Anal. Chem. 2009, 81, 6765–6773
Interlaboratory Trial on the Analysis of Alkylphenols, Alkylphenol Ethoxylates, and Bisphenol A in Water Samples According to ISO/CD 18857-2 E. Stottmeister,*,† O. P. Heemken,‡ P. Hendel,† G. Donnevert,§ S. Frey,| H. Allmendinger,⊥ G. Sawal,X B. Jandel,O S. Geiss,# R. Donau,3 A. Koch,[ I. Heinz,[ M. Ottaviani,4 E. Veschetti,4 W. Hartl,× C. Kubwabo,] C. Benthe,+ V. Tobinski,+ H. Woldmann,∞ and R. Spilker¶ Umweltbundesamt, Heinrich-Heine-Strasse 12, 08645 Bad Elster, Germany, Niedersa¨chsisches Landesamt fu¨r Verbraucherschutz und Lebensmittelsicherheit-Veterina¨rinstitut Oldenburg, Philosophenweg 36, 26121 Oldenburg, Germany, Fachhochschule Giessen-Friedberg, Wiesenstrasse 14, 35390 Giessen, Germany, Bayerisches Landesamt fu¨r Umwelt, Kaulbachstrasse 37, 80539 Mu¨nchen, Germany, Bayer Industry Services, Bayerwerk, 51368 Leverkusen, Germany, Umweltbundesamt, Postfach 33 00 22, 14191 Berlin, Germany, Niedersa¨chsischer Landesbetrieb fu¨r Wasserwirtschaft, Ku¨sten- und Naturschutz, An der Scharlake 39, 31135 Hildesheim, Germany, Thu¨ringer Landesanstalt fu¨r Umwelt und Geologie, Go¨schwitzer Strasse 41, 07745 Jena, Germany, Landeslabor Brandenburg, Gerhard-Neumann-Strasse 2/3, 15236 Frankfurt/Oder, Germany, Hygiene-Institut des Ruhrgebiets, Rotthauser Strasse 19, 45897 Gelsenkirchen, Germany, Istituto Superiore di Sanita`, Viale Regina Elena 299, 00161 Rome, Italy, Umweltbundesamt GmbH, Spittelauer La¨nde 5, 1090 Wien, Austria, Health Canada, 50 Columbine Driveway, PL 0800 C, Ottawa, Ontario, K1A 0K9, Canada, Landeslabor Schleswig-Holstein, Max-Eyth-Strasse 5, 24537 Neumu¨nster, Germany, Im Haselbusch 7, 21614 Buxtehude, Germany, and Sasol Solvents Germany GmbH, Paul-Baumann-Strasse 1, 45764 Marl, Germany ISO/CD 18857-2 (International Organization for Standardization, Geneva) describes a new international standard method for the determination of octylphenol, nonylphenol, their mono- and diethyoxylates, and bisphenol A in nonfiltered samples of drinking, ground, surface, and wastewater. The method is based on the extraction of the analytes from an acidified water sample by solid phase extraction, solvent elution, derivatization, and determination by gas chromatography with mass spectrometric detection. For validation of this method, 14 laboratories from 4 different countries in Europe and Canada participated in an interlaboratory trial to determine the performance characteristics of the method, which are intended for publication in the corresponding standard. The interlaboratory trial was evaluated according * To whom correspondence should be addressed. E-mail: ernst.stottmeister@ uba.de. Fax: +49-37437-76219. † Umweltbundesamt. ‡ Niedersa¨chsisches Landesamt fu ¨ r Verbraucherschutz und Lebensmittelsicherheit-Veterina¨rinstitut Oldenburg. § Fachhochschule Giessen-Friedberg. | Bayerisches Landesamt fu ¨ r Umwelt. ⊥ Bayer Industry Services. X Umweltbundesamt. O Niedersa¨chsischer Landesbetrieb fu ¨ r Wasserwirtschaft, Ku ¨ sten- und Naturschutz. # Thu ¨ ringer Landesanstalt fu ¨ r Umwelt und Geologie. 3 Landeslabor Brandenburg. [ Hygiene-Institut des Ruhrgebiets. 4 Istituto Superiore di Sanita`. × Umweltbundesamt GmbH. ] Health Canada. + Landeslabor Schleswig-Holstein. ∞ Im Haselbusch 7. ¶ Sasol Solvents Germany GmbH. 10.1021/ac900813m CCC: $40.75 2009 American Chemical Society Published on Web 07/23/2009
to ISO 5725-2 and included two duplicate nonfiltered water samples: surface water containing the target compounds in an analyte concentration range from 0.05 to 0.4 µg/L and wastewater containing the target compounds in a concentration ranged from 0.1 to 5 µg/L. The repeatability variation coefficients (within-laboratory precision) varied for all samples and compounds between 1.9 and 7.8%, showing a sufficiently high repeatability of the method. The reproducibility variation coefficients (between-laboratory precision) were found to vary within a satisfactory range of 10.0-29.5% for surface water and 10.8-22.5% for wastewater. The recoveries as a measure of accuracy varied from 98.0 to 144.1% for surface water and from 95.4 to 108.6% for wastewater. The determined concentrations of the samples compared well to the “true” values, thus showing very satisfactory accuracy of the method. In the chromatogram of the surface water sample, a high unresolved background made up of coextractable matrix compounds was apparent. It is conceivable that compounds from this background may be responsible for enhanced recoveries of 144.1% for 4-nonylphenol (mixture of isomers) and of 123.4% for 4-nonylphenol monoethoxylate (mixture of isomers) in the surface water samples. The isotope-marked standard compounds developed in this context proved to be reliable internal standards that allow a precise and accurate quantitation of all compounds specified in ISO/CD 18857-2. The results of the interlaboratory trial confirmed that the analytical method is robust and reliable and can be used as a standard method to analyze the target compounds in water samples. Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
6765
Alkylphenols, such as octylphenol and nonylphenol, alkylphenol ethoxylates, and bisphenol A are synthetic organic chemicals. Widespread use of these compounds in industries, agriculture, and domestic products has resulted in environmental contamination including water, sediment, and biota on the global scale.1-4 Reliable analytical results are required to assess the levels of contamination by these compounds in order to monitor and prevent further contamination and toxic effects. Because of the complex nature of these compounds (there are theoretically approximately 170 isomers of 4-nonylphenol) and very low level of contamination (micrograms per liter, nanograms per liter), a reliable, standard analytical method needs to be evaluated. To ensure the production of valid, comparable data for monitoring programs, e.g., for implementing the European Water Framework Directive,5,6 it is absolutely necessary to develop internationally validated standard methods for these priority pollutants. Among the alkylphenols, nonylphenol and its ethoxylated derivates present a particular analytical challenge because they are technical mixtures of isomers. The technical synthesis of nonylphenol leads to this complex mixture of nonylphenols consisting of isomeric compounds that have different branched nonyl side chains. Their identification is based on the peak pattern (fingerprint) of the chromatogram, although the relationship among the individual peaks of the pattern may differ in samples and standards. Substances coeluting with nonylphenol and its mono- and diethoxylates can interfere in the determination. This may have a large influence on the result, since these three analytes are quantified from the sum of all peaks belonging to the chromatographic pattern. The analyst should use caution to include only those peaks from the sample that are attributable to the analytes of interest. In complex environmental samples, such interferences can occur even if gas chromatography/ mass spectrometry (GC/MS) is used. The difficulty of analyzing these compounds is illustrated by the fact that different interlaboratory trials have resulted in widely varied results.7,8 Another major disadvantage in current analytical procedures is the lack of appropriate isotope-marked internal standards. For these reasons, in April 2006, the German Institute for Standardization (referred to as DIN) submitted a new work-item proposal to the International Organization for Standardization (ISO) to develop an ISO standard method for the determination of alkylphenols, their ethoxylates, and bisphenol A in nonfiltered water samples using solid phase extraction and gas chromatog(1) Kim, Y.-S.; Katase, T.; Sekine, S.; Inoue, T.; Makino, M.; Uchiyama, T.; Yamashita, N. Chemosphere 2004, 54, 1127–1134. (2) Kim, Y.-S.; Katase, T.; Horii, Y.; Yamashita, N.; Makino, M.; Uchiyama, T.; Fujimoto, Y.; Inoue, T. Organohalogen Compd. 2004, 66, 660–667. (3) Bucher, J. R. Environ. Health Perspect. 2009, 117, A96-A97. (4) Heemken, O. P.; Reincke, H.; Stachel, B.; Theobald, N. Chemosphere 2001, 45, 245–259. (5) European Commission. Off. J. Eur. Commun. 2000, L 327. (6) European Commission. Off. J. Eur. Commun. 2001, L 331/1. (7) Sobiecka, E.; Van der Sloot, H.; Hansen, N.; Gawlik, B. M. Project HORIZONTAL Validation Report: Validation of a horizontal standard for the determination of nonylphenol (NP) and nonylphenol - mono- and diethoxylates - by gas chromatography with mass selective detection (GC-MS) in a European Intercomparison Exercise; European Communities, Office for Official Publications of the European Communities: Luxembourg, 2002 (ISBN 978-92-79-07123-2). (8) Loos, R.; Wollgast, J.; Castro-Jiminez, J.; Mariani, G.; Huber, T.; Locoro, G.; Hanke, G.; Umlauf, G.; Bidoglio, G.; Hohenblum, P.; Moche, W.; Weiss, S.; Schmid, H.; Leiendecker, F.; Ternes, T.; Navarro Ortega, A.; Hildebrandt, A.; Barcelo, D.; Lepom, P.; Dimitrova, I.; Nitcheva, O.; Polesello, S.; Valsecchi, S.; Boutrup, S.; Sortkjaer, O.; de Boer, R.; Staeb, J. Trends Anal. Chem. 2008, 27, 89–95.
6766
Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
raphy with mass spectrometric detection. The members of ISO voted on the DIN proposal, and a majority of the voting countries approved the new work. The proposal included a working draft standard developed by DIN working group “Phenols”. This working draft was handed over to the responsible ISO working group WG 17 “Phenols” of the technical committee ISO/TC 147 “Water Quality” and its subcommittee SC 2 “Physical, Chemical, and Biochemical Methods”. The WG 17 developed a committee draft (CD) of the new standard method referred to as ISO/CD 18857-29 and decided to validate it by an interlaboratory trial as the final part of the method development. ISO accepts a new method as fully validated only if the method has been successfully tested in an interlaboratory method performance study. So the purpose of this collaborative assessment was to determine the performance characteristics (accuracy and precision) of the proposed method, to give information on the comparability of the analytical results attained at different laboratories, and to evaluate critical points such as the suitability of the proposed internal standards for quantifying multicomponent analytes (e.g., the technical mixture of nonylphenol isomers) which were specially developed for this analytical approach. Ideal internal standards are isotope-marked molecules of a respective target analyte, which are usually prepared via organic synthesis by exchanging carbon [12C] atoms by [13C] or by exchanging some of the hydrogen atoms by deuterium (Table 1). Physico-chemical properties of such substances are very similar to or nearly the same as of their naturally occurring target analytes, but because of their higher molecular mass (due to the incorporated isotopes) distinction between internal standard and target analyte is possible. Variations during sample preparation are compensated so that methods with especially high analytical accuracy and precision can be developed. The results obtained in the interlaboratory trial will be used for incorporation into the new ISO standard. The study was evaluated according to ISO 5725-210 and included two duplicate nonfiltered water samples: surface water (2 × 1 L) containing the target compounds in an analyte concentration range from 0.05 to 0.4 µg/L and wastewater (2 × 100 mL) containing the target compounds in a concentration range from 0.1 to 5 µg/L. All laboratories were asked to follow the procedure exactly as prescribed in the draft standard. The interlaboratory trial was organized by the Federal Environment Agency (Bad Elster, Germany) and started in November 2007. Water samples were shipped to the participants in the week of November 12, 2007. The organizer also provided a calibration solution containing known amounts of all analytes, a solution containing all internal standards, two spike solutions (separately for surface and wastewater) with unknown concentrations of the analytes, and solid phase extraction cartridges. The participants were asked to spike the water samples immediately prior to extraction to avoid possible losses of the analytes due to chemical and microbial action. Detailed instructions were given to the laboratories spiking their own samples. The deadline for submission of results has been set to December 31, 2007. Participants were asked to submit data using an online data submission system or electronically as Excel reporting sheets using e-mail. The data submission included a detailed questionnaire for additional (9) International Standardization Organization. ISO/CD 18857-2; Geneva, 2008. (10) International Standardization Organization. ISO 5725-2; Geneva, 1994.
Table 1. Target Analytes, Internal Standards (ISTD), Their Abbreviations, and Selected Diagnostic Ions Used for Identification and Quantitation selected diagnostic ions no.
analyte (trimethylsilyl derivative)
abbreviation
target M1a
qualifier M2a
qualifier M3a
ISTD for analyte no.
1 2 3 4 5 6 7 8 9 10 11 12 13 14
4-(1,1,3,3-tetramethylbutyl)phenol 4-(1,1,3,3-tetramethylbutyl)phenol monoethoxylate 4-(1,1,3,3-tetramethylbutyl)phenol diethoxylate 4-nonylphenol (mixture of isomers) 4-nonylphenol monoethoxylate (mixture of isomers) 4-nonylphenol diethoxylate (mixture of isomers) bisphenol A 4-(1,1,3,3-tetramethylbutyl)phenol (ring-13C6)b 4-(1,1,3,3-tetramethylbutyl)phenol monoethoxylate (ring-13C6)b 4-(1,1,3,3-tetramethylbutyl)phenol diethoxylate (ring-13C6)b 4-(3,6-dimethyl-3-heptyl)phenol (ring-13C6)b 4-(3,6-dimethyl-3-heptyl)phenol monoethoxylate (ring-13C6)b 4-(3,6-dimethyl-3-heptyl)phenol diethoxylate (ring-13C6)b bisphenol A-d16b
OP OP1EO OP2EO NP NP1EO NP2EO BPA OP-13C6 OP1EO-13C6 OP2EO-13C6 363 NP-13C6 363 NP1EO-13C6 363 NP2EO-13C6 BPA-d16
207 251 295 207 251 295 357 213 257 301 227 271 315 368
278 322 366 221 265 309 358 284 328 372 269 313 357 369
208 252 296 193, 179 279, 307 323, 351 372 214 213 213 298 342 386 386
1 2 3 4 5 6 7
a
M1 is used for quantification, and M2 and M3 may be used for identification. b Internal standard (ISTD).
information on the measurements. A total of 14 laboratories from four different countries (Austria, Canada, Italy, and Germany) participated. A total of 13 laboratories from 4 countries reported results; one laboratory withdrew due to problems with the GC/ MS instrument. EXPERIMENTAL SECTION Analytical Method. The analytical method applied by all participating laboratories is described in detail in ISO/CD-18857-2 “Water qualitys Determination of selected alkylphenolss Part 2: Gas chromatographic-mass spectrometric determination of alkylphenols, alkylphenol ethoxylates and bisphenol A in non-filtered samples following solid phase extraction and derivatization”. The scope of this method is the determination of octylphenol, nonylphenol, the mono- and diethoxylates of these alkylphenols, and bisphenol A in nonfiltered samples of drinking, ground, surface, and wastewater. The lower limit of the working range depends on the matrix, on the specific compound to be analyzed, and on the sensitivity of the mass spectrometric detection unit. For the compounds in Table 1, the method is applicable to drinking, ground, and surface water in a working range from 0.005 to 0.2 µg/L for OP, OP1EO, and OP2EO, from 0.03 to 0.2 µg/L for NP, NP1EO, and NP2EO, and from 0.05 to 0.2 µg/L for BPA. With dependence on the matrix, the method is also applicable to wastewater in a working range from 0.1 to 50 µg/L for OP, its ethoxylates, and BPA and from 0.5 to 50 µg/L for NP and its ethoxylates. Waste water samples are extracted by passing a 100 mL of sample through the cartridge. Briefly, the method applied consists of the following steps: (1) acidification of the water sample to pH 2 with sulfuric or hydrochloric acid, (2) addition of isotope-marked internal standards, specified in Table 1, (3) solid phase extraction and drying of the cartridge, (4) elution, concentration, and reconstitution of the extract, (5) derivatization with 2,2,2-trifluoro-N-methyl-N-(trimethylsilyl)acetamide (MSTFA), (6) measurement and quantitation of the trimethylsilyl derivatives. Details of the method are given elsewhere.11,12 Sample Characteristics. The wastewater sample for the interlaboratory trial was taken from the outflow of the municipal sewage plant of the small towns Adorf and Bad Elster (Saxony, Germany). The sampling was done by the Federal Environment Agency on November 12, 2007. The suspended-solid content of the wastewater was 4.3 mg/L.
The surface water sample for the interlaboratory trial was taken from the drinking water reservoir Dro¨da, which is the main and the most important drinking water resource in the district Vogtland (Saxony, Germany). The sampling was done by the Federal Environment Agency in collaboration with the State Reservoir Administration of Saxony, Laboratory Plauen, on November 13, 2007. The suspended-solid content of the surface water was 0.87 mg/L. Sample Preparation. Both types of samples were homogenized and then acidified with analytical grade sulfuric acid to pH 2 as described in the draft standard method. The samples were checked for alkylphenols, alkylphenol ethoxylates, and bisphenol A which were not present in the samples from contamination (background) of the water. The samples were shipped simultaneously to all participants using a private courier service. The water samples were not spiked until reaching the participating laboratory. Prior to analysis, the water samples were to be spiked by the participating laboratories as part of the protocol for the study to reduce possible losses of analytes by chemical and microbial action. Detailed instructions were given to the laboratories spiking their own samples to minimize the variability which could come from a mistake in spiking. The protocol that the participating laboratories were required to follow explicitly explained the reason for this approach and stressed again that the objective of the interlaboratory trial is to ensure that the variability in results primarily reflected the typical method capability rather than variability in laboratory or analyst performance. For spiking, 100 µL of two different spiking solutions (provided by the organizer) were added to surface and wastewater samples according to the detailed extra instructions provided. The samples were shaken vigorously, and 100 µL of the internal standard solution was added to each sample. Solid Phase Extraction (SPE). The extraction was done using styrene-divinylbenzene copolymer cartridges (SDB 1) of 6 mL volume and 200 mg sorbent with SPE devices. To prevent a blocking of the cartridge caused by suspended solids, sea sand of ∼0.5 cm bed thickness was placed on the top of the column to act as a filter aid. The conditioning was performed by rinsing the cartridge with 2 × 10 mL aliquots of acetone, followed by 10 mL of water acidified with sulfuric acid to pH 2. Sufficient water was kept in the cartridge (water level just above the packing) to keep Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
6767
Table 2. Overview of All Results from the Interlaboratory Validation of ISO/CD 18857-2 for Alkylphenols, Mono- and Diethoxylates, and Bisphenol A in Spiked Surface Watersa spiked surface water [ng/L] laboratory no.
sample
OP
OP1EO
OP2EO
NP
NP1EO
NP2EO
BPA
1
a b a b a b a b a b a b a b a b a b a b a b a b a b
61 62 49 50 58 56 55 55 38 40 47 45 63 64 41 39 70 68 41 45 53 50 69 70 54 54 50
93 95 85 86 83 87 87 90 68 70 75 76 94 95 56 53 98 95 79 82 88 90 142 146 90 93 88
76 74 68 68 67 69 75 75 44 50 216 175 76 79 60 66 65 66 58 57 65 70 88 93 69 70 68
167 203 204 220 154 162 212 210 157 161 251 279 190 190 313 324 248 192 163 175 169 155 377 341 199 203 150
269 271 250 238 224 202 229 233 189 195 185 167 252 250 301 262 282 335 165 176 163 164 331 338 215 211 190
287 277 299 287 303 297 372 388 252 270 825 901 355 373 303 294 707 641 297 296 261 247 467 519 343 325 290
62 61 89 82 84 84 81 84 55 65
2 3 4 5 6 7 8 9 10 11 12 13 expected value a
85 86 56 54 82 78 124 127 79 83 103 104 86 92 80
Outliers excluded from statistical evaluations are printed in bold.
Table 3. Overview of All Results from the Interlaboratory Validation of ISO/CD 18857-2 for Alkylphenols, Mono- and Diethoxylates, and Bisphenol A in Spiked Waste Watersa spiked waste water [µg/L] laboratory no.
sample
OP
OP1EO
OP2EO
NP
NP1EO
NP2EO
BPA
1
a b a b a b a b a b a b a b a b a b a b a b a b a b
1.72 1.68 1.59 1.57 1.59 1.61 1.51 1.58 1.20 1.30 1.34 1.33 1.65 1.62 0.68 0.59 1.84 1.78 1.30 1.50 1.59 1.63 2.34 2.26 1.63 1.61 1.5
1.35 1.33 1.31 1.30 1.29 1.29 1.25 1.27 1.10 1.10 1.03 1.07 1.41 1.50 0.96 0.85 1.55 1.56 1.30 1.30 1.29 1.33 1.96 1.89 1.32 1.31 1.3
1.01 1.03 0.90 0.90 0.90 0.88 0.85 0.88 0.60 0.70 0.75 0.71 0.96 0.97 0.74 0.82 1.03 1.06 0.70 0.80 0.89 0.91 1.33 1.31 0.93 0.93 0.9
3.94 3.67 4.07 4.39 3.53 3.32 3.55 3.77 2.90 3.00 3.66 3.44 4.00 3.90 6.80 6.80 4.31 4.18 2.70 2.90 3.92 3.61 5.42 4.87 3.84 4.35 3.5
4.74 4.47 4.04 4.17 3.95 3.98 3. 47 3.39 3.20 3.30 3.66 3.46 4.38 4.36 3.40 2.90 4.12 4.24 3.60 3.30 3.81 4.03 6.42 6.09 4.86 4.99 4.1
3.39 3.42 3.92 3.93 3.86 3.82 3.19 3.21 3.10 3.10 3.51 3.38 4.07 4.11 3.10 2.50 5.35 5.65 3.20 3.20 3.49 3.93 5.50 5.35 4.33 4.28 3.8
1.00 1.02 1.23 1.25 1.12 1.13 1.29 1.33 1.00 1.00 0.80 0.80 1.52 1.51 0.79 0.79 1.27 1.20 2.20 2.20 1.31 1.26 1.70 1.68 1.12 1.16 1. 2
2 3 4 5 6 7 8 9 10 11 12 13 expected value a
Outliers excluded from statistical evaluations are printed in bold.
the sorbent activated. The extraction was started immediately after conditioning. The water samples were run through the cartridges with a flow rate of 5-10 mL/min. After extraction, the cartridges were rinsed with 10 mL of water acidified with sulfuric acid to pH 2. Removal of the residual water and drying 6768
Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
of the sorbent packing was done by passing nitrogen with a flow rate of about 500 mL/min through the cartridges for approximately 1 h or by drawing air through the cartridges for the same duration. The end of the drying process can usually be recognized by brightening of the sorbent packing
Table 4. Performance Data from the Interlaboratory Validation of ISO/CD 18857-2 for Alkylphenols, Mono- and Diethoxylates, and Bisphenol A in Spiked Surface and Waste Waters matrix
analyte
no. of laboratoriesa
results useda
outliers [%]
total mean
spiked surface water [ng/L]
OP OP1EO OP2EO NP NP1EO NP2EO BPA
13 11 12 13 13 10 12
26 22 24 26 26 20 24
0 15.4 7.7 0 0 23.1 0
53.7 86.2 68.7 216.1 234.5 306.3 82.8
spiked waste water [µg/L]
OP OP1EO OP2EO NP NP1EO NP2EO BPA
11 12 12 12 12 13 12
22 24 24 24 24 26 24
15.4 7.7 7.7 7.7 7.7 0 7.7
a
1.55 1.27 0.87 3.8 3.91 3.84 1.18
expected value 50 88 68 150 190 290 80 1.5 1.3 0.9 3.5 4.1 3.8 1.2
recovery [%]
CVR [%]
CVr [%]
107.3 98.0 101.0 144.1 123.4 105.6 103.4
19.0 10.0 15.9 29.5 23.3 13.6 23.4
2.5 2.0 3.4 7.8 6.2 3.1 3.8
103.6 97.3 96.5 108.6 95.4 101.1 98.2
10.8 13.9 14.1 16.8 14.5 21.9 22.5
3.5 2.5 4.1 5.5 3.9 4.2 1.9
Outliers excluded.
of the cartridge. The color of the moist adsorbent is brown, and the dry material is light orange. For the elution of the analytes, 1 mL of acetone was added to the dried cartridge and allowed to equilibrate, for example, 10 min. Five aliquots of 1 mL of acetone were used to elute the cartridge. All eluates were collected in a pear shaped flask and evaporated to dryness using a moderate stream of nitrogen blowing on the surface of the extract. Derivatization. The dry residues were reconstituted within 200 µL of acetone. For derivatization of the hydroxyl groups, 25 µL of 2,2,2-trifluoro-N-methyl-N-(trimethylsilyl)acetamide (MSTFA) were added to the acetone solution of the sample extract and mixed gently. After a reaction time of at least 30 s at room temperature, the mixture was transferred to a suitable vial and was ready for measurement. Detection and Quantitation. Extracts were analyzed using gas chromatographic (GC) systems equipped with a mass spectrometric detector (MSD) that was operated in the selected ion monitoring (SIM) mode after electron impact (EI) ionization by all participants. Ion masses for the detection of analytes are listed in Table 1. Compounds were identified by matching both the retention times and the relative intensities of the diagnostic ions (Table 1) of sample components and reference substances. The compounds in the sample were identified, only if the following conditions were fulfilled: (1) the relative or the absolute retention time of the sample component matched that of the authentic compound within a limit deviation of ±0.02 min in the chromatogram of the latest calibration standard, measured under identical conditions; (2) the three selected diagnostic ions (Table 1) were present at the substance specific retention time; and (3) the relative intensities of all selected diagnostic ions measured in the sample do not deviate by more than ±(0.1I + 10) % from the relative intensities determined in the reference standard working solution (I is the relative intensity of the diagnostic ion in the reference standard working solution). For example, three selected diagnostic ions, I1, I2, I3, have the following relative intensities: 100%, 50%, and 15%. Than the maximum allowed deviation for I2 in the sample is (I1 is by definition 100% in both the sample and reference standard) ±(0.1 × 50 + 10) % ) ±15%. That means the relative intensity for I2 shall lie between 35% and 65%. For quantitation of the technical products NP, NP1EO, and NP2EO,
the entire ensemble of main isomers were integrated. In these cases a fixed ratio of the relative intensities of the selected diagnostic ions could not obtained. The concentration of target analytes was calculated based on recoveries of the internal standards. A set of four calibration standard solutions covering the expected concentration range was used for quantitation. Standards and Chemicals. Chemicals used were obtained as follows: OP1EO, OP2EO, NP1EO, NP2EO (Dr. Ehrenstorfer, Germany), OP-13C6, OP1EO-13C6, OP2EO-13C6, 363-NP-13C6, 363NP1EO-13C6, 363-NP2EO-13C6, (Hanse-Analytics, Germany), acetone, sulfuric acid, analytical grade, BPA, BPA-d16 (SigmaAldrich, Germany), OP, NP (Sasol, Germany), MSTFA (Macherey-Nagel, Germany), SPE cartridges, styrene-divinylbenzene copolymer, SDB 1 (Mallinckrodt Baker, Germany). All standard solutions were prepared in acetone. RESULTS This interlaboratory trial on the analysis of alkylphenols, their ethoxylates, and bisphenol A according to ISO 18857-2 was evaluated according to ISO 5725-2. The study included the duplicate analysis of two different types of nonfiltered water samples. The samples were a surface water containing the target compounds in a concentration range from 0.05 to 0.4 µg/L and a wastewater containing the target compounds in a concentration range from 0.1 to 5 µg/L. The results of this study are presented in Tables 2-4. A survey on the performance data of the analytical method is given in Table 4 for both sample types. This table also contains information on the number of results used for the statistical calculations, the expected concentrations of target analytes in the samples, and the recovery rates expressing the accuracy of the method. The precision of the method is determined by the reproducibility variation coefficient CVR as a measure of the comparability between the different laboratories and the repeatability variation coefficient CVr as a measure of internal laboratory precision. Outliers of type B, which have been recognized by statistical evaluation, are printed in bold. These outliers are data pairs whose mean value significantly deviates from the other laboratory means. They are identified by Grubbs’ test. These values were excluded from further calculations. Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
6769
Figure 1. Diagnostic ions used for quantitation (m/z 207) and for confirmation (m/z 221, 193, and 179) of technical nonylphenol.
Accuracy and Precision. As shown in Table 4, the number of results used for statistical evaluations reached from 20 to 26, with outlier rates less than or equal to 2 samples for most of the compounds. The recoveries of all target analytes in the wastewater sample were very close to 100%, covering a range of 95.4-108.6%. The repeatability variation coefficient (within-laboratory precision) CVr was in the range of 1.9-5.5% indicating that the participating laboratories were able to generate similar results applying this method. The reproducibility variation coefficient (betweenlaboratory precision) CVR was, as usual, higher than the repeatability, reaching from 10.8 up to 22.5%, whereby the reproducibility of most compounds was below 20%. It can be concluded that the method is well suited to archive results of high accuracy and sufficient precision for wastewater samples. In case of the surface water sample, the recoveries ranged from 98.0 to 144.1%, with most of the values being close to expected concentrations. Considerable higher recoveries of 6770
Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
144.1% and 123.4% were found only for NP and NP1EO, respectively. It has to be mentioned that NP (besides OP and BPA) usually show the highest blank concentrations in the analytical procedure due to their ubiquitous presence (for example, in plastics, solvents, or detergents). A further problem, which can lead to erroneous results is associated with the quantitation of the technical products NP, NP1EO, and NP2EO. For these compounds, the quantitation is based on the integration of an ensemble of peaks consisting of the different branched isomers. Figure 1 shows the ion traces which are used for quantitation and for confirmation of technical NP. The quantitation of NP includes the integration of all isomers within the time range from 13.30 to 13.95 min (beside this time window there are some more isomers, but these isomers are less abundant and are not used for quantitation). Errors in quantitation can emerge from coeluting compounds, which are mistakenly considered as isomers of the technical
Figure 2. Chromatogram of a calibration standard (A), of a spiked wastewater (B), and of a spiked surface water sample (C). The labeled peaks show the target analytes and the coeluting internal standards.
mixture and integrated with the other isomers. The peak pattern of NP1EO and NP2EO is similar to that of NP, only
the masses of the detected ions are higher, with the mass difference being 44 amu for each ethoxy unit (see diagnostic Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
6771
Figure 3. GC/MS of technical grade nonylphenol and the detailed mass spectrum of isomer 363-NP.
Figure 4. Molecular structures of
13
C-labeled octyl- and nonylphenol and their mono- and diethoxylates.
ions in Table 1). So the same risks in the quantitation of these compounds can arise as in case of NP. The higher the content of matrix loading in the sample extract, the more caution has to be taken by the analyst not to integrate coeluting compounds, which is illustrated in Figure 2 by chromatograms of different sample types (as total ion chromatogram, TIC). Figure 2A shows the chromatogram of a pure calibration standard, where the baseline is close to zero and no peaks except the peaks of the target analytes are detected. In Figure 2B, a chromatogram of the spiked wastewater sample is shown. Only a few more peaks can be seen in this sample compared to the standard. In the chromatogram of the surface water sample (Figure 2C), a high unresolved underground is apparent. This underground is made up by coextractable matrix compounds. It is conceivable that compounds from this underground may have caused interferences, which probably led to enhanced recoveries of 144.1% for NP and of 123.4% for NP1EO in the surface water samples. Substances coeluting with nonylphenol and its monoethoxylate can interfere in the determination. This may have a large influence on the result, since 6772
Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
both analytes are quantified from the sum of all peaks belonging to the chromatographic pattern. There are unresolved peaks which cannot easily properly be assigned to the analyte. This emphasizes again that the analyst should use great caution to include only those peaks from the sample that are attributable to the multicomponent analyte. As in the case of the wastewater sample, the repeatability variation coefficient (within-laboratory precision) CVr was good with values being in the range of 2.0-7.8%. The reproducibility variation coefficient (between-laboratory precision) CVR reached from 10.0 to 29.5%. Again the reproducibility for most compounds was below 20%, while the highest deviation was estimated for NP, probably due to reasons mentioned above. Outliers Discussion. Whenever outliers are recognized in the course of a method validation, the reasons for their occurrence (11) Heemken, O. P.; Amann, N. Analytix 2007, 4, 6–7. (12) International Standardization Organization. ISO 18857-2; Geneva, 2009, in press.
should be clarified. As shown in Tables 2 and 3, laboratory 12 has categorically found too high concentrations for all compounds in both sample types. Six of these values have been recognized to be statistical outliers of type B and were therefore excluded from further calculations. It can be assumed that a systematic error led to these erroneous results. Such systematic deviations can result from adding wrong volumes of the spike solution or by mistakes in preparing the calibration standards or can be caused by errors in diluting the standards. From the other outliers identified, two laboratory means were below the expected values, while five mean values were higher. These seven outliers were generated by four laboratories, whereby four outliers were found in the surface water and three outliers were associated with the wastewater sample. In case of laboratory 6, the outliers were found for the diethoxylates of OP and NP in the surface water sample. In both cases the results were nearly 3-fold higher than the expected values. Because the results for the diethoxylates in the wastewater samples of this laboratory were in good agreement with the expected values, it is more probable that this was a random deviation and not a systematic error. From the remaining three outliers (laboratories 8, 9, and 10), no other conclusions than random errors can be drawn. Internal Standards Used. One reason for the relatively high accuracy and precision of the results as compared to other interlaboratory studies of these compounds7,8 might be attributed to the fact that for each target analyte, the corresponding isotopemarked compound was used as an internal standard. From former interlaboratory trials performed by the DIN working group during the process of method development, it was well-known that the usage of n-isomers as internal standards (n-OP, n-NP, and n-NP1EO) sometimes led to high analytical errors in the quantitation of the target analytes. The reason lies in the different chemical/physical properties of the n-isomers compared with the branched compounds, especially regarding the extraction and elution behavior during SPE and adsorption or discrimination effects of higher boiling compounds in GC. This is particularly substantial for the analysis of the diethoxylates, where suitable internal standards were completely missing so far. For this reason, isotope-marked compounds had been synthesized.11 One of the most prevalent isomers in technical grade NP is 363-NP, the structure and MS spectrum of which appear in Figure 3. The mono- and diethoxylates of NP behave similarly, with 363-NP1EO and 363-NP2EO belonging to the major isomers. The high percentage and the favorable fragmentation pattern of these isomers make them ideal internal standards for the analysis of technical NP and its ethoxylate derivatives. The 13C-ring-labeled 363-isomers of NP and its mono- and diethoxylates and additionally for the analysis of OP, OP1EO, and OP2EO their 13C-ring-labeled analogous were synthesized. An overview of these compounds used as internal standards in this interlaboratory trial is given in Figure 4. For the analysis of BPA, the per-deuterated compound BPA-d16 was used. As shown in course of this study, the recovery rates of the internal standards used were nearly identical to those of the target
analytes, which means that they have high conformability with the chemical/physical properties of the target analytes and therefore own good suitability for usage as internal standards. After this study, Sigma-Aldrich launched the 13C-ring-labeled internal standards used here. CONCLUSION Under ISO/CD 18857-2, a novel method for the determination of octyl- and nonylphenol, their mono- and diethoxlyates, and bisphenol A in water samples has been developed. The method consists of a solid phase extraction, an elution and derivatization step, and the measurement with GC/MSD. For a final evaluation, this method has successfully been validated by an interlaboratory trial. A total of 13 laboratories participated in this study, whereby two different types of water samples (surface and wastewater) were analyzed in duplicate. Results were checked for statistical outliers, which were rejected from further calculations. The results from the interlaboratory trial showed that the method ISO/CD 18857-2 represents a robust procedure for the quantitative analysis of octyl- and nonylphenol and their monoand diethoxlyates in both surface and wastewater samples. Bisphenol A can also be determined by this method. Results obtained were of good accuracy and precision. The occurrence of outliers could be explained by a constant systematic error caused by one laboratory, and the remaining outliers were probably caused by random deviations. In the analysis of alkylphenols and derivates, caution must be taken for contaminations stemming from the laboratory environment, the chemicals and solvents used, as well as the labware. False positive results especially for technical NP, NP1EO, and NP2EO can also result from compounds interfering with the peak pattern of the target analytes that is used for quantitation. This problem may arise when the matrix loading of the sample is high. The isotope-marked standard compounds developed in this context have been proven to be reliable internal standards that allow a precise and accurate quantitation of all compounds specified in ISO/CD 18857-2. Meanwhile, the draft standard method has been approved by the ISO members eligible and interested in casting a vote. The vote passed with broad participation and no negative votes, and now the official text of the standard will be finalized and issued.12 ACKNOWLEDGMENT The authors would like to thank the German Chemical Society (GDCh) for financial support of this interlaboratory trial. Mallinckrodt Baker is kindly acknowledged for providing SPE cartridges during the entire study. The authors also would like to thank the State Reservoir Administration of Saxony, Laboratory Plauen, for assistance with the sampling. All participating laboratories of this study are gratefully acknowledged for their cooperation. Received for review April 16, 2009. Accepted July 9, 2009. AC900813M
Analytical Chemistry, Vol. 81, No. 16, August 15, 2009
6773
