Identification of Unknown Microcontaminants in Dutch River Water

Article pubs.acs.org/est

Identification of Unknown Microcontaminants in Dutch River Water by Liquid Chromatography-High Resolution Mass Spectrometry and Nuclear Magnetic Resonance Spectroscopy J.A. van Leerdam,*,†,# J. Vervoort,‡,⊥,# G. Stroomberg,§ and P. de Voogt†,∥ †

KWR, Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands Wageningen NMR Center, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands § Rijkswaterstaat, Postbus 17, 8200 AA Lelystad, The Netherlands ∥ Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Science Park 904, 1098 XH, Amsterdam, The Netherlands ⊥ Laboratory of Biochemistry, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands ‡

S Supporting Information *

ABSTRACT: In the past decade during automated surface water monitoring in the river Meuse at border station Eijsden in The Netherlands, a set of unknown compounds were repeatedly detected by online liquid chromatography-diode-array detection in a relatively high signal intensity. Because of the unknown nature of the compounds, the consequently unknown fate of this mixture in water treatment processes, the location being close to the water inlet of a drinking water supply company and their possible adverse public health effects, it was deemed necessary to elucidate the identity of the compounds. No data are available for the occurrence of these unknowns at downstream locations. After concentration and fractionation of a sample by preparative Liquid Chromatography, identification experiments were performed using Liquid Chromatography-High Resolution Mass Spectrometry (LC-HRMS) combined with High Resolution Nuclear Magnetic Resonance Spectroscopy (HR-NMR). Accurate mass determination of the unknown parent compound and its fragments obtained in MS/MS provided relevant information on the elemental composition of the unknown compounds. With the use of NMR techniques and the information about the elemental composition, the identity of the compounds in the different sample fractions was determined. Beside some regularly detected compounds in surface water, like caffeine and bisphenol-S, five dihydroxydiphenylmethane isomers were identified. The major unknown compound was identified as 4,4′-dihydroxy-3,5,3′,5′-tetra(hydroxymethyl)diphenylmethane. This compound was confirmed by analysis of the pure reference compound. This is one of the first studies that employs the combination of high resolution MS with NMR for identification of truly unknown compounds in surface waters at the μg/L level. Five of the seven identified compounds are unexpected and not contained in the CAS database, while they can be presumed to be products generated during the production of resins.

■

INTRODUCTION

important to monitor for these emerging substances on a regular basis.1−6 The GC-MS screening technique is only suitable for nonpolar, volatile, or semivolatile compounds. On the basis of retention time and mass spectral information, identification of nontargeted pollutants is performed. Unfortunately, in many cases analysis by GC-MS is not possible. For semipolar and polar compounds, LC is the preferred analytical separation technique, combined with diode-array detection (DAD) and

In countries where surface water is an important source for drinking water production, the quality of this source water needs to be monitored regularly. To detect the presence of compounds, monitoring programs using chromatographic techniques, usually gas chromatography (GC) and liquid chromatography (LC), combined with sensitive detection systems, such as diode array (DA) and mass spectrometry (MS), are used. However, every year many new chemicals are synthesized and produced on a preparative and even industrial scale which may end up in the environment and thereby also in drinking water resources. As these molecules may pose a threat to water quality as well as to drinking water production, it is © 2014 American Chemical Society

Received: Revised: Accepted: Published: 12791

June 10, 2014 October 1, 2014 October 8, 2014 October 8, 2014 dx.doi.org/10.1021/es502765e | Environ. Sci. Technol. 2014, 48, 12791−12799

Environmental Science & Technology

Article

(tandem) MS.7−10 Many drinking water companies in The Netherlands that use surface water as the drinking water resource use an LC-DAD system both for regular chemical monitoring of known compounds and as an early warning system for water intake protection.10−15 In addition, LC combined with MS is a powerful identification technique. With the introduction of high resolution accurate mass spectrometers (HR-MS), both inherent sensitivity in screening and the feature of accurate mass determination of the unknown compound provided useful information about the elemental composition.16,17 Unfortunately, due to the lack of large LC-accurate mass spectral libraries, identification of unknown compounds is often unsuccessful. Applications of LC-HR-MS in the environmental field have been reported, with the majority dealing with identification and confirmation of known target compounds and their degradation products at levels above the 10 μg/ L.18−21 A minimal amount of studies was found dealing with nontargeted analysis in other water types.22,23 Although the HR-MS approach brings the environmental chemists closer to the identification of an unknown contaminant,24 to propose a structure of the unknown compound observed at a level of ng−μg/L in environmental samples is in most instances still beyond their reach.25−27 With the aid of one- and two- (sometimes even three-) dimensional homo- and heteronuclear NMR is in principle possible to elucidate the identity as well as the 3D structure of a unknown organic molecule.28 Although NMR spectroscopy is a general detector for organic molecules, compared to MS, it is a relatively insensitive analytical technique. Hence, in the majority of the cases, it is not suitable for the relatively low concentration levels of compounds found in drinking water resources, such as surface waters and groundwater. Recent studies that employ the combination of accurate mass spectrometry and NMR-techniques focus on food processing, pharmacological control, or wastewater, where the concentrations of the unknown compounds are much higher than those encountered in environmental samples such as surface waters and where the compounds identified were either known compounds (e.g., pesticides or transformation products and impurities related to known products (pharmaceuticals29 or coating agents30). To the best of our knowledge, this combination has not been used for identification of unknown compounds in surface water at the low μg/L level. If degradation products of unknown precursors have to be identified and different matrices are involved, then the task of the analyst is much more demanding. This paper describes a technique study applied to the detection, identification, and structural confirmation of regularly appearing unknown contaminants in a single surface water sample from the river Meuse in The Netherlands. The unknown compounds are still irregularly present today and were first observed by an online LC-DAD monitoring systememployed along the river Meusein 2003. Because of the relatively high nominal concentration of the main compound (∼10 μg/L), unknown health properties and unknown behavior in the water treatment systems at the time of its discovery, the drinking water supply company had to stop the intake of water from the river. As these compounds where repeatedly found in the river causing the closure of the surface water intake, a protocol was devised to collect sufficient sample to allow for an extended structure elucidation study. This study was performed using data from the LTQ-Orbitrap MS and 600 MHz NMR systems after

purification and concentration of the components in the sample.

■

EXPERIMENTAL SECTION Online HPLC-UV Monitoring System. The measuring station in the Meuse at Eijsden (Netherlands) performs a round the clock early warning surveillance for both target and nontarget contaminants. To this end, river water is pumped up, filtered over a 0.2 μm filter, and delivered to the preconcentration units of respective analyzers. For LC-DAD analysis, preconcentration is performed online with a PROSPEKT solvent delivering unit (SPARK Holland, Emmen, The Netherlands), containing a SPE column packed with PLRP-S material (15−25 μm, Polymer Laboratories, Church-Stretton, U.K.). The preconcentration runs for 90 min at a rate of 1 mL/min. The analytical column is a 250 × 4.0 mm2 I.D. Supelcosil LC18-DB packed with 5 μm C18 material (Supelchem, Leusden, The Netherlands). The dimensions of the guard column are 4.0 × 3.0 mm2 packed with ODS (Phenomenex, Utrecht, The Netherlands). The LC-DAD system consists of a Agilent 1100 series (Waldbronn, Germany). Samples were spiked with the internal standard of chloroxuron at a level of 1 μg/L. A linear gradient of water with a 5 mM potassium phosphate buffer (pH 3.0) + 10% acetonitrile (A) and acetonitrile (B) from A-B (0% to 100%) in 60 min is used, with a flow of 1.0 mL/min. The analysis is performed twice every 24 h. As the presence of the contaminants became apparent, a grab sample of 40 L of unfiltered river water was collected for further research. Preparative Liquid Chromatography. In order to successfully identify unknown compounds using both HR-MS and NMR a purification of the unknown compounds was performed. To this end the 40 L grab sample was filtered over sea sand and 2 L aliquots were subsequently extracted separately using SPE-extraction with ENVI-CHROM P (Supelco, Bellafonte, PA, USA)-loaded columns consisting of 6 mL glass tube with frit, containing 250 mg of material. No pH adjustment of the water sample was performed. Methanol (5 mL) was used for desorption of the SPE cartridges. The 20 sample extracts were pooled and concentrated to 10 mL under a gentle stream of nitrogen. The final concentration factor is 4000 times. Chromatograpy of the analytes was carried out on a 5.0 μm Inertsil ODS-3 250 × 4.0 mm2 column (Varian, Middelburg, The Netherlands) at a flow rate of 0.9 mL/min and DADdetection. A linear gradient of methanol (5 to 100%) and ultrapure water was used in 50 min and held at this composition for an additional 10 min, the injection volume was 100 μL. The compounds of interest were identified by their UV-absorption spectra which were compared with those obtained at the measuring station. On the basis of the retention time of the interesting peaks in this chromatogram, a fractionation was performed between tr =15.00 and 27.00 min. A total number of 90 fractions were collected on a 96wells plate. Each fraction corresponds to 8 s of LC pump time (equivalent to 120 μL of LC-solvent). As a result of the small fraction size each chromatographic peak is divided over at least three fractions. The three fractions which contained the major part of the compound were pooled, a subsample (60 μL) was taken for LC-MS analysis, and the rest of the solution (300 μL) was concentrated to dryness by evaporation at 40 °C with a 12792

dx.doi.org/10.1021/es502765e | Environ. Sci. Technol. 2014, 48, 12791−12799


Article

10 min. For negative-ion measurements, a linear gradient of acetonitrile (5 to 100%) and water with 0.05% ammonium formiate was used with the same gradient profile as for positiveion measurements. The analytical column was re-equilibrated for 15 min between consecutive runs. The flow rate of the mobile phase was 0.3 mL/min. The LC column effluent was split via a postcolumn splitter from 0.3 mL/min to 0.125 mL/ min (which is the optimum flow for introduction into the source of the mass spectrometer) and the remaining flow was led to waste. Processing of the MS-Data. In this study, the calculated elemental compositions with a maximum deviation of 5 ppm from the measured exact mass were considered. Parameter settings were as follows: C 0−40, H 0−100, N 0−10, O 0−15, P 0−5, F 0−5; even electron ions for the precursor ions; odd and even electron ions for the product ions. The appropriate numbers of the elements Cl, Br, S and Si were determined from the specific isotope patterns and added if required. Isotope ratio 13 C/12C was checked in order to assign the number of C atoms. The double-bond-equivalent (DBE) parameter was used as an indicator of the stability (degree of π-electron conjugation) of the calculated elemental composition and was set dependent to the existence of a UV signal. Nuclear Magnetic Resonance Spectroscopy. The NMR measurements were performed on a 600 MHz Bruker Avance III NMR spectrometer with 5 mm cryoprobe. The entire samples were dissolved in 0.55 mL CD3OD in 5 mm Wilmad 535-PP-7 NMR tubes. Measurements were performed at 298 K. 1 H NMR spectra were obtained with 16 to 2000 scans depending on the concentration of the unknown compounds in the fraction measured. For the major fraction a 13C NMR spectrum was obtained. In addition, 2D 1H13C HSQC and 2D 1H13C HMBC spectra were measured with either 72 scans for the 1H direction and 800 increments for the 13C direction (HMBC) or 8 scans for the 1H direction and 512 increments for the 13C direction (HSQC). A total of 7 different fractions were measured. The HMBC as well as the HSQC measured were performed with gradient pulses for selection (pulse programmes hmbcgpndqf and hsqcgpph, respectively). Measurement times for the HSQC and HMBC spectra ranged from 4 h (HSQC) to 25 h for the HMBC. Since the compound of interest was present at a low concentration level and in an (aqueous) matrix, it was necessary to both isolate and concentrate the compound. Because of the high sensitivity of 1H NMR, impurities are more easily visible and it was observed that the combined related fractions of the repetitive preparative HPLC showed a background signal related to hydrophobic impurities. These impurities are emerging from the bleeding of the preparative HPLC column and with each purification, a small amount of C-18 related material ends up in the sample, which increases to relatively large amounts when samples are combined. Therefore, each combined sample was purified once more in one analytical HPLC run (Inertsil ODS-3) decreasing the amount of C-18 related column material. After isolation with preparative HPLC, first 1H NMR experiments were performed. Resonances related to this C18-bleeding (0.5 to 2.5 ppm) were ignored in the spectra.

solvent evaporator EZ-2 PLUS (Genevac, Ipswich, U.K.), and shipped to the NMR laboratory. An overview of the selected fractions with the highest UVabsorption at 210 nm which are used for LC-MS and NMRanalysis is shown in Table 1. Table 1. LC-MS Retention Time, Measured Accurate Mass in Full Scan Mode and Proposed Elemental Composition (Including Double Bond Equivalents, DBE, and Mass Error) of the Highest Peaks in the 7 Different Fractions Analyzed by LC-Orbitrap-MSa compound

Tr LCMS (min)

accurate mass [M − H]−

proposed elemental composition

DBE

mass error (ppm)

A B C D E F G

21.15 22.44 23.88 24.37 24.62 25.10 26.19

195.08692 (+) 319.11737 319.11749 317.10190 289.10712 249.02180 455.17020

C8H10N4O2 (+) C17H20O6 C17H20O6 C17H18O6 C16H18O5 C12H10O4S C25H28O8

6 8 8 9 8 8 12

−3.7 −4.2 −3.8 −3.7 −3.5 −3.6 −2.1

a

(+) only signal in the positive ion mode ([M + H]+)

Liquid Chromatography-DAD-Accurate Mass Spectrometry. The LC-DAD-LTQ-Orbitrap MS analysis was performed as described in Hogenboom et al.27 The LCDAD-LTQ-Orbitrap MS system consisted of a Surveyor autosampler model Plus, a Surveyor quaternary gradient LCpump, a Surveyor Photo Diode Array (DAD) model Plus detector, and an LTQ- Orbitrap mass spectrometer (Thermo Electron GmbH, Bremen, Germany). The system was equipped with an Ion Max Electrospray Ionization (ESI) probe and operated in the positive- and negative-ion mode. Full-scan accurate mass spectra (mass range from 100 to 1000 Da) were obtained at high resolution (set at 100 000 fwhm at mass 400 Da by the software), daily external mass calibration was used and the data were acquired and processed by using Xcalibur v.2.0 software. The electrospray source conditions were: capillary voltage 3.6 kV (positive-ion measurements), 2.5 kV (negative-ion measurements), heated capillary temperature 275 °C, and tube lens 70 V. The mass spectrometer was operated in a data-dependent-acquisition (DDA) mode in which both MS and low resolution MSn spectra (n = 5) were acquired without the need to specify parent masses, only the most intense ions were analyzed by MS/MS. The product ions were generated in the LTQ trap at a normalized collision energy setting of 35% and using an isolation width of 2 Da. In a separate experiment with direct injection (i.e., no chromatographic column used) of the sample fraction, the accurate mass of fragment ions from the observed precursor compound (m/z 319.12) were determined using MS2-MS5. The DAD spectral range was set from 200 to 350 nm. Five microliters of the final extract were injected into the LC system consisting of a 100 × 2.0 mm2 i.d. column packed with 3-μm Omnisphere C18 material (Varian-Chrompack, Middelburg, The Netherlands). The guard column was 4.0 × 2.0 mm i.d. packed with pellicular C18 material, 25−35 μm (VarianChrompack). The analytical column and the guard column were maintained at a temperature of 21 °C in a column thermostat. For positive-ion measurements, a linear gradient of acetonitrile (5 to 100%) and water with 0.05% formic acid was used in 40 min and held at this composition for an additional

■

RESULTS AND DISCUSSION Detection of the Unknown Compounds. On the 18th of April 2007, three major peaks were detected at tr = 12.38, 13.54,

12793



Article

Figure 1. Online SPE-LC-DAD chromatogram of 90 mL of river Meuse water. The retention times of the three major peaks are indicated and the corresponding absorption spectra are shown in the inset.

Figure 2. LC-UV chromatogram of the purified sample. The section between 20 and 30 min has been enlarged (inset) and the respective peaks from the samples that were fractionated are indicated.

labeled “A” through “G” based on their retention times. The chromatogram and the ensuing absorption spectra are shown in Figures 2 and 3. The difference in the number of compounds detected by the two techniques can be explained by the fact that the peaks in the original chromatogram are insufficiently resolved by the chromatography applied. The peaks are relatively wide and can easily be made up of multiple compounds.

and 18.75 min in the sample collected between 17.00 and 18.30 h as is shown in Figure 1. Their UV-absorption spectra are also shown in this figure. These compounds were first detected by an automated online liquid chromatography-diode-array detection system which was located on a pontoon in the river. The estimated concentration of the highest peak (13.54 min) related to the internal standard chloroxuron is ∼10 μg/L. The fractionation chromatography revealed the presence of no less than 7 individual compounds. The compounds are 12794



Article

DBE and mass error) of the highest peaks in the fractions analyzed by LC-LTQ-Orbitrap MS. The mass deviations for all seven compounds were negative, which is commonly observed at this mass range in the Orbitrap.31 For compound C, apart from the parent ion also a fragment ion was observed in full scan at m/z 301.10706. To obtain more structural information, additional MS/MS experiments were performed with the compound exhibiting the most intense UV-signal (only done for compound C) resulted in several fragments (see Table 2). The measured accurate mass of the two most intense peaks are m/z 319.11749 and m/z 301.10706. The ion m/z 319.11749 is most likely correlated to the deprotonated molecule, and m/z 301.10706 to a loss of water from the parent molecule. On the basis of the observed accurate mass of the unknown compound, theoretically more than 10 chemical formulas are possible if (by using the Xcalibur 2.0 software) a combination of the following elements is taken into account: C, H, O, N, F, and P. The elements Si, Cl, Br, and S are excluded because of the lack of characteristic isotopic patterns of these elements in the mass spectrum. In addition, since the compound absorbs UV light, the DBE was assumed to be higher than 4.0 (for deprotonated molecules). Assuming that the molecule does not contain nitrogen (since no signal was observed in positive ionization mode), the elemental compositions corresponding to m/z 319.11749 and m/z 301.10706, within a mass error of 5 ppm, are C17H20O6 (DBE = 8) and C17H18O5 (DBE = 8), respectively. For compound C Figure 4 presents the corresponding fragmentation pathway. As shown in Table 2, three MS3 ions (MS1−1−1 to MS1− 1−3), five MS4 ions (MS1−1−1−1 until MS1−1−1−3, MS1− 1−2−1, and MS1−1−3−1) and one MS5 ion (MS1−1−1−1− 1) were detected at high resolution and high mass accuracy. The most likely elemental compositions (i.e., corresponding to the elemental composition of the precursor) within a mass error of 5 ppm are presented. The elemental compositions of the product ions detected are all a part of the proposed elemental composition of the deprotonated ion (C17H20O6), and therefore this elemental composition seems to be correct. Information from 1H NMR Spectra. Although NMR is considered to be the most powerful identification technique, its drawback is its lack of sensitivity. Recent developments in probe design, especially cryoprobe technology, have increased sensitivity for organic samples to the low μg region for 1H NMR or 30−60 μg for 2D 1H13C HMBC spectra.32 The 1H NMR spectrum of the major unknown compound (compound

Figure 3. Absorption spectra of the compounds recorded after fractionation and subsequent purification.

The UV absorption spectra of compounds C and E have similar characteristics (maximum at 280 nm) as the spectra detected by the online measurement. Compounds B, D, and G have also UV absorption spectra with similar characteristics. To obtain more structural information about the 7 unknown compounds, the separated sample fractions after subsequent purification were analyzed by LC-LTQ-Orbitrap MS/MS. Information from High Resolution Mass Spectrometry. In the pooled fractions containing compound A, a signal was observed in the positive ionization mode whereas no signal was observed in the negative mode. For the pooled fractions containing compounds B−G, signals were observed only in the negative ionization mode, maybe due to a lack of proton affinity of the compounds. Therefore, it seems likely that these compounds do not contain any nitrogen atoms. In the negative ion mode, large signals were observed. Table 1 shows the different sample fractions, LC-MS retention times, measured accurate mass and proposed elemental composition (including

Table 2. Fragmentation Pathway of Compound C with Most Intense Fragments (Intensity >20%) and Predicted Elemental Composition MSn

identifier

accurate mass [M − H]−

elemental composition

mass error (ppm)

relative intensity (%)

Δ mass

composition of neutral loss

MS1 MS2 MS3

MS1 MS1−1 MS1−1−1 MS1−1−2 MS1−1−3 MS1−1−1−1 MS1−1−1−2 MS1−1−1−3 MS1−1−2−1 MS1−1−3−1 MS1−1−1−1−1−1

319.11749 301.10706 253.08609 237.09123 235.07550 223.07555 225.09122 235.07544 222.06805 207.08060 208.05275

C17H19O6 C17H17O5 C16H13O3 C16H13O2 C16H11O2 C15H11O2 C15H13O2 C16H11O2 C15H10O2 C15H11O C14H8O2

−3.8 −3.6 −3.7 −3.7 −4.1 −4.1 −3.9 −4.3 −2.6 −2.3 −1.1

100 50 50 100 95 60 100 100 100

18 48 64 66 30 28 18 15 28 15

H and OH OH and CH2OH OH, OH and CHOH CH3 and 3x OH CH2O CO H2O CH3 CO CH3

MS4

MS5

12795



Article

Figure 4. Fragmentation tree of the main compound (compound C). See Table 2 for more details.

Figure 5. Identified compound C in river Meuse water.

C) is shown in Figure S1 (Supporting Information, SI) and the 1 H13C HMBC spectrum in SI Figure S2. The 1H NMR spectrum of compounds D and E are shown in SI Figure S3. The 1H NMR spectrum of compound C shows besides the solvent resonances at 3.32 ppm (methanol) and 4.88 ppm (water) resonances at 6.99 ppm (singlet), 4.68 ppm (singlet), and at 3.81 ppm (singlet) with intensity ratios of 2 to 4 to 1 respectively. The presence of a resonance at 6.99 ppm indicates that an aromatic ring system is present, consistent with LC-UV data. This is corroborated by the 2D 1H13C HMBC spectrum which shows that the 13C resonance for the directly linked carbon of the proton at 6.99 ppm is at 127.0 ppm for 13 C. Moreover, the HMBC spectrum shows that the protons at 3.81 ppm have directly linked carbon (one bond) at 60.6 ppm and that the protons at 4.68 ppm have a directly linked carbon at 40.0 ppm. These latter two carbon resonances do not emerge from aromatic carbon but are from sp3 hybridized carbons with an electrophilic atom attached to the carbon at 60.6 ppm, which is probably an oxygen atom. The HMBC spectrum shows in addition that all proton resonances are indirectly coupled to each other, with 3 or possibly four chemical bonds in between

the hydrogen atoms. In addition it can be seen in the HMBC spectrum that there are two additional aromatic carbon resonances visible, with no hydrogen attached to these carbons. One of the carbon resonances is at 151.3 ppm, which is indicative for an aromatic carbon with oxygen directly attached to it. The other carbon resonance is at 132.8 ppm which is in the aromatic resonance region for carbons with directly bonded carbons. The other fractions analyzed all shown resonances in the aromatic region, with hyperfine structure (HH coupling) for a number of aromatic samples. In addition, all samples showed 1H NMR resonances in the 4.4 to 4.8 ppm region (all singlets) and singlet resonances around 3.8 ppm. No aliphatic 1 H NMR resonances were observed. Structure Proposals and Identification. HR-MS was used to obtain information about the accurate mass of the unknown compounds present in the surface water sample. The accurate mass of the compound C was 319.11737 and its most probable elemental composition C17H20O6. On the basis of the accurate masses of the product ions, OH and CH2OH specific groups were suggested as part of the molecule. Although the accurate mass measurements and elemental composition 12796



Article

Figure 6. Chemical structures of the 7 identified compounds present in river Meuse water.

probable structure based on this information is 4,4′-dihydroxy3,5,3′,5′-tetra(hydroxymethyl)diphenylmethane, as shown in Figures 5 and 6. The 13C NMR spectra of the proposed compound completely agreed with literature data.33 The identity of this compound was subsequently confirmed (level 1 according to Schymanski et al.24 by accurate mass MS and 1H-NMR analysis of the pure reference compound that was obtained by chemical synthesis (Mercachem, The Netherlands). To identify the other compounds in the different fractions, the availability of compound databases and mass spectral libraries which can be searched are very helpful. With the use of an in-house created MS-library with accurate precursor masses and retention times it was possible to identify the main compound in MS-fraction A. In this fraction, only a signal with positive ionization was obtained and the compound was identified as caffeine (C8H10N4O2). Compound F was

calculation of the unknown compound in combination with specific fragments brings one closer to the identity of the unknown compound, techniques such as NMR are still necessary to elucidate the complete chemical structure. When using an elemental formula of C17H20O6, more than 100 different compounds are possible (www.chemspider.com), hence it is clear that NMR is often indispensable for unique identification. The NMR spectrum obtained (see Figure 5) clearly corresponds to a symmetrical structure. Additionally, an aromatic structure with 3 different types of hydrogen atoms in the ratio 1−2−4 was suggested, no aliphatic resonances (CH2CH2 or CH3) were observed. On the basis of the elemental composition of C17H20O6, two identical backbones consisting of aromatic C8H9O3 linked by a CH2-group to cope with the appropriate number of DBEs, was suggested. The three different types of hydrogen were suggested as aromatic hydrogen, hydroxyl (OH), and methylol (CH2OH). The most 12797



■

identified as bisphenol-S (C12H10O4S). Hence, compounds A and F are identified with the reliability corresponding to level 1.24 Both compounds are detected regularly in surface water.26 On the basis of the combined information available from both LC-LTQ-Orbitrap-MS and NMR data, structures of the other unknown compounds are proposed and these are presented in Figure 6. A total number of four isomers and derivatives (i.e., compounds B, D, E, and G) of the main compound (4,4′-dihydroxy-3,5,3′,5′-tetra(hydroxymethyl)diphenylmethane, compound C) were subsequently identified in the different fractions. The identity of these 4 derivatives of the main compound are not confirmed by their pure reference compounds, due to the high costs related to their synthesis. This corresponds to level 2 of the Schymanski et al. classification24 These compounds are found as intermediates in the cure process of cresol phenol-formaldehyde (PF) resins. PF-resins, the first totally synthetic polymers, have wide commercial applications as molding compounds, coatings, wood binders, and high-temperature ablatives.34−37 The first compounds formed in base-catalyzed PF condensation reactions are mono, di-, and trimethylol (CH2OH) derivates of phenol. Condensation of these methylolphenols with themselves or with another molecule of phenol results in dihydroxydiphenylmethanes (HOAr)2CH2 that contain methylol groups.38−40 Additional methylol groups can be added to these dihydroxydiphenylmethanes by further condensation reactions with formaldehyde. Currently, the authors have no tangible information about the possible source of the unknown compounds. It should be noted that upstream of the sampling location a highly industrialized area is located that includes resin production plants. After the identification of compound C, the compound was added to the calibration set used in the monitoring program employing LC-DAD. In the period November 2013- August 2014 the compound was observed in total of 11 cases with a highest observed concentration of 1.6 μg/l. The present study demonstrates that accurate mass HR-MS and NMR can be successfully used jointly to elucidate the structure of unknown and unexpected compounds present in river water at low concentrations. The novelty of the present work is the inclusion of NMR in environmental monitoring. Internet searches revealed that five out of the seven compounds identified do not have a CAS number and are not commercially available. No experimental data on physicochemical properties of these compounds can be found. Considering the high predicted solubility of compound C (>50 g/L), its relatively low predicted log Kow value of −0.2134−37 and those of its derivatives, removal of these compounds during drinking water preparation can be expected to be difficult. Further investigation is required of the removal rates in advanced treatment processes of these synthetic polymethylol phenols that are generated during the production of resins. Although elaboration on the potential toxicity of these compounds is beyond the scope of the present study, these findings are interesting because they are totally unexpected.

■

Article

AUTHOR INFORMATION

Corresponding Author

*Tel: +31 (0)30 60 69 624; fax: +31 (0)30 60 61 165; e-mail: [email protected]. Author Contributions #

These authors contributed equally to this manuscript.

Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Schriks, M.; Heringa, M. B.; van der Kooi, M. M. E.; de Voogt, P.; van Wezel, A. P. Toxicological relevance of emerging contaminants for drinking water quality. Water Res. 2010, 44 (2), 461−476. (2) Petri, M.; Jiang, J. Q.; Maier, M. Proficiency test of non-target screening with gas chromatography mass spectrometry to confirm a detected contamination of raw and drinking water. Water Sci. Technol.: Water Supply 2010, 10, 806−814. (3) Ma, H.; Zhang, H.; Wang, L.; Wang, J.; Chen, J. Comprehensive screening and priority ranking of volatile organic compounds in Daliao River, China. Environ. Monit. Assess. 2014, 1−9. (4) Camino-Sánchez, F. J.; Zafra-Gómez, A.; Ruiz-García, J.; Vílchez, J. L. Screening and quantification of 65 organic pollutants in drinking water by stir bar sorptive extraction-gas chromatography-triple quadrupole mass spectrometry. Food Anal. Methods 2013, 6 (3), 854−867. (5) Civil, W. In Routine trace level screening for a broad range of volatile and semi-volatile organic pollutants in a range of environmental waters using f ull scan GC-MS with a 1000+ compound target screening database and deconvolution software; 2012. (6) Gómez, M. J.; Gómez-Ramos, M. M.; Agüera, A.; Mezcua, M.; Herrera, S.; Fernández-Alba, A. R. A new gas chromatography/mass spectrometry method for the simultaneous analysis of target and nontarget organic contaminants in waters. J. Chromatogr. A 2009, 1216 (18), 4071−4082. (7) Hogenboom, A. C.; Niessen, W. M. A.; Brinkman, U. A. T. The role of column liquid chromatography-mass spectrometry in environmental trace-level analysis: Determination and identification of pesticides in water. J. Sep. Sci. 2001, 24 (5), 331−354. (8) Bobeldijk, I.; Broess, K.; Speksnijder, P.; Van Leerdam, T. Determination of the herbicide amitrole in water with pre-column derivatization, liquid chromatography and tandem mass spectrometry. J. Chromatogr. A 2001, 938 (1−2), 15−22. (9) Reemtsma, T. The use of liquid chromatography-atmospheric pressure ionization-mass spectrometry in water analysis - Part II: Obstacles. TrAC - Trends in Analytical Chemistry 2001, 20 (10), 533− 542. (10) Bobeldijk, I.; Stoks, P. G. M.; Vissers, J. P. C.; Emke, E.; Van Leerdam, J. A.; Muilwijk, B.; Berbee, R.; Noij, T. H. M. Surface and wastewater quality monitoring: Combination of liquid chromatography with (geno)toxicity detection, diode array detection and tandem mass spectrometry for identification of pollutants. Journal of Chromatography A 2002, 970 (1−2), 167−181. (11) Slobodnik, J.; Brouwer, E. R.; Geerdink, R. B.; Mulder, W. H.; Lingeman, H.; Brinkman, U. A. T. Fully automated on-line liquid chromatographic separation system for polar pollutants in various types of water. Anal. Chim. Acta 1992, 268 (1), 55−65. (12) Brinkman, U. A. T.; Slobodnik, J.; Vreuls, J. J. Trace-level detection and identification of polar pesticides in surface water: The SAMOS approach. TrAC - Trends in Analytical Chemistry 1994, 13 (9), 373−381. (13) Slobodnik, J.; Groenewegen, M. G. M.; Brouwer, E. R.; Lingeman, H.; Brinkman Th, U. A. Fully automated multi-residue method for trace level monitoring of polar pesticides by liquid chromatography. J. Chromatogr. 1993, 642 (1−2), 359−370. (14) Purvis, M.; Gledhill, S. J.; Drage, B. E.; Hogenboom, A. C.; Jolly, D. A.; Vreuls, R. J. J.; Brinkman, U. A. T. Determination and

ASSOCIATED CONTENT

S Supporting Information *

(1) 1H-NMR spectrum of compound C. (2) 2D 1H13C HNBC spectrum of compound C. (3) 1H-NMR spectrum of compound D and E. This material is available free of charge via the Internet at http://pubs.acs.org. 12798



Article

monitoring of polar compounds and acidic herbicides using a modified SAMOS system. Int. J. Environ. Anal. Chem. 1999, 74 (1−4), 135−154. (15) De Hoogh, C. J.; Wagenvoort, A. J.; Jonker, F.; Van Leerdam, J. A.; Hogenboom, A. C. HPLC-DAD and Q-TOF MS techniques identify cause of Daphnia biomonitor alarms in the River Meuse. Environ. Sci. Technol. 2006, 40 (8), 2678−2685. (16) Hogenboom, A. C.; Niessen, W. M. A.; Little, D.; Brinkman, U. A. T. Accurate mass determinations for the confirmation and identification of organic microcontaminants in surface water using on-line solid-phase extraction liquid chromatography electrospray orthogonal-acceleration time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 1999, 13 (2), 125−133. (17) Maizels, M.; Budde, W. L. Exact mass measurements for confirmation of pesticides and herbicides determined by liquid chromatography/time-of-flight mass spectrometry. Anal. Chem. 2001, 73 (22), 5436−5440. (18) Hogenboom, A. C.; Niessen, W. M. A.; Brinkman, U. A. T. Characterization of photodegradation products of alachlor in water by on-line solid-phase extraction liquid chromatography combined with tandem mass spectrometry and orthogonal-acceleration time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 2000, 14 (20), 1914−1924. (19) Ibáñez, M.; Sancho, J. V.; Pozo, Ó . J.; Hernández, F. Use of quadrupole time-of-flight mass spectrometry in environmental analysis: elucidation of transformation products of triazine herbicides in water after UV exposure. Anal. Chem. 2004, 76 (5), 1328−1335. (20) Herrera Lõpez, S.; Hernando, M. D.; Gõmez, M. J.; SantiagoMorales, J.; Rosal, R.; Fernández-Alba, A. R. Investigation of galaxolide degradation products generated under oxidative and irradiation processes by liquid chromatography/hybrid quadrupole time-of-flight mass spectrometry and comprehensive two-dimensional gas chromatography/time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 2013, 27 (11), 1237−1250. (21) Gómez, M. J.; Gómez-Ramos, M. M.; Malato, O.; Mezcua, M.; Férnandez-Alba, A. R. Rapid automated screening, identification and quantification of organic micro-contaminants and their main transformation products in wastewater and river waters using liquid chromatography-quadrupole-time-of-flight mass spectrometry with an accurate-mass database. J. Chromatogr. A 2010, 1217 (45), 7038− 7054. (22) Godejohann, M.; Heintz, L.; Daolio, C.; Berset, J. D.; Muff, D. Comprehensive non-targeted analysis of contaminated groundwater of a former ammunition destruction site using 1H-NMR and HPLC-SPENMR/TOF-MS. Environ. Sci. Technol. 2009, 43 (18), 7055−7061. (23) Godejohann, M.; Berset, J. D.; Muff, D. Non-targeted analysis of wastewater treatment plant effluents by high performance liquid chromatography-time slice-solid phase extraction-nuclear magnetic resonance/time-of-flight-mass spectrometry. J. Chromatogr. A 2011, 1218 (51), 9202−9209. (24) Schymanski, E. L.; Jeon, J.; Gulde, R.; Fenner, K.; Ruff, M.; Singer, H. P.; Hollender, J. Identifying small molecules via high resolution mass spectrometry: Communicating confidence. Environ. Sci. Technol. 2014, 48 (4), 2097−2098. (25) Gozlan, I.; Rotstein, A.; Avisar, D. Amoxicillin-degradation products formed under controlled environmental conditions: Identification and determination in the aquatic environment. Chemosphere 2013, 91 (7), 985−992. (26) Bobeldijk, I.; Vissers, J. P. C.; Kearney, G.; Major, H.; Van Leerdam, J. A. Screening and identification of unknown contaminants in water with liquid chromatography and quadrupole-orthogonal acceleration-time-of-flight tandem mass spectrometry. J. Chromatogr. A 2001, 929 (1−2), 63−74. (27) Hogenboom, A. C.; van Leerdam, J. A.; de Voogt, P. Accurate mass screening and identification of emerging contaminants in environmental samples by liquid chromatography-hybrid linear ion trap Orbitrap mass spectrometry. J. Chromatogr. A 2009, 1216 (3), 510−519.

(28) Moco, S.; Vervoort, J.; Moco, S.; Bino, R. J.; De Vos, R. C. H.; Bino, R. Metabolomics technologies and metabolite identification. TrACTrends Anal. Chem. 2007, 26 (9), 855−866. (29) Awasthi, A.; Razzak, M.; Al-Kassas, R.; Greenwood, D. R.; Harvey, J.; Garg, S. Isolation and characterization of degradation products of moxidectin using LC, LTQ FT-MS, H/D exchange and NMR. Anal. Bioanal. Chem. 2012, 404 (8), 2203−2222. (30) Ackerman, L. K.; Noonan, G. O.; Begley, T. H.; Mazzola, E. P. Accurate mass and nuclear magnetic resonance identification of bisphenolic can coating migrants and their interference with liquid chromatography/tandem mass spectrometric analysis of bisphenol A. Rapid Commun. Mass Spectrom. 2011, 25 (9), 1336−1342. (31) Bijlsma, L.; Emke, E.; Hernández, F.; De Voogt, P. Performance of the linear ion trap Orbitrap mass analyzer for qualitative and quantitative analysis of drugs of abuse and relevant metabolites in sewage water. Anal. Chim. Acta 2013, 768 (1), 102−110. (32) Exarchou, V.; Godejohann, M.; Van Beek, T. A.; Gerothanassis, I. P.; Vervoort, J. LC-UV-solid-phase extraction-NMR-MS combined with a cryogenic flow probe and its application to the identification of compounds present in Greek oregano. Anal. Chem. 2003, 75 (22), 6288−6294. (33) Fisher, T. H.; Chao, P.; Upton, C. G.; Day, A. J. Spectral assignments and reference data: A 13C NMR study of the methylol derivatives of 2,4′- and 4,4′-dihydroxydiphenylmethanes found in resol phenol-formaldehyde resins. Magn. Reson. Chem. 2002, 40 (11), 747− 751. (34) Fang, Q.; Cui, H. W.; Du, G. B. Montmorillonite reinforced phenol formaldehyde resin: Preparation, characterization, and application in wood bonding. Int. J. Adhes. Adhes. 2014, 49, 33−37. (35) Hirano, K.; Asami, M. Phenolic resins100 years of progress and their future. React. Funct. Polym. 2013, 73 (2), 256−269. (36) Takeichi, T.; Furukawa, N. Epoxy Resins and Phenol-Formaldehyde Resins; Elsevier: New York, 2012; Vol. 5, pp 723−751 (37) Knop, A.Phenolic Resins in Europe; Springer: New York, 1983; p 160. (38) Grenier-Loustalot, M. F.; Larroque, S.; Grande, D.; Grenier, P.; Bedel, D. Phenolic resins: 2. Influence of catalyst type on reaction mechanisms and kinetics. Polymer 1996, 37 (8), 1363−1369. (39) Grenier-Loustalot, M. F.; Larroque, S.; Grenier, P.; Bedel, D. Phenolic resins: 3. Study of the reactivity of the initial monomers towards formaldehyde at constant pH, temperature and catalyst type. Polymer 1996, 37 (6), 939−953. (40) Grenier-Loustalot, M. F.; Larroque, S.; Grenier, P.; Bedel, D. Phenolic resins: 4. Self-condensation of methylolphenols in formaldehyde-free media. Polymer 1996, 37 (6), 955−964.

12799


Identification of Unknown Microcontaminants in Dutch River Water

Recommend Documents