Quaternary Triphenylphosphonium Compounds: A New Class of

Nov 10, 2015 - (7) Hence, various classes of emerging contaminants such as pharmaceuticals, estrogens, pesticides, ingredients of personal care produc...
11 downloads 21 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Quaternary Triphenylphosphonium Compounds: A New Class of Environmental Pollutants Michael P. Schlüsener, Uwe Kunkel, and Thomas A. Ternes* Federal Institute of Hydrology, Department of Aquatic Chemistry, 56068 Koblenz, Germany S Supporting Information *

ABSTRACT: A nontarget screening using high-resolution mass spectrometry (HRMS) was established to identify industrial emerging contaminants in the Rhine River. With this approach, quaternary triphenylphosphonium compounds (R-Ph3P+) were identified as new emerging contaminants in the aquatic environment. The suggested chemical structures were elucidated by MS fragmentation and chemical databank searches and eventually confirmed via authentic standards. R-Ph3P+ are used worldwide by the chemical industry to synthesize alkenes via the Wittig reaction. In total, five compounds [R = butyl (Bu), R = ethyl (Et), R = methoxymethyl (MeOMe), R = methyl (Me), and R = phenyl (Ph)] were found in German rivers and streams. R-Ph3P+ were detected only in those rivers and streams that received an appreciable portion of wastewater from the chemical industry. Up to 2.5 μg/L Et-Ph3P+ was quantified in a small stream from the Hessian Ried, and in the Rhine, up to 0.56 μg/L Me-Ph3P+ was detected. R-Ph3P+ were also identified in suspended particulate matter and sediments in the Rhine catchment, with MeOMe-Ph3P+ concentrations of up to 0.75 mg/kg and up to 0.21 mg/kg, respectively. Because of the lack of ecotoxicological studies, the environmental risks caused by R-Ph3P+ can be assessed for neither pelagic nor benthic organisms.

■

INTRODUCTION The contamination of water bodies with a multitude of emerging contaminants is one of the key problems facing water management today.1 Although most of these contaminants are present at low concentrations, many of them raise considerable (eco)toxicological concerns, particularly because they are present in complex mixtures.2−4 More than 100000 chemicals are registered in the European Union (data from August 6, 2015), of which 30000, in some cases up to 70000, are used daily.5 Many of these chemicals can enter the aquatic environment via a variety of diffuse (e.g., urban and agricultural runoff)6 and point sources [e.g., wastewater treatment plants (WWTPs)].7 Hence, various classes of emerging contaminants such as pharmaceuticals, estrogens, pesticides, ingredients of personal care products, household chemicals, or biocides have been detected worldwide in rivers and streams.8−13 In recent years, a variety of starting materials of chemical syntheses as well as synthesis byproducts have also been found in the aquatic environment.14 For example, the detection in German rivers of elevated concentrations of hexa(methoxymethyl)melamine, used in the production of coatings and plastics, was attributed to the discharge of wastewater from the automobile industry.15 Tetraglyme, which is used during separation processes and high-temperature reactions, was detected in the Oder River with high spatial and temporal fluctuation. The pollution was caused by the usage of tetraglyme during the removal of SO2 from flue gases.16 The © 2015 American Chemical Society

occurrence of inorganic emerging contaminants can also be attributed to industrial production sites as reported for lanthanum and samarium in the Rhine.17 The high spatial and temporal occurrence of these lanthanides was attributed to the discharges from a company that produces catalysts for petroleum refining. On the other hand, constant contamination levels of industrial chemicals are also known, as reported for 2,4,7,9-tetramethyl-5-decyne-4,7-diol (TMDD), a nonionic surfactant used, e.g., as an industrial defoaming agent. Because of continuous discharges by industrial WWTPs, the concentrations of this compound were constant in the Rhine, with an average concentration of 0.51 μg/L.18 One powerful tool for detecting and identifying emerging organic contaminants in the environment is nontarget analysis.19 This technique covers all substances that are accessible with the chosen analytical detection method. For polar surface water contaminants, the usage of liquid chromatography−high-resolution mass spectrometry (LC− HRMS) systems is currently seen as the most powerful approach.20 As a first result, the nontarget analysis provides only a list of “features” that are defined by the retention times (RTs), the high-resolution mass to charge ratios (m/z), signal Received: Revised: Accepted: Published: 14282

August 18, 2015 October 21, 2015 October 26, 2015 November 10, 2015 DOI: 10.1021/acs.est.5b03926 Environ. Sci. Technol. 2015, 49, 14282−14291

Article

Environmental Science & Technology intensities, and preferably additional MSn spectra. This precursory information has then to be further evaluated and processed with the assistance of different tools such as chemical databases (e.g., Chemspider21), in silico fragmentation software (e.g., Metfrag22), or high-resolution MS/MS libraries (e.g., MassBank23). To prove the correctness of contaminant identification, in most cases, further analytical methods such as nuclear magnetic resonance or the availability of the authentic reference standard are necessary.24 In recent years, several substances, previously not widely regarded to be present in wastewater, were detected via nontarget screening in treated wastewater and in the aquatic environment. In a screening of a WWTP effluent, nine nontarget compounds were identified, of which at least seven are known to be used predominantly in chemical processes.25 In a recent work, various biological transformation products (TPs) of different sartans (angiotensin receptor blockers) were identified in WWTP effluents and receiving waters.26 Additional nontarget studies of biologically treated wastewater confirmed, for example, the presence of an oxidation product of a vulcanization accelerator27 or the presence of glycol ether sulfate surfactants.28 In contrast to the nontarget studies mentioned above, our main aim was not to identify as many compounds as possible (targets, suspects, and nontargets) but rather to use the LC− HRMS measurements as a screening and prioritization tool on a long times series of samples from one sampling site. To this end, nontarget screening via LC−HRMS/MS was performed on daily composite water samples taken for 14 months from the Rhine at Koblenz [river kilometer index (RKI) 590.3]. At this RKI, the Rhine has received wastewater from a multiplicity of different chemical industries and municipal wastewater from a total of ∼26 million people.29 To identify emissions by industrial WWTPs, this study was focused on features whose intensities varied substantially over the time course of the sampling campaign.

water samples were taken from June 14 to 17, 2015, in the center of the rivers directly below the water surface (unless stated otherwise). The sampling sites are listed in Table S1 of the Supporting Information. The samples were immediately cooled to 4 °C and measured within 24 h via LC−HRMS without any additional sample pretreatment. Effluents of the municipal WWTPs were sampled (Table S2 of the Supporting Information), and samples of sediments (SD) and suspended particulate matter (SPM) were taken from the Rhine and its tributaries (see Table S3 of the Supporting Information). SD was taken from the sediment surface (0−20 cm) using a Van Veen grab sampler. SPM was taken by continuous-flow centrifuges within a time period of 6−8 h. All solid samples were freeze-dried and stored at room temperature until they were extracted. Pressurized Liquid Extraction (PLE). Up to 2 g of the freeze-dried suspended solids or sediments was transferred into 10 mL PLE (ASE 350, ThermoFischer, Dreieich, Germany) extraction cells prefilled with quartz sand to reduce the void volume. The cells were sealed with circular cellulose filters (Restek, Bad Homburg, Germany) at both ends. PLE was accomplished by a first extraction with an acidified [1% (v/v) FA] methanol/water mixture [1:1 (v/v)] followed by a second extraction step with pure methanol. For both steps, PLE conditions were as follows: static, 5 min; flush, 150%; purge, 120 s; three cycles; pressure, 100 bar; temperature, 100 °C. The extracts were filled up to 150 mL with ultrapure water (to minimize the proportion of methanol in the extract and avoid broadening peaks during the LC run) and measured via LC− HRMS as described below. All SD and SPM samples were extracted and analyzed in triplicate. LC−HRMS. The LC system consisted of a G1367E auto sampler, a G1330B cooling thermostat for the autosampler, a G1312B binary LC pump, a G1379B membrane degasser, and a G1316A column oven (all from Agilent, Waldbronn, Germany). Separation was achieved using a Zorbax Eclipse Plus C18 column (2.1 mm × 150 mm, 3.5 μm, Agilent) equipped with a Security Guard (2.0 mm × 4.0 mm, AQ C18, Phenomenex, Aschaffenburg, Germany) at 40 °C. The flow rate was set to 300 μL/min using a binary gradient of mobile phase A [ultrapure water with 0.1% (v/v) FA] and phase B [acetonitrile with 0.1% (v/v) FA]. Chromatographic runs were made with the following gradient: from 0 to 1.0 min, 2% B; from 1.0 to 2.0 min, 2 to 20% B; from 2.0 to 16.5 min, 20 to 98% B; from 16.5 to 22.0 min, 98% B; from 22.0 to 22.1 min, 98 to 2% B; from 22.1 to 27.0 min, 2% B. The injection volume was 100 μL for water samples and 10 μL for extracts of SPM and SD. To protect the HRMS system, a postcolumn divert valve (Rheodyne, Darmstadt, Germany) directed the LC flow into the waste from 0.0 to 2.0 min and from 20.0 to 27.0 min. Thus, only peaks eluting between 2 and 20.0 min were directed to the HRMS system. To compensate for the missing flow when the LC flow was discharged into the waste, an additional flow of a 300 μL/min ultrapure water/acetonitrile mixture [1:1 (v/v)] was pumped by an Agilent G1311B quaternary HPLC pump (Agilent). For mass spectrometric detection, a SCIEX (TripleToF 5600, Darmstadt, Germany) hybrid quadrupole time-of-flight mass spectrometer (QToF) system equipped with a DuoSpray ion source and a TurboIonSpray probe for ESI experiments was used in positive and negative ionization mode. The parameters for positive ionization were as follows (deviating values for

■

MATERIALS AND METHODS Chemicals. A list of used chemicals and analytical standards is provided in the Supporting Information. Sampling. To determine the occurrence of organic emerging contaminants in the Rhine, 24 h time-proportional composite samples were taken at the international monitoring station of the ICPR (International Commission for the Protection of the Rhine) at RKI 590.3 near Koblenz, Germany, from January 2014 to March 2015. The water samples were taken by an automated sampling device (SP5 A, MAXX Mess-u. Probenahmetechnik GmbH, Rangendingen, Germany) on the left bank of the Rhine approximately 30−50 cm below the surface water level and stored at 4 °C. Before the measurement, 980 μL of the water sample was transferred to a highperformance liquid chromatography (HPLC) vial and 10 μL of an isotopically labeled internal standard mix containing carbamazepine-13C115N1, and diuron-d6 (cfinal = 0.2 μg/L each) and 10 μL of 10% formic acid (FA) were added. To identify sources of emerging contaminants detected at RKI 590.3, grab samples from the Rhine as well as its major tributaries (Main, Nahe, Neckar, and Lahn) were taken upstream of Koblenz. Furthermore, grab samples were also collected from streams and small rivers in the Hessian Ried located in the northeastern part of the Upper Rhine Plain. Frequently, these rivers and streams have an elevated proportion of treated wastewater (>50%).30 The area receives municipal wastewater from ∼2 million inhabitants. All surface 14283

DOI: 10.1021/acs.est.5b03926 Environ. Sci. Technol. 2015, 49, 14282−14291

Article

Environmental Science & Technology

m/z and RT) at every sampling date is assigned a respective signal intensity that can be used as a proxy for the concentration of this substance. To compensate for long-term alteration of the LC-QToF system, all peak intensities were normalized by the intensity of two internal standards: carbamazepine-13C115N1 (positive ionization mode, feature at m/z 239.087, RT = 9.0 min) and diuron-d6 (negative ionization mode, feature at m/z 237.049, RT = 10.4 min). To search for features with large variations in signal intensity over time, the time trends of the normalized intensities of each feature were visualized using the software package R.32 To further identify discontinuous emissions of organic contaminants, plots of the autocorrelation function of each feature were scanned for periodic variations in the time series. Most pronounced variations of the intensity over time were observed for the feature at m/z 307.125 at a RT of 8.0 min (Figure 1A), with 13 discrete maxima over 14 months. The

negative ion mode in parentheses): ion source gas (GS) 1 and 2, 35 and 45 psi, respectively; curtain gas (CUR), 40 psi; source temperature (TEM), 550 °C; ion spray voltage floating (ISVF), 5500 V (−4500 V); declustering potential (DP), 60 V (−100 V); ion release delay (IRD), 67 ms; ion release width (IRW), 25 ms. The collision energy (CE) was set to 40 eV with a collision energy spread (CES) of 15 eV. The mass spectrometer was recalibrated automatically every five measurements using an automated calibration delivery system (CDS) via the APCI (atmospheric-pressure chemical ionization) probe of the DuoSpray source. A full scan experiment (from 100 to 1200 Da) was performed with an accumulation time of 200 ms. Additional product ion scans (acquisition time of 50 ms, mass range of 30−1200 Da) of the eight highest peaks of the full scan experiment were acquired via independent data acquisition (IDA) using the high-sensitivity mode. Ions were excluded after six independent MS2 experiments for 20 s. Isotopes in the range of 4 Da (±5 ppm) were excluded from IDA experiments. The resulting total cycle time was 650 ms. Ions at m/z 214.090 and 221.190 were excluded in positive mode and ions at m/z 221.075 in negative mode for all IDA experiments because they were background contaminants. MS data acquisition was controlled with Analyst TF version 1.5.1 (SCIEX). Peak Inventory List. Peak picking and alignment were performed with MarkerView version 1.2.1.1 (SCIEX) using the following parameters: lowest RT, 2.0 min; highest RT, 20.0 min; subtraction offset, 10 scans; subtraction multiplication factor, 1.8; noise threshold, 10; minimal spectral peak width, 80 ppm; minimal RT peak width, four scans; RT tolerance, 0.30 min; mass tolerance, 50 ppm; use global exclusion list, false; number of required samples, three; maximal number of peaks, 9000; use raw data area, true. Quantification, Validation Experiments, and Statistical Analysis. Quantification was conducted using the full scan data of the LC-QToF-MS analysis by integrating the peak areas of the extracted ion chromatograms of the exact masses (±0.01 Da) with Multiquant version 3.0 (SCIEX). Calibration was linear from the low nanogram per liter range up to 30 μg/L (Figure S7 of the Supporting Information). The quantification method was validated by spiking three unpolluted surface water samples with the stock solutions of R-Ph3P+ to adjust concentrations of 0.1 and 1 μg/L. In addition, all SD and SPM samples were spiked with 1 μg of each R-Ph 3P+ compound prior to PLE extraction. The limits of quantification (LOQs) were defined at a signal-to-noise ratio of 10:1 in the respective matrix (water or PLE extract). Carbamazepine was simultaneously quantified in the surface waters according to the analytical method described by Richter et al.31 Correlations of the concentrations of substances in the surface waters were calculated on the basis of a two-sided Spearman’s rank correlation test (R package “pspearman”).

Figure 1. Time trends of the normalized peaks areas of two features in daily composite water samples and average discharge of the Rhine at Koblenz (RKI 590.3) from January 2014 to March 2015: (A) m/z 307.125 and RT = 8.0 min and (B) carbamazepine.

normalized peak areas of the maxima (>2.0) were more than 1 order of magnitude higher than the normalized peak areas of the respective minima (1, in total 28 sum formulas (as [M + H+]+) were suggested (see Table S4 of the Supporting Information). The five best matches based on MS and MS/MS spectra were C 11 H 21 FN 4 OP 2 (0.2 ppm), C16 H 23 NOP 2 (−0.5 ppm), C20H19OP (0.6 ppm), C11H16F3N5O2 (−0.9 ppm), and C15H19FN3OP (1.2 ppm). In addition, the substance database STOFF-IDENT (RISK-IDENT, http://risk-ident.hswt.de/ pages/de/start.php, retrieved on July 27, 2015)34 containing high-resolution masses of potential organic water contaminants was used as a platform to identify the elemental composition of the substance at m/z 307.125 and RT = 8.0 min. Setting the mass error to 10 ppm and searching for [M + H+]+ resulted in zero hits. Searching for [M]+ resulted in only one hit, which was methoxymethyltriphenylphosphonium (MeOMe-Ph3P+, CAS Registry No. 4009-98-7 for chloride salt). Subsequently, MeOMe-Ph3PCl was purchased as an authentic standard, and the proposed chemical structure could be confirmed. MeOMe-Ph3P+ is one representative of an array of available R-Ph3P+ compounds that are used by the chemical industry worldwide to synthesize alkenes via Wittig reactions (Figure 2). In the Wittig reaction, an aldehyde or ketone is reacting with a phosphorus ylide to yield olefins.35 By the Wittig reaction, compounds such as vitamin A, β-carotene, prostaglandins, and several pharmaceuticals are synthesized.36 Triphenylphosphine oxide (TPPO) is known as a byproduct of the Wittig reaction. TPPO can also be formed by the oxidation of TPP in the presence of oxygen, i.e., in water. In surface waters, TPPO is further transformed to diphenylphosphine oxide (DPPO).37 Using the fragment matching tool of PeakView, which regards similarities in both neutral losses and fragment masses in the MS2 spectra, the time series of nontarget data from the Rhine at Koblenz were retrospectively evaluated to identify further R-Ph3P+ compounds. On the basis of the fragmentation of MeOMe-Ph3P+ (Figure 3), other R-Ph3P+ should exhibit

Figure 3. MS2 spectrum of the feature at m/z 307.125 and RT = 8.0 min. Fragments 1−3 are common MS2 fragments of R-Ph3P+.

similar MS2 spectra, i.e., fragments consisting of the central phosphorus atom with one, two, or three phenyl rings attached. Hence, the characteristic common fragments of R-Ph3P+ can be assumed to be m/z 108.0123 (C6H5P+), m/z 183.0358 (C 12 H 8 P + ), m/z 185.0515 (C 12 H 10 P + ), m/z 261.0828 (C18H14P+), m/z 262.0906 (C18H15P+), and m/z 263.0984 (C18H16P+) (marked as 1−3 in Figure 3). With this approach, two additional proposed R-Ph3P+ were found in Rhine water at m/z 277.114 and RT = 7.5 min and m/z 291.130 and RT = 7.9 min (Figure S2 of the Supporting Information). Via commercially available authentic standards, the identification of methyltriphenylphosphonium [Me-Ph3P+ (Figure S3 of the Supporting Information, also present in the STOFF-IDENT database)] and ethyltriphenylphosphonium [Et-Ph3P+ (Figure S4 of the Supporting Information)] was confirmed. Because of the three highly apolar phenyl rings and the positive charge at the phosphorus atom, substantial sorption of R-Ph3P+ can be assumed. Therefore, also the SPM samples from the Rhine at RKI 590.3 were searched for characteristic fragments, and in addition to Me-Ph3P+, Et-Ph3P+, and MeOMe-Ph3P+, two further suspected R-Ph3P+ compounds (m/z 319.162 and RT = 8.9 min, and m/z 339.130 and RT = 9.0 min) were observed (Figures S5 and S6 of the Supporting Information). Via commercial standards, the two suspected R14285

DOI: 10.1021/acs.est.5b03926 Environ. Sci. Technol. 2015, 49, 14282−14291

Article

Environmental Science & Technology

Table 1. Identified R-Ph3P+ in Surface Water, Sediments, and Suspended Solids in the Rhine at RKI 590.3 (KOW values were estimated using KOWWIN version 1.6838)

1

All substances were provided by Sigma-Aldrich (Seelze, Germany). 2SD, sediment. 3SPM, suspended particulate matter.

Ph3P+ were confirmed to be butyltriphenylphosphonium (BuPh3P+) and tetraphenylphosphonium (Ph4P+). Thus, in total, five R-Ph3P+ were identified in surface water and SPM samples from the Rhine near Koblenz (Table 1). To the best of our knowledge, the occurrence of R-Ph3P+ in rivers and streams has not been described before in the literature. Validation of the Target Analytical Methods. The limits of quantification (LOQs) in water ranged from 0.01 to 0.03 μg/L and in SD and SPM samples from 0.005 to 0.008 mg/kg (Table S6 of the Supporting Information). The absolute recoveries for all five R-Ph3P+ varied in both matrices between 88 ± 22 and 117 ± 7% (Table S6 of the Supporting Information). Hence, no correction for losses during PLE extraction or due to matrix effects during LC−HRMS was applied. In addition to the R-Ph3P+, TPPO and DPPO were quantified in aqueous samples with recoveries of 100 ± 6 and 87 ± 9%, respectively, while in solid samples, the recoveries were 94 ± 5 and 32 ± 11%, respectively (Table S6 of the Supporting Information). Time Trends of Concentrations and Loads of R-Ph3P+ in the Rhine at Koblenz (RKI 590.3). From 2014 to March 2015, the daily loads of the dissolved Me-Ph3-P+, MeOMePh3P+, and Et-Ph3P+ in the Rhine at RKI 590.3 were calculated by a calibration using authentic standards (Figure 4). Annual

loads of individual R-Ph3P+ ranged from 740 kg/year (for EtPh3P+) and 2000 kg/year (for Me-Ph3P+) to 5400 kg/year (MeOMe-Ph3P+). The loads of MeOMe-PH3P+ exhibited significant fluctuations known from the discontinuous emission of chemicals via industrial WWTPs.16 During July and August 2014, MeOMe-Ph3P+ was not present at RKI 590.3, which is likely caused by a production stop over the summer vacation period of that year (Figure 4B). For Me-Ph3P+ and Et-Ph3P+, fluctuations were less pronounced (Figure 4A,C), which is most probably caused by the smaller total loads of these compounds compared to that of MeOMe-Ph3P+. Occurrence and Emissions of R-Ph3P+. Triphenylphosphonium cations (R-Ph3P+) were found at elevated concentrations in rivers and streams (Table 2) that receive a significant portion of wastewater from the chemical industry. As a consequence, several small streams located in the Hessian Ried were heavily polluted with R-Ph3P+, exhibiting concentrations of up to 2.5 μg/L (Et-Ph3P+) in the Landgraben (Table 2 and Figure S9 of the Supporting Information). The source of R-Ph3P+ in the Hessian Ried is located in the city of Darmstadt, because R-Ph3P+ were detected only in the Landgraben and Schwarzbach (sampled after the confluence with the Landgraben). R-Ph3P+ were not detected in other small streams of the Hessian Ried such as the Mühlbach or the Schwarzbach 14286

DOI: 10.1021/acs.est.5b03926 Environ. Sci. Technol. 2015, 49, 14282−14291

Article

Environmental Science & Technology

while on the left bank at Nierstein at RKI 483, only 0.17 μg/L MeOMe-Ph3P+ were found. Obviously, MeOMe-Ph3P+ was discharged by chemical industries located on the right bank between Worms and Kornsand. Additionally, Et-Ph3P+ and MeOMe-Ph3P+ were detected in the river Neckar at RKI 13, indicating an upstream emission close to the sampling site, because both compounds were not detected in the Neckar at Heidelberg (RKI 24). In contrast, RPh3P+ were not detected in other major tributaries of the Rhine, namely, the rivers Lahn, Main, and Nahe. Hence, industries located on these rivers obviously did not contribute to the pollution with R-Ph3P+ detected in Koblenz during the sampling campaign. However, it must be noted that only grab samples were taken during the sampling campaign for the source tracking of R-Ph3P+. While this sampling methodology is sufficient for identifying the source of a substance, the lack of substance detection in a grab sample is not necessarily evidence of its absence at other time points. Therefore, the possibility that additional sources for R-Ph3P+ exist that were not identified in the performed sampling campaign cannot be excluded. The R-Ph3P+ compounds were also detected in SD and SPM samples over a wide area in the Rhine catchment (Table S5 of the Supporting Information). The concentrations at Koblenz (RKI 590.3/591.4) ranged from 0.010 ± 0.002 mg/kg for BuPh3P+ (in SPM) to 0.75 ± 0.32 mg/kg for MeOMe-Ph3P+ (in SPM). Bu-Ph3P+ and Ph4P+ containing the most lipophilic side chains (R = butyl and phenyl) were not detected in the water phase, but up to 0.025 ± 0.005 and 0.085 ± 0.022 mg/kg were detected in SD and SPM, respectively. This is consistent with the high predicted log KOW values of 5.54 and 5.28 for SD and SPM, respectively (Table 1). In Landgraben sediment, MeOMe-Ph3P+ was even present at concentrations of 1.2 ± 0.3 mg/kg. In addition, the other four R-Ph3P+ were also detected at elevated concentrations, which confirms the high level of pollution of this small stream with R-Ph3P+. The RPh3P+ compounds were not detected in sediments from the River Main, while only Et-Ph3P+ was found upstream of the industrial center of Ludwigshafen/Mannheim. Thus, the contamination of R-Ph3P+ in Koblenz seems to be caused by two major sources; one is located between RKI 410 and RKI 483 and the other discharging through the Hessian Ried. Four of the five identified R-Ph3P+ were additionally found in the sediment of the Rhine at RKI 864.8, ∼275 km downstream of Koblenz and >400 km downstream of the pollution sources. While the sorbed concentrations of R-Ph3P+ were a factor of 3− 4 smaller than in Koblenz, the occurrence of R-Ph3P+ far from the emission sources verifies the long-range transport of these substances via water and SPM. Therefore, in addition to water, SD and SPM are an important matrix for providing evidence of contamination with R-Ph3P+. Sorption of R-Ph3P+. Apparent Kd values for Me-Ph3P+, MeOMe-Ph3P+, and Et-Ph3P+ were calculated on the basis of the corresponding concentrations (December 8, 2014, and January 5, 2015) measured in SPM and surface water of the Rhine at RKI 590.3 [Koblenz (Table S9 of the Supporting Information)]. The elevated average Kd values (4300 L/kg for MeOMe-Ph3P+, 4400 L/kg for Et-Ph3P+, and 1400 L/kg for Me-Ph3P+) confirmed the assessed high sorption affinity of RPh3P+ and a significant exposure potential of benthic organisms to R-Ph3P+. On the basis of the discharge and SPM at the Rhine at Koblenz, the total loads of R-Ph3P+ on December 8, 2014 (values for January 5, 2015, in parentheses), were 4.1 kg/day

Figure 4. Time trends of dissolved daily loads of R-Ph3P+ in the Rhine at Koblenz (RKI 590.3) from January 22, 2014, to March 8, 2015: (A) Me-Ph3P+, (B) MeOMe-Ph3P+, and (C) Et-Ph3P+.

upstream of Trebur, although they contain a very high proportion (>50%) of municipal wastewater (indicated by high concentrations of CBZ). The analysis of four municipal WWTP effluents underlines the conclusion that R-Ph3P+ are primarily discharged by the chemical industry (Table S6 of the Supporting Information). It has to be noted that R-Ph3P+ are used primarily for the Wittig reaction by the chemical industry, but also that other special applications such as their use as phase-transfer catalysts (e.g., for preparation of dyes)39 or to recover technetium from radioactive waste streams40 have been reported. Nevertheless, R-Ph3P+ should not be present in effluents of common municipal WWTPs. However, a contamination of municipal WWTPs cannot be totally ruled out as seen by the detection of 0.70 μg/L Me-Ph3P+ in a WWTP that treats wastewater of a larger university (Table S6 of the Supporting Information). It is very well-known that universities with chemistry faculties perform Wittig reactions, as well. R-Ph3P+ were detected in the Rhine because of the connected chemical industry. In water, SD, and SPM sampled upstream of RKI 443 (upstream of the industrial center of Ludwigshafen/Mannheim), only Et-Ph3P+ was found (Figure 5, Table 2, and Table S8). Hence, there was a substance specific source in the upstream part of the Upper Rhine Plain. On the left bank of the Rhine at Worms (RKI 443), 0.13 μg/L MePh3P+ were found, while this compound was not detectable on the right bank. Thus, it is very likely that Me-Ph3P+ was discharged upstream near Ludwigshafen, a city hosting large chemical industries. On the right bank of the Rhine at Kornsand (RKI 483), 0.56 μg/L MeOMe-Ph3P+ were detected, 14287

DOI: 10.1021/acs.est.5b03926 Environ. Sci. Technol. 2015, 49, 14282−14291

Article

Environmental Science & Technology

Table 2. Concentrations (micrograms per liter) of Identified R-Ph3P+ as well as DPPO, TPPO, and Carbamazepine (CBZ) in Surface Waters in the Rhine Catchment1 river LOQ Rhine Rhine Rhine Rhine Rhine Rhine Rhine Rhine Main Lahn Nahe Neckar Neckar Landgraben Landgraben Landgraben Landgraben Mühlbach Schwarzbach Schwarzbach Schwarzbach Scheidgraben Schlimmer Graben

RKI 410 r 443 l 443 r 483 l 483 r 520 l 550 l 590.3 l 1r 128 2r 24 r 13 r na na na na na na na na na na

city

Me-Ph3P+

Et-Ph3P+

MeOMe-Ph3P+

TPPO

DPPO

CBZ

Brühl Worms Worms Nierstein Kornsand Bingen Oberwesel Koblenz Mainz Nievern Bingen Heidelberg Ladenburg Weiterstadt Griesheim Berkach Trebur Groß-Gerau Nauheim Trebur Ginsheim Dornheim Büttelborn

0.015