsyringe. To overcome this difficulty, a small volume of a miscible organic solvent is typically added to the surfactant-rich phase. The solvents dimethylsulfoxide, dimethylformamide, and pyridine were considered since they also facilitate bimolecular nucleophilic substitution (SN2) reactions such as this TMS derivatization reaction.36 Pyridine proved most effective since it is more volatile and serves as an efficient proton acceptor for derivatization reactions involving silylating agents as BSTFA. Previously, pyridine was reported to be a very effective solvent for BSTFA derivatization reactions37,38 including the derivatization of nonionic surfactants.28 (ii) Effect of Moisture in the Surfactant-Rich Phase on the TMS Reaction. Trimethylsilylation reagents hydrolyze and become nonreactive when exposed to moisture.39 Since the surfactantrich phase following the cloud point extraction step contains appreciable amounts of water, the trimethylsilylation reaction does not proceed to any significant extent. When a sample containing the Triton X-114 surfactant-rich condensate phase (which also contained water) was reacted with BSTFA/pyridine at 80 °C for 5 min and then injected into the GC, the chromatograph obtained exhibited 2 sets of peaks for each of the homologues present in the surfactant preparation (similar to that shown in Figure 3A). Mass spectra analyses of each set of peaks confirmed that one peak corresponded to TMS-Triton X-114 and the other to Triton X-114 (non-derivatized). A significant amount of the non-derivatized Triton X-114 also sorbed to the injector inlet which caused problems with subsequent injections unless cleaned. Even after 1 h reaction at 80 °C, the extent of reaction had not improved. (36) BSTFA Product Bulletin, Pierce Biotechnology, Rockford, IL. (37) Rice, S. L.; Mitra, S. Anal. Chim. Acta 2007, 589, 125–132. (38) Gatidou, G.; Thomaidis, N. S.; Stasinakis, A. S.; Lekkas, T. D. J. Chromatogr. A 2007, 1138, 32–41. (39) Xu, L.; Basheer, C.; Lee, H. K. J. Chromatogr. A 2009, 1216, 701–707.

7118


However, the moisture present in the surfactant-rich extractant phase could be effectively eliminated by subjecting that phase to three heating/centrifugation cycles with the water removed by microsyringe after each cycle followed by the addition of the trimethylsilylation reagent after the final cycle. Figure 3-A shows the chromatogram of a standard solution of Triton X-114 which had not been appreciably derivatized with BSTFA. The peaks pertaining to Triton X-114 (non-derivatized) are designated with the prime mark (′) on the peak numbers whereas the small peaks to the right correspond to those of the TMS-derivatized TX-114. Figure 3-B shows the chromatogram of the derivatized TMS-Triton X-114 (obtained via derivatization of a standard solution of TX-114). As can be observed, the peaks for the derivatized TX-114 are sharper and their retention times are slightly greater (suggesting that the TMS-TX-114 derivatives are slightly less volatile compared to Triton X-114 underivatized). Figure 3-C shows the chromatogram obtained after derivatization of the TX-114 of the surfactant-rich phase under optimized conditions. The retention times of all peaks were very close to those observed for TMS-Triton X-114 standard (Figure 3-B) although they do elute a bit earlier. Mass spectral analysis of the respective peaks indicated that they corresponded with those as labeled in Figure 3-B. It should be noted that broad peaks of very low signal intensity (indicated by the asterisk (*) on the peak numbers) appeared between sharp peaks in Figure 3-C. The retention times of these peaks were different from both those of standard Triton X-114 and/or TMS-Triton X-114 peaks. Mass spectroscopic analysis of the component(s) responsible for these peaks indicated the presence of both the TMS derivatized and nonderivatized forms of the Triton X-114 surfactant. It appears that these low signal intensity but broad peaks arise as a consequence of an azeotropic mixture formed by the presence of both the TMS

Figure 3. GC/MS chromatograms obtained for [A] Triton X-114 solution (in acetonitrile) which had not undergone appreciably reaction with BSTFA/pyridine (the peaks labeled with the numbers with a prime mark (′) represent the underivatized TX-114 while those with just the number represents the trimethylsilylated TX-114 homologues), [B] a standard solution of Triton X-114 (in acetonitrile) which was trimethylsilyated via reaction with BSTFA/pyridine, and [C] the surfactant-rich phase of TX-114 after reaction with BSTFA/pyridine under optimized conditions (the numbers with the asterisk * indicate the low intensity broad peaks observed between the sharp TMS-TX-114 peaks).

derivatized and the non-derivatized form of the Triton X-114 surfactant which essentially coelute. Analytical Characteristics of the Proposed Method. The proposed CPE methodology was applied to the GC/MS analysis of several analyte types to demonstrate feasibility and evaluate the analytical performance of this appraoch. First, an environmental sample solution spiked with 4.00 ppm each of the polycyclic aromatic hydrocarbons (PAHs) acenaphthene, acenaphthylene, anthracene, biphenyl, dibenzofuran, and fluorene was subjected to CPE using Triton X-114. As shown in Figure 4-A, the PAH peaks all appeared within the elution window for TX-114 free from any interference due to the presence of derivatized surfactant peaks, and the PAH peak shapes were acceptable and allowed for precise quantification. Injection of a standard solution of these PAH in acetone that had not been subjected to CPE yielded the chromatogram shown in Figure 4-B and revealed that there is no difference in the respective PAH retention times. All peaks were identified by their mass spectra in these experiments. Repeated injections of the standards and treated CPE extracts also gave no significant fluctuations of the retention times. Good agreement was observed in the mean retention times for each of the six PAHs as determined by GC/MS analysis of the standard solutions and the CPE extracts in which the surfactant-rich phase had been derivatized with BSTFA (Table 2). The relative standard deviations (RSD) obtained were quite good; in fact, the RSDs were slightly better for the CPE samples relative to the standard solutions. Likewise, good reproducibility and agreement between the retention times of the herbicide standards (alachlor, atrazine,

butachlor, hexachlorocyclpentadiene, metolachlor, and simazine) and the same herbicides present in the CPE extract were achieved (see Supporting Information section for the chromatograms). The quantitative performance of the proposed CPE GC/MS method was evaluated using this same six component herbicide test mix along with acenaphthene-d10 as an internal standard. A representative chromatogram is shown in Figure 5 (Table 1 provides a summary of the SIM ions monitored). The herbicides and internal standard gave fairly sharp peaks allowing for the simultaneous determination of these six herbicides. Table 3 summarizes the analytical figures of merit obtained for the proposed CPE GC/ MS method for the determination of these herbicides. Linear relationships were obtained between peak area and herbicide concentration. The results show that a wide linear dynamic range was achieved (ca. 0.5-4000 ppb) with good linearity (correlation coefficients of 0.99) and precision (RSDs in range of 2.1 to 6.7%). The LODs of this proposed CPE GC/MS method (Table 3) ranged from 6.6 ppt (parts per trillion, for metolachlor) to 0.50 ppb (for hexachlorocyclopentadiene). These detection limits are better than or comparable to those reported for the determinaton of these herbicides by a variety of methods. For instance, the detection limits attained for atrazine and simazine via the proposed CPE GC/MS method were 20 and 100 ng/L, respectively. These LODs compare favorably to reported values of 100 and 40 µg/L for atrazine and simazine (determined by liquid-liquid microextraction - HPLC),40 3000 ng/L for atrazine (determined by a fluores(40) Zhou, Q.; Pang, L.; Xiao, J.; Bai, H. Anal. Sci. 2009, 25, 73–76.


7119

Figure 4. GC/MS chromatogram of the polycyclic aromatic hydrocarbons (1) biphenyl, (2) acenaphthylene, (3) acenaphthene, (4) dibenzofuran, (5) fluorene, and (6) anthracene. [A] represents the chromatogram observed after injection of the treated surfactant-rich phase obtained after the TX-114 cloud point extraction of an initial aqueous mixture of these six PAHs (present at a concentration of 4.00 ppm) using the optimized experimental protocol whereas [B] represents the chromatogram obtained after injection of a standard solution of these PAHs (20 ppm each in acetone as solvent). MS operated in the total ion monitoring mode. Table 2. Comparison of PAH Retention Times and Reproducibility retention time (min) (RSDa, %) analyte

standard sampleb

with CPEc

Biphenyl Acenaphthylene Acenaphthene Dibenzofuran Fluorene Anthracene

13.2 (1.8) 14.9 (1.5) 15.7 (1.4) 16.5 (1.4) 18.1 (1.1) 22.6 (1.1)

13.2 (0.7) 14.9 (0.7) 15.7 (0.7) 16.5 (0.8) 18.0 (0.5) 22.6 (0.5)

a Relative Standard Deviation (RSD), n ) 4. b PAH concentration was 20.0 ppm. c Initial PAH concentration was 4.0 ppm.

cence polarization immunoassay),41 600 ng/L for atrazine (determined by liquid phase microextraction - HPLC),42 20-500 ng/L for atrazine (determined by a flow injection enzyme immunoassay),43 and 200 µg/L for atrazine (determined by an ELISA procedure).44 However, the proposed method is not as sensitive for atrazine as a CPE-ELISA method developed by Stangl. et al., who reported a detection limit of 8 ng/L45 or a nanoparticle-based

(41) Choi, M. J.; Lee, J. R.; Eremin, S. A. Food Agric. Immunol. 2002, 14, 107– 120. (42) Zhou, Q. X.; Xie, G. H.; Pang, L. Chin. Chem. Lett. 2008, 19, 89–91. (43) Bjarnason, B.; Bousios, N.; Eremin, S.; Johansson, G. Anal. Chim. Acta 1997, 347, 111–120. (44) Stocklein, W. F. M.; Rohde, M.; Scharte, G.; Behrsing, O.; Warsinke, A.; Micheel, B.; Scheller, F. W. Anal. Chim. Acta 2000, 405, 255–265. (45) Stangl, G.; Weller, M. G.; Niessner, R. Fresenius’ J. Anal. Chem. 1995, 351, 301–304.

7120


solid phase extraction-MS procedure which attained a detection limit of 10 ng/L.46 Table 4 gives the recovery data attained for this herbicide CPE GC/MS method. The recoveries ranged from a low of 50% for simazine to essentially quantitative for butachlor and metolachlor. The precision was also good with the RSDs in the range of 2.0 to 7.0%. Previously, in the context of work aimed at determination of the binding of triazine herbicides to Triton X-114 micelles and to the surfactant-rich phase of TX-114,47 recoveries of about 30% and 50% for simazine and atrazine were reported for the TX-114 CPE step (at [TX-114] ) 0.60%), which is the same trend as found in this study. CPE of herbicides, including atrazine and simazine, using Genapol-X-080 was previously employed as a preconcentration step prior to HPLC using micellar liquid chromatography and an aqueous mobile phase containing that same surfactant.48 In that work, no detection limits were reported, but the enrichment factors achieved ranged from 8 to 65 depending upon the specific herbicide. The effective enrichment factors achieved for the CPE protocol proposed in this work were 17, 25, 30, 32, 33, and 33 for simazine, atrazine, alachlor, hexachlorocyclopentadience, butachlor, and metolachlor, respectively. An additional attractive feature of the proposed CPE approach is that many polar analytes (e.g., those containing an acidic, alcohol, thiol, amide, etc. moiety) possess a labile hydrogen atom (46) Song, Y.; Zhao, S.; Tchounwou, P.; Liu, Y. M. J. Chromatogr., A. 2007, 1166, 79–84. (47) Carabias-Martinez, R.; Rodriguez-Gonzalo, E.; Dominguez-Alvarez, J.; Pinto, C. G.; Hernandez-Mendez, J. J. Chromatogr., A 2003, 1005, 23–34. (48) Halko, R.; Hutta, M. Anal. Chim. Acta 2002, 466, 325–333.

Figure 5. GC/MS chromatogram observed after the injection of the treated surfactant-rich phase following the TX-114 cloud point extraction of a mixture of (1) hexachlorocyclopentadiene, (2) acenaphthene-d10 (the internal standard), (3) simazine, (4) atrazine, (5) alachlor, (6) metolachlor, and (7) butachlor. The initial concentration of all of the herbicide analytes was 4.00 mg/L in the solution prior to its CPE and the MS was operated in the selective ion monitoring mode (see Table 1).

Table 3. Analytical Figures of Merit for CPE-GC/MS Determination of Herbicidesa

Table 4. Recoveries of the Selected Herbicides c

herbicide hexachlorocyclopentadiene simazine atrazine alachlor metolachlor butachlor

LOD linear range correl. (µg L-1) coeff. (R2) RSDb, % (ng L-1) 5–4000 1–4000 0.5–4000 0.5–4000 0.1–4000 0.5–4000

0.9964 0.9926 0.9887 0.9960 0.9983 0.9999

5.8 5.9 4.6 6.7 2.1 6.3

481.5 97.1 19.6 31.2 6.59 33.9

a Data based on the CPE (using Triton X-114 as surfactant) of 5.0 mL of an initial herbicide containing solution followed by GC/MS analysis. The protocol was as outlined in the Experimental Section. b Relative Standard Deviations determined from replicate analysis (n ) 7) of herbicide solutions that contained 1.0 mg/L of the indicated herbicide. c Limit of Detection defined as 3 times the signal-to-noise ratio.

which would also be replaced by the trimethylsilyl group upon reaction with BSTFA/pyridine. Thus, the surfactant derivatization reaction serves to also simultaneously derivatize such analytes that are cloud point extracted prior to their introduction into the GC system. The GC/MS analysis of the non-steroidal antiinflammatory drugs, ibuprofen and flurbiprofen, serves to illustrate this point. Such Profen drugs are routinely analyzed by GC/MS as their trimethylsilyl derivatives.49-51 Injection of a standard (49) Sebok, A.; Vasanits-Zsigrai, A.; Palko, G.; Xaray, G.; Molnar-Perl, I. Talanta 2008, 76, 642–650. (50) Jones, O. A. H.; Voulvoulis, N.; Lester, J. N. Chromatographia 2003, 58, 471–477.

compounda

recoveryb, %

RSDb, %

hexachlorocyclopentadiene simazine atrazine alachlor metolachlor butachlor

95.5 50.2 74.3 90.0 100.0 100.0

6.0 6.0 4.6 7.0 2.0 5.0

a Data based on the CPE (using Triton X-114 as surfactant) of 5.0 mL of an initial herbicide containing aqueous solution followed by GC/ MS analysis. b Solutions were spiked with 100 µg of each herbicide (n ) 7).

solution containing these two profens (underivatized) did not yield any GC/MS peaks even after 90 min. However, two peaks are observed in the chromatogram (see Supporting Information section) after the cloud point extraction of a solution that contained both of these profens in which the TX-114 surfactant in the surfactant-rich extractant phase was reacted with BSTFA. Previously, Han et al. had utilized CPE as a preconcentration step prior to the HPLC determination of flurbiprofen (underivatized) in rat plasma.52 CONCLUSIONS The results demonstrate that it is possible via post-extraction derivatization of the surfactant of the surfactant-rich extractant (51) Rice, S. L.; Mitra, S. Anal. Chim. Acta 2007, 589, 125–132. (52) Han, F.; Yin, R.; Shi, X. L.; Jia, Q.; Liu, H. Z.; Yao, H. M.; Xu, L.; Li, S. M. J. Chromatogr., B 2008, 868, 64–69.


7121

phase to utilize the cloud point extraction technique prior to GC or GC/MS analysis. Using the proposed protocol no deleterious effects upon the GC column nor MS detector system were noticed. The CPE-GC/MS analysis of several model analyte classes yielded very reproducible retention times and excellent quantitative performance. Relative to the other proposed CPE GC approaches, the proposed method offers advantages primarily in terms of convenience and simplicity as less sample manipulations and no additional separation step(s) are required. However, because of the relatively high inlet temperature requirement, the proposed approach will probably not be applicable for GC analysis of thermally labile analytes. Very volatile, quickly eluting (within the first 5 min) analytes might also prove difficult to separate. Such limitations might be overcome by the use of different nonionic surfactant(s) and/or other types of derivatization reagents in the CPE protocol (i.e., ones that lead to a lowering of the GC inlet and column operating temperature requirements). Overall, this

7122


general approach should serve to help expand the scope of CPE as an extraction procedure prior to the GC analysis. ACKNOWLEDGMENT The authors gratefully acknowledge the support of this work in the form of a Japan Society for the Promotion of Science Fellowship (for Research Abroad to Y.T.), WFU Reynolds Research Leave (to W.L.H.) and funding by WFU through the Provost’s Fund for Capital Improvement and Department of Chemistry that allowed for the purchase of the GC/MS system. In addition, we thank Dr. Marcus Wright for GC/MS instrumentation assistance. SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review May 7, 2009. Accepted July 8, 2009. AC9009963

Cloud Point Extraction with Surfactant Derivatization as an Enrichment

Recommend Documents